2009-05-01
[17:02:00.0000] <Philip`>
Hixie: I'm here now

[17:02:01.0000] <Philip`>
Hixie: I don't know; what do browsers do there? :-)

[17:02:02.0000] <Hixie>
forget the earlier thing, i just sent new mail on a different topic

[17:02:03.0000] <Hixie>
actually, same topic, different issue

[17:02:04.0000] <Hixie>
i hope the change i made to the spec still applies

[17:02:05.0000] <Hixie>
it's what you suggested in january

[17:05:00.0000] <Philip`>
Hmm, I've got no idea what the stroke.prune.arc test is doing

[17:07:00.0000] <Hixie>
yeah i wasn't too sure either

[17:07:01.0000] <Philip`>
Hixie: The spec change sounds plausible to me, except for the bit that says "then, then"

[17:08:00.0000] <Hixie>
i don't know what you're talking about *fixy fixy*

[17:08:01.0000] <Philip`>
/me tries to remember what arcTo's arguments are

[17:09:00.0000] <Hixie>
there's an authoring section bit now that explains it nice and friendlyly

[17:09:01.0000] <Philip`>
/me looks at the non-nice and unfriendly section instead

[17:09:02.0000] <Hixie>
i'm sure someone will send me diagrams and stuff explaining what they all do at some point and then the spec will have diagrams too

[17:11:00.0000] <Philip`>
Ah, right - the arcTo bit in stroke.prune.arc is causing a line to be drawn from (50,25) to (50,25) and then to (50,25), and then it should (according to my brain when I wrote the test, which may or may not match the spec or reality or sanity) collapse all the zero-length lines so there's just a single point

[17:11:01.0000] <Philip`>
and then the stroke() won't do anything, and particularly won't draw the round caps/joins

[17:12:00.0000] <Philip`>
And the bit with arc is doing the same, except that it's drawing a curved line between (50,25) and (50,25)

[17:14:00.0000] <Philip`>
Ooh, excellent, "Zero-length line segments must be pruned before stroking a path." and "Subpaths with fewer than two points are ignored when painting the path." makes it sounds like those paths should indeed be not drawn

[17:18:00.0000] <Philip`>
Anyway, changing the spec for the stroke.prune.arc thing seems like a lot of effort to introduce a weird glitch that nobody's going to care about at all

[17:20:00.0000] <Philip`>
So I guess I'd suggest waiting until browsers pass 99% of the tests and want to get 100%, and they'll either fix their handling of this one or they'll complain that it's needlessly hard and the spec should change, and something can be sorted out then

[17:21:00.0000] <Philip`>
(It's quite a spectacularly pointless feature to bother testing, I think)

[17:21:01.0000] <Philip`>
(So I suppose a third possible outcome is that the test could be deleted)

[17:23:00.0000] <Hixie>
nooo don't delete the test

[17:24:00.0000] <Philip`>
I won't delete it in the near future, because my test deployment process requires far too much effort

[17:24:01.0000] <Philip`>
I haven't even got around to uploading fixes for some blatant bugs in a couple of test cases

[17:26:00.0000] <Philip`>
/me goes back to playing Braid, and getting annoyed that the princess keeps exploding

[17:43:00.0000] <Hixie>
shepazu: dude you're talking to yourself on twitter. :-P

[17:54:00.0000] <Hixie>
Philip`: you still there?

[17:54:01.0000] <Hixie>
Philip`: does http://damowmow.com/temp/miter make any sense to you?

[17:55:00.0000] <Philip`>
I'm not here, but will be soon

[18:17:00.0000] <Philip`>
Hixie: Hard to tell without seeing the test case

[18:21:00.0000] <Philip`>
Hixie: But I'm pretty sure the spec is correct

[18:21:01.0000] <Philip`>
http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Ccanvas%20id%3Dc%3E%3C%2Fcanvas%3E%3Cscript%3Eonload%3Dfunction()%7Bvar%20ctx%3Ddocument.getElementById('c').getContext('2d')%3Bctx.lineWidth%3D20%3Bctx.miterLimit%3D1.41%3Bctx.moveTo(10%2C10)%3Bctx.lineTo(50%2C10)%3Bctx.lineTo(50%2C50)%3Bctx.stroke()%7D%3C%2Fscript%3E

[18:22:00.0000] <Philip`>
If you change the 1.41 to 1.42 then it starts drawing the miter

[18:22:01.0000] <Philip`>
and the miter length (from where the lines meet, to the corner of the miter) is lineWidth/2 * sqrt(2)

[18:23:00.0000] <Philip`>
so the relevant factor in the miter limit ratio is lineWidth/2

[18:24:00.0000] <Philip`>
unless I'm horribly mistaken

[18:30:00.0000] <Philip`>
(It's odd that all APIs seem to not mention the half)

[18:30:01.0000] <Philip`>
(but I can't see how I'm wrong)

[18:34:00.0000] <Hixie>
Philip`: so where the spec says the "inside of the join" it should say "the point where the join occurs"?

[18:38:00.0000] <Philip`>
Hixie: I interpret both those phrases in the same way, i.e. the point where the infinitesimally-thin lines intersect

[18:39:00.0000] <Philip`>
Looks like the spec used to say "the point of the join itself (where the lines touch on the inside of the join)"

[18:39:01.0000] <Philip`>
but someone must have changed it

[18:42:00.0000] <Hixie>
i'll strike "inside of the join"

[18:42:01.0000] <Hixie>
i don't understand what it means

[18:44:00.0000] <Philip`>
Hmm, I guess if the miter length was defined as being between the points where the edges of the (thick) stroke intersect, instead of between one of those points and the midpoint, then that'd remove the need to say 'half the line width' instead of 'the line width'

[18:45:00.0000] <Philip`>
but that seems a more complex definition

[18:46:00.0000] <Hixie>
walk me through how you get lineWidth/2 * root-2. i keep getting lineWidth/root-2.

[18:47:00.0000] <Hixie>
the miter length is the square root of (lineWidth/2) times two, right?

[18:47:01.0000] <Hixie>
maths is hard.

[18:47:02.0000] <Hixie>
:-P

[18:49:00.0000] <Philip`>
http://canvaspaint.org/57ce.png

[18:49:01.0000] <Philip`>
The red lines are the two lines being drawn; the blue lines are the outlines of their strokes

[18:49:02.0000] <Philip`>
(so the blue lines are lineWidth/2 away from the red lines)

[18:49:03.0000] <Hixie>
that diagram is exactly what i have on my pad here :-)

[18:50:00.0000] <Philip`>
The green triangle is the miter triangle

[18:50:01.0000] <Philip`>
The yellow line is the miter length

[18:50:02.0000] <Philip`>
The yellow line is the diagonal of a square that's lineWidth/2 x lineWidth/2

[18:50:03.0000] <Philip`>
so its length is sqrt(2) * lineWidth/2

[18:52:00.0000] <Hixie>
yeah apparently my ability to take roots of fractions is, ah, rusty

[18:52:01.0000] <hsivonen>
so I broke visual Hebrew accidentally

[18:53:00.0000] <hsivonen>
I wonder which spec defines it...

[18:53:01.0000] <Philip`>
/me didn't take roots of fractions, he just scaled sqrt(2) by the size of the square :-)

[18:53:02.0000] <Hixie>
clearly i need to take a basic algebra refresher :-P

[18:53:03.0000] <Hixie>
anyway

[18:53:04.0000] <Hixie>
thanks

[18:54:00.0000] <Philip`>
It'd be possible to draw the yellow line twice as long, so it reaches the intersection of the blue lines, which might be how other people define the miter length so that they don't need the divide-by-two

[18:55:00.0000] <Philip`>
Actually, I'm not sure that's true

[18:55:01.0000] <Philip`>
Actually, I think it is

[19:01:00.0000] <MikeSmith>
hsivonen: what's visual Hebrew?

[19:02:00.0000] <Philip`>
Is there non-visual Hebrew?

[19:06:00.0000] <Philip`>
Oh, I suppose there's spoken Hebrew

[19:06:01.0000] <Philip`>
/me completely forgot about that

[19:17:00.0000] <takkaria>
Philip`: too long on IRC. :)

[20:59:00.0000] <Hixie>
Philip`: having eaten dinner i worked out why i was having issues earlier. for some reason it escaped my attention that 1/(root2) == (root2)/2

[20:59:01.0000] <Hixie>
i feel better now

[21:13:00.0000] <boblet>
Hi all, I’m wondering about how to mark up several images with captions which are linked. In HTML4 I’d use a <ul> and wrap the img and text in an <a>. In HTML5 <figure> can’t be inside an <a>. Would I have to wrap the image and legend in separate links?

[21:15:00.0000] <Hixie>
you mean you have several different captioned images, each of which is a link?

[21:15:01.0000] <Hixie>
or you have one captioned link consisting of several images?

[21:16:00.0000] <boblet>
wow, Hixie. I’m honored :) the first. you can see here (at the bottom): http://oli-studio.com/work/wde/200905-html5/

[21:17:00.0000] <Hixie>
hmm

[21:17:01.0000] <boblet>
I figured to just wrap each image/legend in <figure> then wrap the text in <legend>, but that’s invalid: http://oli-studio.com/work/wde/200905-html5+/

[21:17:02.0000] <boblet>
am I over-semanticising? :D

[21:18:00.0000] <Hixie>
i'm not sure those are really <figure>s per se, i'd just use <nav><ul><li><a href="..."><img src="..." alt=""> ...</a></li> ... </ul></nav> personally

[21:19:00.0000] <boblet>
aah I thought as much. so <figure> is more pullquote-style images (in a magazine sense)?

[21:19:01.0000] <Hixie>
yeah

[21:19:02.0000] <Hixie>
more like the kind of thing you'd expect to be labelled "Figure 1." somewhere

[21:20:00.0000] <boblet>
sorted. Thanks for that! OK off to see what else I can HTML5-ify

[21:21:00.0000] <Hixie>
:-)

[21:21:01.0000] <Hixie>
btw the alt="" texts on the "John" images at the top are probably wrong

[21:21:02.0000] <Hixie>
those look decorative, probably best to have title="(what you have now in alt)" with alt="" (blank)

[21:22:00.0000] <boblet>
ooh thanks! I need to read up on alt text recommendations, so thanks for the prompt

[21:24:00.0000] <Hixie>
one way to test is simply to turn off images in firefox

[21:24:01.0000] <Hixie>
and to see whether it still looks like the page you'd wanted it to be

[21:24:02.0000] <Hixie>
basically "what would you do if you weren't allowed to have images"

[21:26:00.0000] <boblet>
hah, cute! the rotated images turn into shrink-wrapped rotated text boxes in FF3.5b

[21:50:00.0000] <Hixie>
anyone have any opinions on <input type=tel>?

[21:51:00.0000] <takkaria>
the only practical use I could see for it would be autofill or maybe on OS X integration with the address book

[21:52:00.0000] <takkaria>
opinion over. :)

[22:17:00.0000] <zcorpan>
Hixie: "+  &lt;h2>Games&lt;/h2> &lt;!-- this starts a second subsection -->" -- s/second/third/

[22:32:00.0000] <Hixie>
zcorpan: thanks

[22:32:01.0000] <Hixie>
takkaria: ta

[22:32:02.0000] <hsivonen>
MikeSmith: visual Hebrew is Hebrew ordered left to right with forced line breaks

[22:32:03.0000] <MikeSmith>
Hixie: I see

[22:33:00.0000] <Hixie>
what do you see?

[22:40:00.0000] <MikeSmith>
I see bursts of color all around me

[22:40:01.0000] <MikeSmith>
with forced line breaks

[22:40:02.0000] <MikeSmith>
I do wonder why that is called "visual Hebrew"

[22:41:00.0000] <MikeSmith>
seems like it should be called reverse-ordered Hebrew or something

[22:42:00.0000] <hsivonen>
MikeSmith: it's a hack that gives the right visual effect but is logically wrong

[22:43:00.0000] <MikeSmith>
hsivonen: how come it broke?

[22:44:00.0000] <MikeSmith>
could it affect other writing systems?

[22:48:00.0000] <hsivonen>
MikeSmith: I don't know why it broke

[22:49:00.0000] <hsivonen>
MikeSmith: visual Hebrew is unique, so I doubt anything else is affected by this particular bug

[22:49:01.0000] <hsivonen>
I suppose I'm responsible for setting a bidi override flag somewhere

[23:00:00.0000] <boblet>
Philip`: I subset the M+ font; “Original font: 586040 bytes (3985 glyphs), Optimized font: 21968 bytes (94 glyphs), Saved 96.3%!” 586KB -> 22KB. Wow

[23:00:01.0000] <boblet>
/me <3 Philip`

[23:00:02.0000] <boblet>
hehe

[23:01:00.0000] <MikeSmith>
Hixie: sorry, that "I see" before was to hsivonen 

[23:01:01.0000] <MikeSmith>
I just took the first "h" that completed in my client

[23:19:00.0000] <Hixie>
heh

[23:35:00.0000] <zcorpan>
Philip`: you should make your font optimizer be keyboard accessible

[00:13:00.0000] <Hixie>
heycam: what is it that exceptions implement, if not an interface?

[00:14:00.0000] <heycam>
Hixie, in Web IDL exceptions are distinct from interfaces

[00:14:01.0000] <heycam>
in the ecmascript binding, they're pretty similar though

[00:14:02.0000] <Hixie>
sure but what's the right terminology?

[00:15:00.0000] <Hixie>
bla bla these exceptions implement the SQLException ...what?

[00:15:01.0000] <heycam>
i'd say "bla bla these exceptions are SQLExceptions"

[00:15:02.0000] <Hixie>
hm interesting

[00:16:00.0000] <Hixie>
okie dokie

[00:16:01.0000] <Hixie>
thanks

[00:16:02.0000] <heycam>
it's not possible for an object to be more than on exception, as currently written

[00:16:03.0000] <heycam>
*than one

[00:16:04.0000] <heycam>
ok but i find this in web idl: "A host exception object  is a host object that implements a given exception. The behavior of a host object that implements more than one exception, or an exception as well as an interface, is not defined."

[00:17:00.0000] <heycam>
so go ahead and say it implements the exception if you want =)

[00:17:01.0000] <Hixie>
i prefer to call the object the exception

[00:17:02.0000] <Hixie>
otherwise we can't throw exceptions

[00:17:03.0000] <Hixie>
but only objects that implement exceptions

[00:17:04.0000] <Hixie>
and that's just confusing :-)

[00:18:00.0000] <heycam>
sure.

[00:18:01.0000] <heycam>
i might reword that bit about implementing exceptions

[00:18:02.0000] <zcorpan>
Philip`: by hiding the radio buttons in some way other than display:none (that preferably works sensibly with spatnav), and then input:focus + span { color:red }

[00:18:03.0000] <Hixie>
heycam: k

[01:27:00.0000] <Hixie>
zcorpan: oh, i misunderstood the svg thing. i thought it was an <img> pointing to svg, not an <svg>.

[01:29:00.0000] <zcorpan>
Hixie: "A class value instead of an align value doesn't seem especially more complex, especially given that it also enables a lot more than alignment (like padding, colours, fonts, etc)." -- td[align=right] { color:green }

[01:29:01.0000] <Hixie>
zcorpan: td.numeric { color: green; } is no more complex

[01:30:00.0000] <zcorpan>
Hixie: but td.numeric { align:right } <td class="numeric"> is more complex than <td align="right">

[01:30:01.0000] <zcorpan>
Hixie: and the former doesn't enable anything that you can't do with the latter

[01:31:00.0000] <Hixie>
by that argument we should allow <font> and bgcolor="" and so on

[01:32:00.0000] <zcorpan>
i'm not necesserily arguing that align should be allowed, i'm just saying your argument against it was moot :-)

[01:33:00.0000] <zcorpan>
align="right" could represent numeric like <small> represents small print

[01:33:01.0000] <Hixie>
*shrug*

[01:34:00.0000] <zcorpan>
(and equally <font color> could represent emphasis)

[01:34:01.0000] <Hixie>
the argument was that there is value to be had from being hardline no-presentational-markup, and that it wasn't notably more complex to use css for this

[01:35:00.0000] <Hixie>
bed time now though

[01:36:00.0000] <Hixie>
nn :-)

[01:36:01.0000] <zcorpan>
nn

[02:07:00.0000] <Philip`>
boblet: I hope that subsetted font doesn't crash any browsers :-)

[02:08:00.0000] <Philip`>
zcorpan: Hmm, good point

[02:08:01.0000] <Philip`>
zcorpan: (I probably need to redesign the whole UI anyway, particularly to support non-English languages sensibly, though I've currently got no idea how to do that)

[02:08:02.0000] <boblet>
Philip`: now he tells me! Testing in progress…

[02:09:00.0000] <Philip`>
boblet: (Actually it's probably more about OSes than about browsers)

[03:14:00.0000] <zcorpan>
annevk5: your blog is broken

[03:20:00.0000] <MikeSmith>
<annevk> @DreamHost. WTF. My site has gone haywire.

[03:20:01.0000] <MikeSmith>
(on twitter)

[03:26:00.0000] <Lachy_>
wtf is <hgroup> for and what problem does it solve?!

[03:29:00.0000] <jgraham>
Lachy: <hgroup> is the new name for <header>

[03:29:01.0000] <jgraham>
It is solving the problem that <header> as previously defined would have been useless in practice

[03:29:02.0000] <jgraham>
Because people would have used it incorrectly too often

[03:29:03.0000] <Lachy>
why would it have been useless?

[03:30:00.0000] <Lachy>
<header> is still in there, and I don't see how the introduction of <hgroup> solves that problem

[03:31:00.0000] <jgraham>
<header> now means something different

[03:31:01.0000] <Lachy>
IIRC, the problem was that people did silly things like <section><header>headings</header><section>content</section></section>

[03:32:00.0000] <jgraham>
Lachy: That was a slightly different problem

[03:33:00.0000] <Lachy>
ok, then I don't remember what the problem with header was. Can you clarify?

[03:43:00.0000] <jgraham>
Lachy: http://lists.w3.org/Archives/Public/public-html/2009Mar/0663.html The problem was that authors assumed <header> was for page-wide header sections

[03:43:01.0000] <jgraham>
 or as a generic replacement for <hx>

[03:47:00.0000] <jgraham>
Hixie: FWIW I'm not convinced that making <header> only take <hx> elements is sufficiently liberal

[03:47:01.0000] <jgraham>
s/header/hgroup/

[03:47:02.0000] <jgraham>
:)

[03:50:00.0000] <annevk5>
zcorpan, I know, krijnh was nice enough to sms me

[03:51:00.0000] <annevk5>
zcorpan, haven't looked into fixing it yet

[03:51:01.0000] <annevk5>
zcorpan, we had this party and all

[03:51:02.0000] <krijnh>
Smoesjes :)

[09:23:00.0000] <gsnedders>
Wow. Next time I go into town it'll be for the ball. Scary.

[09:24:00.0000] <hsivonen>
gsnedders: which dances have you practiced?

[09:25:00.0000] <gsnedders>
hsivonen: Well, I want 

[09:25:01.0000] <gsnedders>
*I went to a ceilidh two weeks ago

[09:26:00.0000] <gsnedders>
/me tries to write in his diary to catch up on the last month before getting read for a night which he'll likely have a fair bit to write about

[09:26:01.0000] <hsivonen>
Celtic ballroom dances?

[09:27:00.0000] <gsnedders>
Gorgeous photos of me in a suit will undoubtedly appear

[09:27:01.0000] <gsnedders>
Well, they aren't really ballroom dances…

[09:27:02.0000] <gsnedders>
But you left your sense of logic at the door, right?

[10:06:00.0000] <jgraham>
/me notes that in modern britian "ball" rarely means something in which actual ballroom dancing takes place

[10:07:00.0000] <Philip`>
Normally it means something spherical

[10:07:01.0000] <Philip`>
and only very tiny people could dance inside them

[11:01:00.0000] <annevk5>
http://bjoern.hoehrmann.de/utf-8/decoder/dfa/ is pretty neat

[16:39:00.0000] <Hixie>
hsivonen: no need to change anything for getElementsByTagName(), right? (case-sensitivity issue with svg in text/html)

[16:46:00.0000] <annevk5>
it should prolly be moved to Web DOM Core in due course

[16:59:00.0000] <gsnedders>
/me collapses


2009-05-02
[19:15:00.0000] <Hixie>
heycam: Is getting Date into WebIDL on your radar?

[19:20:00.0000] <heycam>
Hixie, it's on my radar in as much as i still have that mail flagged to reply to :)

[19:20:01.0000] <heycam>
do you want it so that you can have an ES Date object there?

[19:26:00.0000] <Hixie>
HTML5 uses Date in one of the interfaces, and someone noticed that that doesn't have a definition anyway

[19:27:00.0000] <heycam>
is your intention for it to map to an ES Date object?

[19:27:01.0000] <Hixie>
yes

[19:27:02.0000] <Hixie>
i was using DOMTimeStamp which was defined that way but it changed :-)

[19:27:03.0000] <Hixie>
http://lists.w3.org/Archives/Public/public-webapps/2009JanMar/0458.html

[19:28:00.0000] <heycam>
hmm, there's no entry in http://www.whatwg.org/specs/web-apps/current-work/multipage/infrastructure.html#reflect for Dates

[19:29:00.0000] <heycam>
would you expect myTimeElement.date.setDate(...) to do something?

[19:29:01.0000] <Hixie>
there aren't any DOM attributes of type Date that reflect anything as far as i know

[19:29:02.0000] <heycam>
ones on TimeElement do

[19:29:03.0000] <heycam>
s/T/HTMLT/

[19:30:00.0000] <heycam>
so actually it's just dateTime on HTMLTimeElement that is defined to reflect

[19:30:01.0000] <Hixie>
not as far as i can tell

[19:30:02.0000] <Hixie>
yeah

[19:30:03.0000] <heycam>
ok

[19:30:04.0000] <heycam>
so myTimeElement.date returns a new Date object each time it's got?

[19:32:00.0000] <Hixie>
yeah i guess that would be best

[19:32:01.0000] <Hixie>
that way there's no worry about the object being some weird live thing

[19:32:02.0000] <heycam>
yeah

[19:33:00.0000] <heycam>
(but maybe it'd be nice if a Date object could be read only somehow)

[19:33:01.0000] <heycam>
an ES Date, that is

[19:33:02.0000] <Hixie>
i think i'd rather not introduce new complexity for browsers to screw up

[19:34:00.0000] <heycam>
anyway, it's on my radar (though i don't seem to have a note for it; i'll add one now)

[19:35:00.0000] <heycam>
/me goes to get some breakf.. uh, brunch

[19:42:00.0000] <Hixie>
cool, thanks

[19:42:01.0000] <Hixie>
later

[21:17:00.0000] <mpilgrim>
http://twitter.com/fiddlerelf/status/1674755344

[21:53:00.0000] <MikeSmith>
about JSONDB, etc., proposals for document-based unstructured client-side storage: in what ways are these alternatives to SQL-based structured storage?

[21:54:00.0000] <MikeSmith>
it seems more like they are alternatives to simple name/value storage -- localStorage

[22:00:00.0000] <MikeSmith>
/me re-reads http://blog.vlad1.com/2009/04/06/html5-web-storage-and-sql/

[22:24:00.0000] <MikeSmith>
/me notices that Hixie forgot to set the authors and validators flags in his checkin description for r3054

[22:56:00.0000] <hsivonen>
Hixie: Re: getElementsByTagName

[22:56:01.0000] <hsivonen>
Hixie: the sentence "ames of HTML elements match regardless of case, but elements from other namespaces are treated as in XML (case-sensitively)." is bad

[22:56:02.0000] <Hixie>
yeah i don't know what non-normative text to put

[22:57:00.0000] <hsivonen>
Hixie: the next sentence, which is not equivalent, is OK

[22:57:01.0000] <Hixie>
the next sentence is an implementation conformance criteria

[22:57:02.0000] <Hixie>
so it's not always visible

[22:57:03.0000] <Hixie>
which leaves... nothing

[22:57:04.0000] <Hixie>
for the authors

[22:57:05.0000] <Hixie>
hence the annoying sentence

[22:58:00.0000] <hsivonen>
HTML elements match by lower-casing the argument before comparison

[22:58:01.0000] <Hixie>
MikeSmith: technically, we readded type=tel, it was in an early WF2 draft :-)

[22:58:02.0000] <MikeSmith>
Hixie: ah

[22:58:03.0000] <Hixie>
hsivonen: ok

[00:12:00.0000] <MikeSmith>
hsivonen: how does offline cache relate to parsing

[00:13:00.0000] <MikeSmith>
(in regard to http://twitter.com/hsivonen/status/1669704043)

[00:14:00.0000] <Hixie>
manifest attribute is processed in the parser

[00:15:00.0000] <MikeSmith>
ah

[00:30:00.0000] <archtech>
has anything from HTML% been adopted in IE8/

[00:30:01.0000] <archtech>
Ops, pardon the typos :P HTML5

[00:32:00.0000] <Hixie>
offhand, onhashchange, localStorage, sessionStorage, and postMessage()

[00:32:01.0000] <archtech>
Oh, great, I thought onHashChange was proprietary.

[00:32:02.0000] <archtech>
Thanks, Hixie.

[00:45:00.0000] <MikeSmith>
archtech: I think IE8 also supports navigator.online

[00:45:01.0000] <archtech>
MikeSmith, great :)

[01:20:00.0000] <Hixie>
MikeSmith: yeah but that one was added to html5 because they implemented it and we needed something similar :-)

[01:22:00.0000] <archtech>
Hixie, that's a great approach to standards as well :) it has to work both ways.

[01:23:00.0000] <MikeSmith>
Hixie: huh, didn't know that came from IE originally

[01:23:01.0000] <MikeSmith>
drag-and-drop from IE originally also?

[01:39:00.0000] <MikeSmith>
'Frankly, I'm finding this whole "rock star on the edge" thing is getting old.'

[01:40:00.0000] <MikeSmith>
I think it's actually the real rock starts on the edge who are getting old.

[01:40:01.0000] <MikeSmith>
or who have gotten old

[01:40:02.0000] <MikeSmith>
but I don't know who the rock star is in this particular case

[01:41:00.0000] <virtuelv>
/me goes to read mrlastweek

[01:41:01.0000] <MikeSmith>
virtuelv: that ain't MLW

[01:41:02.0000] <MikeSmith>
MLW differs in terms of humor

[01:42:00.0000] <MikeSmith>
a sense of humor, that is

[01:42:01.0000] <MikeSmith>
that is, in having one 

[01:42:02.0000] <MikeSmith>
sort of

[01:43:00.0000] <MikeSmith>
anyway, I guess the rock star on the edge in this case is mpilgrim 

[01:44:00.0000] <MikeSmith>
mpilgrim: your petty bickering risks undermining the credibility of the entire effort

[01:44:01.0000] <MikeSmith>
so please stop that

[01:44:02.0000] <MikeSmith>
the petty bickering, at least

[01:44:03.0000] <MikeSmith>
please do keep up the rock star part though

[01:49:00.0000] <MikeSmith>
"An end result of such shenanigans is that web page validity no longer means what it used to mean."

[01:49:01.0000] <MikeSmith>
yes. good.

[01:50:00.0000] <MikeSmith>
though we prefer the term "jackassery"

[01:51:00.0000] <MikeSmith>
" In the future, to get a "valid" stamp means that we have to adhere to a small, controlling group of people's interpretation of what the web should be,..."

[01:53:00.0000] <MikeSmith>
here's to hoping that more people do adopt that attitude that "this page is valid <whatever>" stamps are totally worthless

[01:54:00.0000] <MikeSmith>
regardless of what their reasons for adopting that attitude are

[01:54:01.0000] <MikeSmith>
I do like the "This page is valid in crappy biased markup" proposal, though.

[01:54:02.0000] <MikeSmith>
that one is a keeper

[02:06:00.0000] <Lachy>
Hixie, the new <hgroup> element seems completely useless. Why should anyone bother using it around heading elements? What's the difference between:

[02:06:01.0000] <Lachy>
<header><hgroup><h1>Foo</h1><h2>Bar</h2></hgroup> <p>more header content...</header> and

[02:06:02.0000] <Lachy>
<header><h1>Foo</h1><h2>Bar</h2> <p>more header content...</header>

[02:07:00.0000] <annevk5>
counts as a single heading in an outline prolly

[02:08:00.0000] <Lachy>
<header> should do that by itself

[02:09:00.0000] <annevk5>
<header> doesn't influence outline at all

[02:09:01.0000] <Lachy>
oh, he made <header> not be heading content :-(

[02:10:00.0000] <Lachy>
That really sucks

[02:13:00.0000] <billyjackass>
Lachy: your anti-hgroup vehemence is refreshing

[02:14:00.0000] <MikeSmith>
but it's too late

[02:14:01.0000] <MikeSmith>
I've already made a patch that adds <hgroup> to validator.nu

[02:15:00.0000] <MikeSmith>
so it's in the validator, and therefore now the Word of God

[02:16:00.0000] <MikeSmith>
and hsivonen has finally come to his senses about the "no badges" nonsense

[02:16:01.0000] <MikeSmith>
and agreed to start issuing badges

[02:16:02.0000] <annevk5>
/me wonders if http://lists.w3.org/Archives/Public/public-device-apis/2009May/0003.html was ever public knowledge

[02:16:03.0000] <MikeSmith>
the v.nu-issued badge will say, "This page is valid in crappy biased markup"

[02:17:00.0000] <MikeSmith>
but it will be a badge nonetheless

[02:17:01.0000] <annevk5>
"This is good to hear. Geo APIs was booted out of WebApps because of IP  threats from Nokia; has that now changed?"

[02:18:00.0000] <MikeSmith>
annevk5: that's not the reason it was "booted out of WebApps", so it doesn't really matter

[02:18:01.0000] <MikeSmith>
it was never "in" WebApps to begin with, for one thing

[02:20:00.0000] <MikeSmith>
and given that the Geolocation work has been very successful by any rational measure that anybody'd care to use (e.g., multiple implementations shipped already, actual people benefitting from it already)..

[02:21:00.0000] <MikeSmith>
well, I'd hope people would find something better to worry about

[02:21:01.0000] <MikeSmith>
like our petty bickering here

[02:22:00.0000] <MikeSmith>
which is undermining the credibility of the entire effort

[02:28:00.0000] <annevk5>
we have credibility? :p

[02:30:00.0000] <Lachy>
MikeSmith, it's never too late to reverse a mistake introduced into the spec, and hgroup is one of those things that won't last long

[02:30:01.0000] <MikeSmith>
oh, it will last. It has a chapter and verse number already.

[02:31:00.0000] <MikeSmith>
Lachy: are you suggesting we remove verses from the Bible?

[02:31:01.0000] <MikeSmith>
I don't God would like that.

[02:31:02.0000] <MikeSmith>
I mean, he wrote that book.

[02:31:03.0000] <MikeSmith>
indirectly, at least

[02:31:04.0000] <Hixie>
hsivonen is ok with badges now?

[02:31:05.0000] <Hixie>
have you been giving him some of your drugs

[02:32:00.0000] <MikeSmith>
I sense the hand of God in the HTML5 spec. Either that or the hand of Satan. can't tell which quite yet.

[02:32:01.0000] <Lachy>
MikeSmith, sure. Stuff has been added and removed from the bible many times over the past 2000 years (although not so much recently)

[02:33:00.0000] <MikeSmith>
Hixie: I was speaking on hsivonen's behalf. Have not checked with him yet, but I am reasonably confident that he will see the wisdom in the "This page is valid in crappy biased markup" badge.

[02:33:01.0000] <Hixie>
Lachy: <header> has to be allowed to contain <nav>, that's the way people want to use it. Given that, it can't be part of the outlining algorithm, Given that, unless we want every spec on W3C space to have one subsection called "Working Draft - 2 April 2009" or whatever, we need an hx grouping element for subheadings.

[02:34:00.0000] <Hixie>
MikeSmith: uh huh :-P

[02:35:00.0000] <Philip`>
I don't like badges

[02:35:01.0000] <Philip`>
but stickers are great

[02:35:02.0000] <Philip`>
We should have more stickers

[02:36:00.0000] <Lachy>
Hixie, The previous content model for it already did allow <nav> as a descendant.

[02:37:00.0000] <MikeSmith>
Hixie: btw, we are going to use the following at the v.nu badge -

[02:37:01.0000] <MikeSmith>
http://blog.zugara.com/wp-content/uploads/2009/04/leprechaun.jpg

[02:37:02.0000] <Hixie>
MikeSmith: (maybe you could convince him to use "I Check My Markup With validator.nu!" without saying it's valid)

[02:38:00.0000] <Hixie>
Lachy: no it didn't

[02:38:01.0000] <Hixie>
Lachy: it disallowed sectioning content descendants

[02:38:02.0000] <Hixie>
Lachy: nothing can be heading and contain sectioning at the sam e time

[02:38:03.0000] <MikeSmith>
Hixie: yeah (seriously) that would be good. would be good also if W3C validator stopped issuing the badges it does now, and issues an  "I Check My Markup..." thing instead

[02:38:04.0000] <Philip`>
That wouldn't fit on the button

[02:39:00.0000] <MikeSmith>
Philip`: I'm working on that

[02:39:01.0000] <Philip`>
Maybe use an animated GIF with marquee text

[02:39:02.0000] <MikeSmith>
heh

[02:39:03.0000] <Hixie>
parents around here have "My Student Is An Honour Student At Bla Bla High School" as bumper stickers

[02:39:04.0000] <Hixie>
maybe we can have "My Web Page Was Checked With validator.nu!" as bumper stickers

[02:39:05.0000] <Hixie>
give them out at conferences

[02:40:00.0000] <Lachy>
oh, I misread it.

[02:41:00.0000] <Lachy>
also, the 3rd example for <header> now looks a little crazy. Why are the "Important News" and "Games" headings included within the header?

[02:41:01.0000] <MikeSmith>
Lachy: Get thee behind me. Nothing in the Bible has ever been added or removed. Some things in particular *serializations* of the Bible have been added or removed. But the 

[02:42:00.0000] <MikeSmith>
But the abstract model of the Bible has never been altered, and is unalterable.

[02:42:01.0000] <Lachy>
haha

[02:43:00.0000] <Lachy>
MikeSmith, stories were written and added to the bible by hundreds of men over many centuries

[02:43:01.0000] <Hixie>
Lachy: yeah the examples could be improved... send mail or file a bug if you have any better ideas

[02:44:00.0000] <Philip`>
Lachy: But those hundreds of men were just expressing the fundamental truth laid down by God, so it is as if the book came from a single hand

[02:45:00.0000] <MikeSmith>
Philip`: Amen.

[02:45:01.0000] <MikeSmith>
Hallelujah

[02:45:02.0000] <Lachy>
yeah, and all those men who made copies of it by hand and introduced their own "corretions" in the process were just acting on behalf of god too...

[02:46:00.0000] <MikeSmith>
the Holy Spirit chooses Philip` to speak through now and then 

[02:46:01.0000] <Lachy>
:-D

[02:49:00.0000] <Philip`>
Lachy: They were just fixing typos

[02:50:00.0000] <Philip`>
or, uh, scribos

[02:50:01.0000] <Philip`>
because maybe the people writing down the original were a bit drunk at one point and started making up a few extra sections, and so the later people had to fix those mistakes

[02:51:00.0000] <Philip`>
Or maybe they made changes as a way of detecting copyright infringement, like how map makers include fake roads so they can see who's copied their maps

[02:51:01.0000] <Lachy>
yeah, and I'm sure it had nothing to do with needing the bible to suit the needs of the religious leaders at the time, wanting to control the public with it.

[02:52:00.0000] <Lachy>
(copyright didn't exist until the 18th century)

[02:56:00.0000] <Philip`>
"One of the earliest copyright disputes reputedly took place in 557 A.D. between Abbot Finnian of Moville and St. Columba over St. Columba's copying of a Psalter belonging to an Abbot. The dispute over ownership of the copy led to the Battle of Cúl Dreimhne (also known as the Battle of Cooldrumman), in which 3,000 men were killed."

[02:58:00.0000] <Lachy>
sure, but there were no copyright laws at the time.

[03:00:00.0000] <Philip`>
The concept of "if you copy my stuff, I'll send my army to attack your army" seems much more powerful than any copyright law

[03:04:00.0000] <Lachy>
oddly enough, that's similar to the technique being employed today by the RIAA/MPAA, only using lawyers instead of soldiers and slightly less killing

[03:05:00.0000] <Philip`>
So, apart from being completely different, it's similar?

[03:05:01.0000] <annevk5>
why is it impossible to find the anolis spec generator?

[03:05:02.0000] <annevk5>
/me wonders what keyword he's missing

[03:05:03.0000] <Lachy>
http://anolis.gsnedders.com/

[03:05:04.0000] <annevk5>
that's just the source code

[03:05:05.0000] <Philip`>
gsnedders needs more Google juice

[03:06:00.0000] <Lachy>
http://pimpmyspec.net/

[03:06:01.0000] <Philip`>
http://wiki.whatwg.org/wiki/Developing_HTML5_derivatives points at http://pimpmyspec.net/ if that's what you want

[03:09:00.0000] <annevk5>
ta

[03:28:00.0000] <annevk42>
Hixie, why the inconsistency in the event handler attribute tables? first heading lowercase, second heading starts with a capital

[04:22:00.0000] <virtuelv>
WHile being completely off topic: Could someone bash the twitter-folks over the head with the largest cluebat you can find?

[04:24:00.0000] <virtuelv>
Whatever was wrong with loading actual pages using actual URLs, instead of pretending everything is pointing to a  document fragment 

[04:30:00.0000] <annevk42>
looks a lot better than before actually, imo

[04:31:00.0000] <annevk42>
and it's not like you need to share the URLs, they're all bound to your login details anyway

[04:41:00.0000] <virtuelv>
yes, but it actually feels much slower than before

[04:43:00.0000] <virtuelv>
and I see no particular good reason for abolishing loading pages

[04:45:00.0000] <annevk42>
making things look better :)

[04:48:00.0000] <gsnedders>
And when it fails you have to reload everything and start from the newest all over again

[05:48:00.0000] <gsnedders>
/me wonders about automatically getting titles for photos based upon geo data

[05:49:00.0000] <gsnedders>
(i.e., writing code to allow me to be lazy in the future)

[06:56:00.0000] <Philip`>
How exciting

[06:56:01.0000] <Philip`>
Mr & Mrs Last Week posted a link to http://209.85.229.132/search?q=cache:(somewhere you probably don't want to bother looking)&cd=1&hl=en&ct=clnk&gl=uk&client=firefox-a

[06:56:02.0000] <Philip`>
which suggests they're either a Firefox user from the UK, or else copied a Google cache link from someone who was

[06:57:00.0000] <gsnedders>
Damnit! And I always thought they were IE users!

[06:57:01.0000] <Philip`>
(Well, probably the 'view PDF as HTML' link rather than the cache)

[07:14:00.0000] <Philip`>
The 'pv' tool is great

[07:15:00.0000] <Philip`>
If you're doing command-line processing of (as a hypothetical example) a 2GB gzipped file with some decompression and greps and perls and suchlike, you can just stick pv into the pipeline and it'll tell you how fast the processing is going and give you an ETA

[07:17:00.0000] <Philip`>
/me gets a fairly consistent 11 MB/sec, and is unsure whether that's good or not

[07:19:00.0000] <Dashiva>
Disk IO speed would be one possible baseline for comparison

[07:20:00.0000] <Philip`>
cat >/dev/null goes at 722 MB/sec

[07:20:01.0000] <Dashiva>
That seems a bit... high

[07:20:02.0000] <Philip`>
md5sum goes at 228 MB/sec

[07:21:00.0000] <Dashiva>
I suppose if you have the 3Gb SATA

[07:21:01.0000] <Philip`>
It would be a bit high if it was actually reading from physical disk, not from the file cache

[07:21:02.0000] <Dashiva>
No, even then...

[07:21:03.0000] <Philip`>
since it's really just reading from RAM

[07:21:04.0000] <Dashiva>
Yeah, true

[07:22:00.0000] <Philip`>
gzip -cd|md5sum goes at 11 MB/sec

[07:22:01.0000] <Philip`>
so I blame gzip

[07:23:00.0000] <Philip`>
/me assumes the disks are SATA, but doesn't really know

[07:24:00.0000] <Dashiva>
Tried different compression levels?

[07:24:01.0000] <Philip`>
and I don't know how to find out how fast they are

[07:28:00.0000] <Philip`>
The output from gzip -cd goes at ~185 MB/s if I compress with -1

[07:28:01.0000] <Philip`>
The output from gzip -cd goes at ~205 MB/s if I compress with -9

[07:29:00.0000] <Philip`>
and there's a roughly linear increase between those points

[07:29:01.0000] <Philip`>
(The input to gzip is faster with -1, but there's a lot more of it)

[07:34:00.0000] <Philip`>
But I'm way too lazy to wait for it to recompress 35GB of data with gzip -9

[07:34:01.0000] <Dashiva>
Hmm

[07:34:02.0000] <Dashiva>
But you blamed gzip for 11 MB/sec earlier

[07:34:03.0000] <Philip`>
(even though I'm insufficiently lazy that I'll spend two minutes counting that it's 35GB)

[07:34:04.0000] <Dashiva>
Now you're saying it's 180-200

[07:34:05.0000] <Philip`>
11 MB/sec input

[07:35:00.0000] <Philip`>
200 MB/sec output

[07:35:01.0000] <Dashiva>
ah

[07:35:02.0000] <Philip`>
/me should probably be clearer about such things

[07:37:00.0000] <Dashiva>
Well, it makes sense 

[07:40:00.0000] <Philip`>
Oops, the files I've been looking at aren't all text/html

[07:40:01.0000] <Philip`>
I'm sure they were when I last looked, but they certainly haven't changed since then...

[07:43:00.0000] <Philip`>
Hmm, I've got 1500 more content-type headers than pages

[07:45:00.0000] <Philip`>
http://www.agriaffaires.de/gebrauchte/andere-landwirtschaftliche-anhaenger/698421/krampe-big-body-600-premium.html

[07:45:01.0000] <Philip`>
*13* content-type headers?!

[07:49:00.0000] <gsnedders>
Why not? I mean, it's hardly if browsers about any but the last.

[08:56:00.0000] <Dashiva>
In the future, to get a "valid" stamp means that we have to adhere to a small, controlling group of people's interpretation of what the web should be, and I'm just not willing to go there.

[08:56:01.0000] <Dashiva>
Isn't this how it has been all along?

[08:58:00.0000] <jgraham>
Philip`: You written any useful scripts to work with the dotnetdotorg data?

[08:58:01.0000] <Philip`>
Dashiva: No

[08:59:00.0000] <Philip`>
Dashiva: It's been a *different* small, controlling group of people

[09:00:00.0000] <Dashiva>
Right, my bad

[09:02:00.0000] <Dashiva>
I also like the part where it goes "I guess the browser vendors want to cooperate [...] Look forward to lots of browser incompatability"

[09:04:00.0000] <Philip`>
Dashiva: Where does it say that?

[09:04:01.0000] <Philip`>
Also, what "it" is it?

[09:04:02.0000] <Dashiva>
http://realtech.burningbird.net/semantic-web/semantic-web-issues-and-practices/going-non-standard

[09:05:00.0000] <Philip`>
Dashiva: I can't see any mentions of e.g. looking forward to incompatibility in there

[09:05:01.0000] <Philip`>
Oh, wait, maybe it sort of says that in the last-but-one paragraph

[09:05:02.0000] <Philip`>
though not with your words or your spelling error

[09:06:00.0000] <Dashiva>
Yeah, I always typo that

[09:07:00.0000] <Dashiva>
I've started getting ominous right, though!

[09:07:01.0000] <Philip`>
Oh my nous!

[09:08:00.0000] <Dashiva>
I still think omnious sounds more ominous. But it's a lost battle.

[09:08:01.0000] <Philip`>
I find it hard to understand the comment "You know who really makes me mad? The browser companies. They want obfuscation. They want imprecision. They want gaps. They want things in a state of confusion, so they can do what they want, and not have to worry about "standards"."

[09:09:00.0000] <Philip`>
since as far as I'm aware the browser companies formed the WHATWG specifically in response to imprecision and confusion and gaps in the W3C's specification of web technologies

[09:09:01.0000] <Dashiva>
I reckon it's just the usual misunderstanding of how things work

[09:09:02.0000] <Philip`>
and HTML5 defines everything precisely and tries to fill in all the gaps

[09:09:03.0000] <jgraham>
I find it really hard to understand Shelly in general. I really don't understand why she's so angry all the time

[09:09:04.0000] <Philip`>
(though it may still be obfuscated and confusing)

[09:09:05.0000] <Dashiva>
jgraham: Maybe the owls are out to get her

[09:10:00.0000] <jgraham>
Dashiva: Owls deliver post

[09:10:01.0000] <Dashiva>
They also come in the night and tear your legs apart

[09:10:02.0000] <Philip`>
Can owls deliver get too?

[09:11:00.0000] <jgraham>
Philip`: I don't recall Owls doing anything that resembles a get

[09:11:01.0000] <Dashiva>
They get field mice

[09:11:02.0000] <jgraham>
So HTTP over Owl would be non-trivial

[09:11:03.0000] <Dashiva>
Or I suppose that's more of a delete

[09:11:04.0000] <Philip`>
Or a head, depending on how hard they bite

[09:11:05.0000] <Dashiva>
Head is supposed to be safe, isn't it?

[09:12:00.0000] <jgraham>
Idempotent

[09:12:01.0000] <Dashiva>
Okay, then it's okay I guess

[09:12:02.0000] <jgraham>
Maybe if we could train them to only attach starfish and other creatures that can regrow parts of their body

[09:13:00.0000] <jgraham>
*attack

[09:13:01.0000] <Philip`>
jgraham: I have a Perl script that splits the file into more-easily-parseable parallelisable chunks, and a couple of organically grown (i.e. horridly written) Java programs to do grep and to run various analyses on the parsed content

[09:13:02.0000] <Dashiva>
But isn't idempotent defined as repeated applications being the same as one application?

[09:13:03.0000] <Dashiva>
Without saying anything about initial state vs one application

[09:15:00.0000] <Philip`>
jgraham: How about axolotls?

[09:16:00.0000] <Philip`>
(which apparently can regrow limbs)

[09:16:01.0000] <Philip`>
(which I learned via the ever-informative blog at http://fuckyoupenguin.blogspot.com/2009/03/no.html )

[09:17:00.0000] <Philip`>
Dashiva: In that case, DELETE would be considered idempotent

[09:17:01.0000] <Philip`>
since deleting something twice has the same consequence as deleting it once

[09:18:00.0000] <Philip`>
so the term 'safe' seems better than 'idempotent'

[09:18:01.0000] <Dashiva>
Yeah, that's why I said safe :)

[09:20:00.0000] <jgraham>
Ah yes, head is safe + idempotent. Delete is jsut idempotent

[09:20:01.0000] <Philip`>
Of course it's not actually idempotent in an asynchronous environment

[09:21:00.0000] <Philip`>
/me wonders if HTTP has a more subtle definition of 'idempotent'

[09:27:00.0000] <Philip`>
jgraham: (All my other code for working with the data is one-off bash commands, usually some combination of sort and uniq and perl, and I don't have them saved anywhere)

[10:27:00.0000] <gsnedders>
/me isn't entirely happy with his photos from the ball

[10:32:00.0000] <Philip`>
gsnedders: Photoshop

[10:32:01.0000] <gsnedders>
… is where I've spent far too much time trying to make them less shit :)

[10:33:00.0000] <Philip`>
(I suppose I could say "gsnedders: Gimp" to promote OS software, but that might be easy to misinterpret)

[11:24:00.0000] <gsnedders>
http://www.flickr.com/photos/gsnedders/sets/72157617511428181/ — My best attempt.

[11:41:00.0000] <fearphage>
any of you ever used XHR to send an element? what is the use case for that?

[11:41:01.0000] <fearphage>
just investigating

[11:47:00.0000] <Philip`>
http://www.span-raps.com/English2/index.php?option=com_content&task=view&id=51&Itemid=3 - <time w:st="on" hour="13" minute="00"><u><span lang="EN-CA" style="mso-ansi-language: EN-CA">1:00 PM</span></u></time><u><span lang="EN-CA" style="mso-ansi-language: EN-CA">, Saturday, May 15<sup>th</sup>, 2004<p></p></span></u>

[11:47:01.0000] <Philip`>
Hmm, that's the only <time> element I see

[12:15:00.0000] <hsivonen>
interesting accusation on Shelley's blog: "You know who really makes me mad? The browser companies. They want obfuscation. They want imprecision. They want gaps. They want things in a state of confusion, so they can do what they want, and not have to worry about "standards"."

[12:15:01.0000] <Hixie>
Philip` commented on that earlier

[12:43:00.0000] <virtuelv>
hsivonen: oh, assume someone has a position, and attack that

[12:43:01.0000] <virtuelv>
let me look that up in the big book of fallacies

[12:45:00.0000] <virtuelv>
while I'm not involved in HTML5 as such, but other efforts, I can say that what we _don't_ want is imprecision

[12:45:01.0000] <Hixie>
yeah precision is somewhat the point :-)

[12:52:00.0000] <gsnedders>
Let's go back to HTML 4.01!

[12:52:01.0000] <gsnedders>
That's imprecise!

[12:53:00.0000] <gsnedders>
Wait, if we want imprecision, why are we revising it?

[12:55:00.0000] <fearphage>
can anyone help me with a use case/purpose for sending nodes/elements with XHR? or a place i could possibly find an answer for that

[13:14:00.0000] <Hixie>
i wonder what to do about accesskey=""

[13:23:00.0000] <Hixie>
webkit uses ctrl+option

[13:23:01.0000] <Hixie>
firefox uses ctrl

[13:23:02.0000] <Hixie>
(on mac)

[13:23:03.0000] <gsnedders>
Delete it!

[13:23:04.0000] <Hixie>
i never remember what opera's behaviour is

[13:23:05.0000] <Hixie>
something to do with the escape key?

[13:25:00.0000] <Philip`>
Shift+esc

[13:28:00.0000] <Hixie>
interesting

[13:28:01.0000] <Hixie>
just the menu, no direct shortcuts right?

[13:29:00.0000] <krijnh>
Jep

[13:29:01.0000] <Hixie>
interesting, Webkit, Firefox, and Opera all do nothing if the accesskey="" attribute is more than one character long

[13:30:00.0000] <Hixie>
only webkit supports accesskey=" "

[13:30:01.0000] <Hixie>
none of them strip leading spaces

[13:30:02.0000] <Hixie>
hmm

[13:31:00.0000] <Hixie>
this leaves us with some interesting options

[13:32:00.0000] <Hixie>
accesskey="" could be an enumerated attribute with values a..z, 0..9, and some keywords like "search", which could be bound to UA-specific standard bindings

[13:34:00.0000] <krijnh>
What's the benefit over using input type="search" for example?

[13:34:01.0000] <Hixie>
none really

[13:34:02.0000] <Hixie>
but we could have some generic keywords, like "f1" .. "f9", which map to a set of keys unique to each browser which are always available

[13:34:03.0000] <Hixie>
where "f1" is your most important site function

[13:35:00.0000] <Hixie>
though discoverability would be an issue

[13:35:01.0000] <krijnh>
Sounds confusing

[13:36:00.0000] <jgraham>
Hixie: I guess more than just a-z work on keyboards with more characters

[13:37:00.0000] <jgraham>
So that sounds bad

[13:37:01.0000] <jgraham>
gsnedders: 50mm f/1.8 It won't make you a better photographer but may be good for taking better portraits

[13:38:00.0000] <gsnedders>
jgraham: I've been more tempted by the 85mm

[13:38:01.0000] <jgraham>
gsnedders: Well assuming you are rich and always have rather a lot o space the 85mm is supposed to be very nice indeed

[13:39:00.0000] <gsnedders>
I mean the f/1.8 85mm, not the f/1.2

[13:39:01.0000] <jgraham>
But the 50mm is like 80 GBP or less secondhand

[13:39:02.0000] <gsnedders>
Though unsurprisingly I would quite like the f/1.2 :P

[13:40:00.0000] <Hixie>
almost every link on http://www.456bereastreet.com/archive/200601/accesskey_problems_remain_in_xhtml_2/ is now 404

[13:40:01.0000] <jgraham>
gsnedders: Still 300+GBP

[13:40:02.0000] <gsnedders>
Hmm, I remember seeing it for less before. Oh well.

[13:40:03.0000] <Philip`>
http://philip.html5.org/data/accesskey-values.txt

[13:40:04.0000] <Philip`>
(Not very interesting)

[13:40:05.0000] <jgraham>
The fundamental problem with accesskey is that it is a solution for a UI issue that has not been designed by UI designers

[13:40:06.0000] <Philip`>
(but at least it shows most people use single-character accesskeys)

[13:41:00.0000] <jgraham>
This is A Bad Idea

[13:41:01.0000] <krijnh>
Philip`: do the links include titles or something?

[13:41:02.0000] <Hixie>
Philip`: i like accesskey="10"

[13:41:03.0000] <Philip`>
krijnh: Which links?

[13:41:04.0000] <Hixie>
wish i had a "10" key

[13:41:05.0000] <gsnedders>
jgraham: I have been thinking about getting the 50mm just as a stop-gap solution until I have money, but I do so little portrait photography anyway, and the bokeh on the Sigma 17–70mm is nicer IMO

[13:41:06.0000] <krijnh>
Philip`: on which the accesskey attributes are set

[13:42:00.0000] <krijnh>
Perhaps a combination of accesskey and title can be helpful

[13:42:01.0000] <Philip`>
krijnh: Don't know, and I don't have an easy way of extracting that data

[13:42:02.0000] <krijnh>
Ke

[13:43:00.0000] <jgraham>
gsnedders: What is the maximum apature on the sigma? And are you prepared to live wih the fact tha there is a high chance that your sigma lens will have qualiy conrol issues

[13:43:01.0000] <jgraham>
?

[13:43:02.0000] <jgraham>
*aperture

[13:43:03.0000] <gsnedders>
jgraham: f/2.8-3.5 only, which does kinda suck.

[13:43:04.0000] <Hixie>
http://www.w3.org/mid/B2750F31-8ACF-4B46-8D02-44371AEE4C9F⊙ac seems sound

[13:43:05.0000] <gsnedders>
jgraham: Also, it hasn't given me any problems yet

[13:43:06.0000] <Philip`>
gsnedders: The photos would be better if the lighting didn't make everybody look like corpses

[13:44:00.0000] <gsnedders>
Yeah, a lot of them are over-exposed. I've already rectified some of them.

[13:44:01.0000] <jgraham>
gsnedders: Oh, you already have it? Well I guess there's no point in discouraging you from buying it then

[13:44:02.0000] <gsnedders>
jgraham: Yes, I have it. That's what I took all those photos on.

[13:45:00.0000] <gsnedders>
jgraham: I got it to replace the 18–55mm kit lens, and it is certainly a lot better than that

[13:45:01.0000] <jgraham>
gsnedders: A lot of those would also benefit from some fill flash

[13:46:00.0000] <jgraham>
/me would like the 24-105L to replace his kit lens bu it is a bit expensive

[13:46:01.0000] <gsnedders>
Indeed.

[13:46:02.0000] <jgraham>
gsnedders: If you don't already have en external flash, it is well, wel worth getting one

[13:47:00.0000] <gsnedders>
I have been seriously considering getting one…

[13:47:01.0000] <jgraham>
The Speedlite 430EX is perfectly fine

[13:47:02.0000] <jgraham>
As far as I can tell

[13:47:03.0000] <gsnedders>
jgraham: But I can't decide between an external flash and 70–300mm

[13:47:04.0000] <jgraham>
(well I would like lower power sometimes, although in principle it is always possible to arrange for that)

[13:48:00.0000] <jgraham>
gsnedders: What is your use case for the 70-300mm?

[13:48:01.0000] <gsnedders>
jgraham: Things in the distance? :P

[13:48:02.0000] <jgraham>
gsnedders: What things in particular?

[13:49:00.0000] <jgraham>
It makes rather a lot of difference

[13:49:01.0000] <gsnedders>
jgraham: More seriously, there's quite a few things I'd like to be able to photograph while out walking which I can't really get to (e.g., I'm on a mountain and they aren't)

[13:49:02.0000] <jgraham>
So landscapes?

[13:49:03.0000] <gsnedders>
jgraham: yes

[13:49:04.0000] <gsnedders>
(You think you can get a simple answer from me!? Peh!)

[13:50:00.0000] <jgraham>
Oh in that case the 70-300mm is probably a reasonable idea (although I haven't used the canon one for more than a few shots)

[13:50:01.0000] <jgraham>
But external flash opens up whole new worlds of possibility

[13:51:00.0000] <jgraham>
The problem with phoography is a) there are a lot of good things to buy b) They are all expensive and c) None of hem is an effecive substitute for actual talent

[13:52:00.0000] <gsnedders>
/me lacks the talent part :P

[13:52:01.0000] <jgraham>
/me too

[13:52:02.0000] <gsnedders>
/me thinks jgraham is better than him

[13:53:00.0000] <jgraham>
Oh, and the money part :)

[13:53:01.0000] <gsnedders>
Yeah, I lack that too.

[13:53:02.0000] <gsnedders>
But I guess we both have too much we could buy :)

[13:54:00.0000] <jgraham>
/me would also like the 100-400mm. Or maybe one of the long fixed focal length things.

[14:20:00.0000] <inimino>
/me looks forward to a new era of "cool, oversexed, overscripted Ajax applications"

[14:31:00.0000] <gsnedders>
/me would quite like a Lensbaby, and they're fairly cheap, so he may get one soon

[14:31:01.0000] <gsnedders>
/me thought they were pointless and not overly fun until he got to play with Chris Wilson's at TPAC


2009-05-03
[19:51:00.0000] <Hixie>
comments on what i just checked in would be quite useful

[19:51:01.0000] <Hixie>
it is a Possible Solution to the Great accesskey Problem

[19:51:02.0000] <Hixie>
(r3065)

[01:20:00.0000] <annevk5>
Hixie, domintro for document.commands says document.images

[01:21:00.0000] <annevk5>
nm, fixed in the next check in

[01:21:01.0000] <rmccue>
Shouldn't there be some sort of limit on who can change the status on http://www.whatwg.org/ ?

[01:22:00.0000] <annevk5>
it hasn't been abused so far

[01:22:01.0000] <Hixie>
rmccue: why?

[01:22:02.0000] <Hixie>
rmccue: if you feel you shouldn't be allowed to change it... don't change it. :-)

[01:22:03.0000] <rmccue>
Certain people might not realise what it does. :)

[01:23:00.0000] <Hixie>
well we can certainly improve the documentation

[01:24:00.0000] <rmccue>
I figured it would have asked for some sort of authentication before it actually updated, so I apologise for http://twitter.com/WHATWG/status/1685430406

[01:24:01.0000] <Hixie>
no worries

[01:25:00.0000] <Hixie>
i updated the checkbox -- is it clearer now? :-)

[01:26:00.0000] <rmccue>
Checkbox?

[01:28:00.0000] <Hixie>
oh hey the checkbox is just for bots.

[01:29:00.0000] <Hixie>
ok look now :-)

[01:30:00.0000] <rmccue>
Looks awesome, thanks :)

[01:30:01.0000] <Hixie>
np :-)

[01:52:00.0000] <Hixie>
what's a good other name for <command default="">

[02:27:00.0000] <Philip`>
annevk5: That's not true - I've abused it

[04:36:00.0000] <krijnh>
http://realtech.burningbird.net/semantic-web/semantic-web-issues-and-practices/going-non-standard#comment-1369 - Public IRC logs considered harmful

[04:37:00.0000] <krijnh>
Perhaps I should stop logging?

[04:41:00.0000] <krijnh>
"It's in black and white." - or I could just change the colors, of course

[08:41:00.0000] <boblet>
Hi all. Has anyone used hcalendar with html5? did you use <time> for dtstart/dtend?

[08:49:00.0000] <gsnedders>
Hmm… what should the behaviour of a server be if If-Modified-Since has an invalid value?

[08:53:00.0000] <gsnedders>
/me guesses he should read the spec cover to cover, but he never does that

[09:35:00.0000] <gsnedders>
http-date is woefully underspecified

[09:42:00.0000] <Philip`>
/me is shocked

[09:43:00.0000] <gsnedders>
Oh, yeah, I remember now.

[09:43:01.0000] <gsnedders>
2616 and 2616bis both subscribe to Postel's law, without saying what the liberal behaviour should be, so just leave everything undefined.

[09:47:00.0000] <gsnedders>
http://pastebin.com/m1cc1e5d7 is sane insofar as it forbids everything that doesn't match the grammar or is logically invalid

[09:47:01.0000] <gsnedders>
Not that it works, actually. Woops.

[09:51:00.0000] <gsnedders>
http://pastebin.com/mf435844 that should

[09:52:00.0000] <Philip`>
Is the Jul/July inconsistency intentional?

[09:53:00.0000] <Philip`>
And don't you want to check for e.g. $day < 0?

[09:54:00.0000] <Philip`>
"$hour == 24 && $minute != 0 && $second != 0" - should be "$hour == 24 && ($minute != 0 || $second != 0)"

[09:54:01.0000] <Philip`>
"$second > 59" - is it not possible to represent leap seconds?

[09:56:00.0000] <gsnedders>
No; $day < 0 is handled by regex (it can't get there); yes; HTTP-bis is unclear.

[09:56:01.0000] <gsnedders>
(From my reading it is pure GMT so no.)

[09:57:00.0000] <Philip`>
gsnedders: By the way, dotnetdotcom.org appears to give the raw HTTP responses (unlike the pre-parsed data that I had collected), which might be interesting in terms of parsing HTTP

[09:58:00.0000] <gsnedders>
Entry NULL Entry NULL Entry

[09:58:01.0000] <gsnedders>
Hmm…

[09:58:02.0000] <gsnedders>
What if there is null in entry?

[09:58:03.0000] <Philip`>
There isn't

[09:58:04.0000] <Philip`>
I assume they must strip all the nulls, or something

[09:58:05.0000] <gsnedders>
How boring.

[09:58:06.0000] <Philip`>
(but I've not tried to see exactly what they do)

[09:59:00.0000] <Philip`>
(but my parser worked correctly under the assumption that nulls were always delimeters)

[09:59:01.0000] <Philip`>
/me thinks it's a bit of a rubbish storage format, so he converted into sensible length-prefixed strings for processing

[09:59:02.0000] <Philip`>
s//it /

[10:00:00.0000] <Philip`>
(Also split into multiple chunks, to allow parallel processing)

[10:00:01.0000] <gsnedders>
Are there any ABNF parser generators that are easy to use?

[10:02:00.0000] <Philip`>
There's your brain

[10:02:01.0000] <Philip`>
Not sure what else, though

[10:06:00.0000] <gsnedders>
http://www.quut.com/abnfgen/ could be useful

[10:08:00.0000] <MikeSmith>
the datatype definition of the value for the accesskey attribute in the schema needs a name

[10:08:01.0000] <MikeSmith>
keylist? characterlist?

[10:08:02.0000] <MikeSmith>
spec says the value the value must be an ordered set of unique space-separated tokens, each of which must be exactly one Unicode code point in length.

[10:14:00.0000] <Philip`>
It might change in the future to allow keywords, not just characters

[10:14:01.0000] <MikeSmith>
hmm

[10:16:00.0000] <MikeSmith>
then we'd probably also want to change the name of the datatype too

[10:33:00.0000] <jgraham>
/me hopes that krijnh is not seriously considering stpping logging\

[10:34:00.0000] <Philip`>
If he did, I guess somebody else would take over, so it wouldn't be the end of the world for Mr Last Week

[10:36:00.0000] <jgraham>
Yeah, I guess I would start logging or something. Although that wouldn' help me catch up with hings I miss when web faction decide to reboot my server

[10:41:00.0000] <Philip`>
You could get a more reliable server :-)

[10:41:01.0000] <Philip`>
 16:46:03 up 214 days, 17:48,  3 users,  load average: 0.01, 0.01, 0.00

[10:41:02.0000] <Philip`>
says mine

[10:41:03.0000] <Philip`>
(on which I run my IRC client)

[10:42:00.0000] <inimino>
 16:46:23 up 449 days, 10:24, 13 users,  load average: 0.00, 0.05, 0.03

[10:43:00.0000] <inimino>
take that!

[10:43:01.0000] <Philip`>
But I only installed this server 214 days ago :-)

[10:45:00.0000] <inimino>
server companies should sell boxes with pre-cached uptime for bragging rights

[10:46:00.0000] <inimino>
"aged seven years, never turned off" (ok maybe not)

[10:46:01.0000] <Philip`>
"aged seven years, never installed any kernel updates"

[10:47:00.0000] <Philip`>
If I just wanted bragging rights, I could point at:

[10:47:01.0000] <Philip`>
 17:50:15 up 3587 days,  4:16,  7 users,  load average: 0.00, 0.00, 0.00

[10:47:02.0000] <Philip`>
and you would have no way of knowing that that's entirely fake

[10:48:00.0000] <inimino>
you'd have no way of knowing whether I know that 'uptime' truncates the output ;)

[10:49:00.0000] <Philip`>
Truncates to three digits?

[10:49:01.0000] <inimino>
yeah

[10:50:00.0000] <Philip`>
I'm pretty sure I exceeded a thousand days on a Solaris box and it displayed the time just fine :-)

[10:51:00.0000] <inimino>
ah, but that's not formatted like the Solaris uptime output :)

[10:53:00.0000] <Philip`>
  5:56pm  up 3587 days  1:04,  1 user,  load average: 0.77, 0.44, 0.28

[10:53:01.0000] <Philip`>
Happy now? :-p

[10:54:00.0000] <inimino>
"day(s)"

[10:55:00.0000] <inimino>
/me recommends fuschia on chartreuse for the logs

[11:03:00.0000] <Philip`>
inimino: No, it's definitely "days"

[11:04:00.0000] <Philip`>
since I ran the command on the real box, and only fraudulently modified the number

[11:07:00.0000] <gsnedders>
Philip`: But what is real?

[11:07:01.0000] <inimino>
http://ods.com.ua/win/eng/unix/usail/man/solaris/uptime.1.html

[11:08:00.0000] <inimino>
I have no way of knowing how many times that has changed in the last 3587 days, though

[11:13:00.0000] <Philip`>
inimino: My man page is marked as:

[11:13:01.0000] <Philip`>
SunOS 5.9           Last change: 18 Mar 1994                    1

[11:13:02.0000] <Philip`>
and says "day(s)", but the actual command says "days"

[11:13:03.0000] <inimino>
aha!

[11:14:00.0000] <Philip`>
Oh, hmm

[11:14:01.0000] <Philip`>
/usr/bin/uptime says "6:18pm  up xxx day(s),  1:27,  1 user,  load average: 0.03, 0.07, 0.20"

[11:15:00.0000] <Philip`>
but PATH puts /usr/local/bin/uptime first, which is a GNU version

[11:15:01.0000] <inimino>
that explains it

[11:16:00.0000] <Philip`>
/me blames Solaris for having a silly mixture of GNU and non-GNU tools

[11:16:01.0000] <inimino>
/me always found Solaris more tolerable with GNU everything installed

[11:16:02.0000] <Philip`>
/me never worked out how to install software on Solaris

[11:17:00.0000] <inimino>
/me dimly remembers something about pkg-add

[12:39:00.0000] <jgraham>
Philip`: Yeah, a more reliable server would also be nice. But it's hard o optimise over cost, uptime, ease of use, speed, frequency of updates, quality of support, based only on the marketing materials on hosting companies websites and my limited knowledge of what the right tradeoff is

[12:40:00.0000] <jgraham>
Although something not dissimmilar to that was one of the use cases for embedded metadata in HTML 5 so presumaby Hixie will solve that for me any day now

[12:50:00.0000] <Philip`>
jgraham: It would be much easier if hosting companies gave a single quality score on a uniform industry-wide scale for all of their services, and then you could just pick the highest one you can afford

[13:04:00.0000] <hsivonen>
Shelley has a point in the comment that MLW quotes in his latest.

[13:19:00.0000] <jgraham>
hsivonen: What are the examples of Hixie being disrespectful to Sam on IRC?

[13:23:00.0000] <hsivonen>
jgraham: I don't know about Sam.

[13:35:00.0000] <jgraham>
hsivonen: It would be nice if such accusations were backed up with evidence

[13:37:00.0000] <hsivonen>
jgraham: it's the perception that counts. if people get a perception from IRC logs that isn't in line of blog camments/emails, they suspect the blog comments/email

[13:38:00.0000] <Philip`>
/me wonders how often IRC comments are misinterpreted because they're not read in their proper context

[13:41:00.0000] <inimino>
perhaps krijnh could add a row of dancing hamsters to the logs, to impart the proper sense of gravitas

[13:42:00.0000] <krijnh>
Great idea :)

[13:47:00.0000] <hsivonen>
http://blog.killerbees.co.uk/2009/05/xml-co-processor.html

[13:49:00.0000] <jgraham>
hsivonen: I'm not sure what the solution is but it seems like if people were called out for not backing their accusations of bad faith with actual evidence they would be less inclined to make assumptions in line with their existing prejudices and more inclined o read what has actually been said

[13:51:00.0000] <hsivonen>
jgraham: that's possible

[13:52:00.0000] <jgraham>
For example Shelly got the quote from Mark wrong which rather changes the possible set of interpretations

[13:55:00.0000] <Hixie>
As far as I can tell, the reason people think I don't respect RDFa proponents is that they don't want to give use cases (just look at the difficulties we had getting use cases from them) and so when I say "yes but what are the use cases" they take it as some sort of offense

[13:55:01.0000] <Hixie>
whereas tantek is the other way round -- I suggest something, and _he's_ the one who asks for use cases and sample pages

[13:56:00.0000] <Hixie>
(shelley even said this explicitly, she had a blog post whose _title_ specifically said that she was advocating not cooperating with us by not "justifying" rdfa)

[13:56:01.0000] <Philip`>
She was specific about Hixie not bothering to reply to Manu about the RDFa thing, which is good because that made it possible to point out that actually he did (on a different mailing list to where the original post was seen)

[13:56:02.0000] <Philip`>
so actionable criticisms are good :-)

[13:56:03.0000] <Hixie>
manu has been by far the most helpful member of the rdf/rdfa community

[13:57:00.0000] <Hixie>
he and i even had a phone call about this stuff

[13:57:01.0000] <Hixie>
and he's been pretty much the only person to write many detailed use cases

[13:58:00.0000] <Hixie>
(the guy from yahoo! was very helpful too, but he's more of a neutral bystander than an rdfa proponent, from what i understand)

[13:59:00.0000] <Hixie>
anyway, i have one more person to talk to next week about this microdata stuff, and then i'll see what can be done to address these use cases

[13:59:01.0000] <Hixie>
(i wanted to do it last week but said person wasn't available)

[14:58:00.0000] <gsnedders>
"a phone call"!?

[15:02:00.0000] <gsnedders>
Hixie: I'll likely see Avenue Q in June, just because you rave on about it.

[15:11:00.0000] <gsnedders>
/me will also likely go to the Globe twice to see both plays they are producing at the time, As You Like It and Romeo and Juliet

[15:20:00.0000] <Hixie>
gsnedders: cool

[16:03:00.0000] <Dashiva>
"You can not argue this because it is factual opinion."

[16:04:00.0000] <Dashiva>
Axiomatic proof, move over

[16:04:01.0000] <hsivonen>
Dashiva: what are you quoting?

[16:05:00.0000] <Dashiva>
Nothing related, it just reminded me of The Proof.

[16:47:00.0000] <Hixie>
http://identi.ca/notice/3921839

[16:47:01.0000] <Hixie>
now people want to physically beat us :-(

[16:48:00.0000] <weinig>
Hixie: that can only mean you are doing things right

[16:51:00.0000] <Dashiva>
First they flame you, then they assault you, then you reach CR


2009-05-04
[17:41:00.0000] <Hixie>
interesting

[17:42:00.0000] <Hixie>
setTimeout() doesn't seem to throw if you pass it non-code non-string first arguments

[17:49:00.0000] <weinig>
Hixie: we recently changed that WebKit

[17:50:00.0000] <Hixie>
webkit returns undefined

[17:50:01.0000] <Hixie>
firefox actually just works

[17:50:02.0000] <Hixie>
(and doesn't do anything when the timeout comes)

[17:50:03.0000] <weinig>
well, I am not sure we threw, but we now call toString on whatever is passed if it is not a function

[17:50:04.0000] <weinig>
so setTimeout({ toString : function() {  }, 0) now works as it does in firefox

[17:50:05.0000] <Hixie>
that doesn't seem to be the behaviour

[17:51:00.0000] <Hixie>
setTimeout(true, 0); should return a number if that was the case

[17:51:01.0000] <Hixie>
but it returns undefined

[17:51:02.0000] <weinig>
Hixie: this is a very recent change

[17:51:03.0000] <weinig>
Hixie: in the last week or so

[17:51:04.0000] <weinig>
/me checks to make sure it was actually landed

[17:51:05.0000] <Hixie>
/me updates to the latest build in case he last updated just before it landed or something

[17:52:00.0000] <weinig>
yeah, javascript:alert(setTimeout(true, 0)) returns a number in ToT

[17:52:01.0000] <weinig>
in shipping, we return undefined

[17:52:02.0000] <Hixie>
i never test shipping builds

[17:53:00.0000] <Hixie>
shipping is just another word for "obsolete"

[17:53:01.0000] <weinig>
Hixie: a trunk build from a week ago would do the same :)

[17:53:02.0000] <weinig>
/me was playing around with setTimeout and noticed this difference so fixed it

[17:53:03.0000] <Hixie>
yup, it is indeed changed

[17:53:04.0000] <Hixie>
cool

[17:54:00.0000] <weinig>
happy to help :)

[17:55:00.0000] <weinig>
weird coincidence though

[17:56:00.0000] <Hixie>
well in my case it's just a matter of going down a list and this was the next item

[17:59:00.0000] <Hixie>
you do the same with the second argument?

[18:40:00.0000] <Hixie>
can anyone think of any unusual objects in JavaScript that have a [[Call]] thingy defined but aren't what you would consider regular functions?

[18:40:01.0000] <Hixie>
...and which have some detectable side-effect?

[18:42:00.0000] <olliej>
Hixie: huh?

[18:42:01.0000] <olliej>
Hixie: what do you mean?

[18:43:00.0000] <olliej>
Hixie: you mean standard runtime functions that aren't 'pure'?

[18:43:01.0000] <Hixie>
well e.g. select.options can be invoked as a function

[18:43:02.0000] <Hixie>
like select.options('a')

[18:43:03.0000] <Hixie>
as well as being a regular object in its own right

[18:44:00.0000] <Hixie>
with select.options.length, etc

[18:44:01.0000] <olliej>
Hixie: all functions are objects in their own right

[18:44:02.0000] <olliej>
Hixie: anyFunction.length

[18:44:03.0000] <olliej>
Hixie: will return the number of arguments to said function

[18:44:04.0000] <olliej>
Hixie: that said

[18:44:05.0000] <Hixie>
oh i should have said in the DOM, not in JS, my bad

[18:44:06.0000] <olliej>
Hixie: iirc /a/("foo") in js 

[18:44:07.0000] <olliej>
Hixie: oh

[18:44:08.0000] <olliej>
only the dom

[18:45:00.0000] <Hixie>
well in js too is fine

[18:45:01.0000] <Hixie>
does /a/("foo") have any detectable side-effects?

[18:45:02.0000] <Hixie>
does it change state when you invoke it or something?

[18:45:03.0000] <olliej>
Hixie: evaluating a regex puts the results on the RegExp constructor

[18:46:00.0000] <Hixie>
come again?

[18:46:01.0000] <Hixie>
wait, really?

[18:46:02.0000] <olliej>
Hixie: yeah

[18:46:03.0000] <olliej>
i discovered this last week

[18:46:04.0000] <Hixie>
that's awesome

[18:46:05.0000] <olliej>
or maybe two weeks ago

[18:46:06.0000] <olliej>
that made me sad

[18:46:07.0000] <olliej>
Hixie: just a mo

[18:46:08.0000] <olliej>
i'll dig up the details

[18:47:00.0000] <Hixie>
/.+/('eueouue'); RegExp.lastMatch

[18:47:01.0000] <Hixie>
that's super

[18:48:00.0000] <Hixie>
hahaha yes!

[18:48:01.0000] <Hixie>
olliej: thanks, that's what i needed

[18:48:02.0000] <Hixie>
setTimeout(/.*/, 0, 'test'); affects RegExp.lastMatch

[18:48:03.0000] <olliej>
ah ha

[18:48:04.0000] <Hixie>
man in this API ridiculous or what

[18:48:05.0000] <olliej>
Hixie: hehe

[18:48:06.0000] <olliej>
it made me very sad

[18:48:07.0000] <olliej>
Hixie: when i discovered that

[18:48:08.0000] <Hixie>
heycam: i need a new thing in WebIDL which means "this interface is only for JS if you're even thinking of using this in another language please go have a cold shower right now"

[18:49:00.0000] <heycam>
hah

[18:49:01.0000] <Hixie>
maybe, say, as a [HereBeDragons] interface decorator

[18:49:02.0000] <heycam>
can i ask, what interface is this?

[18:49:03.0000] <Hixie>
WindowTimers

[18:50:00.0000] <heycam>
huh.  how come?

[18:50:01.0000] <Hixie>
it has setTimeout and company on it, so that i can reuse them in workers without taking all of Window with it

[18:50:02.0000] <Hixie>
but if you mean "why is setTimeout so bad", consider that its signature is:

[18:50:03.0000] <Hixie>
long setTimeout(in any handler, [Optional] in any timeout, [Variadic] in any args)

[18:51:00.0000] <Hixie>
and the function's behaviour depends on whether handler has a [[Call]] method or not

[18:51:01.0000] <heycam>
so overloading a DOMString and a Callback interface isn't enough?

[18:52:00.0000] <Hixie>
what's the interface that defines a callback that the object /foo/ matches?

[18:52:01.0000] <Hixie>
(but which the string "foo" doesn't)

[18:53:00.0000] <heycam>
not sure what you mean

[18:53:01.0000] <Hixie>
how could overloading a DOMString and a Callback interface be enough?

[18:53:02.0000] <heycam>
currently i think you can overload and distinguish between DOMStrings and an object of an interface

[18:53:03.0000] <Hixie>
setTimeout() takes anything

[18:54:00.0000] <Hixie>
its first argument can be  { toString: function() { return 'alert(0)'; } }

[18:54:01.0000] <Hixie>
and you'll get an alert

[18:54:02.0000] <Hixie>
it can also be window.alert

[18:54:03.0000] <Hixie>
you'll also get an alert

[18:54:04.0000] <heycam>
ok maybe the rules don't exactly fit, since when you have overloading any old type won't get stringified to match the DOMString

[18:54:05.0000] <heycam>
iirc

[18:54:06.0000] <heycam>
but i should check...

[18:56:00.0000] <Hixie>
setTimeout(handler, { toString: function() { return '20.1'; } })  is treated the same as  setTimeout(handler, 20);

[18:56:01.0000] <Hixie>
if you give me the webidl to do that and i'll use it, but i think our time might be better spent just writing a few lines of prose

[18:57:00.0000] <heycam>
yeah

[18:57:01.0000] <heycam>
ok

[18:57:02.0000] <heycam>
you want a [Language=ECMAScript] or something?

[18:57:03.0000] <heycam>
or a term you can use in prose?

[18:58:00.0000] <Hixie>
actually i kinda think [HereBeDragons] would be pretty neat. but i could work with Language=ECMAScript too if you like.

[18:58:01.0000] <Hixie>
i just want something that prevents the idl from compiling

[18:58:02.0000] <heycam>
?

[18:58:03.0000] <Hixie>
so that people who use it have to go figure out wtf is going on

[18:59:00.0000] <heycam>
then perhaps idl is not what you're looking for?

[18:59:01.0000] <Hixie>
possibly

[18:59:02.0000] <heycam>
well, there's a slight advantage to still using it; you get the long return value

[18:59:03.0000] <heycam>
and the association of arguments to the argument names

[19:00:00.0000] <heycam>
but i guess that's about it =)

[19:00:01.0000] <Hixie>
it also implies that the methods are on the object, etc

[19:00:02.0000] <Hixie>
i think i'll keep using it

[19:00:03.0000] <heycam>
yeah

[19:00:04.0000] <heycam>
ok

[19:00:05.0000] <Hixie>
don't worry about it

[19:00:06.0000] <Hixie>
they'll have to implement this anyway i guess

[19:01:00.0000] <Hixie>
so they'll soon find what a... poorly designed API it is

[19:01:01.0000] <heycam>
are you going to supply a big idl file of all the definitions (or a zip file of separate idl files)?

[19:01:02.0000] <Hixie>
(not my first choice of words)

[19:01:03.0000] <Hixie>
wasn't planning on it

[19:01:04.0000] <Hixie>
but people might want it

[19:01:05.0000] <Hixie>
it's pretty easy to make it

[19:01:06.0000] <Hixie>
just strip all the blocks with class=idl

[19:01:07.0000] <heycam>
that's what i do currently (with a bookmarklet)

[19:02:00.0000] <Hixie>
though a few things will fail, e.g. WindowProxy isn't defined in idl

[19:02:01.0000] <Hixie>
only in prose

[19:02:02.0000] <heycam>
hmm

[19:02:03.0000] <heycam>
because it's difficult to specify?

[19:02:04.0000] <Hixie>
(and, well, see above are setTimeout()!)

[19:02:05.0000] <Hixie>
I have no idea how I'd define WindowProxy in IDL

[19:03:00.0000] <Hixie>
it is in every sense indistinguishable from Window

[19:03:01.0000] <Hixie>
and you can't get a hold of the Window object anyway

[19:03:02.0000] <Hixie>
so...

[19:03:03.0000] <heycam>
if it's indistinguishable, and you've got a definition of Window, then it should be easy to come up with a definition for WindowProxy

[19:04:00.0000] <heycam>
(i.e., just copy it)

[19:04:01.0000] <Hixie>
all it does is forward every operation to the underlying Window

[19:05:00.0000] <Hixie>
for example, it's prototype changes when the underlying Window changes

[19:05:01.0000] <Hixie>
its

[19:06:00.0000] <heycam>
does that mean it's impossible to specify in idl though?

[19:06:01.0000] <Hixie>
i think it means that it's not useful to specify in idl

[19:06:02.0000] <heycam>
ok

[19:07:00.0000] <heycam>
so what i'm concerned about is whether it'd be possible to "compile" or process however all the idl for html5

[19:07:01.0000] <heycam>
if there are dangling references to interfaces like WindowProxy...

[19:07:02.0000] <Hixie>
there are bits that need to be provided first

[19:07:03.0000] <Hixie>
but after they are provided, sure

[19:07:04.0000] <Hixie>
(such as WindowProxy)

[19:20:00.0000] <Hixie>
i wonder how to spec the whole "compile the string" thing

[22:07:00.0000] <hsivonen>
I wonder which one of jd's posts has resonated http://twitter.com/jbowtie/status/1691112133

[22:19:00.0000] <ojan>
Hixie: ping

[22:20:00.0000] <Hixie>
here

[22:21:00.0000] <ojan>
Hixie: i'm trying to see if it's worth trying to add undo/redo events to webkit...

[22:21:01.0000] <ojan>
Hixie: seems like they are tied to the UndoManager though...so we'd need to hash all that stuff out before moving forward?

[22:21:02.0000] <Hixie>
pretty much

[22:21:03.0000] <Hixie>
though nobody has suggested a better solution yet

[22:24:00.0000] <ojan>
Hixie: to be sure i'm understanding the intent, onundo/onredo would fire when the user picks undo/redo from a menu, fires the undo/redo execCommands or hits the appropriate key-command, right?

[22:24:01.0000] <Hixie>
yup

[22:34:00.0000] <ojan>
Hixie: i don't quite follow some bits here...

[22:34:01.0000] <ojan>
in 6.11.4

[22:34:02.0000] <ojan>
"If the entry immediately before the undo position is an undo object entry"

[22:35:00.0000] <ojan>
what else could it be?

[22:35:01.0000] <Hixie>
The entries are of two type: DOM changes and undo objects.

[22:35:02.0000] <Hixie>
click "undo transaction history"

[22:35:03.0000] <Hixie>
for more details

[22:36:00.0000] <ojan>
right, i saw that...so, undo/redo events don't fire for DOM changes?

[22:37:00.0000] <Hixie>
the undo/redo events are there just to let scripts perform the changes that need doing for the relevant objects they put on the list

[22:37:01.0000] <ojan>
ah

[22:37:02.0000] <ojan>
that's what i wasn't understanding

[22:37:03.0000] <Hixie>
the whole section is a mess imho

[22:37:04.0000] <Hixie>
i don't really like it

[22:38:00.0000] <Hixie>
(i don't have anything better to suggest though)

[22:38:01.0000] <ojan>
so, the thing you pass in to the add method is any arbitrary JS object of your choosing?

[22:38:02.0000] <Hixie>
right

[22:38:03.0000] <ojan>
i see

[22:38:04.0000] <ojan>
i'm starting to get the intention here now at least :)

[22:39:00.0000] <Hixie>
hehe

[22:39:01.0000] <ojan>
Hixie: i feel like it would be a ton simple to just have the object you pass in be required to have undo/redo methods

[22:40:00.0000] <ojan>
Hixie: gives more flexibility to the web developer in terms of when they setup the undo/redo state

[22:40:01.0000] <ojan>
(i.e. they can do it at the time of the change or at the time of the undo)

[22:40:02.0000] <ojan>
I guess you can do that already with the current API

[22:40:03.0000] <Hixie>
how would you handle the case where an undo operation has no matching redo?

[22:41:00.0000] <Hixie>
i recommend sending a more concrete proposal to the list

[22:41:01.0000] <ojan>
sure

[22:41:02.0000] <Hixie>
it really wouldn't take much to convince me to change the spec

[22:41:03.0000] <ojan>
have you considered undo/redo events for DOM changes?

[22:41:04.0000] <Hixie>
anything even semi-coherent would probably be convincing enough :-)

[22:42:00.0000] <Hixie>
i don't think i had use cases for those

[22:42:01.0000] <Hixie>
so probably not

[22:42:02.0000] <ojan>
ok, i'll try and draft an email then...

[22:42:03.0000] <Hixie>
(i'm not opposed though)

[22:42:04.0000] <ojan>
the use-case is for apps that want to capture the undo/redo and handle it themselves

[22:43:00.0000] <Hixie>
that's not a use case, that's a description of what it can do :-)

[22:45:00.0000] <ojan>
Hixie: as an example, the google rich text editor keeps its own undo stack. right now it doesn't do the right thing if you undo/redo from the menu because the JS code doesn't know that an undo happened. it can't catch and cancel the event.

[22:45:01.0000] <ojan>
Hixie: anyways, I'll put this stuff in an email

[22:46:00.0000] <Hixie>
k

[22:47:00.0000] <Hixie>
(why would an editor _want_ to do things itself, assuming an ideal implementation?)

[22:50:00.0000] <ojan>
Hixie: yeah, that's true, the google editor would use the browser's undo if it didn't suck so much

[22:53:00.0000] <ojan>
Hixie: if you have an editor where the backing store is not HTML, you want to be able to update your sense of the state when an undo/redo happens

[22:54:00.0000] <Hixie>
if you have an editor where the backing store is not HTML, surely you just want to catch any DOM changes

[22:54:01.0000] <Hixie>
not necessarily undo/redo ones in particular

[23:02:00.0000] <ojan>
yes, but you want to be able to know what happened, and in most cases, be able to cancel the event

[23:03:00.0000] <ojan>
in practice, mutation events are far too noisy and hard to work with

[23:03:01.0000] <Hixie>
it's possible that we need to fix mutation events

[23:05:00.0000] <hsivonen>
Hixie: posting them on the event loop or something more drastic?

[23:05:01.0000] <ojan>
Hixie: in practice, mutation events are useless for anything other than a high level sense that something changed

[23:06:00.0000] <Hixie>
hsivonen: i haven't studied the problem, so i don't know

[23:06:01.0000] <ojan>
Hixie: maybe that's a bit harsh, but i've never heard of them actually being used for anything else

[23:06:02.0000] <Hixie>
ojan: it would be good to tell www-dom⊙wo this, as that is where the DOM3 Events spec is being discussed

[23:06:03.0000] <Hixie>
and mutation events are on the table

[23:07:00.0000] <ojan>
/me goes to join yet another mailing list

[23:07:01.0000] <Hixie>
yeah not sure why that's not on public-webapps

[23:08:00.0000] <Hixie>
(in particular, not sure why that's elsewhere but widgets are on public-webapps)

[23:43:00.0000] <ojan>
Hixie: yeah, the more i look at mutation events for this, the more i think that they're just at the wrong granularity for the use-case in question. you actually want something that conveys the semantics of the user-action, not something the just tells you all the details of what changed after the fact

[00:03:00.0000] <ojan>
Hixie: more random questions...

[00:04:00.0000] <ojan>
dataTransfer only supports "Text" and "URL" as types

[00:04:01.0000] <ojan>
is that just because that's what IE does? or is there some reason not to also support HTML?

[00:07:00.0000] <Hixie>
dataTransfer supports any type

[00:07:01.0000] <Hixie>
where does it say it only supports Text and URL?

[00:09:00.0000] <ojan>
under setData/getData it only list Text and URL as options

[00:09:01.0000] <ojan>
i guess it doesn't forbid other options

[00:10:00.0000] <Hixie>
oh those are just two values that are supported for legacy reasons

[00:10:01.0000] <Hixie>
because the value is supposed to be a mime type

[00:10:02.0000] <ojan>
oh, i see...so to support HTML text we would just use the mime type directly?

[00:10:03.0000] <ojan>
i.e. text/html

[00:11:00.0000] <Hixie>
yup

[00:11:01.0000] <ojan>
Hixie: great. we'll attempt adding such to webkit.

[00:11:02.0000] <Hixie>
cool

[00:12:00.0000] <ojan>
thanks for talking through all my random questions :)

[00:12:01.0000] <Hixie>
np

[00:21:00.0000] <Hixie>
feedback i just got from someone not (as far as i know) affiliated with any existing solution for microdata:

[00:22:00.0000] <Hixie>
* they'd like to share event listing data on their city "what's on" guide web site "in a way that retains the event information in a meaningful way"

[00:23:00.0000] <Hixie>
asked why microformats don't work for them, they had several replies that apply equally to microformats and other technologies, and one that is specific to microformats:

[00:23:01.0000] <Hixie>
"There are no obvious parsing tools for people to actually do anything with the data (other than add an event to a calendar)"

[00:30:00.0000] <zcorpan>
Hixie: <li><p>If none of the values (if any) result in an access key being

[00:30:01.0000] <zcorpan>
+   assigned, then the element has no <span>assigned access

[00:30:02.0000] <zcorpan>
+   key</span>.</p></li>

[00:31:00.0000] <zcorpan>
Hixie: why not allow the UA to make up its own shortcut in that case?

[00:31:01.0000] <zcorpan>
Hixie: e.g. based on textContent

[00:31:02.0000] <zcorpan>
Hixie: or based on the keys available in the UA

[00:33:00.0000] <Hixie>
i guess we could...

[00:37:00.0000] <zcorpan>
Hixie: there's some magic required with focusing and accesskey that i don't quite remember

[00:38:00.0000] <zcorpan>
Hixie: e.g. for fieldset you should focus the first descendant form control or something

[00:38:01.0000] <Hixie>
send mail for these issues, i'll look at them in due course

[01:26:00.0000] <zcorpan>
maybe accesskeyLabel should be available to css 'content'

[01:28:00.0000] <zcorpan>
:has-accesskey::after { content:'(' accesskey ')' }

[01:31:00.0000] <hendry>
how are you supposed to express last modified time (mtime) and creation time (ctime) of a page in markup?

[01:33:00.0000] <Philip`>
hendry: <p>Last modified: 2009-05-04.</p>

[01:36:00.0000] <hendry>
Philip`: so no time element? :)

[01:37:00.0000] <Philip`>
hendry: What's the use case for using a special element here? :-)

[01:37:01.0000] <zcorpan>
"You would say just "n" for now, then, when the new feature is wildly deployed, " -- Hixie expects accesskey to be *wildly* deployed?

[01:38:00.0000] <zcorpan>
Philip`: maybe you would add the date to your calendar so you can compare next time you look at the page if it has been modified since you last read it

[01:39:00.0000] <hendry>
Philip`: i'm helping out moving ikiwiki templates to html5. every page has a ctime and mtime in there, and i was wondering how best to mark it up

[01:41:00.0000] <hendry>
Philip`: here is an example of the markup so far http://h.dabase.com/html5-tmp/html5/

[01:48:00.0000] <Lachy>
Philip`, in hendry's example, the benefit of using the time element would be that the browser can, at the user's request, present the date in a more familiar format. (Though, this depends on future UAs implementing such a feature)

[01:50:00.0000] <Lachy>
although, such a feature would be more useful on sites that insist on using ambiguous localised date formats like 04/05/09

[01:52:00.0000] <hendry>
Lachy: thanks, that's a good use case

[02:02:00.0000] <jgraham>
Lachy: Hopefully UAs won't implement such a feature

[02:02:01.0000] <jgraham>
Since I guess it will be more confusing than the status-quo

[02:03:00.0000] <jgraham>
(although I suppose some good implementation is possible like displaying the time as the author wrote it and giving the option to display it in some other format)

[02:04:00.0000] <Philip`>
They could display the localised time in a tooltip, perhaps

[02:04:01.0000] <jgraham>
/me notes that js also has magic $1-$9 variables on the RegExp object which point to the first 9 match groups in the last regexp called

[02:05:00.0000] <jgraham>
I shoul really document that on the web ecmascript page, I guess

[02:07:00.0000] <Lachy>
jgraham, that's one of the major problems solved by having the time element, so I hope they do implement it in some way.

[02:11:00.0000] <jgraham>
Lachy: Mixing up author supplied content and UA supplied content is likely to be really confusing though

[02:11:01.0000] <Philip`>
Lachy: That's a problem solved by having authors write dates like "2009-05-04" instead of "04/05/09", with no need for any markup

[02:14:00.0000] <Philip`>
Anyway there'll be too much legacy content saying <time datetime=2007-10-05>December 3rd 2009</time> for UAs to replace the textual content with something derived from the datetime

[02:17:00.0000] <Lachy>
Philip`, getting authors to write dates in pages only using unambiguous date formats isn't a viable solution in practice.  Letting authors use whatever date format they like like in the page, while providing a way to mark up the date unambiguously and letting UAs expose that to users somehow is

[02:18:00.0000] <Lachy>
jgraham, I'm not proposing that UAs replace the text in the page with the unambiguous format by default, but rather provide a way for the user to access it on request

[02:19:00.0000] <Philip`>
Lachy: Why is getting authors to write dates in pages only using unambiguous date formats not a viable solution in practice?

[02:20:00.0000] <Philip`>
/me very rarely sees pages that say "01/02/03" so it seems like most people already use unambiguous date formats

[02:20:01.0000] <Lachy>
because we have ~15 years of experience showing us that authors still publish ambiguous date formats

[02:20:02.0000] <Lachy>
I see it relatively often

[02:22:00.0000] <jgraham>
/me wonders if authours who use ambiguous date formats would be likely to use <time>

[02:22:01.0000] <Lachy>
here's a page I recall seeing http://news.zdnet.com/2100-9595_22-293658.html scroll down to the comments

[02:25:00.0000] <Lachy>
if authors see practical benefits for users provided by UAs by using it, then they will

[02:26:00.0000] <Lachy>
but if <time> is just seen as useless metadata that doesn't help users at all in major browsers, then it's less likely

[02:27:00.0000] <Philip`>
They would also have to think the aesthetic advantages of writing mm/dd/yy outweigh the practical benefits for users of just writing yyyy-mm-dd with no markup

[02:28:00.0000] <Philip`>
Actually: They would also have to think the aesthetic advantages of writing mm/dd/yy, plus the practical benefits for users of whatever their UA does with <time>, minus the cost of adding and maintaining the extra <time> markup, outweigh the practical benefits for users of just writing yyyy-mm-dd with no markup

[02:29:00.0000] <Philip`>
Also they would have to know about and care about the issue, which I presume many people don't

[02:32:00.0000] <Lachy>
Philip`, from my experience, a lot of american sites do seem to prefer the asthetics of writing mm/dd/yy over the less ambiguous yyyy-mm-dd format

[02:32:01.0000] <olliej>
Lachy: yeah

[02:33:00.0000] <olliej>
Lachy: i've actually taken to writing dd-mmm-yyyy on everything

[02:33:01.0000] <olliej>
simply to avoid confusion

[02:33:02.0000] <Lachy>
and when I was doing web development a few years ago, many of the clients/designers demanded that I use dd/mm/yy (for Australian sites) despite my objectinos

[02:33:03.0000] <olliej>
Lachy: (NZ does dd/mm/yy so i keep getting confused)

[02:33:04.0000] <Lachy>
olliej, I know

[02:33:05.0000] <Lachy>
so does Australia

[02:34:00.0000] <olliej>
Lachy: yeah, you said above

[02:34:01.0000] <olliej>
(i hit enter just after your comment)

[02:34:02.0000] <Lachy>
and, IIRC, the UK (though I could be wrong about that)

[02:37:00.0000] <Rik`>
why mm/dd/yy exists ?

[02:42:00.0000] <olliej>
Rik`: because the US exists

[02:42:01.0000] <Rik`>
yeah but how did they come up with that ?

[02:42:02.0000] <Philip`>
Lachy: (It's dd/mm/yy in the UK)

[02:43:00.0000] <olliej>
Rik`: they went "how does the rest of the world do this? let's do it differently"

[02:43:01.0000] <olliej>
Rik`: you can't talk you don't even have the right names for your months ;)

[02:44:00.0000] <Rik`>
hum ?

[02:44:01.0000] <olliej>
Rik`: i'm joking :D

[02:44:02.0000] <olliej>
Rik`: referring to france not using english month names :D

[02:44:03.0000] <olliej>
for obvious reasons :D

[02:45:00.0000] <Philip`>
"The mm/dd/yy format is used in: Belize, Canada, Federated States of Micronesia, Kenya, Palau, Philippines, Puerto Rico, United States"

[02:45:01.0000] <Philip`>
(says Wikipedia)

[02:45:02.0000] <Rik`>
ohh, like the rest of the world except england, australia and new zealand ? :)

[02:46:00.0000] <olliej>
Rik`: hehe

[02:47:00.0000] <Rik`>
and we're not yet talking about miles and gallons :)

[02:48:00.0000] <olliej>
shudder

[02:48:01.0000] <olliej>
hatehatehate

[02:52:00.0000] <Philip`>
/me wonders whether those are the big gallons or the little gallons

[02:54:00.0000] <Lachy>
I suspect the US went with mm/dd/yy so that the numerical order matched the order when spoken as words, like May 4th 2009

[02:55:00.0000] <Lachy>
whereas Australians tend to say 4th of May, 2009, and use dd/mm/yy

[02:58:00.0000] <Lachy>
wow, apparently the US also do crazy street address formats by putting the apartment number after the street number/name http://en.wikipedia.org/wiki/Endianness#Clarifying_analogy

[03:00:00.0000] <olliej>
Lachy: yeah

[03:00:01.0000] <olliej>
Lachy: that took me a while as well

[03:00:02.0000] <olliej>
Lachy: the US is a bizarre place

[03:01:00.0000] <olliej>
Lachy: oh, and i say 4th May, 2009 :D

[03:01:01.0000] <Lachy>
I'd better get used to it since I'm moving there in July

[03:02:00.0000] <olliej>
Lachy: where abouts?

[03:02:01.0000] <Lachy>
Mountain View, California

[03:02:02.0000] <olliej>
Lachy: ah ha

[03:03:00.0000] <olliej>
Lachy: we should meet up at some point

[03:03:01.0000] <Lachy>
ok

[03:54:00.0000] <annevk42>
http://www.webdirections.org/blog/ask-the-browser-makers-pete-lepage-from-the-internet-explorer-team/ -- see question 5

[04:09:00.0000] <jgraham>
So, say I had a reasonably well structured word document and I wanted to convert it to HTML in a way that roughly preserved the structure but that also added links e.g. from the TOC and from certain terms to their defining instance (which would always be a subsection). WWhat tools could I use to help me achieve that?

[04:13:00.0000] <Lachy>
jgraham, depending on the techniques used to create the Word document, then I've found copying and pasting from Word into Dreamweaver to be a rather successful technique for maintaining structure

[04:14:00.0000] <Lachy>
though it requires the use of the Styles and Formatting task pane in word to do the styles, and the use of the some HTML specific styles (available from the customisation dialog), and there's a Word specific setting in dreamweaver that needs to be turned on to get paragraphs to work right

[04:14:01.0000] <jgraham>
Lachy: Let's assume for the sake of argument that I don't have dreamweaver :)

[04:15:00.0000] <jgraham>
(I could probably download a trial version at home or something but it's not quite convenient)

[04:15:01.0000] <Lachy>
what HTML editor do you have?

[04:16:00.0000] <Lachy>
http://www.google.com/search?q=word+to+HTML+converter

[04:16:01.0000] <jgraham>
Lachy: emacs

[04:16:02.0000] <jgraham>
/me thinks that might not have been the right answer

[04:18:00.0000] <Lachy>
then I can't help you with that.

[04:19:00.0000] <jgraham>
It seems like the oo.o HTML export + a considerable amount of scripting might be a good enough solution

[04:23:00.0000] <Lachy>
have you tried MS Word's own HTML saving feature?

[04:23:01.0000] <Lachy>
it has a clean HTML option

[04:23:02.0000] <jgraham>
Lachy: I don't have MS Word either

[04:23:03.0000] <Lachy>
ok

[04:23:04.0000] <Philip`>
You could copy-and-paste from Word into a JavaScript editing component

[04:23:05.0000] <Philip`>
Oh

[04:24:00.0000] <Lachy>
jgraham, I have Word. If you sent me the file, I could try doing that for you

[04:24:01.0000] <jgraham>
On the other hand, since the document in question is readily publically avaliable it is not a problem if someone else does these things for me and sends me the result

[04:25:00.0000] <jgraham>
Lachy: It is the ES5 spec :) http://wiki.ecmascript.org/lib/exe/fetch.php?id=es3.1%3Aes3.1_proposal_working_draft&cache=cache&media=es3.1:candidate-draft-tc39-2009-025.doc

[04:28:00.0000] <Lachy>
this is going to take a while. I have to do it on my Windows machine at home over VNC, which is itself being ssh-tunnelled over a wifi connection through my imac, so the connection is a little slow

[04:30:00.0000] <jgraham>
Lachy: It's not a problem. Getting a TML version of the document is more of a long-term solution to the issue that I spend too long searching for stuff in the spec :)

[04:30:01.0000] <jgraham>
*HTML

[04:32:00.0000] <Lachy>
ah, ok. Good, then it can wait till after I go have some lunch

[04:34:00.0000] <jgraham>
Lachy: Sure. Take as long as you like. Thanks for the help :)

[05:13:00.0000] <Lachy>
jgraham, http://lachy.id.au/temp/ecma-tc39.zip

[05:16:00.0000] <jgraham>
Lachy: That gives me a 403

[05:20:00.0000] <Lachy>
fixed

[05:23:00.0000] <jgraham>
Lachy: Thanks

[05:24:00.0000] <jgraham>
(at a glance it doesn't seem like the Word output is substantailly better than the Oo.o output; both will need substantial cleanup)

[05:25:00.0000] <jgraham>
Or rather either since I don't intend to do both

[05:30:00.0000] <olliej>
is it just me or does the ES5 spec not document deletability/replacability of standard properties?

[05:33:00.0000] <jgraham>
olliej: What do you mean standard properties? Properties have the Writable:true attribute set unless otherwise specified

[05:34:00.0000] <jgraham>
s/Properties/Properties of built in objects/

[05:34:01.0000] <jgraham>
(see section 15, pg 98 or PDF page 112)

[05:36:00.0000] <jgraham>
(oh and Configurable: true)

[05:42:00.0000] <olliej>
jgraham: yeah, cheers for that

[05:42:01.0000] <olliej>
jgraham: terminology changed from DontDelete to writable:false

[05:46:00.0000] <jgraham>
olliej: Indeed. There's no point in releasing a new spec without a few cosmetic changes to screw with the brains of people familar with the old spec ;)

[05:46:01.0000] <olliej>
jgraham: hehehe

[06:10:00.0000] <jgraham>
zcorpan: YT?

[06:33:00.0000] <Lachy>
looking at a couple of rental sites for Mountain View, they all seem to share a major usability problem! They don't state whether the prices are monthly, fortnightly or weekly! :-(

[06:34:00.0000] <Lachy>
oh, yeah, one of them does actually. Was just hard to find

[06:35:00.0000] <jgraham>
Lachy: Sweden has that underlying poblem writ large; it is assumed that everyone already knows all the rules so they don' bother to state them until you violate them 

[06:38:00.0000] <Lachy>
/me realises that finding an apartment in the right area is hard for someone who knows absolutely nothing about US geography.

[06:39:00.0000] <olliej>
Lachy: typically monthly

[06:39:01.0000] <olliej>
Lachy: and they will almost cetainly be assuming a 12 month lease

[06:39:02.0000] <Lachy>
yeah, 12 month lease is normal

[06:40:00.0000] <jgraham>
Lachy: For where? e.g. here it is rather unusual or illegal or something

[06:40:01.0000] <Lachy>
olliej, what sort of price range is reasonable for that area?

[06:41:00.0000] <olliej>
Lachy: honestly not sure about MV

[06:41:01.0000] <olliej>
Lachy: i have a flatmate and our aparments rent is i think $US2400/mo (in cupertino)

[06:41:02.0000] <Lachy>
jgraham, illegal to offer 12 month leases? wtf?

[06:42:00.0000] <Lachy>
in Australia, leases are generally either 6 or 12 months

[06:42:01.0000] <jgraham>
Lachy: I'm not sure it is actually illegal. It's more that contracts aren't usually (legally?) for fixed periods but are continuous. And it is rather hard for landlords to kick you out unless you don't pay the rent

[06:43:00.0000] <Lachy>
jgraham, are there minimum lease terms imposed?

[06:44:00.0000] <olliej>
jgraham: leases are more about ensuring a minimum time

[06:44:01.0000] <olliej>
rather than a maximum

[06:44:02.0000] <jgraham>
Lachy: I think the only requirement in our contract is that we give 3 months notice if we want to leave

[06:44:03.0000] <jgraham>
So that is a kind of minimum

[06:45:00.0000] <Lachy>
wow, 3 months is an absurdly long notice period

[06:45:01.0000] <olliej>
yeah

[06:45:02.0000] <jgraham>
Yeah, it is pretty long

[06:45:03.0000] <Lachy>
but I think that's like Norway

[06:45:04.0000] <olliej>
when i wasn't in a lease, i only had to give 2 weeks notice

[06:46:00.0000] <Lachy>
in Australia, the term means that neither the tennant or landlord can break the lease before that period, and then after that it continues with a requirement for 2 weeks notice

[06:46:01.0000] <jgraham>
It seems to be the Swedish way. The landlord has no right to get rid of you (wwhich is good) but it is rather hard to leave (which is bad)

[06:47:00.0000] <Lachy>
(except when the tennant does something like significantly damages the property, then the landlord can evict the tennant before the end of the lease and keep the bond)

[06:48:00.0000] <Lachy>
olliej, is it usually a 1 month bond as well?

[06:48:01.0000] <olliej>
yeah

[06:48:02.0000] <Lachy>
ok

[06:48:03.0000] <olliej>
i may have had a bit more

[06:48:04.0000] <jgraham>
(We didn't have to pay any deposit or anything either)

[06:48:05.0000] <olliej>
due to my complete absence of a credit history

[06:49:00.0000] <Lachy>
I should check if Opera will be providing me with temporary accommodation so I can look for an apartment when I get there

[06:49:01.0000] <Lachy>
I'd rather not have to rent something before I get a chance to actually go over there and see the place

[06:49:02.0000] <jgraham>
(and the landlady seems to be part of some collective which negotiates the rent with the tennants association and then assigns the right rent to each apartment based on size, facilities, condition, etc.)

[06:50:00.0000] <jgraham>
(at least that's how I understand it)

[06:50:01.0000] <Lachy>
I like the deal I get in Norway by staying in one of Opera's apartments :-)

[06:50:02.0000] <Rik|work>
in France, you have to give a 3 months notice before leaving but you can arrange with the owner if you find someone to replace you

[07:18:00.0000] <gsnedders>
Germany is similar to France, but it is 1 month, IIRC

[07:19:00.0000] <gsnedders>
Wow. It is almost two hours by bus from Stirling to here

[07:19:01.0000] <olliej>
erk

[07:19:02.0000] <olliej>
bed time i think

[07:19:03.0000] <gsnedders>
/me shakes his head at olliej 

[07:19:04.0000] <olliej>
gsnedders: ?

[07:20:00.0000] <gsnedders>
olliej: Your sleeping his hours…

[07:20:01.0000] <Lachy>
gsnedders, do you need to travel to Stirling?

[07:20:02.0000] <gsnedders>
s/his//

[07:20:03.0000] <olliej>
gsnedders: are awesome? ;D

[07:20:04.0000] <gsnedders>
Lachy: Well, yes, but only from Stirling to here; I'll be coming from London

[07:20:05.0000] <gsnedders>
olliej: :D

[07:20:06.0000] <olliej>
gsnedders: i hate x86 assembly

[07:20:07.0000] <Lachy>
gsnedders, where is "here"?

[07:20:08.0000] <Rik|work>
Stirling, like the engine ?

[07:20:09.0000] <olliej>
i have been up all night fighting it

[07:21:00.0000] <Rik|work>
olliej is a knight fighting evil while you sleep

[07:22:00.0000] <Lachy>
2 hours doesn't seem too long though. Is that about 180 to 200km distance?

[07:22:01.0000] <gsnedders>
Lachy: St Andrews

[07:22:02.0000] <gsnedders>
Lachy: No

[07:22:03.0000] <Lachy>
how far is it?

[07:22:04.0000] <gsnedders>
/me looks up

[07:22:05.0000] <gsnedders>
70km or so, I'd guess

[07:23:00.0000] <gsnedders>
/me has to do 6502 Assembly for a practical computing assessment

[07:23:01.0000] <gsnedders>
How useful!

[07:23:02.0000] <Lachy>
google maps says 83

[07:24:00.0000] <gsnedders>
Yeah, 83km

[07:24:01.0000] <Lachy>
that must be an incredibly slow bus, travelling at an average speed less than 50km/h

[07:25:00.0000] <gsnedders>
The route it takes is unusually sane for a bus around here

[07:25:01.0000] <Lachy>
hmm, I guess it's passing through a fair few towns on the way, which would slow it down

[07:25:02.0000] <gavin_>
average speed of less than 50km for a bus in a city is not "incredibly slow"

[07:25:03.0000] <gsnedders>
A lot of towns.

[07:26:00.0000] <gsnedders>
gavin_: This is certainly not a city whatsoever.

[07:27:00.0000] <Lachy>
gsnedders, is the maximum speed limit for a bus there about 100km/h, outside of towns?

[07:27:01.0000] <gsnedders>
I think it is 60mph, so yeah.

[07:27:02.0000] <Lachy>
like on highways, etc.

[07:27:03.0000] <gsnedders>
100 is 62.3km/h

[07:27:04.0000] <Lachy>
wtf? You measure in miles there?!

[07:27:05.0000] <gsnedders>
Oh sure.

[07:27:06.0000] <Lachy>
why? I though only the US was backwards enough to be using miles

[07:27:07.0000] <gsnedders>
We also measure fuel economy in miles per gallon (imperial gallons, not US gallons), and buy petrol in litres.

[07:28:00.0000] <olliej>
Lachy: i believe the highway speed limit is 60 or 65mph

[07:28:01.0000] <Lachy>
huh?

[07:28:02.0000] <olliej>
Lachy: and i wouldn't plan on relying on buses/public transport

[07:28:03.0000] <gsnedders>
olliej: We aren't talking about CA

[07:28:04.0000] <gsnedders>
olliej: We're talking about somewhere more awesome. :D

[07:28:05.0000] <olliej>
gsnedders: oh?

[07:28:06.0000] <olliej>
gsnedders: not hard

[07:28:07.0000] <olliej>
:D

[07:28:08.0000] <gsnedders>
olliej: Scotland, here.

[07:29:00.0000] <gsnedders>
olliej: Unless you want to find my house in CA :D

[07:29:01.0000] <olliej>
hehe

[07:29:02.0000] <Lachy>
olliej, I've been warned about the poor bus services in CA before.

[07:29:03.0000] <olliej>
erk

[07:29:04.0000] <gsnedders>
http://www.stagecoachbus.com/timetables/23pckttt18082008[1].pdf doesn't ever seem to be overly slow, anywhere, it just seems to be slightly slow for the entire route and it just adds up

[07:30:00.0000] <zcorpan>
jgraham: pong

[07:30:01.0000] <beowulf>
Lachy: what's wrong with miles?

[07:31:00.0000] <Lachy>
beowulf, the fact that they aren't metric, and doing maths with them is hard

[07:31:01.0000] <olliej>
gah

[07:31:02.0000] <olliej>
why am i still up

[07:31:03.0000] <olliej>
night all

[07:31:04.0000] <gsnedders>
Hm, Google Maps says only 16 hours and 18 minutes on foot.

[07:32:00.0000] <Lachy>
beowulf, try making sense out of why there's 12 inches/foot, 3 feet/yard, and 1760 yards/mile, or 5280 feet/mile. There's just no logic there.

[07:32:01.0000] <gsnedders>
Lachy: You actually know that? I don't.

[07:33:00.0000] <Lachy>
(and I had to look those last 2 values up, cause I've got no hope of remembering them)

[07:33:01.0000] <Lachy>
gsnedders, what's the difference between imperial and US gallons?

[07:34:00.0000] <gsnedders>
Lachy: They are a different amount.

[07:34:01.0000] <gsnedders>
Lachy: More detailed than that, I don't know and don't care.

[07:34:02.0000] <Lachy>
/me looks it up

[07:34:03.0000] <gsnedders>
(I have software to convert it to litres, so I don't care.)

[07:35:00.0000] <beowulf>
/me wonders when in real life Lachy needed to know how many feet where in a mile

[07:35:01.0000] <beowulf>
/me lives in Ireland, on one side of the border we measure in miles, on the other, km

[07:36:00.0000] <gsnedders>
/me ought to go to Ireland some time

[07:36:01.0000] <Lachy>
beowulf, luckily, I come from a sensible country that uses the metric system, so I don't need to know. But I might need to when I go to the US

[07:36:02.0000] <gsnedders>
/me is going to be taking a stupid route home from Linköping, via Berlin, London, Southampton, and finally Stirling before getting back to St Andrews

[07:37:00.0000] <Lachy>
gsnedders, so you did get the internship at Opera?

[07:37:01.0000] <gsnedders>
Lachy: Yes

[07:37:02.0000] <Lachy>
cool

[07:37:03.0000] <beowulf>
/me makes note to invite gsnedders to the next whatever camp to talk about html5 or something

[07:37:04.0000] <Lachy>
when?

[07:37:05.0000] <gsnedders>
Lachy: June 29th to August 28th

[07:37:06.0000] <gsnedders>
beowulf: Also: get me money to go there :P

[07:37:07.0000] <Lachy>
ok

[07:38:00.0000] <beowulf>
gsnedders: would sir prefer HSS or regular ferry?

[07:38:01.0000] <gsnedders>
HSS?

[07:38:02.0000] <gsnedders>
http://en.wikipedia.org/wiki/High-speed_Sea_Service?

[07:38:03.0000] <beowulf>
gsnedders: high speed ... service? the catamaran thing

[07:39:00.0000] <gsnedders>
Ah. I wouldn't really care :)

[07:39:01.0000] <beowulf>
:)

[07:39:02.0000] <gsnedders>
Provided I'm not meant to be doing anything else at the time (which is not much next year) I don't care how long it takes

[07:39:03.0000] <gsnedders>
Actually, more to the point, I'm not doing anything that involves me being anywhere specific, I can live fine with my laptop and do what is needed.

[07:40:00.0000] <jgraham>
Lachy: Mathematics with miles: 0.1 miles, 0.5 miles, 1.0 miles 100 miles. Cars on the motorway travel at an average of about 1 mile/minute

[07:40:01.0000] <beowulf>
Lachy: so you'd struggle if i told how how far my house is from here in fields?

[07:41:00.0000] <Lachy>
beowulf, is "fields" a standardised unit of measure?

[07:42:00.0000] <Dashiva>
Fields medal radii

[07:42:01.0000] <Lachy>
Dashiva, what?

[07:43:00.0000] <Dashiva>
I figured that would be standardized :)

[07:43:01.0000] <beowulf>
Lachy, the distance from one hedge to the other?

[07:43:02.0000] <Lachy>
/me assumes beowulf is now just being silly

[07:44:00.0000] <gsnedders>
Only 10:36 from train station to arriving by ferry in Dublin :P

[07:44:01.0000] <gsnedders>
Oh, and five changes :P

[07:45:00.0000] <Lachy>
oh, wow, apparently there are 128 ounces in a gallon (US). That also makes things complicated

[07:46:00.0000] <jgraham>
/me wonders why gsnedders is taking such a convoluted route home

[07:46:01.0000] <gsnedders>
128 is a nice round number!

[07:46:02.0000] <Lachy>
is there a smaller unit than a fluid ounce?

[07:46:03.0000] <Lachy>
gsnedders, only if doing maths in binary

[07:47:00.0000] <jgraham>
Lachy: Sure. But I guess you mean a smaller imperial unit

[07:47:01.0000] <gsnedders>
jgraham: Well, I'm dropping into Berlin to see my sister (and my probable niece), then off to London to spend the night there, to Southampton for uni open day, then off to Stirling for conference on the same evening

[07:47:02.0000] <Lachy>
jgraham, yes

[07:47:03.0000] <gavin_>
you have a probable niece?

[07:47:04.0000] <gsnedders>
Yes.

[07:47:05.0000] <jgraham>
|niece>

[07:48:00.0000] <gsnedders>
She might be a nephew.

[07:48:01.0000] <Lachy>
jgraham, actually, I mean in US units, not imperial units

[07:48:02.0000] <beowulf>
Lachy: a dram?

[07:48:03.0000] <beowulf>
Lachy: not a dram then

[07:48:04.0000] <Lachy>
oh, I forgot about the pint. Better find out how much that is

[07:48:05.0000] <beowulf>
/me knows nothing of US units

[07:49:00.0000] <gsnedders>
Very approx. 500ml

[07:49:01.0000] <jgraham>
Lachy: A pint is the unit that ber comes in. That is all that you need to know

[07:49:02.0000] <jgraham>
*beer

[07:49:03.0000] <gsnedders>
/me thinks going to Dublin would be easier by air :)

[07:51:00.0000] <beowulf>
gsnedders: dublin is expensive coming from the uk, euro/sterling exchange rate and then the recession

[07:51:01.0000] <Lachy>
jgraham, beer should come in either a Schooner or a Middy

[07:52:00.0000] <jgraham>
/me laughs

[07:52:01.0000] <gsnedders>
beowulf: Peh. I have no money any way. What difference does it make? :)

[07:53:00.0000] <Lachy>
jgraham, why is that funny?

[07:54:00.0000] <jgraham>
Uh because they are such silly names? A schooner is a type of boat and, well, if you can't see why "middy" sounds funny there is no hope for you :)

[08:21:00.0000] <gsnedders>
Lachy: Avoid wearing glasses.

[08:25:00.0000] <Lachy>
gsnedders, my choices are: wear glasses, or avoid needing to look at anything that isn't close enough to see clearly

[08:26:00.0000] <gsnedders>
Lachy: I do the latter.

[08:26:01.0000] <Lachy>
/me notes that even the computer screen at arms length in front of me is a little blurry without glasses

[08:29:00.0000] <jgraham>
/me did the latter for rather a long time

[08:31:00.0000] <gsnedders>
/me does have glasses, just only wears them when he can't do the latter

[08:33:00.0000] <Lachy>
I'm getting new lenses when I go to Australia next week. I was really hoping the insurance would cover the cost :-(

[08:47:00.0000] <gsnedders>
/me isn't really convinced by the ending of Pride and Prejudice

[08:49:00.0000] <Lachy>
how does it end?

[08:50:00.0000] <gsnedders>
It just quickly says a lot trying to wrap up all the story arcs, instead of properly ending.

[08:51:00.0000] <jgraham>
/me isn't really convinced that finding only 16 / 1289370 pages with a <html> element is realistic and not, say, a bug in his script

[08:52:00.0000] <jgraham>
(As a collary, the 1289370 number might well be wrong)

[08:59:00.0000] <Lachy>
jgraham, are you searching for pages that use the start tag compared with those that don't?

[09:00:00.0000] <jgraham>
Lachy: No, I'm just testing the code that I wrote to look at the dotnetdotcom data. My conclusion is that the code is broken. This is not really surprising :)

[09:08:00.0000] <gsnedders>
/me thought jgraham was a brilliant programmer ;P

[09:14:00.0000] <jgraham>
Because gsnedders never has bugs

[09:15:00.0000] <gsnedders>
No, never.

[09:15:01.0000] <beowulf>
not closing your lists in an <a> is a bad idea, you get the same blocks within a element thing as from packet boundaries 

[09:18:00.0000] <beowulf>
http://carisenda.com/sandbox/blocka/ 

[09:44:00.0000] <Philip`>
jgraham: There should be something like 425K pages, not 1.3M

[09:45:00.0000] <Philip`>
jgraham: By the way, I hope you're planning to process all the ~20GB of data in Python in finite time :-)

[09:45:01.0000] <Philip`>
Uh

[09:45:02.0000] <Philip`>
*not planning

[09:46:00.0000] <gsnedders>
/me points at the parallelizm module

[09:47:00.0000] <Philip`>
/me points at the orders of magnitude by which Python is slower than fast languages

[09:47:01.0000] <Philip`>
and the non-corresponding orders of magnitudes more processing cores that you can run it on

[09:48:00.0000] <gsnedders>
/me notes he has been wondering about using C to do some processing with the 20GB of data

[09:48:01.0000] <jgraham>
Philip`: I was planning on seeing how hard it was to process in python and then hen that failed to process it using hadoop, perhaps using python for the map/reduce functions. But that may not be a sensible plan :)

[09:49:00.0000] <jgraham>
Also, since the process is mainly IO bound it's not clear why using python is a bad idea

[09:50:00.0000] <jgraham>
/me was following the approach in http://effbot.org/zone/wide-finder.htm which did OK: http://www.tbray.org/ongoing/When/200x/2007/10/30/WF-Results

[09:52:00.0000] <gsnedders>
jgraham: Shouldn't you be doing it in Perl if you're basing it on WF?

[09:52:01.0000] <jgraham>
gsnedders: No. :)

[09:53:00.0000] <jgraham>
/me has to go but wants to find out why he is doing everything wrong :)

[10:45:00.0000] <gsnedders>
/me wonders whether we really need to treat U+000C as whitespace

[10:45:01.0000] <hsivonen>
http://xformstest.org/klotz/shibumiscript/2009/04/tv-raman-on-xforms-and-screen-scraping.html

[10:46:00.0000] <hsivonen>
the post that seemed to be about HTML5 and XHTML2 seems to have comments increasingly about XForms--not XHTML2

[10:59:00.0000] <Philip`>
jgraham: It shouldn't be IO bound for the current dotnetdotcom data - there's not even 4GB, so it should fit in RAM with no IO at all

[11:02:00.0000] <Philip`>
jgraham: Also, IO should be on the order of 100MB/s, which seems much faster than you can decompress and parse data

[11:29:00.0000] <Philip`>
jgraham: (By "IO" I mean raw disk IO)

[11:44:00.0000] <gsnedders>
If you have a text/html document within an object element, is it within a new browsing context?

[11:44:01.0000] <Philip`>
Why wouldn't it be?

[11:44:02.0000] <gsnedders>
/me claims ignorance

[11:47:00.0000] <gsnedders>
Does anyone serve XHTML as application/xml or text/xml?

[11:48:00.0000] <hsivonen>
gsnedders: mozillaquestquest uses text/xml, IIRC

[11:48:01.0000] <gsnedders>
/me was hoping he could just ignore such content, oh well

[11:48:02.0000] <hsivonen>
gsnedders: also, I believe there's MathML content that uses application/xml for IE XSLT hooks

[11:50:00.0000] <rubys>
what do browsers do with xhtml served as appliction/xml or text/xml?

[11:51:00.0000] <hsivonen>
rubys: dispatch on namespace, but at least old versions have issues with what interfaces the document object exposes

[11:51:01.0000] <rubys>
what does IE do?

[11:51:02.0000] <hsivonen>
rubys: also, I'd expect createElement not to be magic

[11:52:00.0000] <hsivonen>
rubys: shows a prettyprint tree or applies XSLT if available

[11:52:01.0000] <gsnedders>
rubys: It treats it as any other XML document, so you can use XSLT to transform it to HTML

[11:52:02.0000] <rubys>
Can the link to the XSLT be in an HTTP header?

[11:53:00.0000] <hsivonen>
unlikely

[11:53:01.0000] <hsivonen>
but I don't know

[11:55:00.0000] <rubys>
I will run a test to verify, but to be clear: webkit + presto + gecko all are thought to dispatch based on namespace when they encounter text/xml ?

[11:55:01.0000] <hsivonen>
rubys: yes, except the document object is created on MIME type

[11:56:00.0000] <hsivonen>
which is a bug per HTML5

[11:56:01.0000] <hsivonen>
I don't know what the implementation reality on that point is today

[11:57:00.0000] <rubys>
I wasn't thinking of starting with HTML, but testing it out on separate SVGs.

[11:57:01.0000] <annevk5>
differs

[11:57:02.0000] <annevk5>
DOM behavior for e.g. createElement differs too

[11:57:03.0000] <hsivonen>
rubys: If you run tests, I suggest testing what interfaces you get on document and testing what document.createElement("foo") returns

[11:58:00.0000] <rubys>
the first thing I would likely try is to see how jquery reacts. :-)

[11:59:00.0000] <rubys>
but I like the idea of svg images being like feeds were a few years ago: displayed as raw xml to backlevel browsers and as something usable by modern browsers.

[12:00:00.0000] <hsivonen>
rubys: you find an XML tree better user experience than a download dialog?

[12:03:00.0000] <Philip`>
hsivonen: You could write a polyglot document that displays the same content via XHTML and via ASCII art in the XML tree

[12:05:00.0000] <hsivonen>
Philip`: should there be validation for that? :-)

[12:21:00.0000] <annevk42>
hmm, great, any form of pasting makes my chromium build crash

[12:21:01.0000] <gsnedders>
/me ought to do physics, but reading Anna Karenina is awfully tempting

[12:22:00.0000] <Philip`>
/me wonders how Larry sees hundreds of messages when loading whatwg.org/issues (presumably without manually clicking on all the links)

[12:22:01.0000] <Philip`>
since I just get the 15 folders, in Chrome 1.0 and Chrome 2.0 and Firefox 3.0 and Opera 9.6

[12:22:02.0000] <annevk42>
I was trying to find out, but Chromium crashes on me

[12:22:03.0000] <hsivonen>
I replied to Larry

[12:23:00.0000] <annevk42>
(when I enter the address manually I get the same as Philip`)

[12:23:01.0000] <Philip`>
hsivonen: I noticed, just before I sent the reply I'd written myself :-)

[12:23:02.0000] <Philip`>
(saying basically the same thing)

[12:23:03.0000] <Philip`>
annevk42: Fortunately Chromium has lots of clever process isolation so that the browser won't crash

[12:24:00.0000] <annevk42>
you know, not working

[12:24:01.0000] <annevk42>
:)

[12:24:02.0000] <hsivonen>
has Chrome 2.0 already been pushed to users?

[12:24:03.0000] <Philip`>
hsivonen: No

[12:24:04.0000] <hsivonen>
Philip`: do you have separate VMs for different versions of Chrome?

[12:24:05.0000] <Philip`>
hsivonen: (I had to manually find and install the beta)

[12:25:00.0000] <Philip`>
hsivonen: No; I tested in Chrome 1.0, then upgraded to 2.0 and tested again

[12:25:01.0000] <hsivonen>
ah

[12:25:02.0000] <Philip`>
/me has no VMs at all

[12:26:00.0000] <Philip`>
mostly because VMware on Linux started panicking my kernel after I updated something

[12:26:01.0000] <Philip`>
and then I decided there were more useful uses of several gigabytes of disk space than a VM which I never use

[12:48:00.0000] <hsivonen>
Hixie: what's your take on http://mxr.mozilla.org/mozilla-central/source/layout/reftests/bugs/228856-2.html ?

[12:53:00.0000] <jgraham>
Philip`: My plan was to decompress the data upfront rather than on the fly. Possibly that isn't sensible (but I'm no sure how the compressed size of the data relative to the ram size matters since presumably it has o be decompressed at some point)

[12:55:00.0000] <Philip`>
jgraham: I think that isn't sensible, because of disk IO speed and RAM buffer sizes

[12:55:01.0000] <Philip`>
The decompressed data will get processed and discarded and garbage-collected very quickly

[12:55:02.0000] <Philip`>
so it won't take up much RAM in total

[12:56:00.0000] <Philip`>
which leaves all the rest available for buffering the disk

[12:56:01.0000] <jgraham>
Philip`: I'm not sure I understand. Maybe I don't understand gzip well enough

[12:57:00.0000] <Philip`>
jgraham: You decompress and process it streamingly, you don't decompress the entire thing into RAM and then process it all at once

[12:58:00.0000] <jgraham>
Philip`: But that limits you to a single process operating on the data, right?

[12:58:01.0000] <jgraham>
(i.e. you can't look at it in chunks)

[12:58:02.0000] <Philip`>
jgraham: If you've got a single gzipped file, then yes

[12:59:00.0000] <Philip`>
so I suggest doing some pre-processing to split it into several chunks, each of which can be gzipped independently

[12:59:01.0000] <Philip`>
(and also change it to a format that can be read more easily than null-terminated fields)

[13:00:00.0000] <Philip`>
and then you can parallelise by having each process process one of the chunks

[13:00:01.0000] <Philip`>
(so make them pretty small chunks, like a hundred megabytes or whatever)

[13:00:02.0000] <jgraham>
Philip`: All of that sounds like a sensible optimisation but essentially not that different to the (broken) code I already have

[13:01:00.0000] <Philip`>
Much more sensible than storing the uncompressed data on disk :-)

[13:01:01.0000] <jgraham>
Which processed the whole file in ~5 minutes. Although that might change a lot when it actually does the right thing

[13:01:02.0000] <jgraham>
Philip`: It's not really that big a file compared to the amount of free disk space I have :)

[13:02:00.0000] <Philip`>
jgraham: The point is that it's big compared to the amount of RAM you have

[13:02:01.0000] <Philip`>
and so it'll be slower to process

[13:03:00.0000] <Philip`>
I suppose unless you've got 4GB of RAM (or you restrict yourself to a subset of the data) you'll have to do some disk IO anyway, and if the OS does FIFO buffering then it'll be the worst possible case, which isn't so good

[13:04:00.0000] <gsnedders>
/me always gets FIFO and FILO mixed up

[13:04:01.0000] <jgraham>
FILO is a type of pastry

[13:04:02.0000] <jgraham>
/me is sure he has said that before

[13:05:00.0000] <Philip`>
gsnedders: It's easy to remember, because people never use the term FILO :-)

[13:05:01.0000] <Philip`>
(They're called stacks instead)

[13:05:02.0000] <gsnedders>
Philip`: Assuming the SQA has logic

[13:05:03.0000] <gsnedders>
(Unlikely)

[13:06:00.0000] <rubys>
LIFO is common, however

[13:06:01.0000] <Philip`>
jgraham: Is that ~5 minutes with some parallelisation?

[13:07:00.0000] <jgraham>
/me can find no evidence in the logs that he has actually made the pastry bun before so he is safe in the knowledge that people will only think him unfunny rather than unfunny and repetitive

[13:07:01.0000] <jgraham>
Philip`: Yes. 4 processes processing 50Mb chunks

[13:07:02.0000] <Philip`>
Pastry bun? :-p

[13:08:00.0000] <jgraham>
Argh

[13:09:00.0000] <Philip`>
jgraham: Hmm, seems to take me ~5 minutes to grep for a string in the 425K pages with no parallelisation

[13:09:01.0000] <Philip`>
(and it's all CPU bound, so it should parallelise embarrassingly)

[13:10:00.0000] <Philip`>
(Actually I suppose it's memory bound, so that's not really true; oh well)

[13:27:00.0000] <Hixie>
hsivonen: is the spec ambiguous?

[13:31:00.0000] <Philip`>
hsivonen: Did you see http://krijnhoetmer.nl/irc-logs/whatwg/20090424#l-503 ?

[13:31:01.0000] <Philip`>
/me wonders if he should report it via the bug tracker or something

[13:34:00.0000] <rubys>
Hixie: what's the status/outlook on RDFa?

[13:34:01.0000] <Philip`>
jgraham: (For comparison: It takes ~70 seconds to read all the pages and extract the HTTP headers, with 4 processes in parallel; and ~400 seconds to parse all the HTML (via SAX2) and extract all the <meta> elements)

[13:35:00.0000] <Hixie>
rubys: just had a meeting with the last group of people i know of who had input they wanted to give on the issue, having lunch now, plan to start going down the use cases this afternoon

[13:36:00.0000] <rubys>
so you will have a proposal by the end of the... week?  month?  Don't need a firm date, just a guess.

[13:37:00.0000] <Hixie>
there are a number of use cases each of which will likely get their own proposals

[13:37:01.0000] <Hixie>
i expect the easiest ones will have proposals by tonight

[13:38:00.0000] <Hixie>
and the more complex ones likely by the end of the week

[13:38:01.0000] <rubys>
thanks!

[13:39:00.0000] <Hixie>
np

[13:45:00.0000] <annevk42>
"Estimated date for last e-mail: 2009-11-01" fun

[13:47:00.0000] <Hixie>
it was april 2010 last week :-)

[13:47:01.0000] <annevk42>
it's making a nose dive atm

[13:48:00.0000] <jgraham>
Philip`: OK, now I have some idea what to aim for :)

[13:49:00.0000] <Hixie>
annevk42: it's gonna flatline this week as i go back to microdata stuff :-)

[13:50:00.0000] <Philip`>
jgraham: (This is on a 2.4GHz quad Core 2)

[14:05:00.0000] <gsnedders>
Where is the estimate?

[14:06:00.0000] <Philip`>
http://tomayko.com/ - <meta http-equiv='X-UA-Compatible' content='YOUR MOM'>

[14:07:00.0000] <annevk42>
hehe

[14:13:00.0000] <annevk42>
W3C QA blog needs karl back; this latest post doesn't even provide some simple pointers

[14:15:00.0000] <annevk42>
Hixie, if you could reply to my email about sync fetch that'd great btw

[14:16:00.0000] <Philip`>
http://philip.html5.org/data/ua-compatible-contents.txt

[14:17:00.0000] <annevk42>
Hixie, especially if you have any tips on how to share an algorithm between a sync and async fetch at the same time (where the algorithm would only dispatch events in the async case, but would share a bunch of common that do not affect the event loop (apart from completely killing the network request; i.e. the current task))

[14:17:01.0000] <Hixie>
annevk42: do you want other events, timeouts, etc, to fire during the sync request?

[14:18:00.0000] <annevk42>
nope

[14:18:01.0000] <Hixie>
k

[14:18:02.0000] <annevk42>
it would be sync very much like alert() is sync I guess

[14:18:03.0000] <annevk42>
but I still need it to follow redirects, throw on network errors, etc. (and dispatch events if invoked as async)

[14:19:00.0000] <Hixie>
i don't think i'd recommend reusing the 'fetch' algorithm then

[14:19:01.0000] <Hixie>
i'd just write a new one

[14:19:02.0000] <Hixie>
that happens to have bits in common

[14:20:00.0000] <Hixie>
not really sure what to suggest, tip wise

[14:20:01.0000] <Hixie>
how can i help you?

[14:22:00.0000] <annevk42>
from http://dev.w3.org/2006/webapi/XMLHttpRequest/#send step 9 and 10 are basically the part that the UA needs to do while fetching regardless of sync/async

[14:22:01.0000] <annevk42>
now I was thinking that if fetch qeued some kind of task I could just wait for that task to complete in the sync case

[14:22:02.0000] <annevk42>
and for async return early

[14:23:00.0000] <Hixie>
the task will not complete if you're waiting

[14:23:01.0000] <Hixie>
the tasks only ever run when the event loop runs

[14:23:02.0000] <Hixie>
this is really just a regular event loop pattern

[14:24:00.0000] <annevk42>
I see

[14:24:01.0000] <annevk42>
writing out everything two times seems like the wrong approach though

[14:26:00.0000] <annevk42>
I suppose writing a sync fetch isn't too hard; hmm

[14:28:00.0000] <annevk42>
does XHR delay the load event?

[14:31:00.0000] <annevk42>
so you'd have ... fetch resource; return send(); follow loading algorithm; ... or ... sync fetch resource; follow loading algorithm; return send()

[14:33:00.0000] <annevk42>
this still requires some kind of event loop for the sync case in the UA of course, but I guess that's always required, because otherwise you'd get stuck

[14:33:01.0000] <annevk42>
but it doesn't have to be explicit in the spec

[14:42:00.0000] <Philip`>
http://philip.html5.org/data/ua-compatible-headers.txt

[14:43:00.0000] <Philip`>
I like the first one

[14:44:00.0000] <gsnedders>
IE=8;FF=3;OtherUA=4: X-UA-Compatible: IE=8;FF=3;OtherUA=4

[14:44:01.0000] <gsnedders>
WTF?

[14:45:00.0000] <annevk42>
Hixie, did the above make some amount of sense?

[14:47:00.0000] <Philip`>
gsnedders: *shrug*

[14:47:01.0000] <Hixie>
annevk42: sorry, was afk. reading.

[14:48:00.0000] <Hixie>
annevk42: you don't need an event loop in the sync case if you just avoid queueing tasks in that case

[14:48:01.0000] <Hixie>
annevk42: i would just have on algorithm, and every now and then say "if sync... else if async ..."

[14:49:00.0000] <annevk42>
Hixie, yeah, that is "follow loading algorithm" :)

[14:49:01.0000] <annevk42>
I suppose I could do it for fetching as well

[14:51:00.0000] <annevk42>
defining the Web sometimes feels like building a house of cards

[14:53:00.0000] <Philip`>
/me wobbles the table

[15:27:00.0000] <Hixie>
oops

[15:27:01.0000] <Hixie>
i just XSSed myself

[15:27:02.0000] <Hixie>
<iframe src="data:text/html,<script>document.write(location)</script>"></iframe> doesn't do quite what you might think it does

[15:47:00.0000] <gsnedders>
/me wonders how it does that

[16:39:00.0000] <annevk5>
Hixie, the attribute having the wrong name in the spec does seem like a problem

[16:40:00.0000] <Hixie>
yeah your e-mail is in the bucket

[16:40:01.0000] <Hixie>
i'm assuming there's something more wrong with it


2009-05-05
[17:29:00.0000] <hsivonen>
Philip`: no, hadn't seen it. filed http://bugzilla.validator.nu/show_bug.cgi?id=503

[17:30:00.0000] <hsivonen>
Hixie: Re: \0 test, probably specs aren't ambiguous. the selector behavior vs. DOM Inspector is just weird

[17:31:00.0000] <Hixie>
ah

[17:32:00.0000] <Hixie>
certain aspects of that test looked wrong given html5's parsing rules fwiw

[17:34:00.0000] <Hixie>
" student is making a timeline of important events in Apple's history. As he reads Wikipedia entries on the topic, he clicks on dates and selects "add to timeline", which causes an entry to be added to his timeline"

[17:35:00.0000] <Hixie>
assuming that the timeline is implemented by a third-party app or web site

[17:35:01.0000] <Hixie>
and assuming that the browser doesn't have built-in knowledge of this app or site

[17:35:02.0000] <Hixie>
and assuming wikipedia doesn't either

[17:35:03.0000] <hsivonen>
Hixie: which parts of the test case looked wrong?

[17:35:04.0000] <hsivonen>
the test case is failing with the HTML5 parser

[17:36:00.0000] <Hixie>
hsivonen: &#0; and ^@ in HTML5 become U+FFFD, not U+0000

[17:36:01.0000] <Hixie>
i guess if wikipedia can output an .ics file somehow, the third party site or app could register as a handler for text/calendar files...

[17:36:02.0000] <hsivonen>
Hixie: isn't that the point of the test?

[17:37:00.0000] <hsivonen>
Hixie: that it's a bug if any of the selectors matches

[17:37:01.0000] <Hixie>
hsivonen: oh i didn't look to see if the test expected the selectors to match, my bad

[17:37:02.0000] <Hixie>
which parts of the test case are failing?

[17:39:00.0000] <hsivonen>
Hixie: http://hsivonen.iki.fi/screen/228856-2.png

[17:39:01.0000] <Hixie>
test case url?

[17:40:00.0000] <hsivonen>
http://mxr.mozilla.org/mozilla-central/source/layout/reftests/bugs/228856-2.html

[17:43:00.0000] <Hixie>
is css also converting things to FFFD?

[17:43:01.0000] <Hixie>
but only at the start of a selector?

[17:43:02.0000] <hsivonen>
Hixie: I have no idea

[17:44:00.0000] <Hixie>
hah, there's a regexp search replace error in the text part of the test (doesn't affect results)

[17:44:01.0000] <Hixie>
&amp;. should be &amp;#

[17:46:00.0000] <Hixie>
hsivonen: i bet that the css parser is turning ^@ into U+FFFD

[17:46:01.0000] <Hixie>
hsivonen: (but not \0)

[23:30:00.0000] <zcorpan__>
annevk42: hey, the php stopped working on simon.html5.org - i guess it's the same issue as you had? how did you fix it?

[23:37:00.0000] <zcorpan>
apparently i had to change AddHandler application/x-httpd-php .php to AddHandler php-cgi .php

[23:40:00.0000] <zcorpan>
although now indexing is broken instead

[23:41:00.0000] <zcorpan>
oh well

[23:42:00.0000] <Hixie>
christ microformats are complicated if you start looking at all value-excerption, value-class, etc

[23:42:01.0000] <Hixie>
/me doesn't understand at all how you parse an hcard

[23:44:00.0000] <takkaria>
with logic

[23:44:01.0000] <takkaria>
and a good grasp of the essential unsatisfactoryness of existence

[23:51:00.0000] <zcorpan>
when i wrote http://simon.html5.org/sandbox/html/w3c-home-in-html5 back in 2007, it only rendered correctly in opera. now it renders correctly in safari and firefox, too

[23:54:00.0000] <zcorpan>
i wonder when it will render correctly in ie

[00:42:00.0000] <Hixie>
man vcard is much more complex than it looks

[00:56:00.0000] <Hixie>
/me sees more e-mails with people suggesting i'm a microformats shill

[00:57:00.0000] <Hixie>
is it a good sign that the microformats people think i'm an rdfa shill while the rdfa people think i'm a microformats shill?

[01:28:00.0000] <takkaria>
Hixie: I'd guess so

[01:33:00.0000] <Philip`>
Hixie: I don't think it's good for everybody to think you're a shill

[01:34:00.0000] <Hixie>
not everybody thinks i'm a shill, only the people who have vested interests in the current topic of discussion

[01:35:00.0000] <Philip`>
Ah, so everybody who's not apathetic thinks you're a shill? :-)

[01:38:00.0000] <Hixie>
no, there are people who are interested in a solution to microdata annotations who don't yet have a horse in the race

[01:38:01.0000] <Hixie>
/me keeps finding bugs in the hcard spec

[01:42:00.0000] <Hixie>
oh wait, this is actually not a bug in hcard, it's crazyness in vcard

[01:42:01.0000] <Hixie>
nevermind!

[03:04:00.0000] <Hixie>
vcard is awesome

[03:04:01.0000] <Hixie>
some of its fields have two states, URI or inline binary

[03:04:02.0000] <Hixie>
but all of its fields have both a type and an encoding

[03:05:00.0000] <Hixie>
URI is uri type, text encoding

[03:05:01.0000] <Hixie>
inline binary in binary type, b encoding

[03:06:00.0000] <Hixie>
the default? binary type, text encoding

[03:06:01.0000] <Hixie>
so if you specify a uri, you must set the type to uri

[03:06:02.0000] <Hixie>
but if you inline the binary data, you must set the encoding to b

[03:12:00.0000] <Philip`>
They should solve the problem by making everything always be a URI

[03:12:01.0000] <Philip`>
and then they can use data: URIs if they want inline binary data

[03:13:00.0000] <jgraham>
/me discovers that writing if (condition); {do something} rather than if (condition) {do something) has rather surprising consequences

[03:13:01.0000] <Philip`>
jgraham: Is it surprisingly that randomly inserting incorrect characters makes code not work? You'd get the same problem if you wrote if (!condition) {do something} :-p

[03:14:00.0000] <Philip`>
*surprising

[03:14:01.0000] <jgraham>
Philip`: It is somewhat surprising if it is a character that is often optional at the end of a line

[03:15:00.0000] <jgraham>
Or rather it is unsurprising but not as easy to spot

[03:16:00.0000] <Philip`>
Oh, I suppose it's more surprising in a crazy language like JS where correct syntax is optional

[03:21:00.0000] <Hixie>
/me is starting to get a headache from the vcard spec

[03:25:00.0000] <annevk42>
/me had that when he tried to figure out how browsing contexts worked for the first time

[03:26:00.0000] <Hixie>
heh

[03:27:00.0000] <Hixie>
the first time those were specced, they were screwed up

[03:27:01.0000] <annevk42>
:p

[05:39:00.0000] <olliej>
who the heck is this giovanni guy who has randomly appeared?

[05:39:01.0000] <annevk42>
I wonder about that too sometimes, but decided not to care too much :)

[05:40:00.0000] <olliej>
annevk42: hee

[05:41:00.0000] <annevk42>
yo!

[05:41:01.0000] <annevk42>
Safari for Ubuntu coming soon? :)

[05:41:02.0000] <Philip`>
He appeared about five months ago

[05:45:00.0000] <jgraham>
annevk42: With Chromium for Ubuntu, why would anyone still care about Safari ;)

[05:51:00.0000] <annevk42>
jgraham, presumably Safari for Ubuntu would have the Mac font settings and such

[05:52:00.0000] <annevk42>
and my current Chromium crashes on paste

[05:52:01.0000] <annevk42>
not fun

[05:52:02.0000] <annevk42>
/me uses Midori instead

[05:52:03.0000] <Philip`>
Just run Windows Safari in Wine

[05:54:00.0000] <olliej>
jgraham: because without safari how would chrome actually render anything :p

[05:55:00.0000] <Philip`>
It could still render the pipes screensaver

[05:56:00.0000] <jgraham>
olliej: Aren't you obliged to continually point out the different between Safari and WebKit?

[05:56:01.0000] <olliej>
jgraham: i gave that up when google started acting like they wrote their own browser 

[05:56:02.0000] <olliej>
:D

[05:57:00.0000] <jgraham>
/me likes the Chromium start up tab "This browser is not ready yet! This is a pre-alpha build of Chromium on Linux. It is woefully incomplete. It's ‘Chromium’, not ‘Google Chrome’"

[05:57:01.0000] <jgraham>
And then later

[05:57:02.0000] <jgraham>
Please don't file bugs, Blogging about it is not helpful

[06:14:00.0000] <Philip`>
http://en.wikipedia.org/wiki/File:BGP_FSM.svg - "W3C ✓ The source code of this SVG is valid" - "Errors found while checking this document as SVG 1.1 + IRI! Result: 16 Errors, 4 warning(s)"

[06:14:01.0000] <Philip`>
Validation badges are a bit useless when people use them for invalid files

[06:17:00.0000] <jgraham>
s/for invalid files//

[06:18:00.0000] <Philip`>
Hmm, I meant something more like "Validation badges are a bit useless given that people use them for invalid files"

[06:18:01.0000] <Philip`>
If they were a guarantee of validity, they'd be reasonably useful, because they'd save you the effort of running the validator when you want to see that a page is valid

[06:20:00.0000] <jgraham>
Philip`: Since you can't rely on people to do that, I maintain that my restatement is equivalent to the original

[06:22:00.0000] <Philip`>
jgraham: I assumed you were reading my statement as more like "Validation badges a bit useless in the specific cases in which people use them for invalid files", and your restatement was not equivalent to, and more correct than, the original

[06:24:00.0000] <Philip`>
(so I was clarifying that my statement was indeed intended to be equivalent to your restatement)

[06:24:01.0000] <Philip`>
(so, um, I think we agree?)

[06:28:00.0000] <zcorpan>
Hixie: should events that don't bubble still be dispatched on window when fired on document?

[06:37:00.0000] <zcorpan>
/me wonders whether the query string has been considered for the svg <param> discussion

[06:53:00.0000] <heycam>
zcorpan, it does (for the exposing of parameters to the parameter referencing mechanism)

[06:53:01.0000] <heycam>
but i see what you mean: maybe we don't need a <param> element; we can just use the query string when referencing documents

[06:54:00.0000] <heycam>
otoh, perhaps other referenceable things (applets?) would only work with <param> elements

[06:54:01.0000] <heycam>
on my third hand, maybe they wouldn't work without modification with <svg:param>s anyway

[07:27:00.0000] <zcorpan>
heycam: you have three hands?

[07:37:00.0000] <Philip`>
I think the hands may be largely metaphorical

[07:41:00.0000] <Philip`>
/me reads a book on network routing design, and finds a bit saying "This design looks good on paper until you try to implement it. It's very difficult to justify all the AS numbers from American Registry for Internet Numbers (ARIN) or other Regional Internet Registries (RIRs) required for such a network. AS numbers are a finite resource that is quickly being depleted."

[07:41:01.0000] <Philip`>
It's both funny and sad that arbitrary integers are a scarce resource

[07:41:02.0000] <jgraham>
/me decides to pick up a few spare when he is next out

[07:42:00.0000] <Philip`>
(AS numbers are kind of like IPv4 addresses, except there's 2^16 times fewer of them)

[07:44:00.0000] <Philip`>
(and just like NAT is a 'solution' to IPv4 scarcity, the book suggests using private AS numbers and then doing some ugly tricks to hide them and look like a single globally-unique AS number to the rest of the world)

[07:46:00.0000] <Philip`>
(and just like IPv6 is a better solution, there's a new extension for 4-byte AS numbers (which is more than anybody will ever need), with approximately zero adoption so far)

[07:48:00.0000] <Philip`>
(I suppose the lesson is that when you design a protocol with a globally-unique bounded number, decide how many bits you'll need if the protocol is wildly successfully and used by the entirety of human civilisation throughout our galaxy, and then double it)

[07:48:01.0000] <jgraham>
The other solution might be to not design protocols that rely on globally unique bounded numbers (but that may be impractical)

[07:50:00.0000] <Philip`>
Good luck convincing Cisco to route packets based on unbounded integers

[07:51:00.0000] <Philip`>
You could build a secretive file-sharing protocol which transfers megabytes of data in the source address of each packet

[07:55:00.0000] <Philip`>
(...And it's kind of hard to avoid the "globally unique" part, since people really like having identifiers with a unique meaning)

[10:16:00.0000] <tantek>
Hixie, feel free to discuss/mention any bugs you find in the hCard spec in #microformats

[10:17:00.0000] <tantek>
see of course http://microformats.org/wiki/hcard-faq and http://microformats.org/wiki/hcard-issues

[10:17:01.0000] <tantek>
(some are known/resolved already, feel free to add "+1s" to any issues you also encounter/agree with)

[10:34:00.0000] <Philip`>
http://www.cssquirrel.com/comic/?comic=16

[10:49:00.0000] <Philip`>
Hixie: Tantek probably shouldn't be underlined in http://www.whatwg.org/specs/web-apps/current-work/multipage/acknowledgements.html

[10:53:00.0000] <jgraham>
Philip`: That's the secret symbol to indicate that tantek is the evil puppetmaster who exerts sole control of the HTML5 spec

[10:54:00.0000] <jgraham>
Although now I have said that, I will have to kill evenyone on the channel. Except tantek of course

[10:56:00.0000] <Philip`>
/me watches tantek perform the Dance of Despair and Disillusionment

[10:58:00.0000] <gsnedders>
Philip`: it's because it's he's a span element, and Hixie is assuming all span elements are incomplete xref

[11:16:00.0000] <takkaria>
Philip`: it does seem the RDFa people see Hixie as having strong bias toward microformats

[11:23:00.0000] <gsnedders>
/me doesn't like either RDFa or µf

[11:24:00.0000] <takkaria>
I'm also not much liking the rampant paranoia people seem to have

[11:29:00.0000] <tantek>
takkaria, at least such rampant paranoia is itself revealed/archived publicly so you can prioritize such input appropriately.

[11:29:01.0000] <gsnedders>
/me waits for that to be quoted by MLW

[11:31:00.0000] <tantek>
takkaria, and regarding biases, Hixie tends to have a bias for productive pragmatism, and microformats certainly have a stated bias for productive pragmatism, so it should not be much of a surprise that some overlap in opinions is the result.

[11:31:01.0000] <takkaria>
sure

[11:32:00.0000] <tantek>
gsnedders, do you like semantic HTML?

[11:32:01.0000] <gsnedders>
taYes

[11:32:02.0000] <gsnedders>
tantek: Yes, even.

[11:32:03.0000] <takkaria>
I have no side in this debate, and some of the accusations being bandied around are just mad

[11:34:00.0000] <tantek>
takkaria, while some of the accusations could certainly be taken as "mad", they do serve two useful functions, 1. the above-noted prioritization enabler, and 2. sometimes even "mad" accusations may actually be a sign of some other issue (other than the one made in a "mad" way) that is worthy of consideration.

[11:35:00.0000] <takkaria>
well, it's quite clear that Shelley doesn't feel valued at all

[11:35:01.0000] <takkaria>
but I think it's a bit extreme to say that Hixie gave you vetting powers over the use cases list

[11:36:00.0000] <tantek>
I believe Hixie has made this point (I don't have a citation) that sometimes even trolls (or more precisely, input of a trolling nature) can be helpful with the issues that are unintentionally point out.

[11:37:00.0000] <takkaria>
yeah, I've seen him make it

[11:37:01.0000] <tantek>
takkaria, such a statement ("gave ... vetting power over") is both extreme, and has the flaw of being a deliberate exaggeration, likely to construct the basis of a strawman argument. http://en.wikipedia.org/wiki/Straw_man (logical fallacy)

[11:38:00.0000] <tantek>
such constructs are not uncommon when arguing theoretical points in (perhaps mostly) academic forums

[11:39:00.0000] <takkaria>
indeed, I do have something of a grasp of argumentative logic and fallacies. :)

[11:39:01.0000] <tantek>
the only logical steps to take are to: 1. point out the strawman fallacy (and any other logical fallacies being made in defense of positions that are typically indefensible with evidence) and 2. (de)prioritize/filter input from such sources accordingly over time.

[11:40:00.0000] <tantek>
takkaria - no lack of grasp was implied :)

[11:42:00.0000] <Hixie>
tantek: the thing is, i find it an especially weird accusation given that most of our conversation was me whining about problems with microformats :-)

[11:42:01.0000] <tantek>
the 2. step is necessary in order to avoid or at least minimize a (perhaps unintentional, but still effective) DOPA attack from such sources. http://tr.im/dopa

[11:46:00.0000] <tantek>
Hixie, it is not that weird, in that the paranoid are quite skilled at inserting however many layers of abstraction / meaning are necessary in order to twist a quote to justifying/bolstering their claim.

[11:46:01.0000] <tantek>
Of course, if the very intent of the paranoid is to push for positions advocating for increasing layers of abstraction / meaning, then resultant accusations exhibit a level of meta-irony that itself has value along the axis of humor.

[11:50:00.0000] <Hixie>
i think it's important to distinguish between certain people who have clearly just been spurned (like whoever does lastweek, and probably shelley) and people who are taking what those people say as gospel (like a lot of other people)

[11:50:01.0000] <Hixie>
the latter group aren't "paranoid", they've just been misled

[11:50:02.0000] <Hixie>
imho

[11:54:00.0000] <tantek>
agreed, that's a useful distinction

[11:56:00.0000] <Hixie>
of course the most useful thing to work out here would be why those who got spurned feel like they got spurned

[11:57:00.0000] <tantek>
Hixie, not sure if that is necessarily "useful" per se, if by "useful" you mean productive, and if by productive, we mean the next highest quality/quantity yield next-actions you could take, or even close. econ / diminishing-returns etc.

[11:58:00.0000] <Hixie>
i think finding out why people feel spurned is important from a moral perspective and from a pragmatic "stop making people run away" perspective

[11:58:01.0000] <tantek>
sometimes you just have let different ideas compete in the market, and let some succeed, while others struggle, wither, and are abandoned. 

[11:59:00.0000] <Hixie>
yeah, but as i don't want mine to wither (since that would mean my life was mostly wasted), i'd rather figure out why it is people are abandoning it, and fix that problem

[12:01:00.0000] <Hixie>
speaking of which, i spoke to yet another group of people working on microdata stuff yesterday, who have neither a horse in the mf space not the rdfa space, and heard yet again the same thing -- "we want to have a generic syntax so that people can mark up stuff that is then exposed to an API, without hard-coding specific class values into the parser"

[12:01:01.0000] <tantek>
Hixie, lest we forget history, it was the browser makers (and you and I included, at the time) who felt "spurned" by W3C staff and advocates for XHTML2+XForms+SVG+namespaces+(insert X-* specs here) at the Web Applications and Compound Documents workshop in 2004-04 in San Jose. http://www.w3.org/2004/04/webapps-cdf-ws/minutes-20040601.html

[12:01:02.0000] <Hixie>
apparently they are considering using a subset of rdfa instead

[12:02:00.0000] <Hixie>
though their faces when they said that indicated that they felt that was more of a situation they were being forced into than a good thing

[12:02:01.0000] <tantek>
right, a priori theoretical solutions rarely take root in bottom-up market-decided dynamics

[12:02:02.0000] <Hixie>
tantek: oh i have been spurned many times in the standards world, that doesn't mean it's ok for me to spurn others :-)

[12:03:00.0000] <tantek>
rather, they tend to be adopted in top-down forced into situations

[12:03:01.0000] <tantek>
large corporations etc.

[12:04:00.0000] <tantek>
Hixie, the point is that both WHATWG and microformats efforts were created because you and I both thought we it was worth the time and effort to explore alternatives to what the W3C was saying was "the right thing".

[12:04:01.0000] <Hixie>
yes... not sure what the relevance is here

[12:04:02.0000] <Hixie>
can you elaborate?

[12:04:03.0000] <tantek>
the market-testing of ideas

[12:05:00.0000] <tantek>
here we are almost 5 years after that infamous workshop (which perhaps we should refer to with some sort of dramatic name like, the Great Web Semantics Schism)

[12:05:01.0000] <Hixie>
the market-testing of ideas is telling me that people want something with the extensibility of rdfa with a syntax nicer than microformats.

[12:05:02.0000] <tantek>
"wants" are not market-testing

[12:06:00.0000] <tantek>
"buys" (what people actually spend time developing and then publishing) are market-testing

[12:06:01.0000] <Hixie>
when people start writing code, i consider that market testing.

[12:06:02.0000] <tantek>
writing code is a start

[12:06:03.0000] <tantek>
look at the stats on the # of abandoned sourceforge projects

[12:06:04.0000] <Philip`>
"gsnedders doesn't like either RDFa or µf" - I'll go one step further, and not like data at all

[12:07:00.0000] <tantek>
"start writing code" is insufficient. actually shipping something, better. interoperating with another implementation, now we're talking.

[12:07:01.0000] <Philip`>
I think we should remove all the data from web pages, and use the saved bandwidth for adding more colours

[12:08:00.0000] <Hixie>
i've spoken to multiple groups during this research, from big companies like yahoo to small companies like manu's, from individuals doing front-line work like livebrum.co.uk to researchers like tbl, and many of these are actively investing time and money into implementing solutions that do both mf and rdfa, and with the exception of people actively working on one or the other, they are complaining about both.

[12:08:01.0000] <tantek>
/me wonders if we should plan a party for 2009-06-01 - the 5 year anniversary of the Great Web Semantics Schism.

[12:08:02.0000] <Hixie>
many of these are in fact shipping (e.g. yahoo, shipping support for both mf and rdfa, livebrum.co.uk, shipping mf support)

[12:09:00.0000] <tantek>
Right, and five years ago, there were numerous vendors shipping interoperable implementations of XForms - I believe the XForms page on the W3C site has record of this.

[12:09:01.0000] <Hixie>
sure, there's nothing wrong with xforms itself.

[12:09:02.0000] <tantek>
And yet, you thought it was worth pursuing WebForms2 and I though it was worth pursuing microformats.

[12:09:03.0000] <Hixie>
it's just not good for web browsers.

[12:10:00.0000] <tantek>
you could say the same thing about RDF

[12:10:01.0000] <Hixie>
i am saying the same thing about rdf and microformats :-)

[12:10:02.0000] <Hixie>
rdfa, rather

[12:10:03.0000] <Hixie>
the people on the front lines that i've spoken to haven't mentioned rdf

[12:11:00.0000] <Philip`>
"Schism" sounds rude

[12:11:01.0000] <Hixie>
(not counting people who are actively working on rdfa, who do mention rdf)

[12:11:02.0000] <tantek>
except that microformats evolved from established well-worn modern web design practices of semantic  HTML (later documented as POSH)

[12:11:03.0000] <Hixie>
(working on the spec, i mean)

[12:11:04.0000] <Hixie>
just because it evolved from POSH doesn't mean it's solving all the relevant problems people are having

[12:13:00.0000] <tantek>
Philip` - anything less than schism would be politically correct watering down.  Nothing short of a schism happened that day almost five years ago. 

[12:13:01.0000] <tantek>
When all the browser vendors and professional web designers/developers found themselves agreeing, and also disagreeing with (most of) the W3C staff and X-* markup language advocates - it became quite clear to everyone in the room that a schism had occured.

[12:14:00.0000] <Hixie>
the big indicator was that Microsoft and Sun were in agreement

[12:14:01.0000] <Hixie>
and that _never_ happens

[12:14:02.0000] <tantek>
and Opera, and Mozilla

[12:14:03.0000] <tantek>
and Redhat as well!

[12:17:00.0000] <tantek>
Hixie - I'll politely remind you that microformats both never deemed to solve "all" problems people are having (leaving out "relevant" as different people have different measures for that), and even made that a non-goal: http://microformats.org/wiki/microformats#microformats_are

[12:17:01.0000] <Hixie>
just because you explicitly set out to not solve all the problems people want solved doesn't mean people will like it :-)

[12:18:00.0000] <Hixie>
the point is everyone i speak to who is using microformats complains that it doesn't provide the flexibility they want and they end up implementing other things too (rdfa, or a subset thereof, typically)

[12:18:01.0000] <Hixie>
in my mind, that tells me there is a problem i need fixing in html5 with a solution that isn't just "use microformats"

[12:19:00.0000] <Hixie>
(it's also not just "use rdfa" given that those same people complain about rdfa, e.g. saying things like "oh we'll ignore the xmlns attributes and just prebind certain prefixes that we recognise")

[12:19:01.0000] <tantek>
or you could punt on the problem altogether in html5

[12:19:02.0000] <takkaria>
HIXIE IN RDFA BIAS SHOCKER

[12:20:00.0000] <Hixie>
i could, but i think it's clear that it's not a minor theoretical problem at this point

[12:20:01.0000] <tantek>
is everything that is not a minor theoretical problem then considered in the scope of something you need to fix in HTML5?

[12:21:00.0000] <Philip`>
It's also clear that it's a difficult problem to solve, and it's not a problem with a clear solution whose usage is being blocked by limitations of the HTML spec

[12:21:01.0000] <tantek>
alternative, punt it from HTML5, and let the market continue to evolve a set of different approaches

[12:21:02.0000] <tantek>
Philip - agreed.

[12:22:00.0000] <Philip`>
On the other hand, many solutions *are* blocked by limitations of the HTML spec because they insist on being valid HTML

[12:23:00.0000] <Philip`>
so perhaps it does really need changes to the spec

[12:23:01.0000] <Hixie>
tantek: things that go into text/html documents are in scope, so i think this is in scope

[12:23:02.0000] <Philip`>
(though maybe just the changes that are needed to facilitate the solution, rather than providing a whole actual solution)

[12:23:03.0000] <tantek>
Hixie - not necessarily - it could just be an extensibility mechanism that you leave "as is".

[12:24:00.0000] <Hixie>
the existing extension mechanisms aren't enough.

[12:25:00.0000] <tantek>
are you requesting proposals for incrementally adding to existing extension mechanisms to fix specific pain points found through real world experience, or ...

[12:26:00.0000] <tantek>
proposals to completely replace existing extension mechanisms with brand new mechanisms based more on attempts to solve universal / global problems?

[12:26:01.0000] <Hixie>
neither, proposals are not something i'm short on right now :-/

[12:26:02.0000] <tantek>
heh

[12:26:03.0000] <Hixie>
for every one use case, people put forward fifteen solutions

[12:26:04.0000] <tantek>
I'm familiar with that problem, but perhaps not as many as 15, more like 3-5.

[12:27:00.0000] <Hixie>
i read 15,000 lines' worth of e-mails and got about 150 lines' worth of use cases out of it

[12:27:01.0000] <Philip`>
I think you should solve the problem by saying that all HTML documents are conforming, and conformance checkers must always return success, and then that'd let people do whatever solutions they want

[12:28:00.0000] <tantek>
getting back to - gsnedders - you said you like semantic HTML right?

[12:30:00.0000] <Philip`>
Alternative solution: All microdata should be centrally collated and processed and presented, like in Wolfram Alpha

[12:30:01.0000] <Philip`>
It's like the Semantic Web, but without the Web

[12:30:02.0000] <Philip`>
which frees you from all the horrid problems of the web

[12:30:03.0000] <Philip`>
so it's bound to work

[12:31:00.0000] <Hixie>
hah

[12:31:01.0000] <Hixie>
one of the scenarios in the e-mail i sent yesterday was basically "i have a bunch of text and want to make wolfram alpha"

[12:32:00.0000] <Hixie>
so it's come up!

[12:33:00.0000] <jgraham>
/me doesn't particularly like semantic markup, at least for some definition of semantic

[12:34:00.0000] <Philip`>
Problem: i have a bunch of text and want to make wolfram alpha. Solution: write five million lines of Mathematica. Easy!

[12:35:00.0000] <Hixie>
it'll take more than that

[12:35:01.0000] <Hixie>
it'll also take five years of painstaking data entry

[12:35:02.0000] <jgraham>
You have to write the right 5 million lines

[12:35:03.0000] <Philip`>
Well, that too

[12:36:00.0000] <Philip`>
jgraham: All you have to do is find the right cellular automaton that will generate those lines, so it's not that much effort really

[12:38:00.0000] <Hixie>
remind me not to put Philip` it charge of solving any important problems :-P

[12:46:00.0000] <Philip`>
/me sees http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-12095 describing some extensions to the <canvas> API

[12:46:01.0000] <Philip`>
(from Chapter 6)

[12:47:00.0000] <Philip`>
(in which lineWidth is basically a gradient with multiple stops, so the width varies along the line)

[12:47:01.0000] <Philip`>
(and lineCaps are basically path objects that get drawn automatically)

[12:55:00.0000] <tantek>
gsnedders, if you like semantic HTML, I'd be interested to know which step along this path you end up not liking things: semantic HTML -> POSH -> microformats as a subset of POSH

[13:11:00.0000] <jgraham>
tantek: What do you mean by semantic HTML? 

[13:12:00.0000] <gsnedders>
tantek: µf seem to be too case-specific, and RDF is too general, basically.

[13:20:00.0000] <tantek>
jgraham - this is a decent summary: http://microformats.org/wiki/semantic-html

[13:22:00.0000] <tantek>
gsnedders the only real difference between POSH, in particular semantic class names and rel values, and microformats is that microformats specify a shared vocabulary (developed via the principles and the process) where as with POSH, there is little or no sharing of vocabulary. Is it the sharing of vocabulary you think is too case-specific?

[13:25:00.0000] <gsnedders>
tantek: No, well, sort of. I see being able to use a generic parser to be a good idea™, but also I think RDF's approach of just mapping everything as object -> property (or whatever the terminology is) is too broad, and that there need 

[13:26:00.0000] <gsnedders>
*needs to be more, um, let me say "non-orthogonal parallel mapping at the same level", which is really clumsy but is the sort of thing I mean

[13:26:01.0000] <gsnedders>
Anyhow, I need to do things more useful :)

[13:27:00.0000] <tantek>
gsnedders, there are lots of "good idea™"s, but in a resource constrained universe (say, a universe bound by the laws of physics), you end up having to prioritize some ideas over others, which leads one to realize, that that prioritization is actually what is important. And ideas that may seem "good" in the abstract (in a vacuum) actually seem not so essential when compared to other ideas that you'd rather spend your limited re

[13:28:00.0000] <tantek>
the notion of  "making it really easy to write a parser" is one such seemingly good idea in the abstract.

[13:30:00.0000] <jgraham>
tantek: That page wasn't really very helpful. It seemed to be a list of html elements

[13:31:00.0000] <jgraham>
I really meant "what characteristics do you think markup has to have to be considered semantic" and, as a followup, "why is having those characteristics a good thing"?

[13:32:00.0000] <jgraham>
http://microformats.org/wiki/posh has some more clues

[13:36:00.0000] <tantek>
jgraham, I prefer to answers first with the immediate pragmatic answers ("what do you mean by semantic HTML") and only after those are discussed/understood, to raise the conversation to an abstract level ("characterists", what is "good")

[13:37:00.0000] <tantek>
in my experience, most people simply want the immediate pragmatic answers to the questions, and pragmatic answers form a better basis for abstract discussion than pure abstract discussion.

[13:37:01.0000] <jgraham>
tantek: Hence the second question being the followup to the first

[13:37:02.0000] <tantek>
precisely, and yes, the POSH page goes into more depth

[13:40:00.0000] <tantek>
jgraham, again, using actual examples to illustrate further, take a look at slide 9 "... the Semantic Spectrum" here: http://tantek.com/presentations/2003SXSW/stylesheets.html

[13:44:00.0000] <jgraham>
So the rules seem to be a) validity, b) Replace some elements (<b>, <br) with other elements, c) use @id rather than <a name=""> d) Use some patterns (only one seems to be documented?) e) Use classnames that identify abstract concepts

[13:46:00.0000] <tantek>
jgraham - many semantic (X)HTML patterns have been documented across the web, and the POSH wiki page serves as an attempt to provide a collection and index of sorts

[13:46:01.0000] <jgraham>
oh and f) "..."

[13:48:00.0000] <jgraham>
tantek: OK, so coming back to my question, a) does not seem to be a question of semantics. It is likely ood for the producer of the page (since valid markup is often easier to debug) and consumers (since the HTML 5 parsing algorithm is not yet widely implemented)

[13:48:01.0000] <jgraham>
*good

[13:48:02.0000] <jgraham>
but not semantics

[13:49:00.0000] <tantek>
HTML4 validators typically help point out problems which often are problems with the semantics of the page

[13:49:01.0000] <tantek>
thus a first "easy" step towards semantic HTML (assuming that's your goal) is to use such validators (e.g. W3C) to find such problems in your code and fix them.

[13:50:00.0000] <tantek>
in theory validity and semantics are orthogonal

[13:50:01.0000] <tantek>
in practice validity helps increase semantics

[13:50:02.0000] <jgraham>
b) Is closer to the issue that I care about. It is very easy for me to make an argument for using certian elements "correctly" (<h1>-<h6>, <th>, <td>) because they are used consistently enough that UAs can process them in a useful way (and many do so)

[13:51:00.0000] <jgraham>
So if you use <td> when you mean <th> it actually hurts users

[13:52:00.0000] <jgraham>
But <strong> is used to mean a whole laundry-list of things and in practice there is no difference between <strong> and <b> in UAs. So what is the argument for caring?

[13:53:00.0000] <jgraham>
Similarly, what is the harm ifI make a bibliography and mark up all the titles of the works using <i>?

[13:53:01.0000] <jgraham>
It's not like any UA could do anything useful with the information if I marked it up in a differnt way

[13:54:00.0000] <jgraham>
Yet people spend hours debating the "right" markup to use in every tny situation

[13:57:00.0000] <tantek>
jgraham, there is some abuse of <strong> no doubt. my own personal anecdotal experience is that there a <strong> *is* a bit better used (less abuse) in practice than <b>. using more semantically named tags also helps reinforce a more semantic mindset when authoring (rather than the "HTML as a way to print pixels" mindset).

[13:57:01.0000] <tantek>
and I have seen some reasonable arguments for repurposing of <i> as a semantic element for bibliographic (and other such) /instances/

[13:58:00.0000] <tantek>
thus yes, there are points of semantic HTML development which are debated, and there are points that are well agreed upon (often as a result of such debate occuring)

[13:58:01.0000] <tantek>
these debates started occurring mostly on blogs in the early 2000s.

[13:59:00.0000] <jgraham>
tantek: I know. I was there :)

[14:01:00.0000] <jgraham>
tantek: If you could get enough people to agree on and use say <i> for a specific purpose so that UAs could actually do something other than guess it should be emphasised in some way then that might be worthwhile (hence <h1>-<h6> being good). But I don't see that as possible

[14:01:01.0000] <tantek>
jgraham, if you have blog posts or presentations on semantic HTML from those early debates/discussion, I encourage you to add links to them to http://microformats.org/wiki/posh#POSH_Blog_posts

[14:02:00.0000] <tantek>
jgraham, I tend to agree with your assessment of <i>.

[14:02:01.0000] <jgraham>
tantek: I think I just commented on other people's blogs (and had somewhat different opinions to the ones I have now)

[14:04:00.0000] <Philip`>
Blogs existed in the early 2000s?

[14:04:01.0000] <Philip`>
Time goes so fast :-(

[14:05:00.0000] <tantek>
jgraham - glad to see you're still interested in the subject :)

[14:06:00.0000] <tantek>
I just started this page a few days ago based on some discussion in #microformats and it could definitely use some (e.g. your) help with more: http://microformats.org/wiki/posh-patterns

[14:06:01.0000] <tantek>
it would be great to have a library of posh patterns that folks could simply pick up and re-use

[14:07:00.0000] <tantek>
because such posh patterns do exist - they just were mostly debated/created before wiki-documenting/listing/indexing such things became established practice

[14:07:01.0000] <jgraham>
tantek: Well I don't hang out in #whatwg to look cool :)

[14:09:00.0000] <Philip`>
/me does

[14:09:01.0000] <Philip`>
(I don't think it works)

[14:14:00.0000] <gsnedders>
Hixie: wrt charset equivalence, do browsers really treat XML which is US-ASCII as Windows-1252?

[14:18:00.0000] <Philip`>
http://realtech.burningbird.net/semantic-web/rdf-and-rdfa/arbitrary-vocabularies-and-other-crufty-stuff - hmm, seems pretty clear to me that formats/protocols without a specific vocabulary still have some value, e.g. they can save you some effort when writing specific tools or learning a specific language

[14:20:00.0000] <Philip`>
and the issue is more about how they don't solve the entire problem, because you still need to write a specific tool or design a specific language on top of those things, and then the problem is whether that generic framework plus the extra task-specific work is more expensive than a different generic framework (e.g. one that's far more minimalist) plus the extra task-specific tools

[14:20:01.0000] <Hixie>
gsnedders: other than IE, i believe so, yes, but test if you want to find out for sure

[14:20:02.0000] <gsnedders>
Yuk.

[14:23:00.0000] <Philip`>
gsnedders: data:text/xml;charset=us-ascii,<html%20xmlns='http://www.w3.org/1999/xhtml'><script>alert('%80'.charCodeAt(0))</script></html> ?

[14:23:01.0000] <jgraham>
Philip`: One of the common themes in embedded metadata conversations is tehnically inclined people saying something equivalent to "and then we will autogenerate the UI"

[14:23:02.0000] <tantek>
Philip, the problems are not merely on the expense (cost)  side, but in experience, more so on the results (benefits) side.

[14:23:03.0000] <tantek>
jgraham - indeed.

[14:24:00.0000] <Philip`>
Opera says 8364; Firefox says "Error: no element found / Source File: jar:file:///opt/firefox/chrome/toolkit.jar!/content/global/bindings/dialog.xml / Line: 1" (?)

[14:24:01.0000] <gsnedders>
Philip`: Odd.

[14:24:02.0000] <gsnedders>
Saf says 8364

[14:25:00.0000] <Philip`>
/me doesn't know why Firefox gives that error message

[14:25:01.0000] <jgraham>
Philip`: Which tends to suggest that people underestimate the extra effort that you have to put in after you have a parser to do something usable.

[14:25:02.0000] <tantek>
one of the benefits that sought but often unsaid (unfortunately) is accurate data, and it turns out, the more human-usable the data (both authoring and viewing) the more accurate the data tends to be.

[14:25:03.0000] <tantek>
jgraham - precisely correct.

[14:26:00.0000] <jgraham>
On the other hand there is value in not having to continually reinvent the lower layers just because the high layer stuff is non-trivial

[14:27:00.0000] <tantek>
you could broaden that aforementioned "common theme" to technically inclined people tending to design for machines (e.g. parsing the content) at the expense of humans (e.g. authoring the content)

[14:27:01.0000] <Philip`>
As someone who's spent substantial amounts of time reverse-engineering weird custom binary file formats for some games, I'm certainly happy when they just use XML, because it saves a load of that low-level work

[14:28:00.0000] <Philip`>
(Of course they then compress the XML in a custom binary format, but at least that's just one format and not half a dozen for all the different types of data)

[14:28:01.0000] <tantek>
sure, having hints at the apparent boundaries and types of data helps a lot

[14:28:02.0000] <Philip`>
It's much easier to process data when it's in a tree of elements/attributes than when it's in a stream of bytes

[14:29:00.0000] <Philip`>
Presumably the RDFa idea is it's much easier to process data when it's in a graph of subject-predicate-object triples than when it's in a tree of elements/attributes

[14:30:00.0000] <tantek>
it also turns out to be often more easy/robust to author data in a tree of elements/attributes that's displayed in some readily viewable tree-like form rather than a stream of bytes.

[14:30:01.0000] <tantek>
so the move from bytes -> tree of data made sense on both the cost and benefit side of the equation

[14:32:00.0000] <Philip`>
You don't author data in a tree, you author it in a stream of bytes that represents a serialised tree

[14:32:01.0000] <tantek>
but at some point, the requirements to add structure actually cost more than humans are willing to do to author it.

[14:32:02.0000] <Philip`>
(Well, I suppose you could have a fancy XML editor that lets you edit in a tree-like UI)

[14:32:03.0000] <Philip`>
(and I suppose some people must use such tools)

[14:33:00.0000] <tantek>
people author outlines (trees) all the time. varying levels of headings and paragraphs etc. are another form of display trees

[14:33:01.0000] <tantek>
so yes, people *do* author in visual tree-like structures beyond just simple flat streams

[14:33:02.0000] <tantek>
where-ever such tree-like structures have already long evolved in human communications

[14:34:00.0000] <tantek>
thus it's easy for people to do, because it's part of human culture to do so

[14:34:01.0000] <tantek>
however, as soon as you talk about attempting to author things as a "graph" (and you're not talking about charts), all of that breaks down. authoring cost goes straight out the window, resulting in much less content. simple economics.

[14:35:00.0000] <Philip`>
Anyway, I guess I was trying to say that generic frameworks like XML and RDF aren't useless, so it's a more complex issue of tradeoffs

[14:36:00.0000] <Philip`>
(and other non-vocabulary-specific generic frameworks might have better tradeoffs for various use cases)

[14:36:01.0000] <Philip`>
and I'll let someone else work out the actual details :-)

[14:38:00.0000] <tantek>
Philip - it's precisely that complexity of the tradeoffs that is all too often ignored, and specifically, the harder to quantify human aspects (ease of authoring, mapping to existing understood and practiced modes of human communication etc.)

[14:38:01.0000] <tantek>
and so you get people debating about generic parsers instead, while the human related problems actually have a much bigger impact on both the overall cost and the benefits from any such system

[14:40:00.0000] <tantek>
it's precisely because of this (technically minded folks ignoring human aspects because such aspects are often much harder to understand and design for), that "What are microformats?" starts with "Designed for humans first and machines second" - http://microformats.org/

[14:40:01.0000] <jgraham>
I can certianly agree that one of the big problems wih RDF(a) is the ifficulty of thinking in the right mode to author it. Proponents seem to say things like "it's just four attributes" whilst failing to understand that the underlying data model is rather abstract and complex

[14:41:00.0000] <Philip`>
Yeah, it's a danger if you just consider the generic framework in isolation, rather than as part of a solution that includes some vocabulary and some vocabulary-specific processing in order to solve particular use cases, because you'll be optimising half of the costs and ignoring the rest

[14:42:00.0000] <Philip`>
jgraham: The underlying data model is trivial - it's just a big bag of subject-predicate-object triples, each of which is a URI

[14:43:00.0000] <Philip`>
The problem seems to be that the mapping between the underlying data model and the serialised byte stream is incredibly complex

[14:43:01.0000] <jgraham>
For some definition of trivial which means "comprehensible if you are the sort of person who could complete a degree in mathematics, linguistics, or a similar subject)

[14:44:00.0000] <Philip`>
s/incredibly/very/ perhaps

[14:44:01.0000] <tantek>
jgraham - agreed.

[14:44:02.0000] <Philip`>
and also the mapping between the underlying data model and the specific problem domain is often complex

[14:46:00.0000] <tantek>
Philip not just mapping to/from the serialized byte stream, but mapping to/from structures long established/evolved in human cultures/communications, and indeed, specific problem domains, and thus specific use cases.

[14:47:00.0000] <Philip`>
jgraham: I remember being taught about nouns and verbs when I was in primary school and it wasn't really that complex, and these triples are basically the same thing :-p (except with URIs, which I wasn't taught about in primary school)

[14:47:01.0000] <gsnedders>
/me wonders why Philip` wasn't taught URIs in primary school

[14:50:00.0000] <Philip`>
gsnedders: Because I'm not a young whippersnapper like you and we didn't have internets in school back then

[14:52:00.0000] <Philip`>
The highlight of our computing classes was modifying the Minesweeper high scores in Windows 3.1

[14:53:00.0000] <Philip`>
/me also cunningly inserted a QuickBASIC program into autoexec.bat so it would play the Teletubbies tune after booting

[14:53:01.0000] <gsnedders>
Man, you guys had Teletubbies!?

[14:54:00.0000] <gsnedders>
It only started in 1997 according to Wikipedia!

[14:54:01.0000] <Philip`>
Okay, maybe it wasn't actually in primary school, it was in the one that comes after primary school and before secondary school

[14:55:00.0000] <jgraham>
Philip`: I'm not that sure primary school level linguisics helps with RDF much

[15:23:00.0000] <franksalim>
Has this happened yet? <sayrer> maybe we should ask for a new WG [http://krijnhoetmer.nl/irc-logs/whatwg/20090414#l-64]

[15:25:00.0000] <Hixie>
not to my knowledge

[15:25:01.0000] <Hixie>
other than the hybi list

[16:18:00.0000] <karlcow>
[16:37]  <tantek> so yes, people *do* author in visual tree-like structures beyond just simple flat streams

[16:18:01.0000] <karlcow>
People author in streams of thoughts (semantics), almost nobody uses the outline mode of MS Office Word. They do not author trees.  


2009-05-06
[18:40:00.0000] <Lachy>
Hixie, Requiem 1.8.10 is out now

[18:43:00.0000] <Hixie>
d'you have a .torrent for it yet?

[18:44:00.0000] <Lachy>
there's on on demonoid

[18:45:00.0000] <Hixie>
k thx

[18:47:00.0000] <Hixie>
seems to be down

[18:48:00.0000] <Lachy>
http://www.demonoid.com/files/details/1908848/10713984/ works for me

[18:49:00.0000] <Hixie>
huh, works from norway

[18:49:01.0000] <Hixie>
and from LA

[18:49:02.0000] <Hixie>
just not from this network

[18:49:03.0000] <Hixie>
weird

[19:46:00.0000] <Hixie>
any bibtex people here?

[20:11:00.0000] <hsivonen>
Hixie: anything specific?

[20:23:00.0000] <Hixie>
yeah

[20:23:01.0000] <Hixie>
"wtf is the format"

[20:23:02.0000] <Hixie>
specifically, how do i quote strings in it?

[20:27:00.0000] <Dashiva>
{String} ?

[20:27:01.0000] <Hixie>
what if it contains a { or a }?

[20:27:02.0000] <Hixie>
(if this is defined anywhere i'd be happy to just read the docs)

[20:27:03.0000] <Dashiva>
I use {} or "" depending on what I need, that's all I know

[20:28:00.0000] <Hixie>
i'm trying to write a bibtex generator

[20:32:00.0000] <Hixie>
but can't find any docs on what the syntax is

[20:32:01.0000] <Hixie>
what are the allowed characters for the key?

[20:32:02.0000] <hsivonen>
Hixie: http://texlipse.cvs.sourceforge.net/viewvc/texlipse/net.sourceforge.texlipse/source/bibtex6.sablecc?revision=1.6&view=markup

[20:33:00.0000] <hsivonen>
Hixie: non-ASCII has hairy LaTeX escaping

[20:34:00.0000] <hsivonen>
can't remember how to escape }

[20:37:00.0000] <hsivonen>
hmm. actually, you can use non-ASCII without escaping

[20:45:00.0000] <Dashiva>
Isn't most things tex just "the implementation is the specification"?

[20:46:00.0000] <Hixie>
looks like i use quotes "like this" and if i want to embed a quote i do it "like {"} this"

[20:46:01.0000] <Hixie>
but i can't see how to embed a lone brace

[20:46:02.0000] <Hixie>
i've seen at least one book whose title was something like "The lone }"

[20:51:00.0000] <Hixie>
maybe {\{} ?

[01:01:00.0000] <jgraham>
Hixie: Without testing I would expect \{ to escape a brace in BibTeX

[01:22:00.0000] <Hixie>
without surrounding {}s?

[01:22:01.0000] <Hixie>
\" doesn't work, it has to be {"}, from what i understand

[01:23:00.0000] <Philip`>
\" is the syntax for putting umlauts on characters, I think

[01:25:00.0000] <Philip`>
/me tests things

[01:25:01.0000] <Philip`>
Looks like \{ doesn't work, but {\{} does

[01:27:00.0000] <Philip`>
Uh, I might be wrong

[01:38:00.0000] <Philip`>
Oh, okay, so you can't actually escape braces

[01:38:01.0000] <Philip`>
What you can do is \newcommand{\leftbrace}{\{} in the LaTeX file that includes the bibliography

[01:38:02.0000] <Philip`>
and then use \leftbrace inside the BibTeX entries

[01:38:03.0000] <Philip`>
(according to some documentation and some testing)

[01:40:00.0000] <Philip`>
(You can escape braces with \{...\} if they're matched, but you need something like \leftbrace if they're unmatched)

[01:42:00.0000] <Hixie>
o_O

[01:42:01.0000] <Hixie>
what kind of wacko language is this, anyway

[01:43:00.0000] <Philip`>
I blame Knuth

[01:46:00.0000] <Philip`>
I think the idea is it's designed to work nicely in the common case, even if that's at significant expense in certain rare cases

[01:46:01.0000] <Philip`>
(where "it" is both TeX and BibTeX)

[01:49:00.0000] <Philip`>
and I've never encountered those rare cases in BibTeX in practice, since all the strings are just title={Some Nice Simple Title} and title={Some Title with {Bits that Need Case Preserving}} and author={Jo{\~a}o} and it works easily

[01:57:00.0000] <Hixie>
Philip`: well hopefully nobody will ever need to convert bibliographic information on a work with a single { to bibtex...

[02:00:00.0000] <Philip`>
I could count the number of times I've seen a work with a single { in its title on the fingers of a hand that has no fingers

[02:00:01.0000] <Philip`>
so that's alright in practice

[02:13:00.0000] <Hixie>
i've seen one :-)

[02:13:01.0000] <Hixie>
but yes

[04:21:00.0000] <annevk42>
/me wonders what [COOKIES] will point to in HTML5

[04:46:00.0000] <hendry>
is there an HTML5 appcache example somewhere? http://twitter.com/ppk/status/1715436905

[04:55:00.0000] <Rik|work>
hendry: http://www.slideshare.net/Berttimmermans/iphone-offline-webapps or http://svay.com/blog/index/post/2009/02/19/Creer-un-client-Twitter-offline-pour-l-iPhone-avec-HTML5 (in french)

[05:31:00.0000] <hendry>
thanks Rik|work 

[05:31:01.0000] <hendry>
there is also http://google-code-updates.blogspot.com/2009/04/gmail-for-mobile-html5-series-using.html

[05:32:00.0000] <hendry>
though personally i was after some minalistic sample

[05:43:00.0000] <jgraham>
There is some documentation on the Mozilla MDC site but it's not quite running code

[07:12:00.0000] <jgraham>
/me winders if hsivonen would consider placing a link to the validator.nu html parser repository somewhere on http://about.validator.nu/htmlparser/

[08:01:00.0000] <philipj>
Hixie, implementing the media resource selection algorithm when checking <source> elements should wait until the DOM is stable (script has finished) but the first set value of the src attribute is the one to use really isn't much fun.

[08:02:00.0000] <philipj>
I'd really prefer if step 2 ("Note: By this point, the algorithm is running asynchronously.") would queue a task to run the rest, i.e. schedule it after scripts have finished

[08:04:00.0000] <annevk42>
philipj, you're prolly better of sending email

[08:38:00.0000] <Philip`>
Hmm, a surprising number of people have contacted me about my font tool, even though I intentionally didn't give a contact email address on the site

[08:46:00.0000] <Lachy>
Hixie, for input type=tel, the spec says "User agents may change the punctuation of values that the user enters." - Does that affect the submitted value, or just the way in which the number is rendered?

[08:48:00.0000] <Lachy>
in my experience, many sites don't accept any punctuation in phone numbers.

[08:48:01.0000] <annevk42>
"values" points to the value concept which means afaict that it affects the submission

[08:49:00.0000] <Lachy>
I guess, if they use type=tel, then hopefully they will fix their back end too

[08:49:01.0000] <annevk42>
it would be nicer if type=tel specified some canonicalized value for submission

[08:50:00.0000] <Lachy>
I don't think that's really possible given the wide variety of formats used for phone numbers around the world

[08:50:01.0000] <Lachy>
and even formats for phone numbers in the same country, depending on whether they're land line, mobile, free call or premium rate numbers

[08:50:02.0000] <annevk42>
international dialing is possible

[08:51:00.0000] <annevk42>
surely canonicalizing phone numbers is possible too then

[08:51:01.0000] <Philip`>
The canonical format could be created with s/[^0-9]//g

[08:51:02.0000] <Philip`>
but that doesn't help when silly Americans use letters instead of numbers in their phone numbers

[08:51:03.0000] <Philip`>
/me wonders if other nationalities do that too

[08:52:00.0000] <annevk42>
though some phone numbers are bound by country

[08:52:01.0000] <Philip`>
Also that wouldn't work when people write "+44 (0)1234 567890"

[08:53:00.0000] <Lachy>
IIRC, the ITU defines official phone number standards, and says something about the format. But that requires the country code to be known, and users don't always enter that into forms

[08:54:00.0000] <Lachy>
Philip`, is the (0) like the area code that's used when you don't explicitly dial the country code?

[08:56:00.0000] <hsivonen>
Hixie: I'm told you can backslash-escape curly braces in .bib

[08:57:00.0000] <jgraham>
Lachy: Yes

[08:58:00.0000] <Philip`>
hsivonen: Only works if they're matched pairs, according to the compiler's error output

[08:59:00.0000] <Lachy>
jgraham, so to dial my UK mobile number from a UK phone, I dial 0 7798 526 965. Or from an international phone using: +44 7798 526 965 ?

[08:59:01.0000] <jgraham>
Lachy: Yes

[09:00:00.0000] <jgraham>
Assuming that is your real mobile number :)

[09:00:01.0000] <Lachy>
it is

[09:00:02.0000] <Lachy>
it's a prepaid SIM card I use whenever I visit London

[09:00:03.0000] <Lachy>
it's cheaper than paying international rates with my Norwegian or Australian SIMs

[09:01:00.0000] <jgraham>
/me wonders when Lachy will start getting abusive calls from Mr Last Week :)

[09:01:01.0000] <Dashiva>
Dramatic readings of select IRC quotes

[09:01:02.0000] <Lachy>
my phone number has been published on my website for years. Haven't received any abusive phone calls yet

[09:01:03.0000] <Philip`>
Maybe nobody abusive reads your website

[09:02:00.0000] <Lachy>
I'm expect MLW will go looking for it now that I've mentioned it's there :-)

[09:02:01.0000] <Philip`>
Not that I'm implying the people reading IRC logs (hello out there!) are abusive

[09:02:02.0000] <Philip`>
(Not that I'm implying they're not, either)

[09:02:03.0000] <Dashiva>
You're implying you're not implying anything

[09:03:00.0000] <jgraham>
/me is confused by the implications of that

[09:03:01.0000] <Lachy>
Philip`, some people consider the abusive ones to be those of us in this channel

[09:04:00.0000] <Philip`>
/me refrains from stating an abusive comment about such people

[09:04:01.0000] <jgraham>
Anyway, the point is that it is amusing to imagine MLW in a phone box somewhere in rural England, feeding his 10ps into the machine to make crank calls to Lachy, whilst using a vocoder to disguise his voice

[09:05:00.0000] <Lachy>
he would have to know when I'm going to be in the UK for that to be effective

[09:05:01.0000] <gsnedders>
I'm sure you'll give away that information in here

[09:05:02.0000] <Lachy>
I haven't yet

[09:06:00.0000] <jgraham>
Lachy: Not now that he knows that your phone number is on your website. Plus you often mention that here, I think

[09:06:01.0000] <Lachy>
(I gave it away elsewhere already though)

[09:06:02.0000] <Philip`>
He could try phoning you every day

[09:06:03.0000] <jgraham>
He could just leave voicemail

[09:06:04.0000] <Lachy>
I don't think I set up my voice mail on that SIM yet

[09:07:00.0000] <Lachy>
I haven't set up voice mail on my norwegian number yet either, but that's cause I don't know how and can't find out cause the instructions are all in norwegian

[09:08:00.0000] <Philip`>
I suppose the problem with all these ideas is that they would merely be abuse against Lachy, and couldn't be justified as "GUERILLA JOURNALSIM" regardless of whether he deserves the abuse

[09:08:01.0000] <Lachy>
I'll be in London this weekend, arriving Saturday and leaving Sunday, on my way to Australia.

[09:08:02.0000] <Lachy>
I'm looking forward to receiving abusive calls now :-)

[09:09:00.0000] <Philip`>
(and a respectable figure like MLW wouldn't stoop to unjustified abuse)

[09:09:01.0000] <gsnedders>
/me has some Lachy mobile number in his phone

[09:09:02.0000] <Lachy>
gsnedders, you probably have my Norwegian number

[09:09:03.0000] <jgraham>
Philip`: No but they would be Crushing the Cabal! 

[09:10:00.0000] <Dashiva>
You can't crush the cabal

[09:10:01.0000] <gsnedders>
Lachy: Yeah, it's Norwegian

[09:10:02.0000] <Dashiva>
At worst you get 52 pickup

[09:10:03.0000] <gsnedders>
/me resists temptation to call Lachy 

[09:11:00.0000] <Lachy>
/me finds gsnedders' number in his address book

[09:11:01.0000] <jgraham>
gsnedders: That is probably good because the cost benefit of proving that you can use a phone vs international calling rates seems bad

[09:11:02.0000] <gsnedders>
/me has already found Lachy's

[09:12:00.0000] <gsnedders>
jgraham: Exactly.

[09:17:00.0000] <Lachy>
http://en.wikipedia.org/wiki/E.164

[09:17:01.0000] <Lachy>
that defines the format for phone numbers

[09:20:00.0000] <Lachy>
The pattern attribute probably shouldn't be allowed to apply to type=tel given that browsers are allowed to automatically adjust the punctuation

[09:20:01.0000] <jgraham>
/me is quite concerned at the idea of browsers automatically adjusting punctuation

[09:21:00.0000] <Lachy>
I'm a little concerned about it. But it's what my phone does when I enter a number into the address book

[09:21:01.0000] <Lachy>
and, for the most part, it seems to apply country specific formatting conventions fairly reliably

[09:22:00.0000] <Philip`>
Is there some list of the use cases for entering phone numbers?

[09:22:01.0000] <Dashiva>
Many registration forms, alas

[09:22:02.0000] <jgraham>
I'm not sure how browsers will be able to tell what punctuation is required. Browsers will need to be compatible but there is no spec

[09:22:03.0000] <Lachy>
Here's a use case: A site wishes to harvest phone numbers for use by telemarketers, and needs a convenient way for users to quickly and easily enter their phone number

[09:22:04.0000] <Philip`>
like, is it for e.g. address books where you want a nice human-readable (and not machine-readable) format, and it's basically plain text but you want to stop people typing in garbage?

[09:23:00.0000] <Philip`>
or is it for machine-readable numbers so a machine can automatically phone you, without any human involvement?

[09:24:00.0000] <jgraham>
Philip`: Probably both

[09:24:01.0000] <Philip`>
In the former case I guess you'd want to allow numbers like "01234 567890 (switchboard; ask for John Smith)" too

[09:24:02.0000] <Lachy>
I think the use case is because it allows for integration with the user's address book, and in mobile browsers with touch screens, focussing the control gives a keypad designed for entering numbers, rather than a normal text entry keypad

[09:24:03.0000] <jgraham>
(and quick access to tehir contacts list)

[09:24:04.0000] <Philip`>
Ah, I guess the autofill case is independent of any validation or canonicalisation

[09:25:00.0000] <Philip`>
s/autofill/fancy input method/

[09:25:01.0000] <Lachy>
I don't think we can stop people typing in garbage, because the phone number conventions use a wide variety of strange punctuation characters, inluding + ( ) # , . (space) and many others

[09:25:02.0000] <jgraham>
It is just a hint to a UA; the server has to be able to deal with the possibility of a garbale value anyway

[09:25:03.0000] <jgraham>
*garbage

[09:26:00.0000] <jgraham>
So having UAs mung the actual value seems bad

[09:27:00.0000] <Lachy>
for a normal desktop browser, yes.

[09:27:01.0000] <Lachy>
but for a user with a browser on a mobile phone, it could be convenient

[09:27:02.0000] <jgraham>
Lachy: Howso?

[09:28:00.0000] <Lachy>
so that the number gets formatted nicely for easier reading, without the user having to manually enter characters that aren't so easy to type on a mobile keypad

[09:29:00.0000] <jgraham>
Lachy: That seems to be something that is between the UA and the user if they want to have a special UI for entering numbers that pretty prints the result before updating the control

[09:30:00.0000] <jgraham>
Not really something that should be in the spec

[09:30:01.0000] <Lachy>
the spec just says it's permitted, not that it's required

[09:31:00.0000] <jgraham>
Lachy: The spec doesn't say every time that a UA is permitted to innovate the UI

[09:31:01.0000] <Philip`>
The spec normatively states that it's permitted, which seems unusual for UI issues

[09:31:02.0000] <Lachy>
yeah, true

[13:53:00.0000] <jgraham>
Hixie: Didn't zcorpan just propose that < be disallowed in _unquoted_ attributes?

[13:54:00.0000] <Hixie>
oh

[13:54:01.0000] <Hixie>
i may have misunderstood

[13:55:00.0000] <Hixie>
i guess that could work

[13:55:01.0000] <Hixie>
though if we keep adding more things that aren't allowed, it's gonna make it hard to know when to quote again

[13:55:02.0000] <Philip`>
That's easy - always quote

[13:56:00.0000] <Philip`>
You already have to pretty much do that in order to ensure compatibility with legacy clients

[13:56:01.0000] <Philip`>
(You need quotes for at least something like [\x00-\x1f"'`] I think)

[13:56:02.0000] <Philip`>
(and >)

[14:04:00.0000] <jgraham>
It saddens me to say that I increasingly find it hard to defend making unquoted attributes conforming

[14:04:01.0000] <jgraham>
Because the quotes are ugly and unnecessary

[14:04:02.0000] <jgraham>
Except when they are really important

[14:05:00.0000] <jgraham>
And the rules are rather more complex then "you just need to quote when there is whitespace or a quote character"

[14:08:00.0000] <Philip`>
I guess it's easy enough if you're writing it by hand, because you won't have funny characters; but if you're writing a serialiser, you either need to be extremely careful (else you'll have bugs like html5lib has, and those bugs could become security issues) or else always quote everything

[14:11:00.0000] <jgraham>
Philip`: Does html5lib have bugs relative to the spec or just relative to legacy clients?

[14:12:00.0000] <Philip`>
jgraham: Just the latter, I think

[14:12:01.0000] <Philip`>
(http://code.google.com/p/html5lib/issues/detail?id=92 and http://code.google.com/p/html5lib/issues/detail?id=93 )

[14:13:00.0000] <Hixie>
serialisers should just quote everything unless they're trying to save bytes

[14:14:00.0000] <jgraham>
Philip`: Yeah I know about those issues. I guess we should quote the exra characters by default. Or maybe just always quote

[14:14:01.0000] <jgraham>
Hixie: Does saving bytes really make a compelling argument, especially given gzip?

[14:15:00.0000] <gsnedders>
I think we should quote them by default when no-quotes, but also quote everything by default

[14:15:01.0000] <Hixie>
jgraham: for some things, yes.

[14:16:00.0000] <jgraham>
Hixie: For example?

[14:16:01.0000] <Philip`>
If serialisers provide a make_output_more_efficient option, people are bound to select that option

[14:16:02.0000] <Hixie>
jgraham: if you're trying to fit something into one packet, and with gzipping you're one byte over, for instance.

[14:16:03.0000] <Philip`>
and even if most people don't select that option, some will, so the serialiser still has to carefully reverse-engineer legacy UAs in order to determine what it still needs to quote in order to prevent XSS attacks

[14:17:00.0000] <Philip`>
(because the spec doesn't provide any guidance)

[14:17:01.0000] <Philip`>
Hixie: In that case, you should just gzip harder!

[14:20:00.0000] <jgraham>
Hixie: That doesn't seem like a case that falls in the 80 part of 80/20. And in particular it doesn't seem probable that there will be a large number of pages that can't be made to fit into a single packet with quotes but do fit in without quotes

[14:22:00.0000] <Philip`>
And those pages can easily sacrifice validity anyway

[14:22:01.0000] <Hixie>
i agree that it's not a huge deal for most people, but i think making it possible is important enough.

[14:23:00.0000] <Hixie>
Philip`: if we're saying they are allowed do it, then we're saying it shouldn't be invalid.

[14:23:01.0000] <Hixie>
that's what "invalid" means

[14:23:02.0000] <Philip`>
Hixie: We're not saying they're allowed to do it

[14:23:03.0000] <Philip`>
but they can do it anyway so that's alright

[14:24:00.0000] <Philip`>
just like we're not saying they're allowed to use <center> simply because it's fewer bytes than <div style="align:center">, but they can do it anyway if they really care about saving bytes

[14:25:00.0000] <Hixie>
conformance is the technical equivalent of law (though without enforcement)

[14:26:00.0000] <Hixie>
just as it's not valid to say "well stealing is against the law but they can still do it if you need to", it's not valid to say "attributes must be quoted but they can still ignore that if they need to"

[14:26:01.0000] <Philip`>
Law without enforcement is just guidelines

[14:27:00.0000] <Hixie>
either we are ok with someone doing something, and we make it conforming, or we have good reasons to not allow it, and we want nobody to do it.

[14:28:00.0000] <Philip`>
That seems quite a valid thing to say - maybe my house is burning down so I can steal a bucket of water from a neighbour, even though it's against the law, and that's fine because there's nothing physically preventing you from doing that in extreme circumstances

[14:29:00.0000] <Hixie>
i disagree; i'd expect the law to allow such extremes explicitly, just like the law allows people to break other people's ribs legally in certain condititions (e.g. doing CPR to save their life)

[14:30:00.0000] <jgraham>
Well as Philip` says we disallow <center> even though it is shorter than <style>.a{text-align:center}</style><span class="a"> because the argument that it leads authors to do harmful things is stronger than the argument that people who need to save bytes should be allowed to write <center>

[14:32:00.0000] <jgraham>
(even though the argument about device-independent markup seems to be somewhat theoretical whereas the argument about quoting can be practically demonstrated)


2009-05-07
[14:21:00.0000] <jgraham>
/me wonders if there is a good way to count all the items in a generator whilst simultaneously filtering a subset of them

[14:21:01.0000] <jgraham>
(in python)

[14:22:00.0000] <jgraham>
(in a way that is cross-process safe)

[16:51:00.0000] <Hixie>
    * Tara runs a video sharing web site for people who want licensing

[16:51:01.0000] <Hixie>
       information to be included with their videos. 

[16:51:02.0000] <Hixie>
man is that a specific audience to target

[16:52:00.0000] <olliej>
Hixie: hehhe

[16:59:00.0000] <zcorpan_>
hmm i wonder how <hgroup> messes up implicit sections

[16:59:01.0000] <zcorpan_>
maybe <hgroup> should have the same rank as its highest child h* element instead of always having rank 1


2009-05-08
[17:00:00.0000] <zcorpan_>
so you can do <h1>foo</h1><h2>bar</h2><hgroup><h3>baz</h3><h4>quux</h4></hgroup>

[17:01:00.0000] <zcorpan_>
where baz is a subsection of bar

[17:01:01.0000] <Hixie>
non-issue in practice, imho, <hgroup> is almost always going to be the first heading

[17:01:02.0000] <Hixie>
and the rest of the time, people can use <section>

[17:01:03.0000] <Hixie>
no?

[17:03:00.0000] <zcorpan_>
would be nice if it worked as people expect when using implied sections

[17:03:01.0000] <Hixie>
i guess

[17:03:02.0000] <Hixie>
...send mail :-)

[17:04:00.0000] <zcorpan_>
jgraham: btw your outliner seems to use the wrong heading in the example above (it uses foo twice)

[17:15:00.0000] <MikeSmith>
Hixie: (or anybody else using tircd) - http://impressive.net/people/gerald/2009/05/xchat-inputcount.pl

[17:16:00.0000] <MikeSmith>
[[

[17:16:01.0000] <MikeSmith>
# xchat-inputcount.pl: add a character count next to xchat's input box.

[17:16:02.0000] <MikeSmith>
# (useful when twittering, to see if you are within 140 characters)

[17:16:03.0000] <MikeSmith>
]]

[17:48:00.0000] <zcorpan_>
↕The start tag SHOULD contain ↕ a ↕version ↕↕ attribute that declares ↕ the ↕version of XHTML in use. The version of this version of XHTML is ↕↕ -//W3C//DTD XHTML 1.1//EN ↕.

[17:48:01.0000] <zcorpan_>
http://www.w3.org/TR/2009/PER-xhtml11-20090507/xhtml11-rec-diff.html

[17:49:00.0000] <zcorpan_>
doctype is MAY

[17:50:00.0000] <Hixie>
i'm amused as to how the main effect HTML5 has had on the XHTML2 WG is that the XHTML2 WG keeps releasing new XHTML 1.x drafts instead of working on XHTML2

[17:51:00.0000] <zcorpan_>
1.1 adds lang=""

[17:52:00.0000] <Hixie>
does it define how it works?

[17:52:01.0000] <Hixie>
http://www.whatwg.org/specs/web-apps/current-work/#the-lang-and-xml:lang-attributes

[17:52:02.0000] <zcorpan_>
The ↑↑ lang ↑↑ attribute is defined in [ ↑↑ HTML4 ↑↑ ]. When this attribute and the ↑↑ xml:lang ↑↑ are specified on the same element, the ↑↑ xml:lang ↑↑ takes precedence. ↑↑ When both ↑↑ lang ↑↑ and ↑↑ xml:lang ↑↑ are specified on the same element, they SHOULD have the same value. ↑

[17:53:00.0000] <Hixie>
so "no" then?

[17:54:00.0000] <Hixie>
i guess they just defer to HTML4

[17:54:01.0000] <Hixie>
as for everything else

[17:56:00.0000] <zcorpan_>
img& usemap (IDREF)

[17:56:01.0000] <Hixie>
they broke it again?

[17:56:02.0000] <Hixie>
am i going to have to update my blog post

[17:56:03.0000] <zcorpan_>
was it ever fixed?

[17:56:04.0000] <Hixie>
http://ln.hixie.ch/?start=1172653243&count=1

[17:57:00.0000] <zcorpan_>
"...new version of this specification which did not fix one of its simplest problems"

[18:03:00.0000] <Hixie>
woah, creativecommons.org is branding BSD and GPL licenses as "cc"?

[18:03:01.0000] <zcorpan_>
hmm, 1.1 made the doctype "may" but in 1.0 it's still "must"

[18:12:00.0000] <zcorpan_>
http://lists.w3.org/Archives/Public/www-style/2009May/0065.html

[18:14:00.0000] <zcorpan_>
/me is confused as to why there should be pseudo-classes for aria

[18:14:01.0000] <zcorpan_>
oh well

[18:14:02.0000] <zcorpan_>
nn

[18:45:00.0000] <hsivonen>
Hixie: CC has called the GPL CC-GPL as early as when CC 1.0 series came out

[18:47:00.0000] <Hixie>
that's... surprising

[18:49:00.0000] <hsivonen>
I think non-program work and program licensing would be more interoperable if insteat of drafting CC-by and CC-(by)-sa CC had promoted CC-BSD and CC-GPL for all kinds of works

[18:49:01.0000] <hsivonen>
*instead

[18:49:02.0000] <Hixie>
yes, imho cc has done much harm through its proliferation of licenses

[18:50:00.0000] <hsivonen>
although I guess it's considered a feature that CC-by-sa doesn't require the preferred form of making modifications to be made available

[18:51:00.0000] <Hixie>
ironically, none of the CC licenses these days are appropriate for what I want at all (since they all require attribution)

[18:51:01.0000] <hsivonen>
given current state of the Internet and codecs, one would want to avoid distributing the preferred form of making modifications to many videos

[18:51:02.0000] <hsivonen>
yeah, attribution on the license level is a big problem

[18:51:03.0000] <hsivonen>
particularly for massively collaborative works like wikis

[18:52:00.0000] <hsivonen>
(when you'd want to integrate external works that don't specify the wiki as the party who receives attribution)

[19:45:00.0000] <roc>
the preferred form of making modifications to videos is the raw data

[22:12:00.0000] <olliej>
Philip`: ping?

[01:01:00.0000] <Philip`>
olliej: Pong

[01:02:00.0000] <olliej>
Philip`: you need to update your canvas compat chart :D

[01:04:00.0000] <Philip`>
olliej: I need to update the tests too

[01:04:01.0000] <olliej>
true

[01:04:02.0000] <Philip`>
particularly so I can discover how badly everyone fails at text rendering :-)

[01:04:03.0000] <olliej>
:p

[01:05:00.0000] <Philip`>
but that'll take quite a bit of time, and I haven't been sufficiently motivated to do it

[01:14:00.0000] <zcorpan_>
hgroup > h1:heading-level(3) ~ h2 { /* same styles as <h4> */ }

[01:16:00.0000] <MikeSmith>
whoa, what is this heading-level() magic?

[01:16:01.0000] <MikeSmith>
is that already supported in browsers?

[01:16:02.0000] <zcorpan_>
no

[01:17:00.0000] <zcorpan_>
it's something i made up this morning

[01:17:01.0000] <MikeSmith>
damn lazy browser developers, get to work implementing that already!

[01:18:00.0000] <zcorpan_>
the outline algorithm needs to be fixed first, at least :)

[01:19:00.0000] <MikeSmith>
zcorpan_: well, I hope your idea doesn't take

[01:19:01.0000] <MikeSmith>
because I like putting stuff into stylesheets like

[01:20:00.0000] <MikeSmith>
> ul > li > ul > li > ul > li > ul > li > ul > li

[01:20:01.0000] <MikeSmith>
when I look at it in vim, it shows up in red and yellow

[01:20:02.0000] <MikeSmith>
like christmas tree lights

[01:24:00.0000] <MikeSmith>
div[class="section"] > div[class="section"] > div[class="section"] > div[class="section"] > div[class="section"] > h2

[01:24:01.0000] <MikeSmith>
... is fun too

[01:24:02.0000] <MikeSmith>
that one adds pink

[02:58:00.0000] <jgraham>
zcorpan_: You can imagine doing something like hgroup > *:not(:heading-level(*)) to match subheadings

[02:58:01.0000] <jgraham>
But it is not really obvious or pretty

[03:04:00.0000] <Philip`>
Hmm, I've never tried WEFT (Microsoft's tool for generating EOTs) until now, but now that I have I can understand why EOTs never became popular

[03:05:00.0000] <Philip`>
I guess whoever wrote it really wanted to be writing an IDE, not a tool that simply needs two inputs (a font file, and the characters you want to include in it) and one output (a new font file)

[03:06:00.0000] <Philip`>
so it's all got customisable UI and it hooks into IE's rendering engine to extract the fonts and characters you're using, and can integrate with something on the web server to upload the new font files and new HTML/CSS and all that stuff

[03:07:00.0000] <Philip`>
plus it's got a project system, and a wizard, and a Windows menu with Tile Vertically and Tile Horizontally and Cascade Windows and all those really useful features

[03:08:00.0000] <jgraham>
Ah, you know you've gone really wrong when you find you have implemented a project system

[03:08:01.0000] <Philip`>
(This is version 5.3.2 of the tool)

[03:09:00.0000] <Philip`>
(from 2003)

[03:09:01.0000] <Philip`>
It's even got a mottled stone background in its toolbar

[03:10:00.0000] <Philip`>
http://www.em2-solutions.com/projs/weft.html - apparently it's those Swedish people's fault

[03:18:00.0000] <zcorpan_>
silly swedish people

[03:19:00.0000] <jgraham>
You're telling me :p

[03:19:01.0000] <zcorpan_>
jgraham: also has weird results when you have invalid markup like <hgroup><center><h1>

[03:20:00.0000] <Philip`>
Also, the fonts it generates seem a bit strange - e.g. if I have a page using normal Calibri, it doesn't draw any ligatures; but if I use WEFT to create an EOT for that page, the new font does draw ligatures

[03:20:01.0000] <Philip`>
except it doesn't really draw them properly - I have the word "fluffily" and it draws a 'fl' ligature, 'ff' ligature, and then it tries to draw 'i' except that glyph isn't in the font so it falls back on a default font instead

[03:21:00.0000] <zcorpan_>
is the "fi" ligature in the font?

[03:21:01.0000] <zcorpan_>
and the "f" glyph?

[03:21:02.0000] <jgraham>
zcorpan_: In the outliner. Weird. Thanks

[03:21:03.0000] <jgraham>
s/./?/

[03:22:00.0000] <Philip`>
(Also it doesn't handle substitutions properly, e.g. i followed by combining acute accent gets drawn like an i with an accent on its left, rather than a dotless i with an accent where the dot would go)

[03:23:00.0000] <Philip`>
zcorpan_: 'f' is; 'i' isn't; 'fi' isn't

[03:23:01.0000] <zcorpan_>
Philip`: then it would have to fall back on a default font anyway (since i and fi aren't present), no?

[03:24:00.0000] <Philip`>
Also, when I tested this yesterday it alternated randomly between two renderings when I refreshed the page, but now it seems more consistent...

[03:24:01.0000] <Philip`>
zcorpan_: No, it should use the 'ffi' ligature

[03:24:02.0000] <zcorpan_>
oh, forgot about ffi

[03:25:00.0000] <Philip`>
I guess it thinks there's an 'ffi' when it's deciding which glyphs to include in the font, but then forgets about the substitution rule for that ligature and so the characters "ffi" get turned into the glyphs 'ff' and 'i' instead of into 'ffi'

[03:26:00.0000] <Philip`>
or something like that

[03:26:01.0000] <Philip`>
Anyway, it lets me conclude that my font subsetting code is not a waste of time repeating what their tool/API already does, because I avoid most of these bugs :-)

[03:27:00.0000] <jgraham>
Philip`: But you don';t have a project mode!

[03:28:00.0000] <Philip`>
jgraham: I'm too lazy to do that, so I'll stick with my checkbox-to-select-font and textarea-to-select-characters and button-to-generate-the-output UI for now

[03:32:00.0000] <Philip`>
/me thinks someone should make an EOT-to-TTF service

[04:45:00.0000] <zcorpan_>
i think smylers might be on to something (re header and hgroup)

[05:06:00.0000] <zcorpan_>
http://forabeautifulweb.com/blog/about/lead_pipe/ - the two paragraphs talking about leading seems to be a reasonable use case for style=""

[05:06:01.0000] <jgraham>
zcorpan_: Yeah, it sounds plausible. My only concern is that it violates "explicit is better than implicit" by making <header> have surprising magic properties

[05:07:00.0000] <jgraham>
(which could be bad if we don't accuratley anticipate how <header> will be used)

[05:08:00.0000] <jgraham>
(although you could argue that the table headers algorithm is also rather magic, but at least that was based on the failure of the explicit solutions + research)

[05:08:01.0000] <jgraham>
(but I guess we can do research on this case)

[05:10:00.0000] <jgraham>
Philip`: If you have time it might be nice to look at how headers are used as children of <div class="header">. Like collect all instances of <div id/class="header"> and filter just <hx> desendants and <div> descendants

[05:24:00.0000] <zcorpan_>
<header><h1>x</h1><nav><h1>y</h1></nav><h1>z</h1></header> - what would you make of this?

[05:34:00.0000] <jgraham>
zcorpan_: I would assume that the second <h1> child of the <header> represented a different subsection

[05:35:00.0000] <jgraham>
So:

[05:35:01.0000] <jgraham>
+--x

[05:35:02.0000] <jgraham>
   +--y

[05:35:03.0000] <jgraham>
+--z

[10:59:00.0000] <Philip`>
jgraham: Haven't you set something up so you can collect that kind of data yourself? :-)

[11:00:00.0000] <Philip`>
by which I mean I probably won't be motivated to find time to look at it myself, since I'd probably have to write lots of Java to extract the right data, and I don't like writing DOM/SAX code in Java

[11:21:00.0000] <gsnedders>
/me sighs at some of the arguments around the intarwebs about HTML 5

[11:21:01.0000] <gsnedders>
Would it not be more productive to send substantial comments about the spec rather than write substantial blog posts bitching about the process (whether fair or not).

[11:24:00.0000] <Philip`>
gsnedders: No, because substantial comments about the spec would get disregarded due to the problems with the process

[11:24:01.0000] <Philip`>
and the solution to such a problem is not to send more substantial comments that will all get disregarded

[11:24:02.0000] <gsnedders>
OK, then bitch to W3C management and not on blogs?

[11:25:00.0000] <gsnedders>
Fork the spec?

[11:25:01.0000] <inimino>
gsnedders: not for people who have disengaged with the process by choice

[11:25:02.0000] <gsnedders>
They all seem more productive things to do in that case.

[11:25:03.0000] <inimino>
forking the spec sounds like work

[11:25:04.0000] <gsnedders>
Bitching on blogs will _not_ fix any problems.

[11:25:05.0000] <inimino>
which doesn't really satisfy the same human urges that bitching does ;)

[11:26:00.0000] <Philip`>
gsnedders: Those are worse options in the cost/benefit ratio, because complaining on blogs and is easy and it might possibly eventually cause some change, whereas the other options require a lot more effort

[11:27:00.0000] <Philip`>
The current evidence indicates it takes years of work for an unofficial fork of HTML to be seen as legitimate

[11:28:00.0000] <inimino>
and that's with the backing of major browser developers

[11:29:00.0000] <Philip`>
It's like the current HTML 5 is an asteroid heading catastrophically towards Earth, and blog comments are like bouncing little rocks off it in an attempt to deflect it onto a course that won't destroy the things those people care about

[11:30:00.0000] <Philip`>
whereas you're suggesting building an entire new asteroid and making sure that hits Earth first and knocks it out of the way so the other asteroid won't have any detrimental impact

[11:32:00.0000] <Philip`>
and your other suggestion is that those people should complain to God so that He declares the asteroid as unholy in the hope that the asteroid will realise the error of its ways

[11:32:01.0000] <gsnedders>
/me stops bitching about the bitchers

[11:32:02.0000] <Philip`>
I like this analogy

[11:33:00.0000] <inimino>
people who see the standards process as a way to strong-arm browser developers are unlikely to find satisfaction in the current process

[11:33:01.0000] <gsnedders>
/me wonders how Content-Type is parsed

[11:34:00.0000] <inimino>
(or in any other process that has a realistic chance to displace it)

[11:34:01.0000] <gsnedders>
Short immediately apparent answer: not how the spec says.

[11:35:00.0000] <inimino>
so I think there is some fallout from the breakdown of unrealistic expectations that had seemed reasonable for a few years

[11:36:00.0000] <gsnedders>
How boring. That page now has sane headers.

[11:37:00.0000] <gsnedders>
Garbage at the end is ignored.

[11:40:00.0000] <jgraham>
Philip`: Er, the point is that I haven't finished writing the data parsing tools 

[11:41:00.0000] <jgraham>
Particularly the more advanced stuff like integrating the validator.nu parser

[11:41:01.0000] <Philip`>
jgraham: Sounds like you've now got an excellent reason to finish them ;-)

[11:41:02.0000] <gsnedders>
Actually, garbage in general is ignored.

[11:42:00.0000] <jgraham>
Philip`: so would it obviously be horrendously slow to pipe sax output from validator.nu to other, saner, programming languages?

[11:43:00.0000] <jgraham>
Or is there some other reason that is a bad idea?

[11:43:01.0000] <Philip`>
jgraham: That does sound horrendously slow

[11:43:02.0000] <Philip`>
jgraham: You could use Jython

[11:43:03.0000] <Philip`>
(which shouldn't be as horrendously slow)

[11:43:04.0000] <jgraham>
Yeah, maybe Jython would be a good idea

[11:44:00.0000] <Philip`>
The overhead of serialisation and IPC and parsing seems like a really bad idea when there's so much data

[11:44:01.0000] <Philip`>
so processing the data as close to the parser as possible is a good thing

[11:44:02.0000] <Philip`>
Ooh, you could use Scala!

[11:45:00.0000] <jgraham>
Philip`: The idea, I guess, would be to pre-filter the data s much as possible (e.g. using regexps)

[11:45:01.0000] <Philip`>
I don't know what that is but people have talked about it so it must be alright

[11:45:02.0000] <Philip`>
jgraham: Regexps are slow :-p

[11:45:03.0000] <jgraham>
Faster than parsing :-p

[11:45:04.0000] <Philip`>
Hmm, not sure how much difference there'd be

[11:46:00.0000] <Philip`>
(particularly for streaming SAX parsing)

[11:46:01.0000] <jgraham>
Yeah but with the SAX parsing you still have to do more work to get useful information out 

[11:47:00.0000] <Philip`>
Depends what information you want to get out

[11:47:01.0000] <Philip`>
/me 's Java stuff uses SAX for some things and DOM for other things, depending on which seemed easier

[11:48:00.0000] <jgraham>
/me generally wants to end up writing processing code in something less painful than java

[11:48:01.0000] <jgraham>
Since it will just be a bunch of one-shot filters

[11:48:02.0000] <Philip`>
I haven't found it painful to write processing code in Java

[11:49:00.0000] <jgraham>
Philip`: You have found it painful enough that you don't want to do this for me :p

[11:49:01.0000] <Philip`>
It helps that the Java code is just extracting information and outputting a giant XML stream, and then I use command-line tools and Perl to do the rest of the analysis and summarisation

[11:50:00.0000] <Philip`>
jgraham: That would be equally painful in any language :-)

[11:50:01.0000] <Philip`>
because I'd have to work out what you actually want, and then how to extract it from a document tree, and then how to present it, and all of those things are language-independent

[12:22:00.0000] <jgraham>
Hmm, Jyhton seems to have the options of a stupidly out of date table version or a reasonably up to date beta :(

[12:22:01.0000] <jgraham>
s//s/

[12:23:00.0000] <Philip`>
The beta worked for me

[12:23:01.0000] <Philip`>
Actually it was an alpha when I used it

[12:25:00.0000] <Philip`>
I found it worked when I basically wrote something like

[12:25:01.0000] <Philip`>
from nu.validator import htmlparser; from org.xml.sax import InputSource; from org.xml.sax.helpers import DefaultHandler

[12:25:02.0000] <Philip`>
class TestContentHandler (DefaultHandler): def startElement(self, uri, localName, qName, attributes): ...

[12:25:03.0000] <Philip`>
parser = htmlparser.sax.HtmlParser(htmlparser.common.XmlViolationPolicy.ALLOW); parser.setContentHandler(TestContentHandler()); parser.parse(InputSource(java.io.ByteArrayInputStream(body)))

[12:25:04.0000] <Philip`>
so it's not too complex really

[12:26:00.0000] <Philip`>
(Uh, and I think 'body' was a byte[] read from disk somehow)

[12:28:00.0000] <jgraham>
Oh well if it is stable, maybe I will try it

[12:29:00.0000] <gsnedders>
/me wonders whether grepping over the entire moz-central is sane

[12:29:01.0000] <jgraham>
gsnedders: No

[12:30:00.0000] <jgraham>
/me wonders why gsnedders is trying to do that

[12:30:01.0000] <Philip`>
Handy hint: Don't use grep in a utf8 locale

[12:30:02.0000] <Philip`>
(It's really slow)

[12:30:03.0000] <jgraham>
Handy hint:mxr.mozilla.org

[12:30:04.0000] <Philip`>
/me therefore uses 'LANG=C grep', which is something stupid like a hundred times faster

[12:30:05.0000] <gsnedders>
/me follows Philip`'s advice of how to be stupid quickly

[14:45:00.0000] <gsnedders>
/me wonders if it is a bad idea to go to the theatre by himself

[14:49:00.0000] <jgraham>
gsnedders: Why would that be a bad idea?

[14:49:01.0000] <gsnedders>
jgraham: geeky loneliness?

[14:49:02.0000] <jgraham>
Nah. Just make sure you are watching Waiting for Godot or something

[14:50:00.0000] <jgraham>
Not Mama Mia!

[14:50:01.0000] <gsnedders>
jgraham: I was thinking more Avenue Q

[14:50:02.0000] <jgraham>
But then watching Mama Mia! is a bd idea in general

[14:50:03.0000] <gsnedders>
Hixie blatantly has far too much influence over me :P

[14:51:00.0000] <jgraham>
gsnedders: Well I guess that isn't as sutiable for watching in the singular as Beckett but I think it will be better than not seeing it at all

[14:52:00.0000] <jgraham>
(Not that I have seen it or anything)

[14:52:01.0000] <gsnedders>
Unless, of course, anyone is in London in the week-beginning 22nd June…

[14:52:02.0000] <jgraham>
gsnedders: Sorry

[14:53:00.0000] <jgraham>
Do you know where you are staying in Linkoping yet?

[14:53:01.0000] <gsnedders>
I got email about one apartment giving me some details about that, yet to get any reply to my reply

[14:54:00.0000] <gsnedders>
(I emailed her on Monday)

[14:55:00.0000] <jgraham>
You spoke to Bibbi then? The mysteries of Swedish letting are still, well, mysterious to me

[14:55:01.0000] <jgraham>
Since our first apartment was found for us and the second one we found very luckily

[14:55:02.0000] <gsnedders>
Bibbi was finding one for me

[14:56:00.0000] <jgraham>
Yeah, that makes sense

[14:56:01.0000] <gsnedders>
Sent me some information about one with photos, and I basically replied saying that it'd be all right, and yet to get a response from her

[14:57:00.0000] <jgraham>
She has been away a lot this week. And I guess since you said it will be OK it doesn't need any action 

[14:57:01.0000] <gsnedders>
I probably also better email her to make sure contract arrived

[16:59:00.0000] <Hixie>
man, i didn't notice the whatwg list had gotten so popular


2009-05-09
[17:00:00.0000] <Hixie>
we're up to 992 members

[17:13:00.0000] <Hixie>
holy batman, check out the amount of metadata on http://www.house.gov/

[17:14:00.0000] <Hixie>
/me fails to find a page that anyone would ever want to do anything with that has this metadata

[17:14:01.0000] <Hixie>
if anyone reading these logs finds pages with this metadata where the metadata would actually be useful, please do let me know

[01:58:00.0000] <jgraham>
/me particularly likes how house.gov has a <meta name=" DC.relation.conformsTo" content=""> where the content attribute lists lots of specificaions including XHTML 1.1, even though it doesn't even pass the machine-checkable conformnce criteria for that specificaion

[02:11:00.0000] <Hixie>
jgraham: it doesn't even pass the machine-checkable conformnce criteria for the "DC.relation.conformsTo" metadata item

[02:12:00.0000] <Hixie>
jgraham: so technically it doesn't _actually_ claim to support those...

[02:12:01.0000] <Hixie>
s/support/conform to/

[02:15:00.0000] <fearphage>
is this a place i can ask xhr questions or is there a better place?

[02:16:00.0000] <Philip`>
DC.relation.conformsTo is a refutation of the idea that there's no value in namespaces being URIs because you can look for the terms on Google instead to find the appropriate specification

[02:44:00.0000] <Philip`>
https://www.bethere.co.uk/ - "Please Note: Copy & paste is not permitted in this field" - why do they go to so much trouble to intentionally disable useful features?

[07:03:00.0000] <Philip`>
So, Opera now has Opera Turbo to optimise bandwidth usage on slow/expensive links

[07:03:01.0000] <Philip`>
The question is: How long will it take until the savings from Turbo exceed the bandwidth I have to use in order to download the latest version of Opera?

[09:40:00.0000] <gsnedders>
http://core.trac.wordpress.org/ticket/9689 — more yay

[12:44:00.0000] <eric_carlson>
Hixie: ping?

[12:58:00.0000] <eric_carlson>
Hixie: I am going to have to run so I will leave my question here for you to find later

[12:58:01.0000] <eric_carlson>
Hixie: the spec says that a media element that isn't in the DOM should not be collected until there is not more audio to play

[12:58:02.0000] <eric_carlson>
Hixie: and mentions paused(), ended(), and rate==0 as things to consider before allowing collection

[12:58:03.0000] <eric_carlson>
Hixie: it doesn't say what to do with a file that is slow to load

[12:58:04.0000] <eric_carlson>
Hixie: eg. 'new Audio("http://....")' may not be playable before the collector runs

[12:58:05.0000] <eric_carlson>
Hixie: so I think the spec should mention this case, but I am not sure what the guidance should be

[12:58:06.0000] <eric_carlson>
Hixie: maybe at least wait until readyState >= HAVE_METADATA?

[12:59:00.0000] <eric_carlson>
Hixie: I will look for you later to chat

[13:39:00.0000] <gsnedders>
What does & do before a variable name in a function definition in C++?

[13:40:00.0000] <Philip`>
Makes it pass-by-reference instead of pass-by-value

[13:41:00.0000] <gsnedders>
What's the difference between pass by reference and passing by pointer?

[13:41:01.0000] <Hixie>
it's an operator that returns a pointer with the address of the variable, right?

[13:41:02.0000] <Hixie>
the address of its operand, i should say

[13:41:03.0000] <gsnedders>
(Which is what * does, right?)

[13:41:04.0000] <Philip`>
i.e. the argument is not (as it would otherwise be) a copy of the parameter's value; it's the same variable

[13:42:00.0000] <Hixie>
oh in a function definition

[13:42:01.0000] <Hixie>
never mind

[13:42:02.0000] <Hixie>
/me shuts up

[13:42:03.0000] <Philip`>
gsnedders: Syntax

[13:42:04.0000] <gsnedders>
Philip`: Then why does one function definition in moz use both? :P

[13:42:05.0000] <Philip`>
The implementation of "void f(T& x)" and "void f(T* x)" is identical; the only difference is that in the first case you can refer to the value as "x", in the second case you have to use "*x"

[13:43:00.0000] <Philip`>
gsnedders: Probably a combination of conventions

[13:44:00.0000] <Philip`>
e.g. it's sometimes conventional to pass output arguments by pointer (so it's explicit)

[13:44:01.0000] <Philip`>
and it's usually conventional to pass complex data types by reference (since it's more efficient than passing by value, and you don't have to worry about all the * syntax)

[13:46:00.0000] <gsnedders>
Parsing of media types is crazy.

[13:49:00.0000] <Philip`>
I suggest building a lookup table for all media type strings

[13:54:00.0000] <gsnedders>
/me doesn't quite understand what Gecko does

[14:22:00.0000] <gsnedders>
Hmm, there seems to be nothing to stop a media type containing multiple forward chars

[16:58:00.0000] <gsnedders>
/me squints

[16:58:01.0000] <gsnedders>
A tourist page about Switzerland referencing Lac Léman, and not Lac de Genevè

[16:58:02.0000] <gsnedders>
* Genève


2009-05-10
[17:06:00.0000] <jwalden>
gsnedders: note that & doesn't necessarily mean implicitly-pointer, consider inlined functions may never even move a value into memory

[17:06:01.0000] <gsnedders>
/me points out his knowledge of C++ is for most intents non-existant

[17:09:00.0000] <Philip`>
Explicit pointers with * might not even move the value into memory, so there's no difference :-)

[17:10:00.0000] <jwalden>
true enough

[17:10:01.0000] <jwalden>
although it may require a smarter compiler to notice that, with references you get it for free

[17:11:00.0000] <Philip`>
I'd except references vs pointers to be purely a frontend feature, and the code that's passed to the optimiser would be identical

[17:11:01.0000] <Philip`>
*expect

[17:15:00.0000] <Philip`>
/me discovers the Status: header in CGI, and wonders where it's meant to be documented

[17:15:01.0000] <Philip`>
(Actually FastCGI but I think it's meant to be the same)

[02:06:00.0000] <williamcwilliams>
Oh man. I just want to thank you good people for cutting HTML 5's doctype declaration to <!DOCTYPE html>. And thank you for helping to phase out DTDs so I don't have to learn its convoluted syntax.

[02:07:00.0000] <Hixie>
our pleasure :-)

[03:40:00.0000] <Hixie>
wow, json allows duplicate names in an object

[03:40:01.0000] <Hixie>
i had no idea

[03:40:02.0000] <Hixie>
(well, they're a SHOULD NOT, but that's it)

[04:03:00.0000] <williamcwilliams>
I can only see trouble coming out of such a thing.

[04:03:01.0000] <Hixie>
i expect it's a mistake, people seem to say SHOULD in specs when they mean MUST all the time

[04:03:02.0000] <Philip`>
Do they say "SHOULD" or "should"?

[04:03:03.0000] <Hixie>
SHOULD in this case

[04:05:00.0000] <Philip`>
/me likes the HTML5 style of SHOULD/MUST/etc usage, since it makes it clear what behaviour needs to be tested

[04:08:00.0000] <Philip`>
Sometimes I think other specifications are mostly written for users (who want to know how they can expect their implementations to behave) and implementors (who want to know how their implementations should behave), neither of whom care much about the difference between 'should' and 'must'

[04:08:01.0000] <Philip`>
whereas something like HTML5 is also written for test writers, who do care about subtle differences between conformance requirements

[04:09:00.0000] <Hixie>
in practice, both of those groups (users and implementors) are hurt by the should/must confusion

[04:09:01.0000] <Hixie>
but yes, in general you are right

[04:11:00.0000] <Philip`>
I'd expect they're hurt mostly by the lack of defined error handling when there's duplicate names, rather than by the use of should/must

[04:18:00.0000] <Hixie>
i just mean they're hurt by lack of interop in general

[04:19:00.0000] <Philip`>
That seems largely independent of any should/must confusion

[04:20:00.0000] <Philip`>
e.g. the JSON spec requires quotes around keys, but lots of JSON implementations let you turn that off because it saves some bytes and it works in some parsers

[04:20:01.0000] <Philip`>
and it seems they don't really care about interoperability

[04:20:02.0000] <jgraham>
should/must confusion that prevents you writing tests hampers interop

[04:21:00.0000] <jgraham>
(also: in the case of JSON there is/was a clause that said roughly 'implementations may have any other behaviour they like'. Which rather undermines the rest of the specification)

[04:24:00.0000] <Philip`>
On an unrelated topic: Automated tests are great, but running them on a wide variety of web browsers is time-consuming. What we could do is sign up to an online ad system that lets you put arbitrary HTML in your ads, and then immediately get hundreds of thousands of testers with no effort and very low financial cost

[04:25:00.0000] <jgraham>
Then we'd have to pay to write tests

[04:26:00.0000] <jgraham>
It would be much better to create our own ad network so that we could inject html tests into other people's ads, and make money at the same time

[04:26:01.0000] <Philip`>
(and the HTML we put in the ads would be executing the tests and then sending the results (plus UA strings etc) back to the central server)

[04:26:02.0000] <Philip`>
jgraham: Uh, not quite sure what you mean

[04:27:00.0000] <Hixie>
/me gets into his flame-proof bunker and sends an e-mail to the list

[04:27:01.0000] <jgraham>
Philip`: Instead of running the tests on some other ad network, set up our own ad network

[04:27:02.0000] <jgraham>
Sell ads on the network -> make money

[04:28:00.0000] <Philip`>
jgraham: But that would be extremely hard

[04:28:01.0000] <jgraham>
Also run tests on the network -> get useful data

[04:28:02.0000] <Philip`>
and would take years to build up

[04:28:03.0000] <Hixie>
abarth has done tests on an ad network

[04:28:04.0000] <Philip`>
and to get on enough sites to get significant numbers of visitors

[04:28:05.0000] <Hixie>
you might want to speak to him

[04:28:06.0000] <zcorpan_>
http://blog.whatwg.org/media/vim-checker.png is a 404 (referenced from http://blog.whatwg.org/vim-checker )

[04:28:07.0000] <jgraham>
Philip`: Hard yes, extremely hard presumably not. And we have 'till 2022 :)

[04:29:00.0000] <Philip`>
/me was inspired by abarth's work on this :-)

[04:32:00.0000] <Philip`>
Hixie: Haven't you seen any World War 2 movies? Bunkers are the worst place to be if you're expecting flames, because someone will break down the door and you'll be trapped and will burn to death and it won't be very nice at all

[04:32:01.0000] <Hixie>
crap

[04:32:02.0000] <Hixie>
where should i hide then?

[04:32:03.0000] <Philip`>
You could go scuba diving

[04:32:04.0000] <Hixie>
/me goes scuba diving in a hurry

[04:33:00.0000] <Hixie>
ahblblublublbolbullblubloblub?

[04:34:00.0000] <williamcwilliams>
Say, I'm peering at the JSON specification via http://www.ietf.org/rfc/rfc4627.txt and it mentions in section 1.1 that it uses "MUST" and "SHOULD" as described in RFC2119. This leads me to believe that Mr. Crockford definitely intended such specific uses of those terms... Sorry if this is a crazy blast from the past here.

[04:35:00.0000] <Hixie>
a lot of spec writers put that boilerplate there without really understanding it, sad to say

[04:35:01.0000] <Hixie>
dunno if this is the case is doug's case

[04:36:00.0000] <Hixie>
but if he really meant SHOULD there, that's mighty odd

[04:37:00.0000] <zcorpan_>
" The names within an object SHOULD be unique."

[04:37:01.0000] <Philip`>
HTML5 had the same boilerplate and then proceeded to use the terms in ways that were not described in RFC2119

[04:37:02.0000] <Hixie>
indeed

[04:37:03.0000] <Philip`>
so it's a common problem :-)

[04:38:00.0000] <williamcwilliams>
That is pretty funny! Yet frustratrating at times, I'm sure.

[04:38:01.0000] <jgraham>
Hixie: Just skimming your new email; is it really necessary to do <span item id=a><span property="com.example.name">a</span></span> rather than something like <span id=a property="com.example.name>a</span>?

[04:39:00.0000] <zcorpan_>
Philip`: in HTML5's case it was more because of RFC2119 not expecting the classes of products that HTML5 was having, right?

[04:39:01.0000] <Philip`>
/me is thinking of the use of phrases like "must only", which have meaning only in English and not in RFC2119

[04:39:02.0000] <Hixie>
jgraham: you need the item="" attribute to declare the new group of name-value pairs

[04:40:00.0000] <jgraham>
Hixie: At the very least it would be good if <span item property> acted like <span item><span property>

[04:40:01.0000] <Hixie>
jgraham: you can't do item="" property="" on the same element in this case, because when you do that it means the new group is the value of the given property on whatever the corresponding "parent" item is

[04:40:02.0000] <zcorpan_>
Philip`: ah, ok

[04:41:00.0000] <Hixie>
jgraham: e.g. <span item="vcard"> <span property="agent" item="vcard"> <span property="...">...</span> </span> </span>

[04:41:01.0000] <Hixie>
jgraham: the middle one there is saying that "agent" is a property of the outer item, whose value is itself an item.

[04:41:02.0000] <Hixie>
luckily, that's basically as complex as this stuff gets.

[04:41:03.0000] <jgraham>
Hixie: Hmm. I will think about this more when I have understood the proposal beter, but my initial reaction is that people will get this wrong and be surprised

[04:42:00.0000] <Hixie>
i don't think the case of those bulbs will be a common case

[04:43:00.0000] <jgraham>
OK. 

[04:43:01.0000] <jgraham>
I'll see what other people say

[04:46:00.0000] <williamcwilliams>
Hixie: Do you mean that if, in your example, the second tag didn't have the the item attribute explicitly set, it would still have item="vcard" like its parent?

[04:47:00.0000] <Hixie>
if the second tag didn't have the item attribute explicitly set, the third one would set a "..." property on the first one, instead of the second

[04:47:01.0000] <Hixie>
so the first one would have two properties, "agent" and "..."

[04:47:02.0000] <Hixie>
instead of the first one having a property "agent" which itself has a property "..."

[04:54:00.0000] <Philip`>
Reversed domain names aren't sufficient for people with shared hosting on a single domain

[04:54:01.0000] <Philip`>
which seems a bit of a pain

[04:54:02.0000] <Hixie>
why not?

[04:54:03.0000] <Hixie>
just stick your path components on the end

[04:54:04.0000] <Philip`>
If everyone on Geocities uses com.geocities.* then there's no convention to avoid conflicts

[04:55:00.0000] <Hixie>
if everyone on Geocities uses microdata, I will hand in my spec editor's badge.

[04:55:01.0000] <Philip`>
If two people on Geocities use com.geocities.* then there's no convention to avoid conflicts

[04:55:02.0000] <Philip`>
and s/Geocities/something like Geocities that isn't dead yet/

[04:55:03.0000] <Hixie>
so buy a domain

[04:55:04.0000] <Hixie>
it's like $5 a year

[04:55:05.0000] <Hixie>
or hope for the best and use com.geocities.philip.foo

[04:56:00.0000] <Hixie>
(or use a uri, those are still allowed too, for teh crazy people who think uris are better!)

[05:00:00.0000] <Philip`>
I haven't read the spec so I don't know what it suggests; I'm just thinking that if it suggests a convention, it should suggest a convention that can work for independent people who share a domain name, even if it's something vague like adding a unique-in-that-domain identifier onto the end

[05:00:01.0000] <Hixie>
the spec doesn't have any informative text on that topic yet

[05:01:00.0000] <Hixie>
i'll make a note to add something to the intro section though

[05:01:01.0000] <Hixie>
thanks

[05:04:00.0000] <Philip`>
If I want to write a page containing data about two cats, and say that cat A is the parent of cat B, and cat B is a child of cat A, can I do that nicely?

[05:05:00.0000] <Hixie>
not yet. I have considered a a <ref> element to do that in a future version, but none of the use cases I had actually needed it, so I punted on it for now.

[05:05:01.0000] <Philip`>
It looks like the new microdata stuff is purely tree-shaped and can't do graphs, and there isn't any way for one item to refer to another

[05:05:02.0000] <Philip`>
since items don't have identifiers

[05:05:03.0000] <Hixie>
items have id="" attributes like any other html element

[05:05:04.0000] <Hixie>
but yes, there's no way to make arbitrary graphs yet

[05:07:00.0000] <Philip`>
Hmm, I suppose the API could be extended to expose $item->id() and $microdata->get_item_by_id('...')

[05:07:01.0000] <Philip`>
but then it won't work if I put each cat on a separate page, and want them to refer to each other

[05:07:02.0000] <Philip`>
(i.e. being in separate items of @docs)

[05:08:00.0000] <Hixie>
none of the use cases needed that

[05:08:01.0000] <Philip`>
so one would have to refer to items by a full URI instead

[05:08:02.0000] <Hixie>
but yeah, if we really wanted to support that, we could just have <ref> support cross-document references too

[05:09:00.0000] <Hixie>
usemap="" showed that wasn't too hot, though

[05:15:00.0000] <Hixie>
ok i just checked in a brief intro section. bed time now. I'm sure when I wake up I'll have record amounts of mail.

[05:15:01.0000] <Hixie>
nn

[05:19:00.0000] <jgraham>
Hixie: It would be nice if the corresponding item algorithm was written in priority order rahter than reverse priority order

[05:19:01.0000] <jgraham>
also: gn

[05:31:00.0000] <Philip`>
Hixie: s/desireable/desirable/

[05:34:00.0000] <Philip`>
Hixie: "Properties can also have values that are URLs. This is achieved using the a element and its href attribute." - but the example uses <img src> instead

[06:45:00.0000] <Philip`>
Is there some DOM API that will return a resolved URL?

[06:45:01.0000] <Philip`>
(and that works in real browsers)

[06:45:02.0000] <Philip`>
given a relative URL as input

[07:09:00.0000] <Philip`>
http://philip.html5.org/demos/microdata/demo.html

[07:10:00.0000] <krijnh>
/me likes the proposal

[07:11:00.0000] <Philip`>
Hmph, Gmail's 'basic HTML' UI doesn't let you select your From address :-(

[07:22:00.0000] <Philip`>
At least the proposal is easy to implement

[07:22:01.0000] <Philip`>
though JQuery is a bit rubbish because its attribute selectors only work half the time

[07:22:02.0000] <krijnh>
They do?

[07:23:00.0000] <Philip`>
$('[item]') finds <div item> in Opera (10) but not in Firefox (3.0); $(...).parents('[item]') doesn't find it in either

[07:23:01.0000] <krijnh>
Weird

[07:23:02.0000] <Philip`>
/me presumes it's to do with optimising some cases with querySelector or whatever it's called

[07:23:03.0000] <krijnh>
Yeah

[07:24:00.0000] <krijnh>
But JohnResig might know :)

[07:24:01.0000] <Philip`>
$('*').filter(function() { return this.getAttribute('item') !== null }) makes me unhappy

[07:24:02.0000] <krijnh>
http://james.padolsey.com/javascript/a-better-data-selector-for-jquery/ - just extend the selector, and create an :item selector

[07:25:00.0000] <Philip`>
That sounds even less efficient

[07:25:01.0000] <krijnh>
It probably is

[07:26:00.0000] <krijnh>
But jQuery users don't care too much about efficiency ;) *runs*

[07:26:01.0000] <Philip`>
I think my code is already O(n^2) but I don't want to make it slower

[07:27:00.0000] <krijnh>
(I'm a jQuery user as well btw)

[07:28:00.0000] <Philip`>
(I think I could make it O(n) without too much trouble, but I'm lazy and this way works)

[07:29:00.0000] <Philip`>
((The problem is finding the properties with a specific corresponding item))

[07:37:00.0000] <Philip`>
Oh, and JQuery's node.is('time') fails in (at least) Opera 10

[07:54:00.0000] <gsnedders>
Woah.  Kinda big commits.

[08:01:00.0000] <Philip`>
gsnedders: I suggest complaining about the solution without bothering to read it first

[08:02:00.0000] <gsnedders>
Philip`: That means not revising for exams that start on Friday

[08:03:00.0000] <Philip`>
gsnedders: Exams happen every year, HTML5 is only going to happen once in the whole of human civilisation

[08:03:01.0000] <Philip`>
You need to get your priorities right

[08:03:02.0000] <gsnedders>
However, HTML 5 will still be ongoing after my exams.

[08:31:00.0000] <jgraham>
Philip`: Ah, you beat me. I have a somewhat working html5lib based implementation which allowed me to conclude that the api is easy to implement

[08:31:01.0000] <jgraham>
But went for a bike ride rather thank finishing it

[08:32:00.0000] <Philip`>
You need to get your priorities right

[08:32:01.0000] <jgraham>
You would have won anyway :)

[08:33:00.0000] <Philip`>
You could still win at producing RDF output

[08:34:00.0000] <jgraham>
Yeah I might do that. Or I might not bother. Who know

[08:34:01.0000] <jgraham>
s

[08:39:00.0000] <Philip`>
http://news.bbc.co.uk/1/hi/programmes/click_online/8040863.stm - "These TV sets are not equipped with full web browsers, as some sites involve lots of small writing that would be difficult to read from a distance." - someone should tell them about the abilities of modern browsers to zoom and change the text size

[08:40:00.0000] <Philip`>
(Sounds like more attempts at walled gardens)

[10:07:00.0000] <jgraham>
"""If there is exactly one element whose corresponding item is item and that has as one of its property names the string "about"""". Does that mean (there is one element whose corresponding item is item) and (the corresponding item has a property called about)"

[10:08:00.0000] <jgraham>
or (of the elements whose corresponding item is item, there is exactly one called about)

[10:09:00.0000] <jgraham>
s/called/with a property called/

[10:10:00.0000] <Philip`>
/me wonders if jgraham is implementing the same thing he's implementing

[10:13:00.0000] <jgraham>
Philip`: Possibly. 

[10:14:00.0000] <Philip`>
I hope you're implementing it in your Python version

[10:15:00.0000] <jgraham>
Philip`: Indeed

[10:15:01.0000] <jgraham>
Although the RDF stuff is much more tedious than the whole rest of the microdata section, so I am tempted to stop and just implement the frontend

[10:16:00.0000] <jgraham>
(where by "much more tedious" I mean "I can't be bothered to work out the anser to the question above")

[10:17:00.0000] <jgraham>
(or implement any of the url functions)

[10:17:01.0000] <jgraham>
(although those are really needed anyway)

[10:37:00.0000] <gsnedders>
8:24 from Southampton Central to Stirling by train, with a 45 minute stop-over in London. Yay :\

[10:40:00.0000] <gsnedders>
Or I could fly Southampton to Edinburgh, then go on by train, which would save me a bit of time but would add hassle…

[11:27:00.0000] <gsnedders>
jgraham: you still around?

[11:28:00.0000] <gsnedders>
Anyone else who knows physics: you around?

[11:28:01.0000] <gsnedders>
Ah, Wikipedia helps.

[11:28:02.0000] <jgraham>
gsnedders: Yes

[11:28:03.0000] <gsnedders>
(i.e., nvm)

[11:28:04.0000] <gsnedders>
(Though I may still need help at some other time)

[11:40:00.0000] <gsnedders>
Anyone got any sane way of remembering which way around the ± is for the Doppler effect (i.e., within equations like f = f_s(\frac{v}{v±v_s}))?

[11:41:00.0000] <gsnedders>
/me should probably conclude he can work this out by thinking about it.

[11:46:00.0000] <gsnedders>
Peh. The 

[11:46:01.0000] <gsnedders>
So if the source and observer and getting closer, f must increase, and then it can be worked out from that.

[11:48:00.0000] <gsnedders>
Is that right?

[11:52:00.0000] <jgraham>
gsnedders: Yeah, things comping toward you go high pitched. If your answer doesn't give that you did it wrong

[11:52:01.0000] <gsnedders>
From Wikipedia: "Craig Bohren pointed out in 1991 that some physics textbooks erroneously state that the observed frequency increases as the object approaches an observer and then decreases only as the object passes the observer." — Huh?

[11:59:00.0000] <jgraham>
gsnedders: They're saying that the frequency doesn't get higher and higher as the object approaches you

[12:01:00.0000] <gsnedders>
OK, that's basically what I concluded after staring at that for five minutes.

[12:01:01.0000] <jgraham>
In unrelated news, I hate easy_install and whatever they call the braindead system it runs on

[12:03:00.0000] <Philip`>
You should be thankful that they didn't call it ezinstall

[12:04:00.0000] <jgraham>
Why the fuck is it trying to extract archives to my web server directory?

[12:06:00.0000] <jgraham>
/me notices Shelly's email from earlier

[13:23:00.0000] <Hixie>
jgraham: is that sentence in the rdf section clearer now?

[13:27:00.0000] <Hixie>
Philip`: to resolve URLs, just use the DOM attributes, the browser will then do it for you

[13:28:00.0000] <jgraham>
Hixie: I don't understand why there must only be a single element with item item

[13:29:00.0000] <Hixie>
then it's still not clear, hold on let me fix it more

[13:30:00.0000] <Hixie>
regenning, look in a few seconds

[13:30:01.0000] <jgraham>
OK

[13:30:02.0000] <jgraham>
(I assume that the actul condition is that for each item there is a single value of a property called about, right?)

[13:31:00.0000] <Hixie>
right

[13:31:01.0000] <Hixie>
regenned

[13:32:00.0000] <jgraham>
Yes, that makes sense now

[13:32:01.0000] <jgraham>
Thanks

[13:32:02.0000] <Hixie>
cool

[13:32:03.0000] <Hixie>
i'm very confused by shelley's e-mails

[13:33:00.0000] <Hixie>
was she wanting me to stop working on the use cases while she redid them?

[13:35:00.0000] <takkaria>
Hixie: I like the microdata draft, seems like it's far more understandable than RDFa

[13:35:01.0000] <Hixie>
sweet

[13:35:02.0000] <Hixie>
i've gotten far more positive feedback (mostly off-list or on IRC) than I expected

[13:35:03.0000] <Hixie>
and virtuall no negative feedback so far

[13:35:04.0000] <gsnedders>
Hixie: It sucks.

[13:36:00.0000] <Philip`>
It's easy to make something more understandable than RDFa; the problem is to make it sufficiently expressive to handle enough of what people want to do with it

[13:36:01.0000] <gsnedders>
(Because it's too long for me to have time to look at now.)

[13:36:02.0000] <gsnedders>
(See, I do actually have a reason!)

[13:36:03.0000] <Hixie>
gsnedders: dude it's actually surprisingly short

[13:36:04.0000] <Hixie>
gsnedders: there's even an incomplete intro section now too

[13:37:00.0000] <gsnedders>
Hixie: OK, the diff was long :P

[13:37:01.0000] <gsnedders>
The fact it renumbered half the TOC didn't helop

[13:37:02.0000] <gsnedders>
*help

[13:37:03.0000] <Hixie>
Philip`: yeah... it handles all the use cases i listed (i checked by making examples for most of them), but i'm sure there are things people didn't mention that will be harder

[13:37:04.0000] <Philip`>
gsnedders: Read the new section rather than the diff

[13:37:05.0000] <Hixie>
gsnedders: oh you want to look at the /source diff not the /index diff

[13:37:06.0000] <Hixie>
gsnedders: and the diff included all kinds of random stuff

[13:37:07.0000] <Hixie>
gsnedders: e.g. new interfaces and stuff

[13:37:08.0000] <Hixie>
gsnedders: which you don't need to read to get the microdata stiff

[13:37:09.0000] <Hixie>
stuff

[13:37:10.0000] <gsnedders>
Hixie: I know. But the email includes both, and /index is always first :P

[13:38:00.0000] <gsnedders>
/me goes back to question 11 of the 2004 exam paper

[13:38:01.0000] <takkaria>
I wonder how hard it would be to transplant hcard into the microdata thing

[13:38:02.0000] <Philip`>
I suppose an obvious complaint is that people who've already put a lot of work into defining RDF-based vocabularies would have to start over again defining a whole new naming system for the vocabulary

[13:38:03.0000] <Hixie>
gsnedders: hah

[13:39:00.0000] <Hixie>
takkaria: surprisingly easy, that's next on my list of things to mail about :-)

[13:39:01.0000] <Philip`>
(and it'd be nice to have a stronger mapping between them)

[13:39:02.0000] <Hixie>
Philip`: nah, they can use uris with this

[13:39:03.0000] <jgraham>
http://james.html5.org/microdata/

[13:39:04.0000] <jgraham>
Work in progress, if it gives the right result it is pure serendipity

[13:40:00.0000] <Hixie>
man you guys rock

[13:40:01.0000] <Philip`>
jgraham: Charset is wrong on test page

[13:40:02.0000] <Hixie>
it's barely been 12 hours and you've already made two separate implementations

[13:41:00.0000] <Philip`>
jgraham: JSON output needs indentation

[13:42:00.0000] <Philip`>
/me almost implemented RDF output too, but discovered it was not trivial to generate N3 when there's blank nodes, and gave up and ate biscuits instead

[13:42:01.0000] <Hixie>
it's not that hard is it? you just do _:nodeN where N is a globally increasing number per blank node

[13:43:00.0000] <Hixie>
basically each time you make a blank node you call it _:nodeN and then pretend it's not blank

[13:43:01.0000] <Philip`>
That's ugly

[13:43:02.0000] <gsnedders>
Hixie: See, it's all work and no µdata :P

[13:43:03.0000] <gsnedders>
s/µdata/play/

[13:44:00.0000] <jgraham>
Philip`: Fixed the pretty printing thing

[13:44:01.0000] <Hixie>
Philip`: it's rdf baby

[13:44:02.0000] <jgraham>
Oh, I think I fixed the other thing too but I'm not sure

[13:44:03.0000] <Philip`>
I wanted to do it like '<> custom:band [ custom:name "Jazz Band"; custom:size "12" ]'

[13:44:04.0000] <Philip`>
(I think that's possibly the right syntax)

[13:45:00.0000] <jgraham>
I got part of the way through doing the RDF thing at which point it became obvious that RDF was going to be more work than the whole microdata section so far

[13:46:00.0000] <jgraham>
So I stopped

[13:46:01.0000] <gsnedders>
rdflib++

[13:46:02.0000] <Philip`>
I suppose I need to generate a graph and then serialise it, rather than printing out a line each time I find a property

[13:46:03.0000] <gsnedders>
/me just tells Philip` to stop implementing RDF himself, as it isn't nice to implement

[13:46:04.0000] <gsnedders>
Also, see #swig

[13:46:05.0000] <jgraham>
/me was hoping it would work to make a list of triples and have something that took a list of triples and printed N3

[13:47:00.0000] <jgraham>
But maybe RDFLib is the way forward

[13:47:01.0000] <Philip`>
gsnedders: I'm not implementing RDF, I'm just implementing something that output text that happens to be a subset of RDF

[13:48:00.0000] <jgraham>
Implementing all the URL munging stuff is still a massive pain though

[13:48:01.0000] <jgraham>
(which I need anyway)

[13:48:02.0000] <jgraham>
(so I can't blame on RDF)

[13:48:03.0000] <Hixie>
URL munging stuff?

[13:49:00.0000] <gsnedders>
Hixie: For less favourable comments, see #swig :P

[13:49:01.0000] <Hixie>
is that archived anywhere?

[13:50:00.0000] <gsnedders>
Hixie: It is, but logs appear to be down

[13:50:01.0000] <gsnedders>
http://swig.xmlhack.com/ is the URL

[13:50:02.0000] <gsnedders>
no idea if it'll come back up

[13:50:03.0000] <Hixie>
can you paste us any feedback then? :-)

[13:51:00.0000] <gsnedders>
Hixie: That's effort :P

[13:51:01.0000] <Philip`>
jgraham: I took the approach of ignoring any URL stuff that wasn't trivial, because I was trying to make a rough proof-of-concept to see how the thing worked, rather than making an actual proper implementation :-)

[13:51:02.0000] <Hixie>
-_-

[13:51:03.0000] <gsnedders>
Hixie: I can, if you help me wit physics! :P

[13:51:04.0000] <Hixie>
objects with mass attract each other

[13:51:05.0000] <Philip`>
What about objects without mass?

[13:51:06.0000] <Hixie>
everything else is complicated

[13:51:07.0000] <gsnedders>
Hixie: I need more complex help :P

[13:52:00.0000] <Hixie>
Philip`: they go at the speed of light so from their perspective time doesn't move so they can't do anything

[13:52:01.0000] <Philip`>
gsnedders: You need to specify your problem in a more complex way than "help me wit[h] physics" if you want more complex help :-p

[13:52:02.0000] <gsnedders>
Hixie: http://pastebin.com/m332bac12

[13:53:00.0000] <gsnedders>
"Calculate the time for the wave to travel between these points (at x = 3.0m and x = 4.0m)" where we have a transverse wave given by y = 8.0 sin(12t-0.50x)

[13:53:01.0000] <Hixie>
gsnedders: let me know if they have any actual feedback :-)

[13:54:00.0000] <Hixie>
correct me if i'm wrong but that wave doesn't move

[13:54:01.0000] <Philip`>
It does if t does, surely

[13:54:02.0000] <Hixie>
since it has a constant amplitude

[13:54:03.0000] <Hixie>
no, that just changes the phase of the wave

[13:54:04.0000] <Hixie>
the wave itself doesn't move

[13:55:00.0000] <gsnedders>
Well, the answer is 0.04°s.

[13:55:01.0000] <gsnedders>
But how do you get there? :P

[13:55:02.0000] <Philip`>
That's just being pedantic :-p

[13:55:03.0000] <Hixie>
it's not pedanticism, it's physics!

[13:55:04.0000] <gsnedders>
But that is true :P

[13:55:05.0000] <Hixie>
if you have waves in water, and you have a buoy on that water, it just bobs up and down

[13:55:06.0000] <Philip`>
I assume the idea is to calculate how long it takes for the peaks to 'travel', or something

[13:55:07.0000] <gsnedders>
It doesn't help me get me to the answer.

[13:55:08.0000] <Hixie>
it doesn't go tearing down the wave like a surfer headed to the beach

[13:55:09.0000] <Philip`>
because it couldn't mean anything else

[13:56:00.0000] <Hixie>
sounds like it's asking the wavelength then

[13:56:01.0000] <Hixie>
or equivalently the frequency

[13:57:00.0000] <Hixie>
which of course you can link using LF = c (wavelength * frequency = propagation speed)

[13:57:01.0000] <Hixie>
er

[13:57:02.0000] <Hixie>
phase speed

[13:57:03.0000] <Hixie>
not propagation speed

[13:57:04.0000] <Philip`>
I guess you simply want sin(12*t1 - 0.5*3) = sin(12*t2 - 0.5*4), so 12*dt = 0.5 and dt=1/24 sec ( = 0.042ish)

[13:59:00.0000] <Philip`>
kind of skipping all the ideas about phase and sin not being invertable, because they make life complex

[13:59:01.0000] <Philip`>
*invertible

[14:02:00.0000] <Philip`>
(Perhaps it's more precise to state that the phase is 12t-0.5x and you want the phases to be equal at the two xs, rather than unjustifiedly asserting that the sins are equal)

[14:02:01.0000] <gsnedders>
Hixie: LF?

[14:02:02.0000] <Hixie>
lambda * f

[14:02:03.0000] <gsnedders>
Ah

[14:03:00.0000] <gsnedders>
/me is probably meant to do this from the phase difference, seeming the first part of this subquestion was working that out

[14:05:00.0000] <gsnedders>
Actually, can it not be done more simply through v = f*lamba and v = s/t?

[14:09:00.0000] <Philip`>
gsnedders: Why do you doubt that it could be?

[14:09:01.0000] <gsnedders>
Philip`: That seems so much simpler than what you and Hixie were saying :P

[14:12:00.0000] <Philip`>
gsnedders: How is it simpler than saying 12*t1-0.5*3 = 12*t2-0.5*4? :-)

[14:13:00.0000] <gsnedders>
/me shrugs

[14:14:00.0000] <jgraham>
Hixie: converting relative urls to absolute urls

[14:19:00.0000] <Philip`>
http://philip.html5.org/demos/microdata/demo.html - now with ugly buggy RDF output

[14:21:00.0000] <Philip`>
Hixie: "property value" is defined with "If the element also has an item attribute: The value is the item created by the element."

[14:21:01.0000] <Philip`>
Hixie: and the JSON thing says "Let value be the property value of element.  If value is an element, then get the object for the element value, ..."

[14:22:00.0000] <Philip`>
which seems inconsistent; the first bit would have to say "The value is the element" instead, or something

[14:26:00.0000] <gsnedders>
Philip`: It can't have RDF output as RDF is an abstract concept :P

[14:35:00.0000] <gsnedders>
/me has an awful lot of learn before the exam

[14:53:00.0000] <Hixie>
jgraham: oh. in JS or server-side?

[14:53:01.0000] <Hixie>
Philip`: yeah, i'll fix that, hold on

[14:56:00.0000] <Hixie>
Philip`: fixed (should be regenned in a few seconds)

[14:57:00.0000] <Philip`>
/me discovers that it's a bad idea to leave Wireshark running in the background, collecting half a million packets and storing them all in RAM

[14:57:01.0000] <jgraham>
Hixie: Server side

[14:57:02.0000] <Hixie>
jgraham: ah. good luck. :-)

[14:57:03.0000] <jgraham>
:(

[14:59:00.0000] <Philip`>
Surely there's a standard Python library for manipulating URIs?

[14:59:01.0000] <jgraham>
Philip`: Yes, but it's getting the right base uri to resolve against and that sort of thing

[15:00:00.0000] <jgraham>
Also, I think in principle to conform with html5 I might need to implement document encoding dependent stuff

[15:00:01.0000] <jgraham>
which is a pity since I have no idea what the document encoding was

[15:01:00.0000] <jgraham>
(alhough I guess I can ind out if necessary)

[15:01:01.0000] <jgraham>
*find

[15:01:02.0000] <Philip`>
Hixie: If I have <link rel="alternate alternate stylesheet stylesheet">, should *all* the tokens get removed and replaced with "alternate-stylesheet"?

[15:02:00.0000] <Philip`>
jgraham: Hmm, I suggest not bothering with those details :-)

[15:02:01.0000] <Hixie>
Philip`: yes; i'll clarify

[15:02:02.0000] <jgraham>
I think I will ignore the encoding-dependant stuff for now

[15:03:00.0000] <gsnedders>
/me doesn't like the µdata proposal

[15:03:01.0000] <gsnedders>
I'll send email when I have time, with some sort of alternative proposal

[15:03:02.0000] <jgraham>
Hixie: In general if I have <link rel="a a b b"> should it be handled the same as <link rel="a b">

[15:04:00.0000] <Hixie>
yeah

[15:04:01.0000] <jgraham>
if a,b!=up

[15:04:02.0000] <jgraham>
Oh that's good because that's what I implemented

[15:05:00.0000] <Philip`>
Nothing in the spec says to remove duplicates, as far as I can see, but I suppose RDF defines a graph with duplicate triples to be equivalent to one with the duplicates removed

[15:05:01.0000] <gsnedders>
Hixie: Short version: I'd prefer something closer to RDFa than the current draft, just without the insanity of RDFa.

[15:06:00.0000] <Hixie>
jgraham: updated the spec -- note the rel=up removal happens before duplicate removal

[15:06:01.0000] <Hixie>
jgraham: and duplicate removal, as Philip` points out, only matters for alternate stylesheet links

[15:06:02.0000] <Hixie>
gsnedders: your feedback or #swig's?

[15:06:03.0000] <gsnedders>
Hixie: mine

[15:06:04.0000] <Hixie>
what do you think is missing?

[15:07:00.0000] <gsnedders>
Hixie: There's nothing missing, it just seems to be reinventing the wheel for no reason.

[15:07:01.0000] <Philip`>
Hixie: If it contains >1 'up', then remove *all* 'up'? (Surely it should keep one?)

[15:07:02.0000] <Hixie>
gsnedders: i tried to stick as closely to RDFa as possible while fixing the problems mentioned in my e-mail

[15:08:00.0000] <Hixie>
Philip`: i can't see how to express the semantic of rel="up up" in rdf

[15:08:01.0000] <Hixie>
Philip`: without a complex set of bnodes

[15:08:02.0000] <gsnedders>
Hixie: As I said, I'll send something more substantial later

[15:08:03.0000] <Hixie>
gsnedders: cool

[15:08:04.0000] <Philip`>
Hixie: Oh, so it's intentionally ignored instead of being made equivalent to rel="up"?

[15:08:05.0000] <Hixie>
Philip`: yeah

[15:08:06.0000] <Philip`>
I suppose that makes some kind of sense

[15:08:07.0000] <jgraham>
gsnedders: I like the current draft in that I can imagine actually being able to author in it without wanting to kill people and then getting everything wrong anyway

[15:09:00.0000] <gsnedders>
Hixie: … that'll probably end up being an entire counter-proposal that nobody will agree with (SWIG people because it isn't RDFa and people here because you didn't write it :P)

[15:09:01.0000] <Hixie>
gsnedders: if you come up with something better, i'll agree with it :-)

[15:10:00.0000] <Hixie>
gsnedders: one of the difficult parts of designing the microdata syntax was getting the DOM API right

[15:10:01.0000] <Hixie>
still not sure i have it quite spot on

[15:10:02.0000] <gsnedders>
Hixie: See, I don't care about the DOM API :P

[15:10:03.0000] <jgraham>
gsnedders: So if you are going to write a counter-proposal please don't add lots of URIs and ways of marking up graphs of data

[15:10:04.0000] <jgraham>
gsnedders: Because they make everything too complicated

[15:10:05.0000] <gsnedders>
jgraham: How do we do anything related to RDF without doing the latter?

[15:11:00.0000] <Hixie>
what's RDF got to do with anything

[15:11:01.0000] <jgraham>
gsnedders: Like the current proposal?

[15:11:02.0000] <Hixie>
very few of these use cases involved having to do RDF

[15:11:03.0000] <Hixie>
and those that did are handled adequately by the RDF conversion section as far as I can tell

[15:11:04.0000] <jgraham>
gsnedders: I would much prefer a solution that hits the 80/20 point now and gets used than one that hits the 99/1 point now bu doesn't get used

[15:12:00.0000] <gsnedders>
/me does find it odd that the proponents of RDF have always given specific use-cases for what they want to do, and not use-cases for generic semantics basically

[15:12:01.0000] <gsnedders>
s/basically.

[15:12:02.0000] <gsnedders>
s#.#//#

[15:12:03.0000] <jgraham>
Especially if the current solution can be extended to do more of the 20 later

[15:13:00.0000] <Hixie>
gsnedders: as far as i can tell, they only do the latter :-)

[15:13:01.0000] <Hixie>
(with a few exceptions)

[15:14:00.0000] <Hixie>
(like Manu)

[15:14:01.0000] <gsnedders>
Hixie: No, they say, "we want RDF", they don't give use-cases for wanting generic semantics

[15:14:02.0000] <Hixie>
(and some of the people on Shelley's blog)

[15:14:03.0000] <Hixie>
gsnedders: oh, right, yes

[15:14:04.0000] <Hixie>
i see what you mean

[15:14:05.0000] <gsnedders>
Hixie: If they want RDF and you ask for use-cases, I'd expect them to give use-cases for generic semantics, not for specific uses of these generic semantics

[15:15:00.0000] <Hixie>
yeah

[15:15:01.0000] <Hixie>
i have heard basically no reason for "generic semantics" as you put it

[15:15:02.0000] <gsnedders>
Generic semantics isn't a very good term…

[15:15:03.0000] <gsnedders>
Extensible semantics?

[15:15:04.0000] <Hixie>
microdata is extensible :-)

[15:15:05.0000] <gsnedders>
Needless to say, you know what I mean :P

[15:18:00.0000] <gsnedders>
/me is unlikely to have time to really write any proposal until after at least his physics exam (in just under three weeks)

[15:18:01.0000] <Philip`>
/me wonders why his router vanishes at 22:13, and doesn't come back until 07:14, and has done this for the past few days

[15:19:00.0000] <Philip`>
Hixie: You could make rel="up up" equivalent to rel="up-up" (and similar for "up up up" etc), so that it doesn't lose data

[15:25:00.0000] <Hixie>
Philip`: *shrug*

[15:25:01.0000] <Hixie>
Philip`: not really the right semantic. and i don't want to make rel="up-up" equivalent. And what do you do with rel="up-up up"...

[15:26:00.0000] <Hixie>
really the right thing is to have a bnode

[15:26:01.0000] <Hixie>
<> <xhv:up> <_:a> . <_:a> <xhv:up> (href) .

[15:26:02.0000] <Philip`>
You do the same thing you do with rel="alternate-stylesheet stylesheet"

[15:26:03.0000] <Hixie>
true

[15:27:00.0000] <Hixie>
and with rel="http://www.w3.org/.../vocab#stylesheet"

[15:27:01.0000] <Hixie>
sigh


2009-05-11
[18:07:00.0000] <takkaria>
things I don't get: people saying that html5:property conflicts with rdfa:property

[18:07:01.0000] <takkaria>
it explicitly doesn't afaics

[18:08:00.0000] <Hixie>
takkaria: if anyone can think of a better attribute name, i'm happy to change it, just to avoid the argument

[18:08:01.0000] <Hixie>
takkaria: but i agree that it's a bit weird for the rdfa guys to be complaining that html5 conflicts with their future plans

[18:08:02.0000] <Hixie>
it's just like when i complained that rdfa conflicted with our future plans :-)

[18:11:00.0000] <takkaria>
I mean, I get that RDFa people want to use RDFa

[18:12:00.0000] <takkaria>
but you can tell RDFa:property and html5:property apart using strchr(content, ":")

[18:12:01.0000] <takkaria>
/me shrugs and goes off to do other things

[18:14:00.0000] <Hixie>
takkaria: RDFa property is not in their own namespace

[18:14:01.0000] <Hixie>
oh, i misunderstood

[18:15:00.0000] <Hixie>
takkaria: they're saying they want to add new features that remove that distinction

[23:54:00.0000] <MikeSmith>
what does the spec say about a case where a form has two controls with the same name?

[23:55:00.0000] <MikeSmith>
as far as I can see, it does not disallow a form to have multiple name instances

[23:56:00.0000] <MikeSmith>
parts of a form to have the same name

[23:59:00.0000] <MikeSmith>
ah, ignore me

[00:00:00.0000] <MikeSmith>
I realize there are of course cases where certain controls need to have the same name

[00:18:00.0000] <erlehmann>
MikeSmith, i think name is obsolete?

[00:45:00.0000] <MikeSmith>
erlehmann: I was meaning to refer to the name attribute on various form controls

[01:17:00.0000] <Philip`>
Hixie: You're not allowed to rename property now, there's already two legacy implementations!

[01:18:00.0000] <Hixie>
hah

[01:19:00.0000] <Philip`>
/me can't think of any names other than "key"

[01:20:00.0000] <zcorpan_>
Hixie: the microdata solution and the <dl> example gives more reason to introduce <di> (you duplicate the <dt> content in <dd><meta>)

[01:21:00.0000] <Hixie>
zcorpan_: yeah, i did notice that this made <di> have a reason for being

[01:21:01.0000] <Hixie>
was hoping no-one else would notice too :-P

[01:21:02.0000] <zcorpan_>
/me replies :)

[01:37:00.0000] <zcorpan_>
Hixie: will <link rel="alternate stylesheet" ...> and <link rel="alternate-stylesheet"> produce the same RDF output?

[01:38:00.0000] <zcorpan_>
s/">/" ...>/

[01:39:00.0000] <zcorpan_>
     <p property="com.damowmow.desc">Siamese color-point.

[01:39:01.0000] <zcorpan_>
     <img property="com.damowmow.img" alt="" src="/images/erwin.jpeg">

[01:39:02.0000] <zcorpan_>
Hixie: was the img intended to be a child of the p?

[01:49:00.0000] <zcorpan_>
Philip`: item=" " generates two types in your impl (both empty string)

[01:57:00.0000] <Philip`>
zcorpan_: Possibly fixed now

[01:59:00.0000] <Hixie>
zcorpan_: yes, they'll produce the same output.

[01:59:01.0000] <jgraham>
Ah so, unsurprisingly, "I cannot live with" has become a magic phrase

[01:59:02.0000] <Hixie>
zcorpan_: not sure what to do about it. (so will <link rel="http://..../#alternate-stylesheet">)

[02:00:00.0000] <zcorpan_>
Hixie: (true)

[02:01:00.0000] <zcorpan_>
maybe it's not a problem

[02:01:01.0000] <Hixie>
i think it is

[02:01:02.0000] <zcorpan_>
Philip`: yep, cool

[02:01:03.0000] <Hixie>
but i don't know what to do about it

[02:01:04.0000] <Hixie>
short of making those explicitly non-conforming, but even then

[02:02:00.0000] <zcorpan_>
can you produce a URL that cannot be produced otherwise?

[02:02:01.0000] <Hixie>
not as far as i can tell

[02:02:02.0000] <zcorpan_>
#alternate stylesheet

[02:02:03.0000] <zcorpan_>
with a space :)

[02:04:00.0000] <Philip`>
#ALTERNATE-STYLESHEET

[02:04:01.0000] <Hixie>
zcorpan_: heh

[02:04:02.0000] <Hixie>
Philip`: you can still do that as a url, right?

[02:05:00.0000] <Philip`>
Hixie: No, because they're lowercased first

[02:05:01.0000] <Philip`>
(I think)

[02:06:00.0000] <Hixie>
Philip`: urls are? that's a problem if so

[02:06:01.0000] <Hixie>
i don't intend urls to be lowercased

[02:07:00.0000] <Philip`>
"split the value of the element's rel attribute on spaces, obtaining list of tokens.  Convert each token in list of tokens to lowercase."

[02:07:01.0000] <jgraham>
Hixie: BTW, as a general comment it is rather confusing when the spec uses "lowercase" to mean "ascii only lowercase"

[02:07:02.0000] <jgraham>
(I know it is defined and there is alink and everything, but it is still confusing)

[02:08:00.0000] <Hixie>
jgraham: file a bug, i can change that terminology easily enough

[02:08:01.0000] <Philip`>
/me wonders if he's badly confused about things

[02:08:02.0000] <Philip`>
Hixie: As far as I can see, rel="http://..." will never produce any triples at all

[02:09:00.0000] <Philip`>
(because there's only "For each token token in list of tokens that contains neither a U+003A COLON character (:) nor a U+002E FULL STOP character (.), generate the following triple:")

[02:09:01.0000] <zcorpan_>
/me just noticed that Philip`'s impl didn't generate a triple for urls

[02:10:00.0000] <Philip`>
/me notes that his implementation does not attempt to be perfectly correct

[02:10:01.0000] <Philip`>
(but if there are things that are easy to fix then I can fix them)

[02:11:00.0000] <Hixie>
Philip`: oh right, for rel="" it's only values, i forgot

[02:11:01.0000] <Hixie>
Philip`: ok, will use uppercase

[02:19:00.0000] <zcorpan_>
/me updates html5-elements

[02:19:01.0000] <Hixie>
Philip`: done

[02:25:00.0000] <zcorpan_>
/me notes that the order is different in the Amanda examples, in Philip`'s impl

[02:28:00.0000] <jgraham>
zcorpan_: Order of hat?

[02:29:00.0000] <jgraham>
Argh. "what"

[02:29:01.0000] <zcorpan_>
jgraham: the "band" and "name" properties

[02:29:02.0000] <zcorpan_>
"In this example, the outer item represents a person, and the inner one represents a band:"

[02:29:03.0000] <zcorpan_>
and "This example is the same as the previous one, but all the properties are separated from their items:"

[02:30:00.0000] <zcorpan_>
Hixie: you could either move the second div to below the name paragraph or clarify that the order of the properties will be different

[02:30:01.0000] <Philip`>
Properties have no order

[02:30:02.0000] <Philip`>
(in the JSON version, at least)

[02:30:03.0000] <jgraham>
What Philip` said

[02:31:00.0000] <zcorpan_>
ok

[02:31:01.0000] <Philip`>
The output order is just an irrelevant artifact of the JS Object implementation and of the JSON serialiser

[02:31:02.0000] <jgraham>
At least I hope that is true because if if is, it's much easier to come up with nice API design

[02:32:00.0000] <jgraham>
And if it's not, my implementation is wrong :)

[02:33:00.0000] <zcorpan_>
the order of attributes in html is also supposed to be irrelevant, but browsers still have to represent them in the source order to plugins for web compat

[02:34:00.0000] <zcorpan_>
iirc

[02:34:01.0000] <jgraham>
zcorpan_: Right. But hopefully this case is different

[02:35:00.0000] <Philip`>
Hixie: The first example in the spec using 'getItem' looks a little broken

[02:35:01.0000] <Philip`>
(at least in the multipage version)

[02:35:02.0000] <zcorpan_>
it's only different until people start to rely on the order of a popular implementation

[02:35:03.0000] <Philip`>
because it does "i < something" instead of "i &lt; something"

[02:36:00.0000] <Philip`>
If all implementations implement it in the way the spec describes, they'll give the same order for the same input

[02:36:01.0000] <Philip`>
so there's no problem

[02:36:02.0000] <Philip`>
(and it doesn't matter that different inputs give different orders)

[02:36:03.0000] <jgraham>
Actually I think I would prefer it if the DOM APIs were unordered

[02:36:04.0000] <jgraham>
/me hadn't really looked at those

[02:37:00.0000] <Philip`>
(and since the order is an implementation detail (though one that may be constrained by compatibility concerns), it can be ignored in the introduction section)

[02:37:01.0000] <Philip`>
jgraham: You don't want to allow enumeration of properties?

[02:37:02.0000] <zcorpan_>
Philip`: it's a problem for implementations that assume that the order is irrelevant and do optimzations that will change the order

[02:38:00.0000] <jgraham>
Philip`: Yeah, not sure what to do about enumeration

[02:38:01.0000] <jgraham>
But having IndexGetter seems bad

[02:39:00.0000] <jgraham>
Since that just encourages people to rely on order

[02:39:01.0000] <Philip`>
Having IndexGetter seems good, because it's a useful feature

[02:39:02.0000] <Philip`>
and the spec could just require implementations to sort properties alphabetically, or whatever

[02:39:03.0000] <jgraham>
(enumeration seems very useful and as along as it is clearly stated that the order is arbitary, maybe it's not a problem)

[02:40:00.0000] <jgraham>
Philip`: Why?

[02:40:01.0000] <jgraham>
(is IndexGetter useful?)

[02:40:02.0000] <Philip`>
jgraham: Um, because I failed to understand what IndexGetter is

[02:40:03.0000] <zcorpan_>
Philip`: sorting alphabetically would have worked but is too late for html attributes

[02:40:04.0000] <Philip`>
and I was just thinking of enumeration

[02:40:05.0000] <Hixie>
the collections are ordered in tree order

[02:40:06.0000] <zcorpan_>
Philip`: still not too late for microdata :)

[02:41:00.0000] <Hixie>
zcorpan_: not sure what you mean about the divs earlier

[02:41:01.0000] <jgraham>
Hixie: Enforcing ordering is bad if you want to back the implementation with a hash table

[02:41:02.0000] <zcorpan_>
Hixie: <div subject="amanda" property="band" item id="jazzband"></div> starts the "band" property before seeing the "name" property

[02:41:03.0000] <zcorpan_>
Hixie: so the order is different from the previous example

[02:42:00.0000] <Hixie>
jgraham: collections are already ordered in the rest of the platform, so it won't need new code as far as i can tell

[02:42:01.0000] <Hixie>
zcorpan_: hm, yeah, might be best to not have that change

[02:42:02.0000] <Hixie>
Philip`: uppercased and fixed < issue

[02:42:03.0000] <jgraham>
Hixie: But this is something that non-browsers are likely to implement

[02:43:00.0000] <jgraham>
So having a requirement to be ordered also burdens them

[02:43:01.0000] <Philip`>
Interoperability between non-browsers seems unimportant

[02:43:02.0000] <Hixie>
jgraham: the dom api? i doubt it. why would they use that api? it's got all kinds of weird JSisms

[02:43:03.0000] <Philip`>
since the problem is just when two browsers run one script that makes assumptions that are only valid in one

[02:44:00.0000] <Philip`>
whereas it seems unlikely anyone will run the same code on two non-browser implementations and expect them to be identical

[02:44:01.0000] <jgraham>
Hixie: I guess if it only applies to the DOM API then I can't really object. I don't want people to have the general idea that items are ordered collections of name:value pairs

[02:45:00.0000] <jgraham>
s/I/I just/

[02:51:00.0000] <Hixie>
jgraham: maybe we should define some other api for other languages or something to prevent people from trying to implement it outside the dom

[02:51:01.0000] <Hixie>
jgraham: or maybe if html5lib gets an api for it and we publicise that, people will try to duplicate that one and it won't be an issue

[02:52:00.0000] <Hixie>
jgraham: i'm happy to put something in the spec if you think it'd help

[02:52:01.0000] <jgraham>
Hixie: Not worth worrying about at this point, I guess

[02:53:00.0000] <Hixie>
k

[02:53:01.0000] <zcorpan_>
/me wonders whether the microdata solution will get bad press because of its name cf predefined class names vs microformats before

[02:56:00.0000] <Hixie>
I'm happy to rename it if anyone has a better idea

[02:56:01.0000] <Hixie>
i wasn't really intending to call it "microdata", that's just what it is

[02:58:00.0000] <zcorpan_>
HRDF?

[02:58:01.0000] <zcorpan_>
i.e. Hixie-RDF

[02:59:00.0000] <Hixie>
heh

[02:59:01.0000] <Hixie>
it's not RDF per se

[03:00:00.0000] <Hixie>
any more than it is JSON or whatever

[03:00:01.0000] <zcorpan_>
that doesn't really matter :)

[03:01:00.0000] <Hixie>
it matters a little!

[03:01:01.0000] <jgraham>
FWIWW microdata seems like a totally fine name to me

[03:03:00.0000] <Philip`>
I don't understand why it's micro

[03:03:01.0000] <Hixie>
tiny bits of data

[03:03:02.0000] <zcorpan_>
/me wonders whether document.items and element.item will clash with web pages' scripts

[03:03:03.0000] <Hixie>
zcorpan_: :-/ hope not, hard to know ahead of time though

[03:03:04.0000] <Philip`>
Hixie: But it could be big bits of data, and lots of them

[03:04:00.0000] <Hixie>
Philip`: they can't really be that big, while still being useful.

[03:04:01.0000] <Hixie>
granted there could be many

[03:04:02.0000] <Hixie>
not sure how that stops them being small though :-)

[03:04:03.0000] <Philip`>
I suppose the most prominent feature is that it's data embedded into the markup, but a term like "embedded data" doesn't sound great to me

[03:04:04.0000] <zcorpan_>
<html item>

[03:05:00.0000] <Philip`>
Hixie: Any data could be viewed as a collection of lots of small bits of data, so that's hardly unique to the microdata proposal :-)

[03:05:01.0000] <zcorpan_>
i think it's primary feature is that it's easily extractable data while ignoring other content

[03:06:00.0000] <zcorpan_>
so maybe "Easily Extractable Data" (EED)

[03:06:01.0000] <Philip`>
zcorpan_: If that was the primary feature, you could use <script type="text/json">easily extractable data in JSON form</script> which would be much easier

[03:06:02.0000] <Philip`>
It's more significant that it's mixed up with the markup

[03:06:03.0000] <zcorpan_>
Philip`: "Embeeded Easily Extractable Data" (EEED)

[03:07:00.0000] <zcorpan_>
s/e/d//

[03:07:01.0000] <zcorpan_>
er

[03:07:02.0000] <zcorpan_>
s#/##

[03:07:03.0000] <Hixie>
zcorpan_: i don't like naming things in a way that presumes that a design criteria was successfully met ("easily")

[03:08:00.0000] <Philip`>
Call it Embedded Data With A Related Dom api, or Edward for short

[03:08:01.0000] <Hixie>
zcorpan_: because too often i have seen technologies called something like that only to have failed at that goal, and it just makes them easy targets for jokes :-)

[03:08:02.0000] <Hixie>
Philip`: heh

[03:08:03.0000] <zcorpan_>
Edward is a good name :)

[03:09:00.0000] <zcorpan_>
though the data-* stuff is also embedded data with a dom api

[03:10:00.0000] <Philip`>
Web Data

[03:15:00.0000] <Philip`>
Semantic Web 2.0

[03:16:00.0000] <Philip`>
By the way, implementing the RDF/N3 output thing made me dislike prefixes, because I couldn't work out what prefixes to use

[03:16:01.0000] <Philip`>
Searching for http://purl.org/dc/terms/ suggested that people use a random mixture of dcterms:, dct: and dc:

[03:17:00.0000] <Philip`>
and I don't like having to make uneducated choices between such things

[03:17:01.0000] <zcorpan_>
Philip`: randomize which prefix to use

[03:17:02.0000] <Hixie>
prefixes are a disaster. When I'm forced to use them I intentionally mint new ones like "a", "b", "c", etc.

[03:17:03.0000] <Hixie>
to prevent people from thinking they have special meanings.

[03:19:00.0000] <jgraham>
I think I will implement rdf output using rdflib (assuming I can work out how to use it). Hopefully that will give me magic random prefixes

[03:22:00.0000] <zcorpan_>
/me wonders whether form controls with microdata was considered

[03:23:00.0000] <Philip`>
I guess the conflict is between the viewpoint in which prefixes are just a syntactic shortcut (and so you can use whichever prefix you fancy, and there's no need to standardise them or have any conventions or design them carefully), and the viewpoint in which they're critical to human understanding of the data (since people think of them as being names with a colon in the middle, and try to use consistent meaningful prefixes, etc)

[03:23:01.0000] <Philip`>
It'd be fine if it was all one way, or all the other way

[03:25:00.0000] <zcorpan_>
Hixie: <input type=image src=foo property=bar>

[03:25:01.0000] <Philip`>
(e.g. if humans thought at that level of abstraction, or if humans never had to look at prefixes at all; or (the other way) if the prefixes really were just part of the name and were standardised etc)

[03:25:02.0000] <Hixie>
zcorpan_: <input> always uses the textContent

[03:25:03.0000] <Hixie>
zcorpan_: there were no use cases that needed form controls to be usefully linked into the microdata system

[03:26:00.0000] <zcorpan_>
i'm sure someone can mint a use case for that :)

[03:26:01.0000] <Philip`>
That could be a nice way of submitting structured data to a server

[03:27:00.0000] <Hixie>
not really

[03:27:01.0000] <Philip`>
without having to come up with unique names for every single input control in your page, and reconstructing that back into a tree structure on the server

[03:27:02.0000] <Hixie>
user input rarely maps directly to structured data schemas

[03:28:00.0000] <Philip`>
(particularly if the form controls are generated automatically, like in repetition templates)

[03:50:00.0000] <zcorpan_>
Hixie: maybe you could solve the naming/RDFa conflict debate and solve the potential Web script incompatibility problem by prefixing all relevant attributes with something

[03:53:00.0000] <Hixie>
any ideas?

[03:53:01.0000] <zcorpan_>
data- is taken :P

[03:54:00.0000] <Hixie>
indeed

[03:54:01.0000] <zcorpan_>
though data-* isn't much used yet so could be redesigned

[03:54:02.0000] <zcorpan_>
not that there's anything wrong with data-*

[03:56:00.0000] <zcorpan_>
"rdf"

[03:57:00.0000] <zcorpan_>
<div rdfitem><span rdfproperty="name">, document.rdfItems, element.rdfContent etc

[03:57:01.0000] <jgraham>
/me wonders if that is supposed to be a serious suggestion

[03:57:02.0000] <Hixie>
as noted before, this doesn't really have any more to do with rdf than json or anything else :-)

[03:58:00.0000] <jgraham>
Because it seems like a really bad idea :)

[03:58:01.0000] <zcorpan_>
Hixie: as noted before, that only matters a little

[03:58:02.0000] <zcorpan_>
jgraham: why is it bad?

[03:58:03.0000] <Hixie>
zcorpan_: :-P

[03:59:00.0000] <jgraham>
If you really want prefixes it should be really short like md-property

[03:59:01.0000] <Philip`>
Prefix all the names with µ

[03:59:02.0000] <Hixie>
it matters enough, imho

[03:59:03.0000] <Hixie>
Philip`: html attributes have to be ascii

[03:59:04.0000] <zcorpan_>
"md-" could work

[03:59:05.0000] <Hixie>
Philip`: (to make hsivonen happy)

[03:59:06.0000] <zcorpan_>
though it's no shorter than "rdf" :)

[03:59:07.0000] <Hixie>
md- is ugly

[04:00:00.0000] <jgraham>
zcorpan_: Because this isn't a generic RDF serialisation and all dong things that irritate the RDF people even more will not help anyone

[04:00:01.0000] <Philip`>
Hixie: Prefix them with org.whatwg.microdata-

[04:00:02.0000] <Hixie>
hah

[04:00:03.0000] <jgraham>
Plus it is silly to put rdf on a pedestal

[04:00:04.0000] <Philip`>
jgraham: "and all dong things"?

[04:01:00.0000] <Philip`>
/me 's error-correcting parser fails

[04:01:01.0000] <jgraham>
Philip`: s/all dong/doing/

[04:01:02.0000] <Philip`>
Ah

[04:01:03.0000] <jgraham>
I guess I started a different clause

[04:01:04.0000] <jgraham>
Hixie: md- is indeed ugly but anything longer than about 2-3 characters is irritating

[04:02:00.0000] <Hixie>
0 characters works for me :-)

[04:02:01.0000] <jgraham>
/me would prefer no prefix, really

[04:02:02.0000] <zcorpan_>
/me is slightly scared about web compat

[04:02:03.0000] <zcorpan_>
cf formaction=""

[04:03:00.0000] <Philip`>
zcorpan_: Unintentional uses of the attributes won't cause UAs to misbehave

[04:03:01.0000] <zcorpan_>
Philip`: i guess

[04:03:02.0000] <Hixie>
zcorpan_: let's address it when it's a problem

[04:04:00.0000] <jgraham>
It seems really unlikely that anyone is using the attributes in just the way needed to produce spurious microdata

[04:04:01.0000] <zcorpan_>
but it might make it hard to use on existing pages that already have <div id="item"> or something

[04:04:02.0000] <Philip`>
(http://philip.html5.org/data/attr-count-pages-dotbot.txt suggests most of the attributes aren't used too frequently

[04:05:00.0000] <Philip`>
)

[04:05:01.0000] <zcorpan_>
Philip`: the concern is about scripts having document.items matching <div id="items"> etc

[04:06:00.0000] <Philip`>
Oh, okay

[04:06:01.0000] <zcorpan_>
so implementing the dom api might make existing pages to stop working

[04:12:00.0000] <Hixie>
we can always make id=items overwrite .items

[04:17:00.0000] <Philip`>
id="items" is about the 1209th most common value

[04:18:00.0000] <Philip`>
(on 355 out of ~425K pages)

[04:24:00.0000] <Hixie>
ok, intro section at http://www.whatwg.org/specs/web-apps/current-work/#microdata is written

[04:25:00.0000] <Hixie>
it doesn't cover everything, and i'll add more examples later

[04:25:01.0000] <Hixie>
but it covers the basics

[04:28:00.0000] <Philip`>
"favorite-color" - urgh :-(

[04:30:00.0000] <Philip`>
"A valid reversed DNS identifier is a string that consists of a series of IDNA labels in reverse order (i.e. starting with the top-level domain), the prefix of which, when reversed and converted to ASCII, corresponds to a registered domain."

[04:30:01.0000] <Philip`>
but e.g. org.example.name is unlikely to correspond (directly) to a registered domain

[04:31:00.0000] <Philip`>
(since it'd be silly to register a subdomain for every single property you ever use)

[04:31:01.0000] <Philip`>
Oh

[04:31:02.0000] <Philip`>
I might have missed the bit saying "prefix"

[04:31:03.0000] <Philip`>
Seems a confusing definition since it's not at all clear what the prefix of "org.example.name" is

[04:32:00.0000] <Philip`>
"Types are identified in three ways:" - I only count two

[04:40:00.0000] <Hixie>
all the lists of ways you can identify things have "predefined types" commented out

[04:40:01.0000] <Hixie>
will fill that in later this week (maybe tomorrow)

[04:41:00.0000] <Hixie>
not sure how to fix the reversed dns identifier definition to be easier to understand

[04:41:01.0000] <Hixie>
bet time now

[04:41:02.0000] <Hixie>
nn

[04:42:00.0000] <Hish>
hi. is there a special tag which would help me to display page numbers, date, author, or filename on each page? can I use header/footer for this or is there something else? thanks.

[04:44:00.0000] <zcorpan_>
Hish: you mean when printed?

[04:46:00.0000] <Hish>
zcorpan_: yes

[04:48:00.0000] <zcorpan_>
Hish: i think you can do it with css but browser support is probably flaky

[04:49:00.0000] <zcorpan_>
Hish: position:fixed will repeat things on every page but might not be suitable for proper page headers

[04:49:01.0000] <zcorpan_>
Hish: see printed media, generated content and css3

[04:50:00.0000] <jgraham>
You might b e able to get it working using Prince

[04:50:01.0000] <Hish>
zcorpan_: thx. so the bottomline is that this is CSS related and there are no special tags for this.

[04:50:02.0000] <Hish>
Prince???

[04:51:00.0000] <zcorpan_>
Prince is a piece of software that is used to generate the PDF version of the html5 spec

[04:51:01.0000] <zcorpan_>
supports this kind of stuff

[04:52:00.0000] <jgraham>
Generally it's a HTML+CSS -> PDF converter

[04:54:00.0000] <Hish>
is it a proposal or an already working piece of code?

[04:54:01.0000] <zcorpan_>
the latter

[04:54:02.0000] <Hish>
sorry. forget it. didn't read your last comment, zcorpan_

[04:55:00.0000] <Hish>
do you have a url?

[04:57:00.0000] <jgraham>
http://www.princexml.com/

[05:45:00.0000] <zcorpan_>
Hixie: "This example creates a big list with a nested list for each item on the page, each with of all the property names used in that item./p>" markup typo

[05:46:00.0000] <zcorpan_>
a markup error that a validator wouldn't catch :(

[05:47:00.0000] <zcorpan_>
nor a spell-checker, presumably

[05:47:01.0000] <Philip`>
But you'd catch it when reading the page

[05:48:00.0000] <zcorpan_>
Philip`: after having done markup validation, you'd expect you would have caught all markup errors

[05:51:00.0000] <zcorpan_>
maybe a validator could issue a warning if there was a ">" before an implied tag, or similar

[05:51:01.0000] <jgraham>
zcorpan_: Your expectation would be wrong, however

[05:51:02.0000] <zcorpan_>
clearly

[05:52:00.0000] <zcorpan_>
i'd like to know if it's possible to make it right a bit more often, though

[05:52:01.0000] <jgraham>
/me wonders ho "kind" means something like "property"

[05:52:02.0000] <jgraham>
*how

[05:53:00.0000] <Philip`>
/me looks for synonyms

[05:53:01.0000] <Philip`>
We could replace it with the "realestate" attribute

[05:54:00.0000] <Philip`>
Since we already have a "subject" attribute, property could be renamed to "predicate"

[05:55:00.0000] <jgraham>
Philip`: Noooooooooooooooooooo

[05:56:00.0000] <Philip`>
Since the attribute is determining how the value is related to the item, it could be called "rel"

[05:56:01.0000] <jgraham>
Tip 1 for writing a usable markup language: don't use natural language words that most native speakers of the natural langauge don't understand 

[05:57:00.0000] <jgraham>
as part of the language

[05:59:00.0000] <Philip`>
Everyone should know what predicates are! You can't even do higher-order logic without them

[05:59:01.0000] <jgraham>
Oh noes!

[05:59:02.0000] <jgraham>
As gsnedders would say]

[06:02:00.0000] <zcorpan_>
Philip`: your impl doesn't generate a triple for the <title>

[06:04:00.0000] <Philip`>
zcorpan_: See line 186 or thereabouts

[06:04:01.0000] <Philip`>
// TODO: title

[06:06:00.0000] <zcorpan_>
/me notes that RDF uses title.textContent while document.title uses just the <title>s child text nodes

[06:11:00.0000] <Philip`>
zcorpan_: (Added title)

[06:13:00.0000] <zcorpan_>
Philip`: i get <> dct:title "HTML 5 microdata demo" . regardless of what i write as input

[06:14:00.0000] <Philip`>
zcorpan_: That's because that's the first title element on the page

[06:14:01.0000] <zcorpan_>
Philip`: maybe you could use an iframe?

[06:15:00.0000] <Philip`>
zcorpan_: Then I'd have to figure out how to make jQuery give me elements from within that iframe instead of the current document

[06:16:00.0000] <zcorpan_>
Philip`: you could replace title[0] with title[1]

[06:17:00.0000] <Philip`>
zcorpan_: I'd prefer to be implementing what the spec says, so any bugs are in the demo setup rather than in the parser code

[06:21:00.0000] <zcorpan_>
Philip`: btw http://simon.html5.org/tools/js/innerhtml-viewer/getInnerHTML.js has an ascii-lowercase function

[06:26:00.0000] <jgraham>
function toAsciiLowerCase(input) {return input.replace(/[A-Z]/g, function(char) {return char.toLowerCase()})}

[06:28:00.0000] <Philip`>
Hmph, '_:n2 c:org.w3.name "HTML5" .' isn't valid because it's not a proper QName

[06:29:00.0000] <Philip`>
or something like that

[06:29:01.0000] <Philip`>
http://www.rdfabout.com/demo/validator/ complains "Invalid token" so I assume it's right and I'm wrong

[06:32:00.0000] <jgraham>
/me wwonders how he got spidermonkey to fill up empty array elements with characters

[06:32:01.0000] <jgraham>
doing something like ["a", "b"].concat(["c",,"d"])

[06:41:00.0000] <Philip`>
N3 output should now be correcter

[06:42:00.0000] <Philip`>
and conciser

[06:42:01.0000] <Philip`>
except for the bits that had to become verboser in order to be correcter

[06:47:00.0000] <zcorpan_>
c:org.w3.name is not a proper QName?

[06:48:00.0000] <Philip`>
It's not a proper "QName or something like that", where the "something" is (([A-Z_a-zÀ-ÖØ-öø-˿Ͱ-ͽͿ-῿‌-‍⁰-↏Ⰰ-⿯、-퟿豈-﷏ﷰ-�𐀀-�][\-0-9A-Z_a-z·À-ÖØ-öø-ͽͿ-῿‌-‍‿-⁀⁰-↏Ⰰ-⿯、-퟿豈-﷏ﷰ-�𐀀-�]*)?:)?[A-Z_a-zÀ-ÖØ-öø-˿Ͱ-ͽͿ-῿‌-‍⁰-↏Ⰰ-⿯、-퟿豈-﷏ﷰ-�𐀀-�][\-0-9A-Z_a-z·À-ÖØ-öø-ͽͿ-῿‌-‍‿-⁀⁰-↏Ⰰ-⿯、-퟿豈-﷏ﷰ-�𐀀-�]*

[06:48:01.0000] <Philip`>
according to http://www.w3.org/2000/10/swap/grammar/n3-report.html

[06:49:00.0000] <zcorpan_>
so no dots

[06:49:01.0000] <aja>
is audio element still around? wondered why it wasn't in microdata additions like img, object, video, & a are

[06:50:00.0000] <zcorpan_>
it's still around

[06:50:01.0000] <aja>
/me thought so

[06:50:02.0000] <Philip`>
zcorpan_: Lots of dots, but not "." in particular

[06:51:00.0000] <Philip`>
aja: Sounds like an oversight

[06:51:01.0000] <jgraham>
aja: Seems like it makes as much sense as <video> so probably an omisssion

[06:51:02.0000] <jgraham>
accidential omission

[06:51:03.0000] <aja>
can someone pls mention to Hixie......not gonna online rest of day

[06:52:00.0000] <aja>
be online, that is

[06:52:01.0000] <Philip`>
aja: You could email it somewhere

[06:52:02.0000] <Philip`>
but I guess Hixie will read the IRC logs anyway

[06:53:00.0000] <aja>
k...will f/u if i dont see an update.....later all

[06:54:00.0000] <zcorpan_>
Philip`: maybe you should not use any prefixes to prevent such bugs?

[06:54:01.0000] <Philip`>
zcorpan_: That would make it unnecessarily hard to read

[06:56:00.0000] <Philip`>
Also, I don't care much about bugs - readability is more useful

[07:01:00.0000] <Philip`>
"<[object Object]> c:about _:n0 ." - hmm

[07:30:00.0000] <Philip`>
It's confusing that <span item><span item><span property=foo>bar</span></span></span> generates no properties

[07:39:00.0000] <Philip`>
/me sees http://chatlogs.planetrdf.com/swig/2009-05-11.html#T10-27-03 has a few more HTML5-related comments

[07:51:00.0000] <jgraham>
Twitter really, really sucks for, well, as far as I can tell, for pretty much everything. But in particular for all the myriad types of communication where reasoned argument is needed

[07:51:01.0000] <fearphage>
jgraham: for instance?

[07:53:00.0000] <gavin>
fearphage: you two should continue this discussion on twitter

[07:53:01.0000] <fearphage>
you may have a point

[07:56:00.0000] <jgraham>
fearphage: Ironically it is hard to explain on IRC for rather similar reasons. But look e.g. at rubys's titter page and then try to follow any of the threads backward

[07:57:00.0000] <jgraham>
*twitter

[07:58:00.0000] <jgraham>
And wonder if the medium is not distorting the message in a rahter harmful way, by allowing people to be irrational by enforced brevity

[07:58:01.0000] <jgraham>
So that reasoned argument being replaced by pithy one-liners is the norm

[07:58:02.0000] <jgraham>
and is totally expected and so unremarkable

[08:01:00.0000] <fearphage>
jgraham: that's twitter's ui failing really. it should show threaded views of conversations

[08:03:00.0000] <zcorpan_>
and not have a silly 140 char limit

[08:03:01.0000] <Philip`>
It doesn't have a silly 140 char limit - it has a sensible 140 char limit

[08:04:00.0000] <Philip`>
If it didn't have that limit then it would be a blog, and everyone has a blog already so there wouldn't be any point in it

[08:06:00.0000] <jgraham>
Philip`: Indeed. But the 140 character limit is both the distinguishing design feature and the reason it sucks

[08:06:01.0000] <jgraham>
Well that and the UI

[08:07:00.0000] <jgraham>
But the UI problem is a distant decond

[08:07:01.0000] <jgraham>
second

[08:07:02.0000] <zcorpan_>
if it sucks so much, you could come up with a competing alternative and make people switch and earn lots of money

[08:08:00.0000] <jgraham>
zcorpan_: I don't deny that people like it. But people like taking heroin but that sucks too

[08:08:01.0000] <jgraham>
So I don't know what to do

[08:08:02.0000] <jgraham>
Apart from maybe to ignore it

[08:09:00.0000] <zcorpan_>
jgraham: give up and take heroin, then twitter about it

[08:14:00.0000] <zcorpan_>
would maybe be nice to be able to give RDF/XML or n3 as input and convert it to html5 microdata

[08:15:00.0000] <Philip`>
Is that possible?

[08:16:00.0000] <zcorpan_>
maybe not losslessly

[08:16:01.0000] <Philip`>
And: Is that possible in general, or only in very limited situations e.g. where all the predicates are the xhtml/custom namespace?

[08:17:00.0000] <jgraham>
Why is that useful?

[08:17:01.0000] <zcorpan_>
didn't say it would be useful, only maybe nice 

[08:18:00.0000] <zcorpan_>
e.g. if you're familiar with microdata syntax but not rdf

[08:19:00.0000] <Philip`>
<meta item id="n0"> <meta subject="n0" property="about" content="http://subject/"> <meta subject="n0" property="http://predicate/" content="value">

[08:19:01.0000] <Philip`>
Would that work?

[08:19:02.0000] <Philip`>
as a general way to encode (non-nested) triples

[08:20:00.0000] <Philip`>
It seems useful to be able to write data using existing RDF vocabularies, so you can easily reuse all the work that's gone into defining them

[08:23:00.0000] <zcorpan_>
how many are subscribed to www-archive?

[08:23:01.0000] <Philip`>
How many read it via web archives?

[08:23:02.0000] <Philip`>
(since that mode of usage is the reason for the group's existence, after all)

[08:24:00.0000] <Philip`>
Oh, hmm, content="http://..." won't work because that's not an absolute URL

[08:26:00.0000] <Philip`>
Actually... Is it?

[08:26:01.0000] <Philip`>
"If name is not an absolute URL, ..." - how can it ever be an absolute URL? It's always just a string token

[08:28:00.0000] <Philip`>
I can't tell whether (absolute) URLs are meant to be a type distinct from strings, or whether they're just any string that happens to resolve into itself with no errors when you use the URL resolution algorithm

[08:29:00.0000] <jgraham>
/me thought the latter

[08:30:00.0000] <Philip`>
I assumed the former, but I guess it must be the latter, but then I have no idea whether <span property=foo>http://google.com/</span> should be an RDF resource (<http://google.com/> in N3) or still just a string ("http://google.com/")

[08:31:00.0000] <jgraham>
I guess a string

[08:31:01.0000] <jgraham>
But clarity would be nice

[08:31:02.0000] <Philip`>
It makes sense to me that any content in href/src/etc should be a resource, whereas any other content should just be a string

[08:31:03.0000] <jgraham>
That's what I assume

[08:31:04.0000] <Philip`>
If you make it a string then you lose the whole graphiness of RDF, which would be silly

[08:31:05.0000] <Philip`>
Wait

[08:31:06.0000] <Philip`>
No you don't, I'm confused again

[08:32:00.0000] <jgraham>
/me doesn't know

[08:32:01.0000] <Philip`>
I'm sure the <span>http:// ought to be a string, because anything else is crazy

[08:32:02.0000] <jgraham>
But it makes sense that if you want to make a link to another resource you have to say <a href="http://my.other.resource">

[08:32:03.0000] <Philip`>
but property="http://..." ought to be a URL, because anything else is crazy

[08:33:00.0000] <Philip`>
and href="http://..." ought to be a URL, because anything else is crazy

[08:33:01.0000] <jgraham>
Yes

[08:33:02.0000] <Philip`>
so the spec must be wrong because it uses the same terminology for property and for values

[08:34:00.0000] <jgraham>
/me wishes the spec called values "values" rather than "property values"

[08:35:00.0000] <jgraham>
Philip`: Does it? Where

[08:35:01.0000] <jgraham>
?

[08:42:00.0000] <Philip`>
<meta item id="n0"> <link subject="n0" property="about" href="http://subject/"> <link subject="n0" property="http://predicate/" href="http://value/"> <meta subject="n0" property="http://predicate/" content="value"> should work, I think

[08:42:01.0000] <Philip`>
(though not in my implementation)

[08:43:00.0000] <Philip`>
jgraham: Um...

[08:43:01.0000] <jgraham>
Hmm, so it is necessary to keep tabs on where a property came from to determine if it is a url or a string that looks like a URL

[08:43:02.0000] <Philip`>
jgraham: I'm still just being confused

[08:43:03.0000] <jgraham>
I have a bug there too, then

[08:44:00.0000] <jgraham>
Philip`: This only really applies to the RDF serialization, right?

[08:44:01.0000] <jgraham>
The json serialization seems to have data loss...

[08:44:02.0000] <Philip`>
In order to produce RDF output (but not JSON), concept-property-value needs to remember whether it returns a URL or a string (which may or may not look like a URL)

[08:44:03.0000] <Philip`>
and then you use <...> or "..." depending on which it is

[08:45:00.0000] <Philip`>
but for 'type' and 'name' you just have a string and look to see whether it looks sufficiently like an absolute URL to be treated like one

[08:45:01.0000] <jgraham>
Philip`: Is that a spec quote?

[08:46:00.0000] <Philip`>
function looks_like_an_absolute_url(str) { return str.match(/^(([a-zA-Z][0-9a-zA-Z+\-\.]*:)\/{0,2}[0-9a-zA-Z;\/?:@&=+$\.\-_!~*'()%]+)?(#[0-9a-zA-Z;\/?:@&=+$\.\-_!~*'()%]+)?$/); }

[08:46:01.0000] <Philip`>
That'll do for me, I guess

[08:46:02.0000] <Philip`>
jgraham: Is what a spec quote?

[08:46:03.0000] <jgraham>
"In order to produce RDF output..."

[08:47:00.0000] <Philip`>
jgraham: Regardless of your answer, my answer is no

[08:47:01.0000] <Philip`>
The spec doesn't say anything about this for the RDF output, because it ignores types entirely

[08:47:02.0000] <jgraham>
OK, so the spec is wrong here

[08:47:03.0000] <jgraham>
In some sense

[08:48:00.0000] <jgraham>
s/wrong/incomplete/

[08:48:01.0000] <Philip`>
unless you implicitly read some typing in the bits where it says "the value is an absolute URL"

[08:48:02.0000] <Philip`>
(which I do, because that's sensible)

[08:48:03.0000] <jgraham>
(yeha, I did that too)

[08:49:00.0000] <Philip`>
So, anyway, the spec should clarify that, but otherwise it seems like it's capable of representing arbitrary subject/predicate/object triples as long as you ignore all the extra bits like types and languages

[08:49:01.0000] <Philip`>
so you could do an RDF-to-text/html converter

[08:50:00.0000] <jgraham>
/me isn't sure about types and names as urls

[08:51:00.0000] <jgraham>
Oh wait. I think it's OK

[08:51:01.0000] <jgraham>
I'm being silly

[08:51:02.0000] <jgraham>
Probably

[08:52:00.0000] <zcorpan_>
how does RDF/XML distinguish between URLs and strings?

[08:52:01.0000] <jgraham>
So: Any type or name that looks like a URL is a URL. Any value that somes from a URLish place is a URL but any value that comes from text is text

[08:57:00.0000] <Philip`>
zcorpan_: <predicate rdf:resource="object"/> vs <predicate>object</predicate>

[08:58:00.0000] <Philip`>
zcorpan_: according to '<#n1> <#n2> <#n3> . <#n4> <#n5> "#n6" .' in http://www.rdfabout.com/demo/validator/

[09:01:00.0000] <zcorpan_>
Philip`: ok

[09:03:00.0000] <zcorpan_>
The underlying triples

[09:03:01.0000] <zcorpan_>
:n1 :n2 :n3 .

[09:03:02.0000] <zcorpan_>
:n4 :n5 "#n6" .

[09:04:00.0000] <Philip`>
(In N3, the empty prefix is <#>)

[09:04:01.0000] <Philip`>
(by default)

[09:37:00.0000] <zcorpan_>
will someone send email about the url vs string issue?

[09:41:00.0000] <jgraham>
I will if no one else does, but not before I have had a serious attempt at implementing that part of the spec

[09:41:01.0000] <jgraham>
Which might be this evening but is more likely not to be

[09:43:00.0000] <gsnedders>
/me should probably buy Lightroom at academic prices before he leaves school

[09:43:01.0000] <jgraham>
gsnedders: You can still do that at university you realise?

[09:44:00.0000] <gsnedders>
jgraham: But if I'm not going to uni for a year, I then couldn't do it for a year :P

[09:44:01.0000] <Philip`>
You could ask somebody still in school/university to buy it for you

[09:44:02.0000] <csarven>
Is there a document somewhere that outlines XML Namespaces as helpful (as opposed to http://wiki.whatwg.org/wiki/Namespace_confusion ) ? Thought it'd be good to have data for both extreme cases, unless, of course, if Namespaces is 100% confusing and useless for all developers.

[09:45:00.0000] <Philip`>
csarven: Namespaces helpful? That's blasphemy!

[09:45:01.0000] <jgraham>
csarven: You may have come to the wrong place :)

[09:45:02.0000] <csarven>
Unless I'm mistaking, the existing approach is hardly scientific. :)

[09:46:00.0000] <Philip`>
/me is not aware of documents listing the benefits of namespaces over other approaches, but assumes there must be some somewhere

[09:46:01.0000] <jgraham>
(But I guess the TAG will have something about how they are wonderful somewhere)

[09:46:02.0000] <csarven>
s/mistaking/mistaken

[09:47:00.0000] <jgraham>
csarven: If I didn't have to go, I would probably argue with that statement :)

[09:49:00.0000] <Philip`>
I guess the obvious advantages are that they make it possible to create unambiguous globally-unique names (using the DNS infrastructure to avoid conflicts between independent groups), that can be dereferenced to find related information about their meaning, and that can be expressed concisely in XML syntax in a way that's easy to read and write by hand but can still express arbitrarily complex arrangements of namespaced data

[09:49:01.0000] <Philip`>
Also, colons are cute

[09:50:00.0000] <zcorpan_>
/me finds http://mailman.ic.ac.uk/pipermail/xml-dev/1999-January/008189.html

[09:52:00.0000] <Philip`>
That's, um, a lot of syntax

[09:53:00.0000] <Philip`>
<?xml:begin-ns name="D" urn="http://www.daisoft.com/schema-D"?><D:baz></D:baz><?xml:end-ns name="D"?>

[09:53:01.0000] <Philip`>
vs <D:baz xmlns:D="http://www.daisoft.com/schema-D"></D:baz>

[09:54:00.0000] <Philip`>
XML Namespaces could have been much worse than it is :-)

[09:55:00.0000] <csarven>
Would I get yelled at if I started a Wiki page on 'Namespaces' which covers both pros and cons?

[09:55:01.0000] <Philip`>
Not at all

[09:58:00.0000] <gsnedders>
csarven: I think the current page is more to avoid making the same points over and over again about the conventional wisdom

[10:04:00.0000] <csarven>
gsnedders Is the decision for HTML5 based on http://wiki.whatwg.org/wiki/Namespace_confusion ? i.e., a subset of views on XML Namespaces

[10:05:00.0000] <gsnedders>
csarven: No

[10:06:00.0000] <Philip`>
There are non-confusion-related issues with Namespaces in text/html, like the inconsistency between text/html parsers and XML parsers when names contain colons, that make it less desirable too

[10:07:00.0000] <csarven>
That's good. So, how was the decision made? Is there more evidence supporting the disadvantages of NS than advantages?

[10:07:01.0000] <csarven>
I'm just curious about the data gathered for both cases

[10:09:00.0000] <zcorpan_>
/me wonders which decision is being discussed

[10:09:01.0000] <csarven>
Pardon me if I'm repeating/rehashing something which has been discussed, I don't wish to take people's time with this. Kindly point me to a URI which answers such questions.

[10:20:00.0000] <Philip`>
/me updates his RDF output to be much less wrong

[10:30:00.0000] <Philip`>
csarven: I don't think there's a URI giving the rationale

[10:30:01.0000] <zcorpan_>
Philip`: "_:n2 a <http://www.w3.org/1999/xhtml/custom#org.w3.spec> ." - what's the "a"?

[10:31:00.0000] <Philip`>
zcorpan_: Shorthand for <http://www.w3.org/1999/02/22-rdf-syntax-ns#type>

[10:31:01.0000] <zcorpan_>
ah

[10:31:02.0000] <Philip`>
Maybe rdf:type would be less confusing

[10:32:00.0000] <Philip`>
but I'm assuming it was added to N3 (and Turtle) because they wanted it to be used

[10:34:00.0000] <zcorpan_>
are there more magic identifiers in n3?

[10:35:00.0000] <Philip`>
http://www.w3.org/DesignIssues/Notation3.html

[10:36:00.0000] <Philip`>
but http://www.dajobe.org/2004/01/turtle/ (sort of a subset of N3) only has 'a' and not the others

[10:37:00.0000] <Philip`>
/me hopes those are the most proper specifications

[10:38:00.0000] <Philip`>
csarven: When you say "decision", which one do you mean? (The decision to use full URIs / reverse domain labels instead of CURIEs for the microdata property attribute?)

[10:38:01.0000] <zcorpan_>
n3 is more complicated than i had thought

[10:41:00.0000] <zcorpan_>
/me stops reading and goes on to do something different

[10:43:00.0000] <csarven>
Philip` Moreso where WHATWG stands with NS and how was that decision made i.e., based on which data? I'm only aware of http://wiki.whatwg.org/wiki/Namespace_confusion

[10:44:00.0000] <takkaria>
namespaces aren't an end in themselves, though, they're an implementation detail for certain use cases

[10:44:01.0000] <takkaria>
namespaces are explicitly used, e.g. for SVG and MathML content, the spec says those elements go in the right namespace

[10:45:00.0000] <Philip`>
csarven: I don't think the WHATWG has a specific standing on Namespaces - sometimes they're considered to be appropriate, sometimes other solutions are considered to be better

[10:59:00.0000] <csarven>
takkaria Philip` Thanks.

[11:02:00.0000] <Philip`>
(However, I have heard rumours that XML Namespaces cause babies to grow horns and were the cause of World War 2, so to stay safe I think we should avoid them)

[11:08:00.0000] <gsnedders>
/me rather dislikes uni of Edi for re-inventing email for the sake of applicants

[11:09:00.0000] <Philip`>
Maybe they didn't re-invent email, but invented it before it had actually been invented, and then stuck with their legacy system

[11:52:00.0000] <Philip`>
Hmm, the serialised JSON output can be O(2^n) in size compared to the markup

[11:52:01.0000] <Philip`>
<meta item id="n0"> <meta subject="n0" property="a b" item id="n1"> <meta subject="n1" property="a b" item id="n2"> <meta subject="n2" property="a b" item id="n3"> <meta subject="n3" property="p" content="x"> etc

[13:01:00.0000] <tantek>
Philip - the "space separated set" feature of profile, class, rel (and property in the example you gave) attributes is quite powerful, and helps with more succinct expression of semantics.

[13:02:00.0000] <tantek>
I think it's a consequence of content-centric design rather than syntax-centric (or even model-centric) design.

[13:03:00.0000] <Hish>
hi. can anybody enlighten me regarding footnotes (http://www.whatwg.org/specs/web-apps/current-work/multipage/text-level-semantics.html#footnotes)? I understand the three approaches but when it comes to editing a document with long footnotes, then doing copy/paste the part with the <a href> link won't change the footnode content. any ideas?

[13:05:00.0000] <jgraham>
/me wonders if anyone who has used rdflib is around and if so wheher they could point him to some simple instructions for creating a graph and serializing it

[13:06:00.0000] <jgraham>
Preferbly one that, unlike the officil documentation, doesn't require me to already understand the api

[13:32:00.0000] <Philip`>
jgraham: Read the source code - it's Python so it must be easy to read

[13:33:00.0000] <Philip`>
tantek: It would just be nice if it was possible to parse a page without being vulnerable to DOS attacks :-)

[13:38:00.0000] <jgraham>
Philip`: Not sure that reading the source will allow me to trivially work out how all the design elements are expected to fit together (which is basically the problem)

[13:38:01.0000] <jgraham>
Also: I just read the xml:base spec and oh boy do the authors hate conformance criteria

[13:38:02.0000] <Philip`>
jgraham: Maybe the source code contains comments giving high-level documentation of how all the design elements are expected to fit together

[13:39:00.0000] <Philip`>
although if it does, they'll probably be three years out of date and have no correspondence to the current code, and those comments won't exist anyway

[13:39:01.0000] <Philip`>
so maybe that's not the best plan

[13:40:00.0000] <Philip`>
/me decides that in general it's impossible to serialise microdata as N3 without explicit blank nodes

[13:40:01.0000] <Philip`>
(even if you use the [...] syntax whenever possible)

[13:41:00.0000] <Philip`>
because it's possible to have two predicates using the same blank node, but impossible to represent that non-explicitly, as far as I can tell

[14:05:00.0000] <krijnh>
http://www.grauw.nl/blog/entry/513

[14:08:00.0000] <Dashiva>
So the only conceivable reason to remove namespaces is to spite RDFa, obviously

[14:09:00.0000] <gsnedders>
Duh.

[14:10:00.0000] <jgraham>
Philip`: Given there are no docstrings anywhere, I have doubts about the number of comments in the code

[14:12:00.0000] <jgraham>
Hmm that is the same comment that Laurens posted at simonwillison.net. Unrelatedly, my comment there seems to have become trapped in the moderation queue. Or maybe it is a write only queue

[14:25:00.0000] <jgraham>
Does the html 5 definition of absolute url make any sense?

[14:26:00.0000] <jgraham>
An absolute url is one that resolves to itself

[14:26:01.0000] <jgraham>
But resolving a url requires something to resolve it against which is not specified

[14:31:00.0000] <Philip`>
jgraham: An absolute URL doesn't need anything to resolve it against

[14:31:01.0000] <Dashiva>
Would namespaces be less of a mess if tutorials used random strings for prefixes?

[14:32:00.0000] <Philip`>
Dashiva: No

[14:35:00.0000] <jgraham>
Philip`: Right but the algorithm doesn't allow for that

[14:35:01.0000] <Dashiva>
"And the copy and paste thing is less relevant than you all seem to imply. First of all, it would take no more than ten minutes to ensure people understand what example foaf: means."

[14:35:02.0000] <Dashiva>
Oh boy

[14:36:00.0000] <jgraham>
Dashiva: What are you quoting?

[14:37:00.0000] <Dashiva>
Fifth comment on http://realtech.burningbird.net/semantic-web/semantic-web-issues-and-practices/holding-on-html5

[14:37:01.0000] <Dashiva>
Also interesting that RDFa chooses to ignore part of the CURIE spec

[14:44:00.0000] <Philip`>
Which part is that?

[14:54:00.0000] <Dashiva>
Philip`: Without prefix


2009-05-12
[17:01:00.0000] <othermaciej>
I guess I should re-add this channel to my autojoin list

[17:03:00.0000] <Hixie>
so in #webkit othermaciej semi-accidentally suggested prop=""

[17:03:01.0000] <Hixie>
instead of property=""

[17:03:02.0000] <Hixie>
and that made me think maybe prop="" or propname="" would be good

[17:06:00.0000] <Philip`>
itemprop=""

[17:06:01.0000] <Hixie>
oooh

[17:06:02.0000] <Hixie>
itemprop="" is good

[17:12:00.0000] <Philip`>
Hixie: The RDF output has no way of identifying what's an item - you just get a load of blank nodes, some of which aren't used as objects in any triple and therefore can be assumed to be top-level items, which is quite indirect

[17:12:01.0000] <Hixie>
should we rename element.properties also if we use itemprop="" and element.itemprop?

[17:12:02.0000] <Philip`>
so maybe it'd be good to generate a triple like '<> <http://www.w3.org/1999/xhtml/something#top-level-item> _:n' for each top-level item _:n, so that the relationship between items and pages is explicit (and so e.g. it's easy to iterate over them all)

[17:13:00.0000] <othermaciej>
renaming element.properties to element.itemprops or element.itemProperties would be weird

[17:13:01.0000] <othermaciej>
because in that case, the element presumably is the item, and it's redundant to ask for the item's item properties

[17:13:02.0000] <Hixie>
ok

[17:14:00.0000] <othermaciej>
I was going to say "what other kind would it have?" but it would indeed have CSS properties

[17:14:01.0000] <othermaciej>
still, I don't think "item properties" vs "CSS properties" is a good distinction, so if disambiguation is needed, I'd say "data properties"

[17:14:02.0000] <othermaciej>
or something like that

[17:15:00.0000] <othermaciej>
but that could be confused with the data- attribute mechanism

[17:16:00.0000] <Hixie>
well i just renamed property to itemprop

[17:19:00.0000] <Hixie>
Philip`: yeah, that's not a bad idea...

[17:19:01.0000] <Philip`>
http://philip.html5.org/demos/microdata/demo.html - now with itemprop

[17:20:00.0000] <Hixie>
that was quick

[17:20:01.0000] <Hixie>
you guys rock

[17:21:00.0000] <Philip`>
"guys"? I'm only one person :-p

[17:21:01.0000] <Hixie>
you and james :-)

[17:21:02.0000] <Philip`>
/me is happy to take all of the credit anyway

[17:21:03.0000] <Hixie>
you are especially awesome, if that helps :-)

[17:24:00.0000] <Hixie>
Philip`: added an ...#item triple

[17:24:01.0000] <Hixie>
let me know what you think

[17:24:02.0000] <Hixie>
if you like it as is i'll commit it

[17:24:03.0000] <Hixie>
otherwise let me know if you'd prefer a different name

[17:25:00.0000] <Philip`>
I don't like how my code has a line saying "value = value.value;"

[17:25:01.0000] <Hixie>
hah

[17:25:02.0000] <inimino>
Hixie: s/ancestors/descendants/ in "Properties don't have to be given as ancestors of the element with the item  attribute."

[17:26:00.0000] <Hixie>
inimino: thanks, fixing

[17:26:01.0000] <Philip`>
Hixie: Is the ...#item thing visible in a place which does not require me downloading megabytes of spec?

[17:26:02.0000] <Philip`>
(since my current internet connection costs 1.5 pence per megabyte, and I don't want to just throw money away)

[17:27:00.0000] <Hixie>
i just added the following as step two in the "generate the triples for an item" algorithm:

[17:27:01.0000] <Hixie>
Generate the following triple:

[17:27:02.0000] <Hixie>
subject  the document's current address predicate  http://www.w3.org/1999/xhtml/vocab#item object  subject 

[17:27:03.0000] <Hixie>
with some more newlines

[17:27:04.0000] <Philip`>
That doesn't distinguish top-level items

[17:28:00.0000] <Hixie>
oh true

[17:28:01.0000] <Hixie>
oops

[17:28:02.0000] <Philip`>
Maybe it'd fit better in the place that currently says "(Nothing is done with the returned subject.)"

[17:28:03.0000] <Hixie>
yeah i'm putting it as a subalgorithm there

[17:28:04.0000] <Hixie>
is #item ok?

[17:29:00.0000] <Philip`>
I don't know if conflicting with <meta name=item> is fine

[17:30:00.0000] <Hixie>
and rel=item

[17:30:01.0000] <Hixie>
and explicit propitem=""s with that value

[17:30:02.0000] <Hixie>
i think it's fine, because it doesn't require implementations to do anything (like the rel thing implied it did)

[17:31:00.0000] <Hixie>
it's up to the author not to make spurious statements

[17:31:01.0000] <Hixie>
they could just as easily throw in spurious ones in to the triple store later

[17:32:00.0000] <Philip`>
I also don't know whether it's sensible to use that vocabulary space

[17:32:01.0000] <Philip`>
(Maybe it is; I've got no idea)

[17:32:02.0000] <Hixie>
no idea either

[17:33:00.0000] <Hixie>
inimino: fixed

[17:33:01.0000] <Hixie>
Philip`: ok, checking it in

[17:35:00.0000] <Hixie>
http://groups.google.com/group/thosch/browse_thread/thread/3aa1dc61c148c145/edfeb6432267af37

[17:36:00.0000] <Hixie>
i've no idea what that last paragraph means, but it's good for us...

[17:36:01.0000] <Philip`>
Hixie: s/descendents/descendants/

[17:37:00.0000] <Hixie>
man between that and "occurrence" (which i reliably spell "occurance") i really can't spell

[17:37:01.0000] <Philip`>
http://philip.html5.org/demos/microdata/demo.html - now with "<> xhv:item _:n0 ." etc

[17:39:00.0000] <Philip`>
(I really to output proper N3 blank-node syntax when possible...)

[17:39:01.0000] <Philip`>
*need

[17:41:00.0000] <Philip`>
Hixie: Are you aware that computers nowadays are capable of looking up words in dictionaries and automatically highlighting when you get it wrong?

[17:41:01.0000] <Hixie>
my dictionary says "descendents" is right

[17:41:02.0000] <Hixie>
and i don't bother with spell checking the spec because about 40% of the words are made up anyway

[17:43:00.0000] <Philip`>
http://encarta.msn.com/dictionary_1861603981/descendant.html - "Descendant is the only spelling of the noun, denoting somebody or something related to an ancestor, or something based on an earlier thing. Descendent is the usual spelling of the adjective, which means "going downward.""

[17:44:00.0000] <Hixie>
i didn't say my dictionary was right

[17:44:01.0000] <Philip`>
It's right that the word exists

[17:45:00.0000] <Hixie>
with an s on the end?

[17:46:00.0000] <Philip`>
Easy way to remember the spelling of the noun: Imagine a person whose children and grandchildren are all ants

[17:46:01.0000] <Philip`>
and they are his descendants

[17:47:00.0000] <Philip`>
Just be careful not to imagine a person whose children and grandchildren are talking trees, because that'll give entirely the wrong idea

[17:47:01.0000] <Hixie>
i know it has an a, the problem is i have motor memory now

[17:48:00.0000] <Philip`>
Hmm, I suppose a group of things which are descendent could be called descendents

[17:48:01.0000] <inimino>
Hixie: thanks

[17:48:02.0000] <Philip`>
/me decides to resolve the issue by going to bed

[17:48:03.0000] <Hixie>
nn

[17:49:00.0000] <Philip`>
jgraham: (By the way, your software is very legacy now)

[18:03:00.0000] <iand>
Hixie: if you're looking for implementation feedback on microdata, I have marked up a version of my homepage at http://iandavis.com/2009/microdata/

[18:04:00.0000] <iand>
it's based on my existing page http://iandavis.com/blog/about which uses eRDF

[18:05:00.0000] <iand>
I used property attribute, not itemprop, so I could test with http://philip.html5.org/demos/microdata/demo.html

[18:05:01.0000] <Philip`>
That thing now only accepts itemprop, not property

[18:06:00.0000] <iand>
Philip`: heh, I can't keep up

[18:06:01.0000] <iand>
hang on

[18:06:02.0000] <iand>
ok, just updating

[18:06:03.0000] <iand>
there's one thing I can't get to work satisfactorily

[18:07:00.0000] <iand>
I want to say <http://iandavis.com/id/me> <http://xmlns.com/foaf/0.1/weblog> <http://iandavis.com/blog>

[18:07:01.0000] <iand>
and <http://iandavis.com/blog> <http://purl.org/dc/terms/title> "Internet Alchemy" . 

[18:07:02.0000] <Hixie>
iand: cool, looking...

[18:08:00.0000] <iand>
but I can't get that, currently I have a blank node instead of the blog url

[18:08:01.0000] <Hixie>
(btw i based this microdata thing in part on the ideas in eRDF, it was useful)

[18:08:02.0000] <iand>
Hixie: thanks

[18:09:00.0000] <iand>
if I remove the item attibute then I get the triple <http://iandavis.com/id/me> <http://xmlns.com/foaf/0.1/weblog> <http://iandavis.com/blog>

[18:09:01.0000] <Hixie>
christ man, that page is going to be like the most microdata-irific page of the year

[18:09:02.0000] <iand>
but the title of the blog is then attatched to  <http://iandavis.com/id/me>

[18:09:03.0000] <iand>
Hixie: it was a fairly easy conversion from erdf

[18:09:04.0000] <Hixie>
cool

[18:09:05.0000] <Hixie>
ok let's see

[18:10:00.0000] <Hixie>
your first triple you can do using the under-documented "about" magic word:

[18:11:00.0000] <Hixie>
<span item><link itemprop="about" href="http://iandavis.com/id/me"><link itemprop="http://xmlns.com/foaf/0.1/weblog" href="http://iandavis.com/blog"></span>

[18:11:01.0000] <Hixie>
(gonna add more about "about" later today or tomorrow)

[18:11:02.0000] <Hixie>
oh to do both of those you'd do:

[18:11:03.0000] <iand>
yeah I think I get about, but I don't seem to be able to nest it

[18:11:04.0000] <iand>
not sure if that was just a limitation of the demo implementation

[18:12:00.0000] <Philip`>
I'm not aware of any significant ways the demo implementation diverges from the spec

[18:12:01.0000] <Hixie>
<div item><link itemprop="about" href="http://iandavis.com/id/me"><div itemprop="http://xmlns.com/foaf/0.1/weblog" item><link itemprop="about" href="http://iandavis.com/blog"><meta itemprop="http://purl.org/dc/terms/title" content="Internet Alchemy"></div></div>

[18:12:02.0000] <Hixie>
that should do it

[18:12:03.0000] <Hixie>
unless i typoed something

[18:13:00.0000] <iand>
ah, using meta

[18:15:00.0000] <Hixie>
Philip`: two bugs in your impl with that ode

[18:15:01.0000] <Hixie>
Philip`: you don't seem to support <meta>

[18:15:02.0000] <Hixie>
Philip`: and you don't remove the c:abouts

[18:15:03.0000] <Hixie>
oh that's a bug in the spec

[18:15:04.0000] <Hixie>
oops

[18:15:05.0000] <Philip`>
Works for me

[18:15:06.0000] <Philip`>
in Firefox 3.0

[18:15:07.0000] <Hixie>
oh could be a safari bug

[18:15:08.0000] <Hixie>
makes sense

[18:16:00.0000] <Philip`>
and Opera 10.0

[18:16:01.0000] <Philip`>
I haven't tested in Safari

[18:16:02.0000] <iand>
Hixie: with the solution above the title of the blog is embedded in content attribute - any way to make that work with it as visible text, i.e. for compatibility with html4 browsers?

[18:16:03.0000] <Philip`>
jQuery causes various browser-compatibility problems

[18:17:00.0000] <Hixie>
iand: sure, you <span itemprop="http://purl.org/dc/terms/title">Internet Alchemy</span> instead of the <meta>

[18:17:01.0000] <Hixie>
s/you/use/

[18:17:02.0000] <iand>
ok

[18:17:03.0000] <Philip`>
/me should replace it with pure DOM code when possible

[18:17:04.0000] <iand>
/me hacks a bit

[18:17:05.0000] <Philip`>
Hixie: Nobody ever told me to remove c:about :-)

[18:18:00.0000] <Hixie>
yeah adding that now

[18:20:00.0000] <Philip`>
Assuming you're not adding something crazy, the demo implementation does it now

[18:20:01.0000] <Hixie>
spec has it now too

[18:20:02.0000] <Hixie>
just added "If name is equal to the string "about", skip this name"

[18:20:03.0000] <Hixie>
to "For each name name in element's property names, run the following substeps:"

[18:20:04.0000] <Hixie>
as substep 1

[18:21:00.0000] <Philip`>
Ah, good, that's not crazy

[18:24:00.0000] <iand>
aha http://iandavis.com/2009/microdata/ gives the triples I want now

[18:24:01.0000] <iand>
source is URI crazy of course

[18:30:00.0000] <iand>
thanks Philip` for the microdata demo - it was very helpful

[18:33:00.0000] <Hixie>
iand: what's your opinion overall?

[18:34:00.0000] <Hixie>
(of the microdata idea)

[18:34:01.0000] <iand>
ignoring any conflict with rdfa etc, this would meet my needs

[18:34:02.0000] <iand>
I'd like an abbreviation mechanism, but I can live without it

[18:35:00.0000] <iand>
there seems to be no treatment of language?

[18:35:01.0000] <iand>
and datatypes are limited to the html5 ones

[18:35:02.0000] <iand>
but for me personally, datatypes were never strong goal in the design of erdf either

[18:35:03.0000] <Philip`>
I think the only datatypes are string and URL

[18:36:00.0000] <Philip`>
or at least that's how I've implemented it

[18:36:01.0000] <iand>
datetime as well

[18:36:02.0000] <Philip`>
That's a string, according to my code :-)

[18:36:03.0000] <iand>
:)

[18:36:04.0000] <Philip`>
/me wonders if he should handle it differently

[18:37:00.0000] <Hixie>
yeah, there's no real typing per se

[18:37:01.0000] <iand>
the JSON output doesn't distinguish between a URI in href/src and a URI in content - in RDF it makes a difference (i.e. literal vs resource)

[18:37:02.0000] <Hixie>
none of the use cases listed really needed it

[18:37:03.0000] <iand>
not sure if it's important for the JSON

[18:38:00.0000] <Hixie>
i wish i could find an abbreviation mechanism that didn't involve prefixes

[18:38:01.0000] <iand>
I guess a big area that is missing for RDFa folks will be XMLLiterals, i.e. values of triples that contain markup

[18:38:02.0000] <Hixie>
i looked at xml literals, but the use cases again didn't really need it, so i punted on it

[18:38:03.0000] <othermaciej>
if the tag representing an item property contains markup and not just text, is it flattened to text?

[18:38:04.0000] <Hixie>
othermaciej: yes

[18:39:00.0000] <Hixie>
(the goal is very much not to map rdf to html, but to solve specific problems that were raised)

[18:39:01.0000] <othermaciej>
sure, I don't dispute that goal

[18:39:02.0000] <iand>
Hixie: yes, I get that

[18:39:03.0000] <Hixie>
othermaciej: the parenthetical was to iand, not you, sorry :-)

[18:39:04.0000] <othermaciej>
I don't think being able to represent every feature of RDF is essential

[18:39:05.0000] <othermaciej>
ok

[18:40:00.0000] <othermaciej>
anyway, I think being able to map *from* microdata *to* RDF is the important direction

[18:40:01.0000] <iand>
Hixie: did you look at the erdf usage of <link rel="schema.foaf" href="http://xmlns.com/foaf/0.1/" /> for abbreviation

[18:40:02.0000] <iand>
I borrowed it from dublin core conventions

[18:41:00.0000] <Hixie>
yeah, but it's still a prefix

[18:41:01.0000] <Philip`>
You could use suffixes instead

[18:41:02.0000] <Hixie>
the problem is with separating the identifier into two parts

[18:41:03.0000] <iand>
could you allow some sort of mapping of URIs to shortnames to be declared in the document?

[18:42:00.0000] <Hixie>
whether that's prefixes, or an import mechanism, the problem still exists

[18:42:01.0000] <Hixie>
i think long term the real solution will be the use of shorter vocab terms like com.example.foo and the use of predefined vocabs

[18:42:02.0000] <Hixie>
which use short names

[18:42:03.0000] <iand>
i.e. some way to declare that itemprop="name" maps to http://xmlns.com/foaf/0.1/name

[18:43:00.0000] <Hixie>
well for certain terms we can predefine them

[18:43:01.0000] <Hixie>
but in general import mechanisms have many of the same problems as prefixes

[18:44:00.0000] <Hixie>
maybe i should register term.md or some such, and let people register terms there on a first-come-first-served basis

[18:44:01.0000] <Hixie>
md.term.foo

[18:45:00.0000] <iand>
problem is that there are millions of possible predicates... 

[18:46:00.0000] <Philip`>
Say that any name that doesn't contain a '.' should have the string "http://tinyurl.com/" prepended and then be treated as a URL

[18:46:01.0000] <Hixie>
heh

[18:46:02.0000] <Hixie>
that's basically what we do now but with the w3.org prefix

[18:46:03.0000] <Philip`>
since they've already got the unique-identifier-to-URL mapping database set up

[18:47:00.0000] <Philip`>
so you still get the benefits of resolvable identifiers

[18:47:01.0000] <Hixie>
term.md is $150

[18:47:02.0000] <Hixie>
that seems excessive

[18:47:03.0000] <Philip`>
The thing with the w3.org prefix has no mechanism for avoiding conflicts, whereas TinyURL guarantees a unique mapping

[18:48:00.0000] <Hixie>
well that's the same as getting a domain where people can register domains

[18:48:01.0000] <Hixie>
er

[18:48:02.0000] <Hixie>
terms

[18:48:03.0000] <Philip`>
Hixie: You could use that domain's name without bothering to register it

[18:49:00.0000] <iand>
I don't think RDF people will get much value out of the reverse dns shortening, com.xmlns.foaf.0.1.name isn't a lot different to http://xmlns.com/foaf/0.1/name

[18:49:01.0000] <iand>
so I woudnt do anything special for them

[18:49:02.0000] <iand>
er them includes me

[18:50:00.0000] <Philip`>
I don't fancy using identifiers like uk.ac.cam.cl.www.pjt47.desc

[18:50:01.0000] <iand>
as I said, the microdata spec works for me, barring a few wrinkles like languages

[18:51:00.0000] <Philip`>
(partly since it's awfully ugly, partly since it won't be applicable for the rest of eternity)

[18:52:00.0000] <Hixie>
iand: what's the languages wrinkle?

[18:52:01.0000] <Philip`>
(but I don't have any other domain names that are likely to be relevant for eternity (or the next decade, which is basically eternity) either)

[18:52:02.0000] <Dashiva>
Philip`: Isn't that problem the same whether you use URLs or rewerse domains?

[18:52:03.0000] <iand>
Hixie: I'd like to say the title of my blog is in french

[18:52:04.0000] <iand>
I tried adding a lang attribute on the span but it was ignored

[18:53:00.0000] <iand>
it should really create a literal object with that language 

[18:53:01.0000] <Philip`>
Dashiva: Yes, so reverse domains don't provide any benefit but they're a bit harder to write (I have to think very carefully to avoid writing ac.uk.cam or uk.cam.ac etc)

[18:53:02.0000] <iand>
<span itemprop="http://purl.org/dc/terms/title" lang="fr">Internet Alchemy</span>

[18:53:03.0000] <iand>
should give

[18:54:00.0000] <iand>
_:x <http://purl.org/dc/terms/title> "Internet Alchemy"@fr

[18:54:01.0000] <iand>
I have no idea if that's idiomatic for html5

[18:54:02.0000] <Philip`>
/me goes to bed before someone causes him to implement that language feature

[18:54:03.0000] <Hixie>
Philip`: hah

[18:54:04.0000] <Hixie>
Philip`: nn (again)

[18:55:00.0000] <Hixie>
iand: what's the magical phrase i should use to require that? Right now I say to "generate a triple" with a subject, predicate, object

[18:55:01.0000] <Hixie>
iand: do i just add "language" as one of the fields that the "triple" has?

[18:55:02.0000] <Hixie>
iand: (how does rdfa do it?)

[18:56:00.0000] <Hixie>
"if there is a value for [current language] then the value of the [plain literal] should include this language information"

[18:56:01.0000] <Hixie>
hm

[18:56:02.0000] <Hixie>
i can do that

[18:56:03.0000] <Hixie>
oh hey, a typed literal can't have a language?

[18:57:00.0000] <iand>
no

[18:57:01.0000] <iand>
they're exclusive

[18:57:02.0000] <iand>
"A plain literal is a string combined with an optional language tag"

[18:57:03.0000] <iand>
"A typed literal is a string combined with a datatype URI."

[18:57:04.0000] <Hixie>
/me tries to work out the difference between a url and a literal of some kind

[18:58:00.0000] <iand>
what do you need to know?

[18:59:00.0000] <iand>
when you generate a triple, the object is a URI ref if it has come from href/src attributes and a literal otherwise

[19:00:00.0000] <iand>
the literal may have a datatype of xsd:datetime if the datetime attribute is present

[19:00:01.0000] <iand>
otherwise it will be a plain literal with a language tag taken from the value of the enclosing lang attribute if any

[19:01:00.0000] <Dashiva>
Philip`: They have the benefit of not being URLs, and you save the http:// part too.

[19:07:00.0000] <iand>
ok, I'm off now- sleep calls

[19:07:01.0000] <Hixie>
nn

[19:07:02.0000] <Hixie>
and thanks for the help

[19:08:00.0000] <Hixie>
iand: ok, included language information

[19:08:01.0000] <Hixie>
i didn't do <time> because that would be much more complex than is reasonable

[19:09:00.0000] <Hixie>
(we'd have to check if it had a time, a date, a timezone, acting differently in each case; we'd have to reserialise, we'd have to handle invalid input cases, etc)

[19:55:00.0000] <tantek>
Hixie, what methodology are you using to choose new attribute names? (e.g. "property" vs. "prop" vs. "itemprop")

[19:56:00.0000] <tantek>
(and while you're at it, why not consider "iprop"? it's shorter, has that catchy "i-" start etc. you could even then rename your use of "typeof" to "itype" for some indication of it being related. or even "iroot" 1/2 ;)

[19:58:00.0000] <othermaciej>
iprop sounds pithy

[20:00:00.0000] <Dashiva>
"Spec things first, bikeshed later"

[20:11:00.0000] <tantek>
Dashiva - hence it helps to have a methodology (or at least principles to follow) for such things (like picking names) to avoid religious/bikeshed arguments.

[20:39:00.0000] <Dashiva>
tantek: You could also say that bikeshedding is probably going to happen anyway, for a feature like this.

[20:46:00.0000] <tantek>
Dashiva, indeed it is difficult to "avoid" 100% per se, one can only hope to reduce and perhaps somewhat minimize, especially over time.

[20:50:00.0000] <Dashiva>
Hmm, I suppose the design principles have worked like that. There are some arguments still, but they don't seem to last long.

[21:10:00.0000] <Hixie>
tantek: i use the "try to make things clear yet functional" methodology

[01:01:00.0000] <Hixie>
does anyone understand the proposal in http://www.w3.org/mid/9b3b213a0905111517o19503e0lea499ddd0547467⊙mgc ?

[01:05:00.0000] <othermaciej>
O_O

[01:05:01.0000] <othermaciej>
o_O

[01:06:00.0000] <othermaciej>
O_o

[01:07:00.0000] <Hixie>
that was my response

[01:07:01.0000] <Hixie>
glad it's not just me

[01:07:02.0000] <jgraham>
<aol>me too</aol>

[01:34:00.0000] <krijnh>
http://twitter.com/sandervdv/status/1765852544 - Hixie: you're not coming to Amsterdam? :)

[01:34:01.0000] <Hixie>
trying not to travel at all this year, so that i can get html5 done

[01:35:00.0000] <Hixie>
(i did eventually reply to sandervdv after i saw his twitter -- turns out his e-mail had looked too much like spam)

[01:35:01.0000] <Hixie>
(and i'd ignored it)

[01:38:00.0000] <krijnh>
Jammer :)

[01:49:00.0000] <Philip`>
Hixie: The new language thing shouldn't apply to 'object', but that's left out of the list

[01:51:00.0000] <Hixie>
really? i distinctly remember including <object>

[01:51:01.0000] <Hixie>
it's there

[01:51:02.0000] <Hixie>
between link and source

[01:52:00.0000] <Philip`>
Um...

[01:52:01.0000] <Philip`>
Good point

[01:53:00.0000] <Philip`>
Demo now does languages (with very simplified processing of lang attributes)

[01:53:01.0000] <Hixie>
is the processign hard?

[01:55:00.0000] <Philip`>
It would be harder if I cared about restricting it to HTML elements, or about handling xml:lang or Content-Language

[01:55:01.0000] <Philip`>
unless I'm missing something that makes it trivial

[01:55:02.0000] <Hixie>
ah ok

[01:55:03.0000] <Hixie>
unfortunately i don't think we expose the language magic to the dom yet

[02:09:00.0000] <Hixie>
hsivonen: you should coallesce any parse errors generated by an end tag having attributes into one error

[02:32:00.0000] <Philip`>
http://broadcast.oreilly.com/2009/05/the-bold-and-the-beautiful-two.html

[02:36:00.0000] <Philip`>
("Two new drafts out at W3C from the HTML 5 effort: HTML 5: The Markup Language (hat-tip Micah) and HTML 5: A vocabulary and associated APIs for HTML and XHTML ... The first one is a model of the kinds of standards-writing we need ...")

[02:41:00.0000] <Hixie>
/me comments

[02:43:00.0000] <Philip`>
It's rare for me to find a blog post about HTML5 that you haven't already commented on

[02:44:00.0000] <Hixie>
heh

[02:59:00.0000] <othermaciej>
it "make simple objective statements that can be trivially implemented" eh?

[03:03:00.0000] <Hixie>
he means implemented in a schema

[03:03:01.0000] <Hixie>
for a lot of people, that's the only meaning of implementation they use

[03:05:00.0000] <Philip`>
/me wonders if many people are just allergic to writing code

[03:30:00.0000] <zcorpan_>
http://simon.html5.org/html5-elements - now with itemprop

[03:37:00.0000] <zcorpan_>
Philip`: you could place the <title> in the textarea to make the demo work correctly

[03:39:00.0000] <Philip`>
zcorpan_: Wouldn't that make the page invalid?

[03:40:00.0000] <zcorpan_>
Philip`: yes but it would make the demo work correctly

[03:40:01.0000] <Philip`>
Also, <title> is really boring so I can't imagine many people care about testing it much :-)

[03:40:02.0000] <zcorpan_>
/me tested it

[03:41:00.0000] <zcorpan_>
Philip`: if you care about validation you can move elements around with script

[03:41:01.0000] <Philip`>
That would be a disgusting hack :-p

[03:41:02.0000] <zcorpan_>
isn't the whole thing a hack anyway?

[03:43:00.0000] <Philip`>
Yes, but not a disgusting hack

[03:47:00.0000] <Philip`>
zcorpan_: Changed it now so the <title> is in the textarea

[03:47:01.0000] <Philip`>
(which makes Firefox not use it as the title for the page)

[03:56:00.0000] <zcorpan_>
Philip`: file a bug :)

[04:05:00.0000] <zcorpan_>
"s2.2 I don't see what the need for gratuitously departing from SGML and XML is, allowing <!doctype html> rather than <!DOCTYPE html>." - isn't sgml case insensitive there?

[04:08:00.0000] <Philip`>
zcorpan_: I find it hard to care sufficiently to file a bug :-(

[04:09:00.0000] <zcorpan_>
Philip`: you could bug hsivonen to fix it while he's touching all other code related to the parser in some way

[04:10:00.0000] <zcorpan_>
Philip`: (i think there is magic glue between the parser and document.title in gecko)

[04:19:00.0000] <Philip`>
Is Opera Turbo meant to actually work?

[04:19:01.0000] <Philip`>
As far as I can see, it only works when I've got it switched off and am on a fast connection

[04:21:00.0000] <virtuelv>
young van kesteren's new name is apparently "Ann": http://www.oreillynet.com/xml/blog/2007/02/ecs_moving_from_wellformed_xml.html

[04:22:00.0000] <virtuelv>
Philip`: yes, it's supposed to work (and as far as I've been able to determine, it does, but it's still in alpha)

[04:23:00.0000] <virtuelv>
(And, even from Opera's rather fat pipes here, I'm seeing significant speedups on a site like http://db.no/

[04:23:01.0000] <virtuelv>
)

[04:23:02.0000] <Philip`>
virtuelv: I can tell when it's active because all the images turn into ugly JPEGs :-)

[04:25:00.0000] <Philip`>
It seems to be working now (when I'm on a wireless LAN), but I don't think I've been able to get it to do anything at all when I'm on a dialup (mobile broadband) connection

[04:25:01.0000] <Philip`>
/me will have to experiment more

[04:30:00.0000] <zcorpan_>
ECS doesn't seem to support the kind of markup people use on the web - nested divs

[04:31:00.0000] <zcorpan_>
it also doesn't whine when you make this mistake: <p>Foo <code>bar<code> baz.</p>

[05:20:00.0000] <Hixie>
hsivonen is going to hate me

[05:20:01.0000] <Hixie>
my next checkin introduces only about a bazillion conformance criteria

[05:20:02.0000] <Hixie>
(wip at http://www.whatwg.org/specs/web-apps/current-work/#md-vcard)

[05:21:00.0000] <Hixie>
ok i'll continue this tomorrow

[05:21:01.0000] <Hixie>
nn

[05:24:00.0000] <Philip`>
hsivonen should like having bazillions of conformance criteria - it significantly raises the barriers to entry for competing conformance checkers, thus ensuring he retains market dominance

[05:27:00.0000] <Philip`>
/me hopes predefined vocabularies will be moved out of the main spec document

[07:14:00.0000] <hsivonen>
Hmm. Why does N-Triples force everything to be ASCII instead of being UTF-8?

[07:14:01.0000] <hsivonen>
it seems annoying to have to escape all non-ASCII

[07:15:00.0000] <Philip`>
Does anyone use N-Triples?

[07:16:00.0000] <Philip`>
http://www.w3.org/DesignIssues/Notation3 says "N3 files are encoded in UTF-8 (See RFC2279), in normalized in Normalization Form C."

[07:16:01.0000] <Philip`>
http://www.dajobe.org/2004/01/turtle/ says "A Turtle document is a Unicode[UNICODE] character string encoded in UTF-8. Unicode codepoints only in the range U+0 to U+10FFFF inclusive are allowed."

[07:16:02.0000] <hsivonen>
I don't know, but escape for the requirement to escape non-ASCII, N-Triples seems like it's the cleanest format for expressing RDF

[07:17:00.0000] <hsivonen>
I wonder how one could produce a Unicode codepoint that isn't in that range

[07:19:00.0000] <jgraham>
hsivonen: Presumably using some higher version of the unicode standard that allows more codepoints

[07:19:01.0000] <hsivonen>
jgraham: ok

[07:20:00.0000] <hsivonen>
I was thinking of implementing the microdata RDF conversion by having Java model classes for microdata and making them expose Jena RDF interfaces, but the Jena RDF interfaces have way too much stuff to implement

[07:20:01.0000] <Philip`>
/me wonders how you represent literals' languages in N-Triples

[07:20:02.0000] <hsivonen>
Philip`: langString  ::=  '"' string  '"' ( '@' language  )?

[07:22:00.0000] <Philip`>
Oh, right, I was looking at http://www.w3.org/2001/sw/RDFCore/ntriples/

[07:28:00.0000] <Philip`>
Hmm, if I click a fragment link in a page then the subjects of all the triples change

[07:28:01.0000] <Philip`>
(because they're set to the document's current address)

[07:36:00.0000] <hsivonen>
Also, it seems that an absolute URL is always ASCII per spec

[07:36:01.0000] <hsivonen>
so it seems IRIs that aren't also URIs can't be used as property names

[07:39:00.0000] <zcorpan_>
hsivonen: it could if they are ASCIIfied in the RDF output algorithm

[07:40:00.0000] <hsivonen>
zcorpan_: but RDF allows IRI-ish property names, right?

[07:42:00.0000] <zcorpan_>
hsivonen: dunno, i thought you were referring to the RDF spec

[07:42:01.0000] <hsivonen>
zcorpan_: I was referring to HTML5 microdata

[07:43:00.0000] <hsivonen>
zcorpan_: HTML5 defines 'absolute URL' via resolving the URL

[07:43:01.0000] <zcorpan_>
ok

[07:43:02.0000] <hsivonen>
zcorpan_: and resolving seems to URIfy 

[07:44:00.0000] <zcorpan_>
does RDF make a difference between property names that use percent escaping and those that don't?

[07:45:00.0000] <hsivonen>
zcorpan_: my understanding is that they are different

[07:45:01.0000] <hsivonen>
zcorpan_: but I'm not an RDF expert

[07:46:00.0000] <hsivonen>
zcorpan_: IIRC, they are like XML namespaces: compared as strings

[07:46:01.0000] <zcorpan_>
hsivonen: btw, the n3 syntax allows \UHHHHHHHH escapes for unicode code points

[07:46:02.0000] <zcorpan_>
hsivonen: ok

[07:46:03.0000] <hsivonen>
zcorpan_: yes, but it's human-unfriendly if the text is mostly non-ASCII

[07:47:00.0000] <hsivonen>
(assuming a text editor--not a dedicated RDF editor)

[07:47:01.0000] <zcorpan_>
hsivonen: it was more re how to use characters above U+10FFFF

[07:47:02.0000] <hsivonen>
oh

[07:47:03.0000] <hsivonen>
but those aren't Unicode code points by definition :-)

[07:48:00.0000] <zcorpan_>
true

[07:49:00.0000] <zcorpan_>
still seems weird that it has eight instead of six "H"es

[07:49:01.0000] <Philip`>
It's just copying Python

[07:50:00.0000] <zcorpan_>
/me didn't know python had the same weirdness

[07:52:00.0000] <Philip`>
I might be wrong

[07:53:00.0000] <Philip`>
Hmm, maybe I'm not; \Uxxxxxxxx works in Python 2.5

[07:54:00.0000] <Philip`>
/me can't find when it was introduced

[07:56:00.0000] <hsivonen>
Hixie: my photostream on Flickr and set views are examples of pages that have multiple media items with different licenses

[07:56:01.0000] <hsivonen>
(my photos are either ARR or CC-by on a case-by-case basis)

[08:12:00.0000] <Philip`>
http://philip.html5.org/demos/microdata/demo.html - now with fancier and more readable (and hopefully syntactically valid and representing the correct triples) N3 output

[08:12:01.0000] <hsivonen>
I wonder what charter this artifact is being produced and published under: http://www.w3.org/MarkUp/DTD/html4-rdfa-1.dtd

[08:14:00.0000] <hsivonen>
Philip`: is there a way to feed content to your script using <textarea>?

[08:15:00.0000] <Philip`>
hsivonen: Uh... Copy-and-paste?

[08:15:01.0000] <Philip`>
Maybe I'm not sure what you mean

[08:15:02.0000] <hsivonen>
ooh. I see the textarea now

[08:15:03.0000] <jgraham>
Philip`: Your UI is not that obvious :)

[08:15:04.0000] <hsivonen>
the layout of the page is so cluttered that I missed it

[08:15:05.0000] <Philip`>
jgraham: Very true :-)

[08:15:06.0000] <jgraham>
Philip`: Especially in Firefox where the json doesn't wrap

[08:16:00.0000] <Philip`>
Suggestions and/or implementations of improved UIs would be welcome

[08:17:00.0000] <hsivonen>
Philip`: it would help if the JSON output didn't overlap the input

[08:18:00.0000] <Philip`>
It doesn't, when I look at it in Firefox 3.0

[08:18:01.0000] <Philip`>
Is your screen 320 pixels wide?

[08:18:02.0000] <hsivonen>
1650 px wide window

[08:19:00.0000] <Philip`>
How is the JSON wide enough to reach the input box? It should only be a few dozen characters wide...

[08:19:01.0000] <hsivonen>
layout code checked out of trunk a bit over a week ago

[08:20:00.0000] <jgraham>
Philip`: I guess the JSON in firefox 3.5 no longer has line breaks

[08:20:01.0000] <hsivonen>
Philip`: it's all on one line

[08:20:02.0000] <Philip`>
Oh

[08:20:03.0000] <Philip`>
That's not good

[08:20:04.0000] <Philip`>
Does it have a native implementation of window.JSON that differs from the scripted one?

[08:21:00.0000] <jgraham>
Philip`: Yes

[08:21:01.0000] <Philip`>
That's not good

[08:22:00.0000] <jgraham>
Philip`: Yes it is. Just not in this case

[08:23:00.0000] <Philip`>
So much for "don't break the web"

[08:23:01.0000] <Philip`>
/me is just using the standard JSON2.js

[08:24:00.0000] <Philip`>
Hmm, looks like ES3.1 specifies something that does support indentation

[08:24:01.0000] <Philip`>
Maybe FF3.5 just hasn't implemented it?

[08:25:00.0000] <Philip`>
Uploaded a new version that uses the non-native JSON implementation

[08:30:00.0000] <hsivonen>
Philip`: thanks

[08:31:00.0000] <hsivonen>
Philip`: your impl doesn't deal with <time> elements that have the datetime in element content

[08:31:01.0000] <Philip`>
"If the element is a time element with a datetime attribute: The value is the value of the element's datetime attribute."

[08:31:02.0000] <Philip`>
I think I'm doing what the spec says

[08:32:00.0000] <hsivonen>
oh. 

[08:32:01.0000] <Philip`>
Maybe I'm not

[08:32:02.0000] <Philip`>
since I forgot the "with a datetime attribute" bit

[08:34:00.0000] <Philip`>
hsivonen: Fixed

[08:34:01.0000] <hsivonen>
thanks

[08:35:00.0000] <Philip`>
This N3 output is actually surprisingly readable now

[08:35:01.0000] <Philip`>
particularly for input like http://iandavis.com/2009/microdata/

[08:37:00.0000] <jgraham>
Philip`: You have rather low standards for what is readable :)

[08:40:00.0000] <Philip`>
jgraham: I'm comparing it to the old non-nested version that was full of explicit blank nodes and a seemingly random ordering of triples, and I think it's a significant step up from that :-)

[08:45:00.0000] <jgraham>
Philip`: Oh well I aagree that it may not be the most unreadable syntax it is possible to imagine

[08:45:01.0000] <jgraham>
But the whole concept of using URIs to identify things other than webpages torpedos readability from the start

[08:47:00.0000] <Philip`>
/me can't easily imagine a syntax that would be much more readable, expressing the same data

[08:49:00.0000] <jgraham>
Philip`: Assuming the same underlying concepts, macro expansions would make it more readable like def:ian davis=><http://iandavis.com/id/me> 

[08:50:00.0000] <jgraham>
Ian Davis a <http://xmlns.com/foaf/0.1/Person> 

[08:50:01.0000] <jgraham>
(obviously you would need more syntax somehow to distinguish macros from other things)

[08:51:00.0000] <jgraham>
Like ref:(Ian Davis) a <http://xmlns.com/foaf/0.1/Person> 

[08:52:00.0000] <jgraham>
(and this would be bad for other reasons like copy and paste)

[09:02:00.0000] <Philip`>
jgraham: That should be rdf:type rather than a, based on the current implementation :-)

[09:02:01.0000] <Philip`>
/me wonders why they didn't use a keyword like "isa" so that it would actually make some intuitive sense

[09:03:00.0000] <Philip`>
/me wonders if jgraham is looking at an old version of the output...

[09:05:00.0000] <jgraham>
Philip`: Oh, I see. The page had the old output at the bottom, so I was looking at that instead of the new output further down the page

[09:05:01.0000] <jgraham>
s/The page/The page with the embedded microdata/

[09:06:00.0000] <jgraham>
I guess the new version is a bit more readable

[09:06:01.0000] <Philip`>
Ah

[09:08:00.0000] <Philip`>
Evidence suggests my UI is rubbish :-(

[09:08:01.0000] <Philip`>
which is not too surprising since my entire UI design basically consists of style="width: 50%; float: right"

[09:09:00.0000] <hsivonen>
I wonder if the current microdata spec is too loose when it comes to ensuring that microdata can be mapped to RDF/XML

[09:09:01.0000] <hsivonen>
specifically, I don't see requirements of certain strings being limited to NCNames

[09:16:00.0000] <Philip`>
hsivonen: What strings need to be limited to that?

[09:17:00.0000] <Philip`>
/me thought everything could just be expressed as an unabbreviated URI in an attribute value

[09:17:01.0000] <hsivonen>
Philip`: oh. I thought properties needed to become qnames in RDF/XML

[09:18:00.0000] <Philip`>
hsivonen: Maybe they do; I know nothing about RDF

[09:22:00.0000] <Philip`>
hsivonen: (Even if that's the case, I don't see why it ought to be restricted to the subset that's mappable to RDF/XML)

[09:24:00.0000] <Philip`>
hsivonen: (Presumably RDF people don't like being restricted to that subset, since N3 and Turtle are intentionally and explicitly more expressive than RDF/XML)

[09:24:01.0000] <hsivonen>
ah, OK

[09:25:00.0000] <hsivonen>
I find that surprising

[09:26:00.0000] <Philip`>
(http://www.dajobe.org/2004/01/turtle/ - "The recommended XML syntax for RDF, RDF/XML ([RDF-XML]) has certain restrictions imposed by XML and the use of XML Namespaces that prevent it encoding all RDF graphs (some predicate URIs are forbidden and XML 1.0 forbids encoding some Unicode codepoints). These restrictions do not apply to Turtle.")

[09:26:01.0000] <Philip`>
(which makes it sound like it's a feature of Turtle rather than a bug or a hazard)

[09:28:00.0000] <hsivonen>
Hixie: it might be a good idea to note that the RDF graphs produces by the algorithm may not be serializable as RDF/XML

[09:28:01.0000] <Philip`>
They might not be serialisable as N3/Turtle either

[09:28:02.0000] <Philip`>
e.g. if you have lang="foo bar"

[09:29:00.0000] <Philip`>
or if you have itemprop=">"

[09:34:00.0000] <Philip`>
/me wonders why so much time has been spent discussing the RDF extraction of microdata, and very little about the JSON

[09:36:00.0000] <hsivonen>
I wonder what "html:abbr" is about in http://mxr.mozilla.org/mozilla-central/source/toolkit/components/microformats/src/Microformats.js#345

[09:43:00.0000] <Philip`>
hsivonen: http://www.codingforums.com/archive/index.php/t-49227.html ?

[09:43:01.0000] <Philip`>
(several posts down)

[09:47:00.0000] <hsivonen>
Philip`: html:abbr as an IE workaround? that's sad

[09:50:00.0000] <Philip`>
Doesn't seem to be a commonly used pattern in practice, but maybe whoever was writing that code liked it or saw a site that relied on it

[10:02:00.0000] <Philip`>
http://waffle.wootest.net/2009/05/11/rdf-2/ - "The Microdata section of the HTML5 draft might be one of its finest moments."

[10:07:00.0000] <jgraham>
"""I'm starting to tune out the HTML 5 meta-discussions [...] the emo-drama that people insist on manufacturing… “is such a buzz-kill."""

[10:08:00.0000] <Philip`>
Emu drama?

[10:08:01.0000] <Philip`>
Oh

[10:11:00.0000] <jgraham>
Philip`: BTW, I expect the lack of discussion about the JSON format is because JSON is more interesting to people who just do things rather than people who architecture things

[10:12:00.0000] <Philip`>
Shouldn't there at least be someone complaining that 'item.properties.foo[0]' is much harder to type than 'item.foo' and in most cases will be equivalent?

[10:12:01.0000] <hsivonen>
jgraham: where's 'emo-drama' quoted from?

[10:12:02.0000] <Philip`>
hsivonen: http://shawn.medero.net/

[11:23:00.0000] <tantek>
hsivonen, regarding "I wonder what html:abbr is about in Microformats.js" - see http://microformats.org/wiki/hcard-parsing#all_properties - and I invite you to #microformats for any additional comments.

[11:30:00.0000] <Philip`>
tantek: That seems to just talk about the normal element named "abbr", not the crazy element named "html:abbr"

[11:31:00.0000] <tantek>
Philp - is that some sort of oblique namespaces joke? I don't quite understand.

[11:33:00.0000] <Philip`>
tantek: No - it's about the element with localName == "html:abbr"

[11:34:00.0000] <Philip`>
(which is impossible to get when you parse XML, but is what you get if you parse "<html:abbr>" in a text/html parser that's not IE)

[12:13:00.0000] <Groovy>
since when google.com uses <!doctype html> ?

[12:32:00.0000] <Philip`>
Groovy: Since several months ago, at least

[12:33:00.0000] <Philip`>
Groovy: (though only on a few of their pages)

[12:56:00.0000] <Wolfman2000>
Afternoon.  I'm trying to make some sense of the new HTML5 tags, and I want to be sure I understand their semantics.  <nav> vs. <menu>: it seems that both are to be used for navigation links...and yet only <menu> directly accepts <li>.  What is the proper way of using those tags?

[13:10:00.0000] <jgraham>
Wolfman2000: You probably want <nav>

[13:11:00.0000] <jgraham>
<menu> is designed for building application menus but it isn't really supported yet

[13:11:01.0000] <jgraham>
/me goes to check he isn't talking rubbish

[13:12:00.0000] <gsnedders>
nav is for primary navigation, menu is for something that I've never needed and thus can't remember what it is for

[13:12:01.0000] <Wolfman2000>
The next question then: <article> vs <section>.  It seems that there are many sections per article, btu there can be many articles per page.  Is an article supposed to represent an entire page, or just the content usually between <header> and <footer>?  (I can rephrase if needed)

[13:14:00.0000] <jgraham>
Wolfman2000: You can have multiple <article>s per page

[13:15:00.0000] <jgraham>
e.g. <article><h1>My first news item</h1></article> <article><h1>My second news item</h1></article>

[13:15:01.0000] <Wolfman2000>
...and what if the items are not really...well, news/blog related?

[13:15:02.0000] <jgraham>
No problem

[13:16:00.0000] <Wolfman2000>
My original understanding: using a two column layout, <header> at top, <article> for main column, <nav>/<menu> for navigation on secondary column, <footer> on bottom.  Is this still valid thinking?

[13:16:01.0000] <jgraham>
That seems reasonable

[13:17:00.0000] <Wolfman2000>
...long as I'm not required to have <header> and <footer> in <article>/<section>, I may be able to make something work.

[13:17:01.0000] <Wolfman2000>
I'm currently redesigning one of my websites, and I've been convinced to at least look into HTML5.  It seems to reduce my need for IDs at least.

[13:21:00.0000] <Philip`>
http://google.com/support/webmasters/bin/answer.py?answer=146898 - someone should tell them about <hX> elements instead of <strong>

[13:22:00.0000] <Philip`>
Aren't their CURIEs resulting in URIs like "http://rdf.data-vocabulary.orgreview"?

[13:24:00.0000] <gsnedders>
Philip`: I suggest Philip`.

[13:25:00.0000] <Philip`>
gsnedders: Huh?

[13:25:01.0000] <gsnedders>
Philip`: To tell them about hX

[13:25:02.0000] <jgraham>
Philip`: If only there were a google employee around who could file a bug

[13:25:03.0000] <Philip`>
gsnedders: They don't seem to provide any way to provide feedback

[13:25:04.0000] <Philip`>
so complaining on IRC is the most effective thing to do

[13:37:00.0000] <gsnedders>
jgraham: I guess a polarizing filter darkens the sky because the light in unpolarized and hence polarizing it makes it less intense?

[13:41:00.0000] <Philip`>
Hah

[13:41:01.0000] <Philip`>
One example says xmlns:v="http://rdf.data-vocabulary.org/"

[13:41:02.0000] <jgraham>
gsnedders: Well that is true

[13:41:03.0000] <Philip`>
All the others say xmlns:v="http://rdf.data-vocabulary.org"

[13:42:00.0000] <jgraham>
But it's not really the relevant fact

[13:43:00.0000] <jgraham>
gsnedders: The point is that the light from the sky is partially polarised (because it is scattered light from the atmosphere and scattering polarises light in a way that depends on the angle of scattering)

[13:43:01.0000] <Philip`>
<spanspan property="v:locality">Springfield</span>

[13:43:02.0000] <Philip`>
I think they need to do a bit of QA on these help pages

[13:44:00.0000] <jgraham>
gsnedders: And as a consequence, with the polarizer set to the right angle you can eliminate almost all of the scattering from the sky

[13:44:01.0000] <jgraham>
s/scattering/light/

[13:44:02.0000] <gsnedders>
jgraham: ah

[13:44:03.0000] <jgraham>
(or t least from those regions of the sky that are heavilly polarised)

[13:45:00.0000] <jgraham>
gsnedders: Is this a physics question or a photography question?

[13:45:01.0000] <gsnedders>
jgraham: a physics of photography question :)

[13:45:02.0000] <jgraham>
Philip`: But evryone knows that http://example.com/ === http://example.com

[13:45:03.0000] <gsnedders>
jgraham: (It's random curiosity, really, but from a physics POV)

[13:46:00.0000] <jgraham>
gsnedders: Oh that's OK. If it's curiosity I don't need to worry to much about being right

[13:46:01.0000] <Philip`>
jgraham: I do hope Google's implementation doesn't just look for the prefix "v:"

[13:46:02.0000] <gsnedders>
jgraham: Apart from generic URI parsers that don't know about what RFC2616 defines on-top about scheme specific normalization

[13:46:03.0000] <gsnedders>
jgraham: :P

[13:47:00.0000] <jgraham>
gsnedders: "parsers" fail to meet the definition of "everybody" since "body" here specifically means "human body"

[13:47:01.0000] <jgraham>
Philip`: Didn't Hixie say that the Google implementation of some microdata stuff just used regexps

[13:48:00.0000] <Wolfman2000>
...if only the Tidy HTML Validator extension I had recognized HTML5, it would make things easier...

[13:48:01.0000] <gsnedders>
jgraham: yes

[13:48:02.0000] <jgraham>
Wolfman2000: http://validator.nu

[13:48:03.0000] <Wolfman2000>
jgraham: the page I'm working on is not live yet

[13:49:00.0000] <jgraham>
http://html5.validator.nu/ is better actually

[13:49:01.0000] <jgraham>
Wolfman2000: Either change the "Address" dropdown to "textarea"

[13:49:02.0000] <jgraham>
or go through some pain to get a local install

[13:49:03.0000] <Wolfman2000>
...gotta love javascript powered fields

[13:50:00.0000] <Philip`>
jgraham: No

[13:50:01.0000] <Philip`>
jgraham: "Google's handling of RDF blocks for license declarations is all done with regular expressions instead of actually parsing the namespaces"

[13:53:00.0000] <jgraham>
Philip`: That sounds quite close if you s/microdata/license declarations/

[13:56:00.0000] <Philip`>
jgraham: For some reason I thought you said "RDFa" in your statement

[13:56:01.0000] <Philip`>
but the evidence indicates that you didn't

[14:02:00.0000] <Wolfman2000>
...and with the help of the new HTML5 tags (and some abuse of child selectors), I've trimmed out the <divs> and id=s from my main page.  The other pages...will take some time.

[14:03:00.0000] <Philip`>
Wolfman2000: Has everything broken awfully in IE? :-)

[14:03:01.0000] <Wolfman2000>
Philip`: javascript workaround. :P

[14:04:00.0000] <Wolfman2000>
honestly, I don't really care about IE

[14:06:00.0000] <Philip`>
(Looking at http://rdf.data-vocabulary.org/rdf.xml, I guess xmlns:v="http://rdf.data-vocabulary.org" and xmlns:v="http://rdf.data-vocabulary.org/" are *both* wrong, and it ought to be xmlns:v="http://rdf.data-vocabulary.org/#")

[14:07:00.0000] <Wolfman2000>
...right, form elements should keep their IDs.  That's fine.

[14:07:01.0000] <Philip`>
(but presumably Google can add some hacks in their parser to make it all work)

[14:10:00.0000] <Hixie>
hsivonen: yeah but all your photos have their own pages, so the photostreams and set views don't have to be machine-readable

[14:13:00.0000] <Wolfman2000>
I'd like confirmation via this picture.  http://img509.imageshack.us/my.php?image=picture1kpe.png  Left sidebar uses <nav> & <ul> for its links.  Center, horizontal links use <ul> by itself, no <nav>.  Am I using the tags properly?

[15:12:00.0000] <Hixie>
anyone remember where Philip`'s parser state diagram is?

[15:16:00.0000] <Dashiva>
I don't know if it's the most recent, http://canvex.lazyilluminati.com/misc/states10.png

[15:20:00.0000] <Hixie>
Dashiva: sweet, thanks

[15:22:00.0000] <Hixie>
oh hey, the very next comment where i needed those for was philip giving links to them

[15:22:01.0000] <Dashiva>
haha

[15:25:00.0000] <Hixie>
so when yahoo came out with their "rdfa support", some people on the rdfa list quickly poked holes in it showing it wasn't exactly following the specs

[15:26:00.0000] <Hixie>
i wonder how long it'll be before someone tries to test google's rdfa support

[15:35:00.0000] <roc>
not in SVG format? Shame!

[15:35:01.0000] <Hixie>
the one philip linked to was

[15:52:00.0000] <Hixie>
i like how shelley complains that the discussion on microdata is split between whatwg and public-html

[15:53:00.0000] <Hixie>
given that she was the one who split it

[15:53:01.0000] <Hixie>
(it was all on whatwg until she cross-posted it to public-html)

[15:59:00.0000] <KevinMarks>
@hixie, well it didn't take long: http://iandavis.com/blog/2009/05/googles-rdfa-a-damp-squib

[16:02:00.0000] <Philip`>
Today I have learnt that Gmail's basic HTML view completely ignores one's attempts to have multiple sender addresses

[16:02:01.0000] <Philip`>
/me switches to the expensive view and resends the message that got bounced...

[16:03:00.0000] <othermaciej>
Google inventing their own scheme is a demonstration of the main problem with RDF

[16:03:01.0000] <othermaciej>
it's a mechanism with no standard vocabulary, and tools won't really be able to do useful things with your data unless you agree on a vocabulary

[16:03:02.0000] <othermaciej>
thus defeating the goal of decentralized extensibility

[16:03:03.0000] <KevinMarks>
I thought that was the main advantage of RDF? That's what I'm told when I talk about microformats and agreement being key

[16:04:00.0000] <KevinMarks>
parsing is easy, agreement is hard

[16:05:00.0000] <othermaciej>
RDF doesn't require any agreement on vocabulary so long as you don't need to communicate information

[16:06:00.0000] <Philip`>
Hixie: It's hard to poke holes in Google's RDFa support when it seemingly provides no way to test it, other than owning a site which Google's algorithms and policies deem appropriate to be parsed

[16:07:00.0000] <roc>
It's useful to support multiparty agreements on vocabulary without going through a central registry

[16:07:01.0000] <roc>
*how* useful is unclear

[16:08:00.0000] <gsnedders>
KevinMarks: The problem is to communicate anything you need agreement. It's the hard part you need to get right, the easy part is less important. RDF gets the priority the wrong way around, IMO.

[16:09:00.0000] <jwalden>
lazyirc: what's the URL of the websockets spec?

[16:10:00.0000] <roc>
is it my imagination or is Ian Davis saying "They’ve ... thrown away the best [part of RDFa] – the decentralized vocabularies of terms." ... when his real problem is that they didn't use FOAF, i.e., that they *exploited* decentralization?

[16:10:01.0000] <KevinMarks>
gsnedders: exactly. Thats the point of the microformats process, to converge agreement. 

[16:10:02.0000] <gsnedders>
jwalden: http://lmgtfy.com/?q=websockets+api&l=1

[16:11:00.0000] <gsnedders>
roc: Yeah, I saw that too.

[16:11:01.0000] <jwalden>
gsnedders: there's only a w3.org version?  no whatwg version that's latest-svn?

[16:11:02.0000] <gsnedders>
jwalden: No whatwg version

[16:11:03.0000] <othermaciej>
gsnedders: that site is awesome

[16:11:04.0000] <jwalden>
boo-urns

[16:12:00.0000] <gsnedders>
othermaciej: Fun fact: I had to Google to find its URL :)

[16:12:01.0000] <Philip`>
roc: They aren't exploiting decentralisation, because they're centralising everything to data-vocabulary.org

[16:13:00.0000] <roc>
I suppose that's one way of looking at it

[16:14:00.0000] <Philip`>
They're running everything in their own little walled garden, where you use Google's vocabularies to interact with Google's tools

[16:14:01.0000] <Philip`>
(Well, maybe not "little", more like "gigantic", but the point is otherwise similar)

[16:15:00.0000] <roc>
oh, Ian Davis is also suggesting that it's a deliberate attempt by Google to destroy RDFa

[16:15:01.0000] <Philip`>
Decentralisation would mean anybody could make their own vocabularies and have them interact usefully with Google's tools, I guess

[16:16:00.0000] <Philip`>
(e.g. I could define my own vocabulary that's a subclass of one Google supports, and then they'd parse the RDF that describes my vocabulary and support it too)

[16:16:01.0000] <Philip`>
(perhaps)

[16:16:02.0000] <Philip`>
(though actually I know basically nothing about RDF, so this might be entirely wrong)

[16:17:00.0000] <KevinMarks>
how? how on earth would they support displaying it in snippets if they don't know what you mean by it?

[16:19:00.0000] <Philip`>
KevinMarks: You can define your vocabulary to be a subclass of theirs, and then they can treat your data as if it was in their vocabulary

[16:19:01.0000] <Philip`>
I'm assuming that's kind of the point of using RDF to describe RDF vocabularies

[16:20:00.0000] <gsnedders>
Fun fact: the W3C TR list RDF didn't conform to its own RDFs

[16:20:01.0000] <KevinMarks>
this is the point in the conversation where I start thinking that RDF actually stands for Recursive Descent Forever

[16:26:00.0000] <Hixie>
Philip`: i assure you that whatever else we might be doing, we're not fetching the rdf vocab definitions and doing things with subclassing

[16:29:00.0000] <Philip`>
Surely it's not that hard - you can have "<#me> rdf:type :Man" and ":Man rdfs:subClassOf :Person" and then add some more triples saying that x rdf:type y and y rdfs:subClassOf z implies x rdf:type z

[16:29:01.0000] <Philip`>
and then you dump it all into your triple store and crank the handle and an answer comes out

[16:30:00.0000] <Philip`>
I just made up all that up so it's probably not true but that's how I'd do if I wasn't constrained by sanity

[16:31:00.0000] <Philip`>
Hixie: Indeed, so it just seems like a complex syntactic alternative to microformats and a bit pointless

[16:31:01.0000] <roc>
is http://www.whatwg.org/specs/web-apps/current-work/ loading for anyone else?

[16:31:02.0000] <Hixie>
wfm

[16:33:00.0000] <gsnedders>
et moi

[16:34:00.0000] <roc>
sigh

[16:35:00.0000] <Hixie>
/me is confused as to what the point of the "label" type in vcard is

[16:36:00.0000] <KevinMarks>
label is unstructured address

[16:38:00.0000] <Hixie>
right but what's the point?

[16:39:00.0000] <Hixie>
interestingly in norm's vcard ontology, he puts the label inside the address

[16:39:01.0000] <KevinMarks>
maybe you don't have a structured address, just a string

[16:40:00.0000] <KevinMarks>
thats how we mapped it into PortableContacts

[16:40:01.0000] <KevinMarks>
the original spec talks about using it to print an address label

[16:41:00.0000] <KevinMarks>
"This property specifies the addressing label for physical delivery to the person/object associated with the vCard. The property is intended to include the information necessary to create a formatted delivery address label."

[16:42:00.0000] <KevinMarks>
and has CR/LF's in it

[16:42:01.0000] <Philip`>
(Are RDFa property attributes case-sensitive? If so, the Google help pages got that wrong too ("dtreviewed" vs "dtReviewed"))


2009-05-13
[17:13:00.0000] <Hixie>
i wonder where kevin quoted that from

[17:14:00.0000] <Hixie>
doesn't seem to be in rfc2426

[17:15:00.0000] <Philip`>
Hixie: There's this clever new web site known as a "search engine", which lets you answer questions like that

[17:16:00.0000] <Philip`>
e.g. http://www.imc.org/pdi/vcard-21.txt

[17:17:00.0000] <Hixie>
i tried that

[17:17:01.0000] <Philip`>
(That looks like a bad conversion of http://www.imagescript.com/atg/resources/vcard-21.doc)

[17:17:02.0000] <Hixie>
before asking :-)

[17:17:03.0000] <Hixie>
apparently the string i used was broken in some way

[17:18:00.0000] <Hixie>
weird

[17:19:00.0000] <Philip`>
A likely excuse

[17:22:00.0000] <Hixie>
i wonder why that text got lost in the conversion to rfc

[18:32:00.0000] <Hixie>
wtf, vcard contains the lat long position 37.386013;-122.082932

[18:32:01.0000] <Hixie>
which when i look it up

[18:32:02.0000] <Hixie>
turns out to be within cycling distance of where i'm sitting

[18:32:03.0000] <Hixie>
freaky

[18:49:00.0000] <franksalim>
that's 0.7 miles from me

[19:13:00.0000] <karlcow>
[18:07]  <othermaciej> Google inventing their own scheme is a demonstration of the main problem with RDF

[19:13:01.0000] <karlcow>
not at all. The problem here is not RDF but power. Small communities can still invent their own vocabularies and that is good. Google here is creating a new vocabulary AND they have the power to impose it. 

[19:17:00.0000] <Hixie>
ok, author conformance criteria for the vCard vocabulary are now defined: http://www.whatwg.org/specs/web-apps/current-work/#vcard

[19:17:01.0000] <Hixie>
(it's basically the same as hCard, but defined in detail)

[19:22:00.0000] <heycam>
"The item with the type vcard has an fn  property value consists of a string with exactly one sequence of space characters, and it neither at the immediate start nor the immediate end of the string."

[19:22:01.0000] <heycam>
missing a word (maybe "occurs")

[19:26:00.0000] <Hixie>
thanks

[19:27:00.0000] <Hixie>
wow that whole sentence was broken in many ways

[00:54:00.0000] <hsivonen>
hmm. so even the super-smart googlers make mistakes with Namespaces in their RDFa documentation

[01:22:00.0000] <hsivonen>
I wonder what kind of havoc will ensue now that Google uses the de facto VML prefix for another purpose

[01:52:00.0000] <hsivonen>
"<p><strong>Blast 'Em Up Review</strong></p>

[01:52:01.0000] <hsivonen>
"

[01:53:00.0000] <hsivonen>
It's a semantic heading, because it uses <strong> and <strong> is semantic

[02:30:00.0000] <Philip`>
/me tries posting somewhere that uses reCAPTCHA, and finds that it rejects two attempts even though the image was pretty clear and he's almost entirely certain he typed the words in correctly

[02:48:00.0000] <Philip`>
http://radar.oreilly.com/2009/05/google-announces-support-for-m.html - xmlns:v="http://rdf.data-vocabulary.org "

[02:49:00.0000] <Philip`>
Honestly, how hard can it be to get a namespace string right?

[02:50:00.0000] <jgraham>
Philip`: Too hard, it seems :)

[02:50:01.0000] <hsivonen>
Philip`: very, very hard, evidently

[02:50:02.0000] <hsivonen>
I have a special program that hooks into accessibility APIs for inputting namespace URIs

[02:51:00.0000] <hsivonen>
(I don't use it for anything else. really.)

[02:51:01.0000] <hsivonen>
(Typinator)

[02:51:02.0000] <jgraham>
(partially because curies doing simple concatenation violates all sorts of assumptions that people have e.g. http://foo/ and http://foo being assumed the same but treated differently)

[02:52:00.0000] <Philip`>
Looks like Google's updated their help pages to include the trailing "/" now

[02:52:01.0000] <hsivonen>
jgraham: in fairness, that not a CURIE bug per se. It's an issue that goes back to Namespaces itself

[02:53:00.0000] <jgraham>
hsivonen: Really? Maybe I don't fully understand either then

[02:53:01.0000] <jgraham>
Well I guess you just mean that a namespace has to be a literal string

[02:53:02.0000] <hsivonen>
jgraham: xmlns='http://foo' and xmlns='http://foo/' are distinct namespaces ever since Namespaces

[02:53:03.0000] <Philip`>
/me notes that people may make exactly the same mistakes in mistyping HTML5 microdata itemprop values

[02:54:00.0000] <hsivonen>
Philip`: not if the vocabulary designer has the good sense not to mint URI-based vocabularies

[02:58:00.0000] <Philip`>
hsivonen: People mistype non-URI strings too

[02:58:01.0000] <Philip`>
So, maybe not exactly the same mistakes

[02:58:02.0000] <Philip`>
but mistakes with the same effects :-)

[02:59:00.0000] <Hixie>
not quite the same effect

[02:59:01.0000] <Hixie>
it'll break the one occurrence

[02:59:02.0000] <Hixie>
not the whole page

[02:59:03.0000] <hsivonen>
Philip`: I see that as an argument to bring microdata into the sphere of validation instead of keeping it as an anything goes zone with possibly 'spurious triples'

[02:59:04.0000] <jgraham>
Philip`: I think it is a usability issue to use strings that people expect to behave in one way (like they do in their browser address bar) in a context where they are treated entirely differently (as a pure literal)

[03:08:00.0000] <Philip`>
As far as I can tell, the biggest problem with Google's microformats/RDFa support is that they don't provide any way to test it, so it's impossible to tell whether you've got it right or not

[03:11:00.0000] <Philip`>
(...where "you" is a person who's tried to mark up data on their page)

[03:14:00.0000] <hsivonen>
I want to test what happens if

[03:14:01.0000] <hsivonen>
1) I don't declare the prefix v

[03:14:02.0000] <hsivonen>
2) I declare it to another URI

[03:14:03.0000] <hsivonen>
3) I declare the right URI to be bound to another prefix

[03:16:00.0000] <hsivonen>
Considering that Y! hard-coded some prefixes, I think it's too early to say that Google hasn't done the same

[03:24:00.0000] <aja>
is there an example html5 vcard anywhere?

[03:25:00.0000] <aja>
/me wants to make sure he's reading microdata stuff correctly

[03:28:00.0000] <Philip`>
aja: There isn't one yet, as far as I'm aware

[03:29:00.0000] <jgraham>
aja: Make one :)

[03:32:00.0000] <aja>
not sure whether/where to use item= or itemprop= 

[03:35:00.0000] <Philip`>
<div item=vcard> and then everything else is itemprop

[03:35:01.0000] <Philip`>
as far as I can tell

[03:36:00.0000] <Philip`>
(The spec says "vcard" is a (predefined) type, and everything else is a (predefined) property name)

[03:37:00.0000] <jgraham>
FWIW I think itemprop is a terrible name

[03:37:01.0000] <jgraham>
and that seems like one good reason why

[03:37:02.0000] <aja>
Philip`: that's what i thought, too

[03:37:03.0000] <Philip`>
jgraham: Got any better suggestions? :-)

[03:38:00.0000] <jgraham>
Philip`: property ;)

[03:38:01.0000] <jgraham>
Or key

[03:39:00.0000] <jgraham>
propertyname is rather long

[03:39:01.0000] <aja>
was originally property...changed to itemprop shortly thereafter (why? i dunno)

[03:40:00.0000] <jgraham>
aja: Because property has theoretical conflicts with future versions of RDFa

[03:40:01.0000] <aja>
makes sense

[03:51:00.0000] <Hixie>
aja: http://junkyard.damowmow.com/387

[03:51:01.0000] <Hixie>
except s/property/itemprop/

[03:52:00.0000] <aja>
gotcha....tks

[03:53:00.0000] <aja>
/me was pondering how to do include-pattern sorta things

[03:53:01.0000] <Hixie>
aja: subject="" is the only way of doing include-like things yet

[03:53:02.0000] <Hixie>
i was thinking of adding <ref/> in a future version

[03:53:03.0000] <Hixie>
but probably not in html5

[03:54:00.0000] <aja>
hmmm....interesting

[03:58:00.0000] <aja>
Hixie: might subject being possibly plural work?

[03:59:00.0000] <Hixie>
how do you mean?

[04:00:00.0000] <aja>
eg...any org name up on a site's header....being able to be used in multiple vcards further down the page

[04:02:00.0000] <aja>
or single vcard being used on multiple calendar enties...that sorta thing

[04:04:00.0000] <Hixie>
for the latter i think <ref> makes more sense

[04:04:01.0000] <Philip`>
<div item=vcard id=me>...</div> <div item=calendar><meta itemprop=owner object=me>...</div>

[04:04:02.0000] <Hixie>
(if we eventually add it)

[04:04:03.0000] <Hixie>
and for the former, well... use <meta>, i guess

[04:05:00.0000] <Hixie>
do we have any other attributes that take space-separated lists of ids?

[04:05:01.0000] <Hixie>
Philip`: that's more or less what <Ref> would be

[04:07:00.0000] <aja>
only html profile for GRDDL, that i'm aware of, off type of my head.....,.and that's meta now anyway

[04:08:00.0000] <aja>
n/m...ids

[04:11:00.0000] <Philip`>
/me likes how Vodafone has developed open source Linux software in Python for their devices

[04:11:01.0000] <Philip`>
(It's not officially supported, but it seems much better than what the other mobile providers have done on Linux (i.e. nothing, as far as I can tell))

[04:13:00.0000] <jgraham>
/me likes how Vodaphone have developed Linux software for the mobile broadband thing he has

[04:13:01.0000] <aja>
no Android apps site out there yet? ;)

[04:13:02.0000] <Philip`>
jgraham: That's the specific software I'm thinking of :-)

[04:13:03.0000] <Philip`>
Too bad I can't get it to actually work, though

[04:13:04.0000] <Hixie>
/me is starting to get a headache from the iCalendar spec

[04:15:00.0000] <Hixie>
i like how vcard and icalendar are almost compatible but not quite

[04:15:01.0000] <Hixie>
e.g. they have subtly different formats for times and dates

[04:15:02.0000] <jgraham>
Philip`: It seems to both work and not work for me. The whole UI seems to be broken. The actual part that allows me to connect to the internet seems to work

[04:15:03.0000] <Hixie>
or e.g. how iCalendar requires the PRODID line but vcard doesn't

[04:16:00.0000] <Philip`>
jgraham: I managed to connect without that software at all, just using normal PPP stuff, so that's alright

[04:17:00.0000] <hsivonen>
It's a small miracle that calendaring apps interoperate at all

[04:17:01.0000] <Philip`>
Do calendaring apps actually interoperate at all?

[04:17:02.0000] <Hixie>
iCalendar isn't that bad

[04:18:00.0000] <Hixie>
it's better than many web specs on which we have workable (if not stellar) interop

[04:18:01.0000] <Hixie>
doesn't define error handling, but nobody did in the 90s

[04:18:02.0000] <jgraham>
Philip`: Oh well maybe that's what's working for me. I didn't try without the software because I assumed it would fail

[04:18:03.0000] <aja>
hsivonen: latest tryserver builds still for html5 parser?

[04:19:00.0000] <hsivonen>
Philip`: iCal interoperates with Nokia's S60 calendar app and partially interoperates with Ericsson's T series calendaring app

[04:19:01.0000] <hsivonen>
aja: nope

[04:19:02.0000] <hsivonen>
aja: the latest are for the HTML5 namespace correctness patch

[04:20:00.0000] <aja>
hsivonen: the one in your blog entry crashed horribly on win/xp.....linux was fine, though

[04:20:01.0000] <hsivonen>
I should probably pull the old links from the blog entries...

[04:27:00.0000] <hsivonen>
Hixie: when you say "RDF quad", do you mean lang/datatype as the fourth item in the tuple?

[04:27:01.0000] <Hixie>
no, i mean provenance

[04:28:00.0000] <Hixie>
the language and datatype annotations are considered part of the object literal

[04:28:01.0000] <hsivonen>
ok

[04:35:00.0000] <hsivonen>
what's happening with Origin in the IETF?

[04:39:00.0000] <hsivonen>
Hixie: have you addressed all the recent hgroup feedback in the spec? that is, is the outline algorithm stable now?

[04:41:00.0000] <Hixie>
i have at least 16 messages on the topic that i haven't dealt with yet

[04:41:01.0000] <hsivonen>
ok

[04:42:00.0000] <Hixie>
btw if the absolute URL algorithm URIfies, that's a bug

[04:42:01.0000] <Hixie>
are you sure it does?

[04:43:00.0000] <hsivonen>
I'm not quite sure, but I read it twice and I think it does

[04:43:01.0000] <Hixie>
sounds like a bug

[04:43:02.0000] <hsivonen>
/me goes read it the third time

[04:43:03.0000] <Hixie>
it might un-INDA and leave the path as IRI

[04:43:04.0000] <Hixie>
(which would also be a bug)

[04:43:05.0000] <Hixie>
it should ToUnicode() imho

[04:45:00.0000] <hsivonen>
Hixie: should script-inserted xml:base be considered in HTML documents?

[04:45:01.0000] <hsivonen>
doh.

[04:45:02.0000] <hsivonen>
there's a note that says yes

[04:45:03.0000] <hsivonen>
there's a bug in Gecko, then

[04:45:04.0000] <Hixie>
spec says yes

[04:45:05.0000] <Hixie>
the spec goes to some lengths to make base URI changes not be expensive

[04:46:00.0000] <hsivonen>
Gecko checks document HTMLness and returns early

[04:46:01.0000] <Hixie>
if you look at the source, there are comments in a bunch of places about implications certain rules have wrt base URI changes

[04:46:02.0000] <hsivonen>
Hixie: it seems to me that steps 6. and 7.1. modify the URL in the algorithm for resolving

[04:47:00.0000] <hsivonen>
8., too

[04:48:00.0000] <hsivonen>
argh. vCard-RDF uses upper-case local names for properties

[04:48:01.0000] <hsivonen>
worse, it also used mixed case

[05:04:00.0000] <Hixie>
hsivonen: send mail to the list (and connoly⊙wo), i'm pretty sure we want the absolute url to be an IRI, off-hand. Not sure what the implications are though.

[05:07:00.0000] <jgraham>
Hixie: I might be missing something trivial but the "resolve a url" algorithm takes the url to be resolved + either an absolute url or an element

[05:07:01.0000] <jgraham>
But the definiton of "absolute url" just says "if the url resoves onto itself"

[05:07:02.0000] <jgraham>
which doesn't specify what you are resolving it against

[05:08:00.0000] <jgraham>
(those bits in quotation marks are not actual quotes)

[05:08:01.0000] <Hixie>
it should probably say explicitly that it means "if it resolves onto itself for any input base url" or some such

[05:08:02.0000] <Hixie>
again, danc is the guy to talk to here

[05:09:00.0000] <Hixie>
he's editing that bit

[05:09:01.0000] <jgraham>
OK

[05:09:02.0000] <Hixie>
ok bed time now

[05:09:03.0000] <Hixie>
nn

[05:09:04.0000] <jgraham>
goodnight

[05:09:05.0000] <Philip`>
/me wonders if the question about whether 'absolute URL' is a datatype or is a property of a string has been raised or resolved yet

[05:10:00.0000] <jgraham>
/me has been wondering something similar

[05:10:01.0000] <Philip`>
/me is currently too lazy to raise it

[06:00:00.0000] <Philip`>
Hmm, the images on the Last Week blog don't work too well for me

[06:01:00.0000] <Philip`>
since they say "You are not able to access this service because Content Control is in place. If you're 18 years or over, you can remove Content Control by contacting your mobile service provider's customer support team."

[06:06:00.0000] <virtuelv>
Philip`: are you saying that your mobile ISP forces a porn filter on you?

[06:06:01.0000] <virtuelv>
that sounds australian

[06:21:00.0000] <Philip`>
virtuelv: Seems like it

[06:22:00.0000] <Philip`>
though it's not really "forces" if I could ask them to remove it

[06:22:01.0000] <Philip`>
but it sounds like they make it awkward to remove

[06:22:02.0000] <Philip`>
and I don't care that much about seeing Mr Last Week's images

[06:23:00.0000] <jgraham>
Philip`: You are not missing much. Indeed it is like the opposite of missing much

[06:25:00.0000] <virtuelv>
Philip`: I concur with jgraham's conclusion

[06:25:01.0000] <virtuelv>
let's just say that he goatse'd the validator icon for the purpose of annoying mpilgrim 

[09:04:00.0000] <Wolfman2000>
...umm...sort of a silly question here...is there any reason why <table> tags no longer have the summary attribute?

[10:04:00.0000] <Philip`>
Wolfman2000: It's because of the argument that <table summary> is often used badly, and rarely used well, and is replaceable by alternatives (like <legend> or <p>) that are accessible to users of graphical browsers too

[10:05:00.0000] <Wolfman2000>
Philip`: ...and what of those that use the handicapped browsers?

[10:05:01.0000] <Philip`>
Wolfman2000: They can access content in <legend> and <p> too

[10:05:02.0000] <Wolfman2000>
True.  Still...don't tables use <caption>?

[10:05:03.0000] <Philip`>
Um

[10:05:04.0000] <Philip`>
I might mean <caption>, not <legend>

[10:08:00.0000] <webben>
Wolfman2000: i.e. the contention is that an explanation of how a complex table is organized is useful to everybody.

[10:09:00.0000] <Philip`>
The argument also involves the idea that the better way to improve accessibility is by making it as simple and as automatic as possible, instead of by adding complexity through accessibility-specific hooks for non-graphical browsers

[10:10:00.0000] <webben>
Wolfman2000: It's notable that using ARIA you could point at the p with aria-describedby.

[10:10:01.0000] <Philip`>
(because, on average, people use those hooks in incorrect ways (e.g. <img longdesc="some text instead of a URL">) or useless ways (e.g. <table summary="Layout table with 3 columns and 2 rows">), and either harming the user experience or just wasting their own time)

[10:11:00.0000] <Wolfman2000>
...that reminds me Philip`.  The new way for alt text on images is supposed to be <img>Alt Text</img>, right?

[10:11:01.0000] <Philip`>
Wolfman2000: No

[10:11:02.0000] <webben>
Wolfman2000: Still alt=

[10:11:03.0000] <Philip`>
Wolfman2000: <img> is still an empty element

[10:12:00.0000] <Philip`>
(except in XHTML2, I guess)

[10:12:01.0000] <Wolfman2000>
...must be thinking of XHTML2

[10:12:02.0000] <webben>
yep

[10:13:00.0000] <Philip`>
HTML parsers won't understand <img>...</img>, and can't be changed to understand it without breaking existing pages

[10:14:00.0000] <Philip`>
/me notes that some people argue that the benefits of <table summary> are high and the costs are low and therefore it's worth keeping it in HTML5

[10:17:00.0000] <Philip`>
(Also some people seem to focus on the ability to make a page accessible, rather than on the likelihood of a random page written by an average person being made slightly more accessible)

[10:17:01.0000] <Wolfman2000>
I admit I'm more on the "accessible" side

[10:18:00.0000] <Philip`>
Nobody is on the anti-accessible side :-)

[10:21:00.0000] <Wolfman2000>
Philip`: at times I question that.

[11:21:00.0000] <gsnedders>
/me is wondering how to sanely mark up drama

[11:24:00.0000] <Philip`>
Is this like Shakespearean drama or emo drama?

[11:24:01.0000] <gsnedders>
Shakespearean

[11:25:00.0000] <gsnedders>
Philo: "And is become the bellows and the fan/To cool a gypsy's lust." (I.1.9–10)

[11:25:01.0000] <gsnedders>
Problem: marking that up sanely, and getting sane presentation of that.

[11:25:02.0000] <gsnedders>
However, my exam is on Friday so I don't really have time to care.

[11:31:00.0000] <Philip`>
Hmm, Firefox 2 moves <link> and <meta> into <head>; Firefox 3 only moves <meta>

[11:48:00.0000] <gsnedders>
For some reason my spell-checker doesn't know these Shakespearean names.

[14:12:00.0000] <jgraham>
/me discovers he has accidnetially caused a very odd spec discussion on a mailing list he wasn't aware of

[14:19:00.0000] <Philip`>
/me demands details

[14:19:01.0000] <Hixie>
jgraham: uri?

[14:27:00.0000] <jgraham>
Philip`: It turns out that there is a es5-discuss mailing list in addition to the es-discuss mailing list I knew about

[14:28:00.0000] <jgraham>
After reorting a possible compatibility issue on es-discuss a thread started on es5-discuss which ends with Mark Miller and Douglas Crockford suggesting removing string indexing from ecmascript 5 

[14:29:00.0000] <jgraham>
As an alternative to introducing a single compatibiliy-preserving type check 

[14:29:01.0000] <Philip`>
They should go the whole hog and remove strings entirely

[14:29:02.0000] <inimino>
there is or has been in the past es-discuss, es3x-discuss, es4-discuss, and es5-discuss

[14:30:00.0000] <inimino>
(at least)

[14:30:01.0000] <jgraham>
inimino: And people complain about whatwg vs public-html

[14:30:02.0000] <jgraham>
At least those two have different participati

[14:30:03.0000] <jgraham>
on requirements

[14:32:00.0000] <jgraham>
Hixie: https://mail.mozilla.org/pipermail/es5-discuss/2009-May/002490.html

[14:33:00.0000] <Hixie>
thx

[15:03:00.0000] <jgraham>
/me wonders if wikipedia is supposed to be used as software api documentation

[16:10:00.0000] <jgraham>
/me has a certianly-broken implementation of N3 output now

[16:16:00.0000] <Hixie>
http://www.w3.org/mid/1igm05tt25ip8kg10hureru72agmo2n2c9⊙hbhd has the most references of any e-mail i've ever seen in w3c space

[16:26:00.0000] <jgraham>
That is rather impressive

[16:38:00.0000] <Philip`>
/me gets the feeling that Björn has much more fun with the formatting of his emails than with their content

[16:38:01.0000] <Hixie>
...why does the iCalendar example for RESOURCE imply that the meeting is going to need a racoon?!

[16:39:00.0000] <Hixie>
what kind of weird-ass meeting were the iCalendar guys going to?!

[16:42:00.0000] <Philip`>
Perhaps it was a meeting to discuss plans to overthrow Cyril Sneer

[16:45:00.0000] <Hixie>
can someone tell me if http://www.ietf.org/rfc/rfc2445.txt says whether the STATUS line can be included more than once per vevent?

[16:45:01.0000] <Hixie>
it looks like it allows it

[16:45:02.0000] <Hixie>
but i may be missing some key line somewhere that says that the default is no duplicates?

[16:51:00.0000] <Dashiva>
A French raccoon to boot...

[16:51:01.0000] <Hixie>
i'm thinking the word has a double meaning but if it does i don't know it

[16:52:00.0000] <Dashiva>
Maybe karlcow knows


2009-05-14
[17:05:00.0000] <tantek>
Hixie, from reading RFC2445 4.8.1.11 "Purpose: This property defines **the** overall status or confirmation for the calendar component." (**emphasis** added)

[17:06:00.0000] <tantek>
Thus at first glance I would conclude at most **one** STATUS per VEVENT

[17:06:01.0000] <Hixie>
sure but is there a conformance criteria anywhere saying how many you can give?

[17:06:02.0000] <Hixie>
some of them are explicit

[17:07:00.0000] <tantek>
Hixie, compare with the next section 4.8.1.12 Summary which says "Purpose: This property defines **a** short summary or subject for the calendar component."

[17:07:01.0000] <Hixie>
yes but e.g. some say "The property can be specified once"

[17:07:02.0000] <Hixie>
and others say "This property can be specified in"

[17:07:03.0000] <tantek>
lacking explicit conformance criteria - such differences in property definition are another source of intent we have to go by

[17:07:04.0000] <tantek>
also, by what "makes sense"

[17:07:05.0000] <tantek>
e.g. it makes sense for an event to have at most one STATUS

[17:08:00.0000] <Hixie>
my goal here is to merely duplicate the exact conformance criteria of the rfc

[17:08:01.0000] <Hixie>
without interpretation

[17:08:02.0000] <tantek>
no matter how you read it you will be doing some interpretation

[17:08:03.0000] <Hixie>
without interpretation beyond the literal

[17:09:00.0000] <tantek>
and it also makes sense for an event to have multiple SUMMARY value, perhaps of different lengths 

[17:09:01.0000] <Philip`>
Why not just say the data represented in the markup must follow the conformance criteria of the RFC (and not explicitly say what they are), and let validator developers work out what it means?

[17:10:00.0000] <Philip`>
/me remembers reading things recently about someone attempting to make an iCal validator that was actually useful in practice and would tell you when your calendar would work in common applications, based on the feed validator

[17:10:01.0000] <Hixie>
Philip`: i couldn't work out a way to say it that made sense

[17:11:00.0000] <Hixie>
i hate how icalendar requires you to declare whether your value is a datetime or a date

[17:12:00.0000] <Hixie>
why on earth doesn't it just assume it's a datetime if there's a date and a time and assume it's a date if there's just a date?

[17:16:00.0000] <Hixie>
wohay!

[17:17:00.0000] <Hixie>
i found a part of the spec that actually says what can be included once and what can't!

[17:18:00.0000] <tantek>
Hixie do share which section #

[17:21:00.0000] <Hixie>
4.6.1 Event Component

[17:21:01.0000] <Hixie>
this really is a quite well-written spec

[17:21:02.0000] <Hixie>
no implied conformance criteria or anything

[17:26:00.0000] <tantek>
good find Hixie

[17:43:00.0000] <Hixie>
weird

[17:43:01.0000] <Hixie>
'duration' can take P1DT1S but not P1DT1H1S

[17:43:02.0000] <Hixie>
if you give the hours you have to give the minutes in order to give the seconds

[17:44:00.0000] <tantek>
I'm sure someone made an argument like "makes parsing easier" ;)

[17:45:00.0000] <Hixie>
i don't see how, in this case

[18:02:00.0000] <Hixie>
/me fails to understand the purpose and use of RECURRENCE-ID in iCalendar

[18:09:00.0000] <Hixie>
i wonder whether to support all the recurring event properties

[18:22:00.0000] <Philip`>
http://www.jenitennison.com/blog/node/104 - hmm, that "really simple scenario" seems to be basically what I was talking about last night, with fetching vocabulary definitions to discover subclass relationships

[18:24:00.0000] <Philip`>
(which is scary)

[18:24:01.0000] <Hixie>
"These are the kind of pipe dreams that I used to ridicule semantic web folk about" indeed

[18:27:00.0000] <Philip`>
/me wonders what happens when Amazon builds its vocabulary as a subclass of Google's, and then Google changes its own vocabulary to be a subclass of Amazon's

[18:44:00.0000] <Dashiva>
Philip`: They're both right, obviously

[18:45:00.0000] <Hixie>
jesus, iCalendar has a lot of examples for rdate

[18:47:00.0000] <Dashiva>
"They will now have a stake in answering the difficult questions around trust, confidence, accuracy and time-sensitivity of semantic information."

[18:47:01.0000] <Dashiva>
Are there even answers to those questions?

[18:47:02.0000] <Hixie>
it's google's job to answer questions around trust, confidence, accuracy and time-sensitivity of semantic information

[18:47:03.0000] <tantek>
Hixie, rdate is really complex. And not clear that all the complexity is desired or needed.

[18:47:04.0000] <Hixie>
Dashiva: that's pretty much what google's area of expertise is

[18:48:00.0000] <Hixie>
tantek: indeed

[18:48:01.0000] <tantek>
Hixie, I suggest you take a look at the updates (removals) to rdate etc. that iCal-BASIC is doing.

[18:48:02.0000] <tantek>
I was keeping up with the "ietf-calsify" mailing list for a while, but after a while was unable to do so.

[18:48:03.0000] <tantek>
http://lists.osafoundation.org/pipermail/ietf-calsify/

[18:48:04.0000] <tantek>
you may have better luck

[18:51:00.0000] <Hixie>
tantek: thanks, noted

[19:10:00.0000] <Hixie>
ok, finished adding the vEvent vocabulary to the html5 microdata section

[00:18:00.0000] <hsivonen>
Philip`: where were you quoting from in http://realtech.burningbird.net/semantic-web/semantic-web-issues-and-practices/holding-on-html5#comment-1455 ?

[00:36:00.0000] <zcorpan_>
Philip`: my copy of webkit moves <meta>, but not <link>, to head

[00:36:01.0000] <zcorpan_>
wonder if someone has filed bugs

[00:41:00.0000] <Hixie>
shelley missed the point in Philip`'s comment

[00:41:01.0000] <Hixie>
which is that people _must_ understand namespaces to use them

[00:42:00.0000] <Hixie>
but that they can treat the reverse dns names as opaque magic incantations and it'll still work

[00:52:00.0000] <aja>
Hixie: ping (typo nit)

[00:53:00.0000] <Hixie>
aja: here

[00:53:01.0000] <aja>
geo says 'dix' instead of 'six' digits

[00:54:00.0000] <Hixie>
thanks

[00:54:01.0000] <Hixie>
will fix

[00:55:00.0000] <aja>
np....what other uF-like micro's to come?

[00:56:00.0000] <Hixie>
aja: vevent (i added today), bibtex (doing that now), and vcard (already in) are all i'm planning on adding right now

[00:56:01.0000] <Hixie>
those were the ones i needed for the use cases that came up

[00:57:00.0000] <aja>
/me was hoping for reviews :)

[00:58:00.0000] <aja>
and even hitem......yikes, item=item

[01:00:00.0000] <hsivonen>
https://twitter.com/arun/statuses/1787778325

[01:00:01.0000] <aja>
guess those are probably far from 80% cases

[01:05:00.0000] <aja>
hsivonen: more like a grue, IMHO

[01:10:00.0000] <zcorpan_>
Hixie: "They are based on the vocabulary defined in the iCalendar specification, which should be consulted for more information about how to interpret values." - i'm a bit confused as to how to conform when one "should consult" another spec

[01:10:01.0000] <Hixie>
aja: nobody mentioned reviews anywhere in the use cases as far as i recall

[01:11:00.0000] <Hixie>
zcorpan_: what kind of conformance?

[01:11:01.0000] <Hixie>
(author, browser, tool, etc)

[01:12:00.0000] <zcorpan_>
Hixie: well that's also not so clear to me, who is the "should" intended for

[01:13:00.0000] <Hixie>
oh, oops, that shouldn't be a should

[01:13:01.0000] <Hixie>
my bad

[01:13:02.0000] <aja>
think twas similar wording in vcard, too, fwiw

[01:14:00.0000] <zcorpan_>
phew, i thought for a second it was just me (after pf suggesting html5 should say "should consult wcag" for alt requirements)

[01:14:01.0000] <zcorpan_>
(which also made no sense to me)

[01:15:00.0000] <Hixie>
yeah that was a mistake

[01:18:00.0000] <aja>
at least pf didn't say MUST :)

[01:20:00.0000] <zcorpan_>
aja: actually, after saying that "should consult" didn't make sense to me, they suggested instead "must consult"

[01:21:00.0000] <aja>
oh well

[01:21:01.0000] <Hixie>
was that a ua conf criteria? cos that would be hard to test!

[01:21:02.0000] <zcorpan_>
Hixie: i think it would be an authoring conf criteria

[01:22:00.0000] <zcorpan_>
still equally hard to test though

[01:22:01.0000] <zcorpan_>
since one can consult a spec but then ignore it

[01:22:02.0000] <Hixie>
i guess you ask the author "did you consult it?" and if they say no, their document is invalid?

[01:22:03.0000] <Hixie>
but if they say yes, then regardless of what they did, that criteria is met.

[01:23:00.0000] <zcorpan_>
yep

[01:24:00.0000] <Hixie>
man, bibtex is not well-defined

[01:25:00.0000] <gsnedders>
Hixie: What? You've found _any_ definition? You've done better than I.

[01:26:00.0000] <aja>
...and has lotsa scary curly braces

[01:26:01.0000] <Hixie>
gsnedders: if you count tutorials and the like, i've found dozens

[01:26:02.0000] <Hixie>
all slightly different

[01:26:03.0000] <Hixie>
in subtle ways

[01:27:00.0000] <gsnedders>
Hixie: Oh, sure, I found tutorials and the like. Never any actual spec, though.

[01:27:01.0000] <Hixie>
i also found the original "spec"/documentation

[01:27:02.0000] <Hixie>
from like 85

[01:27:03.0000] <gsnedders>
Oh, that.

[01:27:04.0000] <gsnedders>
That's useless.

[01:27:05.0000] <Hixie>
but iirc it failed to define anything

[01:27:06.0000] <gsnedders>
It defines what the fields mean.

[01:27:07.0000] <gsnedders>
It doesn't define the syntax or anything useful like that.

[01:30:00.0000] <Hixie>
yeah

[01:32:00.0000] <gsnedders>
This is why for Anolis I gave up and used refer (troff's format)

[01:32:01.0000] <hsivonen>
gsnedders: since when do you give up on reverse engineering stuff and use something else?

[01:32:02.0000] <gsnedders>
hsivonen: When I don't need one specific format badly enough

[01:33:00.0000] <gsnedders>
hsivonen: Also: when I have real deadlines

[01:33:01.0000] <Hixie>
bibtex is more widely used as far as i can tell

[01:33:02.0000] <zcorpan_>
so who'll write up a Web bibtex?

[01:33:03.0000] <Hixie>
me

[01:33:04.0000] <Hixie>
doing it now

[01:33:05.0000] <Hixie>
literally as we speak

[01:33:06.0000] <zcorpan_>
cool

[01:33:07.0000] <Hixie>
i'm up to "note"

[01:33:08.0000] <Hixie>
going down alphabetically :-)

[01:34:00.0000] <gsnedders>
Hixie: I take it you aren't worrying about real-BibTeX syntax?

[01:35:00.0000] <tantek>
gsnedders, you mean bibtex like for citations?

[01:36:00.0000] <gsnedders>
tantek: No, I mean what goes in the actual .bib file

[01:36:01.0000] <Hixie>
gsnedders: define "worrying"

[01:36:02.0000] <Hixie>
gsnedders: i'm using the vocabulary for microdata

[01:36:03.0000] <Hixie>
gsnedders: and defining how you convert that _to_ bibtex

[01:37:00.0000] <Hixie>
which is far easier than parsing it

[01:37:01.0000] <gsnedders>
Hixie: Needing to define syntax; so you don't need to actual parse BibTeX

[01:37:02.0000] <tantek>
Hixie, why did you choose bibtex when there are so many citation formats to choose from? http://microformats.org/wiki/citation-formats

[01:37:03.0000] <hsivonen>
Hixie: and _from_ is Someone Else's Problem?

[01:38:00.0000] <Hixie>
tantek: it was the best of a bad bunch (i used the microformats.org page as a guide)

[01:40:00.0000] <tantek>
What evaluation function did you use to determine "best" (or even an approximation thereof) ?

[01:40:01.0000] <hsivonen>
Hixie: is it deliberate or accidental that HTML5 defines a different tagName upper-casing algorithm than what browsers actually use?

[01:40:02.0000] <Hixie>
tantek: i forget, it was about 2 weeks ago.

[01:41:00.0000] <Hixie>
tantek: a combination of factors including deployment, support in software that would be used by the people who's use cases i was trying to address, breadth of support, etc

[01:41:01.0000] <Hixie>
(breadth as in how many types of works were supported)

[01:41:02.0000] <tantek>
ok good to know. that's been one of the blocking items for the citation microformat

[01:41:03.0000] <Hixie>
hsivonen: no

[01:42:00.0000] <hsivonen>
tantek: fwiw, I'd pick BibTeX by popularity which includes stuff like ACM offering snippets in .bib

[01:42:01.0000] <tantek>
too many existing (not quite compatible) citation formats to choose from

[01:42:02.0000] <Hixie>
stuff like ACM offering snippets in .bib was one of the factors

[01:42:03.0000] <Hixie>
in fact export to bibtex is quite a commonly available feature

[01:43:00.0000] <Hixie>
even google scholar does it

[01:43:01.0000] <hsivonen>
Hixie: although with those snippets, you need to be able convert from .bib

[01:44:00.0000] <hsivonen>
Hixie: was "no" to deliberate or accidental?

[01:44:01.0000] <Hixie>
hsivonen: not deliberate as far as i'm aware (sorry)

[01:44:02.0000] <hsivonen>
Hixie: thanks

[01:46:00.0000] <gsnedders>
/me finishes reading Björn's email

[01:47:00.0000] <gsnedders>
That was, uh, impressive.

[01:49:00.0000] <hsivonen>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/105

[01:49:01.0000] <hsivonen>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/106

[01:49:02.0000] <hsivonen>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/107

[01:49:03.0000] <hsivonen>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/108

[01:49:04.0000] <Hixie>
oh yes, that _is_ intentional

[01:49:05.0000] <Hixie>
it was based on something you said, actually

[01:50:00.0000] <Hixie>
namely that the parser should only have magic in the 7bit ascii range

[01:50:01.0000] <hsivonen>
It seems that Gecko and WebKit both have different lower-casing in the parser and in the DOM

[01:50:02.0000] <hsivonen>
yay

[01:51:00.0000] <hsivonen>
Hixie: only having 7bit magic in parser matches Gecko&WebKit, yes

[01:51:01.0000] <hsivonen>
Opera is different...

[01:53:00.0000] <Hixie>
(i don't think this is an area with great interop issues)

[01:54:00.0000] <hsivonen>
See http://software.hixie.ch/utilities/js/live-dom-viewer/saved/109 in Safari...

[01:54:01.0000] <hsivonen>
the DOM view

[01:54:02.0000] <hsivonen>
ß get uppercased to SS

[01:55:00.0000] <hsivonen>
I wonder what upper/lowercasing Gecko uses in the DOM

[01:55:01.0000] <hsivonen>
it's neither ASCII nor full Unicode

[01:56:00.0000] <Hixie>
w(document.getElementsByTagName("gross").length) == 0 in that example

[01:56:01.0000] <Hixie>
good times

[01:56:02.0000] <Hixie>
w(document.getElementsByTagName("GROSS").length) == 0 too

[01:57:00.0000] <hsivonen>
clearly, WebKit lowercases argument instead of doing Unicode fold case comparison

[01:57:01.0000] <Hixie>
didn't this get resolved recently?

[01:58:00.0000] <hsivonen>
the spec says *something* but I don't recall the discussion of getting there taking data about the DOM side of implementation into account

[01:59:00.0000] <hsivonen>
I'd like to see an informed sign-off from people who call the DOM shots for various code bases before I try to change the algorithms Gecko uses here

[02:00:00.0000] <hsivonen>
I'll send email

[02:03:00.0000] <Hixie>
i seem to recall the discussion involved bz and mjs

[02:03:01.0000] <Hixie>
and them signing off on what the spec says now

[02:04:00.0000] <hsivonen>
ah. ok

[02:05:00.0000] <othermaciej>
what was the topic?

[02:06:00.0000] <othermaciej>
oh I'd have to review what the spec says

[02:06:01.0000] <hsivonen>
othermaciej: browsers currently use different upper/lowercasing algorithms in DOM operations from what the spec says

[02:07:00.0000] <othermaciej>
I think what makes sense is to ascii-lowercase (not Unicode) the parameter to getElementsByTagName for HTML elements in HTML documents only

[02:07:01.0000] <othermaciej>
and maybe also lowercase the tag name in such cases

[02:07:02.0000] <othermaciej>
(I think I was in favor of that but bz was not)

[02:07:03.0000] <othermaciej>
(if no lowercasing then you have to use localName instead of tagName to DTRT for parser-generated nodes)

[02:08:00.0000] <Hixie>
the spec says, for html docs only, "compare the given argument in a case-sensitive manner, but when looking at HTML elements, the argument must first be converted to lowercase"

[02:08:01.0000] <Hixie>
and boris signed off on it: http://lists.w3.org/Archives/Public/public-html/2009May/0020.html

[02:09:00.0000] <othermaciej>
and does "converted to lowercase" mean unicode or ASCII lowercase?

[02:09:01.0000] <Hixie>
Converting a string to lowercase means replacing all characters in the range U+0041 .. U+005A (i.e. LATIN CAPITAL LETTER A to LATIN CAPITAL LETTER Z) with the corresponding characters in the range U+0061 .. U+007A (i.e. LATIN SMALL LETTER A to LATIN SMALL LETTER Z).

[02:09:02.0000] <hsivonen>
Hixie: you really should change the term in the spec to include the qualifier ASCII

[02:09:03.0000] <Hixie>
someone filed a bug on it

[02:09:04.0000] <Hixie>
it'll happen in due course

[02:10:00.0000] <Hixie>
rather busy with microdata right now

[02:10:01.0000] <Hixie>
:-)

[02:10:02.0000] <othermaciej>
so I see a result of 1 in my WebKit ToT build for the case hsivonen posted

[02:11:00.0000] <othermaciej>
is that not right?

[02:11:01.0000] <hsivonen>
othermaciej: which test case?

[02:11:02.0000] <othermaciej>
does the spec say to compare to tagName or to localName?

[02:11:03.0000] <Hixie>
what's wrong in webkit for the last example hsivonen posted is that the DOM view shows .tagName says "SS"

[02:11:04.0000] <hsivonen>
othermaciej: at least Safari 4 public beta uses Unicode upper/lowercasing for DOM ops

[02:12:00.0000] <othermaciej>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/109

[02:12:01.0000] <othermaciej>
is it the uppercasing of the tagName that's wrong?

[02:12:02.0000] <hsivonen>
othermaciej: yes

[02:12:03.0000] <Hixie>
a better example is http://software.hixie.ch/utilities/js/live-dom-viewer/saved/110

[02:13:00.0000] <Hixie>
compare the result of getEBTN() (1, correct) to the DOM (shows two GROSSes, which is wrong)

[02:13:01.0000] <othermaciej>
fair enough

[02:13:02.0000] <Hixie>
wrong per html5, that is

[02:13:03.0000] <Hixie>
i don't think this is anything to worry about

[02:14:00.0000] <othermaciej>
I think our getElementByTagName compares localName, not tagName

[02:14:01.0000] <othermaciej>
which is why the DOM display is inconsistent with the results of gEBTN

[02:14:02.0000] <Hixie>
right

[02:15:00.0000] <Hixie>
the DOM display is using a Unicode ToUpper() but should use just the ASCII one per html5 today

[02:15:01.0000] <othermaciej>
is it correct to compare to localName? (I would think so, despite the name of the API)

[02:15:02.0000] <Hixie>
yes

[02:15:03.0000] <hsivonen>
does tagName exist in the WebKit data model

[02:15:04.0000] <hsivonen>
it doesn't in the Gecko data model

[02:15:05.0000] <othermaciej>
I don't know what you mean by "exist"

[02:16:00.0000] <hsivonen>
othermaciej: does an element have a pointer to a string/atom object that represents tagName?

[02:16:01.0000] <othermaciej>
no

[02:17:00.0000] <othermaciej>
it knows how to compute it from the QName

[02:17:01.0000] <Hixie>
appaarently it doesn't know how to do it properly :-P

[02:17:02.0000] <othermaciej>
(which is different for html elements in an html document)

[02:17:03.0000] <hsivonen>
Gecko also stores local and prefix and computes tagName if JS asks for it

[02:19:00.0000] <othermaciej>
really the only bug is using a unicode uppercasing instead of an ascii uppercasing

[02:19:01.0000] <Hixie>
yep

[02:20:00.0000] <othermaciej>
though it also looks to me like getElementsByTagName is doing a unicode lowercasing

[02:20:01.0000] <othermaciej>
which is also a bug, but not revealed by hsivonen's example

[02:20:02.0000] <othermaciej>
I could have sworn we added operations to do ASCII-only upper/lower

[02:20:03.0000] <Hixie>
and you'll get an imperceptible perf boost by following html5, horrah

[02:24:00.0000] <hsivonen>
othermaciej: it was revealed by one of my earlier examples

[02:24:01.0000] <zcorpan_>
wait, getElementsByTagName uses localName?

[02:24:02.0000] <zcorpan_>
/me has probably got that wrong in Web DOM Core then

[02:24:03.0000] <hsivonen>
zcorpan_: I sure hope so

[02:25:00.0000] <hsivonen>
/me hasn't tested

[02:25:01.0000] <othermaciej>
amusingly, DOM 3 Core is unclear

[02:27:00.0000] <zcorpan_>
firefox uses nodeName

[02:27:01.0000] <zcorpan_>
opera and webkit use localName

[02:27:02.0000] <zcorpan_>
ie uses the equivalent of localName, too

[02:27:03.0000] <zcorpan_>
/me will fix web dom core

[02:29:00.0000] <hsivonen>
I guess I should file a Gecko bug, then

[02:30:00.0000] <Philip`>
hsivonen: http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2009-May/019681.html

[02:31:00.0000] <hsivonen>
Philip`: thanks

[02:36:00.0000] <zcorpan_>
http://simon.html5.org/specs/web-dom-core#dom-document-getelementsbytagname fixed

[02:37:00.0000] <zcorpan_>
that just made getElementsByTagName more useful in compound documents, like selectors

[02:39:00.0000] <gsnedders>
zcorpan_: What's a filter?

[02:39:01.0000] <zcorpan_>
gsnedders: "A NodeList is a collection, except that NodeLists can be static (as opposed to live) if a specification defines them to be static. [HTML5]"

[02:40:00.0000] <gsnedders>
um, how am I meant to know that is what a filter refers to?

[02:40:01.0000] <zcorpan_>
gsnedders: i guess i should reference html5 for that

[02:54:00.0000] <Philip`>
jgraham: <type 'exceptions.NameError'>: global name 'reflib' is not defined

[02:54:01.0000] <Philip`>
in microdata.py:193

[02:55:00.0000] <Philip`>
using http://iandavis.com/2009/microdata/ as input

[02:55:01.0000] <hsivonen>
http://blogs.adobe.com/jd/2009/05/building_upon_untested_assumpt.html

[02:55:02.0000] <jgraham>
Philip`: Fixed?

[02:56:00.0000] <jgraham>
(I literally only tested the test document as input and I only checked that the output existed not that it made sense so, as I said, there are bugs)

[02:56:01.0000] <Philip`>
jgraham: Seems to be happier now

[02:57:00.0000] <Philip`>
"[<microdata.SimpleUrl object at 0x2b07e4e47490>]" _3:title "Internet Alchemy";

[02:59:00.0000] <Philip`>
Hmm, the output seems to be only vaguely correlated to what it should be

[03:00:00.0000] <Philip`>
(even on test.html)

[03:01:00.0000] <Philip`>
So I agree there are bugs :-)

[03:39:00.0000] <gsnedders>
"Summary: The HTML5 editor says he's fighting a battle against Flash. But he doesn't explain why, so it's hard for us to help him get better." — jd

[03:50:00.0000] <aja>
anyone know if chromium pre-alpha nightlies are pulling webkit trunk? says 530.10

[03:51:00.0000] <Hixie>
chromium is usually a bit behind

[03:51:01.0000] <aja>
/me obviously knows nada abouit webkit version numbering

[03:51:02.0000] <Hixie>
they merge to trunk every few days/weeks (not sure the schedule)

[03:51:03.0000] <Hixie>
(at least they used to, they were aiming to just be on tip always, dunno if they got there yet)

[03:52:00.0000] <Hixie>
there's a #chromium channel which might be more appropriate :-)

[03:52:01.0000] <aja>
yeah,that's why i was wondering

[04:30:00.0000] <zcorpan>
wonder if html5 should be published again once the microdata stuff is "done"

[04:31:00.0000] <zcorpan>
to promote wider review of the microdata stuff

[04:31:01.0000] <Hixie>
probably

[04:31:02.0000] <Hixie>
i wonder when sam will do that review he said he'd do

[04:31:03.0000] <Hixie>
i thought he said he was going to do it last month

[04:32:00.0000] <Hixie>
we sent out the whatwg call for review and everything, but i don't think he ever did anything with it

[04:34:00.0000] <zcorpan>
Hixie: re running out of punctuation characters to use, there's "|"

[04:34:01.0000] <zcorpan>
Hixie: also, s/tetx/text/

[04:35:00.0000] <jgraham>
/me wonders hy Hixie is running out of punctuation characters

[04:35:01.0000] <jgraham>
*why

[04:40:00.0000] <zcorpan>
don't remember what he'd use them for, and can't find it in the logs, but i think it was something with abbreviating microdata names, and "." and ":" were taken

[04:45:00.0000] <Hixie>
zcorpan: yeah, | might work. It's what CSS uses, I guess.

[04:45:01.0000] <Hixie>
jgraham: basically if we want to register prefixes, we'd need a prefix punctuation character

[04:46:00.0000] <Hixie>
jgraham: so e.g. "foaf" could be registered in the wiki

[04:46:01.0000] <Hixie>
foaf|name

[04:46:02.0000] <Hixie>
item="foaf|Person" ... itemprop="foaf|name"

[04:46:03.0000] <Hixie>
could work i guess

[04:46:04.0000] <Hixie>
zcorpan: (fixed tetx, thanks)

[04:47:00.0000] <Philip`>
Or use "-"

[04:47:01.0000] <hsivonen>
the openid rel stuff uses .

[04:48:00.0000] <Hixie>
Philip`: too many one-word terms use "-"

[04:48:01.0000] <Hixie>
hsivonen: "." is in uris and reversed dns thingies

[04:48:02.0000] <jgraham>
It has to be a non-uri character, right?

[04:48:03.0000] <zcorpan>
a uri will have a colon and a reversed dns would have at least two dots

[04:48:04.0000] <zcorpan>
whereas a prefixed name would have just one dot

[04:49:00.0000] <Hixie>
true

[04:49:01.0000] <Philip`>
Could get rid of the reversed DNS thing, and then '.' would be free

[04:49:02.0000] <jgraham>
using 1 vs multiple dots seems really bad

[04:50:00.0000] <Hixie>
Philip`: i don't want to require that people use ugly URIs or require that they prefix things.

[04:50:01.0000] <Hixie>
with registered prefixes

[04:50:02.0000] <jgraham>
Would / work?

[04:51:00.0000] <Philip`>
Hixie: They can still use non-globally-unique strings in that case

[04:51:01.0000] <zcorpan>
# ?

[04:51:02.0000] <jgraham>
Philip`: I think the reverse dns thing is nice

[04:52:00.0000] <zcorpan>
item="foaf/Person" ... item="foaf#Person"

[04:52:01.0000] <jgraham>
Philip`: Also if you want ot report bugs on my microdata implementation, please do :)

[04:52:02.0000] <Philip`>
The reversed DNS thing only seems to be useful when they want a globally unique identifier, and they don't want to register a prefix, and they think "z.y.x.a" is significantly less ugly than "http://x.y.z/a"

[04:52:03.0000] <Philip`>
jgraham: It's too broken for me to bother reporting specific bugs - you should just look at it yourself ;-)

[04:52:04.0000] <jgraham>
Gah

[04:53:00.0000] <Philip`>
(at least the RDF bit)

[04:53:01.0000] <jgraham>
I was hoping that you would have noticed some particular class of underlying brokenness which suggested a simple but far reaching bug

[04:54:00.0000] <Philip`>
jgraham: As far as I can tell, the underlying brokenness is that your RDF extraction is totally broken :-p

[04:54:01.0000] <Philip`>
but it might still be a simple fix in the implementation

[04:56:00.0000] <Philip`>
jgraham: e.g. "Amanda" and "Jazz Band" aren't properties of whatever item they're meant to be properties of

[04:56:01.0000] <jgraham>
Philip`: That wasn't too helpful :p (seriously though I will try to look at it this evening or so)

[04:56:02.0000] <Philip`>
and all the top-level items should be in vocab:item

[04:57:00.0000] <Philip`>
and there's no rdf:type anywhere

[04:58:00.0000] <jgraham>
Is rdf:type something that was added more recently than Sunday? I didn't actually read the spec again since then

[04:58:01.0000] <Philip`>
It's always been there

[04:58:02.0000] <Philip`>
(vocab:item was new, though)

[04:59:00.0000] <Hixie>
ah well i figure we'll punt on the registered prefixes for now

[04:59:01.0000] <Hixie>
i'm sure we can find something to make it work if we need to later

[04:59:02.0000] <Hixie>
first though i must sleep

[04:59:03.0000] <Hixie>
nn

[04:59:04.0000] <Philip`>
jgraham: Also, in things like '<http://www.w3.org/1999/xhtml/custom#com.damowmow.img> "http://james.html5.org/microdata/hedral.jpeg"' I think the object should be a resource rather than a string

[05:00:00.0000] <jgraham>
Philip`: That sounds reasonable

[05:00:01.0000] <Philip`>
jgraham: Also: """Silver is  years old and refuses to eat alone, always waiting for either Yellow or Blue to eat with him.11"""

[05:01:00.0000] <jgraham>
Interesting

[05:01:01.0000] <Philip`>
Also looks like <time> isn't being handled properly

[05:03:00.0000] <jgraham>
Philip`: Thanks

[05:09:00.0000] <zcorpan>
Hixie: btw, you ought to change (or remove) the fallback text in http://damowmow.com/playground/demos/flash/001.html since i see people pointing to it still and hence people copy-and-paste verbatim and the web is full of "FAIL (the browser should render some flash content, not this)."

[05:10:00.0000] <zcorpan>
s/verbatim/and change only the parts needed to make it work/

[05:13:00.0000] <zcorpan>
Hixie: also the cc should look like <!--[if !IE]>--> ... <!--<![endif]--> in order to be valid in html5

[06:54:00.0000] <hsivonen>
hmm. it seems to me that http://rdfa.info/wiki/Rdfa-profiles would render new content written with RDFa profiles incompatible with existing RDFa consumers

[06:54:01.0000] <hsivonen>
just like HTML5 microdata would be incompatible with existing RDFa consumers

[06:57:00.0000] <Philip`>
I guess changing xmlns:*="" to prefix="..." would be similarly incompatible

[07:00:00.0000] <Philip`>
(So it'd be nice to sort out any such issues before everyone starts deploying RDFa more than they have already)

[07:33:00.0000] <hsivonen>
I'm completely puzzled why http://mxr.mozilla.org/mozilla-central/source/layout/reftests/bugs/444375-1-ref.html renders differently with the HTML5 parser even though the DOM looks the same to me

[08:03:00.0000] <gsnedders>
Is anyone expecting @autosave or @results to be valid on input?

[08:06:00.0000] <Philip`>
/me decides that the microdata DOM API is crazy, because sometimes it'll return a string and sometimes it'll return an array and you have no way of predicting which it's going to be and your code will break horribly when you expect something to specified multiple times and it's only specified once

[08:06:01.0000] <Philip`>
s/to spec/to be spec/

[08:32:00.0000] <Lachy>
/me finally found a few minutes to look at the new microdata section

[08:32:01.0000] <Lachy>
looks interesting, but only read the introductory text so far


2009-05-15
[00:58:00.0000] <tantek>
Quick question - the archives for the channel should be here right? http://krijnhoetmer.nl/irc-logs/whatwg/20090515

[01:00:00.0000] <Hixie>
yup

[01:00:01.0000] <Hixie>
but krijnh wasn't in the channel these past few hours so some stuff will have been missed

[01:56:00.0000] <Hixie>
jesus, the win7 installer makes the linux installers look positively consistent

[02:02:00.0000] <Hixie>
ok seriously

[02:02:01.0000] <Hixie>
even the VERY FIRST DIALOG of the installer mixes the win2k skin and the vista skin

[02:02:02.0000] <Hixie>
wtf

[02:04:00.0000] <Hixie>
the first window is win2k-style (with non-native buttons) and can be dragged around, the second window in the installer is aero-style with buggy button focus outlines

[02:04:01.0000] <Hixie>
-_-

[02:07:00.0000] <othermaciej>
Hixie: just wait until you find the stuff with Windows 98 skins

[02:07:01.0000] <othermaciej>
or Win 3.1

[02:07:02.0000] <Hixie>
really?

[02:07:03.0000] <Hixie>
in win7 proper?

[02:07:04.0000] <othermaciej>
the Windows UI is an archeological treasure trove

[02:07:05.0000] <othermaciej>
well I dunno about Win7

[02:07:06.0000] <othermaciej>
in Vista you can still find retro dialogs

[02:08:00.0000] <othermaciej>
it is possible they fixed it all for 7, though I suspect some of the dialogs were in programs where MS may have lost the source code or something

[02:08:01.0000] <Hixie>
if the win7 installer is any indication, they didn't.

[02:10:00.0000] <Philip`>
Hixie: In an all-array API, when you're dealing with data of a form you already know, all you have to do is write "item.properties.name[0]" instead of "item.properties.name", which doesn't seem too hard, and then your code won't randomly inexplicably fail when you make small changes to your data

[02:10:01.0000] <Hixie>
yeah

[02:10:02.0000] <Hixie>
it's really ugly

[02:10:03.0000] <Hixie>
having [0]s everywhere

[02:10:04.0000] <Hixie>
and i always forget to include them

[02:10:05.0000] <Hixie>
(e.g. with gEBTN i always forget the [0])

[02:12:00.0000] <jgraham>
/me always forgets the [0] too but still prefers APIs which are consistent over ones where you always have to put in explicit type checks

[02:12:01.0000] <Hixie>
you still have to type check even with the [0]

[02:12:02.0000] <Hixie>
(you have to check for null)

[02:12:03.0000] <Hixie>
so it doesn't actually gain you much

[02:12:04.0000] <Philip`>
"[0]" is less ugly than "if (!x instanceof Array) x = [x];"?

[02:13:00.0000] <Hixie>
yes, but "" is less ugly than "[0]"

[02:13:01.0000] <Hixie>
and that's the common case

[02:13:02.0000] <Hixie>
the generic case isn't the common case

[02:13:03.0000] <Philip`>
It seems a very common pattern to have some property which can occur one or more times

[02:13:04.0000] <Philip`>
and that's the case where (in the current API) you can't tell whether you're going to get back an array or a string

[02:13:05.0000] <Philip`>
and you have to do the instanceof thing

[02:14:00.0000] <Hixie>
so what api changes would you want?

[02:14:01.0000] <jgraham>
(FWIWW the interal API in my implementation effectively uses property:[values] where [values] is a list of one or more values, so it is more like the API Philip` is proposing

[02:15:00.0000] <jgraham>
)

[02:15:01.0000] <Hixie>
i could make HTMLPropertyCollection always return a PropertyNodeList 

[02:15:02.0000] <Philip`>
I haven't thought about this in any detail so I don't know if there's a better solution, but changing HTMLPropertyCollection.namedItem to always return an array (and maybe an empty array instead of null) would make me happier

[02:16:00.0000] <Philip`>
The JSON serialisation already makes everything an array, so it's no different to that

[02:16:01.0000] <Philip`>
(s/array/PropertyNodeList/ or whatever)

[02:16:02.0000] <jgraham>
Yeah that sounds like a good solution at first glance

[02:18:00.0000] <jgraham>
/me seems to have got his senses confused there

[02:19:00.0000] <Hixie>
ok, done

[02:21:00.0000] <jgraham>
(BTW I made some changes to my implementaion so that it has a slightly different set of bugs. I expect it is still very broken though)

[02:30:00.0000] <Hixie>
ok, updated the examples too

[02:44:00.0000] <aja_>
fixed the vcard example, too?

[02:45:00.0000] <Hixie>
what was wrong with it?

[02:45:01.0000] <aja_>
/me had system hange after mentioning it to you earlier

[02:45:02.0000] <aja_>
class=adr instead of itemprop

[02:46:00.0000] <Hixie>
yeah fixed that when you said it

[02:46:01.0000] <Hixie>
none of this is checked in yet

[02:46:02.0000] <aja_>
the jack bauer vcard.....afred's okay

[02:46:03.0000] <aja_>
alfred

[02:47:00.0000] <aja_>
cools

[02:47:01.0000] <aja_>
geez...can't friggin type today

[02:51:00.0000] <Hixie>
gotta love how win7 has a magical 16x16 hit box in the top left of windows that has no visible affordance but that shows a menu if you click it

[02:51:01.0000] <gsnedders>
Doesn't that exist going back to at least 95?

[02:52:00.0000] <gsnedders>
Hixie: Also: stop giving your cats drugs. They're crazy enough as is.

[02:53:00.0000] <Hixie>
in win95, there's an icon there

[02:53:01.0000] <gsnedders>
/me nowadays only uses the school's crippled version of Windows, and his memory isn't that great

[02:54:00.0000] <gsnedders>
/me only needs a text editor anyway :P

[02:59:00.0000] <Philip`>
http://en.wikipedia.org/wiki/File:Windows_2.0.png - they've had that top-left-corner icon for a long time

[03:03:00.0000] <Hixie>
i like(d) the icon, my complaint is just that now they have a ui element with no visual affordance

[03:06:00.0000] <Philip`>
Maybe it's for compatibility with programs that automatically open the menu by simulating a mouse click in the top-left 16x16 pixels of a window

[03:06:01.0000] <Hixie>
man, win7 has all kinds of weird bugs

[03:06:02.0000] <Hixie>
e.g. the UAC window has the wrong cursor for the path label

[03:06:03.0000] <Hixie>
Philip`: i'm sure it is

[03:06:04.0000] <Hixie>
Philip`: or compatibility with legacy users that do that, ilke me

[03:06:05.0000] <Hixie>
like, even

[03:07:00.0000] <Hixie>
and there's no way to enable UAC on high but have it not dim the desktop...? weird

[03:12:00.0000] <Philip`>
Does the screen still flash black when it brings up the UAC box?

[03:13:00.0000] <Hixie>
it does something similar

[03:13:01.0000] <Hixie>
fades it on display, and then fades back poorly on dismissal

[03:21:00.0000] <Hixie>
looks like actually most windows do have an icon, it's just explorer windows that don't

[03:23:00.0000] <virtuelv_>
Hixie: aren't path labels in Windows >= Vista just buttons?

[03:23:01.0000] <jgraham>
The default windows 7 UI is so /so/ ugly

[03:24:00.0000] <jgraham>
(that was irrelevant but still)

[03:24:01.0000] <Philip`>
virtuelv_: They turn into text fields if you click them in the right way, if I remember correctly

[03:25:00.0000] <virtuelv_>
Gnome/Nautilus is also doing the buttons thing, but never pretends to be both

[03:26:00.0000] <virtuelv_>
if you want to make it editable, you either press ctrl-l, or a button

[03:26:01.0000] <virtuelv_>
at which point, it becomes an <input type="text">

[03:28:00.0000] <Hixie>
virtuelv_: didn't do anything when i clicked it

[03:28:01.0000] <Hixie>
virtuelv_: but it had the hand cursor

[03:28:02.0000] <Hixie>
btw win7 seems to come with more cronjobs preconfigured than even linux installs

[03:28:03.0000] <Hixie>
which is impressive

[03:28:04.0000] <Hixie>
(33)

[03:30:00.0000] <Hixie>
[This content is preliminary and subject to change.]

[03:30:01.0000] <Hixie>
surely that shouldn't appear in a release candidate :-)

[03:30:02.0000] <Hixie>
(help for the memory checker tool)

[03:48:00.0000] <Hixie>
is there any end-user software that uses RDF? something that isn't a semantic web tool per se but actually does tasks a typical end user actually wants to do

[03:48:01.0000] <Hixie>
like booking plane flights or whatever

[03:52:00.0000] <zcorpan>
<form><span></form><input> does not seem to match browsers - browsers associate the input with the form, afaict

[03:52:01.0000] <jgraham>
Hixie: I don't know but it might help if you defined "uses" better. Do you mean "uses for input", "uses as its internal data model", "has the option to serialize to" or something else?

[03:53:00.0000] <Hixie>
uses for input

[03:56:00.0000] <zcorpan>
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/114

[04:00:00.0000] <jgraham>
Hixie: Maybe http://www.mkbergman.com/?page_id=325 is helpful (I selected category:harvester which I have to tell you since apparently they don't give queries unique uris)

[04:03:00.0000] <zcorpan>
though it says "null" in ie8

[04:03:01.0000] <zcorpan>
so html5 matches ie8

[04:05:00.0000] <zcorpan>
wait now it's null/undefined in webkit and firefox too... must have done something wrong before

[04:05:01.0000] <zcorpan>
aha

[04:05:02.0000] <zcorpan>
<form><span>foo</form><input>

[04:06:00.0000] <zcorpan>
makes it non-null in firefox

[04:06:01.0000] <zcorpan>
but still null in webkit and ie

[04:07:00.0000] <zcorpan>
<form><div></form><input> is non-null in webkit

[04:13:00.0000] <Hixie>
jgraham: i don't understand what that is

[04:13:01.0000] <Hixie>
jgraham: but it doesn't look like something my mum would use

[04:13:02.0000] <jgraham>
Hixie: It's a list of semantic web tools

[04:13:03.0000] <Hixie>
oh

[04:13:04.0000] <jgraham>
Well RDF tools I guess

[04:13:05.0000] <gsnedders>
Hixie: Well, in my mum's case, the whole web doesn't look like something she would use :P

[04:14:00.0000] <jgraham>
(it doesn't seem to work in Opera)

[04:14:01.0000] <Hixie>
jgraham: does it not work in safari?

[04:14:02.0000] <gsnedders>
Actually, s/web/computer/

[04:14:03.0000] <Hixie>
/me opens firefox

[04:14:04.0000] <jgraham>
Hixie: I don't know and finding out would mean starting my wwindows virtual machine which is... slow

[04:15:00.0000] <Hixie>
hm, works in ff

[04:15:01.0000] <Hixie>
doesn't work in safari.

[04:17:00.0000] <jgraham>
Hixie: To be fair nothing that I have seen listed thee looks like something my Mum would use

[04:17:01.0000] <jgraham>
(and my Mum is rather more computer literate than gsnedders' Mum)

[04:18:00.0000] <Hixie>
i can't even find a _category_ that my mum might use

[04:18:01.0000] <Hixie>
there has to be some average-end-user software SOMEWHERE that uses RDF as its input

[04:19:00.0000] <gsnedders>
Hixie: Google, nowadays? :P

[04:19:01.0000] <gsnedders>
Time for me to head off to school to do some last minute revision before a three-and-a-half hour long English exam. Yay. :\

[04:19:02.0000] <jgraham>
gsnedders: Good luck

[04:19:03.0000] <Hixie>
i've already looked at how google handles rdf internally, i'm looking for more data points

[04:20:00.0000] <gsnedders>
Hixie: Well, it's some average-end-user software, no? :P

[04:20:01.0000] <Hixie>
yeah but i'm looking for something else

[04:20:02.0000] <jgraham>
Hixie: There just aren't people publishing RDF for the kind of things that ordinary people do. So there is no need to build tools

[04:20:03.0000] <Hixie>
there are tools for end users that use xml

[04:20:04.0000] <Hixie>
why not tools for end users that use rdf?

[04:22:00.0000] <jgraham>
(I can vaugely imagine my Mum using something that allowed her to annotate web pages and there seem to be several RDF-based tools for doing that but I'm not really sure why the RDF is needed or whether, in fact my Mum would use one of those even in the unlikely event that it had perfect UI)

[04:23:00.0000] <Hixie>
would or would not?

[04:24:00.0000] <jgraham>
?

[04:24:01.0000] <Hixie>
"in fact my Mum would use one of those even..."

[04:24:02.0000] <Hixie>
my dad does all kinds of things with data graphs, like geneology, which i presume is well suited to being handled in RDF if anything is

[04:24:03.0000] <jgraham>
oh s/infact/in fact,/

[04:24:04.0000] <Hixie>
but i'm not aware of any end-user software that does geneology using rdf

[04:24:05.0000] <jgraham>
er s/in fact/in fact,/

[04:25:00.0000] <Hixie>
that still doesn't make much sense :-)

[04:26:00.0000] <Hixie>
(most geneology software seems to use GEDCOM as an interchange format)

[04:26:01.0000] <Philip`>
Are you missing the "whether"? :-)

[04:26:02.0000] <Hixie>
oh, i get it now

[04:26:03.0000] <Hixie>
thanks :-)

[04:26:04.0000] <pesla>
Does anyone of you know wether its possible to load HTML from a file on a FTP server using Javascript?

[04:26:05.0000] <pesla>
(Maybe a bit offtopic)

[04:27:00.0000] <Hixie>
it's possible if the javascript is itself running in a page from the same ftp server, i believe

[04:27:01.0000] <Hixie>
just use xhr

[04:28:00.0000] <pesla>
Is there cross browser support for xhr already?

[04:28:01.0000] <Hixie>
sure

[04:28:02.0000] <pesla>
Thx for the suggestion, didn't come to my mind yet ;)

[04:29:00.0000] <Hixie>
oh hey, that's exciting, i have a meeting in 7 hours and i haven't slept yet

[04:29:01.0000] <Hixie>
crap

[04:29:02.0000] <Hixie>
/me rushes to bed

[04:29:03.0000] <Hixie>
nn!

[04:30:00.0000] <jgraham>
/me guesses Hixie no longer wants to know that http://www.jenitennison.com/blog/taxonomy/term/34 has a lot of references to a geneology web 2.0 app using RDF but, afaict no actual link

[04:35:00.0000] <beowulf>
what's the point of the RDF in HTML5 stuff?

[04:36:00.0000] <Philip`>
There isn't RDF in HTML5 stuff

[04:36:01.0000] <beowulf>
oh

[04:36:02.0000] <Philip`>
There's just data in HTML5 stuff, and you can represent that data with RDF if you really want to

[04:38:00.0000] <beowulf>
why would you do that, then?

[04:39:00.0000] <jgraham>
Want to represent data in RDF? Not sure but some people seem to like it

[04:39:01.0000] <jgraham>
(I guess it might be appropriate for some applications where your model is a graph of interconnected nodes)

[04:40:00.0000] <beowulf>
i've worked on a couple of sites that used the UK eGMS standard for data in html, but i've never really understood to what end this activity was aimed

[04:42:00.0000] <jgraham>
beowulf: Isn't this the sort of thing that goverments just like even if it has no actual value? At least that was always the impression I got...

[04:44:00.0000] <beowulf>
jgraham: that's the impression i have, should i have the same impression of the microsyntaxes section of html5? (sorry if i've confused two seperate techs)

[04:46:00.0000] <jgraham>
beowulf: I can imagine some cases where it has the potential to be useful. e.g. if academic journals stopped using their own made up classnames for citation data and used the bibtex microdata spec instead it would be trivial to write a tool that extracted tose citations for inclusion in a LaTeX document

[04:46:01.0000] <Philip`>
<meta name="e-GMS.accessibility" content="NEED ICRA RATING. PLACE IN WEBENGINE.CFG" /> - hooray for metadata

[04:46:02.0000] <jgraham>
Which I would probably have done if people had been doing that two years ago

[04:47:00.0000] <jgraham>
s/done/writeen a tool to do/

[04:47:01.0000] <jgraham>
argh

[04:49:00.0000] <Philip`>
In fact, "NEED ICRA RATING. PLACE IN WEBENGINE.CFG" is the second most common value for e-GMS.accessibility, coming just behind "A" and just ahead of "Triple-A"

[04:49:01.0000] <Philip`>
(with Double-A falling in last place)

[04:50:00.0000] <Philip`>
jgraham: Why wouldn't you have just clicked the "view as BibTeX" links that online journals seem to all provide?

[04:50:01.0000] <jgraham>
Philip`: IIRC Nature doesn't

[04:50:02.0000] <jgraham>
and some others too

[04:50:03.0000] <Philip`>
Oh

[04:51:00.0000] <jgraham>
At one point Monthly Notices of The Royal Astronomical society got a new website that removed all the useful features of the old one, including BibTeX data

[04:51:01.0000] <Philip`>
/me almost never uses anything other than the ACM Portal and sometimes CiteSeer and Google Scholar

[04:52:00.0000] <beowulf>
Philip`: so most cases or that metadata are wrong and useless, and yet you're probably the only one who noticed

[04:52:01.0000] <beowulf>
s/or/for

[04:52:02.0000] <jgraham>
(but fortunatley NASA provided the Atrsophysics Data System which almost always had BibTeX data, but it did mean finding the article again)

[04:53:00.0000] <beowulf>
/me is still scared from having to read the eGMS stuff, never mind use it

[04:53:01.0000] <jgraham>
beowulf: metadata is sure to be wrong if no one is using it and merely likely to be wrong otherwise

[04:53:02.0000] <Philip`>
beowulf: (Admittedly this is only 11 pages, so the margin of error in the results are quite high)

[04:53:03.0000] <Philip`>
*margins

[04:54:00.0000] <Philip`>
((by which I mean this is 425K pages, of which 11 use e-GMS.accessibility))

[04:55:00.0000] <Philip`>
(and a handful of others use e-GMS for some keywords)

[04:56:00.0000] <Philip`>
(Actually, all of those handful are on the site http://www.youngscot.org/ which repeats the huge list of keywords five times on every page)

[04:57:00.0000] <Philip`>
(Metadata is great in theory)

[04:59:00.0000] <beowulf>
like a laser, then?

[05:01:00.0000] <Philip`>
HTML5 microdata seems to be more about replacing microformats with something that can parsed with a generic parser

[05:01:01.0000] <Philip`>
s/can/can be/

[05:01:02.0000] <Philip`>
and that doesn't have to squeeze itself into attributes where it doesn't really fit

[05:02:00.0000] <Philip`>
and tha has a decentralised mechanism for avoiding name conflicts

[05:02:01.0000] <Philip`>
*that

[05:07:00.0000] <Philip`>
/me can never decide whether something relying on DNS is centralised or decentralised

[05:08:00.0000] <hober>
I think @class has the same story as @item insofar as disambiguation is concerned. Which is to say, those concerned about disambiguation can choose sufficiently disambiguous names for their own vocabularies.

[06:01:00.0000] <jgraham>
"I will never support microdata, because no case has been made for its existence." That seems logically flawed

[08:15:00.0000] <Philip`>
XMLLiteral handling in text/html seems a bit inconsistent at the moment - for the same input in three implementations I get:

[08:15:01.0000] <Philip`>
  "Test<i>Test<br/>Test</i>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>

[08:15:02.0000] <Philip`>
  "Test<i>Test<br>Test</i>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>

[08:16:00.0000] <Philip`>
  "Test<I>Test<BR>Test</I>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>

[08:16:01.0000] <Philip`>
  "TestTestTestTest"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>

[08:16:02.0000] <zcorpan>
3 impl, 4 results?

[08:17:00.0000] <Philip`>
Yes (since some were JS implementations, and vary between browsers)

[08:17:01.0000] <zcorpan>
ah

[08:21:00.0000] <zcorpan>
Philip`: btw, had you sent email about triples changing after following an internal link? should the fragment be omitted from "the document's current address"?

[08:21:01.0000] <Philip`>
Using the same implementations on XHTML, I get:

[08:21:02.0000] <Philip`>
  "Test<i xmlns=\"http://www.w3.org/1999/xhtml\">Test<BR/>Test</i>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>

[08:21:03.0000] <Philip`>
  "Test<i xmlns="http://www.w3.org/1999/xhtml">Test<BR xmlns="http://www.w3.org/1999/xhtml"/>Test</i>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>

[08:22:00.0000] <Philip`>
zcorpan: Oh, I forgot about that and haven't sent anything

[08:25:00.0000] <Philip`>
/me notes that those XHTML bits are correct, because the input had a "<BR/>" in it

[08:28:00.0000] <hsivonen>
where's the parsing spec for XML literals (that don't seem to have a root element)?

[08:41:00.0000] <hsivonen>
http://www.w3.org/TR/rdf-concepts/#section-XMLLiteral

[08:42:00.0000] <hsivonen>
wow. is that really saying that XMLLiterals must be exclusively canonicalized to be conforming?

[08:44:00.0000] <hsivonen>
it seems there's no explicit parsing spec but it's implied that the precessing is similar to setting innerHTML in XHTML

[09:07:00.0000] <zcorpan>
wouldn't it be <span item><span itemprop=w3c.time propvalue="12-25-2009"> ?

[09:09:00.0000] <hsivonen>
zcorpan: why the reverse endianness?

[09:09:01.0000] <hsivonen>
or permuted rather than reversed

[09:12:00.0000] <Philip`>
zcorpan: It would (according to the current spec)

[09:12:01.0000] <Philip`>
(since itemprops correspond to the nearest ancestor item, and an element is not its own ancestor)

[09:14:00.0000] <zcorpan>
hsivonen: copy-paste from what jonas wrote

[10:22:00.0000] <gsnedders>
w00t! English! Done!

[10:24:00.0000] <hsivonen>
gsnedders: did it go well?

[10:25:00.0000] <Philip`>
gsnedders: Did you fail miserably?

[10:31:00.0000] <gsnedders>
hsivonen: The first part, definitely. The second part, not so well, but I think well enough.

[10:32:00.0000] <gsnedders>
Philip`: Hence no.

[10:32:01.0000] <gsnedders>
Philip`: Also, seeming I apparently had the best dissertation it'd be hard to fail badly, seeming I scarcely had to get any marks to pass

[10:32:02.0000] <hsivonen>
gsnedders: btw, there's no "XHTML WG"

[10:32:03.0000] <hsivonen>
(Re: twitter)

[10:33:00.0000] <gsnedders>
Peh. XHTML2 WG. You know what I mean.

[10:36:00.0000] <Philip`>
gsnedders: Were the people marking your dissertation impressed by the markup?

[10:36:01.0000] <gsnedders>
Philip`: It wasn't markup :P

[10:37:00.0000] <Philip`>
gsnedders: I thought that's what you were spending all your time doing? :-)

[10:37:01.0000] <gsnedders>
Philip`: No. Markup for my notes for English in general

[10:37:02.0000] <gsnedders>
Also: my fingers hurt from writing for three hours so I don't want to type too much

[10:38:00.0000] <Philip`>
Type with other appendages

[10:41:00.0000] <Philip`>
/me gets an impression from the "Link rot is not dangerous" topic that namespace URIs are quite a fragile foundation

[10:42:00.0000] <Philip`>
so they suggest building other structures on top of that, like caching and redirecting and hardcoding override lists and reminding people not to accidentally let their domains expire and making local subclasses

[10:42:01.0000] <hsivonen>
Philip`: it seems to be that believing in Follow your Nose and believing in Link Rot not being dangerous are contradictory beliefs but you can pick either one and argue coherently

[10:43:00.0000] <Philip`>
and I suppose it makes me wonder instead whether it'd be a good reason to not use that foundation at all

[10:44:00.0000] <Philip`>
(though I don't know what other foundations would be better)

[10:45:00.0000] <hsivonen>
Philip`: Java-style FQNs avoid the temptation to succumb to Follow Your Nose

[10:46:00.0000] <Philip`>
hsivonen: They still suffer from a mild form of Link Rot, because if I start using com.example.foo and then accidentally let com.example expire and someone else registers it and builds a new application based on that domain then they might conflict with my names and people will think my names belong to them

[10:47:00.0000] <Philip`>
The problem is that an object with a certain lifetime is depending on an object with a potentially shorter lifetime

[10:48:00.0000] <Philip`>
and C++ teaches us that such relationships will eventually cause horrible pain and memory corruption

[10:48:01.0000] <hsivonen>
Philip`: more likely, the owner of com.example is acquired by another company or the product is made open source under org.apache

[10:49:00.0000] <Philip`>
That too

[10:49:01.0000] <hsivonen>
but yeah, Java FQNs suck because code outlasts organization structure

[10:49:02.0000] <Philip`>
I wonder if all the com.sun.* Java classes will be renamed to com.oracle.*

[10:53:00.0000] <Philip`>
Has anyone (in distributed systems in general, not just the web) solved the problem of having identifiers that are globally unique and eternally unique, and that don't depend on a centralised registry, and that are human-readable?

[10:54:00.0000] <Philip`>
(I suppose I consider mildly distributed registries, like the IP number assignment system, to be similar enough to a centralised registry that it's not acceptable either)

[10:55:00.0000] <Philip`>
If not, I guess the question is which of those requirements to sacrifice

[10:55:01.0000] <hsivonen>
Philip`: depends on probablility of collision

[10:55:02.0000] <hsivonen>
Philip`: openid.delegate seems to strike the right balance

[11:05:00.0000] <hober>
Python package names seem to strike a good balance

[11:09:00.0000] <hsivonen>
hober: yeah, they are basically the same as openid.* and openid2.* rel values

[11:10:00.0000] <Philip`>
But Python lets you manually override the mapping from identifier to implementation (e.g. by changing PYTHONPATH or moving libraries into the current directory) if you need to resolve conflicts in exceptional cases

[11:10:01.0000] <Philip`>
just like XML Namespace prefixes do

[11:11:00.0000] <Philip`>
Without some mechanism for conflict resolution, it's not a robust way to avoid problems caused by naming conflicts

[11:12:00.0000] <hsivonen>
Philip`: how often do python names actually collide?

[11:15:00.0000] <Philip`>
It's not terribly unlikely that I would e.g. create a physics simulation program with a file called "atom.py" containing my Atom class, and say "import atom"

[11:15:01.0000] <Philip`>
and it would collide with the 'atom' package (for RFC 4287 style Atom)

[11:16:00.0000] <hober>
I think python package names, or java FQNs, etc. provide enough disambiguation

[11:16:01.0000] <Philip`>
(The conflict there is easily resolved because Python will look for ./atom.py before it looks for /usr/lib/python2.5/site-packages/atom, if I remember correctly)

[11:16:02.0000] <hober>
People seem to want more disambiguation than they really need, to get some kind of warm-fuzzy feeling of false security

[11:17:00.0000] <Philip`>
Why is it false security? (as opposed to unnecessary (but true) security)

[11:17:01.0000] <hober>
But given author error, cut-and-paste, etc. the straightjacket solutions to disambiguation don't really get you any more disambiguation than the "sufficiently-interesting prefix" solution

[11:27:00.0000] <gsnedders>
Is there any way to have a form that can be submitted to multiple URLs depending on what button is pressed, without script?

[11:29:00.0000] <gsnedders>
I guess @value of the submit button

[11:52:00.0000] <Philip`>
"link rot within the RDF world is an extremely rare and unlikely occurrence"

[11:52:01.0000] <Philip`>
That sounds like a testable observation

[11:52:02.0000] <Philip`>
and the test seems to refute that observation

[11:54:00.0000] <jgraham>
/me had python package name conflicts when he tried to build V8 so it does happen and can be confusing

[11:54:01.0000] <Philip`>
I see several 404s and several servers-not-responding, from the few dozen mosts common RDF namespace URIs

[11:54:02.0000] <Philip`>
s/mosts/most/

[11:56:00.0000] <Philip`>
(There are so many servers-not-responding that my download script is taking forever :-/ )

[12:06:00.0000] <Philip`>
I really should have reduced curl's timeout

[12:51:00.0000] <gsnedders>
/me dreams of a world in which HTML was sane

[13:03:00.0000] <gsnedders>
http://direct.tesco.com/search/default.aspx?search=%EF%BF%BF

[13:03:01.0000] <gsnedders>
Oh how boring. They fixed it.

[13:03:02.0000] <Dashiva>
"Unexpected error" is a fix? :)

[13:03:03.0000] <gsnedders>
Dashiva: Well, at least what it returns is well-formed

[13:04:00.0000] <Dashiva>
Fair enough

[13:19:00.0000] <jgraham>
/me wonders ifmoving the predefined microformats into their own small specs would keep people happy whilst having few downsides

[13:22:00.0000] <hsivonen>
jgraham: what WG would publish those?

[13:22:01.0000] <jgraham>
HTML WG? I guess it might be outside our charter... 

[13:23:00.0000] <gsnedders>
webapps. Everything that doesn't have a proper WG goes there. :P

[13:23:01.0000] <jgraham>
(In practice I can't imagine that Sam would dicourage it at least)

[13:28:00.0000] <Philip`>
Is Shelley failing to notice that any random person could start using xmlns:foo="http://www.w3.org/foo" even if the W3C has got nothing to do with it? I don't see why URLs would prevent that whereas reversed domains wouldn't

[13:28:01.0000] <Philip`>
hsivonen: Publish them on wiki.whatwg.org

[13:30:00.0000] <gsnedders>
Can you reply to multiple emails at once in Thunderbird?

[13:30:01.0000] <gsnedders>
/me needs a sane email client

[13:30:02.0000] <gsnedders>
/me is on the edge of giving up on GUIs and moving to alpine

[13:32:00.0000] <Dashiva>
Philip`: I'd guess so. Sacred URLs seem to be a common thing in semweb :)

[13:32:01.0000] <hsivonen>
is there a corollary for Godwin's law for "I can't wait for the lawsuits on this one."?

[13:37:00.0000] <jgraham>
I have concluded that there is no point in discussing things with Shelly. She has stated that she is not interested in technical merit and will not change her opinion regardless of facts so I don't see that she has anything productive to contribute

[13:38:00.0000] <jgraham>
Or at least there is nothing productive to be gained from calling her out when she is making falicious arguments because she is uninterested in facts, only in seeing her chosen solution suceed

[13:41:00.0000] <jgraham>
(I guess there might be something gained by the side effect of someone who is interested in rational discussion seeing the issues but I think overall it is better to discourage people who are unwilling to be reasonable)

[13:44:00.0000] <Philip`>
Even if someone will not be convinced by any arguments, they might still bring up relevant points that are worth discussing

[13:45:00.0000] <Philip`>
so it seems useful to respond to those points, but not to respond if you're just repeating past arguments rather than finding anything new

[13:47:00.0000] <jgraham>
Philip`: That was what I ws trying to say in the parenthetical above

[13:48:00.0000] <jgraham>
Although I'm not sure I am right that the value of discouraging pure advocacy outweighs the value of discussing all relevent points when those two goals conflict

[13:49:00.0000] <Philip`>
jgraham: Indeed, but I started responding during a respawn break in TF2 before your parenthetical, and didn't finish responding until the next respawn

[13:50:00.0000] <jgraham>
gsnedders: http://mailtweak.mozdev.org/tweaks.html#collect looks like it might help you

[14:04:00.0000] <Hixie>
oh hey, shelley quit in a huff again

[14:05:00.0000] <Hixie>
it must be 1pm

[14:37:00.0000] <Philip`>
http://bnode.org/blog/2009/05/15/could-microdata-work-better-for-me-than-rdfa - "it took me quite some time to admit that (intuitively desirable) URI abbreviations in HTML do have negative practical implications"

[16:31:00.0000] <gsnedders>
http://twitter.com/brucel/statuses/1810429136

[16:31:01.0000] <gsnedders>
Oh lovely.

[16:43:00.0000] <Philip`>
Just for reference purpopses: the context for jgraham's remarks was http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2009-May/019767.html, which seems to me to be a rant

[16:44:00.0000] <Philip`>
which isn't necessarily a bad thing, but in this case it makes reasonable debate seem very hard to achieve

[16:54:00.0000] <Philip`>
(...At least, I assume that was the context)


2009-05-16
[17:14:00.0000] <gsnedders>
Oh how boring. ian.hixie.ch has changed to something far less interesting.

[17:17:00.0000] <hober>
Hixie, re your tweet: http://microformats.org/wiki/hcard#Organization_Contact_Info

[17:22:00.0000] <Hixie>
@hober yeah but what do you do in real vCard?

[17:22:01.0000] <Hixie>
i can't find any docs on it

[17:22:02.0000] <Hixie>
what do hCard processors do when ouputting vCard?

[17:22:03.0000] <Hixie>
that uses that pattern?

[17:29:00.0000] <Hixie>
hmm, Address Book.app outputs N:;;;;

[17:30:00.0000] <Hixie>
fine i'll do that

[17:30:01.0000] <gsnedders>
Address Book.app does some weird things

[17:30:02.0000] <Hixie>
what other apps export vcard that i have access to?

[18:30:00.0000] <Hixie>
woo, there's now an algorithm for going from HTML to vCard http://www.whatwg.org/specs/web-apps/current-work/#extracting-a-vcard

[20:25:00.0000] <Hixie>
olliej: with() is useful in some very rare cases where you know exactly what the object's properties are and you want to do a lot of work with them (e.g. matrix maths)

[20:27:00.0000] <othermaciej>
using with() for matrix math would be terrible

[20:29:00.0000] <olliej>
Hixie: with is never ever good

[20:29:01.0000] <olliej>
Hixie: it defeats all optimisations, and results in absolutely absurd semantics

[20:30:00.0000] <olliej>
Hixie: and you need to know all properties on the object and its prototype chain

[20:31:00.0000] <othermaciej>
with is pretty much a bad idea in every possible way

[02:11:00.0000] <jgraham>
Philip`: It was more emails like http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2009-May/019751.html which say things like "I will never support microdata, because no case has 

[02:11:01.0000] <jgraham>
been made for its existence.

[02:11:02.0000] <jgraham>
"

[02:13:00.0000] <jgraham>
I assumed it was uncontroversial that saying you would never support something implied you were not prepared to consider arguments why you should support something

[02:14:00.0000] <jgraham>
(in this case one could, perhaps, claim that "because no case has been made" allows for the possibility of a case being made, but that is clearly  false satement because a case was made, just one that Shelley happeed to disagree with)

[02:19:00.0000] <hsivonen>
jgraham: I tried to point out in the CSSquirrel comments that if you say you don't care about the outcome, people pribably should be expected to focus on the feedback of people who do care about the outcome

[02:20:00.0000] <hsivonen>
jgraham: similarly, if you declare you won't budge from position X, it's not worthwhile for others to try to convinve you to accept an alternative solution X'.

[02:26:00.0000] <othermaciej>
CSSquirrel?

[02:28:00.0000] <hsivonen>
othermaciej: http://www.cssquirrel.com/2009/05/04/comic-update-html5-manners/#comment-23463

[02:56:00.0000] <hsivonen>
hmm. AOL not on the list at http://www.w3.org/Consortium/Member/List

[03:01:00.0000] <Philip`>
jgraham: Oh, okay

[03:12:00.0000] <Hixie>
 i hope the vcard output algorithm makes sense

[03:12:01.0000] <Hixie>
it's not quite what i implemented in JS

[03:13:00.0000] <hsivonen>
http://lists.w3.org/Archives/Public/www-archive/2009May/0038.html (last paragraph)

[03:16:00.0000] <Philip`>
Hixie: It seems to be defined at a very inconvenient level of abstraction - someone implementing the vCard output would have to have access to the DOM and the item-extraction algorithm, rather than it being layered on top of a generic microdata extraction algorithm

[03:17:00.0000] <Hixie>
how so?

[03:21:00.0000] <Philip`>
Someone who's writing a generic microdata parsing library doesn't want to have to implement the vCard output (and iCal and BibTeX and ...) themselves, because the point is to be generic and not hard-code lots of vocabularies

[03:22:00.0000] <Hixie>
sure but what's dom-dependent about it?

[03:22:01.0000] <Philip`>
and someone who's working with vCard data will want to use the generic microdata parsing library rather than rewriting everything themselves, and doesn't want to have to hook into internal details of the library (like whether the <title> element is going to be exposed via DOM or SAX or something else)

[03:24:00.0000] <Hixie>
does the same not apply to the RDF output?

[03:24:01.0000] <Philip`>
so there should be some clearly specified API between them, e.g. using JSON as an intermediate format (and giving a single JSON-to-vCard or RDF-to-vCard or something-else-to-vCard conversion algorithm, depending on what API is chosen)

[03:25:00.0000] <Hixie>
that seems excessively round-about

[03:25:01.0000] <Philip`>
Hixie: The RDF output is not vocabulary-specific, so that should go in the generic library (which can happily deal with all the issues like extracting <title> elements)

[03:26:00.0000] <Hixie>
i guess i could just define things that such an API is required to expose

[03:27:00.0000] <Philip`>
I don't think it has to be a serialised API like JSON, but it seems useful to have some clearly specified API that can be easily implemented by the generic library and is powerful enough to let vocabulary designers do whatever they want without having to modify the generic library

[03:28:00.0000] <Hixie>
which would be the list of top-level items, for each item: types, properties; for each property: names, the value, and if value is an item, or else if tag is time, or else if tag is a url element; the document's title; 

[03:28:01.0000] <Hixie>
(that's it so far i think)

[03:29:00.0000] <Philip`>
And the document's current address

[03:29:01.0000] <Hixie>
and the json and rdf outputs would need slightly more (e.g. json output needs to be able to walk the dom to find the applicable citation)

[03:29:02.0000] <Hixie>
well presumably the document's current address is input to the microdata api

[03:29:03.0000] <Hixie>
not output

[03:29:04.0000] <Hixie>
assuming this isn't a scripted environment

[03:30:00.0000] <Philip`>
/me thinks the JSON and RDF outputs are fine where they are, because they're not vocabulary-specific and so we don't have to worry about the scalability of implementing a thousand of them in the library

[03:45:00.0000] <Hixie>
i wonder why iCalendar isn't an official RFC2425 profile

[03:53:00.0000] <hsivonen>
Hixie: I also think that microdata-to-Foo conversion algorithms should operate on a microdata data model instead of escaping into the DOM on an ad hoc basis

[03:57:00.0000] <Hixie>
it's not really ad hoc

[03:58:00.0000] <Hixie>
assuming you're not talking about json or rdf outputs

[03:58:01.0000] <Hixie>
which kinda are

[03:58:02.0000] <Hixie>
what do you think of the new list at http://www.whatwg.org/specs/web-apps/current-work/#converting-html-to-other-formats

[04:16:00.0000] <hsivonen>
Hixie: the bulleted list lacks the node language of nodes that are itemprops

[04:17:00.0000] <hsivonen>
Hixie: btw, my on-list comment about 'about' and <figure> is likely misguided

[04:18:00.0000] <Hixie>
do i ever export the node language?

[04:18:01.0000] <hsivonen>
you did the day before yesterday

[04:18:02.0000] <hsivonen>
/me re-reads

[04:18:03.0000] <Hixie>
that's only in the RDF or JSON serialisations

[04:18:04.0000] <Hixie>
which aren't covered by this

[04:18:05.0000] <Hixie>
and need far more than the list says

[04:18:06.0000] <Hixie>
RDF only i think, in fact

[04:19:00.0000] <hsivonen>
"Otherwise, if element is not one of the URL property elements, let value be a plain literal, with the language information set from the language  of the element, if it is not unknown."

[04:19:01.0000] <hsivonen>
yes, RDF only

[04:20:00.0000] <Hixie>
(i've added it to the list but commented out)

[04:22:00.0000] <hsivonen>
special-casing "about" makes me wonder if there should be a dedicated boolean attribute itemabout for flagging the aboutness node instead of having a reserved word

[04:23:00.0000] <hsivonen>
aside: why oh why didn't microformats.org write algorithms like this down in 2006 or so?

[04:24:00.0000] <Hixie>
all of the one-word keywords are "reserved words"

[04:25:00.0000] <othermaciej__>
I would love microformats so much more if they'd defined parsing rules and clear data extraction algorithms

[04:25:01.0000] <Hixie>
i guess "about" is still magical even for item=""

[04:25:02.0000] <Hixie>
though only for RDF

[04:25:03.0000] <Hixie>
i don't want to add an attribute just to make it possible to name bnodes in the rdf serialisation, that's excessive

[04:26:00.0000] <Hixie>
(it's not like most people are going to be exporting to rdf, imho)

[04:26:01.0000] <othermaciej>
do generic RDF tools do anything useful?

[04:26:02.0000] <othermaciej>
(I honestly don't know)

[04:27:00.0000] <othermaciej>
It does seem that in some cases people use generic RDF tools to extract information which they then process in a vocabulary-specific way

[04:27:01.0000] <hsivonen>
Hixie: the use case I'm thinking of is not naming bnodes (not a use case) but associating key-value pairs with an external resource to give metadata about it

[04:28:00.0000] <Hixie>
<img itemprop=about src=... alt=""> seems better than having a magical boolean attribute

[04:28:01.0000] <Hixie>
the boolean attribute doesn't reduce the magic required, either

[04:28:02.0000] <Hixie>
sicne you still have to do error handling for dupes, not urls, etc

[04:29:00.0000] <hsivonen>
Hixie: ok. I withdraw my feedback about 'about'

[04:29:01.0000] <Hixie>
and if you exclude the RDF case, "about" isn't magic, it's just a property name common to many vocabularies

[04:29:02.0000] <Hixie>
othermaciej: no idea, i've been looking for end-user RDF-consuming software without any luck for the past few days

[04:30:00.0000] <hsivonen>
Hixie: you should make property names that start with http://www.w3.org/1999/xhtml/custom# non-conforming

[04:30:01.0000] <othermaciej>
I think the kind of tool I described couldn't possibly be an end-user tool, or at least not a good one

[04:30:02.0000] <hsivonen>
Hixie: otherwise, there can be conforming property names that are distinct in JSON but not in RDF

[04:31:00.0000] <othermaciej>
because you can't make a good UI for displaying "generic" data

[04:31:01.0000] <Hixie>
/me wonders what level of arcane magic he's going to have to do to output "rdate" properly in the iCalendar output and ponders just not supporting it in a clever way, requiring that people who use its longer forms use the iCalendar syntax directly instead of using <time>

[04:31:02.0000] <Hixie>
hsivonen: is that a "real problem" as they say?

[04:31:03.0000] <hsivonen>
Hixie: as real as many other conformance reqs

[04:32:00.0000] <Hixie>
hsivonen: fair enough.

[04:32:01.0000] <Hixie>
othermaciej: at least not with current technology

[04:32:02.0000] <Hixie>
hsivonen: i hate having rdf-specific rules in the non-rdf part of this

[04:32:03.0000] <Hixie>
hsivonen: isn't there a better solution?

[04:32:04.0000] <hsivonen>
Hixie: also, an easy error to make if one tries to put a once extracted RDF graph back into HTML

[04:33:00.0000] <hsivonen>
Hixie: can't think of one right now

[04:34:00.0000] <hsivonen>
http://twitter.com/JeniT/status/1806014001

[04:35:00.0000] <Hixie>
i wish we had a way for people to test how google interpreted rdfa, so that any bugs (if any) could be shaken out

[04:37:00.0000] <Philip`>
I also wish we had a specification that said what was a bug and what was correct

[04:38:00.0000] <Hixie>
i'm sure shane and the rdfa community will address your comments and fix the problems you pointed out

[04:38:01.0000] <Hixie>
hsivonen: ok, banned those prefixes

[04:38:02.0000] <hsivonen>
Hixie: thanks

[04:39:00.0000] <Hixie>
ok since i'm not doing magic for rdate, icalendar output is going to be easy

[04:44:00.0000] <Philip`>
Hixie: The comments seem unresolvable in an HTML4+RDFa specification, and nobody has shown great interest in writing an HTML5+RDFa specification, so I don't know where the comments would get addressed

[04:46:00.0000] <Philip`>
At least Shane's statement that "The goal here is that all implementations extract the same collection of triples from a given document." sounds exactly like the goal of an HTML5+RDFa specification (as long as it's not restricted to valid documents, or to static documents)

[04:46:01.0000] <Hixie>
well i presume he'd have to define HTML4 parsing and DOM representations

[04:47:00.0000] <Philip`>
I think that counts as "unresolvable" :-p

[04:48:00.0000] <Hixie>
it's obviously not unresolvable since we did it for html5

[04:48:01.0000] <Hixie>
and it's defined for microdata

[04:48:02.0000] <Hixie>
so it seems reasonable to expect the same of other proposals

[04:48:03.0000] <Hixie>
i don't see how you could achieve his stated goals, either

[04:48:04.0000] <Philip`>
s/unresolvable/practically unresolvable/

[04:49:00.0000] <Hixie>
it's impractical not to resolve it

[04:49:01.0000] <Philip`>
Sure

[04:50:00.0000] <Philip`>
so HTML4+RDFa is impractical

[04:50:01.0000] <Philip`>
and seems quite pointless, because all it really provides is a DTD, and you could use an XHTML DTD instead if you just want validation

[04:51:00.0000] <hsivonen>
Philip`: it provides a DTD and it tells you to ignore certain output of the DTD-based validator

[04:52:00.0000] <Philip`>
But his stated goal suggests that he might be receptive to an HTML5+RDFa specification, since it's the only way to achieve his stated goals, so that suggests it could be possible to move in the right direction in the future

[04:53:00.0000] <Philip`>
s/repetition//

[04:56:00.0000] <Philip`>
(But that seems unlikely to happen unless someone familiar with HTML5 writes it, because anyone else would find it very hard to understand the HTML5 processing model in enough detail without spending ages learning it)

[04:57:00.0000] <Hixie>
it would not be inappropriate for people working on core w3c technologies to learn how the web's most core language works

[05:00:00.0000] <othermaciej>
I'm not sure everyone agrees that HTML is the Web's most core language

[05:00:01.0000] <Hixie>
most deployed language?

[05:01:00.0000] <othermaciej>
that seems harder to dispute :-)

[05:01:01.0000] <Philip`>
Defining an HTML processing model for RDFa is kind of diving in the deep end - it's not a nice gradual learning curve, and so people are more likely to ignore it and stick with what they're familiar with

[05:01:02.0000] <othermaciej>
(though people have certainly tried to claim that public deployment numbers are not representative of what you'd find if you could survey intranets)

[05:06:00.0000] <Hixie>
othermaciej: imho "the web" doesn't include intranets.

[05:07:00.0000] <hsivonen>
http://www.guha.com/mcf/vocab.html

[05:07:01.0000] <othermaciej>
I would generally agree with that sentiment, if only for methodological reasons

[05:07:02.0000] <hsivonen>
Hixie: but it does include your local hard drive?

[05:07:03.0000] <othermaciej>
it is very easy to make unverifiable claims about the contents of intranets

[05:07:04.0000] <Hixie>
no, why would it?

[05:08:00.0000] <othermaciej>
I do think specific claims about intranet content that are supported by evidence may be worth considering

[05:08:01.0000] <hsivonen>
Hixie: my memory may be failing me

[05:09:00.0000] <othermaciej>
even if not at the same weight as the public Web

[05:09:01.0000] <othermaciej>
since a controlled environment has less need for open-ended interoperability

[05:09:02.0000] <Philip`>
hsivonen: You should use ECC memory

[05:09:03.0000] <hsivonen>
Philip`: or krijnh's logs

[05:09:04.0000] <othermaciej>
so maybe it's just me, but Sam often seems to have more interesting things to say about what's going on in the HTML WG on his blog than in public-html

[05:14:00.0000] <Hixie>
ok, icalendar output support is done

[05:14:01.0000] <Hixie>
i'll do bibtex probably sunday

[05:15:00.0000] <Hixie>
http://www.whatwg.org/specs/web-apps/current-work/#icalendar

[05:23:00.0000] <Hixie>
nn

[05:23:01.0000] <hsivonen>
Hixie: my menory was failing, sorry: http://krijnhoetmer.nl/irc-logs/whatwg/20081201#l-255

[05:37:00.0000] <hsivonen>
http://benlog.com/articles/2009/05/14/incremental-benefit-and-bursts-of-innovation/

[11:28:00.0000] <gsnedders>
Maths help needed: if f(x) = cos 2x, find f'(\frac{2\pi}{3})

[11:28:01.0000] <gsnedders>
f'(x) = -2sin 2x, right?

[11:29:00.0000] <gsnedders>
And then if I set x = \frac{2\pi}{3} I don't get the answer in the back of book when I enter it in to my calculator :(

[11:29:01.0000] <Philip`>
gsnedders: http://www.wolframalpha.com/input/?i=cos+2x

[11:30:00.0000] <Philip`>
That says d/dx(cos(2 x)) = -2 sin(2 x)

[11:30:01.0000] <gsnedders>
i.e., is agrees with me

[11:31:00.0000] <Philip`>
then http://www.wolframalpha.com/input/?i=sin%284pi%2F3%29 says -sqrt(3)/2

[11:31:01.0000] <Philip`>
so you'd get sqrt(3)

[11:31:02.0000] <gsnedders>
Where does the negative go?

[11:31:03.0000] <Philip`>
Which negative?

[11:32:00.0000] <gsnedders>
-sqrt(3)/2

[11:32:01.0000] <Philip`>
It goes at the front

[11:32:02.0000] <gsnedders>
:P

[11:32:03.0000] <Philip`>
Do you mean something else?

[11:32:04.0000] <gsnedders>
Why is the final answer not negative?

[11:32:05.0000] <Philip`>
Because you multiply by -2

[11:33:00.0000] <gsnedders>
Now, cheating and using wolfram alpha doesn't actually help me learn this.

[11:33:01.0000] <Dashiva>
You can still use Philip`'s steps

[11:33:02.0000] <Philip`>
I don't think you need Wolfram Alpha to tell you that -2 * -1/2 = +1

[11:33:03.0000] <gsnedders>
OK, no

[11:33:04.0000] <Dashiva>
Derivative, then calculate for just sin, then multiply by -2

[11:34:00.0000] <Dashiva>
I suppose the hard step is memorizing sin for the various pi thirds :)

[11:35:00.0000] <gsnedders>
Oh, duh

[11:35:01.0000] <gsnedders>
My calculator is in degrees

[11:35:02.0000] <Philip`>
Memorising sin is easy - it's always 0, 1, 1/2 or sqrt(3)/2, or sometimes negative, and sin(pi/3)=sin(30 deg)=1/2 and all the others fit around it in the obvious places

[11:36:00.0000] <Philip`>
Um

[11:36:01.0000] <Philip`>
That's not right is it

[11:36:02.0000] <Philip`>
sin(pi/6)=sin(30 deg)=1/2

[11:37:00.0000] <gsnedders>
Hmm, Wolfram|Alpha says it's raining here. I would disagree with that, looking out of the window.

[11:37:01.0000] <Dashiva>
Philip`: Yeah, I use the same rule

[11:37:02.0000] <Dashiva>
And cos is just sin flipped

[11:39:00.0000] <Philip`>
http://www.wolframalpha.com/input/?i=%ef%bf%bf - that's the most useful response I've ever seen to that query string

[11:39:01.0000] <Philip`>
It's a little silly how the encodings table is an image, though

[11:40:00.0000] <Dashiva>
But it isn't correct, is it? :)

[11:40:01.0000] <Philip`>
Why not?

[11:41:00.0000] <Dashiva>
It should give pages about input sanitation and such

[11:42:00.0000] <Philip`>
Why should it do that? It's not a search engine

[11:43:00.0000] <Dashiva>
I wouldn't expect a search engine to, though

[11:43:01.0000] <gsnedders>
Dashiva: What would you expect a search engine to do? Throw a fatal error?

[11:44:00.0000] <Philip`>
http://www.wolframalpha.com/input/?i=p%3Dnp - alas, there are some things it can't compute yet

[11:44:01.0000] <Dashiva>
gsnedders: Give pages about BOM

[11:44:02.0000] <Dashiva>
"Functionality for this topic is under development..."

[11:44:03.0000] <Dashiva>
They'll have the answer soon, in other words

[11:44:04.0000] <gsnedders>
hah

[11:47:00.0000] <Philip`>
Hmm, given a short list of primes it lists primes as a possible closed form, but it doesn't work for more complex sequences like what http://www.research.att.com/~njas/sequences/ does :-(

[11:54:00.0000] <gsnedders>
Our maths teacher showed us that in class

[11:56:00.0000] <Philip`>
I've found it to be very occasionally very useful

[11:58:00.0000] <gsnedders>
(This was the former head of maths, now retired, who always went off on tangents from his lesson plans, and taught us an amazing amount of stuff in the course out of order, and an amazing amount of stuff totally randomly)

[12:03:00.0000] <gsnedders>
http://www.wolframalpha.com/input/?i=Ben+Lomond — I always thought it was a mountain…

[12:15:00.0000] <Philip`>
http://www.wolframalpha.com/input/?i=whatwg.org - "daily visitors: ~50,000" - really?

[12:18:00.0000] <Philip`>
http://www.wolframalpha.com/input/?i=w3.org - the "HTML element hierarchy" looks pretty

[12:26:00.0000] <gsnedders>
Philip`: Well, it says SimplePie.org has 61,000 per day, whereas stats tracker on the website says around 800

[12:28:00.0000] <hsivonen>
http://www.wolframalpha.com/input/?i=python

[12:30:00.0000] <hsivonen>
Philip`: why am I not seeing daily visitor stats on wolfram|alpha?

[12:32:00.0000] <Philip`>
http://www.wolframalpha.com/input/?i=sin%282x%29+cos%282y%29+from+0+to+10 - the plots are nice, but I can't work out how to change the y range

[12:33:00.0000] <Philip`>
hsivonen: I get "Web statistics for all of whatwg.org:" in the third bit, which has the visitor stats

[12:36:00.0000] <hsivonen>
3rd bit?

[12:41:00.0000] <Philip`>
The third section in the results page

[12:49:00.0000] <hsivonen>
I see no such thing

[12:51:00.0000] <Philip`>
Hmm

[12:52:00.0000] <Philip`>
What do you see?

[13:27:00.0000] <gsnedders>
/me has a breakneck revision timetable for maths

[13:27:01.0000] <gsnedders>
This is gonna be fun :\

[13:57:00.0000] <Hixie>
hsivonen: heh, that irc log had above-average quality of content

[14:18:00.0000] <Hixie>
hsivonen: validator.nu bug report http://www.reddit.com/r/programming/comments/8kfux/why_everyone_should_dump_ie/c09layo

[14:20:00.0000] <gsnedders>
/me shouldn't try and use Wolfram|Alpha to help him with maths, because he doesn't have it on Thursday

[14:20:01.0000] <Philip`>
Hixie: Sounds like you need to make the spec smaller

[14:25:00.0000] <gsnedders>
(Though having some method of calculating stuff that shows all the steps to get there is damned useful)

[14:29:00.0000] <Hixie>
Philip`: hah

[14:30:00.0000] <gsnedders>
Hixie: Maybe remove the parsing section?

[14:31:00.0000] <Philip`>
Hixie: You could remove all the whitespace

[14:32:00.0000] <gsnedders>
I mean, html5lib's serializer has an option for that :P

[14:33:00.0000] <Hixie>
i think i use that, no?

[14:39:00.0000] <Philip`>
There's all the wasted space between words too

[14:41:00.0000] <Philip`>
You don't need a whole character for that, you could just CamelCase everything

[14:43:00.0000] <gsnedders>
Hixie: no, you don't

[14:44:00.0000] <Hixie>
k

[14:44:01.0000] <gsnedders>
It will, however, make the index diffs totally unreadable

[15:04:00.0000] <Hixie>
they're not particularly readable now


2009-05-17
[02:20:00.0000] <hsivonen>
Hixie: the size of the spec prevents me from porting to AppEngine unless Google changes their limits, too :-)

[04:57:00.0000] <gsnedders>
If a function exists in the domain (0, 2], then it can take values 0 < x ≤ 2, right?

[04:57:01.0000] <gsnedders>
/me can't remember which way around round and square brackets are, and can't find it in his notes

[05:02:00.0000] <nessy>
yes, that's right

[05:02:01.0000] <nessy>
thought we learnt to us ] instead of ( on the left

[05:03:00.0000] <nessy>
makes it easier to remember

[05:06:00.0000] <gsnedders>
nessy: thx

[05:41:00.0000] <gsnedders>
When does a system of three equations have an infinite number of solutions?

[05:43:00.0000] <gsnedders>
/me finds that in his own notes

[06:59:00.0000] <drostie>
gsnedders: when they have more than three equations, or when they have three variables but don't describe three linearly-independent planes.

[06:59:01.0000] <drostie>
sorry, more than three variables.

[10:26:00.0000] <hendry>
could someone please review my html5 blog markup? http://natalian.org/ # also, how do I get <article> to work in FF or Webkit?

[10:32:00.0000] <hsivonen>
hendry: do you have display: block;?

[10:34:00.0000] <hendry>
hsivonen: ah, of course. :)

[10:35:00.0000] <hendry>
hsivonen: was a bit confusing how the border came up twice tho' there was one <article>

[10:36:00.0000] <hendry>
is there some html5 blog template? not sure what to do with the comments

[10:55:00.0000] <akamike>
hendry: I was thinking about how to mark-up comments a couple of days ago. I was wondering if asides would be appropriate for each comment, rather than article. While each comment can be an article, that would make it an article of an article.

[10:55:01.0000] <akamike>
An aside makes more sense since each comment is a piece of information that is related to the article but not necessarily key to that article. You don't have to read the comments to get the basic value of the article, but they can add additional value.

[11:04:00.0000] <gsnedders>
hendry: article for each comment

[11:04:01.0000] <gsnedders>
hendry: having each article nested within the post's article

[11:09:00.0000] <hendry>
gsnedders: Ok. What about tags?

[11:09:01.0000] <gsnedders>
hendry: span? :P

[11:09:02.0000] <gsnedders>
hendry: If a list of tags, ul

[11:10:00.0000] <hendry>
gsnedders: should tags be in the article or not, aside or what what

[11:10:01.0000] <gsnedders>
hendry: They should just be in the post's article

[11:11:00.0000] <hendry>
gsnedders: ok, let me know if you have any other suggestions on natalian.org please :)

[11:11:01.0000] <gsnedders>
hendry: That means bothering to look :P

[14:25:00.0000] <gsnedders>
/me should buy tickets for Avenue Q soon


2009-05-18
[20:37:00.0000] <mpilgrim>
i've completed my rel=feed research ( c.f. http://krijnhoetmer.nl/irc-logs/whatwg/20090416#l-507 )

[20:38:00.0000] <mpilgrim>
i sampled 3 billion web pages from google's latest index

[20:38:01.0000] <mpilgrim>
weeding out errors like rel='RSS 2.0 feed' and false positives like rel='service.feed',

[20:38:02.0000] <mpilgrim>
i found exactly 1 page that uses rel='feed' according to specification and to the exclusion of any other autodiscovery mechanism

[20:38:03.0000] <mpilgrim>
http://seiji.asia/

[20:39:00.0000] <mpilgrim>
and they have a visible link on their page that also links to their feed

[20:40:00.0000] <mpilgrim>
so there would be little harm in removing rel=feed support from the only browser that actually supports it

[20:40:01.0000] <mpilgrim>
and little harm in removing it from html 5

[00:15:00.0000] <zcorpan>
hendry: surely <div class="inlineheader"> <span class="header"> should be <h2>?

[00:16:00.0000] <zcorpan>
hendry: you could use <footer> in the posts, too

[01:43:00.0000] <jgraham>
Hixie: http://wiki.dbpedia.org/DBpediaMobile looks like it might be an end-user application that does something with RDF

[01:44:00.0000] <jgraham>
(although DBpedia seems to be down at the moment since all the other demo pages are not working for me)

[01:45:00.0000] <Hixie>
interesting

[01:45:01.0000] <Hixie>
i wonder how it handles types

[01:45:02.0000] <Hixie>
if you have data in the wrong type

[01:45:03.0000] <Hixie>
does it notice?

[01:45:04.0000] <Hixie>
e.g. if it has a date to display but it's not a date

[01:45:05.0000] <Hixie>
but a string formatted as a date

[01:57:00.0000] <Hixie>
i am amused that xml, which has a namespacing mechanism, has the */*+xml notation for MIME types, despite not needing it (due to the namespaces), but JSON, which does need it (you could have different types of JSON data that you can only differentiate out of band) only has one MIME typpe.

[02:06:00.0000] <annevk5>
<div item><img itemprop="image" src="google-logo.png" alt="Google"></div> -- what if you want an itemprop=name to equal Google here?

[02:14:00.0000] <Philip`>
<div item><meta itemprop=image content=Google><img src=google-logo.png alt=Google></div>

[02:14:01.0000] <annevk5>
k

[02:14:02.0000] <Philip`>
(which is ugly)

[02:15:00.0000] <annevk5>
yeah

[02:15:01.0000] <Philip`>
(and doesn't work in browsers that move <meta> to <head>)

[02:17:00.0000] <Hixie>
but which, unlike pretty much every other solution i considered, will actually be understood!

[02:18:00.0000] <annevk5>
the colon restriction for itemprop does not seem necessary

[02:18:01.0000] <Hixie>
what colon restriction?

[02:18:02.0000] <annevk5>
when item has no tokens itemprop cannot contain a colon

[02:19:00.0000] <Hixie>
sure it can

[02:19:01.0000] <Hixie>
it just has to be a url

[02:19:02.0000] <annevk5>
ah ok

[02:19:03.0000] <annevk5>
just thought of that :)

[02:20:00.0000] <annevk5>
from what I've seen so far microdata looks nice, but it seems most companies are going with some variant of RDFa...

[02:21:00.0000] <Philip`>
It seems many are going with RDFa+microformats, i.e. they don't really care what the input syntax is and will just use whatever they can easily convert into their internal data structures

[02:21:01.0000] <Hixie>
the most recent checkin is what microdata is really about

[02:21:02.0000] <Hixie>
all the rest is basically noise

[02:36:00.0000] <jgraham>
Hixie: So you punted on escaping lone braces in the bibtex output?

[02:36:01.0000] <Hixie>
i was told by users of bibtex that if they were quoted there was no need

[02:36:02.0000] <Hixie>
so i used quotes

[02:37:00.0000] <Hixie>
but really someone should get tex and reverse engineer wtf the spec should say

[02:37:01.0000] <jgraham>
Oh, I thought we decided that didn't work

[02:37:02.0000] <Philip`>
I'm almost entirely certain it didn't work

[02:38:00.0000] <Philip`>
(BibTeX understands it fine, but LaTeX breaks when trying to parse BibTeX's output, I think)

[02:38:01.0000] <jgraham>
(IIRC you had to define \lonerightbrace or similar)

[02:41:00.0000] <jgraham>
(Maybe the easiest solution is to ask some of the mathml people for review here? Isn't David Carlisle one of the LaTeX3 people?)

[02:42:00.0000] <Philip`>
(We're not allowed to ask experts!)

[02:42:01.0000] <jgraham>
(that was sort of a rhetorical question since I just checked his homepage)

[02:42:02.0000] <Philip`>
(Experts are dangerous because they might know things)

[02:42:03.0000] <hsivonen>
Hixie: can one *input* RDF into DBpedia?

[02:43:00.0000] <Hixie>
jgraham: well if it needs a macro, screw that

[02:43:01.0000] <Hixie>
jgraham: we've already got much bigger problems with newlines in values and the like, if i'm not mistaken

[02:44:00.0000] <Hixie>
hsivonen: presumably

[02:44:01.0000] <hsivonen>
Hixie: how?

[02:44:02.0000] <Hixie>
no idea

[02:45:00.0000] <jgraham>
Hixie: Oh the newline thing could be interesting. Is it still the case that property values are taken from unnormalised text content?

[02:45:01.0000] <annevk5>
wasn't DBpedia created by scraping Wikipedia?

[02:45:02.0000] <hsivonen>
annevk5: yes. My understanding was that RDF was only the output of DBpedia, not input.

[02:45:03.0000] <jgraham>
annevk5: I think it is (in theory) frequently recreated by scraping wikipedia

[02:45:04.0000] <Hixie>
jgraham: currently, at least. i haven't looked at the feedback yet.

[02:46:00.0000] <annevk5>
hsivonen, mine too

[02:46:01.0000] <annevk5>
hsivonen, also seems that if you can create it automatically wikipedia could embed the necessary data itself

[02:46:02.0000] <jgraham>
http://wiki.dbpedia.org/Documentation

[02:48:00.0000] <Philip`>
http://www4.wiwiss.fu-berlin.de/dbpedia/dev/ontology.htm looks scary

[02:55:00.0000] <hsivonen>
Philip`: it's a tree, though

[02:56:00.0000] <hsivonen>
I'm amused that an ontology uses vague semantics like "Adult Actor" instead of clear terms like "Porn Actor"

[02:57:00.0000] <Philip`>
"The authors of HTML would like us to use <em> (emphasis) instead of <i> and <strong> instead of <b>. However, the frowned-upon <i> and <b> tags are more convenient to type, and have a clearer semantics (does emphasis mean italic or bold?)." (http://chaos.troll.no/~shausman/api-design/api-design.pdf)

[02:58:00.0000] <Hixie>
wow there's so much wrong with that i don't even know where to start

[02:59:00.0000] <hsivonen>
Hixie: that being the API design guide?

[02:59:01.0000] <Hixie>
Philip`'s quote

[03:00:00.0000] <Philip`>
hsivonen: They might just be reflecting the terminology used by Wikipedia (e.g. http://en.wikipedia.org/wiki/Template:Male_adult_bio)

[03:00:01.0000] <hsivonen>
Hixie: why? <em> and <strong> are simply longer synonyms for <i> and <b>

[03:01:00.0000] <Hixie>
no, they're not

[03:01:01.0000] <Hixie>
and the "authors of HTML" wouldn't like people to use <em> instead of <i>

[03:01:02.0000] <Hixie>
and <i> isn't frowned upon

[03:01:03.0000] <Hixie>
and it doesn't have clearer semantics

[03:01:04.0000] <Hixie>
and it doesn't mean italics or bold

[03:02:00.0000] <Hixie>
anyway, bed time for me now :-)

[03:02:01.0000] <Philip`>
That argument seems to be kind of ignoring reality

[03:03:00.0000] <jgraham>
Philip`: Hixie's argument?

[03:04:00.0000] <Philip`>
/me was just interested in how the quote's use of "semantics" seems to mean the opposite of what it usually means in the context of HTML

[03:04:01.0000] <Philip`>
jgraham: Yes

[03:04:02.0000] <jgraham>
Philip`: Yeah. I tend to agree with hsivonen on this point

[03:27:00.0000] <gsnedders>
are http://stuff.gsnedders.com/dif1.jpg and http://stuff.gsnedders.com/dif2.jpg right?

[03:27:01.0000] <gsnedders>
(This is maths)

[03:29:00.0000] <jgraham>
gsnedders: d/dx(2y dy/dx) = 2 (dy/dy)^2 d^2y/dx^2

[03:29:01.0000] <jgraham>
er

[03:30:00.0000] <gsnedders>
Where is this?

[03:30:01.0000] <jgraham>
d/dx(2y dy/dx) = 2 (dy/dy)^2 + 2y d^2y/dx^2

[03:30:02.0000] <gsnedders>
/me realizes his copy is still in the scanner

[03:30:03.0000] <jgraham>
dif2.jpg

[03:31:00.0000] <gsnedders>
Where does the ^2 come from?

[03:31:01.0000] <gsnedders>
(the first one)

[03:31:02.0000] <Philip`>
"dy/dy"?

[03:32:00.0000] <gsnedders>
That too

[03:32:01.0000] <Philip`>
/me assumes that should be dy/dx

[03:32:02.0000] <gsnedders>
/me thinks he needs a better maths teacher than jgraham :P

[03:32:03.0000] <gsnedders>
/me could go into school and see if any maths teacher is around

[03:32:04.0000] <gsnedders>
Apart from the fact that break starts now…

[03:32:05.0000] <jgraham>
gsnedders: You know I have other things to do right :)

[03:33:00.0000] <gsnedders>
jgraham: Yes :)

[03:33:01.0000] <Philip`>
gsnedders: d/dx(ab) = da/dx*b + a*db/dx, and you have a=2y and b=dy/dx so the first bit is d(2y)/dx*dy/dx = 2(dy/dx)^2

[03:33:02.0000] <jgraham>
gsnedders: What Philip` said

[03:34:00.0000] <Philip`>
gsnedders: (You had d(2y)/dx = 2)

[03:34:01.0000] <gsnedders>
Philip`: Like, what I have on the right hand side of the page?

[03:35:00.0000] <jgraham>
(also: you can "check dimensions" on this sort of thing e.g. if y was a distance and x was a time then the two terms should both have units of distance^2/time^2)

[03:35:01.0000] <jgraham>
(but remember that d^2 y has the same dimensions as y, not as y^2)

[03:35:02.0000] <Philip`>
gsnedders: Yes, except without the error

[03:36:00.0000] <Philip`>
gsnedders: You're saying "v' = 2", which is misleading because that's dv/dy and you want dv/dx instead

[03:37:00.0000] <gsnedders>
Philip`: u, not v :P

[03:37:01.0000] <Philip`>
gsnedders: Same thing :-p

[03:37:02.0000] <gsnedders>
Philip`: u' = 2 dy/dx

[03:37:03.0000] <gsnedders>
right?

[03:37:04.0000] <Philip`>
Yes

[03:38:00.0000] <gsnedders>
/me thought he probably try and write some notes to prove he actually understands this, and blatantly more or less does, just can't actually do it :P

[03:39:00.0000] <gsnedders>
(So yes, that's my messy handwriting)

[03:39:01.0000] <Philip`>
You should have done it all in MathML

[03:39:02.0000] <gsnedders>
I'm not hand-wriing MathML.

[03:39:03.0000] <gsnedders>
*writing

[03:39:04.0000] <jgraham>
gsnedders: On the other one, have you written dy/dx = dx/dy*dy/dx? Becuase that seems wrong...

[03:40:00.0000] <gsnedders>
That's equal to 1.

[03:40:01.0000] <gsnedders>
WTF?

[03:40:02.0000] <gsnedders>
/me tries to work out what he's on about

[03:40:03.0000] <Philip`>
That must mean y=x

[03:40:04.0000] <jgraham>
(maybe you meant dy/dx = dy/dk*dk/dx or something)

[03:40:05.0000] <Philip`>
It's all a conspiracy! They were the same variable all along

[03:41:00.0000] <othermaciej_>
I wonder if there are other solutions to that differential equation

[03:43:00.0000] <gsnedders>
jgraham: Ah, it's because you differentiate a term with respect to y then multiple by dy/dx

[03:43:01.0000] <gsnedders>
jgraham: That's what I'm on about

[03:43:02.0000] <gsnedders>
Also, it has reached that point of the morning when the sun shines straight through the south facing window above me blinding me

[03:43:03.0000] <gsnedders>
Apple blatantly should make brighter laptop screens

[03:44:00.0000] <Philip`>
You could turn it around

[03:44:01.0000] <Philip`>
or use curtains

[03:44:02.0000] <gsnedders>
But the other side of the desk is against the wall

[03:44:03.0000] <gsnedders>
Curtains sound like a better idea, but that means bothering to stand up to shut them

[03:44:04.0000] <othermaciej>
I wish we could make laptop screens as bright as the iphone screen

[03:44:05.0000] <Philip`>
Put it on a flat surface (e.g. the floor) behind you rather than on the desk

[03:45:00.0000] <jgraham>
I wish I understood what gsnedders just said to tell him if he is talking nonsense or not

[03:54:00.0000] <gsnedders>
jgraham: After that comment, I went to find more books of notes…

[03:55:00.0000] <gsnedders>
I ought to find out if we have to hand them back…

[03:56:00.0000] <gsnedders>
/me notices that according to his revision timetable he has no more time for differentiation

[04:25:00.0000] <hsivonen>
Yay. My site is an example of a real-world site for browser development: http://khtml-konqueror.blogspot.com/2009/05/kdekdelibskhtml_8374.html

[04:26:00.0000] <Philip`>
The title of that blog doesn't fill me with great confidence in KHTML

[04:28:00.0000] <othermaciej>
I'm not sure the remaining KHTML hackers have a lot of confidence either

[06:04:00.0000] <Philip`>
Hmm, xmlns:* in text/html interacts badly with XMLLiterals in RDFa

[06:04:01.0000] <Philip`>
because one would expect XMLLiterals to be produced by serialising the DOM to XML

[06:04:02.0000] <Philip`>
but if it's got attributes called "xmlns:*" in no namespace, it can't be serialised to XML

[06:05:00.0000] <Philip`>
s/in RDFa/, in RDFa/

[06:48:00.0000] <hsivonen>
how annoying that HTML5 parsing breaks Gecko reftests that incidentally rely on Gecko's HTML parsing peculiarities

[06:50:00.0000] <jgraham>
hsivonen: That seems weird. What fraction of the reftests are broken for that reason?

[06:51:00.0000] <hsivonen>
jgraham: I don't know yet

[06:51:01.0000] <jgraham>
OK, well do you know what fraction are broken at all?

[06:51:02.0000] <jgraham>
(since that would, at least, be an upper bound)

[06:52:00.0000] <hsivonen>
jgraham: no, I don't 

[06:52:01.0000] <hsivonen>
(I don't know how to get the total # of tests)

[07:12:00.0000] <gsnedders>
hsivonen: How about the # that are broken?

[07:12:01.0000] <gsnedders>
(or at least fail)

[07:14:00.0000] <hsivonen>
gsnedders: 36 reftest failing 34 undiagnosed so far

[07:14:01.0000] <hsivonen>
*reftests

[07:14:02.0000] <gsnedders>
At least you can manually check that number in a not totally insane amount of time

[07:16:00.0000] <Philip`>
/me would be much more worried if replacing the parser *didn't* cause a load of tests to fail

[07:22:00.0000] <hsivonen>
/me notes that both Gecko and IE8 don't hoist whitespace from after </body> into body

[07:22:01.0000] <hsivonen>
but WebKit and Opera do hoist

[07:22:02.0000] <hsivonen>
I wonder if that's Web-relevant

[07:38:00.0000] <hsivonen>
I wonder when Google is starting to do stuff like Wonder Wheel in SVG instead of Flash

[07:38:01.0000] <hsivonen>
s/starting/going to start/

[07:45:00.0000] <jgraham>
hsivonen: Maybe 3422 reftests? That is a really bad guess though and just comes from counting lines that don't start # in files called reftest.list using my extremely dubious bash skills

[07:52:00.0000] <hsivonen>
I guess reftests written solely for Gecko give a hint of how subtle dependencies there'd be if Web authors tested in only one browser

[07:52:01.0000] <Philip`>
And if Web authors were intentionally trying to highlight bugs in browsers

[07:52:02.0000] <hsivonen>
Philip`: the failing reftests so far have failed unintentionally

[07:53:00.0000] <hsivonen>
Philip`: i.e. they've tried to test something other than the thing that caused the failure

[07:53:01.0000] <jgraham>
Philip`: I was under the impression that web authors _were_ trying to highlight bugs in browsers. I mean the state of the web stretches "never attribute to malice what can be adequtely explained by incompetence" to beyond the limit.

[07:53:02.0000] <jgraham>
;)

[07:54:00.0000] <hsivonen>
so far the failures have been stuff like lone </p> being different, whitespace nodes getting pushed around differently and <colgroup> getting inferred differently

[08:05:00.0000] <hsivonen>
I expect trouble with colgroup inference on sites that use <col> and don't care about testing in IE anymore

[09:19:00.0000] <zcorpan_>
hsivonen: my copy of minefield-html5 fails http://simon.html5.org/test/html/parsing/form/form-association.html

[09:20:00.0000] <zcorpan_>
hsivonen: btw i've asked before that <colgroup> parsing (and the content model) be changed to allow lone <col>

[09:20:01.0000] <zcorpan_>
Philip`: maybe you should point that out on the list re "one if statement"

[09:27:00.0000] <Philip`>
/me gets bored and tries to find out how hard it'd be to write an RDFa-for-HTML5 specification

[09:31:00.0000] <jgraham>
You must be *really* bored...

[09:32:00.0000] <Philip`>
Given how many people are implementing it, it'd be kind of nice if there was some interoperability, and currently there isn't much

[09:38:00.0000] <Philip`>
http://philip.html5.org/docs/rdfa/ - in how many ways is this horridly broken and/or rubbish?

[09:42:00.0000] <zcorpan_>
Philip`: you should join the xhtml2 wg and rewrite all their specs

[09:43:00.0000] <Philip`>
zcorpan_: That doesn't seem especially useful for all of them, only for ones that are being actively used :-p

[09:43:01.0000] <Philip`>
Also, that would be too much work and I'd get bored before then

[09:44:00.0000] <zcorpan_>
Philip`: you should join the xhtml2 wg and rewrite all their specs that are being actively used until you get bored

[09:44:01.0000] <Philip`>
The latter condition would only last for a few days, I expect

[09:45:00.0000] <zcorpan_>
wait, didn't you start doing this because you were bored?

[09:45:01.0000] <Philip`>
Yes

[09:46:00.0000] <Philip`>
and soon I'll think of something else to do instead

[09:46:01.0000] <Philip`>
(That occasion might coincidentally coincide with the release of the TF2 sniper/spy update)

[09:50:00.0000] <zcorpan_>
Philip`: should "valid URL" be "valid URL that is an absolute URL"?

[09:52:00.0000] <Philip`>
zcorpan_: Apparently not, as far as I can tell from http://www.w3.org/TR/rdfa-syntax/

[09:52:01.0000] <zcorpan_>
ok

[09:53:00.0000] <Philip`>
It talks about resolving all URIs, and doesn't say they have to be absolute

[09:53:01.0000] <Philip`>
even for xmlns - "Note that it is generally considered a good idea not to use relative paths in namespace declarations, but since it is possible that an author may ignore this guidance, it is further possible that the URI obtained from a CURIE is relative. However, since all URIs must be resolved relative to [base] before being used to create triples, the use of relative paths should not have any effect on processing."

[09:54:00.0000] <zcorpan_>
seems silly to allow relative urls imho :)

[09:54:01.0000] <danbri>
the relative paths in xmlns thing was a huge war in XML land, about 9 nears ago...

[09:55:00.0000] <Philip`>
zcorpan_: I agree :-)

[09:55:01.0000] <danbri>
"seems silly to allow relative urls" is a good summary :)

[09:55:02.0000] <Philip`>
This isn't just about xmlns, though

[09:56:00.0000] <danbri>
yup

[09:56:01.0000] <Philip`>
It includes things like <div about="#me">

[09:56:02.0000] <Philip`>
which I suppose you do want to allow relative URLs for

[09:56:03.0000] <danbri>
but re xml, this did blow up. http://lists.w3.org/Archives/Public/xml-uri/  http://lists.w3.org/Archives/Public/xml-uri/2000May/ 

[09:56:04.0000] <danbri>
(...1700 messages in two months)

[09:57:00.0000] <Philip`>
I guess it's impossible to require anything of xmlns values, but requiring CURIEs to be absolute (after prefix processing) seems sane

[09:57:01.0000] <danbri>
yup

[09:58:00.0000] <Philip`>
danbri: Only 1700?

[09:58:01.0000] <danbri>
plus a few offlist i'm sure :)

[09:58:02.0000] <Philip`>
public-html had 1773 in *one* month :-)

[09:58:03.0000] <danbri>
i consider rdfa with curies to be about as fragile as Javascript ... but without the built-in "aw crap it doesn't work, what did I screw up?" corrective that comes with separating .js usage from .js declarations

[09:59:00.0000] <danbri>
...so i think we need some utilities that replace that part of the lifecycle. "aww crap it doesn't work in google/yahoo" might operate the same, if their indexes updated in real time or there were better checkers...

[10:00:00.0000] <Philip`>
Or just an HTML validator?

[10:00:01.0000] <Philip`>
Hmm, someone needs to invent a real-time search engine indexer update protocol

[10:00:02.0000] <danbri>
there are so many ways of making technically valid stuff that's still not v useful

[10:01:00.0000] <danbri>
but yeah, integration with normal validation ideally

[10:01:01.0000] <Philip`>
Maybe you could set up an <event-source> server that the indexers connect to, so you can immediately send them all your updates

[10:01:02.0000] <danbri>
http://www.sitemaps.org/ is sorta related, but i guess those sitemaps are just periodically polled

[10:02:00.0000] <Philip`>
What kinds of problems would still be technically valid?

[10:02:01.0000] <Philip`>
Typos in namespace URIs is the obvious one, I suppose

[10:02:02.0000] <danbri>
using an unfashionable property or old style URI

[10:02:03.0000] <danbri>
yeah 

[10:02:04.0000] <Philip`>
(I'm assuming the validator would be clever enough to detect references to undeclared prefixes etc)

[10:03:00.0000] <danbri>
eg. if someone validates using http://purl.org/dc/elements/1.0/Creator I'd like an easy way for them to learn that http://purl.org/dc/elements/1.0/creator is more widely used, and http://purl.org/dc/elements/1.1/creator even more widely used

[10:03:01.0000] <danbri>
(similar things hold for microformat idioms, since different constructs become fashionable over time, or are consumed by different apps)

[10:05:00.0000] <Philip`>
(Quite a few of the errors in http://philip.html5.org/data/rdf-namespace-status.txt are just plain typos - people find these things unusually tricky)

[10:07:00.0000] <danbri>
rdf-namespace-status.txt is cool

[10:07:01.0000] <danbri>
i had a script somewhere that tried to extract from dns/whois the time left for each popular domain

[10:08:00.0000] <danbri>
was quite fiddly as i recall, maybe there's a better library out there to try somewhere

[10:08:01.0000] <Philip`>
whois+grep? :-)

[10:09:00.0000] <Philip`>
Hmm, I guess they're not a very consistent format

[10:10:00.0000] <Philip`>
xmlns.com appears to have its whois information in YAML

[10:10:01.0000] <Philip`>
(via Gandi)

[10:11:00.0000] <danbri>
whois+grep --- yeah exactly what i was doing initially

[10:11:01.0000] <danbri>
then i think i found some perl api 

[10:11:02.0000] <danbri>
yaml, cool hadn't seen that

[10:12:00.0000] <danbri>
so one issue is figuring out from the domain "xxx.yyy.zzz.aaa.bbb.ccc" which one to look up

[10:12:01.0000] <Philip`>
Hmm, good point

[10:12:02.0000] <danbri>
foo.co.uk you look up foo.co.uk

[10:12:03.0000] <danbri>
foo.co.com you look up co.com ...

[10:12:04.0000] <Philip`>
http://publicsuffix.org/ seems too much like hard work

[10:12:05.0000] <danbri>
yeah that's the site i was just thinking of

[10:13:00.0000] <danbri>
i guess you can crawl down and look 'em all up

[10:13:01.0000] <Philip`>
What did you want that data for?

[10:13:02.0000] <danbri>
i'd like to make a "health check" site for namespaces, ... so people have some sense of QA before they commit to using one

[10:13:03.0000] <Philip`>
I guess most people would leave it until the last minute before renewing their registration, and it's hard to tell the difference between that and an abandoned domain

[10:14:00.0000] <danbri>
'cos it's all well and good being decentralised, but not if it's going to leave holes in the data in 10 years time cos someone's random experiment accidentally got worldwide use

[10:14:01.0000] <danbri>
last minute - yeah, also i'm told some registrars (eg. in Germany) don't let you register lots of years

[10:14:02.0000] <Philip`>
You could solve that problem by putting the RDF schema in a data: URL and use that as the namespace, instead of http://

[10:15:00.0000] <danbri>
so for eg dbpedia.org seemed to be perillously close to expiration, but the owner/admin had it on some rolling auto-debit arrangement 

[10:15:01.0000] <danbri>
heh, that'll really help with our consiseness problem :)

[10:16:00.0000] <Philip`>
Then you could define that namespace URI as an entity in an external DTD

[10:16:01.0000] <Philip`>
and import that into your document and use the entity reference

[10:16:02.0000] <Philip`>
which would avoid verbosity

[10:17:00.0000] <Philip`>
Then anyone publishing a vocabulary can put a DTD containing their schema data: URI onto their server

[10:17:01.0000] <Philip`>
The advantage is that you can move it onto a different host without changing the namespace that your data uses

[10:18:00.0000] <danbri>
heh

[10:18:01.0000] <Philip`>
The disadvantage is that it's clearly a stupid idea

[10:18:02.0000] <danbri>
how could the schema reference the properties defined within it? 

[10:18:03.0000] <danbri>
btw http://prefetch.net/code/domain-check looks handy

[10:19:00.0000] <Philip`>
I guess it could still reference them with "#"

[10:19:01.0000] <Philip`>
If it wants to reference anything in another schema, it can just import that schema's DTD

[10:20:00.0000] <danbri>
it's mad but worth writing up!

[10:21:00.0000] <Philip`>
Also it ties in nicely to the discussion on the WHATWG list about having browsers download external DTDs

[10:21:01.0000] <danbri>
/me gotta run (eats...)

[11:14:00.0000] <Philip`>
Why does http://validator.nu/?doc=http://philip.html5.org/docs/rdfa/ not like my <ins>es and <del>s?

[11:34:00.0000] <Philip`>
http://www.robweir.com/blog/2009/05/battle-for-odf-interoperability.html - "Standards are written documents -- text -- and as such they require interpretation. [...] When interpreting a document using an purposive view, you look at the purpose, or intent, of a document in its full context, and interpret the text harmoniously with that intent." - sounds like an excuse for writing vague specs

[13:04:00.0000] <gsnedders>
Philip`: A valid CURIE should probably have to have the prefix it uses defined

[13:04:01.0000] <Philip`>
gsnedders: My thing already requires that

[13:05:00.0000] <Philip`>
at least in the sense that it's got a big red box saying it should require that

[13:05:01.0000] <Philip`>
unless I missed something

[13:05:02.0000] <Philip`>
which is quite possible

[13:05:03.0000] <gsnedders>
Oh, no.

[13:05:04.0000] <gsnedders>
I just can't read.

[13:05:05.0000] <Philip`>
Hmm, maybe I should list "ability to read" in the prerequisites for reading the document

[13:15:00.0000] <jgraham>
Speaking of the public suffix list, I should hunt down gerv

[13:19:00.0000] <Philip`>
Hunt him down and kill him?

[13:23:00.0000] <jgraham>
Perhaps

[13:38:00.0000] <Traveler0>
can anyone help me with hymns requiem? 

[13:40:00.0000] <Traveler0>
currently using 1.8.7 and itunes 8.0.2.2 and ever time i try and decrypt a file it just says "failed"

[13:44:00.0000] <jgraham>
/me wonders how  Traveler0 ended up in this channel

[13:45:00.0000] <jgraham>
Traveler0: In general this seems like an unlikely place to get help for that particular problem. The main Requium user here isn't around at the moment

[13:47:00.0000] <annevk5>
the logs have a pagerank that is too high

[13:47:01.0000] <annevk5>
at least for certain topics :)

[13:48:00.0000] <jgraham>
annevk5: I assumed that but I couldn't reverse engineer the right search terms :)

[13:52:00.0000] <Traveler0>
jgraham: yes i found this from google

[13:53:00.0000] <Traveler0>
when does he usually chat?

[13:57:00.0000] <jgraham>
Traveler0: He's not really a hardcore user or anything. He jsut mentions it sometimes :) And I don't really know when he will be around, afaik he is away at the moment

[14:01:00.0000] <Traveler0>
from what the logs say Lachy is the one talking about it

[14:46:00.0000] <gsnedders>
Traveler0: He's a bit of a loner really, talking to nobody but himself about it :P

[14:49:00.0000] <Traveler0>
gsnedders, well i'd really like to have a few words with him

[14:50:00.0000] <Traveler0>
do i have is correct nick?

[14:53:00.0000] <gsnedders>
Traveler0: Yes, but I doubt he can really help

[14:54:00.0000] <gsnedders>
Whenever I send a txt I get odd sounds from the speakers here

[14:55:00.0000] <Traveler0>
gsnedders i get crazy feed back to for some reason

[15:15:00.0000] <roc>
hsivonen: I'd guess that reftests actually would under-estimate problems found on the real Web

[15:15:01.0000] <roc>
since most of the people who write reftests are quite clueful

[15:16:00.0000] <roc>
if I do say so myself

[15:16:01.0000] <roc>
you can easily get the total number of reftests ... just run them, it's in the window title

[15:16:02.0000] <Traveler0>
what is a reftests ?

[15:17:00.0000] <roc>
then again, in the context of reftests, a "problem" means "does not render pixel-accurately the same as the reference page"

[15:17:01.0000] <roc>
which is more strict than really matters on the Web

[15:17:02.0000] <roc>
http://weblogs.mozillazine.org/roc/archives/2008/12/reftests.html

[15:36:00.0000] <Hixie>
i love when people say "why can't we use prefixes? people understand css!"

[15:36:01.0000] <Hixie>
leaving quite alone the problem with ignoring evidence (people fail to understand prefixes, so that's more relevant than css), the sad fact is people don't understand css either

[15:36:02.0000] <Hixie>
most stylesheets just do .foo { properties... }

[15:37:00.0000] <Hixie>
with no more compelx selectors than that

[15:40:00.0000] <zcorpan>
Hixie: that's still indirection, though. using style="" is without indirection

[15:41:00.0000] <inimino>
I think the main problem is that prefixes look meaningful, but aren't supposed to be

[15:42:00.0000] <annevk5>
interesting point

[15:51:00.0000] <Traveler0>
Hixie, do you use requiem?

[16:15:00.0000] <zcorpan>
http://www.w3.org/TR/html-design-principles/#degrade-gracefully s/irrelevant/hidden/

[16:20:00.0000] <annevk5>
it's not like someone is actually maintaining that document

[16:53:00.0000] <Hixie>
http://www.w3.org/mid/7B28DCC4-BFAB-4BC3-9C96-7B37BF196349⊙gc


2009-05-19
[17:03:00.0000] <annevk5>
I'm not quite sure I follow the logic in that e-mail. A draft cannot change because of lack of implementations because it did not change as a result of receiving comments?

[17:06:00.0000] <zcorpan>
annevk5: the email says there were implementations

[17:11:00.0000] <othermaciej_>
annevk5: the indirection isn't quite the same

[17:11:01.0000] <othermaciej>
in a typical use of namespaces, you see "foo:bar", but that really means the ordered pair {http://something.foo.com/FooML, bar}

[17:11:02.0000] <othermaciej>
and the foo has no meaning

[17:11:03.0000] <annevk5>
The more I read that e-mail the less I get it. Oh well.

[17:12:00.0000] <othermaciej>
but binding CSS to an element with class="foo", there is no "more real" meaning than foo, and you won't be misled by looking for literal foo

[17:12:01.0000] <othermaciej>
for example, if you have a rule foo { color: green; } it's an unlikely mistake to try to find all green text by looking for everything with class foo in any document

[17:14:00.0000] <annevk5>
I agree that the CSS comparison does not quite work.

[17:31:00.0000] <othermaciej>
I think the difference is in CSS, you have an indirection to binding style, whereas XML namespaces introduce indirection to binding meaning

[18:23:00.0000] <annevk5>
http://gawker.com/5258524/ o_O

[18:26:00.0000] <Hixie>
i wonder why some of the classificatiosn are censored

[19:13:00.0000] <annevk5>
fwiw, application/microdata+json will probably not fly with the IETF

[19:14:00.0000] <annevk5>
though I could be mistaken I think s/+/-/ will be easier

[19:14:01.0000] <Hixie>
i'm just trying to be consistent with +xml

[19:14:02.0000] <Hixie>
i'm happy to change it to whatever makes them happy

[19:30:00.0000] <annevk5>
should subject="" be allowed to allow explicit assocation with the top-level microdata items?

[19:30:01.0000] <annevk5>
(it already works that way in user agents)

[19:30:02.0000] <Hixie>
say what?

[19:31:00.0000] <annevk5>
e.g. <span item> <span subject itemprop=name>X</span> </span>

[19:32:00.0000] <Hixie>
subject="" prevents the element from ever having a corresponding item

[19:32:01.0000] <Hixie>
(subject="" literally empty, i mean)

[19:34:00.0000] <annevk5>
"Otherwise, there is no corresponding item." ... "The list of elements that create items but do not themselves have a corresponding item forms the list of top-level microdata items."

[19:35:00.0000] <annevk5>
(what I was trying to point out was that subject="" is currently disallowed)

[00:21:00.0000] <hsivonen>
roc: the reftest failures don't suggest any lack of clue--just lack of cross-browser testing

[00:21:01.0000] <hsivonen>
roc: that is, test is for behavior x but also depends on behavior y where y varies among Gecko and IE or Gecko and WebKit

[00:22:00.0000] <hsivonen>
where behavior y is obscure and counter-intuitive

[00:47:00.0000] <Hixie>
annevk5: oh, i see what you're saying

[00:48:00.0000] <Hixie>
annevk5: i don't really think we should allow that, in fact it's probably better just to call out empty subject="" as an error

[03:56:00.0000] <hsivonen>
is there a way to test the precedence of the MS-maintained IE8 mode blacklist without being able to serve arbitrary content at any of the blacklisted domain?

[03:57:00.0000] <hsivonen>
that is, is there a way to edit the blacklist locally?

[03:57:01.0000] <hsivonen>
(an easy way that is)

[03:58:00.0000] <hsivonen>
I wonder if evaluating a bookmarklet when a given site has loaded inherits its blacklist properties

[03:58:01.0000] <othermaciej>
I bet it would

[03:59:00.0000] <othermaciej>
particularly if you use innerHTML to inject content into the site

[03:59:01.0000] <hsivonen>
othermaciej: innerHTML is no good, because the meta thingy works only if the parser sees it before a script has executed on the page

[04:00:00.0000] <othermaciej>
document.write perhaps?

[04:00:01.0000] <hsivonen>
document.write might work

[04:00:02.0000] <othermaciej>
or inject an iframe

[04:00:03.0000] <othermaciej>
supposedly subframes inherit the blacklist state of the top level document

[04:00:04.0000] <hsivonen>
kurafire claims that frames inherit the IE8/EmulateIE7 setting from parent

[04:00:05.0000] <jgraham>
hsivonen: Does it need to be a test that wworks online? Can't you just use /etc/hosts to point the domain at a site of your choice?

[04:01:00.0000] <jgraham>
Or does the balcklist not work like that?

[04:01:01.0000] <hsivonen>
jgraham: good point

[04:01:02.0000] <hsivonen>
jgraham: thanks

[04:05:00.0000] <Philip`>
You could modify the blacklist file in a hex editor

[04:05:01.0000] <Philip`>
(Probably won't work, though)

[04:11:00.0000] <hsivonen>
I'm appalled at the shady sites that doing a live search for windows hosts file turns up in IE8

[04:11:01.0000] <hsivonen>
I wouldn't get junk like that searching for Linux or Mac things

[04:12:00.0000] <Philip`>
Hmm, the recent update to the IE8 blacklist seems to have changed very little

[04:12:01.0000] <Philip`>
Removed: ca.gov daft.ie nexopia.com onet.pl picnik.com pricegrabber.com rightmove.co.uk yandex.ru

[04:12:02.0000] <Philip`>
Added: sat.gob.mx

[04:13:00.0000] <hsivonen>
google search is much more useful

[04:13:01.0000] <Philip`>
(Those are the changes from a couple of weeks ago, vs the initial release)

[04:21:00.0000] <hsivonen>
OK. I had the MS blacklist in the wrong place in my flowchart

[04:22:00.0000] <hsivonen>
umm. no, I didn't

[04:22:01.0000] <hsivonen>
hmm. either I have tested wrong, kurafire has tested wrong, or I have misunderstood what he said

[04:23:00.0000] <Philip`>
Or it's non-deterministic

[04:23:01.0000] <jgraham>
Or some combination of the above

[04:30:00.0000] <hsivonen>
whee! IE=9.9 and IE=99 are indeed different

[04:33:00.0000] <hsivonen>
oops. my test script is wrong for 9.9

[04:34:00.0000] <hsivonen>
9.9 and 99 are indeed the same

[05:27:00.0000] <hsivonen>
it turns out that the IE8 mode sniffing is even more complex than I previously thought

[05:27:01.0000] <hsivonen>
it takes into account the compatibility modeness of the frame parent

[05:28:00.0000] <hsivonen>
the parent can be in IE5 mode and the parent still remembers how it got to the IE5 mode

[05:28:01.0000] <Rik|work>
hsivonen: http://farukat.es/journal/2009/05/245-ie8-and-the-x-ua-compatible-situation

[05:28:02.0000] <hsivonen>
Rik|work: yes, that's what inspired me to test

[05:29:00.0000] <hsivonen>
Rik|work: my testing shows that X-UA-Compatible still allows you to override the mode from frame parent

[05:29:01.0000] <hsivonen>
for iframes at least

[05:30:00.0000] <Rik|work>
i work in a company that got in the compatibility list, we never received the mail to inform us

[05:56:00.0000] <hsivonen>
whew. I've done my share of community service today and have revised http://hsivonen.iki.fi/doctype/ and linked flowcharts

[05:56:01.0000] <hsivonen>
(several times until I got them right. I think)

[06:02:00.0000] <Philip`>
That is a little bit crazy

[06:04:00.0000] <hsivonen>
Philip`: what's crazy? revising or IE's behavior?

[06:04:01.0000] <hsivonen>
that's two hours of my life I'll never get back

[06:05:00.0000] <Philip`>
IE's behaviour

[06:05:01.0000] <Philip`>
and hence the flowcharts

[06:05:02.0000] <Philip`>
Two hours isn't much when you consider how much time you waste sleeping :-)

[06:15:00.0000] <annevk5>
sleeping is not actually wasting time though

[06:16:00.0000] <jgraham>
Indeed sleeping can often be more enjoyable than the rest of the day was :)

[06:32:00.0000] <hsivonen>
/me found a reftest that actually tests Gecko-specific weirdness: http://mxr.mozilla.org/mozilla-central/source/content/html/document/reftests/bug448564-2_malformed.html

[06:49:00.0000] <annevk5>
seems that http://my.opera.com/ODIN/blog/2009/05/19/opera-mama-a-sneak-peek-at-headings-images-and-summary confirms that summary="" indeed sucks on the Web

[06:52:00.0000] <annevk5>
https://twitter.com/robburns1/status/1809470040 "the problem is that the Microsoft/Opera cartel seeks to use HTML5 & WhatWG to undermine the web regardless of W3C endorsement"

[06:53:00.0000] <annevk5>
I guess that goes nicely with the vast Apple/Google conspiracy (they even have shared board members under investigation!!)

[06:55:00.0000] <Philip`>
I didn't think Microsoft and Opera were getting along well enough to join a cartel with each other, what with the lawsuits and suchlike

[06:55:01.0000] <Dashiva>
It's a typo, he meant Opera/Google

[06:56:00.0000] <Philip`>
Isn't Mozilla in on this?

[06:56:01.0000] <Dashiva>
Google owns Mozilla, duh

[06:58:00.0000] <Philip`>
I think Google should buy Opera

[06:58:01.0000] <Philip`>
just to complete the set

[06:58:02.0000] <Philip`>
Well, not complete, but make more complete than it would otherwise be

[06:58:03.0000] <Philip`>
It'd be great if they were paying for development of three competing browsers

[07:07:00.0000] <annevk5>
whoa, Google and Safari get the encoding right for http://thebjoernhoehrmannproject.org/ -- nice

[07:13:00.0000] <Philip`>
Why wouldn't they get it right?

[07:14:00.0000] <roc>
if there's a conspiracy, I don't want to be left out

[07:16:00.0000] <Philip`>
/invite roc #whatwg-secret-treehouse

[07:17:00.0000] <Philip`>
Whoops

[07:17:01.0000] <annevk5>
Philip`, they used the text/html algorithm

[09:33:00.0000] <Philip`>
http://broadcast.oreilly.com/2009/05/the-assertions-in-html-5.html

[09:39:00.0000] <annevk5>
Seems he somewhat misunderstood how things came to be :)

[09:41:00.0000] <Philip`>
I'm not sure that what he says is wrong

[09:42:00.0000] <Philip`>
because the constraints in H:TML are largely based on RELAX NG and (maybe?) Schematron implementations

[09:42:01.0000] <Philip`>
which is what he seems to say

[09:43:00.0000] <annevk5>
Not really: "(I don't know whether the designers of HTML limited themselves to Schematron assertions or to the subset in drafts of XSD 1.1.)"

[09:44:00.0000] <jgraham>
Philip`: But by basing his analysis on those parts of H:TML that have been extracted from schema in validator.nu, he implicitly misses anything that could not be written as a schema constraint

[09:45:00.0000] <jgraham>
(for example the constraint on table cells)

[09:46:00.0000] <Philip`>
annevk5: Oh, I missed that bit

[09:46:01.0000] <jgraham>
I guess it is hard to say his conclusions are wrong since the article just ends with no conclusions

[09:47:00.0000] <Philip`>
jgraham: Seems like a useul thing to point out in a comment :-)

[10:49:00.0000] <gsnedders>
What does (a,b)|c mean? the (a.b) is the greatest common divisor, but what about the |?

[10:51:00.0000] <gsnedders>
divisibility.

[11:02:00.0000] <annevk5>
Hmm, for cases such as http://www.w3.org/mid/65D2E21C-44E4-4E0B-B7D3-C635A8854A7D⊙ac it seems better to go with HTML5 directly

[11:03:00.0000] <annevk5>
A lot less markup, far easier for authors, and since neither solution has much adoption yet it doesn't really matter it seems.

[11:37:00.0000] <gsnedders>
/me should just give up all hope of ever managing to do any sort of maths related subject, seeming he can't do maths

[11:37:01.0000] <gsnedders>
Oh well, I guess Ill do English

[11:39:00.0000] <Philip`>
Maths is just like English, but with numbers

[11:42:00.0000] <inimino>
and with fewer right answers

[11:48:00.0000] <jgraham>
I guess maths actually has more right answers since there must be a countably infinite number of possible english essays but one could consruct a maths question with a uncountably infinite number of right answers

[11:49:00.0000] <inimino>
fewer right answers as a percentage of answers that are likely to be given, to questions that are likely to be asked

[11:50:00.0000] <inimino>
see, this is like English, where you have to interpret ;-)

[11:52:00.0000] <jgraham>
I'm not really sure how one takes percentages of infinite quantities

[11:53:00.0000] <inimino>
both the set of questions likely to be asked and answers likely to be given are finite sets

[11:54:00.0000] <inimino>
(though I guess "Lebesgue measure" might also be a right answer to that question)

[11:54:01.0000] <jgraham>
I have mo idea how you determined that. It seems that there are an infinite number of possible maths questions at least

[11:55:00.0000] <jgraham>
I guess for english questions there is a limit imposed by length

[11:55:01.0000] <inimino>
but surely not all of them are infinetely likely to be asked

[11:55:02.0000] <inimino>
erm

[11:55:03.0000] <inimino>
s/infinetely/equally/

[11:56:00.0000] <inimino>
for any expressible question or answer in maths there is just as hard a limit on length

[11:57:00.0000] <jgraham>
That seems like a more reasonable objection

[11:58:00.0000] <inimino>
an implication is that answers or questions that can't be expressed in the expected lifetime of the universe are somewhat less likely to be given or asked

[12:00:00.0000] <jgraham>
I wonder if one can still represent an infinite number of mathematical problems using some notation that allows the expression of an infinite number of possiblities in finite space

[12:00:01.0000] <jgraham>
I'm not quite sure that would work though

[12:01:00.0000] <inimino>
sounds hard

[12:54:00.0000] <hsivonen>
http://lists.w3.org/Archives/Public/www-tag/2009May/0056.html

[15:05:00.0000] <gsnedders>
hmm… I don't get what PCRE is doig…

[15:05:01.0000] <gsnedders>
*doing

[15:06:00.0000] <gsnedders>
/\x{D800}/u never matches

[15:06:01.0000] <gsnedders>
http://bugs.php.net/bug.php?id=47526

[15:17:00.0000] <Hixie>
these conspiracy theories are very confusing

[15:17:01.0000] <Hixie>
who am i supposed to be getting my bribe from this month again?

[15:19:00.0000] <gsnedders>
Hmm, html5lib has kinda died since the end of March

[15:43:00.0000] <gsnedders>
hmm, PHP 5.3-RC3-dev takes 0.58s less than 5.2.9 to tokenize the spec

[15:43:01.0000] <gsnedders>
(taking it down to 6.37s)

[15:53:00.0000] <gsnedders>
/me gawps at PHP's memory usage when storing arrays

[15:53:01.0000] <gsnedders>
We're talking ~70 bytes of overhead per array value

[15:54:00.0000] <gsnedders>
Which with what I'm doing is 25 times what storing the raw value would

[15:54:01.0000] <gsnedders>
*sighs*

[15:54:02.0000] <Philip`>
/me wonders how gsnedders is storing raw values that are 2.8 bits each

[15:54:03.0000] <Philip`>
*bytes

[15:54:04.0000] <gsnedders>
/me notes he wasn't being exact

[16:11:00.0000] <annevk5>
http://lists.w3.org/Archives/Public/www-html/2009May/0022.html is this bit of HTML5 so intuitive or did he actually look into it?

[16:16:00.0000] <Hixie>
annevk5: hah, awesome

[16:17:00.0000] <Hixie>
annevk5: you should let him know html5 already does that :-)

[16:41:00.0000] <annevk5>
Hixie, I'm not on that list anymore

[16:42:00.0000] <annevk5>
Hixie, out of curiosity, do you keep some notes somewhere other than /issues/ on what you're planning to work / working on?

[16:42:01.0000] <Hixie>
i have my company-internal okr list which says what i plan to have done by the end of the quarter

[16:43:00.0000] <Hixie>
but basically i just do what people want me to do

[16:43:01.0000] <Hixie>
so if you have something you need doing, let mek now

[16:43:02.0000] <Hixie>
and i'll get to it after microdata

[16:47:00.0000] <annevk5>
what's left for microdata? the remaining comments?

[16:48:00.0000] <Philip`>
Hmm... Microdata can be used to represent tree-shaped data structures

[16:48:01.0000] <Philip`>
HTML documents are tree-shaped data structures

[16:49:00.0000] <Philip`>
Therefore, I think we should have a way to encode HTML documents inside microdata attributes

[16:50:00.0000] <Hixie>
annevk5: quite a few use cases not yet addressed

[16:50:01.0000] <Hixie>
still got lots of mail to go through

[16:50:02.0000] <Philip`>
<div item=p><meta itemprop=style content=background:yellow></div> etc

[16:50:03.0000] <Hixie>
well, not mail, but use cases


2009-05-20
[17:14:00.0000] <dolske>
Hixie: why do media elements not get UA-provided controls unless the |controls| attribute is explicitly set?

[17:14:01.0000] <dolske>
it seems to be a problem people keep stumbling over, and feeds have an annoying habit of stripping it too.

[17:16:00.0000] <Hixie>
how else would you do it?

[17:16:01.0000] <Hixie>
basically it was the least bad solution we could find

[17:17:00.0000] <dolske>
have UA-controls be the default, unless a nocontrols attribute was set?

[17:18:00.0000] <Hixie>
how would you handle the case where nocontrols is set but js is disabled?

[17:19:00.0000] <dolske>
have the UA ignore the attribute, and provide controls anyway?

[17:19:01.0000] <annevk5>
I think it was mostly this way because the original design didn't have controls at all.

[17:20:00.0000] <Hixie>
dolske: so you would have elements with nocontrols="" have controls?

[17:21:00.0000] <dolske>
only when JS is disabled, since the page obviously can't provide its own.

[17:21:01.0000] <Hixie>
seems like the same as a controls="" attribute, except for being a negative attribute (historically a source of confusion) and one that isn't always accurate

[17:23:00.0000] <dolske>
Doesn't seem any worse than today with no controls attribute set and JS disabled... I'd expect a UA to enable the default controls in that case too.

[17:23:01.0000] <Hixie>
right but at least the source doesn't lie about what's going on there :-)

[17:24:00.0000] <Hixie>
it's early days yet, i'd expect feed sanitisers to fix their code in due course

[17:24:01.0000] <Hixie>
i'm surprised they're letting <video> through at all, that's probably a sign of a serious security bug

[17:24:02.0000] <dolske>
It seems a little less ambigious when JS is *enabled*, though... with "controls" you can't know for a fact that the page is going to supply its own, maybe something stripped the attribute. It seems unlikely for someone to accidentally add a "nocontrols" attribute.

[17:25:00.0000] <Hixie>
i don't see why anything would strip the controls="" value specifically but not add a nocontrols="" attribute unless there is already a security bug, as noted

[17:27:00.0000] <gsnedders>
/me wonders how to make php html5lib more conformant without losing perf.

[17:27:01.0000] <dolske>
I suppose, although there's still author error...

[17:27:02.0000] <gsnedders>
/me wonders about turning the whole tokenizer into a switch($state) statement, hence avoiding function call overhead (which is non-neglible)

[17:29:00.0000] <annevk5>
the UA should probably have a right click menu with options regardless

[17:41:00.0000] <Hixie>
yeah the browser should always make the controls available, even if the author said not to -- the attribute is just a way to say that the browser should definitely provide them

[17:41:01.0000] <Hixie>
as opposed to only if the user wants them

[17:41:02.0000] <Hixie>
that's another reason i don't really like "nocontrols"

[17:41:03.0000] <Hixie>
it's more like "no-controls-unless-you-want-them"

[21:38:00.0000] <zcorpan__>
"Yalin Wang and Jianying Hu have written an interesting paper on table

[21:38:01.0000] <zcorpan__>
type detection. Here's a link: http://portal.acm.org/citation.cfm?coll=GUIDE&dl=GUIDE&id=511478 They developed an algorithm and tested it on about 14000 tables." - http://markmail.org/message/mmlgkpck35u66qjp

[21:47:00.0000] <zcorpan>
hmm, not free to read the paper

[22:06:00.0000] <zcorpan>
http://www.w3.org/mid/1242739380.4531.275.camel@localhost

[23:25:00.0000] <heycam>
close ACTION-2561

[23:25:01.0000] <heycam>
sorry, ww

[01:11:00.0000] <Philip`>
zcorpan: http://www.research.ibm.com/people/j/jyhu/www2002.pdf

[01:14:00.0000] <zcorpan>
Philip`: thanks

[01:19:00.0000] <hsivonen>
whoa. I've failed to reference to email where James Clark calls the BS on XSD in my thesis

[01:21:00.0000] <hsivonen>
hmm. that sentence needs some parentheses to parse correctly

[01:21:01.0000] <hsivonen>
I've failed to reference in my thesis the email where James Clark calls the BS on XSD

[01:22:00.0000] <hsivonen>
this one http://www.imc.org/ietf-xml-use/mail-archive/msg00217.html

[01:23:00.0000] <hsivonen>
Hixie: were you planning on writing a microdata version of ccREL?

[01:24:00.0000] <Hixie>
no, as discussed here: http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2009-May/019668.html

[01:24:01.0000] <Hixie>
though it has been suggested that being able to link a rel=license (or similar) directly to an <img> or <video> would be beneficial (though without the extra baggage of ccREL, we don't want to be interpreting licenses in microdata or anything)

[01:24:02.0000] <Hixie>
dunno if that will involve microdata though

[01:24:03.0000] <Hixie>
i don't really see that it can to be honest

[01:25:00.0000] <Hixie>
well i guess it could

[01:25:01.0000] <Hixie>
but it'd be a bit awkward

[01:25:02.0000] <Hixie>
i dunno

[01:25:03.0000] <Hixie>
we'll see

[01:25:04.0000] <hsivonen>
Hixie: it seems to me that attribution-name, attribution-url and license could be properties on an item

[01:25:05.0000] <hsivonen>
and more-rights

[01:26:00.0000] <Hixie>
i don't see any reason to make the attribution-name and attribution-url machine readable

[01:26:01.0000] <Hixie>
such a thing would become intractable quickly and is not really needed by most licenses anyway

[01:26:02.0000] <Hixie>
(most licenses need the license and/or copyright and/or disclaimer included, not the attribution)

[01:27:00.0000] <Hixie>
(and i don't see why any of that is of benefit to computers searching through this stuff)

[02:05:00.0000] <hsivonen>
is it intentional that accesskey allows astral characters?

[02:07:00.0000] <Hixie>
yes

[02:08:00.0000] <hsivonen>
ok. thanks

[02:08:01.0000] <pesla>
np

[02:08:02.0000] <Hixie>
hsivonen: you might want to warn about it of course :-)

[02:08:03.0000] <Hixie>
"this seems unlikely to be a useful key"

[02:10:00.0000] <zcorpan>
what's Document.load?

[02:10:01.0000] <zcorpan>
seems to be undefined in my browsers

[02:12:00.0000] <Hixie>
ok time for bed

[02:12:01.0000] <Hixie>
nn

[02:19:00.0000] <hsivonen>
/me thought both Gecko and Opera had DOM Load/Save

[02:20:00.0000] <hsivonen>
from the era when W3C specs were considered presumptively righteous and MS inventions like XHR presumptively dirty

[02:23:00.0000] <zcorpan>
Hixie: the character separating prefix and property should be the underscore in order to find the definition using google

[02:24:00.0000] <jgraham>
Bt oh boy would that be ugly

[02:24:01.0000] <jgraham>
*But

[02:24:02.0000] <jgraham>
Also: the paper on table classification looks rather interesting

[02:26:00.0000] <zcorpan>
it would be interesting to look at tables that have nested tables but otherwise has the characteristics of a genuine table

[02:42:00.0000] <zcorpan>
http://gsnedders.html5.org/outliner/process.py?url=http%3A%2F%2Fwww.w3.org%2F - alt text is not used

[02:44:00.0000] <jgraham>
/me becomes slightly nervous of the conclusions in the paper when it starts talking about <ROWSPAN> and <COLSPAN> tags

[02:45:00.0000] <jgraham>
Also, it would be interesting to know how biased they are by working with the Java swing XML parser which presumably does not implement robust HTML parsing

[02:47:00.0000] <Philip`>
Also this was in 2002, when the world was different

[02:47:01.0000] <zcorpan>
figure 2b has empty tbody elements

[02:49:00.0000] <jgraham>
Philip`: Maybe someone has doen a followup? Can you search for papers that cite this one?

[02:50:00.0000] <Philip`>
jgraham: Can't you search for them? :-p

[02:50:01.0000] <Philip`>
http://portal.acm.org/citation.cfm?coll=GUIDE&dl=GUIDE&id=511478#citedby

[02:50:02.0000] <Philip`>
http://scholar.google.com/scholar?num=30&hl=en&lr=&safe=off&cites=1801764763188325408

[02:51:00.0000] <Philip`>
etc

[02:53:00.0000] <jgraham>
Philip`: Maybe. I don't know what ACM lets you do without paying but I don know that you can do whatever :)

[02:54:00.0000] <Philip`>
I don't have any special access now since I'm still at home :-)

[02:54:01.0000] <jgraham>
Philip`: You could proxy yourself through cam though :)

[02:54:02.0000] <Philip`>
The page zcorpan linked to has all the information, and only the PDF itself is restricted, as far as I know

[02:55:00.0000] <Philip`>
jgraham: I could, but I'm too lazy to do that :-p

[02:57:00.0000] <zcorpan>
jgraham: the google search result points to the pdfs directly

[02:57:01.0000] <Philip`>
(Also, I haven't figured out how to set up an actual proxy, so I use rdesktop to download the files locally on the remote server, and then copy via sftp)

[03:00:00.0000] <jgraham>
Philip`: Can't you do something like ssh -D (mor options that I can't remember)

[03:03:00.0000] <jgraham>
/me takes this oppertunity (again) to point out how insane the idea of pay-per-subscription scientific journals are

[03:03:01.0000] <Philip`>
jgraham: I could, but the last half of your command line is where the problem is

[03:04:00.0000] <Philip`>
(Also, configuring the browser to use that proxy requires knowledge and effort)

[03:04:01.0000] <hsivonen>
my hypothesis is that university libraries could run the system more cheaply by paying the salaries of the journal editors than by subscribing to the journals

[03:04:02.0000] <Philip`>
It does seem a bit silly when you can just Google for the paper's title and download the PDF directly

[03:05:00.0000] <Philip`>
and I've only seen that method fail with with really ancient (like 1990s) papers

[03:05:01.0000] <hsivonen>
the whole system seems to be supported by the inertia of universities insisting on giving more academic credit for publishing in one of these behind-the-paywall journals

[03:05:02.0000] <jgraham>
Philip`: I used the method in http://ubuntu.wordpress.com/2006/12/08/ssh-tunnel-socks-proxy-forwarding-secure-browsing/

[03:06:00.0000] <Philip`>
jgraham: Oh, that doesn't sound too complex

[03:07:00.0000] <hsivonen>
the inertia with Type 3 fonts in LaTeX-originating PDFs is great, too

[03:08:00.0000] <jgraham>
It seems like the paper above has ~15% of tables being "genuine" tables whereas there is a Google (and others) paper that finds ~1.1% are "relational tables"

[03:08:01.0000] <jgraham>
(the latter apper is called "WebTables: Exploring the Power of Tables on the Web" if you want to find it)

[03:09:00.0000] <roc>
hsivonen: the academics are actually changing relatively fast

[03:09:01.0000] <roc>
see the recent MIT decision for example

[03:09:02.0000] <hsivonen>
roc: what did MIT decide?

[03:09:03.0000] <jgraham>
hsivonen: The problem with fonts in LaTeX is that almost no LaTeX users understand how TeX fonts work

[03:09:04.0000] <roc>
basically the entire faculty is only allowed to publish in journals that allow public access

[03:09:05.0000] <hsivonen>
roc: cool

[03:09:06.0000] <jgraham>
hsivonen: e.g. me

[03:10:00.0000] <hsivonen>
the anecdotes I've heard recently have been about water engineering, marine biology and cancer research

[03:10:01.0000] <roc>
also you see on a regular basis the entire editorial board of a 'normal' pay-per-view journal resigning and setting up a new open-access journal with a slightly different name

[03:11:00.0000] <Philip`>
The instructions I got for some paper submission thing was they required a PS version of the paper as well as PDF, because most LaTeX users put the wrong type of fonts in the PDF

[03:11:01.0000] <hsivonen>
roc: nice

[03:12:00.0000] <hsivonen>
too bad ACM, Elsevier and Springer still hold the locks to back catalog

[03:16:00.0000] <jgraham>
roc: I have never seen that. Also making faculty do something is interesting because (in many cases) the faculty don't have to keep publishing much 

[03:17:00.0000] <hsivonen>
(the anecdotes I mentioned above were anecdotes about paywall craziness)

[03:19:00.0000] <roc>
here's the MIT thing

[03:19:01.0000] <roc>
http://www.bitsbook.com/2009/03/mit-adopts-an-open-access-policy/

[03:20:00.0000] <hsivonen>
thanks

[03:20:01.0000] <jgraham>
roc: Oh it doesn't actually require that they publishin an open access journal. It just requires that they *also* publish the information in the (open access) MIT repository

[03:21:00.0000] <jgraham>
Which is not that different to putting the article on arXiv 

[03:21:01.0000] <hsivonen>
/me wonders why MIT ops to disallow selling the articles for profit

[03:21:02.0000] <roc>
yes but the journal has to *allow* open access

[03:21:03.0000] <hsivonen>
if they are already available on the Web, why bother disallowing business model innovation that could spread the ideas even further

[03:22:00.0000] <roc>
most traditional journals don't

[03:22:01.0000] <roc>
many of them have been turning a blind eye to personal Web publication

[03:23:00.0000] <jgraham>
roc: Where does it say that? Of course the journal has to allow you to also put the article in another place but since they were all allowing that for arXiv anyway it doesn't seem that helpful

[03:23:01.0000] <jgraham>
(maybe arXiv isn't so prevelant in other fields so there it might be a big change)

[03:23:02.0000] <roc>
most traditional journals, at least in CS-related fields that I'm familiar with, technically required you to sign over copyright, at which point you lose the right to publish your article on your own Web site

[03:23:03.0000] <hsivonen>
so the policy is a great improvement, but I don't understand MIT's fascination with the *-nc flavors of CC licensing and now adopting a policy that doesn't allow the Provost to apply Free licenses to the papers

[03:24:00.0000] <roc>
it's true that people were just ignoring that

[03:25:00.0000] <roc>
but the MIT policy is flat-out incompatible with copyright assignment to the journal, unless the journal explicitly allows the open-access MIT publication

[03:25:01.0000] <jgraham>
roc: IIRC the blackwells publishing copyright form that I had to sign used to say that you were signing all the rights to blackwells and that you should try to disseminate the paper using arXiv

[03:25:02.0000] <jgraham>
I never understood how the cluses were compatible

[03:25:03.0000] <jgraham>
*clauses

[03:25:04.0000] <roc>
me neither

[03:26:00.0000] <hsivonen>
It's insane that strictly speaking one needs to ask a journal company for permission to publish a compilation of articles as a PhD thesis

[03:26:01.0000] <jgraham>
I think recently they might have changed it so that you retained your own copyright

[03:27:00.0000] <roc>
could be, I've been out of the loop

[03:27:01.0000] <roc>
anyway, licensing and other details aside, there's a clear desire by MIT faculty and various other academic groups to make their work open access

[03:50:00.0000] <zcorpan>
hsivonen: when elements get cloned, lots of unhelpful messages are generated

[03:51:00.0000] <zcorpan>
hsivonen: consider <div><div><div><ul><b><li></ul>x</div>x</div>x</div>

[03:52:00.0000] <zcorpan>
hsivonen: all "End tag div seen but there were unclosed elements." messages are just adding noise

[03:54:00.0000] <zcorpan>
hsivonen: or <ul><b><li></ul>x<p><p><p>

[03:56:00.0000] <zcorpan>
hsivonen: http://www.chasejobs.co.uk/index.php?option=com_catsone has such markup

[03:59:00.0000] <zcorpan>
/me filed a bug

[03:59:01.0000] <gsnedders>
zcorpan: The spec doesn't say to use alt text. Blame Hixie 

[04:00:00.0000] <zcorpan>
gsnedders: does the spec say to use textContent?

[04:00:01.0000] <jgraham>
Doesn't the spec just say to use textContent

[04:00:02.0000] <jgraham>
Oh, you already asked that

[04:01:00.0000] <gsnedders>
zcorpan: No, it just says to use the element. Doesn't say how.

[04:01:01.0000] <gsnedders>
zcorpan: And textContent seems a sane soltion

[04:01:02.0000] <zcorpan>
gsnedders: well then it's your bug to not show the image or its text equivalent

[04:02:00.0000] <gsnedders>
zcorpan: Not showing it seems perfectly reasonable though

[04:02:01.0000] <zcorpan>
i disagree :)

[04:04:00.0000] <zcorpan>
it seems reasonable to be able to use <h1><img src=... alt="Foo"></h1> and expect the image or "Foo" to show up in the outline

[04:04:01.0000] <jgraham>
zcorpan: It seems entirely unreasonable for an image to show up in an outline :)

[04:05:00.0000] <jgraham>
(But quite reasonable for "Foo" to show up)

[04:05:01.0000] <gsnedders>
zcorpan: Get a better textContent spec'd :P

[04:06:00.0000] <zcorpan>
1. Replace all descendant "img" elements with a new text node that has the .data set to the element's alt attribute. 2. Use textContent

[04:07:00.0000] <jgraham>
zcorpan: "Replace" is reeally the wrong word. You don't want to mutate the document :)

[04:08:00.0000] <zcorpan>
jgraham: for gsnedders' purpose, mutating the document seems ok, but he can implement it in another way that is equivalent :)

[04:09:00.0000] <gsnedders>
In Anolis I think I do actually copy the nodes already for TOC

[04:10:00.0000] <hsivonen>
hrm. the simples microdata to XML mapping I can think of would use OPMLish text-in-attributes antipattern

[04:11:00.0000] <hsivonen>
because XML element content can't distinguish between empty string and absent string

[04:14:00.0000] <gsnedders>
/me laughs at Hixie getting a reference to Avenue Q in yet another email

[04:19:00.0000] <zcorpan>
jgraham: without mutation: The textContentWithAlt attribute, on getting, must return a concatenation of the data of all the descendant Text nodes, as well as the value of the alt attribute of all the descendant img elements, of the context node, in tree order.

[04:20:00.0000] <zcorpan>
where attributes are said to be in no namespace and elements in the xhtml namespace

[04:29:00.0000] <zcorpan>
hmm, document.body.onload doesn't seem to be the same as window.onload in gecko and webkit

[04:32:00.0000] <zcorpan>
Hixie: the live dom viewer seems to give js error in ie8

[04:35:00.0000] <zcorpan>
window.onload and document.body.onload seems to be the same in ie

[04:50:00.0000] <Philip`>
gsnedders: I vote in favour of cross-document cross-reference support in Anolis

[04:51:00.0000] <Philip`>
(For http://philip.html5.org/docs/rdfa/ I hacked it so <span title="html5:resolve-a-url"> would go to the right place)

[04:51:01.0000] <Philip`>
((i.e. http://whatwg.org/html5/#resolve-a-url, and then the multipage magic sends you to the correct page))

[04:51:02.0000] <Philip`>
(but I guess that's not a very general solution)

[04:52:00.0000] <Philip`>
(and it doesn't work for #linkTypes because the references get lowercased :-( )

[05:03:00.0000] <gsnedders>
Philip`: I'm lazy, OK?

[05:05:00.0000] <gsnedders>
Philip`: I'll get around to it, OK? :P

[05:05:01.0000] <gsnedders>
Philip`: The two Anolis things I want to get done before I start at Opera are merging in biblio, and xdoc xref

[05:08:00.0000] <hsivonen>
gsnedders: will biblio use BibTeX microdata?

[05:08:01.0000] <jgraham>
gsnedders: I want microdata support in anolis

[05:08:02.0000] <jgraham>
Not for any particular reason, just because

[05:08:03.0000] <gsnedders>
hsivonen: No, it uses refer, because I concluded BibTeX was impossible to sanely parse at all

[05:08:04.0000] <hsivonen>
gsnedders: but you can now store bib data in .html with sane parsing!

[05:09:00.0000] <gsnedders>
hsivonen: When I wrote it you couldn't :P

[05:09:01.0000] <hsivonen>
/me mumbles something about off-the-shelf libraries being available for Java

[05:10:00.0000] <hsivonen>
gsnedders: if you had used Java, you could have used a .bib parser already there

[05:10:01.0000] <gsnedders>
/me mumbles something about off-the-shelf libraries being unbelievably buggy in Python

[05:10:02.0000] <gsnedders>
I could've tried Jython, but I'm heavily reliant upon lxml, which being a CPython extension won't work

[05:10:03.0000] <gsnedders>
hsivonen: I don't know Java, which is another problem.

[05:13:00.0000] <zcorpan>
gsnedders: web dom core has xdoc xrefs to html5

[05:14:00.0000] <gsnedders>
/me goes back to working on solving diophantine equations

[05:15:00.0000] <gsnedders>
(well, linear Diophantine equations)

[05:49:00.0000] <zcorpan>
hsivonen: itemprop="http://example.org/foo/#"

[05:49:01.0000] <hsivonen>
zcorpan: don't do that :-)

[06:04:00.0000] <annevk5>
zcorpan, how does <frameset onload> work when there's multiple framesets?

[06:04:01.0000] <annevk5>
come to think of it, that question applies to <body onload> too

[06:06:00.0000] <hsivonen>
what boxes did I miss? http://hsivonen.iki.fi/web-stack/

[06:07:00.0000] <hsivonen>
I notice I don't have Atom or RSS there, but they aren't involved when content is loaded into a browsing context

[06:10:00.0000] <annevk5>
browsers do implement the sniffing rules so they sort of are

[06:11:00.0000] <gsnedders>
Yeah, but nothing else really does

[06:11:01.0000] <gsnedders>
Most feed readers ignore content-type

[06:13:00.0000] <zcorpan>
annevk5: the body or frameset element (or window object) that gets the attribute set last overrides

[06:15:00.0000] <annevk5>
also when the element is not inserted yet?

[06:15:01.0000] <annevk5>
/me wonders if this is defined

[06:16:00.0000] <zcorpan>
annevk5: yep. the spec just says "... and with corresponding content attributes and DOM attributes exposed on the body element"

[06:16:01.0000] <zcorpan>
(which should say "body and frameset elements")

[06:18:00.0000] <zcorpan>
when the attribute is set, it overrides the element's ownerDocument's window.onload

[06:18:01.0000] <annevk5>
seems weird for non inserted elements to have an effect

[06:19:00.0000] <zcorpan>
/me points at topic

[06:19:01.0000] <annevk5>
meh

[06:25:00.0000] <jgraham>
hsivonen: It is not quite obvious why HTML, SVG, etc. are one side or the other of DOM

[06:26:00.0000] <jgraham>
(they should sit inside DOM in some sense)

[06:27:00.0000] <Philip`>
Does anyone have cross-document cross-references that aren't referring to HTML5?

[06:28:00.0000] <jgraham>
Philip`: You, surely?

[06:29:00.0000] <Philip`>
jgraham: I don't have any cross-references like that

[06:29:01.0000] <Philip`>
I just have links

[06:29:02.0000] <Philip`>
which, uh, I suppose are kind of cross-references

[06:30:00.0000] <Philip`>
/me presumes there's some good reason to have special syntax rather than just using <a href> everywhere

[06:31:00.0000] <annevk5>
yeah, from XHR2 to CORS

[06:31:01.0000] <annevk5>
HTML5 will have those too

[06:31:02.0000] <annevk5>
well, might

[06:50:00.0000] <hsivonen>
jgraham: they sit on top of the DOM

[06:52:00.0000] <hsivonen>
/me considers putting Unicode under everything but bitmaps and video

[06:55:00.0000] <jgraham>
hsivonen: In what sense?

[06:55:01.0000] <jgraham>
(It depends what you mean by HTML, SVG, etc.)

[06:56:00.0000] <hsivonen>
jgraham: in the sense HTML5 is a DOM language

[06:56:01.0000] <annevk42>
Philip`, the reason is not having to type <a href> everywhere

[06:56:02.0000] <hsivonen>
jgraham: and MathML, SVG and ARIA are in implementation practice, too

[06:57:00.0000] <Philip`>
annevk42: Typing <a href> doesn't seem like a great burden

[06:57:01.0000] <jgraham>
hsivonen: Sure. I think it is a bit confusing though, especially if you don't already have that wordview. In particular if you view HTML as a stream of bytes

[06:57:02.0000] <Philip`>
annevk42: but I guess typing the content of the href does

[06:58:00.0000] <Philip`>
jgraham: I thought the diagram was meant to represent how things are, rather than how some people misunderstand them to be

[07:00:00.0000] <jgraham>
Philip`: If the goal is to educate those people then doing so in a way that causes confusion is not the best approach

[07:01:00.0000] <jgraham>
hsivonen: Also, events are generally not exposed directly to the user

[07:02:00.0000] <hsivonen>
jgraham: what should I have between Events and the User?

[07:02:01.0000] <hsivonen>
/me considers being politically incorrect and using labels Ogg and Theora

[07:02:02.0000] <jgraham>
hsivonen: Javascript :)

[07:02:03.0000] <hsivonen>
perhaps adding WAV and PCM

[07:03:00.0000] <jgraham>
(Events seems to live within js/dom)

[07:03:01.0000] <hsivonen>
jgraham: having JS between user and events doesn't seem right

[07:03:02.0000] <hsivonen>
should I put mouse/keyboard there?

[07:03:03.0000] <hsivonen>
USB?

[07:03:04.0000] <hsivonen>
gotta go

[07:04:00.0000] <jgraham>
Oh, I see. I wasn't thinking of the flow of information going that way

[07:04:01.0000] <jgraham>
Since it goes the opposite way everywhere else

[07:04:02.0000] <jgraham>
(this is the problem with the "stack" concept)

[07:05:00.0000] <Philip`>
Human Interface Devices?

[07:05:01.0000] <annevk42>
it's much more a graph than stack

[07:05:02.0000] <Philip`>
/me realises that he doesn't know how to parse XML in Perl

[07:06:00.0000] <Philip`>
(except using XML::Simple, which is no good for parsing XHTML)

[07:07:00.0000] <Philip`>
Hmm, I wonder if XML::LibXML is sane

[07:10:00.0000] <Philip`>
Hmm, it seems to hang when parsing

[07:11:00.0000] <Philip`>
Oh, great, it's trying to download a zillion XHTML module DTDs

[07:11:01.0000] <gsnedders>
EPIC FAIL.

[07:12:00.0000] <jgraham>
Philip`: Surely you can turn that off?

[07:12:01.0000] <Philip`>
jgraham: I can once I know it's going to be doing that

[07:12:02.0000] <Philip`>
/me apologies to the w3.org server hosting people for accidentally contributing to the DDOS against them

[07:13:00.0000] <Philip`>
*apologises

[07:39:00.0000] <mpilgrim>
oh look, more false dichotomies from rubys: http://lists.w3.org/Archives/Public/public-html/2009May/0161.html

[07:40:00.0000] <mpilgrim>
/me wonders if rubys ever read the mailing list archives before he was appointed from on high

[07:40:01.0000] <gsnedders>
/me waits for mpilgrim to make another appearance on MLW's infamous blog

[07:41:00.0000] <mpilgrim>
the title of the spec was settled a long time ago

[07:43:00.0000] <mpilgrim>
/me is inexplicably reminded of the classic "bungee boss" dilbert strip

[07:44:00.0000] <mpilgrim>
"Hi-I'm-Your-New-Boss-Let's-Change-Everything-Before-I-Get-Reassigned-Oops-Too-Late-Goodbye"

[07:44:01.0000] <mpilgrim>
sproing

[07:44:02.0000] <mpilgrim>
He was like a mentor to me

[07:44:03.0000] <Dashiva>
Do you expect he'll be reassigned?

[07:45:00.0000] <mpilgrim>
no one seems to know why or how he was assigned in the first place

[07:45:01.0000] <mpilgrim>
so i couldn't say

[07:45:02.0000] <Philip`>
mpilgrim: You should volunteer to be chair

[07:46:00.0000] <Philip`>
I can't imagine any ways in which that could fail to work

[07:46:01.0000] <mpilgrim>
lately the working group seems to have been infested with bungee trolls

[07:48:00.0000] <gsnedders>
Hmm… according to this textbook, -2\frac{dy}{dx}-1 = \frac{2(x+y)}{x} - 1

[07:48:01.0000] <mpilgrim>
"Hi-I-Care-Deeply-About-The-Future-Of-The-Web-Let's-Change-Everything-Before-I-Get-Exasperated-Oops-Too-Late-Goodbyte"

[07:48:02.0000] <gsnedders>
How is the first term not negative?

[07:48:03.0000] <gsnedders>
(on the RHS)

[07:49:00.0000] <mpilgrim>
"We specifically need people to put forward suggestions for @profile and @summary.  If no such proposals are produced, such issues will be summarily closed." <-- noted for future reference

[07:50:00.0000] <Philip`>
gsnedders: How do you go from dx to x?

[07:50:01.0000] <gsnedders>
Philip`: I'm not entirely sure about that either

[07:50:02.0000] <Dashiva>
Is that an identity or an equation to solve?

[07:50:03.0000] <mpilgrim>
/me isn't sure rubys understands the biggest sticking point with the spec license

[07:50:04.0000] <gsnedders>
Identity

[07:51:00.0000] <Dashiva>
Then I'm with Philip`, it seems to be missing something

[07:54:00.0000] <jgraham>
/me joins Philip` and Dashiva 

[07:55:00.0000] <Philip`>
Fortunately nobody seems have cared about the bit in my document that says "This is blatant copyright infringement."

[07:56:00.0000] <gsnedders>
Or using the W3C stylesheet

[07:56:01.0000] <jgraham>
/me wonders what mpilgrim thinks the biggest sticking point is

[07:56:02.0000] <Philip`>
gsnedders: That too

[08:01:00.0000] <gsnedders>
http://secret.gsnedders.com/school/id.jpg

[08:01:01.0000] <gsnedders>
(This is blatant copyright infringement.)

[08:02:00.0000] <Philip`>
gsnedders: The question makes much more sense when we can see that dy/dx = -(x+y)/x

[08:03:00.0000] <gsnedders>
Oh, so they substitute for it

[08:03:01.0000] <gsnedders>
That makes sense

[08:03:02.0000] <Philip`>
(by rearranging the expression on the first line)

[08:05:00.0000] <jgraham>
/me rhetorically wonders if it is worth pointing out to Shelley that she doesn't get to define who "participated" in HTML 5

[08:06:00.0000] <Dashiva>
/me rhetorically avoids answering

[08:06:01.0000] <mpilgrim>
jgraham: they don't own it. it isn't theirs.

[08:07:00.0000] <jgraham>
/me doesn't know who is supposed to be owning anything

[08:07:01.0000] <mpilgrim>
traditionally, choosing a license is reserved for the copyright owner

[08:07:02.0000] <jgraham>
Oh, I follow now

[08:07:03.0000] <Philip`>
Doesn't the W3C traditionally own the copyright on the specs it publishes?

[08:08:00.0000] <gsnedders>
Philip`: The W3C host sites do, not W3C

[08:08:01.0000] <Philip`>
"Copyright ©1997-1999 W3C® (MIT, INRIA, Keio), All Rights Reserved.", "Copyright © 2009 W3C® (MIT, ERCIM, Keio), All Rights Reserved.", etc

[08:10:00.0000] <mpilgrim>
Philip`: traditionally, yes

[08:19:00.0000] <gsnedders>
http://secret.gsnedders.com/school/dif2.jpg

[08:19:01.0000] <gsnedders>
is that right?

[08:23:00.0000] <Philip`>
gsnedders: Looks about right to me

[08:23:01.0000] <Philip`>
though you could have cancelled out the 2s much earlier, to save a lot of writing :-)

[08:24:00.0000] <gsnedders>
That is true.

[09:02:00.0000] <mpilgrim>
i love how rubys seems to think that there is a stable of potential spec editors just chomping at the bit to be given a chance to publish

[09:02:01.0000] <mpilgrim>
we've been desperately trying to find editors for years

[09:03:00.0000] <mpilgrim>
nobody ever steps up and sticks with it long enough to matter

[09:04:00.0000] <mpilgrim>
/me notes, somewhat hypocritically, that he is not volunteering either

[09:05:00.0000] <jgraham>
mpilgrim: You seem to be assuming naivity which is not necesssarily true

[09:05:01.0000] <Philip`>
I don't think he thinks there is, which is why he includes the condition that if there aren't editors for a document then the group will drop that document, rather than saying the group will block on issues that don't have the support of an editor

[09:07:00.0000] <mpilgrim>
jgraham: well, like the rest of us, i sincerely hope that he DOES have a stable of potential spec editors

[09:08:00.0000] <mpilgrim>
though i think it's a bit cruel to make the call for editors without acknowledging how much work it will be

[09:08:01.0000] <jgraham>
mpilgrim: I interpreted his email more as "if you want to push an idea that you can't get someone else to spec you can spec it yourself (or stop complaining)"

[09:08:02.0000] <mpilgrim>
otherwise we'll just end up with more orphaned drafts

[09:09:00.0000] <mpilgrim>
/me rereads it

[09:10:00.0000] <mpilgrim>
i think we're both right

[09:10:01.0000] <mpilgrim>
it's "put up or shut up"

[09:10:02.0000] <gsnedders>
That's how I read it.

[09:11:00.0000] <mpilgrim>
but it's not as if we've been overwhelmed with a plethora of competently edited specs that we're ignoring

[09:12:00.0000] <annevk42>
I wish we had more editors for Web specs

[09:13:00.0000] <mpilgrim>
perhaps the spec fairy will visit us

[09:14:00.0000] <jgraham>
Is the spec fairy the one who gives you money wwhen you give her a spec?

[09:14:01.0000] <jgraham>
Because that is another issue...

[09:14:02.0000] <mpilgrim>
hmm, good point, the analogy doesn't work

[09:15:00.0000] <mpilgrim>
anyway, if rubys can get people to put in the work and produce a half-decent spec, more power to him

[09:15:01.0000] <jgraham>
(actually I think it sort of works for one of the problems which is that there just aren't a lot of people willing to pay for decent web specs, despite the huge value that they have)

[09:16:00.0000] <jgraham>
(it is a tragedy of the commons type thing where it is in everyone's interest for someone else to do it)

[09:16:01.0000] <jgraham>
(so Hixie is one of the very few people paid to edit specs full time)

[09:19:00.0000] <mpilgrim>
actually, most w3c groups are composed entirely of employees of member companies

[09:19:01.0000] <mpilgrim>
who are presumably paid to collaborate/discuss/edit their respective specs

[09:19:02.0000] <gsnedders>
Just not FT

[09:20:00.0000] <jgraham>
In many cases not just not FT but not enough of the time to make a good result

[09:20:01.0000] <mpilgrim>
not sure about that either

[09:20:02.0000] <mpilgrim>
at ibm there was a whole department of people who did standards work full-time

[09:21:00.0000] <mpilgrim>
but i agree with the fundamental point, that spec editing is hard and few people have the time + energy + resources + expertise to do it

[09:22:00.0000] <Philip`>
+ motivation

[09:22:01.0000] <jgraham>
If Hixie an't being paid to work full time on HTML 5 it would have died da long time ago

[09:23:00.0000] <Philip`>
(I guess there are people who could edit specs but don't care enough about it to do so)

[09:25:00.0000] <Philip`>
jgraham: Sounds like an unsustainable model for spec development

[09:26:00.0000] <jgraham>
Philip`: Right, but what else do you suggest?

[09:26:01.0000] <Philip`>
jgraham: No idea

[09:27:00.0000] <jgraham>
Unless you can make enough money from the spec itself to hire editors. But that doesn't seem likely

[09:28:00.0000] <Philip`>
It seems like companies don't mind paying their own employees to write their own specs

[09:28:01.0000] <Philip`>
e.g. OOXML

[09:29:00.0000] <Philip`>
but it seems much harder to get them to contribute similarly to open collaborative processes

[09:30:00.0000] <mpilgrim>
adsense for specs

[09:30:01.0000] <mpilgrim>
you heard it here first

[09:30:02.0000] <Philip`>
You could say that anybody writing a web page using your spec MUST provide a link to the spec

[09:31:00.0000] <Philip`>
Then you'll get massive PageRank and can sell links

[09:31:01.0000] <mpilgrim>
presumably that would work on all specs except rel=nofollow

[09:31:02.0000] <mpilgrim>
;)

[09:32:00.0000] <Philip`>
The problem is the editor would need up-front funding to develop the spec, and would only get any return once it's finished and widely used

[09:36:00.0000] <mpilgrim>
spec on spec

[09:37:00.0000] <Philip`>
There's always the ISO model of charging people to download the spec

[09:38:00.0000] <jgraham>
Philip`: I may not have mentioned that due to it being evil and wrong

[09:38:01.0000] <Philip`>
Also you could hide some patented technologies in the spec, and then sue Microsoft for a billion dollars a few years later

[09:39:00.0000] <mpilgrim>
submarine specs

[09:39:01.0000] <Philip`>
(See e.g. MP3)

[09:39:02.0000] <mpilgrim>
pretty sure we have rules against that

[09:39:03.0000] <annevk42>
patent policy doesn't protect against that if the spec writer is just a person

[09:40:00.0000] <annevk42>
at least, I believe it doesn't

[09:40:01.0000] <mpilgrim>
as opposed to spec writers who are superhuman?

[09:40:02.0000] <mpilgrim>
i'm confused

[09:40:03.0000] <annevk42>
spec writers employed by legal entities who could be counter sued

[09:40:04.0000] <mpilgrim>
no, i mean the w3c has rules against that

[09:40:05.0000] <Philip`>
Patent policy doesn't help if the patent is owned by somebody outside the W3C

[09:41:00.0000] <gsnedders>
s/W3C/WG/

[09:41:01.0000] <Philip`>
so the editor could get his/her friend to patent the technology and then sue Microsoft

[09:41:02.0000] <mpilgrim>
pretty sure we have rules against that too, but whatever

[09:42:00.0000] <Philip`>
There may be rules, but you can break the rules

[09:42:01.0000] <mpilgrim>
oh dear god: http://lists.w3.org/Archives/Public/public-html/2009May/0170.html

[09:42:02.0000] <Philip`>
If the punishment is getting thrown out of the W3C, but you get a billion dollars, it's pretty good overall

[09:43:00.0000] <mpilgrim>
it's really amazing that the w3c ever publishes anything at all

[09:45:00.0000] <Philip`>
Most WGs aren't as objectionable as the HTML WG

[09:49:00.0000] <mpilgrim>
most standards aren't as important as HTML 5 either

[09:49:01.0000] <mpilgrim>
and most of the ones that are, aren't standardized in public

[09:51:00.0000] <jgraham>
There seems to be an awful lot of "over my dead body" around publication issues in the HTMLWG

[09:51:01.0000] <jgraham>
which doesn't bode well

[09:52:00.0000] <beowulf>
didn't the guy who raised the objection to the design principals quit the wg? 

[09:55:00.0000] <Philip`>
Philip Taylor?

[09:55:01.0000] <Philip`>
If it was him, then yes

[09:56:00.0000] <hober>
hsivonen: mind if I use (with attribution) your web stack diagram in a blog post of mine?

[09:58:00.0000] <annevk4>
jgraham, there's not too much "over my dead body" considering the size of the group

[09:59:00.0000] <annevk4>
jgraham, I don't think the people who disagree strongly with certain decisions has grown since the start

[10:00:00.0000] <annevk4>
I listened to one of the TED talks yesterday where the speaker suggested that if nobody disagreed with what you're doing you're doing it wrong.

[10:01:00.0000] <Philip`>
I'm not sure I'd agree with that statement

[10:03:00.0000] <annevk4>
http://www.ted.com/index.php/talks/lang/eng/seth_godin_on_the_tribes_we_lead.html

[10:05:00.0000] <annevk4>
It does resonate with me that if you try to change the status quo people will disagree.

[10:05:01.0000] <annevk4>
And it's pretty clear we try to change the status quo

[10:05:02.0000] <gsnedders>
I thought most of what we did was standardizing the status quo

[10:06:00.0000] <Philip`>
Lack of disagreement might just mean that nobody knows what you're doing

[10:06:01.0000] <beowulf>
or it might mean you're being given a lot of slack in the hope something might happen

[10:06:02.0000] <Philip`>
And when you reach an extreme where everybody disagrees with you, you're almost certainly wrong

[10:07:00.0000] <annevk4>
I don't really see any of that applying to HTML5

[10:07:01.0000] <Philip`>
so it's not a particularly helpful guideline, since it doesn't tell you whether you're experiencing too much disagreement or too little

[10:10:00.0000] <annevk4>
Philip`, it seems pretty evident that as far as the standards community is concerned people are aware of HTML5

[10:10:01.0000] <annevk4>
Philip`, that said, I'm not sure it should be used as guideline

[10:11:00.0000] <gsnedders>
http://rossching.com/little-bribes/ is really impressive

[10:47:00.0000] <jgraham>
annevk4: If you look at people who have been actively involved, there seems to be quite a lot of people making objections and not being amenable to discussion or compromise

[10:48:00.0000] <annevk4>
My perception is different. Maybe it's worth gathering data? Then again, not sure what we could usefully do with the outcome.

[10:49:00.0000] <archtech>
Have you guys seen the ECMA5 presentation: http://www.youtube.com/watch?v=Kq4FpMe6cRs

[10:49:01.0000] <jgraham>
archtech: Not yet but I plan to watch it

[10:49:02.0000] <archtech>
I like the juxtaposition of HTML5 and ECMAScript5 :) The version where things finally are well done ;)

[10:49:03.0000] <jgraham>
annevk4: I expect I am being too negative

[10:50:00.0000] <jgraham>
It wouldn't be the first time

[10:50:01.0000] <jgraham>
It will be interesting to see what happens when we try to transition to LC

[10:52:00.0000] <gsnedders>
/me thinks we need DOM5 and CSS5 then Web5 will be complete!

[10:52:01.0000] <Philip`>
Hmm, rdfQuery throws an exception if you have <link rel=bogus>

[10:52:02.0000] <Philip`>
which isn't very forward compatible

[10:52:03.0000] <Philip`>
gsnedders: Don't forget HTTP5

[10:52:04.0000] <gsnedders>
And TCP/IP5

[10:52:05.0000] <Philip`>
Also we need User5

[10:53:00.0000] <gsnedders>
And Earth5

[10:53:01.0000] <annevk4>
and URL5 etc.

[10:56:00.0000] <beowulf>
and IE5. No, wait...

[11:53:00.0000] <archtech>
In the presentation we learn arguments object is removed, but nothing is there to replace it.

[11:53:01.0000] <archtech>
We have no default params, so weird.

[11:57:00.0000] <inimino>
archtech: the arguments object isn't removed

[11:58:00.0000] <archtech>
inimino, I misunderstood, can you please tell me what is their strategy?

[11:58:01.0000] <inimino>
I haven't seen the presentation so I'm not sure what was being referred to

[11:59:00.0000] <inimino>
but have no fear, 'arguments' is alive and well in ES5

[12:00:00.0000] <archtech>
inimino, to clarify myself, what the presentation seems to say is, it will be in ES5 no-strict mode, but any call to arguments, arguments.callee, arguments.caller will throw immediately in strict mode

[12:00:01.0000] <inimino>
I'll check

[12:01:00.0000] <archtech>
 javascript

[12:01:01.0000] <archtech>
ops

[12:05:00.0000] <inimino>
archtech: arguments.caller and .callee will now throw

[12:06:00.0000] <archtech>
Arguments will not?

[12:06:01.0000] <inimino>
no

[12:06:02.0000] <archtech>
I see, thanks.

[12:07:00.0000] <archtech>
Now if only they agreed on private/protected flags for properties as well ;)

[12:27:00.0000] <Hixie>
jgraham: if i wasn't being paid to work on html5, my social life would be far less interesting, but html5 would still be happening

[12:30:00.0000] <Philip`>
So Google is effectively paying for your social life?

[12:30:01.0000] <Philip`>
Sounds like a good trick you've got worked out there

[12:30:02.0000] <Hixie>
it's a pretty awesome trick

[12:33:00.0000] <annevk42>
/me more or less catched up with email and is wondering about XMLHttpRequest fetch again

[13:20:00.0000] <jgraham>
Hixie: I would prefer it if we didn't have to do that experiment :)

[13:23:00.0000] <mpilgrim>
oh look, john foliot is going on again about "experts"

[13:25:00.0000] <mpilgrim>
the WCAG WG was chartered on february 5, 2004: http://www.w3.org/2004/04/wcag-charter

[13:26:00.0000] <mpilgrim>
on december 11, 2008, they published "Techniques for Web Content Accessibility Guidelines 2.0": http://www.w3.org/TR/2008/NOTE-WCAG20-TECHS-20081211/

[13:27:00.0000] <mpilgrim>
it includes an entire section on "Using alt attributes on img elements": http://www.w3.org/TR/WCAG-TECHS/H37.html

[13:27:01.0000] <mpilgrim>
which lists a grand total of 2 examples

[13:27:02.0000] <mpilgrim>
only 1 of which contains actual markup

[13:28:00.0000] <mpilgrim>
HTML 5 has 14 sections devoted to using alt attributes on img elements:

[13:28:01.0000] <mpilgrim>
1. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#a-link-or-button-containing-nothing-but-the-image

[13:28:02.0000] <mpilgrim>
2. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#a-phrase-or-paragraph-with-an-alternative-graphical-representation:-charts,-diagrams,-graphs,-maps,-illustrations

[13:28:03.0000] <mpilgrim>
3. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#a-short-phrase-or-label-with-an-alternative-graphical-representation:-icons,-logos

[13:28:04.0000] <mpilgrim>
4. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#text-that-has-been-rendered-to-a-graphic-for-typographical-effect

[13:28:05.0000] <mpilgrim>
5. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#a-graphical-representation-of-some-of-the-surrounding-text

[13:28:06.0000] <mpilgrim>
6. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#a-purely-decorative-image-that-doesn't-add-any-information

[13:28:07.0000] <mpilgrim>
7. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#a-group-of-images-that-form-a-single-larger-picture-with-no-links

[13:28:08.0000] <mpilgrim>
8. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#a-group-of-images-that-form-a-single-larger-picture-with-links

[13:28:09.0000] <mpilgrim>
9. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#a-key-part-of-the-content

[13:28:10.0000] <mpilgrim>
10. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#an-image-not-intended-for-the-user

[13:29:00.0000] <mpilgrim>
11. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#an-image-in-an-e-mail-or-private-document-intended-for-a-specific-person-who-is-known-to-be-able-to-view-images

[13:29:01.0000] <mpilgrim>
12. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#general-guidelines

[13:29:02.0000] <mpilgrim>
13. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#guidance-for-markup-generators

[13:29:03.0000] <mpilgrim>
14. http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-0.html#guidance-for-conformance-checkers

[13:29:04.0000] <annevk42>
So reading WCAG-TECHS and looking at http://www.whatwg.org/specs/web-apps/current-work/images/parsing-model-overview.png I wonder why John thinks the alternative text is too long

[13:30:00.0000] <annevk42>
as WCAG-TECHS clearly says to include the words from the image in the alternative text

[13:30:01.0000] <krijnh>
/me wonders if mpilgrim would also paste all this if this channel wasn't logged :)

[13:30:02.0000] <Philip`>
On the other hand, the WCAG section on alt is readable, whereas the HTML 5 version is a stupidly large and complex amount of content that will just get totally ignored and/or misunderstood

[13:30:03.0000] <Philip`>
See e.g. http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2009-May/019631.html

[13:30:04.0000] <mpilgrim>
with a total of 32 examples

[13:31:00.0000] <mpilgrim>
"experts" are valuable if-and-only-if they share their expertise

[13:31:01.0000] <mpilgrim>
the accessibility "experts" -- who were given ALL THE TIME IN THE WORLD to share their expertise -- failed to do so

[13:32:00.0000] <mpilgrim>
so they don't get to complain that somebody came in later to fill the GAPING VOID OF KNOWLEDGE that they left

[13:37:00.0000] <Philip`>
http://www.web3d.org/x3d/wiki/index.php/X3D_and_HTML5 - "The X3D Working Group will participate in the HTML Working Group for the purpose of best integrating X3D with HTML."

[13:59:00.0000] <annevk42>
/me wonders how http://www.w3.org/mid/000601c9d984$a79d63b0$f6d82b10$@org will work out

[14:13:00.0000] <jgraham>
/me wonders if turning all bibliography properties into URIs won't destroy usability

[14:14:00.0000] <jgraham>
s/all/some/

[14:14:01.0000] <Philip`>
Abbreviate the URIs with a "biblio:" prefix and it'll be perfectly usable

[14:14:02.0000] <jgraham>
:p

[14:15:00.0000] <jgraham>
/me remembers that he tried to use zotero once but it seemed so complex that he didn't bother to learn to use it properly

[14:18:00.0000] <Philip`>
http://www.librocket.com/ - "HTML/CSS Game Interface Middleware" - "libRocket uses the time-tested open standards XHTML1.0 and CSS2.0 (while borrowing features from HTML5 and CSS3)"

[14:18:01.0000] <Philip`>
/me can't find any references to what features were borrowed from HTML5

[14:29:00.0000] <tantek>
mpilgrim, so what you're really saying is that "experts" are valuable if-and-only-if they make their expertise accessible to others.

[14:50:00.0000] <mpilgrim>
tantek: haha. but seriously, i still firmly believe what i wrote in 2004: http://diveintomark.org/archives/2004/08/16/specs

[14:50:01.0000] <mpilgrim>
"experts" are morons who write about their mistakes

[14:51:00.0000] <mpilgrim>
there isn't anything intrinsic to being an expert

[14:51:01.0000] <mpilgrim>
you're only as "expert" as what you share

[14:53:00.0000] <tantek>
heh - yes, classic essay

[14:54:00.0000] <tantek>
mpilgrim - and sharing in email is not particularly accessible IMHO - so if expertise isn't shared on a web page with a permalink (email archives notwithstanding because they don't search/index well), then it's not sufficiently shared.

[14:55:00.0000] <tantek>
mpilgrim - the IE5Mac/Tasman team pretty much followed the "Angels" description in your essay when developing/shipping IE5Mac

[14:55:01.0000] <tantek>
(modulo some interpretation of "thoroughly" - not sure how much is enough there, right?)

[15:03:00.0000] <mpilgrim>
IE5/Mac totally rocked

[15:03:01.0000] <mpilgrim>
for its time

[15:24:00.0000] <annevk42>
o_O http://code.google.com/p/chromium/issues/detail?id=988#c9

[15:31:00.0000] <gsnedders>
:o


2009-05-21
[17:58:00.0000] <roc>
I wonder if that X3D activity is a response to canvas3d

[17:58:01.0000] <roc>
ah yes

[17:58:02.0000] <olliej>
roc: hehehe

[17:59:00.0000] <roc>
yippee

[17:59:01.0000] <olliej>
roc: what is x3d?

[17:59:02.0000] <roc>
it's like the SVG of 3D

[17:59:03.0000] <roc>
except, I suspect, worse

[17:59:04.0000] <roc>
I think it evolved from VRML

[18:00:00.0000] <olliej>
roc: heheh

[18:00:01.0000] <olliej>
worse the svg?

[18:00:02.0000] <olliej>
wow

[18:00:03.0000] <olliej>
/me hides

[18:06:00.0000] <Philip`>
roc: It is indeed evolved from VRML

[18:07:00.0000] <Philip`>
You can express it in 'classic' VRML syntax as well as in XML syntax

[18:07:01.0000] <roc>
excellent

[18:07:02.0000] <roc>
two syntaxes for a format that's not even really deployedyet

[18:08:00.0000] <Philip`>
There's quite a number of independent implementations

[18:08:01.0000] <Philip`>
If you're careful, you can even write a scene that works in at least two implementations

[18:09:00.0000] <Philip`>
(It's also an ISO standard)

[18:09:01.0000] <Hixie>
aaah, why are we talking about x3d

[18:09:02.0000] <Hixie>
/me tries to get to a safe place

[18:10:00.0000] <Philip`>
I'm not sure X3D is particularly crazy or anything

[18:10:01.0000] <Hixie>
but why did it come up!

[18:10:02.0000] <Hixie>
actually i know nothing about 3d

[18:10:03.0000] <Philip`>
Because I saw http://www.web3d.org/x3d/wiki/index.php/X3D_and_HTML5

[18:10:04.0000] <Hixie>
though people seem to keep asking me about stuff to do with it

[18:11:00.0000] <Philip`>
(and because somebody emailed me about the X3D-in-canvas3d prototype I wrote ~16 months ago)

[18:11:01.0000] <Hixie>
jesus, everybody and their dog wants to put their vocabulary in html5

[18:11:02.0000] <Philip`>
Hixie: You've added lots of other things into HTML5 you know nothing about :-)

[18:12:00.0000] <Hixie>
i'm confident in saying that i know more about everything in html5 than i do about 3d anything

[18:12:01.0000] <Hixie>
but that admittedly doesn't raise the bar very high

[18:12:02.0000] <Philip`>
3D is just like 2D, except in a different direction

[18:13:00.0000] <Hixie>
2D doesn't have things like lights

[18:13:01.0000] <Hixie>
and cameras

[18:13:02.0000] <Philip`>
Depending on your level of abstraction, 3D doesn't have those either :-)

[18:14:00.0000] <Philip`>
Cameras are just transformation matrices, and lights are just vectors you feed into your shaders, and it's easy when you copy-and-paste the scary maths from somewhere

[18:14:01.0000] <Hixie>
woosh

[18:14:02.0000] <Philip`>
I'm sure it's easier to understand than <keygen> :-)

[18:15:00.0000] <Hixie>
actually <keygen> was easy to understand, it just wasn't documented

[18:16:00.0000] <Hixie>
i wonder how we should allow people to annotate their HTML documents with a datetime representing the last substantial update

[18:16:01.0000] <Hixie>
(or if we should not)

[18:17:00.0000] <Philip`>
Sounds like a job for the microdata hammer

[18:17:01.0000] <tantek>
one existing solution: they could use hAtom to markup the page as an entry with an "updated" datetime.

[18:19:00.0000] <Hixie>
i don't want to make html5 hardcode any magical class values; last time i did that i got yelled at so much it had to be removed :-)

[18:20:00.0000] <Philip`>
/me would wonder what a "substantial update" is, but decides instead to go to bed

[18:20:01.0000] <tantek>
no need to hardcode any magical class values in html5 - simply don't bother to solve the problem in html5 and let people use the existing solution

[18:20:02.0000] <Hixie>
this is actually part of some other problem i'm solving

[18:21:00.0000] <Hixie>
(namely adding Atom as one of the formats you can export to in http://www.whatwg.org/specs/web-apps/current-work/#converting-html-to-other-formats )

[18:27:00.0000] <roc>
I was quite clearly to say "not really deployed" instead of "not implemented"

[18:27:01.0000] <roc>
I was quite *careful*

[18:27:02.0000] <roc>
I wish I knew another language so I could pretend English wasn't my first language

[18:27:03.0000] <Hixie>
i wonder if the x3d-in-html5 effort is related to the people who asked me if collada could be put in html5 <canvas> recently

[18:28:00.0000] <Hixie>
if <article pubdate=""> gives the datetime that the article was published, what would be a good attribute name for the datetime that the article was last modified?

[18:29:00.0000] <Hixie>
(or alternatively, what would be good attribute names for published datetime and updated datetime respectively; it doesn't have to be pubdate)

[18:36:00.0000] <Hixie>
eh screw it, i'll just have only pubdate

[18:36:01.0000] <Hixie>
oh hey, updates can already be annoted using <ins>/<del>

[18:37:00.0000] <Hixie>
so i can just get the date from that

[00:05:00.0000] <hsivonen>
hober: using the diagram is ok

[00:51:00.0000] <Hixie>
anything interesting going on in the world of web specs?

[00:52:00.0000] <othermaciej>
you tell us!

[00:54:00.0000] <Hixie>
too many other things going on today to get around to spec work sadly

[01:03:00.0000] <hsivonen>
Hixie: did you add the DC creator stuff just because you could or was it in response to a use case?

[01:04:00.0000] <hsivonen>
Hixie: it violates theoretical purity

[01:04:01.0000] <Hixie>
just because i could

[01:04:02.0000] <Hixie>
it does?

[01:04:03.0000] <Hixie>
i was trying to convey as many of the document's semantics as possible

[01:04:04.0000] <Hixie>
so that the conversion loses as little as possible

[01:04:05.0000] <othermaciej>
what theoretical purity?

[01:04:06.0000] <Hixie>
(it still loses a lot)

[01:05:00.0000] <hsivonen>
yes. If you build an object model for microdata that contains enough data to output into all the outputs, you need to have an isAddressThatAppliesToDocument flag on items

[01:06:00.0000] <Hixie>
the RDF output is not about microdata

[01:06:01.0000] <hsivonen>
oh?

[01:06:02.0000] <Hixie>
it's a conversion of HTML to RDF

[01:06:03.0000] <Hixie>
which includes microdata

[01:06:04.0000] <Hixie>
but is not limited to it

[01:07:00.0000] <Hixie>
the Atom output i'm doing now barely mentioned microdata at all

[01:07:01.0000] <Hixie>
mentions, i should say

[01:07:02.0000] <hsivonen>
hmm.

[01:07:03.0000] <Hixie>
the vCard, vEvent, and BibTeX outputs are microdata-specific

[01:07:04.0000] <Hixie>
maybe i should reorganise that section to separate them

[01:07:05.0000] <hsivonen>
I was thinking of doing SAX-to-model builder

[01:08:00.0000] <hsivonen>
and a model to XML-represented-as-SAX converter 

[01:08:01.0000] <hsivonen>
that could feed into a RELAX NG validator whose error messages have been munged

[01:08:02.0000] <Hixie>
sounds useful

[01:09:00.0000] <hsivonen>
(retaining Locator data, of course)

[01:09:01.0000] <Hixie>
i can split the RDF output into two, one for microdata and one for documents

[01:09:02.0000] <Hixie>
(with the output of the latter being a superset of the former's)

[01:10:00.0000] <hsivonen>
It think rel / <meta name> stuff should be considered part of microdata and be reflected in the JSON output

[01:10:01.0000] <Hixie>
why?

[01:10:02.0000] <hsivonen>
because we can :-)

[01:10:03.0000] <Hixie>
microdata is just items

[01:10:04.0000] <Hixie>
because it's simpler

[01:11:00.0000] <Hixie>
simpler > possible

[01:11:01.0000] <hsivonen>
seriously, though, it makes it possible to use JSON as a document metadata conveyor without having to resort to RDF

[01:11:02.0000] <Hixie>
(microdata isn't metadata, it's just data)

[01:12:00.0000] <Hixie>
i don't mind adding the metadata to the JSON stuff

[01:12:01.0000] <Hixie>
though that affects what gets dragged and dropped

[01:12:02.0000] <Hixie>
which i'm not sure is what we want

[01:16:00.0000] <Hixie>
hmm... <h1> or <title> for atom:title?

[01:17:00.0000] <Hixie>
i'm thinking <h1>, with <hgroup><h2> being atom:subtitle

[01:34:00.0000] <Philip`>
Hixie: Seems probably unlikely that people from the X3D community would be asking for COLLADA support, since they're completely separate (and partly overlapping) technologies

[01:34:01.0000] <Hixie>
ah

[01:34:02.0000] <Hixie>
i wonder why 3d is suddenly so important

[01:35:00.0000] <Philip`>
(As I understand it, they're sort of competing technologies in certain aspects, but the X3D people now recognise that COLLADA is popular and isn't going to go away so they're trying to live with it rather than fight against it, but it's not what they really want to support)

[01:36:00.0000] <Philip`>
3D has always been important :-)

[01:36:01.0000] <Hixie>
collada, from what i understand, is a disaster

[01:36:02.0000] <Philip`>
but now the technology is capable of supporting it decently (perhaps)

[01:37:00.0000] <Hixie>
if i'm not mistaken, collada is about 15 different technologies all jammed into one spec because nobody could agree to remove their proposal

[01:37:01.0000] <Philip`>
I've used COLLADA (for loading models and animations into a game engine), and it worked quite nicely for what I wanted

[01:38:00.0000] <Philip`>
It's not intended for connecting arbitrary pairs of producers and consumers, like HTML - it's intended to fit between specific tools in a pipeline, and the tools can pick which subset to support based on what features they want

[01:40:00.0000] <Philip`>
e.g. you'll export a COLLADA file from 3ds Max, and run it through a custom tool that converts it into some efficient custom geometry format, and maybe have some other tool to extract texture references, and another to get some physics data, and whatever

[01:41:00.0000] <Philip`>
(and maybe some reusable tools in the middle to optimise geometry or something)

[01:43:00.0000] <Philip`>
and so you can just write tools specifically for the dialect that's generated at the beginning of the process, and you don't need to worry about any unused features because they're irrelevant until you want to start using them

[01:46:00.0000] <Hixie>
Philip`: since i've been asked to put it in html, that's the environment in which i was considering its appropriateness

[01:52:00.0000] <hsivonen>
Hixie: did you actually mean to use a single 'author' or 'editor' field in the HTML version of BibTeX

[01:52:01.0000] <hsivonen>
Hixie: that's horribly ugly

[01:52:02.0000] <Hixie>
i meant to use whatever bibtex uses

[01:53:00.0000] <hsivonen>
Hixie: comparend to having multiple and munging to and from .bib's " AND "

[01:53:01.0000] <Hixie>
i was originally going to do that but it means hard-coding them in the conversion

[01:53:02.0000] <hsivonen>
Hixie: I suggest special casing those two, because .bib's intra-field format makes no sense for DRY HTML

[01:53:03.0000] <Hixie>
which means hard-coding a non-localised string

[01:54:00.0000] <hsivonen>
Hixie: consider it a reserved word in .bib

[01:54:01.0000] <hsivonen>
Hixie: letting 'and' leak into HTML content in whatever language is worse, much worse

[01:54:02.0000] <Hixie>
i guess

[01:54:03.0000] <Hixie>
/me isn't a fan of bibtex

[01:55:00.0000] <Hixie>
i wish there was a better vocabulary to use here

[01:55:01.0000] <Hixie>
but the alternatives are all ten times worse

[01:55:02.0000] <hsivonen>
particularly when foo and bar and baz doesn't make DRY sense even in English

[01:55:03.0000] <Hixie>
Refer for example has the worst vocabulary terms ever

[01:55:04.0000] <Hixie>
and most of the others have much more domain-specific usage

[01:55:05.0000] <Hixie>
or no usage to speak of

[02:00:00.0000] <hsivonen>
the microdata representation sucks big time for format like "Doe, J.", but that's because munging of personal names sucks

[02:00:01.0000] <hsivonen>
and people should just put "John Doe" in their bibliographies

[02:02:00.0000] <Philip`>
Hixie: Yeah, it does seem pretty inappropriate as a general standard model/animation format

[02:03:00.0000] <Philip`>
The whole being-written-in-XML thing makes it inappropriate for content delivery anyway

[02:04:00.0000] <Hixie>
hsivonen: thanks for the bug

[02:04:01.0000] <Hixie>
collada is xml?

[02:04:02.0000] <Philip`>
Uh... Yes

[02:04:03.0000] <Philip`>
I guess you didn't look at it in any detail :-p

[02:08:00.0000] <Philip`>
(It's not terrible in terms of space/time efficiency, since the main issue is just that it stores numbers as space-separate strings of decimals rather than as 32-bit floats, but it's (intentionally) not great)

[02:08:01.0000] <Philip`>
s/e/ed/

[02:09:00.0000] <Hixie>
i said i knew nothing!

[02:09:01.0000] <Hixie>
wow i had no idea collada was xml

[02:09:02.0000] <Hixie>
isn't that, like, dumb

[02:09:03.0000] <Hixie>
surely 3d modelling is one of those things that you'd want a binary format for

[02:10:00.0000] <Philip`>
There are lots of binary format, but they're almost universally awful

[02:10:01.0000] <Philip`>
e.g. the only widely-supported one is 3DS, which comes from the days of DOS and is limited to 64K vertices and supports almost no features

[02:10:02.0000] <Philip`>
and there are some newer proprietary binary ones

[02:12:00.0000] <Philip`>
The aim was to make something extensible and relatively easy to understand, and not to make something that's as efficient as possible (because everyone will have different rendering engines and want their own custom formats for that), and XML was a better way to achieve that than a new binary format

[02:13:00.0000] <Hixie>
fair enough

[02:15:00.0000] <othermaciej>
3D canvas will probably get implementation traction

[02:15:01.0000] <othermaciej>
although I don't quite get why Mozilla decided to take it to Khronos instead of a Web standards group

[02:16:00.0000] <Philip`>
Possibly because they need involvement from hardware/driver developers in order to make it work securly, and all those people are in Khronos

[02:16:01.0000] <Philip`>
s//e/

[02:39:00.0000] <Hixie>
ok bed time nn

[02:39:01.0000] <annevk42>
g'night

[03:09:00.0000] <zcorpan_>
if you remove the "green", you still have "left border", so even ignoring markup the requirements are met

[03:11:00.0000] <zcorpan_>
i guess i would just be contributing noise if i pointed that out on the list

[03:12:00.0000] <zcorpan_>
though i wonder if john would be satisfied if the border was changed from green to black

[03:14:00.0000] <zcorpan_>
oh maciej already said so

[03:14:01.0000] <annevk42>
I doubt borders are exposed to e.g. screen readers though

[03:15:00.0000] <zcorpan_>
annevk42: the requirements didn't say anything about that

[03:16:00.0000] <annevk42>
Private email suggests that's not a concern :/

[03:16:01.0000] <Philip`>
The quote from WCAG makes it sound like you don't have to actually make it work in e.g. screen readers, you can just convey the information with markup (and presumably class="example" is markup) and it's somebody else's problem if they can't actually access that information through their UA

[03:18:00.0000] <zcorpan_>
"WCAG 2 - 1.4.1 Use of Color: Color is not used as the only visual means of conveying information, indicating an action, prompting a response, or distinguishing a visual element. (Level A);" - this is met, there is still a border as a visual means

[03:18:01.0000] <othermaciej>
the WCAG rules lawyering is completely off point

[03:18:02.0000] <annevk42>
That was my point, yes. Nobody has replied to that yet as far as I can tell.

[03:18:03.0000] <annevk42>
There are quite a lot of specifications that use class="example" in this way...

[03:18:04.0000] <hsivonen>
Philip`: I think mpilgrim had a blog post about that kind of accessibility enablement

[03:18:05.0000] <othermaciej>
the way examples are indicated is indeed inadequate, but for reasons that have nothing to do with John's complaint

[03:19:00.0000] <zcorpan_>
"WCAG 1 - 2.1 Ensure that all information conveyed with color is also available without color, for example from context or markup. [Priority 1];" - there is no information conveyed with color. information is conveyed with a border

[03:19:01.0000] <annevk42>
E.g. http://www.w3.org/TR/css3-namespace/

[03:19:02.0000] <othermaciej>
also it is lame of him to complain about something I myself noticed and proposed to change

[03:19:03.0000] <zcorpan_>
"and Section 508 § 1194.22 c) Web pages shall be designed so that all information conveyed with color is also available without color, for example from context or markup." - same here

[03:19:04.0000] <othermaciej>
and to give some huge lecture on how bad it is

[03:19:05.0000] <zcorpan_>
so WCAG2 is met, and WCAG1 and section 508 are not appliciable

[03:20:00.0000] <othermaciej>
annevk42: the way examples render there looks much better to me

[03:20:01.0000] <othermaciej>
like the way they say "Example"

[03:20:02.0000] <annevk42>
othermaciej, that's a matter of style, not substance, in this case

[03:20:03.0000] <othermaciej>
annevk42: I would say rather that it's a matter of style, not markup

[03:20:04.0000] <othermaciej>
but in this case the style affects the substance

[03:21:00.0000] <othermaciej>
b/c I think in retrospect it was unclear to people that the examples are just examples, and separate from the statements of principle

[03:21:01.0000] <annevk42>
fair enough

[03:22:00.0000] <zcorpan_>
othermaciej: in some sections the examples are clearly marked as such, e.g. "2.2.1. Examples"

[03:23:00.0000] <zcorpan_>
the first two batches of examples

[03:24:00.0000] <othermaciej>
zcorpan_: good point! maybe using that style in most places would be sufficient

[03:24:01.0000] <othermaciej>
also, sadly, some of the design principles are still missing examples

[03:25:00.0000] <annevk42>
I think you rewrote the first two principles to be more lengthy and then the group decided to publish

[03:25:01.0000] <othermaciej>
also I wonder if perhaps "Evolution Not Revolution" could be retitled "Incremental Improvement" and thus made to encapsulate Hixie's proposed "Baby Steps" principle

[03:25:02.0000] <othermaciej>
yeah

[03:25:03.0000] <othermaciej>
this document could use some polish

[03:25:04.0000] <othermaciej>
the thing is, certain parties may turn any revision process into a rathole

[03:25:05.0000] <othermaciej>
I re-read the votes on the invidual principles

[03:25:06.0000] <annevk42>
s/may/will/ I think

[03:26:00.0000] <othermaciej>
and found that of all the objections, nearly all came from a few people who in turn objected to more than half the principles

[03:26:01.0000] <othermaciej>
so basically we have a set of Design Principles that most of the group agrees on, and then a handful of people who object to almost all of them

[03:26:02.0000] <othermaciej>
it's hard to see how to increase consensus in that light

[03:26:03.0000] <othermaciej>
you could water down the documet, but then the many people who largely agree with it could object

[03:29:00.0000] <gsnedders>
Hixie wanted visible (symbolic, IIRC) marks for things like examples, created by Anolis. I haven't done that yet.

[03:29:01.0000] <annevk42>
ah, so John Foliot can blame you!

[03:30:00.0000] <Philip`>
Only if the design principles document uses Anolis

[03:30:01.0000] <gsnedders>
Oh, we're talking about a11y of the design principles doc now?

[03:31:00.0000] <gsnedders>
/me is way behind on the ml

[03:31:01.0000] <Philip`>
I hope we are, otherwise I'm totally lost

[03:32:00.0000] <annevk42>
I thought it was about both

[03:33:00.0000] <Philip`>
/me shrugs

[03:35:00.0000] <zcorpan_>
hmm, the headings are blue

[03:36:00.0000] <zcorpan_>
wonder if that's also a violation of wcag?

[04:01:00.0000] <hsivonen>
whoa! role=math really made it into ARIA last call

[04:01:01.0000] <hsivonen>
I'd love to see two interoperable implementations of that one

[04:01:02.0000] <annevk42>
funny

[04:02:00.0000] <annevk42>
I was just reading Member-only ARIA discussions as I encountered something ARIA-related in my inbox

[04:02:01.0000] <om_sleep>
what does role=math mean?

[04:02:02.0000] <zcorpan_>
i wonder why ARIA discussions are still happening in Member-only lists

[04:02:03.0000] <annevk42>
Not too optimistic about my comments getting addressed adequately

[04:02:04.0000] <hsivonen>
om_sleep: it means that the the thing represents math and its text equivalent is MathML or TeX

[04:03:00.0000] <om_sleep>
I see

[04:03:01.0000] <om_sleep>
so you'd put it on an image that's an equation?

[04:03:02.0000] <om_sleep>
or something like that

[04:03:03.0000] <hsivonen>
which makes me ask: TeX with which implicitly defined macros?

[04:03:04.0000] <zcorpan_>
i also still wonder why MathML can't stand on its own and needs role="math"

[04:03:05.0000] <om_sleep>
anyway

[04:03:06.0000] <om_sleep>
bedtime

[04:03:07.0000] <hsivonen>
om_sleep: you basically dump fooTeX stuff in alt

[04:03:08.0000] <om_sleep>
zcorpan_: presumably the idea here is that the UA doesn't implement MathML but the assistive technology does

[04:04:00.0000] <zcorpan_>
hsivonen: i asked that to be defined before

[04:04:01.0000] <om_sleep>
(!)

[04:04:02.0000] <hsivonen>
I wonder how top on the list of e.g. JAWS developers a TeX subsystem is

[04:04:03.0000] <zcorpan_>
om_sleep: even in that case, MathML can stand on its own -- it has a namespace and a tag name to identify itself as "math"

[04:05:00.0000] <hsivonen>
zcorpan_: oh yeah, the first SHOULD makes no sense

[04:05:01.0000] <hsivonen>
/me starts drafting feedback

[04:09:00.0000] <zcorpan_>
"In order to be perceivable, images SHOULD also be labeled by text that describes the math formula as it should be spoken, using the aria-describedby attribute."

[04:09:01.0000] <zcorpan_>
wait, if the author should write how the formula should be spoken, what does role="math" help?

[04:10:00.0000] <zcorpan_>
oh well

[04:11:00.0000] <zcorpan_>
interesting that there are are both SHOULDs and shoulds

[04:11:01.0000] <webben>
Hmm. Yep that's a pretty confusing bit of spec.

[04:13:00.0000] <zcorpan_>
when role="math" was new, i asked "how would it be implemented?"

[04:13:01.0000] <zcorpan_>
evidently, that's not a great concern

[04:18:00.0000] <hsivonen>
email sent

[04:30:00.0000] <zcorpan_>
hsivonen: where did you send it?

[04:31:00.0000] <zcorpan_>
found it

[05:13:00.0000] <hsivonen>
am I understanding correctly that aria-label exists because title attribute is badly supported by AT?

[05:15:00.0000] <hsivonen>
or is the purpose to suppress the tooltip?

[05:20:00.0000] <hsivonen>
hmm. the conformance of aria-checked is interesting

[05:20:01.0000] <hsivonen>
it must be present but may have the value "undefined"

[05:24:00.0000] <zcorpan_>
hsivonen: my impression is that it exists because it's hard or annoying to position <h2> elements off-screen with css

[05:24:01.0000] <hsivonen>
zcorpan_: ah. 

[05:25:00.0000] <hsivonen>
zcorpan_: do you know why aria-required is global but aria-expanded isn't?

[05:25:01.0000] <hsivonen>
the former seems less universally applicable than the latter

[05:25:02.0000] <zcorpan_>
hsivonen: no

[05:25:03.0000] <zcorpan_>
hsivonen: (...and that Lynx users can be ignored)

[05:26:00.0000] <hsivonen>
If Lynx users wish to become unignored, Lynx should catch up with a lot of Web tech that isn't even tied to bitmap rendering

[05:27:00.0000] <hsivonen>
someone should take Gecko or WebKit and replace the bitmap renderer with a character grid renderer

[05:27:01.0000] <annevk42>
I wonder that given the time it takes to get ARIA deployed and given that IE is updating as well if it would not have been more productive to argue for the sane solutions from the start...

[05:28:00.0000] <annevk42>
Of course, with hindsight things are easier.

[05:29:00.0000] <hsivonen>
well, assuming that IE8 doesn't support role=math, feel free to argue for a saner solution for that one

[05:30:00.0000] <hsivonen>
the obvious solution is MathML-in-text/html, but for that to work, IE and WebKit need MathML support

[05:30:01.0000] <annevk42>
you've already done that no?

[05:31:00.0000] <hsivonen>
annevk42: I effectively argued for the reading of the math as words being used as alt

[05:31:01.0000] <hsivonen>
the major use case is math on wikipedia

[05:31:02.0000] <annevk42>
another problem with this silo approach to comments is that the WG can back each other up while people commenting are all alone

[05:32:00.0000] <hsivonen>
I guess they should have the resources to feed their math to AsTeR in order to generate the words

[05:32:01.0000] <hsivonen>
at least in English

[05:32:02.0000] <zcorpan_>
is alt text for math images covered in html5?

[05:32:03.0000] <annevk42>
e.g. following the progress on each other's comments is much harder, etc.

[05:33:00.0000] <annevk42>
zcorpan_, yeah, use MathML :)

[05:35:00.0000] <hsivonen>
I'm updating v.nu's ARIA support

[05:36:00.0000] <hsivonen>
it seems that PF hoisted nearly everything except aria-expanded into global states/props, and I don't understand the logic behind that move at all

[05:37:00.0000] <hsivonen>
also, I don't like the design pattern for aria-checked, but I guess it's too late to complain about that one

[05:37:01.0000] <annevk42>
hmm Hixie complicated things so much with the storage mutex and task queues

[05:37:02.0000] <annevk42>
/me grmbls

[05:41:00.0000] <hsivonen>
http://twitter.com/myakura/statuses/1867322694

[05:43:00.0000] <annevk42>
while certainly practical, do we really want to alienate yet another bunch of people? :)

[05:50:00.0000] <Dashiva>
Bikeshed a new name instead?

[05:50:01.0000] <hsivonen>
Maybe the ODF proponents shouldn't have bragged about how short their spec is: http://blogs.msdn.com/dmahugh/archive/2009/05/09/1-2-1.aspx

[05:54:00.0000] <annevk42>
"Text: The specific conversion is implementation-defined"

[05:54:01.0000] <annevk42>
yay

[05:55:00.0000] <hsivonen>
they should have defined .doc5, .xls5 and .ppt5 instead of these ODF and OOXML things

[05:55:01.0000] <Dashiva>
<insert joke about ODF5>

[05:55:02.0000] <Dashiva>
/me shakes fist at hsivonen

[05:58:00.0000] <annevk42>
I guess the conversion is impl specific because it can depend on locale and such as well

[05:59:00.0000] <annevk42>
Sort of glad I'm not involved with that though admittedly we have similar issues with e.g. character encoding

[06:04:00.0000] <annevk42>
also check the comments on that MS blog post btw

[06:09:00.0000] <annevk42>
hmm: http://www.schneier.com/blog/archives/2009/05/on_the_anonymit.html

[06:30:00.0000] <Dashiva>
"So that everyone can be  sure about which parts are which,  the normative parts use specific phrases like “shall” and “shall not” to clearly label the things the standard actually requires you to do."

[06:30:01.0000] <Dashiva>
Reminds me of that specs vs blogs post :)

[06:32:00.0000] <annevk42>
if I'm to believe that person it sounds like ODF 1.1 has been written in HTML4 style

[06:38:00.0000] <annevk42>
That last email from Laura is interesting. Rather than saying what she thinks she selectively highlights parts from past surveys. (I.e. not mentioning that there was 5 to 1 agreement for the principle but rather that people disagreed and as well as strongly disagreed with the principle.)

[06:38:01.0000] <annevk42>
I seem to recall this happening more often, but I could be wrong

[06:40:00.0000] <annevk42>
And the last bit seems simply misleading as Larry was talking about the name of the draft which was a separate question from the one she highlights.

[06:57:00.0000] <Philip`>
Hmm, people are saying things related to me on Twitter but I've got no way to respond

[06:57:01.0000] <Philip`>
/me guesses he should just send email or something

[06:57:02.0000] <annevk42>
they are?

[06:58:00.0000] <Dashiva>
Philip`: Post it anywhere else and get someone to RT it ;)

[06:59:00.0000] <zcorpan_>
hsivonen: "Actually, I’ve noticed that on this page the social bookmark links create this error: “WHITESPACE in QUERY” Er, what?" - http://www.tomleadbetter.co.uk/blog/entry/html5/

[07:01:00.0000] <zcorpan_>
that page also has <input type="url" id="url" name="url" value="http://" /> which is not valid

[07:01:01.0000] <annevk42>
ah, burningbird + samruby

[07:01:02.0000] <annevk42>
seems we're following similar twitter queries :)

[07:03:00.0000] <Philip`>
I'm not really following any, I just randomly look at random people's Twitter pages sometimes :-)

[07:03:01.0000] <zcorpan_>
hsivonen: and "</l↩</ul>" which could have slightly clearer messages maybe

[07:04:00.0000] <Philip`>
I assume proper following would require a Twitter account, which is far too advanced for me when I haven't even got a blog yet

[07:04:01.0000] <Philip`>
(I haven't even got a proper homepage yet, so I'm ~15 years behind the times)

[07:06:00.0000] <annevk42>
/me hits the next issue in understanding task queues

[07:07:00.0000] <zcorpan_>
"The contact form also has several ARIA attributes, and if you look at it in Opera, you’ll see it has a nifty built-in validation."

[07:07:01.0000] <hsivonen>
what kind of validations does Opera have for ARIA?

[07:07:02.0000] <hsivonen>
or is that about WF2 validation?

[07:07:03.0000] <zcorpan_>
hsivonen: none, it's type="url" etc

[07:07:04.0000] <zcorpan_>
yep

[07:57:00.0000] <annevk42>
sigh

[07:57:01.0000] <annevk42>
event loops differ between workers and browsing contexts

[07:58:00.0000] <hsivonen>
how?

[07:58:01.0000] <annevk42>
in browsing contexts they're always associated with a Document

[07:58:02.0000] <annevk42>
so when a Document gets garbage collected all the associated events get too (I think)

[07:59:00.0000] <annevk42>
this means that XHR cannot tie into existing event loops or has to tie into different ones depending on whether XMLHttpRequest Document exists

[07:59:01.0000] <annevk42>
neither is nice

[08:00:00.0000] <hsivonen>
I have tried not to look at GC interaction of native and JS objects

[08:00:01.0000] <hsivonen>
I'm assuming magic happens while I'm not looking

[08:00:02.0000] <annevk42>
it would probably be much easier if Hixie sorted all this out though one problem with that is he'd be the only spec writer to understand it

[08:05:00.0000] <Philip`>
Do implementors understand it?

[08:05:01.0000] <hsivonen>
mpilgrim: fwiw, I read Sam's process post as "put money where your mouth is (by producing camera-ready text) or stop complaining"

[08:07:00.0000] <hsivonen>
mpilgrim: and the license thing matters, because camera-ready text only works if it doesn't come with copyright strings attached

[08:07:01.0000] <annevk42>
Philip`, I think so, given the recent WebKit comments on Workers

[08:08:00.0000] <annevk42>
Philip`, but I'm not a 100% sure

[08:08:01.0000] <annevk42>
hmm, midori crashes all the time and the current chromium-browser cannot load any page whatsoever -- FAIL

[08:30:00.0000] <Philip`>
http://linuxcentre.net/adobe-has-issued-a-dmca-removal-request-for-rtmpdump/ - hooray for DRM

[08:31:00.0000] <Philip`>
(rtmpdump is a tool for saving Flash streaming video to disk)

[08:32:00.0000] <Philip`>
I guess I'll have to delete the version off my disk, and stop downloading TV programmes that are broadcast for free anyway

[09:11:00.0000] <Dashiva>
/me is catching up on public-html mail from the two last weeks...

[09:11:01.0000] <Dashiva>
Can anyone tell me, is there a happy ending?

[09:12:00.0000] <annevk42>
anything going wrong?

[09:13:00.0000] <annevk42>
hehe, I like how I found http://www.w3.org/TR/qaframe-spec/ again

[09:14:00.0000] <annevk42>
thanks for writing that karlcow!

[09:14:01.0000] <Dashiva>
Well, so far there seems to be a lot of heat about rdfa

[09:16:00.0000] <Philip`>
Someone tried to start writing a concrete document on how to use RDFa in HTML5

[09:16:01.0000] <Philip`>
Don't know if it's a good idea or not

[09:17:00.0000] <gsnedders>
w00t!

[09:17:01.0000] <gsnedders>
Maths exam over!

[09:17:02.0000] <Dashiva>
There was one line, about how the RDFa community would gladly ensure incorporating RDFa was as painless as possible

[09:18:00.0000] <Dashiva>
And I was wondering if that was the same RDFa community insisting on namespaces and CURIEs

[09:19:00.0000] <Philip`>
Dashiva: I think some people in that community have expressed a willingness to consider alternate prefix-binding mechanisms rather than xmlns:*

[09:19:01.0000] <Philip`>
which would avoid the text/html-specific pain

[09:21:00.0000] <Dashiva>
That sounds good

[09:21:01.0000] <Philip`>
I don't know whether they want to incompatibly change RDFa-in-XHTML to use that same mechanism, though

[09:22:00.0000] <Philip`>
or change it and leave the old deprecated mechanism in there too, so that there's three ways to specify prefixes

[09:22:01.0000] <Philip`>
or do something else

[09:23:00.0000] <Dashiva>
Mhm

[10:35:00.0000] <gsnedders>
Time to do more work on html5lib-php, me thinks

[10:35:01.0000] <gsnedders>
(Like working on the 23 failing tokenizer tests)

[10:49:00.0000] <gsnedders>
/me fixes 10

[10:50:00.0000] <annevk4>
the phone discussion just now was interesting

[10:50:01.0000] <annevk4>
apparently Larry thinks that the design principles we drafted do not reflect how HTML5 is written though he did not elaborate much on why when asked

[10:51:00.0000] <gsnedders>
http://www.w3.org/2009/05/21-html-wg-minutes.html

[10:51:01.0000] <smedero>
well he did refer to table @summary and head @profile as his examples

[10:51:02.0000] <smedero>
i think i kinda got what he meant... but it felt like he was dancing about something else he really wanted to say

[10:52:00.0000] <smedero>
he did note he needed some time to formulate his comments...

[10:53:00.0000] <annevk4>
still not really clear to me how @profile is related to caving cowpaths, but fair enough

[10:55:00.0000] <annevk4>
I think the problem with the HTML WG meeting is that a lot of time people start thinking while the telcon is happening rather than planning ahead of time what they want to say whenever the next telcon is

[10:56:00.0000] <annevk4>
This causes a lot of lost time for the other participants if someone is semi-brainstorming while everyone is dialed in

[10:56:01.0000] <smedero>
indeed.

[11:02:00.0000] <gsnedders>
So I think all the tokenizer test cases html5lib-php fails are either disputable or the parse errors thrown in 9.2.4 Tokenization (and not subsections)

[11:03:00.0000] <Philip`>
Why are they disputable?

[11:03:01.0000] <gsnedders>
Philip`: The spec doesn't say what order parse errors have to be thrown in

[11:04:00.0000] <gsnedders>
(Or more to the point, for null bytes and non-characters, when they have to be thrown)

[11:05:00.0000] <gsnedders>
html5lib-php throws all those parse errors first, then does the actual tokenizing

[11:07:00.0000] <Philip`>
hsivonen: "Thora" typo in diagram

[11:07:01.0000] <Philip`>
gsnedders: The test cases should have ignoreErrorOrder:true (or something like that) in those cases

[11:14:00.0000] <Dashiva>
"Microdata will only work in HTML5/XHTML5. XHTML 1.1 and yes, 2.0 will be around for years, decades."

[11:14:01.0000] <Dashiva>
So we're still on the "If it doesn't validate, it doesn't work" page?

[11:15:00.0000] <gsnedders>
Dashiva: Yes

[11:20:00.0000] <Dashiva>
"Reversed domains are as ugly as CURIEs (but at least CURIEs resolve to something useful [...]"

[11:21:00.0000] <Dashiva>
Just avoiding the redundant http:// makes them a lot prettier in my book :)

[11:21:01.0000] <gsnedders>
s/resolve/might resolve/

[11:22:00.0000] <Dashiva>
"Are you thinking that someone will be using different namespaces but the same prefix? Come on -- do you really think that will happen?"

[11:23:00.0000] <Dashiva>
Well, if prefixes never collide, why do we bother with namespaces? Why not just use the globally unique prefixes?

[11:25:00.0000] <hober>
Dashiva: indeed.

[11:30:00.0000] <annevk4>
Dashiva, can you please provide pointers with these quotes?

[11:39:00.0000] <Dashiva>
Sure

[11:39:01.0000] <Dashiva>
Any particular one you want right now?

[11:40:00.0000] <Dashiva>
@annevk

[11:40:01.0000] <annevk4>
1 and 3 but 2 might be fun as well for completeness

[11:48:00.0000] <gsnedders>
http://code.google.com/p/html5lib/source/detail?r=8f225971195ed9f2e6a8e28abaa01dc35100decb

[11:48:01.0000] <gsnedders>
w00t fun

[11:48:02.0000] <gsnedders>
I <3 languages with shit Unicode support.

[11:48:03.0000] <annevk4>
PHP6!

[11:50:00.0000] <gsnedders>
Um, have we been talking about that for as long as DNF yet? :P

[11:50:01.0000] <gsnedders>
/me can remember when there was talk of getting an alpha out by the end of 2005

[11:50:02.0000] <gsnedders>
(As of today, we are still pre-alpha)

[11:52:00.0000] <Dashiva>
Hmm, can I easily get a link from the whatwg archives from a message-id?

[11:52:01.0000] <annevk4>
no :(

[11:53:00.0000] <annevk4>
maybe we should write a script that does enables that

[11:53:01.0000] <Dashiva>
First one: http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2009-May/019751.html

[11:54:00.0000] <Hixie>
Dashiva: if you have the body of the e-mail, searching for that on google works

[11:55:00.0000] <gsnedders>
10 test failures…

[11:56:00.0000] <Dashiva>
Second: http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2009-May/019749.html

[11:58:00.0000] <Dashiva>
Third: http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2009-May/019752.html

[11:59:00.0000] <Dashiva>
annevk4: Would also be nice if the "by date" view segmented the list by day

[12:03:00.0000] <annevk4>
Dashiva, hmm, I might just be compelled enough to try to do something

[12:03:01.0000] <annevk4>
been a while since I wrote Python

[12:14:00.0000] <Madness>
A quick question about html5 and links around stuff: Suppose I have a link with overflowing content. Should the clickable area be limited o the a element or extended to the overflow too?

[12:16:00.0000] <Hixie>
Madness: the click on any text node or box that is a descendant of the element would bubble to the element and then cause the link to be followed

[12:17:00.0000] <Madness>
Thank you.

[12:22:00.0000] <Hixie>
there's a (!@$%*! registration fee for the tpac meeting?

[12:22:01.0000] <Hixie>
are you kidding me?

[12:23:00.0000] <Hixie>
we have enough trouble getting people to attend the htmlwg meeting in the first place

[12:24:00.0000] <annevk42>
how much?

[12:25:00.0000] <Hixie>
$50

[12:25:01.0000] <annevk42>
per day or for the whole event?

[12:25:02.0000] <Hixie>
event

[12:25:03.0000] <Hixie>
er

[12:25:04.0000] <Hixie>
day!

[12:25:05.0000] <Hixie>
jesus!

[12:25:06.0000] <Hixie>
that's even worse!

[12:26:00.0000] <annevk42>
so I guess it's more than food costs then

[12:26:01.0000] <Hixie>
i can't justify $350 for a meeting i don't even want to go to in the first place!

[12:26:02.0000] <Hixie>
that's insane!

[12:26:03.0000] <annevk42>
you paid more last year

[12:26:04.0000] <annevk42>
well, your employer

[12:27:00.0000] <Hixie>
i only paid travel last year

[12:27:01.0000] <Hixie>
and hotel

[12:27:02.0000] <annevk42>
and food

[12:27:03.0000] <Hixie>
i have to pay for food even if there's n meeting

[12:27:04.0000] <smedero>
is that info on a members only list?

[12:27:05.0000] <Hixie>
http://www.w3.org/2009/11/TPAC/overview.html#Registration

[12:27:06.0000] <annevk42>
I usually pay a lot more for food when there's a meeting :)

[12:28:00.0000] <smedero>
lovely. following links around the w3.org site, I ended up here: http://www.w3.org/2002/09/TPOverview.html

[12:28:01.0000] <smedero>
heh

[12:28:02.0000] <annevk42>
"The registration fee will increase to 75 USD per person, per day after a date approximately 4-6 weeks prior to the meeting."

[12:29:00.0000] <smedero>
yowsers

[12:29:01.0000] <annevk42>
fortunately the dollar ain't worth shit

[12:29:02.0000] <annevk42>
:p

[12:29:03.0000] <Hixie>
yeah well i'm paid by the $ so...

[12:29:04.0000] <Hixie>
also, google already paid the insane 5-digit membership fee this year

[12:30:00.0000] <Hixie>
so charging again for the meeting is crazy

[12:35:00.0000] <annevk42>
/me investigates cron jobs for the alternate mail index

[13:08:00.0000] <hsivonen>
is the HTML WG having a meeting at TPAC?

[13:12:00.0000] <smedero>
I thought so...

[13:12:01.0000] <smedero>
Didn't Chris Wilson send something around?

[13:13:00.0000] <gsnedders>
He was asking about interest for a meeting, IIRC

[13:13:01.0000] <gsnedders>
/me isn't really sure, as it's highly unlikely he'll be there, and thus doesn't care

[13:37:00.0000] <gsnedders>
What file do I add tests to for the tokenizer?

[13:38:00.0000] <gsnedders>
What's the diff between test1, test2, test3 and test4?

[13:39:00.0000] <Philip`>
One of them is (originally) automatically generated, and full of weird short inputs, so you probably shouldn't use that one

[13:39:01.0000] <Philip`>
There's no clear difference between the others, as far as I'm aware

[13:40:00.0000] <Philip`>
so pick one that has some existing tests that are vaguely similar to the one you're adding

[13:40:01.0000] <Philip`>
or pick one at random

[13:42:00.0000] <Philip`>
or think up a new system that makes sense, and rewrite all the tests into a different layout

[13:47:00.0000] <annevk4>
Philip`, any chance you can put <meta name=robots content=index,nofollow> on your survey pages?

[13:47:01.0000] <annevk4>
Philip`, I'm getting reqeusts from spammers if they can put spam links on your page

[13:48:00.0000] <annevk4>
Philip`, e.g. http://philip.html5.org/data/charsets.html

[13:49:00.0000] <annevk4>
Philip`, it's not too serious fwiw


2009-05-22
[17:12:00.0000] <karlcow>
[11:18]  <annevk42> hehe, I like how I found http://www.w3.org/TR/qaframe-spec/ again

[17:12:01.0000] <karlcow>
[11:18]  <annevk42> thanks for writing that karlcow!

[17:12:02.0000] <karlcow>
that was a Team work :) but thanks 

[18:08:00.0000] <Hixie>
any atom experts here?

[18:08:01.0000] <Hixie>
should an atom <atom:content> node contain a redundant instance of the entry's title?

[20:23:00.0000] <mpilgrim>
hixie: there's nothing stopping you, but it's highly unusual

[20:24:00.0000] <mpilgrim>
(re: atom:content)

[20:25:00.0000] <Hixie>
k

[01:20:00.0000] <Hixie>
ok

[01:20:01.0000] <Hixie>
that concludes my first attempt at dealing with microdata-related use cases

[01:36:00.0000] <KevinMarks>
looks like I tuned in just too late

[01:50:00.0000] <Hixie>
KevinMarks: http://lists.w3.org/Archives/Public/public-html/2009May/0207.html

[02:02:00.0000] <hsivonen>
now the readers of public-html who aren't subscribing to whatwg need to deal with the almost-always-broken next/prev navigation of the whatwg archives if they want to inspect the threads

[02:06:00.0000] <othermaciej>
is anyone here who was at the html wg telecon today?

[02:55:00.0000] <Hixie>
i love how people talk about the content of my e-mails instead of the content of the spec

[02:56:00.0000] <Hixie>
who cares if html5 can replace atom or not -- the spec doesn't take a position on the issue

[02:56:01.0000] <Hixie>
it just defines how to go from text/html to atom

[02:58:00.0000] <hsivonen>
Hixie: if it's only for conversion within a server, it's not an interop issue that belongs in the spec

[02:59:00.0000] <Hixie>
who mentioned a server?

[02:59:01.0000] <hsivonen>
Hixie: hence, if *you* have put it in the spec, one assumes that the purpose is that one party publishes an HTML page and another converts it to Atom

[02:59:02.0000] <hsivonen>
which looks a lot like minting a new feed format to me

[03:00:00.0000] <Hixie>
i think y'all are making a mountain out of a molehill

[03:00:01.0000] <Hixie>
it makes sense imho for a format spec to define how it maps to other specssssss

[03:00:02.0000] <Hixie>
specs, even

[03:00:03.0000] <Hixie>
/me pokes at his s key

[03:01:00.0000] <hsivonen>
which reminds me that I should rewrite my HTML-to-feeds converter

[03:01:01.0000] <hsivonen>
the current one is so brittle I can't make changes to it

[03:18:00.0000] <Hixie>
ok bed time... tomorrow i guess i'll have to figure out some solution to the shared workers message port issue

[03:58:00.0000] <Philip`>
http://www.w3.org/News/2009#item79 - "W3C Rescinds Four Proposed Edited Recommendations for XHTML Documents"

[03:58:01.0000] <Philip`>
/me wonders if that has happened before for any other documents

[04:10:00.0000] <gsnedders>
It does kinda tell the XHTML2 WG that they can't just ignore comments, though.

[04:18:00.0000] <hsivonen>
gsnedders: part of the Process-wise problem is, though, that others have moved on and are not bothering to send them comments

[04:18:01.0000] <gsnedders>
Indeed.

[04:19:00.0000] <hsivonen>
gsnedders: so they could follow the Process but still publish stuff that hasn't had proper review

[04:24:00.0000] <hsivonen>
the W3C Process seems to lack a provision for that case

[04:35:00.0000] <gsnedders>
As has been implied before, MLW has now admitted he was at TPAC.

[05:16:00.0000] <othermaciej>
gsnedders: do you think we're giving him too much attention maybe?

[05:17:00.0000] <gsnedders>
/me shrugs

[05:17:01.0000] <othermaciej>
I would say that for instance more attention than 0 is too much

[05:17:02.0000] <gsnedders>
I don't think that'll make any difference though

[05:18:00.0000] <othermaciej>
trolls like attention

[05:19:00.0000] <annevk42>
I was at the telcon, FWIW

[05:20:00.0000] <othermaciej>
annevk42: I just wanted to ask if there was any decision about my proposals for addressing the Formal Objection to the design principles

[05:21:00.0000] <othermaciej>
annevk42: I concluded based on the minutes that there wasn't

[05:21:01.0000] <annevk42>
It seemed that people were mostly uninterested working on it

[05:21:02.0000] <othermaciej>
instead people took it as an opportunity to make unrelated complaints

[05:21:03.0000] <hsivonen>
wonders if objections cease blocking "group consensus" and turn into lack of "community consensus" (or something) when the objector leaves the group

[05:21:04.0000] <annevk42>
Dave Singer would talk to you about it though

[05:22:00.0000] <othermaciej>
all I wanted was a response to my suggestion

[05:22:01.0000] <othermaciej>
if people have unrelated complaints about the document they should post them to the list

[05:23:00.0000] <othermaciej>
or if they think Philip TAYLOR's objection is not worth addressing at all they can say that too

[05:23:01.0000] <othermaciej>
oh well

[05:23:02.0000] <annevk42>
Ah, the response to that was that while it was ok, Laura Carlson had concerns in reply to your email so there was no lazy consensus or some such

[05:23:03.0000] <annevk42>
I think rubys put that on the mailing list already

[05:24:00.0000] <othermaciej>
Laura was one of the very few people to object to any of the principles

[05:25:00.0000] <othermaciej>
it seems like the few people who objected to any of them also objected to a lot of them

[05:29:00.0000] <annevk42>
I'm not sure how we declare consensus  but I get the feeling 1 objector is enough

[05:32:00.0000] <othermaciej>
I reviewed the voting and if people still largely hold their positions, consensus may be impossible

[05:32:01.0000] <othermaciej>
it seems that a large majority of the group agreed with all or nearly all of the principles

[05:32:02.0000] <othermaciej>
and a few people disagreed with all or nearly all

[05:32:03.0000] <othermaciej>
that makes it unlikely that any small changes would reduce the degree of disagreement

[05:33:00.0000] <othermaciej>
that being said, I'd still like to address the Formal Objection as best I can, since Sam asked for it to be addressed

[05:34:00.0000] <othermaciej>
(my point is kind of that deleting a few of the principles won't make the objectors happy, and would probably make some of the people in favor unhappy, while deleting most or all of them would make the document pointless)

[05:37:00.0000] <annevk42>
I think some people, i.e. Laura and Larry, expect the Design Principles to be used as rules, rather than as means of explaining

[05:37:01.0000] <annevk42>
And therefore they want full consensus on them or something. I don't believe that'll ever work.

[05:38:00.0000] <othermaciej>
well, when we originally put them in the wiki, their intent was to be descriptive, not prescriptive

[05:39:00.0000] <othermaciej>
that they would describe the design approach that had been used to draft HTML5 so far

[05:39:01.0000] <othermaciej>
and with the hope that these principles would be useful to others

[05:40:00.0000] <othermaciej>
I think they do help clarify things at times by letting complex repeating themes be expressed in a short way by referencing the principles

[05:40:01.0000] <othermaciej>
but sometimes they are used to argue opposite sites of the same point

[05:40:02.0000] <othermaciej>
or sometimes people reject the premise

[05:41:00.0000] <othermaciej>
I do admit though that I get tired of rehashing basic arguments about error handling and working with existing Web content

[05:42:00.0000] <othermaciej>
so maybe it would be nice to have consensus about those kinds of things

[05:42:01.0000] <annevk42>
I was happy to find the W3C specification guidelines and it citing error handling as being important

[05:42:02.0000] <annevk42>
Nobody replied when I pointed that out though.

[05:43:00.0000] <othermaciej>
incidentally I don't think Larry's distinction between "Technical Specification" and "Applicability Statement" matches IETF practice

[05:43:01.0000] <othermaciej>
I almost pointed it out on the list, but then I concluded that debating the fine points of IETF-specific terms of art was not really on topic

[05:45:00.0000] <annevk42>
I find TS and AS to be very vaguely defined so you can give almost any meaning to them you want to

[05:45:01.0000] <othermaciej>
I think that the intent was that things like http://tools.ietf.org/rfc/rfc793.txt are Technical Specifications and things like http://tools.ietf.org/rfc/rfc1122.txt are Applicability Statements

[05:45:02.0000] <othermaciej>
note that both of them specify various kinds of error handling

[05:45:03.0000] <annevk42>
How Larry comes to the statement that error handling is not part of TS is somewhat beyond me though (although some other people applauded him for it...)

[05:45:04.0000] <othermaciej>
so wether error handling is specified is irrelevant

[05:46:00.0000] <othermaciej>
certainly that's not justified by the BCP he linked

[05:46:01.0000] <othermaciej>
anyway, I think the difference is meant to be between things like RFC793 that describe a protocol or format, and things like RFC1122 that give requirements for deploying combinations of protocols/formats/etc in a particular environment

[05:47:00.0000] <hsivonen>
I think the concept of "Technical Specification" isn't helpful for HTML5

[05:47:01.0000] <othermaciej>
having conformance criteria, error handling, or rules for specific classes of software does not exclude you from being a TS

[05:47:02.0000] <othermaciej>
that too

[05:47:03.0000] <othermaciej>
I didn't realize at first that he was using an IETF-specific term of art

[05:47:04.0000] <othermaciej>
now that I know that, it doesn't seem worthwhile to debate with him

[05:50:00.0000] <annevk42>
I'm trying to debate with him, but he's not actually replying :)

[06:10:00.0000] <Philip`>
Hmm, OpenOffice isn't very good for forms written in Word whose column layout is done using spaces

[06:12:00.0000] <othermaciej>
I am amazed at the complexity of http://hsivonen.iki.fi/doctype/ie8-mode.png

[06:14:00.0000] <annevk42>
Hmm, I wonder if Google will implement Ruby for just Chromium or also WebKit. I reckon it has to be in WebKit. (On a side note, it seems they're looking into both XHTML Ruby and HTML5 Ruby and the CSS3 Ruby module (which I believe has outstanding issues, but nobody ever started working on).)

[06:14:01.0000] <hsivonen>
othermaciej: and it doesn't even visualize the decision of which UA string to use and whether the compat mode button gets hidden

[06:14:02.0000] <othermaciej>
Google is implementing Ruby?

[06:14:03.0000] <hsivonen>
othermaciej: I should probably make it clearer what it means to have a compat mode frame parent

[06:14:04.0000] <othermaciej>
as in the markup?

[06:14:05.0000] <othermaciej>
I thin it would be hard to implement outside the layout engine

[06:15:00.0000] <othermaciej>
but I haven't really heard from them about it

[06:15:01.0000] <hsivonen>
othermaciej: http://code.google.com/p/chromium/issues/detail?id=4016#c3

[06:15:02.0000] <annevk42>
http://lists.w3.org/Archives/Public/www-style/2009May/0189.html

[06:15:03.0000] <annevk42>
and there's another message on some HTML list

[06:16:00.0000] <othermaciej>
well perhaps he'll drop a giant patch bomb when he's done

[06:16:01.0000] <othermaciej>
or maybe the people who do WebKit text stuff already know about his work

[06:16:02.0000] <othermaciej>
/me shrugs

[06:16:03.0000] <annevk42>
http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2009-May/019838.html

[06:21:00.0000] <annevk42>
I wonder if W3C management would rescind SVG 1.2 as well if Bjoern starts complaining about e.g. http://lists.w3.org/Archives/Public/public-iri/2005Jun/0000.html

[06:26:00.0000] <othermaciej>
too bad Bjoern didn't turn out to be as good at editing specs as he is at finding technical flaws

[06:30:00.0000] <Dashiva>
hsivonen: I don't see an arrow for IE=6, how does that behave?

[06:31:00.0000] <Philip`>
jgraham: You need to implement microdata-to-{vCard,iCalendar,BibTeX,Atom} soon, so that people can test the spec and so that I won't feel any compulsion to implement anything myself

[06:31:01.0000] <hsivonen>
Dashiva: I thought that wouldn't be very relevant to test

[06:32:00.0000] <hsivonen>
Dashiva: although now that you mention it, it is an interestingly unobvious case

[06:33:00.0000] <annevk42>
Hmm, why can't <article pubdate> be taken from <time> somewhere within the <article>. That'd be nicer

[06:33:01.0000] <Philip`>
I would guess that it's the same as IE=7, because the IE6 engine was replaced with the IE7 engine rather than being kept alive in parallel

[06:33:02.0000] <Philip`>
but guessing based on logic seems to be quite dangerous when talking about IE modes

[06:36:00.0000] <hsivonen>
It seems to me that the only values one should use are EmulateIE7 and Edge, and the former should only be used temporarily to migrate to the latter

[06:38:00.0000] <Dashiva>
But if you use Edge, you can't get to quirks mode at all?

[06:38:01.0000] <Dashiva>
Or should those arrows be goin to "Quirky or No doctype"?

[06:39:00.0000] <Philip`>
http://philip.html5.org/data/ua-compatible-contents.txt / http://philip.html5.org/data/ua-compatible-headers.txt - lots use IE=7, lots use IE=EmulateIE7, basically nobody uses IE=edge

[06:41:00.0000] <Philip`>
(plus the usual dozy copy-pasters and typoers)

[06:42:00.0000] <hsivonen>
Dashiva: you shouldn't use the quirks mode

[06:42:01.0000] <Dashiva>
Well, yes, but what if you do?

[06:43:00.0000] <hsivonen>
Dashiva: just don't :-)

[06:43:01.0000] <Dashiva>
Maybe you forget to include a doctype on one of your pages

[06:43:02.0000] <Dashiva>
Does IE8 use almost standards or quirks then?

[06:43:03.0000] <hsivonen>
Dashiva: IE8 standards as shown in the diagram

[06:43:04.0000] <Dashiva>
okay

[06:44:00.0000] <hsivonen>
Dashiva: so you should always have a standards-mode or almost standards-mode doctype

[06:44:01.0000] <hsivonen>
since otherwise you can put other browsers in quirks without putting IE8 to quirks

[06:44:02.0000] <Dashiva>
Yeah, that's what I was wondering if would happen. It seemed odd.

[06:45:00.0000] <hsivonen>
I suppose the desirable semantics would be EmulateIEEdge

[06:45:01.0000] <Philip`>
Isn't that what you get if you don't specify any x-ua-compatible at all?

[06:46:00.0000] <hsivonen>
Philip`: then you get the button

[06:46:01.0000] <Philip`>
Ah

[06:46:02.0000] <hsivonen>
Philip`: or whatever the frame parent did

[06:46:03.0000] <Philip`>
Oh

[06:46:04.0000] <hsivonen>
Philip`: or now whatever the parent did, but you inherit its compat modeness

[06:47:00.0000] <Philip`>
Uh

[06:48:00.0000] <hsivonen>
s/now/not/

[06:48:01.0000] <Philip`>
Ah

[06:49:00.0000] <hsivonen>
the fun part is that you can end up in the IE 5.5 mode with compat modeness or without, so knowing that frame parent is quirks is not enough knowledge to decide frames

[06:49:01.0000] <hsivonen>
you need to consider *how* the frame parent got there

[06:50:00.0000] <hsivonen>
perhaps I should add decisions to hide the button and to propagate compat mode to child frames as boxes to the flowchart

[06:51:00.0000] <Philip`>
It's very nice of Microsoft to have fixed the unfortunate bug wherein it's nearly possible for a person to understand how their page is going to be rendered

[06:51:01.0000] <annevk42>
I'm not too interested in making adjustments to my site to cater to this mess

[06:51:02.0000] <annevk42>
Though I understand a large part of the Web has already done so :/

[06:52:00.0000] <Philip`>
annevk42: But you're not interested in making adjustments to your site to cater for any other released (or beta?) browser either :-)

[06:53:00.0000] <annevk42>
Ah yeah, I forgot my site renders like crap anyway :D

[06:53:01.0000] <hsivonen>
Philip`: scary how big a proportion copied and pasted from http://intertwingly.net/blog/2008/01/22/Best-Standards-Support and got a bogus result

[06:55:00.0000] <hsivonen>
also, it's sad how much FF=3 and OtherUA=4 get copied and pasted

[06:55:01.0000] <hsivonen>
must remember to cite this when someone naively doubts the rampant cargo cultism in Web authoring

[07:00:00.0000] <gsnedders>
/me stabs PHP for not even having a queue structure

[07:03:00.0000] <Philip`>
gsnedders: Can't you just use an array?

[07:03:01.0000] <Philip`>
and push and shift

[07:03:02.0000] <Philip`>
(That's how it works in Perl anyway)

[07:03:03.0000] <gsnedders>
Shifting is O(n) though :(

[07:04:00.0000] <Philip`>
That's fine if n is small

[07:05:00.0000] <Dashiva>
Use a circular buffer instead of shifting? :)

[07:05:01.0000] <gsnedders>
Philip`: Also, arrays have a 70 byte overhead per value in PHP, which can get quite big.

[07:07:00.0000] <Philip`>
gsnedders: I don't think PHP having a queue structure would solve that problem

[07:07:01.0000] <Philip`>
because they'd presumably implement it in the same way

[07:08:00.0000] <gsnedders>
SPLQueue in PHP 5.3 doesn't have that overhead, but is in a disablable extension (which is why it's called PHP Standard Library, obviously).

[07:08:01.0000] <Philip`>
If I remember correctly, the Python html5lib just uses indexes into constant strings, since that was more efficient than mutable lists

[07:08:02.0000] <Philip`>
(for the input stream, anyway)

[07:09:00.0000] <gsnedders>
Yeah, PHP html5lib does too

[07:12:00.0000] <Madness>
hsivonen, iinm when you feed IE8 the html5 doctype it automatically uses the superstandard layout engine.

[07:12:01.0000] <Philip`>
/me would call it the "less buggy than all the others" layout engine, not the "superstandard" one :-p

[07:13:00.0000] <hsivonen>
Madness: if your domain isn't blacklisted, framed by a compat mode site, etc., etc.

[07:14:00.0000] <hsivonen>
Madness: the main benefit of IE=Edge is getting rid of the button, though, if one happens to care about the button

[07:14:01.0000] <Madness>
Hm, intresting.

[07:19:00.0000] <hsivonen>
what should I do when SVN seems to stop tracking a file that is supposed to be under version control?

[07:20:00.0000] <gsnedders>
"seems" how?

[07:20:01.0000] <Madness>
Check for exclusions?

[07:20:02.0000] <gsnedders>
svn status?

[07:20:03.0000] <hsivonen>
svn stat doesn't mention it after I edit it

[07:20:04.0000] <gsnedders>
Check your ignores

[07:20:05.0000] <hsivonen>
and svn add says it's already under VC

[07:20:06.0000] <hsivonen>
can't see it in ignores

[07:20:07.0000] <gsnedders>
But that shouldn't really have any effect if it is already tracked

[07:20:08.0000] <gsnedders>
*affect

[07:20:09.0000] <gsnedders>
*effect

[07:22:00.0000] <Philip`>
Did you do something crazy like copy a .svn directory from one place to another?

[07:22:01.0000] <hsivonen>
Philip`: I copied my whole local sandbox to another directory

[07:23:00.0000] <hsivonen>
surely .svn has the good sense to be relative and not depend on absolute fs location?

[07:23:01.0000] <hsivonen>
ignores look reasonable

[07:24:00.0000] <Philip`>
It's fine if you copy the entire working copy tree, but I've seen it get horribly confused if you copy a subdirectory to another location within the tree

[07:24:01.0000] <hsivonen>
I haven't done that

[07:24:02.0000] <Philip`>
Does svn diff notice any changes?

[07:24:03.0000] <hsivonen>
it doesn't

[07:24:04.0000] <Philip`>
Have you definitely actually changed the file?

[07:25:00.0000] <hsivonen>
Ooh!

[07:26:00.0000] <hsivonen>
no, I hadn't. *blush*

[07:26:01.0000] <gsnedders>
/me quite often does that

[07:27:00.0000] <hsivonen>
I had edited the wrong copy in a neighboring dir

[07:27:01.0000] <Philip`>
Ah, that could explain the behaviour you see :-p

[07:27:02.0000] <gsnedders>
/me closes as notabug

[07:28:00.0000] <hsivonen>
thanks

[08:23:00.0000] <annevk42>
I could see myself turning http://thebjoernhoehrmannproject.org/ into Atom-compatible XHTML5

[08:24:00.0000] <Philip`>
annevk42: For what purpose?

[08:24:01.0000] <annevk42>
So that other people can track it of course :)

[08:24:02.0000] <annevk42>
And since I don't want to maintain an Atom feed or write a Perl script

[08:31:00.0000] <Philip`>
annevk42: web.archive.org indicates that it was updated 3 times in over a year, so I'm not sure it really needs hundreds of feed readers pinging it every 30 minutes looking for updates :-p

[08:33:00.0000] <annevk42>
Philip`, that's not for you to decide :)

[08:46:00.0000] <Philip`>
annevk42: Of course not - it's for Hixie to decide

[08:49:00.0000] <Dashiva>
I thought it was for the community to build consensus about

[08:57:00.0000] <Philip`>
It is, except we don't have consensus on which community needs to build that consensus, so we go with the one-person community consisting of Hixie

[08:58:00.0000] <Dashiva>
I don't think there's consenus on that

[08:58:01.0000] <Dashiva>
+s

[09:05:00.0000] <Philip`>
Dashiva: It doesn't matter as long as Hixie consents with it

[09:19:00.0000] <Philip`>
The bit of HTML 5 that talks about xml:lang is horribly confusing

[09:45:00.0000] <Philip`>
http://philip.html5.org/data/meta-robots.txt - hmm, not much consistency there

[09:46:00.0000] <Philip`>
"index, follow, jogos, game, online, flash, jogos em flash, jogos online, jogos multiplay, multiplay, Anti Pacman" - don't think they've quite got the hang of it yet

[09:50:00.0000] <Philip`>
annevk42: Added <meta name="robots" content="index, nofollow"> to my HTML pages now

[10:37:00.0000] <hober>
Is Sam back on board with the DOM Consistency principle? http://www.w3.org/mid/4A16D275.10202⊙in

[10:40:00.0000] <Philip`>
Where did he indicate he was against it?

[11:08:00.0000] <hober>
Philip`: http://intertwingly.net/blog/2009/04/08/HTML-Reunification

[11:19:00.0000] <Philip`>
hober: Ah, right

[14:10:00.0000] <Hixie>
wow safari does 'autofocus'

[14:10:01.0000] <Hixie>
sweet

[14:35:00.0000] <Philip`>
http://developers.slashdot.org/article.pl?sid=09/05/22/1634215

[14:36:00.0000] <Philip`>
http://developers.slashdot.org/comments.pl?sid=1242425&cid=28058191 - "When's the last time you laid out a site without a table element on every page? Hell, it's almost always the next thing to follow <body> on my pages." - hmm, I think the last time I did that was about 1998

[16:11:00.0000] <othermaciej>
Hixie: apparently Larry doesn't like that you say "must" instead of "is"

[16:12:00.0000] <Hixie>
and i don't like that specs say is instead of must, so...

[16:15:00.0000] <Philip`>
You could say "It must be the case that the behaviour is ..." to combine the best of both worlds

[16:15:01.0000] <Hixie>
lol

[16:18:00.0000] <othermaciej>
right

[16:18:01.0000] <othermaciej>
passive voice

[16:19:00.0000] <annevk42>
I think you can make a point that the browsing context stuff needs to be called out, but nobody has done yet. I somewhat doubt most people got that far...

[16:19:01.0000] <annevk42>
And with browsing context stuff I also mean things like navigation and the page load processing model because that is not at all specific to HTML

[16:20:00.0000] <othermaciej>
I do think "processing requirements" is a category not really captured by the current subtitle

[16:20:01.0000] <othermaciej>
and it's fair enough to say some of the processing requirements go beyond what just goes without saying

[16:22:00.0000] <Hixie>
the subtitle is a waste of time

[16:22:01.0000] <Hixie>
you'll notice the whatwg spec doesn't have one

[16:23:00.0000] <hober>
It's a giant bikeshed that people use to vent all of their other problems with the direction of the spec.

[16:24:00.0000] <annevk42>
yeah, e.g. Larry somehow didn't note the section called "Scope"

[16:24:01.0000] <annevk42>
typically that's where you state what the spec is about, not in the title...

[16:25:00.0000] <Philip`>
hober: If it's giant bikeshed then you need to be pretty careful about what colour you paint it, otherwise it'll be an eyesore that's visible for miles around

[16:25:01.0000] <Hixie>
maybe i should change the subtitle to just "A Technical Specification"

[16:26:00.0000] <Philip`>
Or "A Specification"

[16:26:01.0000] <Philip`>
Could you have specifications that aren't technical?

[16:27:00.0000] <Philip`>
(in the standard meaning of "technical", not the technical IETF meaning of it)

[16:27:01.0000] <annevk42>
"A Work of Fiction" would also be applicable until 2022 :p

[16:28:00.0000] <Hixie>
"Based on a true story!"

[16:28:01.0000] <annevk42>
:)

[16:28:02.0000] <annevk42>
http://twitter.com/jdowdell/statuses/1887218782 ?

[16:29:00.0000] <Philip`>
HTML 5: A Young Lady's Illustrated Primer

[16:34:00.0000] <hober>
Philip` ftw

[16:34:01.0000] <Hixie>
Philip`: maybe once we have diagrams...

[16:34:02.0000] <Philip`>
There's at least, uh, three that I can think of

[16:35:00.0000] <Philip`>
I guess one per hundred pages isn't enough to really count as "Illustrated"

[16:35:01.0000] <Hixie>
right

[16:36:00.0000] <gsnedders>
/me ponders trying to do a presentation with that title

[16:36:01.0000] <gsnedders>
/me then realizes he can't draw, and thus would have major problems with illustrations

[16:38:00.0000] <Philip`>
I suppose there's also the issue that young ladies aren't really the target audience for the spec

[16:38:01.0000] <Philip`>
gsnedders: Just do them all in xfig

[16:38:02.0000] <gsnedders>
I'm totally uncreative

[16:40:00.0000] <inimino>
all the more reason to practice

[16:40:01.0000] <gsnedders>
I gave up a long time ago.

[16:41:00.0000] <inimino>
unless you tried for at least ten years with no progress, you gave up too soon :-)

[16:41:01.0000] <gsnedders>
Well,  must've done art (at school) for around ten years :P

[16:42:00.0000] <inimino>
if you can read and write and do maths you can learn to draw

[16:44:00.0000] <Philip`>
I had a How To Draw Dinosaurs book a while ago

[16:44:01.0000] <Philip`>
I could cope with some of bits where you just trace the shapes that are in the book

[16:44:02.0000] <Philip`>
but that was about my limit

[16:47:00.0000] <inimino>
I recommend a life drawing course

[16:48:00.0000] <inimino>
I've never seen anyone that took one and did not improve considerably

[16:49:00.0000] <Philip`>
Improvement does not imply that the end state is any good :-p

[16:50:00.0000] <othermaciej>
Hixie: I agree, subtitle is a waste of time


2009-05-23
[21:26:00.0000] <othermaciej>
so what's the story with EventSource and server-sent events these days?

[21:29:00.0000] <Hixie>
it's stable

[21:29:01.0000] <Hixie>
not aware of any outstanding issues

[01:49:00.0000] <hsivonen>
http://lists.w3.org/Archives/Member/tag/2009May/0066.html

[01:55:00.0000] <Hixie>
http://lists.w3.org/Archives/Public/public-xhtml2/2009May/0024.html

[01:59:00.0000] <hsivonen>
context: http://lists.w3.org/Archives/Member/tag/2009May/0051.html

[02:21:00.0000] <hsivonen>
http://lists.w3.org/Archives/Member/w3c-ac-forum/2009AprJun/0112.html

[02:28:00.0000] <Hixie>
there's a perception that wg groups fail to meet their deadlines because they always do, because the timetables the w3c put forward are wildly unrealistic, because they consider realistic timetables to be too likely to turn people away

[02:37:00.0000] <mhausenblas>
/me seldom agrees with Hixie, but 'they consider realistic timetables to be too likely to turn people away' deserves a +1 ;) 

[02:39:00.0000] <Hixie>
when the htmlwg was chartered, i said what the html5 timetable was

[02:39:01.0000] <Hixie>
last call in oct 2009, etc

[02:39:02.0000] <othermaciej>
which was clearly unreasonable, to anyone who had any understanding of the matter

[02:40:00.0000] <othermaciej>
at this point I am not sure if we will get to Last Call this decade, given Sam's way of talking about it...

[02:40:01.0000] <Hixie>
they said that was never going to work and instead used this: http://www.w3.org/2007/03/HTML-WG-charter#deliverables

[02:40:02.0000] <Hixie>
last call june 2008, CR by sept last year

[02:41:00.0000] <Hixie>
whatwg's html5 will reach LC in oct 2009

[02:41:01.0000] <Hixie>
i can't speak as to the w3c's :-)

[02:59:00.0000] <zcorpan>
"Add a <code title="">type</code> attribute whose value is the string "<code title="">text/html</code>" to <var title="">x</var>"

[02:59:01.0000] <zcorpan>
Hixie: ^ that seems unnecessary and might be incorrect

[03:00:00.0000] <Hixie>
i thought i removed the type attribute in the case where it was wrong

[03:01:00.0000] <zcorpan>
hmm the next step inserts another <link rel="alternate">

[03:01:01.0000] <Hixie>
oooops

[03:01:02.0000] <Hixie>
looks like i have an extra one

[03:02:00.0000] <zcorpan>
isn't type="" optional anyway?

[03:02:01.0000] <Hixie>
yeah but seems good to include it if you know it

[03:03:00.0000] <Hixie>
ok removed the extra one

[03:03:01.0000] <Hixie>
bed time now

[03:04:00.0000] <Hixie>
uh

[03:04:01.0000] <Hixie>
hsivonen: 

[03:04:02.0000] <Hixie>
Possible conformance errors:

[03:04:03.0000] <Hixie>
"Stream length exceeds limit.": non-document-error io: http://www.whatwg.org/specs/web-apps/current-work/source-validation

[06:40:00.0000] <gsnedders>
/me sighs

[06:40:01.0000] <gsnedders>
Splitting out the input stream from the tokenizer in PHP html5lib is a lot of work

[07:30:00.0000] <gsnedders>
/me finds a fairly major bug

[07:30:01.0000] <gsnedders>
Though actually with only CDATA and RCDATA elements having ASCII names it isn't actually a bug that can ever be hit

[07:31:00.0000] <annevk4>
that doesn't like you discovered a bug then :)

[07:34:00.0000] <gsnedders>
/me wonders whether he should change the tokenizer to allow R?CDATA elements to have non-ASCII names, but seeming it's an anti-optimization, decides against it

[08:44:00.0000] <Philip`>
gsnedders: Why do you want to split it out?

[08:45:00.0000] <gsnedders>
Philip`: To have multiple different InputStream classes, esp. one for files

[08:48:00.0000] <Philip`>
gsnedders: Ah

[08:55:00.0000] <gsnedders>
Huh… line 751 tokenizer.py

[08:55:01.0000] <gsnedders>
Will that not match D followed by any permutation of the characters OTYPE (with repetition allowed)?

[08:56:00.0000] <gsnedders>
s/O/OC/

[08:57:00.0000] <Philip`>
gsnedders: No

[08:58:00.0000] <gsnedders>
Oh, duh.

[08:58:01.0000] <gsnedders>
/me realizes how dumb that was

[08:58:02.0000] <Philip`>
gsnedders: It loops with expected=('o', 'O') first and checks the next character against that, etc

[08:58:03.0000] <gsnedders>
Exactly.

[09:04:00.0000] <gsnedders>
The difficulty of doing this is mainly because the PHP impl assumes it is a string so it has look-ahead and look-behind for the entire input data.

[09:08:00.0000] <Philip`>
I don't think the Python version ever did that, but it did assume it could arbitrarily consume and unconsume characters whenever it fancied

[09:12:00.0000] <takkaria>
yeah, if you can optimise your tokeniser never to use look-behind, that can give you good performance benefits

[09:19:00.0000] <gsnedders>
We need to look-behind per spec, but we can't look behind in, e.g., an http file resource.

[09:23:00.0000] <gsnedders>
/me finally makes it to the final method

[09:30:00.0000] <Philip`>
gsnedders: Why do we need to use look-behind per spec?

[09:30:01.0000] <gsnedders>
Oh, no, we don't.

[09:30:02.0000] <gsnedders>
It's just both PHP and Python impl consume too much :P

[09:30:03.0000] <Philip`>
It needs look-ahead, which I guess would look like look-behind from a different perspective

[09:30:04.0000] <gsnedders>
Yeah, indeed.

[09:35:00.0000] <gsnedders>
The Python deviates from the spec quite a lot when consuming an entity.

[09:38:00.0000] <Philip`>
The spec isn't written a way that can be efficiently implemented

[09:38:01.0000] <Philip`>
*in

[14:04:00.0000] <hsivonen>
"Gregory: What I need to do is to proselytize the use of RDFa for accessibility"

[14:04:01.0000] <hsivonen>
http://www.w3.org/2009/05/13-xhtml-minutes.html

[14:14:00.0000] <Dashiva>
"Gregory: I would like to a DTD for XHTML 1.1+RDFa+Aria... +role"

[14:15:00.0000] <Dashiva>
DTD proliferation :)

[14:18:00.0000] <Philip`>
Sounds like a scalability problem for the W3C, if the number of DTDs they have to host is exponential in the number of technologies they specify

[14:21:00.0000] <gsnedders>
It's factorial, not exponential.

[14:22:00.0000] <gsnedders>
Stop trying to paint the W3C in good light :P

[14:22:01.0000] <Philip`>
I'm pretty sure it's exponential

[14:23:00.0000] <Philip`>
If you have n technologies, then for each technology you can decide to either add it to the DTD or not

[14:23:01.0000] <Philip`>
which means you have a choice between 2 values, n times

[14:23:02.0000] <Philip`>
which results in 2^n possible choices

[14:24:00.0000] <gsnedders>
Permutations

[14:24:01.0000] <gsnedders>
You have the number of permutations of specs you have.

[14:25:00.0000] <Philip`>
No you don't

[14:25:01.0000] <Philip`>
Order doesn't matter

[14:25:02.0000] <gsnedders>
Ah, true.

[14:26:00.0000] <gsnedders>
Duh.

[14:26:01.0000] <Philip`>
And it's not combinations (nCr) either, because r is not restricted to a single value

[14:26:02.0000] <gsnedders>
Yeah, true.

[14:27:00.0000] <Philip`>
(Obviously it wouldn't be exactly 2^n, because nobody's going to want the DTD that contains zero technologies)

[14:27:01.0000] <Philip`>
(but it'll be O(2^n))

[14:32:00.0000] <gsnedders>
2^n-1, no?

[14:33:00.0000] <Philip`>
Nooooo!

[14:33:01.0000] <Philip`>
Constants are ugly

[14:33:02.0000] <gsnedders>
:D

[14:33:03.0000] <Philip`>
Just call it O(2^n) :-p

[14:38:00.0000] <Dashiva>
Order matters if UAs do string matching

[14:39:00.0000] <Dashiva>
Poor code :(

[14:41:00.0000] <Philip`>
They can use a canonical ordering

[14:41:01.0000] <gsnedders>
\sum_{r=1}^{n} nPr then

[14:41:02.0000] <gsnedders>
Ow.

[14:41:03.0000] <Philip`>
so that there aren't multiple orderings of the same set of technologies

[14:45:00.0000] <Dashiva>
What about when users start writing the doctype names by hand?

[14:45:01.0000] <Philip`>
Throw a fatal error if they didn't use the canonical ordering

[15:23:00.0000] <Hixie>
i am baffled

[15:24:00.0000] <Hixie>
when i said last call was october 2009 in 2006, html5 was accused of glacial progress

[15:24:01.0000] <Hixie>
when i say last call is october 2009 in 2009, html5 is accused of going too fast

[15:25:00.0000] <gsnedders>
Logic? Peh.

[15:25:01.0000] <gsnedders>
/me throws it out

[15:28:00.0000] <inimino>
that's only baffling if it's the same people making both sets of accusations

[15:34:00.0000] <gsnedders>
http://www.ietf.org/internet-drafts/draft-sneddon-atom-export-01.txt — feedback welcome.

[15:37:00.0000] <gsnedders>
/me wonders what to do about IRI for atom:category@scheme

[15:40:00.0000] <Philip`>
gsnedders: It would be nice if the abstract said what it was actually about

[15:40:01.0000] <Philip`>
(In particular, what it's meant to be exporting, and from where)

[15:41:00.0000] <hober>
gsnedders: sounds like a description of the backing store I use for my blog :)

[15:45:00.0000] <gsnedders>
/me makes another tweet probably nobody will understand


2009-05-24
[23:25:00.0000] <Hixie>
i'm no longer checking html5 for validity because validator.nu won't let me check it -- says it's too long :-(

[23:25:01.0000] <Hixie>
hsivonen: any chance you could increase the limit again :-)

[00:21:00.0000] <othermaciej>
annevk42: are you around?

[01:08:00.0000] <hsivonen>
Hixie: I'll fix when I get to my work computer

[01:09:00.0000] <Hixie>
sweet :-)

[01:09:01.0000] <hsivonen>
I hadn't realized RDFa had this random restriction: http://www.w3.org/TR/rdfa-syntax/#col_Metainformation

[01:10:00.0000] <hsivonen>
Did I read right that Shane called UAs that lowercase names "legacy"?

[01:11:00.0000] <othermaciej>
hsivonen: well they haven't updated to XHTML2 yet

[01:11:01.0000] <othermaciej>
so obviously

[01:12:00.0000] <Philip`>
hsivonen: Which restriction are you referring to?

[01:12:01.0000] <hsivonen>
Philip`: not generating triples for rels not on that list

[01:13:00.0000] <Philip`>
hsivonen: Ah

[01:13:01.0000] <Philip`>
Seems a really bad idea in terms of forward-compatibility

[01:14:00.0000] <Philip`>
(At least one implementation (rdfQuery) throws a fatal error if you have a rel keyword that's not on the list, which can't be good)

[01:14:01.0000] <hsivonen>
Philip`: are you implying compliance isn't good? :-)

[01:16:00.0000] <Philip`>
hsivonen: Compliance doesn't require implementations to throw fatal errors, just to ignore the value, as far as I can tell

[01:16:01.0000] <hsivonen>
ah

[01:33:00.0000] <mhausenblas>
/me wondering what in http://www.w3.org/TR/rdfa-syntax/#col_Metainformation is random to hsivonen 

[01:36:00.0000] <hsivonen>
mhausenblas: the admittance of non-HTML4 relations seems arbitrary

[01:36:01.0000] <mhausenblas>
ahm. and if you compare that to http://wiki.whatwg.org/wiki/RelExtensions  -  this is good practice, then, right? ;)

[01:37:00.0000] <hsivonen>
mhausenblas: license is there. also role. nofollow isn't there

[01:37:01.0000] <mhausenblas>
if nofollow isn't there then it's a bug, IMHO

[01:37:02.0000] <mhausenblas>
I'll check and bring it up at next telecon, thanks

[01:38:00.0000] <hsivonen>
mhausenblas: the whatwg wiki doesn't make future-limiting processing restrictions

[01:38:01.0000] <mhausenblas>
/me is not in the position to judge, but I personally like and prefer the 'open' vocabulary world

[01:39:00.0000] <Philip`>
/me will refrain from making any suggestions on handling rel="alternate stylesheet" similarly to how HTML5 microdata handles it (as a single token)

[01:39:01.0000] <hsivonen>
a wiki page seems more 'open' than a list created by a TF

[01:39:02.0000] <mhausenblas>
experience tells that even if you allow people to come up with their own voc, over time the economics enforce few vocs per topic to survive 

[01:40:00.0000] <Philip`>
/me also refrains from commenting on rel="STYLESHEET" etc, which seemingly won't be accepted by an RDFa processor

[01:40:01.0000] <mhausenblas>
hsivonen: open in the sense of everyone can publish everywhere everything (why a single Wiki?)

[01:41:00.0000] <hsivonen>
mhausenblas: to avoid the step where you have overlapping vocabularies

[01:41:01.0000] <mhausenblas>
hsivonen: in my world this does not happen as each voc lives in its own namespace

[01:41:02.0000] <Philip`>
Regardless of where valid rel values are defined, they're going to be defined somewhere outside of RDFa, and so if RDFa defines it own parallel list then it's going to get out of sync at some point

[01:41:03.0000] <Philip`>
s/it/its

[01:42:00.0000] <mhausenblas>
I guess the Java and other communities have experienced this for years and it seems to work fine

[01:42:01.0000] <mhausenblas>
I have absolutely no idea why people oppose namespaces (note: I'm not talking about XMLNS per se) 

[01:45:00.0000] <mhausenblas>
/me BRB, gotta grab some coffee ;)

[01:54:00.0000] <hsivonen>
mhausenblas: I meant semantically overlapping

[01:55:00.0000] <hsivonen>
Philip`: already got out of sync

[01:55:01.0000] <hsivonen>
(even for XHTML2!)

[01:55:02.0000] <mhausenblas>
hsivonen: ok. but I don't see the problem there

[01:56:00.0000] <mhausenblas>
take for example Goolge. they decided not to reuse the already available vocs such as FOAF, etc.

[01:56:01.0000] <mhausenblas>
fine. no problem. see for example http://iandavis.com/blog/2009/05/googles-rdfa-a-damp-squib#comment-1407

[01:57:00.0000] <hsivonen>
mhausenblas: some people who seem to care more about RDF than about RDFa cheerleading see it as a problem

[01:57:01.0000] <hsivonen>
mhausenblas: http://www.jenitennison.com/blog/node/104 (incl. comments)

[01:58:00.0000] <mhausenblas>
hsivonen: not sure if I understand this. RDFa is just an RDF serialisation

[01:59:00.0000] <mhausenblas>
yes, I know Jeni's post and have left my 0.02€ at http://lists.w3.org/Archives/Public/public-lod/2009May/0095.html

[01:59:01.0000] <Philip`>
It sounds like Google is using it as a serialisation of a non-RDF data format, rather than of RDF

[01:59:02.0000] <Philip`>
(and they're using microformats as a different serialisation of the same internal non-RDF data format)

[01:59:03.0000] <mhausenblas>
Philip`: there you might be right ;)

[02:00:00.0000] <hsivonen>
Philip`: indeed, but it's stille cheered as a success for RDFa. go figure

[02:00:01.0000] <mhausenblas>
the thing is: I'm interested in the Web of Data where data comes from whatever nicely machine-processable format

[02:01:00.0000] <hsivonen>
Philip`: anything that has the slightest appearance of using RDFa is cheered as a success, regardless of conformance or RDF model alignment

[02:01:01.0000] <Philip`>
hsivonen: Kind of like cheering when Google starts using "<!doctype html>" on some of their pages? :-)

[02:02:00.0000] <hsivonen>
Philip`: cheering for just the privacy page is embarrassing

[02:02:01.0000] <othermaciej>
it does seem like there is some degree of fanboyism around RDFa

[02:02:02.0000] <hsivonen>
Philip`: but at least it validated as HTML5 instead of just using the doctype

[02:02:03.0000] <othermaciej>
like excitement about its superficial aspects instead of about the goals it is meant to achieve

[02:03:00.0000] <mhausenblas>
hsivonen: IIRC the microformats community even had a party celebrating it, but this is not the point (and I don't care)

[02:03:01.0000] <othermaciej>
I'll be excited when Web pages use new HTML5 tags/attributes/APIs

[02:04:00.0000] <hsivonen>
mhausenblas: celebrating what?

[02:04:01.0000] <othermaciej>
the doctype is not remotely exciting

[02:04:02.0000] <mhausenblas>
hsivonen: that Google supports microformats

[02:04:03.0000] <Philip`>
It seems there's some degree of fanboyism around every technology

[02:04:04.0000] <hsivonen>
mhausenblas: wow

[02:05:00.0000] <mhausenblas>
to be fair Gentlemen, I think each and everyone of us who has invested couple of years into a technology is happy when there is an uptake

[02:05:01.0000] <mhausenblas>
saying this is not the case is a simply dishonest 

[02:05:02.0000] <othermaciej>
I like it when people use WebKit

[02:05:03.0000] <mhausenblas>
s/is a/is

[02:05:04.0000] <Philip`>
I guess it still could be useful for each group to point out the unjustified fanboyism of each other group, even if we're all doing it

[02:05:05.0000] <mhausenblas>
Philip`: +1

[02:06:00.0000] <othermaciej>
but I have distinctly mixed feelings if they use it in a way where they make significant changes which they don't contribute back

[02:06:01.0000] <othermaciej>
I think what's most important is to be aware of your own unjusfified fanboyism

[02:06:02.0000] <othermaciej>
and try to keep it from influencing technical decisions in a bad way

[02:06:03.0000] <Philip`>
It seems the kind of situation that makes self-awareness difficult, which is why it's helpful to have other people point it out

[02:09:00.0000] <Philip`>
(...though preferably in a way that's not too rude or condescending, I guess)

[02:09:01.0000] <mhausenblas>
as for me I just say: whatever data comes in there I'm happy (bonus if it links to other data, as this is the Web, right?)

[02:09:02.0000] <mhausenblas>
two examples: just started reading the wonderful RESTful Web Services book -  awesome. so much useful stuff in it (and I see perfectly how linked data connects to it)

[02:09:03.0000] <mhausenblas>
other example: started to work with Erik on  RESTful RDF (http://dret.typepad.com/dretblog/2009/05/rest-and-rdf-granularity.html)

[02:10:00.0000] <mhausenblas>
Philip`: yes. being polite and honest doesn't exclude each other ;)

[02:12:00.0000] <mhausenblas>
ok, back to some hacking ... cya laters

[02:23:00.0000] <othermaciej>
"These are the kind of pipe dreams that I used to ridicule semantic web folk about back before they body-snatched me!"

[02:23:01.0000] <othermaciej>
now that's a high degree of self-awareness

[03:03:00.0000] <mhausenblas>
(just FYI, the discussion here earlier motivated me to do a quick write-up for starting to collect RDF anti-patterns)

[03:03:01.0000] <mhausenblas>
see http://chatlogs.planetrdf.com/swig/2009-05-24.html#T09-00-52

[03:03:02.0000] <mhausenblas>
(back to hacking now, really ;)

[04:52:00.0000] <annevk42>
othermaciej, am now

[04:53:00.0000] <othermaciej>
annevk42: I was going to ask you to remind me how to check out and generate the Design Principles draft

[04:53:01.0000] <othermaciej>
I don't think I have a checkout of the spec here

[04:57:00.0000] <annevk42>
http://dev.w3.org/cvsweb/

[04:58:00.0000] <annevk42>
it's in html5/html-design-principles

[04:58:01.0000] <annevk42>
you want to edit Overview.src.html

[04:59:00.0000] <annevk42>
mhausenblas, nofollow is defined in the HTML5 spec itself

[04:59:01.0000] <annevk42>
mhausenblas, so it doesn't need to be on RelExtensions

[04:59:02.0000] <mhausenblas>
ok

[05:07:00.0000] <othermaciej>
what's the right place to send EventSource feedback?

[05:07:01.0000] <annevk42>
public-webapps⊙wo

[05:28:00.0000] <othermaciej>
hmm I can't seem to remember my w3c cvs password

[05:29:00.0000] <othermaciej>
guess I'll have to ask a cvs admin for help

[05:32:00.0000] <annevk42>
it works with an ssh keypair or whatever that's called

[06:48:00.0000] <gsnedders>
/me wonders why the wifi has been so slow

[06:48:01.0000] <gsnedders>
Suddenly it jumped just as I said that from 10 to 200 KB/s

[06:48:02.0000] <gsnedders>
And back down

[06:49:00.0000] <gsnedders>
/me is trying to download the spec :P

[06:53:00.0000] <hsivonen>
whoa Anolis adds more bytes than the split-out sections take away

[06:54:00.0000] <gsnedders>
Back down to a number of bytes per second

[07:28:00.0000] <hsivonen>
Hixie: increased the limit to 4 MB

[07:41:00.0000] <karlcow>
[04:06]  <othermaciej> it does seem like there is some degree of fanboyism around RDFa

[07:41:01.0000] <karlcow>
s/RDFa/whatwg|html5|hixie|$fan/ ($fan = pick your own fav topics of a community), which makes the sentence stupid

[07:41:02.0000] <karlcow>
[04:08]  <Philip`> It seems there's some degree of fanboyism around every technology

[07:41:03.0000] <karlcow>
fully agreed. 

[08:10:00.0000] <Philip`>
karlcow: It doesn't make the sentence stupid, it just makes it less general than it perhaps should be

[08:19:00.0000] <gsnedders>
/me guesses he ought to do school work again

[08:19:01.0000] <gsnedders>
/me sighs

[08:25:00.0000] <annevk4>
 /whois Hixie

[08:26:00.0000] <annevk4>
/me tries to figure out why HTTP is no longer functioning

[08:31:00.0000] <gsnedders>
/me wonders what IRI to use for atom:category@scheme in a spec

[08:32:00.0000] <zcorpan_>
... he refers to a note from Simon Pieters

[08:32:01.0000] <zcorpan_>
... I am sure we addressed that."

[08:33:00.0000] <zcorpan_>
i have not received a reply to the email that björn cited

[09:10:00.0000] <gsnedders>
Nice. Back down to 50% test cases failing.

[09:44:00.0000] <jgraham>
/me wonders if he has missed anything, sees a depressing amount of email

[10:11:00.0000] <gsnedders>
/me is thankful html5lib has so many test cases with how much he's broken it

[10:18:00.0000] <jgraham>
gsnedders: Did anyone ever tell you that the order of the errors isn't significant in the tokenizer tests

[10:18:01.0000] <gsnedders>
jgraham: No

[10:18:02.0000] <jgraham>
Oh well they should have done

[10:18:03.0000] <jgraham>
At least I think that is true

[10:18:04.0000] <gsnedders>
This doesn't account for 500 test cases failing, though :D

[10:19:00.0000] <jgraham>
No. But It might account for some ailing sonce you fix the major issues

[10:19:01.0000] <gsnedders>
I think it fixed one, the order of errors.

[10:25:00.0000] <jgraham>
Philip`: I plan to implement microdata-to-*

[10:26:00.0000] <gsnedders>
/me gets down to 61 errors

[10:27:00.0000] <gsnedders>
Huh. Somehow it goes from afterAttributeValueQuoted to attributeName

[10:29:00.0000] <gsnedders>
Oh, wait. I hadn't re-run the tests since I fixed that illogic.

[10:32:00.0000] <gsnedders>
And down to 25…

[10:49:00.0000] <Philip`>
jgraham: I told him that the order of errors isn't significant when the test case has ignoreErrorOrder:true or whatever it is

[10:56:00.0000] <gsnedders>
Which is different to what jgraham asked.

[11:01:00.0000] <Philip`>
gsnedders: Indeed

[11:01:01.0000] <Philip`>
The idea is that the tests which don't have ignoreErrorOrder ought to have predictable error order in any sane implementation, so you should test the order in order to detect bugs, though technically it's not required by the spec

[11:06:00.0000] <gsnedders>
2000 line diff. Fun.

[11:07:00.0000] <jgraham>
/me notes that a PHP implementation might not be considered sane :)

[11:08:00.0000] <gsnedders>
/me agrees

[11:09:00.0000] <gsnedders>
/me has been adding things to his post entitled "PHP Grievances" while working on html5lib over the past few days :)

[11:14:00.0000] <gsnedders>
Philip`: Oh, and re: yesterday, there is look-behind in the data-state.

[11:17:00.0000] <takkaria>
gsnedders: there doesn't have to be

[11:17:01.0000] <gsnedders>
There isn't in Python or PHP implementations, dunno about Ruby…

[11:17:02.0000] <gsnedders>
takkaria: And seeming I just changed the PHP impl. to be like that… :D

[11:17:03.0000] <gsnedders>
http://code.google.com/p/html5lib/source/detail?r=2aa8163f5b86662263bd609f81e40c324d574e42

[11:18:00.0000] <takkaria>
I wonder what we did in hubbub for that

[11:20:00.0000] <takkaria>
oh yeah

[11:21:00.0000] <takkaria>
basically, hubbub uses an inputstream which doesn't support ungetting or peeking backward

[11:22:00.0000] <gsnedders>
Ow.

[11:22:01.0000] <takkaria>
acutally it's a highly efficient implementation

[11:22:02.0000] <takkaria>
we have a current position in the inputstream and then we have an offset number of bytes we are into the buffer

[11:23:00.0000] <gsnedders>
Oh, sure. But it's not going to be fun to implement parts of the spec with :)

[11:23:01.0000] <takkaria>
indeed not

[11:23:02.0000] <takkaria>
but it's not much pain really

[11:35:00.0000] <gsnedders>
So splitting out the input stream was around a 0.1s hit when tokenizing the spec

[11:35:01.0000] <gsnedders>
(Now around 6.1s)

[12:01:00.0000] <takkaria>
that's not terrible

[12:04:00.0000] <Philip`>
gsnedders: Oh, I forgot about that bit

[12:04:01.0000] <Philip`>
but it's only four characters, and only needed in rare circumstances, so it's not much of a problem to add a special look-behind buffer

[12:05:00.0000] <Philip`>
/me supposes it'd be possible to optimise it to a single int keeping track of the state, instead of keeping all the characters

[12:21:00.0000] <takkaria>
it largely depends on how performant string manipulation is in PHP

[12:23:00.0000] <hsivonen>
/me notes that the tokenizer can be implemented without lookagead or lookbehind

[12:23:01.0000] <hsivonen>
*lookahead

[12:23:02.0000] <takkaria>
at some point I should really look at the v.nu parser and see how you remove lookahead

[12:24:00.0000] <takkaria>
because I can imagine it being a fairly decent performance benefit

[12:55:00.0000] <gsnedders>
takkaria: Basically, I think the biggest limit when dealing with the data state is going to be function call overhead.

[12:55:01.0000] <gsnedders>
takkaria: Function call overhead is a non-negilable amount of time when tokenizing the spec, so basically have to rely as much as possible on language constructs and not functions.

[12:57:00.0000] <Philip`>
Write the whole parser as a single giant regular expression, then you won't have any non-native function calls

[13:02:00.0000] <gsnedders>
Philip`: "The maximum length of a compiled pattern is 65539 (sic) bytes  if  PCRE

[13:02:01.0000] <gsnedders>
       is compiled with the default internal linkage size of 2."

[13:02:02.0000] <gsnedders>
Philip`: Also: "All values in repeating quantifiers must be less than 65536."

[13:06:00.0000] <takkaria>
gsnedders: I'm sure that the function call overhead is a pretty big portion, but string use will be a factor

[13:06:01.0000] <takkaria>
hubbub tries to delay string handling to as late as possible by using counters and indexes into buffers instead of doing string ops

[13:06:02.0000] <takkaria>
and then does the string ops when tokens are emitted

[13:07:00.0000] <takkaria>
I dunno, it just might be an avenue you could look at if you're intending on optimising further

[13:07:01.0000] <gsnedders>
http://stuff.gsnedders.com/phphtml5lib.pdf

[13:07:02.0000] <gsnedders>
takkaria: We work one byte at a time

[13:07:03.0000] <gsnedders>
takkaria: (Basically, we use a single byte long string)

[13:07:04.0000] <Philip`>
Do you have anything like charsUntil?

[13:08:00.0000] <Philip`>
(from Python's html5lib)

[13:08:01.0000] <gsnedders>
Philip`: Yes

[13:08:02.0000] <Philip`>
Does that work one byte at a time?

[13:08:03.0000] <gsnedders>
No

[13:08:04.0000] <gsnedders>
Well, I guess within PHP it does :P

[13:08:05.0000] <gsnedders>
(Actually, I think PHP just uses libc for strspn, so within libc it does)

[13:09:00.0000] <Philip`>
Okay, so when you say you work one byte at a time I should ignore you :-)

[13:09:01.0000] <gsnedders>
Look at that graph: the big drop was when I added that.

[13:09:02.0000] <gsnedders>
Apart from when we grab a lot of stuff at once, we work one byte at a time :D

[15:05:00.0000] <Philip`>
Does SGML precisely define some concept like case-insensitivity?

[16:02:00.0000] <gsnedders>
/me pulls out his battered second hand copy of the SGML Handbook

[16:04:00.0000] <Philip`>
Hmm, I prefer my books to be covered with breadcrumbs

[16:05:00.0000] <gsnedders>
I think not

[16:06:00.0000] <gsnedders>
No, it does

[16:06:01.0000] <gsnedders>
Uppercase letters (A-Z, numbers 65–90) map to their lowercase equivalents (a–z, numbers 97–122).

[16:07:00.0000] <gsnedders>
You have to look in about five different places to find that, but it seems well enough defined.

[16:07:01.0000] <Philip`>
Ah, thanks

[16:11:00.0000] <gsnedders>
Hmm, I have an unkillable Terminal

[16:12:00.0000] <gsnedders>
Like, running kill -KILL against it has no effect

[16:15:00.0000] <Philip`>
That's easy to fix - reboot

[16:15:01.0000] <gsnedders>
That means terminating everything else, though

[16:16:00.0000] <Dashiva>
How do you know it's unkillable? E.g. can you still use it?

[16:16:01.0000] <gsnedders>
Dashiva: No

[16:16:02.0000] <gsnedders>
Oh well, reboot it is.

[16:17:00.0000] <Philip`>
Can't you just ignore it?

[16:17:01.0000] <gsnedders>
I can't open stuff now either. WTF?

[16:17:02.0000] <gsnedders>
brb

[16:24:00.0000] <gsnedders>
wee… sanity returns!

[16:26:00.0000] <Dashiva>
You might want to question the sanity of your computer if a stubborn terminal can prevent programs from opening

[16:27:00.0000] <gsnedders>
It's my computer, so of course it fails to meet the definition of sane.

[16:32:00.0000] <Philip`>
/me is reminded of the fun times when his gamin server dies, and so any program that tries to connect to it on startup hangs, and eventually when he realises the problem and restarts the daemon suddenly dozens of copies of Krusader and KPDF spring into existence as they get unblocked


2009-05-25
[23:00:00.0000] <takkaria>
hmm

[23:26:00.0000] <othermaciej>
I don't know why I keep replying to Larry

[03:30:00.0000] <hsivonen>
annevk4: why doesn't CORS allow multiple enumerated allowed origins?

[03:31:00.0000] <hsivonen>
annevk4: being able to specify only one foreign origin in Access-Control-Allow-Origin is inconvenient if you want to serve fonts for a couple of origin but don't want to enable random bandwidth leeches

[03:33:00.0000] <hsivonen>
too late to change it now, of course

[03:36:00.0000] <hsivonen>
http://realtech.burningbird.net/web/page-markups/loose-set-notes-rdfa-microdata-and-html5#comment-1494

[03:43:00.0000] <Hixie>
does chaals think that consensus means unanimity?

[03:48:00.0000] <Philip`>
"To me, consensus seems to be the process of abandoning all beliefs, principles, values and policies. So it is something in which no one believes and to which no one objects."

[03:53:00.0000] <jgraham>
You know things are bad when Margret Thatcher quotes seem apropos

[03:55:00.0000] <webben>
Hixie: Possibly. I don't think http://lists.w3.org/Archives/Public/public-html/2009May/0299.html clearly implies that though.

[03:55:01.0000] <webben>
(if that's what you're thinking of)

[04:01:00.0000] <annevk42>
hsivonen, because we didn't want to introduce complicated features

[04:01:01.0000] <annevk42>
hsivonen, you can just as well do it conditionally

[04:08:00.0000] <Dashiva>
because, even though testable, in practice

[04:08:01.0000] <Dashiva>
the implementation conformance requirement is unnecessary

[04:08:02.0000] <Dashiva>
http://lists.w3.org/Archives/Public/public-html/2009May/0307.html

[04:08:03.0000] <Dashiva>
Is this really still an issue... gah

[04:11:00.0000] <webben>
/me finds it odd that vagueness is promoted as a good way to do "extensibility".

[04:11:01.0000] <annevk42>
I'm surprised nobody has pointed out that our charter allows for non-consensus decisions

[04:12:00.0000] <webben>
annevk42: I wonder which sense of "consensus" the charter is using.

[04:13:00.0000] <webben>
also does the charter basically mean that failing consensus, and after recording opinions, it's the chair's decision?

[04:13:01.0000] <webben>
oh i see the link explains

[04:14:00.0000] <annevk42>
http://www.w3.org/2007/03/HTML-WG-charter#decisions

[04:14:01.0000] <annevk42>
It actually also supports the editing model Hixie has been using. I.e. editor puts forward a proposal which is then refined by group discussion and gets killed or more refined if there is not sufficient support.

[04:15:00.0000] <webben>
hmm... http://www.w3.org/2005/10/Process-20051014/policies#Votes doesn't seem to say what happens if there is a hung vote.

[04:19:00.0000] <hsivonen>
annevk42: I don't know how to do it conditionally with Apache and doing it conditionally doesn't mix well with long-term caching of fonts

[04:20:00.0000] <othermaciej>
his http://lists.w3.org/Archives/Public/public-html/2009May/0323.html definitely sounds like Chaals thinks consensus == unanimity

[04:23:00.0000] <webben>
Yep.

[04:24:00.0000] <Dashiva>
It does sound like it, but based on what chaals has said before I think he just means that he should put a really good effort in to satisfy everyone first

[04:24:01.0000] <Dashiva>
*that we should

[04:25:00.0000] <webben>
W3C could probably use a page like http://en.wikipedia.org/wiki/Wikipedia:What_is_consensus%3F

[04:28:00.0000] <annevk42>
hsivonen, the cache would be per-origin

[04:29:00.0000] <annevk42>
hsivonen, you'd need to specify Vary: Origin

[04:30:00.0000] <othermaciej>
I hate citing the Process document to people

[04:30:01.0000] <hsivonen>
annevk42: seems more complex than one line in Apache config :-(

[04:31:00.0000] <hsivonen>
well, waiting for mod_cors, I guess 

[04:32:00.0000] <hsivonen>
I suppose this could already be done with variables on conditional mod_headers directives

[04:32:01.0000] <othermaciej>
Dashiva: actually, now that I read his email closer it sounds like he's blurring the distinction between W3C-style consensus and unanimity

[04:32:02.0000] <hsivonen>
but I've never studied how those work

[04:33:00.0000] <annevk42>
I think it can be done with the set environment variable stuff

[04:33:01.0000] <annevk42>
Probably with similar lines rubys used for the IE8 crap that now leaked on the Web

[04:33:02.0000] <annevk42>
Why not make the fonts available for everyone?

[04:33:03.0000] <hsivonen>
annevk42: well, that particular case suggests that Apache env vars aren't really copy-paste-friendly material

[04:34:00.0000] <hsivonen>
annevk42: I was going to blog on how to optimize @font-face bandwidth without subsetting fonts

[04:34:01.0000] <hsivonen>
and not letting random sites leech seems like a natural part of it

[04:34:02.0000] <annevk42>
I believe only Gecko enforces the same-origin restriction on them by the way

[04:35:00.0000] <hsivonen>
annevk42: I expect Firefox 3.5 to gain enough market share to deter leeches

[04:36:00.0000] <Philip`>
/me would like to see how much a 500KB font can be optimised without subsetting :-)

[04:37:00.0000] <hsivonen>
Philip`: a 1 MB font can be optimized quite a bit

[04:37:01.0000] <annevk42>
lol, did othermaciej, hsivonen, and I just wrote essentially the same email?

[04:38:00.0000] <Philip`>
annevk42: You are all manifestations of the hivemind

[04:39:00.0000] <othermaciej>
similar, not quite identical

[04:40:00.0000] <Philip`>
hsivonen: How about 10MB fonts, then? :-)

[04:43:00.0000] <hsivonen>
Philip`: dunno. The largest fonts I use are 1.5 MB: http://hsivonen.iki.fi/fonts/libertine/4.4.1/

[04:43:01.0000] <hsivonen>
Philip`: small fonts compress nicely, too: http://hsivonen.iki.fi/fonts/liberation/1.04.93.devel/

[04:45:00.0000] <Philip`>
Hmm, http://en.wikipedia.org/wiki/Kochi_font (9MB for one .ttf) seems to be the largest font I have, excluding the ones I stole from Windows and OS X

[05:21:00.0000] <gsnedders>
http://w3.org/mid/4A1A4AF8.1030001⊙iaaj — never miss monthly deadlines again!

[05:23:00.0000] <annevk42>
http://fberriman.com/2009/05/24/html5-microdata-over-cooked/

[05:23:01.0000] <hsivonen>
jgraham: added info on how to check out trunk to http://about.validator.nu/htmlparser/

[05:24:00.0000] <jgraham>
hsivonen: Thanks

[05:24:01.0000] <annevk42>
hsivonen, did you see http://schepers.cc/?p=117 ?

[05:24:02.0000] <hsivonen>
annevk42: yes. my comment is in the moderation queue.

[05:24:03.0000] <hsivonen>
annevk42: I think it's night time on shepazu's time zone

[05:24:04.0000] <annevk42>
myakura's twitter stream is nice

[05:25:00.0000] <hsivonen>
annevk42: http://pastebin.ca/1433802

[05:26:00.0000] <annevk42>
I just checked, even the 1999 Process document does not say objections need to be put *in* the document. They just need to be archived and referencable. http://www.w3.org/Consortium/Process/Process-19991111/activities.html#WGArchiveMinorityViews

[05:29:00.0000] <annevk42>
Hmm, Doug doesn't point to your original objection

[05:30:00.0000] <annevk42>
oh, but searching for W3C Technology Stack gives it back as the second unindented result :)

[05:32:00.0000] <hsivonen>
also, http://www.google.com/search?q=Web+Technology+Stack and http://www.google.com/search?q=Browser+Technology+Stack

[05:33:00.0000] <annevk42>
"official doctrine of the WHATWG" :)

[05:35:00.0000] <hsivonen>
I wasn't aware that the WHATWG had official doctrines

[05:35:01.0000] <Philip`>
Maybe we should set up a wiki page defining the official doctrines

[05:36:00.0000] <hsivonen>
should Hixie's IRC remarks be considered to be 'ex cathedra'?

[05:38:00.0000] <annevk42>
I think Doug misunderstood that you didn't want to make a diagram for the Web but just one for browsers

[05:38:01.0000] <jgraham>
I thought Hixie was more like Christ himself. His IRC comments are like the sermon from the mount

[05:39:00.0000] <annevk42>
what is HID btw?

[05:39:01.0000] <hsivonen>
annevk42: perhaps I should have revised the URL to match the title...

[05:39:02.0000] <hsivonen>
annevk42: but cool URIs don't change now

[05:39:03.0000] <hsivonen>
annevk42: Human Interface Devices

[05:40:00.0000] <Philip`>
jgraham: We need to work out these issues before Hixie dies, otherwise we won't know whether to wait for his resurrection or form a conclave to elect a new Hixie

[05:41:00.0000] <Philip`>
Of course it's more likely that Hixie is Brian

[05:41:01.0000] <gsnedders>
Philip`: Or we could just wait or his second coming.

[05:41:02.0000] <gsnedders>
*for

[05:42:00.0000] <jgraham>
Philip`: The whole resurrection thing is overrated anyway because he just disappears off again straight afterwards. So we probably need to go with the whole "elect Hixie's representative on earth" thing anyway

[05:43:00.0000] <gsnedders>
And then we'll end up reforming WHATWG so we no longer have a representative of Hixie on earth, and just follow Hixie himsel.

[05:43:01.0000] <Philip`>
I think I'm going to stick with the Hixie-as-Brian idea, because that's got much better songs

[05:44:00.0000] <gsnedders>
"He grew, grew, grew and grew, grew up to be/A man, a man called Hixie."

[05:45:00.0000] <jgraham>
"He's not the messiah, he's a very naughty boy"

[05:45:01.0000] <jgraham>
Somehow that is even funnier in this context

[05:46:00.0000] <gsnedders>
/me is listening to Brian Song by Monty Python from Monty Python Sings

[05:59:00.0000] <othermaciej>
Sam is really confusing me now

[06:05:00.0000] <annevk42>
othermaciej, I think in part it may have been differences between the various Process document versions

[06:06:00.0000] <annevk42>
othermaciej, just skimmed through the two and things have changed

[06:06:01.0000] <othermaciej>
that explains his public-html email

[06:06:02.0000] <othermaciej>
it does look like he was using a stale version

[06:07:00.0000] <othermaciej>
I'm confused by his www-archive email

[06:10:00.0000] <annevk42>
if you're interested in contradictions btw, search for "willful violation" throughout the HTML5 specification

[06:11:00.0000] <hsivonen>
is there some kind of simple script that opens a local HTTP "proxy" that responds with zero-length 200 or 404 to every request?

[06:12:00.0000] <annevk42>
"That may be because there's no way to register them. So claiming extensions aren't possible isn't helpful, when reality proves they are possible." -- is it just me or does this sound weird coming from Julian?

[06:13:00.0000] <othermaciej>
annevk42: I'm mainly interested in which ones Sam thinks are critical

[06:13:01.0000] <othermaciej>
but thanks for pointing that out

[06:14:00.0000] <hsivonen>
annevk42: what Julian says makes sense to me

[06:14:01.0000] <othermaciej>
sounds like the "url" willful violations make sense

[06:14:02.0000] <othermaciej>
hsivonen: it makes sense to me too, but I agree with annevk42 that it sounds unlike Julian's usual prescriptivist point of view

[06:16:00.0000] <othermaciej>
annevk42: it

[06:16:01.0000] <othermaciej>
annevk42: it's pretty enlightening to look at these

[06:17:00.0000] <othermaciej>
annevk42: would be an interesting exercise to catalogue and classify them if Sam's intent is really sincere

[06:17:01.0000] <othermaciej>
it sounds like many will be fixed by new or updated specs to refefence

[06:18:00.0000] <othermaciej>
there are surprisingly few

[06:20:00.0000] <othermaciej>
the only contradiction with a W3C specification is charmod

[06:20:01.0000] <othermaciej>
and it would be hard to disagree with that one

[06:29:00.0000] <gsnedders>
/me wonders whether to commit a change to php html5lib which changes the tokenizer from multiple methods to one big switch statement

[06:30:00.0000] <gsnedders>
Takes the time to tokenize the spec down from 7.0s to 5.7s

[06:30:01.0000] <Philip`>
What reasons are there against committing it?

[06:30:02.0000] <gsnedders>
It's a bit more messy

[06:31:00.0000] <Philip`>
This is HTML, it's meant to be messy

[06:32:00.0000] <jgraham>
gsnedders: Will it make the code significantly harder to maintain?

[06:32:01.0000] <Philip`>
It's code that shouldn't need to be extended significantly in the future (since the spec isn't going to change much), so maintenance isn't the most critical issue

[06:32:02.0000] <gsnedders>
jgraham: I expect that depends on who you ask :)

[06:33:00.0000] <jgraham>
Philip`: In practice the spec has changed a bit and I would expect some further changes as browsers implement this stuff

[06:34:00.0000] <jgraham>
gsnedders: Assume I am asking you

[06:34:01.0000] <gsnedders>
jgraham: No, it's only a small bit worse.

[06:35:00.0000] <gsnedders>
/me pushes

[06:35:01.0000] <Philip`>
/me falls

[06:35:02.0000] <gsnedders>
Wait, I should probably run the test suite :P

[06:36:00.0000] <gsnedders>
Oh, good. Nothing broke.

[06:36:01.0000] <jgraham>
gsnedders: BTW it seems like it would be a much more useful use of your time to make a treebuilder rather than work on endless refactorings of the tokenizer

[06:36:02.0000] <gsnedders>
jgraham: But it's 10 times quicker now!

[06:36:03.0000] <Philip`>
gsnedders: Yes, but it's useless

[06:36:04.0000] <Philip`>
because nobody just wants tokens :-p

[06:37:00.0000] <gsnedders>
Philip`: The test suite does :P

[06:37:01.0000] <jgraham>
gsnedders: The test suite is not an end user

[06:37:02.0000] <Philip`>
The test suite doesn't like being anthropomorphised

[06:38:00.0000] <hsivonen>
meanwhile, I'm confused by the perf results I'm seeing with HTML5-enabled Gecko builds

[06:38:01.0000] <jgraham>
gsnedders: It's not until you have a treebuilder that people can do thinks like implement rdfa parsers with your code

[06:38:02.0000] <gsnedders>
jgraham: You could just work with tokens :P

[06:39:00.0000] <gsnedders>
jgraham: Also, it's more fun working with something that actually runs/

[06:39:01.0000] <Philip`>
gsnedders: You can't just work with tokens unless you write your own treebuilder

[06:39:02.0000] <Philip`>
hsivonen: Confused because it's too fast or because it's too slow?

[06:40:00.0000] <hsivonen>
Philip`: too slow and suddenly *way* too slow

[06:40:01.0000] <Philip`>
Ah

[06:43:00.0000] <jgraham>
hsivonen: A regression or a new measurement?

[06:44:00.0000] <hsivonen>
the measurement is supposedly the same

[06:45:00.0000] <jgraham>
hsivonen: So a regression?

[06:54:00.0000] <hsivonen>
I suppose so. but I have no idea how my own changes could have caused it. I did merge in a lot of trunk code, though.

[06:56:00.0000] <zcorpan>
Hixie: heya

[06:57:00.0000] <annevk42>
not awake I think

[06:57:01.0000] <zcorpan>
hmm right

[07:08:00.0000] <othermaciej>
I don't think I can reply to Sam's latest email without further pissing him off

[07:11:00.0000] <othermaciej>
specifically http://lists.w3.org/Archives/Public/www-archive/2009May/0067.html to be specific

[07:41:00.0000] <jgraham>
/me is not sure wtf there is a discussion about the title

[07:45:00.0000] <othermaciej>
I kinda think it is because of the reason I suggested that made Sam so angry

[07:59:00.0000] <annevk42>
othermaciej, oh, I just replied as well

[08:00:00.0000] <annevk42>
othermaciej, I thought you were gonna leave it

[08:07:00.0000] <othermaciej>
annevk42: I replied to only the technical portion

[08:07:01.0000] <othermaciej>
will decide whether to address the other part later

[08:13:00.0000] <zcorpan>
alt text for http://hsivonen.iki.fi/web-stack/ is interesting

[08:15:00.0000] <annevk42>
http://twitter.com/uberChick/statuses/1909606174 -- I guess that's why we didn't do it

[08:16:00.0000] <jgraham>
zcorpan: In what way interesting?

[08:17:00.0000] <zcorpan>
jgraham: not the specified alt text but what is appropriate alt text, considering html5 and wcag etc

[08:18:00.0000] <zcorpan>
i.e. what would you use as alt

[08:19:00.0000] <annevk42>
given the processing model example in HTML5 I'd prolly just describe the diagram

[08:19:01.0000] <annevk42>
WCAG suggests the same; all text of the image should be in alt

[08:21:00.0000] <Dashiva>
Laura sure loves citing :)

[08:22:00.0000] <jgraham>
me? Well I guess if I was being good, I would go read the spec... Which would lead me to section 4.8.2.1.2 A phrase or paragraph with an alternative graphical representation: charts, diagrams, graphs, maps, illustrations. But I would quickly realise that was unreasonably hard so I would look at 4.8.2.1.9 A key part of the content, realise that was still too hard

[08:22:01.0000] <jgraham>
and so write something short like hsivonen did

[08:23:00.0000] <annevk42>
Dashiva, she cites a lot of discussion but then when she starts making statements she doesn't cite anything

[08:23:01.0000] <annevk42>
Dashiva, I personally don't find it useful

[08:24:00.0000] <Dashiva>
annevk4: This is true

[08:24:01.0000] <hsivonen>
I initially started outlining the idea for the diagram as text and decided that it was easier to draw it when the bulleted list started ballooning

[08:44:00.0000] <othermaciej>
hsivonen: I don't know how much you care about correct stacking of your browser technology stack diagram, but shouldn't XHR be in some way on top of XML

[08:44:01.0000] <othermaciej>
?

[08:45:00.0000] <othermaciej>
also I'm unsure why Events are on top of JavaScript; in general it seems unclear what layers JavaScript should be on top of or below

[08:45:01.0000] <Philip`>
And on JSON

[08:45:02.0000] <Philip`>
(XHR, that is)

[08:45:03.0000] <Philip`>
unless JSON is on XHR

[08:45:04.0000] <othermaciej>
XHR doesn't depend technologically on JSON

[08:45:05.0000] <Philip`>
or unless something else is on both JSON and XHR

[08:46:00.0000] <othermaciej>
the emerging native JSON APIs might be worth calling out

[08:46:01.0000] <Philip`>
Unicode doesn't depend technologically on HTTP / URL, either

[08:46:02.0000] <hsivonen>
othermaciej: it seems to me that XHR is forefost a JS API for HTTP and the shortcut XML parsing is a minor point

[08:46:03.0000] <Philip`>
so I presume that's not what the diagram is showing

[08:46:04.0000] <othermaciej>
fair enough but then maybe it should sit right on top of http

[08:46:05.0000] <jgraham>
othermaciej: My understanding is that "Events" means user-initiated events (although I guess it should still be on top of DOM)

[08:47:00.0000] <hsivonen>
othermaciej: mainly the issue is that laying out the boxes in 2 dimensions restricts what boxes can touch each other

[08:47:01.0000] <othermaciej>
on second thought I guess unicode interposes in the same  way as for text, but unicode on top of http is not a dependency it's a pipeline

[08:47:02.0000] <othermaciej>
yeah

[08:47:03.0000] <Dashiva>
3D canvas to the rescue

[08:47:04.0000] <othermaciej>
diagrams like this end up either unreadable or inaccurate or both

[08:47:05.0000] <hsivonen>
all the specs that are above Unicode really deal with byte layer too and have Unicode inside the parsing rules

[08:48:00.0000] <othermaciej>
the comparison to the W3C diagram is interesting

[08:49:00.0000] <othermaciej>
not sure if XPath is worth citing in your version

[08:50:00.0000] <othermaciej>
/me is looking at WebCore directory names

[08:50:01.0000] <Philip`>
Dashiva: That wouldn't solve the fundamental problem, which is that the topology of finite-dimensional space imposes restrictions on diagrams that are based on adjacencies between non-intersecting shapes

[08:50:02.0000] <othermaciej>
/me flushes with embarassment and moves right on past wml/

[08:51:00.0000] <Philip`>
Seems much easier to solve it by just drawing a graph with potentially intersecting edges

[08:51:01.0000] <othermaciej>
plugins are part of the browser technology stack but I guess don't have the same "open web" flavor

[08:52:00.0000] <Dashiva>
Philip`: But a static image of such a graph would also be pretty confusing. With canvas you could let the user navigate around it :)

[08:53:00.0000] <hsivonen>
I guess adding XPath would require putting Selectors there as a box as well

[08:54:00.0000] <hsivonen>
and yes, NPAPI would belong in the diagram. can't figure out where, though

[08:54:01.0000] <Philip`>
Dashiva: It wouldn't necessarily be pretty confusing - it'd probably be mostly a treeish structure like hsivonen's current diagram, but with a few extra edges in weird directions

[08:54:02.0000] <hsivonen>
Canvas 2D could be called out on ints own right, but it would have to be between JS and Visual Rendering

[08:54:03.0000] <othermaciej>
hsivonen: maybe you should use RDF and make the Browser Technology Graph

[08:55:00.0000] <othermaciej>
clearly a stack is not an adequate data structure for this use case

[08:55:01.0000] <hsivonen>
I guess a graph would show the intertwingularity better

[08:55:02.0000] <Philip`>
/me would probably write .dot files rather than RDF

[08:55:03.0000] <hsivonen>
Word template?

[08:56:00.0000] <jgraham>
Philip`: You could write HTML5 microdata, autoconvert it to RDF then to .dot and finally to SVG, which you could in turn embed inside the original HTML document

[08:56:01.0000] <othermaciej>
I don't know what HID is

[08:56:02.0000] <jgraham>
hsivonen: graphviz

[08:57:00.0000] <hsivonen>
othermaciej: Human Interface Device

[08:57:01.0000] <hsivonen>
pointing device / text input was too long

[08:58:00.0000] <annevk42>
hsivonen, I hope someone will appeal to some higher court? (re twitter)

[08:59:00.0000] <othermaciej>
now I feel less embarassed that I don't know

[08:59:01.0000] <othermaciej>
the granularity of specifications is kind of arbitrary

[09:00:00.0000] <othermaciej>
which is unfortunate for a diagram like this

[09:00:01.0000] <othermaciej>
editing is a pretty big subsystem in WebKit but it has no spec to call its own

[09:01:00.0000] <annevk42>
mostly HTML5 I reckon

[09:02:00.0000] <othermaciej>
yeah although HTML5 is not nearly detailed enough about editing behavior to allow for interoperability I think

[09:02:01.0000] <othermaciej>
man the w3c version of the web technology stack has an impressive collection of acronyms I've never heard of

[09:02:02.0000] <hsivonen>
annevk42: I don't know how the case will proceed in the courts. there's a parallel request to an ombudsman who sits aside from the courts and the parliament

[09:10:00.0000] <annevk42>
a Parliamentary Ombudsman?

[09:10:01.0000] <hsivonen>
annevk42: yes

[09:10:02.0000] <annevk42>
apparently the concept differs a lot per country

[09:11:00.0000] <annevk42>
seems it would be annoying if that person agreed with the judge

[09:11:01.0000] <hsivonen>
the most basic problem is that Finland lacks a constitutional court and the parliament oversees the constitutionality of its own acts which is a joke

[09:15:00.0000] <annevk42>
so in theory the parliament has unlimited power?

[09:16:00.0000] <othermaciej>
the UK parliament has theoretically unlimited power too, so it's not unprecedented

[09:17:00.0000] <hsivonen>
annevk42: I think a couple of consecutive parliaments together have unlimited power (to the extent the EU doesn't sue)

[09:17:01.0000] <othermaciej>
courts will not even uphold a law that tries to restrict future acts of parliament

[09:17:02.0000] <annevk42>
othermaciej, according to Wikipedia that will soon stop as the UK will get a supreme court

[09:17:03.0000] <annevk42>
http://en.wikipedia.org/wiki/Supreme_Court_of_the_United_Kingdom

[09:20:00.0000] <othermaciej>
yeah it will have a Supreme Court but I don't think that will establish judicial review of acts of parliament, since there is no written constitution, just conventions

[09:22:00.0000] <annevk42>
/me finds http://en.wikipedia.org/wiki/Democracy_Index#2008_ranking

[09:23:00.0000] <hsivonen>
can an old version of CHUD block Mac OS X update?

[09:24:00.0000] <hsivonen>
I wonder what I should do when an OS X update fails to install

[09:27:00.0000] <hsivonen>
http://theory.uchicago.edu/~harvey/harveyomega/

[09:49:00.0000] <Philip`>
hsivonen: Doesn't work very well in Opera :-(

[09:49:01.0000] <Philip`>
(The page reloads after pressing enter)

[09:49:02.0000] <Philip`>
/me blames all the Opera people here

[11:09:00.0000] <gsnedders>
http://digg.com/tech_news/Tim_Berners_Lee_on_the_next_Web?t=25847831#c25848908 — hah! Too true!

[11:57:00.0000] <gsnedders>
jgraham: What was that graphing software you were saying I should use

[11:58:00.0000] <Philip`>
Excel?

[11:59:00.0000] <gsnedders>
Philip`: No

[12:02:00.0000] <gsnedders>
Veusz

[12:03:00.0000] <gsnedders>
http://krijnhoetmer.nl/irc-logs/whatwg/20090220#l-561

[12:07:00.0000] <gsnedders>
I think something is wrong with the OS X binary.

[12:12:00.0000] <Philip`>
/me wonders if Larry is underestimating the extent to which trivial details end up causing sites to break in some browsers

[12:19:00.0000] <Philip`>
Hmm, I've even written code myself that tests if img.width != 0 and if so then it uses both img.width and img.height

[12:20:00.0000] <Philip`>
(though it does check img.complete first)

[12:20:01.0000] <Philip`>
(and just uses width==0 to test whether it completed successfully or unsuccessfully)

[12:39:00.0000] <annevk4>
hmm @reply in twitter means the reply will not appear for the people following you?

[12:39:01.0000] <annevk4>
wtf

[12:39:02.0000] <Dashiva>
Well, sometimes

[12:40:00.0000] <Dashiva>
I haven't been able to figure out when it doesn't appear

[12:42:00.0000] <annevk4>
maybe only when you are not following one of them

[12:42:01.0000] <annevk4>
that seems to be it

[12:42:02.0000] <annevk4>
I don't like it

[12:43:00.0000] <Dashiva>
Ah, that seems to fit my observations, yes

[12:43:01.0000] <gavin_>
people following you don't get @replies you send to people that they aren't themselves following

[12:44:00.0000] <annevk4>
sucks

[12:44:01.0000] <gavin_>
that's what the #fixreplies fiasco was about

[12:44:02.0000] <gavin_>
used to be a user pref to control whether you saw them

[12:46:00.0000] <annevk4>
I thought the whole idea was that if I followed someone I got to see what they do and what keeps them busy

[12:47:00.0000] <annevk4>
oh well

[12:50:00.0000] <gavin_>
I guess the reasoning is that making it a user pref is costly (need to check all follower prefs for each tweet)

[12:50:01.0000] <gavin_>
and making it the default is sometimes confusing/undesired

[12:50:02.0000] <gavin_>
so it was just removed

[12:51:00.0000] <Madness>
http://wiki.whatwg.org/wiki/RelExtensions hm, no icon or shortcut icon defined...

[12:52:00.0000] <annevk4>
that's in HTML5 itself

[12:53:00.0000] <Madness>
Meaning it's not an extension?

[12:53:01.0000] <annevk4>
I suppose

[12:53:02.0000] <annevk4>
maybe we should still list them there for completeness

[12:54:00.0000] <Madness>
Yes please, so silly heads like me won't get puzzled. :)

[12:57:00.0000] <jgraham>
gsnedders: wfm

[13:08:00.0000] <gsnedders>
jgraham: http://stuff.gsnedders.com/veusz.tiff

[13:09:00.0000] <gsnedders>
Version 1.2 works though

[13:12:00.0000] <annevk4>
Hixie, stylesheet should come after sidebar

[13:23:00.0000] <gsnedders>
OK, does anyone have suggestions for non-sucky non-buggy graphing software?

[13:36:00.0000] <hsivonen>
I'm baffled by being accused of divisive rhetoric when the accusation talks about WHATWG doctrine and implies I assumed sinister motives when I didn't. Hmm. http://schepers.cc/?p=117

[13:36:01.0000] <annevk42>
Arguably the definition of "available" is somewhat flaky

[13:37:00.0000] <annevk42>
E.g. how can one know the image is valid if it is only partially downloaded. And if valid is a requirement for being available, how it can be available before being completely downloaded.

[13:37:01.0000] <annevk42>
(Apart from the fact that supported image type and valid image are indeed not defined.)

[13:41:00.0000] <Philip`>
hsivonen: Maybe you're just much more subtle in your use of divisive rhetoric, so that it can't be easily seen from individual quotes

[13:44:00.0000] <hsivonen>
Philip`: oh? my original observation was that the W3C's PR materials weren't being inclusive of HTML and XHR, which makes it seem like the W3C hasn't quite accepted what it invited

[13:45:00.0000] <annevk42>
Apparently they just haven't updated their marketing materials

[13:45:01.0000] <hsivonen>
clearly

[13:45:02.0000] <Dashiva>
I'm baffled that a single calf is being called a cow path

[13:45:03.0000] <annevk42>
It wasn't about CR or Rec; as Ivan points he made the diagram in 2004-5 and back then HTML5 and XHR didn't happen

[13:46:00.0000] <hsivonen>
annevk42: it has been updated since then, and the surrounding material has been updated since then

[13:46:01.0000] <hsivonen>
I don't assume that things were kept out deliberately, but it's PR material that isn't up-to-date

[13:46:02.0000] <annevk42>
But the diagram (other than removing one part) remained the same right?

[13:47:00.0000] <hsivonen>
annevk42: no

[13:47:01.0000] <annevk42>
oh ok

[13:47:02.0000] <hsivonen>
annevk42: XSL, RDFa and GRDDL have been added

[13:47:03.0000] <annevk42>
ok

[13:48:00.0000] <annevk42>
I tend to agree that calling you divisive for pointing out something you consider flawed is not very nice

[13:49:00.0000] <Philip`>
hsivonen: It may also be that you're not using divisive rhetoric at all, so I didn't mean to imply you were :-)

[13:50:00.0000] <annevk42>
by schepers' own logic WICD shouldn't be there btw

[13:50:01.0000] <annevk42>
and if CR does qualify I'm missing XBL2

[13:54:00.0000] <hsivonen>
http://search.twitter.com/search?q=http%3A%2F%2Fschepers.cc%2F%3Fp%3D117

[13:57:00.0000] <annevk4>
fun, they applaud the post and we critique it; guess we're not making much process on that front then

[16:34:00.0000] <Hixie>
hm, larry is right. the wg isn't chartered to produce the current html5 draft. i guess we'd better start over.

[16:36:00.0000] <annevk4>
you don't think it's still evolved from HTML4 if you look at the features?

[16:36:01.0000] <Hixie>
nope, not evolved at all, it's a whole new language. woe is us we made a terrible mistake.

[16:37:00.0000] <Philip`>
According to http://www.w3.org/html/wg/ we have a week before we're scheduled to reach CR - that seems plenty of time to download a copy of the HTML4 spec and then paste some of the new features into it

[16:37:01.0000] <annevk4>
Hixie, mu

[16:37:02.0000] <annevk4>
Philip`, could try to give HTML4 a proper CR phase and see what happens :)

[16:38:00.0000] <Hixie>
well i sure hope someone volunteers to do that, i'm too tired from writing the last draft, which turned out to be totally the wrong thing and which we are now forced to throw away.

[16:38:01.0000] <Hixie>
maybe larry can write it

[16:39:00.0000] <othermaciej>
I didn't read Larry's walls of text yet

[16:39:01.0000] <othermaciej>
should I?

[16:40:00.0000] <annevk4>
this is about a much shorter email it seems

[16:40:01.0000] <annevk4>
subject line ending in "(from scratch)"

[16:41:00.0000] <annevk4>
I think I'll go read something and go to bed

[16:41:01.0000] <annevk4>
not really interested in another roundtable of discussing this; not too sure why I put effort into it today


2009-05-26
[17:31:00.0000] <Hixie>
ok, today's e-mail has been dealt with

[17:35:00.0000] <othermaciej>
Hixie: I do think that sometimes conformance requirements stated in algorithmic form can be harder to work with, if you are not implementing the algorithm literally

[17:35:01.0000] <Hixie>
i agree

[17:35:02.0000] <othermaciej>
but to me that takes a back seat to precision

[17:36:00.0000] <othermaciej>
for example, for img.width and img.height

[17:36:01.0000] <othermaciej>
there's probably the following important criteria:

[17:36:02.0000] <Hixie>
if there are any algorithms that someone can find a non-algorithmic way to describe the same requirements, let me know

[17:36:03.0000] <othermaciej>
- the height and width should be either 0 or the correct value

[17:36:04.0000] <othermaciej>
- the height and width should only change from 0 to the final value barring the final load, no intermediate different values

[17:37:00.0000] <othermaciej>
- height and width should be available no later than the "load" even

[17:37:01.0000] <othermaciej>
*event

[17:37:02.0000] <othermaciej>
- height and width should be available at the same time

[17:37:03.0000] <othermaciej>
I'm not sure all of these can be easily inferred from the algorithm, simple though it is

[17:38:00.0000] <othermaciej>
for parsing it is likely impossible to express equivalent requirements in declarative rather than as-if imperative form

[17:38:01.0000] <othermaciej>
I haven't really studied the spec with an eye to this

[17:38:02.0000] <othermaciej>
I would expect improvements along these lines, and pointing out where the spec overconstrains implementations, will happen in CR

[17:39:00.0000] <othermaciej>
Larry seems to underestimate how different the internals of different browser engines are, at least when it comes to details

[17:42:00.0000] <Hixie>
i think your height/width definitions miss some things

[17:43:00.0000] <Hixie>
e.g. that the values are the rendered dimensions not the actual dimensions

[17:43:01.0000] <Hixie>
when the image is rendered

[17:43:02.0000] <othermaciej>
it wasn't intended to be complete

[17:43:03.0000] <Hixie>
ok

[17:43:04.0000] <Hixie>
the problem is that making it complete ends up taking three times more room than what we have now

[17:43:05.0000] <othermaciej>
I was just thinking about some things that I am pretty sure are needed for interoperability

[17:43:06.0000] <Hixie>
and it's not clear that it's better

[17:43:07.0000] <othermaciej>
it wasn't suggested to be an alternative

[17:43:08.0000] <Hixie>
yeah

[17:44:00.0000] <Hixie>
larry seems to want an alternative though

[17:44:01.0000] <othermaciej>
it was a thought experiment, to see if the current conformance requirements imply the things that are needed for interop

[17:44:02.0000] <Hixie>
ah i see

[17:44:03.0000] <othermaciej>
my impression is that the most important for non-rendered images is that you have width and height by the time load occurs

[17:45:00.0000] <othermaciej>
I think an image has to be "available" by the present definition no later than the time the "load" event is fired

[17:45:01.0000] <othermaciej>
I think width and height becoming available at the same time likely matters for interop, even though Larry thought that was an overconstraint

[17:46:00.0000] <Hixie>
yes, 'load' fires by definition after the image is 'available'

[17:46:01.0000] <Hixie>
per spec

[17:47:00.0000] <othermaciej>
he may have a point that the spec overconstrains things needlessly at times, though his specific example does not seem to me a very good one

[17:47:01.0000] <othermaciej>
plus I do think implementors often point it out when there are overconstraints

[17:48:00.0000] <othermaciej>
(consider e.g. concerns arising from multiprocess rendering and various serial behavior rules)

[17:48:01.0000] <othermaciej>
but still, I'd prefer overconstraint to the sever underspecification of HTML4

[17:49:00.0000] <othermaciej>
the cost of an unnecessary conformance requirement is less than the cost of a missing one

[17:49:01.0000] <othermaciej>
if one assumes the spec will continue to be maintained

[17:50:00.0000] <Hixie>
i'm sure the spec is both over- and under-constrained in parts

[17:50:01.0000] <Hixie>
since i am sadly not perfect :-)

[17:50:02.0000] <othermaciej>
anyway, even if I disagree with Larry on many things, I think his substantive commentary using a concrete example is much more helpful than title trolling

[17:50:03.0000] <Hixie>
it's certainly a step up

[17:50:04.0000] <annevk4>
Hixie, I mentioned earlier over IRC that "valid image" and still being downloaded are somewhat at odds with each other

[17:50:05.0000] <Hixie>
i can actually address his concerns when he mentions them

[17:50:06.0000] <othermaciej>
as was Sam's identification of a specific issue bothering him

[17:51:00.0000] <Hixie>
annevk4: i removed "valid image" from the spec a few minutes ago

[17:51:01.0000] <othermaciej>
since it turns out that issue is already slated to be fixe

[17:51:02.0000] <annevk4>
Hixie, I suppose that helps :)

[17:51:03.0000] <Hixie>
the content-sniffing thing?

[17:51:04.0000] <Hixie>
adam said he's doing some work on it next week

[17:51:05.0000] <Hixie>
iirc

[17:52:00.0000] <Hixie>
and then we can submit it for rfc or something

[17:52:01.0000] <Hixie>
i don't fully understand how the ietf works

[17:52:02.0000] <Hixie>
did you know that once you submit an rfc, you can no longer fix it?

[17:52:03.0000] <Hixie>
it's amazing there are only a few thousand rfcs so far, given that

[17:52:04.0000] <Hixie>
i guess errors just go unfixed for years

[17:53:00.0000] <othermaciej>
so there's no provision for errata, you have to make a new RFC?

[17:54:00.0000] <othermaciej>
that seems kind of wacky, it would be like locking down a W3C spec at FPWD

[17:55:00.0000] <othermaciej>
and yeah it is amazing how rarely there are updates to fix errors

[17:55:01.0000] <annevk4>
I saw you can report errata against an RFC in some fashion (not sure if it's official)

[17:56:00.0000] <Hixie>
i believe they consider RFC publication more like CR

[17:59:00.0000] <annevk4>
e.g. with URIs it took several RFCs to become a STD

[17:59:01.0000] <othermaciej>
yeah but it seems like even the "Proposed Standard" state is one where one would expect further changes are needed

[17:59:02.0000] <othermaciej>
there's nothing wrong per se with using a new RFC number for that

[17:59:03.0000] <othermaciej>
just surprising it hasn't happened more

[18:01:00.0000] <othermaciej>
interesting, RFC2068 was a Proposed Standard but RFC2616 which obsoleted it isn't yet (a decade after coming out)

[18:10:00.0000] <Hixie>
annevk4: any idea what happens when you set xhr.responseXML.body.onload ?

[18:10:01.0000] <Hixie>
(re http://www.w3.org/Bugs/Public/show_bug.cgi?id=6515 )

[18:13:00.0000] <annevk4>
it seems that by the time you can access that it has already fired

[18:13:01.0000] <annevk4>
but no, I haven't tested

[18:15:00.0000] <Hixie>
fired? i would have presumed that no events ever fire on those documents

[18:15:01.0000] <Hixie>
by the ua, anyway

[18:15:02.0000] <Hixie>
i shall poke around

[18:18:00.0000] <annevk4>
you once made a testcase with form submission in such a document and expected submit to fire

[18:18:01.0000] <annevk4>
and it did

[18:18:02.0000] <annevk4>
iirc

[18:19:00.0000] <Hixie>
ah

[18:20:00.0000] <annevk4>
http://twitter.com/heycam/status/1918008204 hehe

[18:20:01.0000] <Hixie>
i don't understand why this is an issue for html but not svg

[18:21:00.0000] <Hixie>
did the svgwg get feedback along the lines of "you should have a subtitle that says you're a browser spec"?

[18:21:01.0000] <heycam>
SVG Tiny 1.2: A specification that defines an XML-based vector graphics language, associated APIs, and other APIs for non-graphics related things that we needed but weren't specified elsewhere

[18:21:02.0000] <heycam>
nope

[18:21:03.0000] <annevk4>
Hixie, it did get feedback that it did too much

[18:22:00.0000] <heycam>
but not that it needed to call that out in the title

[18:22:01.0000] <heycam>
(and in the end we did rip a lot of stuff out)

[18:22:02.0000] <annevk4>
right

[18:23:00.0000] <annevk4>
but don't tell people they're using the title as an argument for something else because they don't like that

[18:27:00.0000] <annevk4>
hmm yeah, SVG 1.2 didn't call out any of those things in title/subtitle/abstract

[18:28:00.0000] <annevk4>
but I don't really have a problem with calling out page load processing in the subtitle/abstract

[18:38:00.0000] <Hixie>
i was considering coming up with a boatload of alternative subtitles

[18:38:01.0000] <Hixie>
and just cycling through them using a script

[18:38:02.0000] <Hixie>
so every time you load the spec it gives you a new one

[18:39:00.0000] <othermaciej>
that would be lols

[18:45:00.0000] <annevk4>
not exactly a new concept: http://krijnhoetmer.nl/irc-logs/ :)

[18:55:00.0000] <annevk4>
/me takes another stab at sleeping

[21:55:00.0000] <Hixie>
oooh tough one

[21:55:01.0000] <Hixie>
marquee has some custom event handler attributes

[21:55:02.0000] <Hixie>
do i add them to the general set?

[21:55:03.0000] <Hixie>
or do i make marquee an exception?

[21:56:00.0000] <Hixie>
you know, i think the thing i find annoying about a lot of the people who keep complaining about html5 in various ways is that they never give advice for cases like this

[21:56:01.0000] <Hixie>
so it feels like they're complaining and complaining but not ever being constructive in areas that aren't their pet peeve

[22:01:00.0000] <sayrer>
http://www.rfc-editor.org/errata.php

[22:01:01.0000] <sayrer>
http://www.rfc-editor.org/errata.php

[22:01:02.0000] <sayrer>
fwiw

[22:01:03.0000] <sayrer>
most errata lists I've encountered seem to be informally 

[22:02:00.0000] <sayrer>
maintained

[22:02:01.0000] <sayrer>
like, by people who actually need to consult them

[22:08:00.0000] <Hixie>
doesn't seem to have any serious errata for any of the rfcs i looked at

[22:34:00.0000] <Hixie>
i feel like leif is trying to make a point but i have no idea what that point is

[22:35:00.0000] <Hixie>
i wish he'd just come out and say it instead of quoting me and implying his point

[22:49:00.0000] <sayrer>
Hixie, there is a provision for errata, that's all

[22:49:01.0000] <sayrer>
and, you can sort of just make them

[22:49:02.0000] <sayrer>
like the HTTP errata that were maintained for years

[22:49:03.0000] <Hixie>
indeed

[22:49:04.0000] <Hixie>
errata are a poor way of maintaining specs, though, as I learnt with CSS

[22:50:00.0000] <sayrer>
sure, when the http-bis thing started

[22:50:01.0000] <sayrer>
figuring out what to do with the errata was a problem

[22:51:00.0000] <sayrer>
dang, larry found a section of the spec I would like to delete

[22:51:01.0000] <sayrer>
I guess the little google summaries might count

[22:58:00.0000] <othermaciej>
the Outline section?

[22:58:01.0000] <othermaciej>
I am not really sure what that is for

[22:58:02.0000] <othermaciej>
what kind of client produces or displays an outline?

[22:59:00.0000] <Hixie>
accessibility tools, amongst others

[22:59:01.0000] <Hixie>
but that section is present not so much for tools, so much as to make sure that it is well-defined what the heading is for each part of the document

[22:59:02.0000] <othermaciej>
VoiceOver doesn't make an outline and if it did, I'm not sure it would necessarily want to use that algorithm to compute it

[23:00:00.0000] <Hixie>
one of the questions that kept coming up with html4 was "what exactly is my document structure"

[23:00:01.0000] <othermaciej>
perhaps "what the heading is for each part of the document" has more immediate relevance

[23:04:00.0000] <sayrer>
well-defined sounds pretty circular

[23:05:00.0000] <sayrer>
Hixie, interesting use of passive voice. who was asking the question?

[23:13:00.0000] <Hixie>
many people

[23:13:01.0000] <Hixie>
i didn't take names

[23:14:00.0000] <othermaciej>
really in practical terms this section is only defining semantics, not conformance requirements for typical UAs

[23:14:01.0000] <othermaciej>
but doing it in algorithm terms

[23:14:02.0000] <Hixie>
yes

[23:14:03.0000] <Hixie>
i'm happy to phrase it in non-algorithmic terms if anyone can come up with a sane way of describing it that way

[23:14:04.0000] <Hixie>
i couldn't work out how to do it

[23:17:00.0000] <othermaciej>
I don't find the definition of this particular piece of semantics very important

[23:17:01.0000] <othermaciej>
I can see how it might be hard to define in purely declarative terms without potentially missing cases

[23:23:00.0000] <hsivonen>
pasting some URLs for to have a record in the log:

[23:24:00.0000] <hsivonen>
http://twitter.com/jdowdell/status/1608188445

[23:24:01.0000] <hsivonen>
http://twitter.com/jdowdell/status/1352144527

[23:24:02.0000] <hsivonen>
http://twitter.com/jdowdell/status/1344855975

[23:24:03.0000] <hsivonen>
http://twitter.com/jdowdell/status/1464529893

[23:24:04.0000] <hsivonen>
(end of flood)

[23:26:00.0000] <othermaciej>
/me is not sure he gets the theme there

[23:26:01.0000] <hsivonen>
seems like jd considers the variability of HTML runtimes a problem, so I guess HTML5 should err on the side off well-defined behavior

[23:38:00.0000] <othermaciej>
oh you're quoting the Adobe trollblogger?

[23:44:00.0000] <hsivonen>
othermaciej: I'm quoting the Adobe Flash blogger who seems to blog a lot about HTML these days

[23:49:00.0000] <MikeSmith>
johnny one-note

[23:49:01.0000] <othermaciej>
hsivonen: after reading some of his recent posts, I stand by my prior assertion

[23:49:02.0000] <othermaciej>
MikeSmith: do you have administrative access to the w3c CVS server?

[23:51:00.0000] <MikeSmith>
othermaciej: yeah

[23:51:01.0000] <MikeSmith>
sorta

[23:52:00.0000] <MikeSmith>
I have access to the actual filesystem where the cvs ,v files are

[23:56:00.0000] <othermaciej>
yeah, he's basically Adobe's Asa Dotzler

[00:01:00.0000] <MikeSmith>
another possible competitor in this race -

[00:01:01.0000] <MikeSmith>
http://twitter.com/mattmay/status/1884053829

[00:02:00.0000] <MikeSmith>
but I think he'll need to try harder than that if he really wants to win it

[00:02:01.0000] <othermaciej>
it makes Adobe look sad and desparate to try to fight against HTML having more features

[00:04:00.0000] <Hixie>
i don't think matt is fighting html having more features in that twitter

[00:05:00.0000] <Hixie>
since leaving the w3c would make progress on html5 far quicker and easier

[00:05:01.0000] <Hixie>
if anything, he's fighting _for_ html having more features

[00:06:00.0000] <othermaciej>
I wonder why John likes the catchphrase "browser brands" so much

[00:06:01.0000] <othermaciej>
why does he say that instead of "browsers"? is that an Adobe thing?

[00:09:00.0000] <hsivonen>
othermaciej: the impression I get is that he wants to portray browsers as different chrome designs

[00:09:01.0000] <othermaciej>
but wouldn't that portray the engines as essentially interchangeable?

[00:09:02.0000] <othermaciej>
which is contrary to his point?

[00:09:03.0000] <hsivonen>
othermaciej: I suppose

[00:10:00.0000] <hsivonen>
othermaciej: although I think the point is that you pick your favorite toolbar and run Flash in the space below it

[00:10:01.0000] <othermaciej>
ah

[00:14:00.0000] <othermaciej_>
"browser brand" is not a very common phrase outside his blog

[00:14:01.0000] <othermaciej_>
but yeah I can see how he might want to take the "browsers are just Flash loaders" position

[00:31:00.0000] <MikeSmith>
hsivonen: when you have a few minutes, I could use a bit of help figuring out how to write a test for valid surrogate pairs in accesskey values

[00:36:00.0000] <zcorpan_>
Hixie: yt?

[00:40:00.0000] <Hixie>
zcorpan_: am now

[00:54:00.0000] <Dashiva>
http://lists.w3.org/Archives/Public/public-html/2009May/0375.html seems to be a great example of how to abuse the design principles

[01:01:00.0000] <Hixie>
it's more an example of not understanding the principles

[01:01:01.0000] <Hixie>
it might be worth just removing the names of the principles and just numbering them

[01:03:00.0000] <othermaciej>
tempting! but would make it hard to cite them

[01:04:00.0000] <Hixie>
that's the point :-)

[01:04:01.0000] <Hixie>
it would also mean that they couldn't be mischaracterised based on their names

[01:05:00.0000] <Hixie>
but seriously, the biggest problem with the design principles is that people have tried to apply them like blind rules

[01:05:01.0000] <Hixie>
instead of taking them as what they are -- principles of design

[01:07:00.0000] <Dashiva>
Could give them code names. Easier to remember without having names to infer meaning from

[01:08:00.0000] <othermaciej>
people also object to the fact that I said they are just guidelines and rules of thumb not absolute laws

[01:08:01.0000] <othermaciej>
the one that bugs me the most abuse-wise is "Pave the Cowpaths"

[01:08:02.0000] <othermaciej>
where opponents of it keep claiming it is being misused

[01:08:03.0000] <othermaciej>
in a way that, as far as I can tell, no one in favor of the principle actually does

[01:11:00.0000] <Hixie>
people always do that

[01:11:01.0000] <Hixie>
it's the same as the way people quote my "handling-people" file and say "look! he's doing this and that!" when they are losing an argument

[01:11:02.0000] <hsivonen>
looks like removing the head of the pin generates more heat than deciding how many angels can dance on it

[01:11:03.0000] <hsivonen>
(profile)

[01:11:04.0000] <othermaciej>
how dare you use the 2 + 2 = 4 principle to claim that 1 + 1 is an odd number!

[01:11:05.0000] <othermaciej>
that's an abuse of the principle!

[01:11:06.0000] <othermaciej>
"... wait, what?"

[01:12:00.0000] <hsivonen>
Hixie: does one lose an argument if one catches someone else than you practicing "handling people"?

[01:13:00.0000] <Hixie>
there's never any benefit to pointing it out, since people would always just deny it

[01:13:01.0000] <hsivonen>
/me looked up the sabotage manual from Gutenberg the other day

[01:13:02.0000] <Hixie>
that file is just useful as a way to know what to look for

[01:13:03.0000] <Hixie>
since once you spot someone doing something from that list, you can then apply appropriate countermeasures

[01:13:04.0000] <Hixie>
instead of falling into their trap

[01:15:00.0000] <zcorpan_>
Hixie: we're implementing spellcheck, so if you're going to simplify the algorithm, we'd appreciate if you did it soonish :)

[01:16:00.0000] <Hixie>
noted

[01:18:00.0000] <othermaciej>
crap, I forgot that IE means I can't use CSS generated content

[01:19:00.0000] <zcorpan_>
othermaciej: ie8 supports it

[01:19:01.0000] <othermaciej>
would it be ok for W3C documents to be IE8-and-up?

[01:19:02.0000] <zcorpan_>
sure

[01:19:03.0000] <zcorpan_>
i'm sure the html5 spec doesn't work correctly in ie :)

[01:21:00.0000] <zcorpan_>
othermaciej: although you should expect people to complain because they use ie7 and a screen reader and now you're against blind people

[01:21:01.0000] <othermaciej>
well

[01:22:00.0000] <othermaciej>
I could preface every example with the words "Example:" by hand

[01:22:01.0000] <othermaciej>
but that seems annoying

[01:22:02.0000] <othermaciej>
was trying to decide whether to do that instead of a :before rule

[01:23:00.0000] <zcorpan_>
othermaciej: the html5 spec uses generated content to insert "Note:" and "Warning:"

[01:24:00.0000] <Hixie>
the spec gen script will eventually do this for us

[01:24:01.0000] <Hixie>
once gsnedders is out of school or whatever

[01:25:00.0000] <zcorpan_>
when you're in school you have plenty of time

[01:25:01.0000] <othermaciej>
would anyone like to request quick fixes to the Design Principles or point out errors, while I am editing?

[01:25:02.0000] <Hixie>
baby steps!

[01:25:03.0000] <Hixie>
baby steps!

[01:26:00.0000] <zcorpan_>
othermaciej: i sent an email before

[01:26:01.0000] <othermaciej>
Hixie: oh that's right

[01:27:00.0000] <zcorpan_>
othermaciej: s/irrelevant/hidden/

[01:27:01.0000] <othermaciej>
Hixie: I was thinking of changing Evolution not Revolution to "Incremental Improvement" so it can include the spirit of "baby steps" without causing a fight over the name "baby steps"

[01:27:02.0000] <othermaciej>
zcorpan_: thanks!

[01:27:03.0000] <othermaciej>
Hixie: I think that's too substantive a change to sneak in with minor edits though

[01:27:04.0000] <Hixie>
i don't care what you call it, but the key point is not adding so many new features that implementors get lost behind and never catch up

[01:28:00.0000] <othermaciej>
Hixie: I think you might already be ahead of the curve on that...

[01:28:01.0000] <Hixie>
yeah

[01:28:02.0000] <othermaciej>
although there aren't many things I would call for removing at the moment

[01:28:03.0000] <Hixie>
hence how important it is!

[01:28:04.0000] <Hixie>
it was on my feedback on that survey n years ago

[01:28:05.0000] <Hixie>
on the list of things i'd like to see before we publish

[01:28:06.0000] <Hixie>
if that helps :-)

[01:30:00.0000] <othermaciej>
I guess I can include it as part of addressing the old feedback

[01:31:00.0000] <zcorpan_>
/me ponders about an example to Solve Real Problems

[01:34:00.0000] <Hixie>
Ollie: hi :-)

[01:35:00.0000] <othermaciej>
would anyone care to help me proofread http://dev.w3.org/cvsweb/~checkout~/html5/html-design-principles/Overview.html?rev=1.24 before I note it on the list?

[01:35:01.0000] <Ollie>
Hixie: Hi there :)

[01:35:02.0000] <Hixie>
othermaciej: "The working group is working on a new version of HTML not yet published under TR" is no longer accurate

[01:36:00.0000] <Dashiva>
gsnedders: For extra credit, show this to the teacher: http://2.media.tumblr.com/9pDtThz04nxemo71EGVHHWQgo1_500.gif

[01:36:01.0000] <hsivonen>
othermaciej: I'll read

[01:37:00.0000] <hsivonen>
othermaciej: it's no longer the *First* public WD

[01:37:01.0000] <hsivonen>
othermaciej: what Hixie said

[01:38:00.0000] <othermaciej>
fixed what Hixie said, changed front matter to say "an Editor's Draft"

[01:38:01.0000] <hsivonen>
othermaciej: it's probably unnecessary to put the vote tally up there

[01:39:00.0000] <othermaciej>
where is there a vote tally?

[01:39:01.0000] <Hixie>
in the SotD

[01:39:02.0000] <hsivonen>
othermaciej: yes. the next paragraph after that is odd, too

[01:39:03.0000] <othermaciej>
should I just remove that part?

[01:40:00.0000] <othermaciej>
I think I will

[01:40:01.0000] <othermaciej>
removed both those paragraphs

[01:40:02.0000] <hsivonen>
thanks

[01:42:00.0000] <Hixie>
othermaciej: i don't think "Evolution Not Revolution" really conveys the "do things slowly" approach i think is important

[01:42:01.0000] <othermaciej>
Hixie: I don't think it does either!

[01:42:02.0000] <othermaciej>
I would like to edit it to do so in addition to changing the title

[01:42:03.0000] <Hixie>
k

[01:42:04.0000] <othermaciej>
changing the name rather

[01:42:05.0000] <othermaciej>
I think one principle can cover both concepts

[01:42:06.0000] <othermaciej>
don't boil the ocean, and don't try to do too much at one go

[01:42:07.0000] <othermaciej>
if I may speak in shorthand

[01:49:00.0000] <hsivonen>
othermaciej: looks good. However, I notice that you didn't bother to annotate each priciple by saying if it applies to conforming language or supported language

[01:49:01.0000] <othermaciej>
I guess that's another thing I should fix as part of the next pass

[02:41:00.0000] <MikeSmith>
use of Unicode surrogate pairs is restricted to content encoded in UTF-16, right?

[02:42:00.0000] <Hixie>
in theory

[02:45:00.0000] <hsivonen>
MikeSmith: scripts could introduce lone surrogates to pages that had any encoding

[08:10:00.0000] <annevk42>
so I follow a link to sitepoint and say yes to answering some survey

[08:10:01.0000] <annevk42>
first question "Based on today's visit, how would you rate your site experience overall?" ...

[08:11:00.0000] <annevk42>
second "Which of the following best describes the primary purpose of your visit?" doesn't list "following a link" as one of the options...

[08:11:01.0000] <annevk42>
third "Were you able to complete the purpose of your visit today?" only gives yes/no

[14:15:00.0000] <gsnedders>
Oh well, I guess we should change the interface then :D

[14:17:00.0000] <jgraham>
Dashiva: good point

[14:18:00.0000] <hsivonen>
cgriego: not deployed. I'll deploy it in 12 hours or so

[14:19:00.0000] <gsnedders>
jgraham: LOL

[14:19:01.0000] <gsnedders>
jgraham: I do like how you're using my implementation of the outlining algorithm as an example and not your own

[14:45:00.0000] <ezyang>
I think the doctype syntax is also ambiguous 

[14:46:00.0000] <ezyang>
Since if I only specify systemId, it's not distinguishable from publicId 

[14:49:00.0000] <ezyang>
Anyway, pushed the DOM -> test format converter. 

[14:49:01.0000] <ezyang>
Probably the most dirty piece of code I've written in a long while. 

[14:59:00.0000] <jgraham>
gsnedders: Yours has the advantage of being relatively well tested and well commented

[15:00:00.0000] <gsnedders>
jgraham: It's almost entirely untested, it's just people rely upon it!

[15:00:01.0000] <gsnedders>
(Those idiots)

[15:02:00.0000] <ezyang>
Developers want shiny things NAO. 

[15:03:00.0000] <jgraham>
gsnedders: It is known to work in some situations

[15:03:01.0000] <jgraham>
In other news we should write some tests for this stuff

[15:04:00.0000] <gsnedders>
jgraham: "news"? :D

[15:05:00.0000] <jgraham>
News of the form "MPs claim everything they can on expenses" i.e. everyone already knew, it has been explicit for months and yet it is still "news"

[15:10:00.0000] <ezyang>
Excellent! 369 failures, but at least it doesn't fatally error out. 

[15:11:00.0000] <gsnedders>
ezyang: BTW, do you have exams? When?

[15:11:01.0000] <ezyang>
It looks like there are some stray newlines 

[15:11:02.0000] <ezyang>
gsnedders: I'm done with exams ;-) 

[15:11:03.0000] <gsnedders>
ezyang: re your latest tweet, @id :)

[15:11:04.0000] <gsnedders>
ezyang: Ah, I still have one to go.

[15:11:05.0000] <ezyang>
Ok. So they totally got rid of it? 

[15:11:06.0000] <ezyang>
Heh 

[15:11:07.0000] <gsnedders>
yeah

[15:11:08.0000] <ezyang>
Good luck mate 

[15:12:00.0000] <gsnedders>
Like, computing exam. The one I haven't done any work for all year. The one I got 77/80 in the coursework. The one I barely need to pass to get a A.

[15:12:01.0000] <ezyang>
Oh hoho 

[15:12:02.0000] <gsnedders>
I had physics exam… I was thirty minutes late, because I thought it was on Thursday, until I got a phone call from school!

[15:13:00.0000] <gsnedders>
s/…/today…/

[15:13:01.0000] <ezyang>
You're scott free. Play video games or work on html5lib or something 

[15:13:02.0000] <ezyang>
Eek. At least they called. 

[15:13:03.0000] <gsnedders>
ezyang: Hixie's answer would be work on Anolis :)

[15:13:04.0000] <ezyang>
I'm still not quite sure what anolis is. 

[15:13:05.0000] <gsnedders>
An HTML document post-processor :P

[15:14:00.0000] <gsnedders>
(Sorry, that's what I call it because I couldn't really work out what to call it.)

[15:14:01.0000] <ezyang>
Ohh, like XHTML Compiler? 

[15:14:02.0000] <gsnedders>
What?

[15:14:03.0000] <ezyang>
It like, takes HTML files and turns them into more awesome HTML files 

[15:15:00.0000] <gsnedders>
What it does is take HTML files and turn them into more awesome HTML files with things like a table of contents and cross-references.

[15:15:01.0000] <ezyang>
Yeah. I've done this before. 

[15:15:02.0000] <ezyang>
That's neat. What's the language? 

[15:15:03.0000] <gsnedders>
Python.

[15:15:04.0000] <gsnedders>
It's kinda slow.

[15:16:00.0000] <gsnedders>
Well, for the biggest user of it it is.

[15:16:01.0000] <ezyang>
Slow == ? 

[15:16:02.0000] <gsnedders>
But that's Hixie, and HTML 5 is a kinda big file. :)

[15:16:03.0000] <ezyang>
Aha. 

[15:16:04.0000] <ezyang>
Yeah, HTML 5 is big. 

[15:16:05.0000] <gsnedders>
It's been a while since I've run it against HTML 5 myself. I can't remember how long it takes.

[15:17:00.0000] <gsnedders>
I'll probably someday get around to rewriting it in some other language, possibly Java (seeming that has an HTML 5 parser for it, and is quick, once the VM is running).

[15:18:00.0000] <ezyang>
Ummm... I'm not convinced you actually want to rewrite it in Java. 

[15:18:01.0000] <Hixie>
"once the VM is running" isn't helpful since the VM would have to be started each time in most scenarios :-)

[15:19:00.0000] <ezyang>
Anyway, I presume you're loading the document entirely into memory and then doing transforms? 

[15:20:00.0000] <jgraham>
gsnedders: Just wait for the python VM nd/or Hixie's computer to get faster

[15:20:01.0000] <Hixie>
actually i use your computer :-P

[15:20:02.0000] <gsnedders>
jgraham: Hixie doesn't own the computer. :P

[15:20:03.0000] <jgraham>
Oh

[15:21:00.0000] <gsnedders>
Hixie: But on a webservice, like v.nu, it's fine, because the start-up time doesn't matter.

[15:21:01.0000] <gsnedders>
ezyang: Yes.

[15:21:02.0000] <Hixie>
if jgraham can set it up that way :-)

[15:21:03.0000] <jgraham>
Well wait for my sebserver to get faster which still should happen rather faster than HTML 5 grows

[15:21:04.0000] <jwalden>
ezyang: hey, I think I remember seeing your name in occasional skims of r-h-t over the past year, are you living in random?  (I was, graduated last year)

[15:21:05.0000] <jgraham>
/me is filled with fear at the hought of trying to run java on the server

[15:22:00.0000] <jgraham>
I mean I only get about 1 in 2 java applications to run on my own machine

[15:23:00.0000] <ezyang>
jwalden: Yeah! 

[15:23:01.0000] <jwalden>
ezyang: clam, or elsewhere?

[15:23:02.0000] <jwalden>
I was on clam first, third, fourth years, pecker third

[15:24:00.0000] <jgraham>
In no way is MIT a weird cult. At all. Honest.

[15:24:01.0000] <gsnedders>
Hixie: I would likely do so :)

[15:27:00.0000] <Dashiva>
/me bakes jgraham on a slow roast

[15:27:01.0000] <jgraham>
/me is not sure what he did o deserve being roasted

[15:29:00.0000] <Dashiva>
You mentioned cult after clams had been brought up

[15:29:01.0000] <gsnedders>
/me had people at TPAC trying to bully him into applying to places like MIT

[15:30:00.0000] <ezyang>
gsnedders: We've touched on this topic before, but do you know where you're going? 

[15:31:00.0000] <gsnedders>
ezyang: No. Well, Opera for this summer.

[15:31:01.0000] <ezyang>
Nice! 

[15:31:02.0000] <ezyang>
So you get to do the runaround this winter :-P 

[15:31:03.0000] <jgraham>
Can you roast clams?

[15:32:00.0000] <gsnedders>
ezyang: Seeming you're from the US you might have a clue: when would I have to do SAT by?

[15:33:00.0000] <ezyang>
Erm, why do you want to take the SAT? 

[15:33:01.0000] <ezyang>
As a non-US person, you can't get away with, say, IBs? 

[15:33:02.0000] <jgraham>
/me knows you have to take some special exams to go to grad school in the US

[15:34:00.0000] <jwalden>
MIT was awesome, crazy being in a place where so many other people are just as weird as you are :-D

[15:34:01.0000] <gsnedders>
ezyang: At places like MIT, no.

[15:35:00.0000] <jwalden>
last I heard IBs were never really in style, and even APs are going out these days

[15:35:01.0000] <ezyang>
I don't know the answer to your question offhand, since I did my SATs really early. 

[15:35:02.0000] <ezyang>
collegeboard.com can probably tell you when the last time for seniors to take it is. You'll want to get your scores in before the application date. 

[15:36:00.0000] <ezyang>
Also, if you don't know about it already, College Confidential is a really awesome resource. I'm a crufty college student turning Sophomore, so I've basically forgotten everything about the process. 

[15:37:00.0000] <jgraham>
gsnedders: January 

[15:38:00.0000] <gsnedders>
jgraham: Says where?

[15:38:01.0000] <jgraham>
The internet is a wonderful thing

[15:38:02.0000] <jgraham>
http://www.mitadmissions.org/topics/apply/international_applicants_helpful_tips/index.shtml#testingrequirements

[15:41:00.0000] <hendry>
evening, anyone have any thoughts on Website backups to share? http://kitenet.net/~joey/blog/entry/website_backup_RFC/

[15:41:01.0000] <ezyang>
Awesome! 84 passes. That's totally commit worthy. :-D 

[15:42:00.0000] <ezyang>
(285 failures, btw) 

[15:42:01.0000] <ezyang>
gsnedders: Shall I commit? 

[15:42:02.0000] <gsnedders>
ezyang: Yup

[15:43:00.0000] <ezyang>
Oh, wait, it's really chatty. Let me remove those var_dumps 

[15:46:00.0000] <ezyang>
Pushed. gsnedders, have fun! 

[15:46:01.0000] <gsnedders>
Anyhow, time for me to go sleep.

[15:46:02.0000] <ezyang>
(me, I'm going outside for a walk) 

[15:46:03.0000] <ezyang>
Ahaha 

[15:46:04.0000] <ezyang>
Good night! 

[15:46:05.0000] <gsnedders>
Actually, more realistically, eat a few biscuits, read Anna Karenina for a couple of hours, then sleep.

[16:08:00.0000] <Hixie>
#$%^*@$^@

[16:09:00.0000] <Hixie>
/me finally admits to himself that he needs to reverse engineer <marquee>

[16:09:01.0000] <Hixie>
i was really hoping to be able to define html5 without doing this

[16:11:00.0000] <ezyang>
What needs to be reverse engineered about it? 

[16:12:00.0000] <Hixie>
how does loop=4 work?

[16:12:01.0000] <Hixie>
if you change the behavior, does the looping restart?

[16:13:00.0000] <ezyang>
Aha 

[16:13:01.0000] <ezyang>
That doesn't sound too difficult, albeit slightly time consuming 

[16:15:00.0000] <Hixie>
it's not difficult, no :-)

[16:25:00.0000] <Hixie>
great, all the browsers act weirdly and differently from each other

[16:25:01.0000] <Dashiva>
And oddly enough I suspect that's not an ironic "great", but actually a positive thing?

[16:26:00.0000] <ezyang>
Hixie: In cases like this, is documenting existing behavior or deciding on a standard the first priority? 

[16:31:00.0000] <Hixie>
ezyang: what's the difference?

[16:31:01.0000] <Hixie>
Dashiva: it was ironic, though i guess it is actually good too, gives me more flexibility :-)

[16:45:00.0000] <jwalden>
browsers don't actually create anything new, they just copy like mad all the time :-)

[16:45:01.0000] <jwalden>
so writing a browser standard is just more copying :-)

[16:49:00.0000] <othermaciej>
we were all kinda trying to reverse engineer IE's <marquee>

[16:49:01.0000] <othermaciej>
if you can reverse engineer more accurately then win


2009-05-27
[17:12:00.0000] <gmiernicki>
has anyone thought about updating <blink> ?

[17:12:01.0000] <gmiernicki>
i think that tag needs some work

[17:21:00.0000] <ezyang>
I'm looking at an old implementation of tree building, and I see that it's divided into "phases". Was this old vocabulary that was removed from the spec later? 

[17:22:00.0000] <Hixie>
yeah

[17:23:00.0000] <ezyang>
I wonder how harmless it is. 

[17:23:01.0000] <Hixie>
the spec has changed quite a bit since then

[17:23:02.0000] <Hixie>
in ways that matter

[17:24:00.0000] <ezyang>
Ok. 

[17:24:01.0000] <annevk2>
you want to review all the old code, but you can probably get away with reusing parts of it

[17:24:02.0000] <ezyang>
This probably explains why less than a fourth of the tests are passing :-) 

[17:30:00.0000] <annevk2>
http://twitter.com/masinter/statuses/1929353498 ?

[17:33:00.0000] <annevk2>
someone has been sending off-topic emails or something?

[17:34:00.0000] <annevk2>
/me wonders if a certain Dean E has been sending emails

[17:35:00.0000] <annevk2>
(certainly seemed like Shelley got one from him)

[17:40:00.0000] <Hixie>
i sent him mail saying that calling the last 5 years of my technical life "not a technical specification" was personally offensive

[17:40:01.0000] <Hixie>
dunno about the other two!

[17:41:00.0000] <Hixie>
(he didn't seem to understand why i was offended, which is a little disconcerting, but oh well)

[17:45:00.0000] <ezyang>
"after after frameset" haha 

[17:45:01.0000] <Hixie>
yeah don't look too closely at the terminology in the html5 spec

[17:46:00.0000] <Hixie>
sometimes i backed myself into a corner and the results are embarassing :-)

[17:46:01.0000] <Hixie>
that's one example :-)

[00:05:00.0000] <MikeSmith>
does the Khronos Group not provide public access to their WG/mailing-list discussions?

[00:06:00.0000] <MikeSmith>
I'm trying to find an info page and/or mailing list for the‘Accelerated 3D on Web’group that was announced last month

[00:08:00.0000] <othermaciej>
I don't think they do

[00:08:01.0000] <othermaciej>
this is one reason I am puzzled that Mozilla chose it as the venue for Canvas 3D

[00:10:00.0000] <MikeSmith>
I guess going to Khronos makes sense if it's intended to be a spec for exposing OpenGL to javascript

[00:11:00.0000] <othermaciej>
but the hard part is the Web platform aspect (how to make the API safe without being slow)

[00:11:01.0000] <othermaciej>
not the 3D aspect

[00:11:02.0000] <othermaciej>
and Khronos is poor on process openness afaik

[00:14:00.0000] <MikeSmith>
hmm

[00:14:01.0000] <MikeSmith>
well, that's going to make things interesting

[00:19:00.0000] <othermaciej>
whatwg IRC gets results: http://blogs.adobe.com/jd/2009/05/_then_they_call_you_names.html

[00:19:01.0000] <MikeSmith>
"How Much Does it Cost to Get Involved?" section of http://www.khronos.org/members/ says "Free for Developers"

[00:19:02.0000] <MikeSmith>
but that page doesn't define what a "developer" is

[00:20:00.0000] <othermaciej>
maybe you just have to sign up then

[00:22:00.0000] <MikeSmith>
"A "troll" is someone who uses the anonymity possible on the Internet to harass others -- it is not someone who takes named responsibility for asking reasonable questions."

[00:24:00.0000] <annevk42>
what does "public-yet-pragmatically-inaccessible" mean?

[00:25:00.0000] <othermaciej>
that means this is a smoke-filled room

[00:25:01.0000] <othermaciej>
even though it is open to the public and logged on the Web

[00:25:02.0000] <annevk42>
I see

[00:27:00.0000] <annevk42>
wtf was has difficult question?

[00:27:01.0000] <annevk42>
s/has/his/

[00:28:00.0000] <othermaciej>
he's just asking the hard question of why the Web platform should support things like video or animation when Flash already does it fine

[00:28:01.0000] <othermaciej>
aren't we just fragmenting the Web by providing a way to do those things without Flash?

[00:28:02.0000] <othermaciej>
and aren't we polluting the hypertext purity of HTML by trying to bend it to the purpose of Rich Internet Applications?

[00:28:03.0000] <othermaciej>
shouldn't Web apps written using the Open Web platform remain poor?

[00:29:00.0000] <othermaciej>
why the class warfare? why would they want to be rich?

[00:29:01.0000] <MikeSmith>
heh

[00:29:02.0000] <sayrer>
there are some fair points there

[00:30:00.0000] <othermaciej>
sure, but

[00:30:01.0000] <othermaciej>
no matter what difficult questions he asks, the answer always seems to be the same: not to add functionality to HTML and to just use Flash instead

[00:31:00.0000] <sayrer>
well ok, but that bias is pretty obvious

[00:31:01.0000] <MikeSmith>
I wish he would just get back to pure unadultered trolling, instead of I-am-not-trolling

[00:32:00.0000] <sayrer>
for example, there are many Mozilla community members who don't see h264 video elements as a meaningful advance over flash movies

[00:32:01.0000] <sayrer>
but the interoperability story is worse, on balance

[00:32:02.0000] <sayrer>
although you do get on some devices where flash isn't present

[00:33:00.0000] <MikeSmith>
who cares about those device?

[00:33:01.0000] <MikeSmith>
devices

[00:33:02.0000] <othermaciej>
a surprising number of people care about devices such as the iPhone

[00:33:03.0000] <othermaciej>
I think the potential advance of <video> is twofold:

[00:33:04.0000] <MikeSmith>
othermaciej: yeah, I was being facetious

[00:34:00.0000] <MikeSmith>
I was channeling

[00:34:01.0000] <sayrer>
it's not a nice situation, morally. many handset makers are members of the mpeg consortium

[00:35:00.0000] <othermaciej>
(1) as to openness, I think <video> + H.264 is an improvement, because a patent-encumbered open standard is still an open standard; and there is the opportunity for expiration of certain key patents to help matters further

[00:35:01.0000] <sayrer>
open standards should be distributable by open source projects

[00:35:02.0000] <sayrer>
h.264 fails that test

[00:37:00.0000] <othermaciej>
(2) I think one of the most interesting aspects is the technology story, that video can be freely mixed with your other content and doesn't have to be in a special multimedia box

[00:37:01.0000] <sayrer>
yes, that is a theoretical advantage

[00:37:02.0000] <othermaciej>
after all, much of programming today is about connecting things together, not raw capabilities

[00:37:03.0000] <sayrer>
but so far the interoperability is pretty bad

[00:37:04.0000] <Philip`>
Browsers should improve their plugin integration so that Flash doesn't have to be in a special plugin box

[00:39:00.0000] <othermaciej>
I am more inclined at present to put Flash in a hermetically sealed box

[00:39:01.0000] <othermaciej>
so that it can stop crashing the browser so much

[00:39:02.0000] <sayrer>
yeah... like a different process

[00:44:00.0000] <annevk42>
oh look, the mainframe argument came back

[00:46:00.0000] <othermaciej>
everything has a constituency

[00:51:00.0000] <jgraham>
re: http://twitter.com/masinter/statuses/1929353498 it may be me that is being mocked for calling him impolite (asserting that something was useless because he personally couldn't think of the uses)

[00:57:00.0000] <hsivonen>
MPEG-4 as an "open standard" is persuasive against single-vendor ad hoc formats but still not good enough, because it is patent-encumbered

[00:57:01.0000] <othermaciej>
I don't think calling someone impolite is an ad hominem argument

[00:57:02.0000] <othermaciej>
unless you say that because they are impolite, the position they hold must be wrong

[01:06:00.0000] <heycam>
/me wonders why sam's latest mail is sent in UTF-7 and says "application/xhtml왩" instead of "application/xhtml+xml"

[01:09:00.0000] <annevk42>
heycam, maybe because your mail client sucks? wfm

[01:09:01.0000] <heycam>
could be

[01:10:00.0000] <Philip`>
I see "+xml" in the raw message, but I guess that gets decoded funny as UTF-7

[01:10:01.0000] <heycam>
yeah

[01:11:00.0000] <Philip`>
In Thunderbird I get 왩

[01:11:01.0000] <heycam>
i'm using mutt

[01:11:02.0000] <heycam>
(and as the author says, all mail clients suck...)

[01:11:03.0000] <MikeSmith>
heycam: yeah, I get the 왩 in mutt too

[01:11:04.0000] <heycam>
i guess thunderbird encoded incorrectly (since tb is what sam used it seems)

[01:12:00.0000] <othermaciej>
is he sending UTF-7 as a way to make a subtle point about recent list discussion?

[01:14:00.0000] <heycam>
heh that was my first thought, but then i couldn't work out what that point would be

[01:17:00.0000] <Philip`>
He was replying to a message that was also UTF-7, so I'd guess his mail client copied that by default

[01:20:00.0000] <hsivonen>
URL to UTF-7 email?

[01:20:01.0000] <heycam>
http://lists.w3.org/Archives/Public/public-html/2009May/0469.html

[01:20:02.0000] <hsivonen>
heycam: thanks

[01:20:03.0000] <heycam>
it interestingly caused the list archive to omit the links to senders/receivers

[01:22:00.0000] <heycam>
oh not omitted, just included verbatim as invalid html  (<span id="to"><dfn>To</dfn>: Manu Sporny <msporny⊙dc></span>)

[01:48:00.0000] <hsivonen>
Hixie: it seems you should have adopted the "HTML won" or "HTML is winning" line in the WaSP interview without talking about "4"

[01:50:00.0000] <annevk42>
lol yeah; I'm still wondering what Leif is trying to proof, but it sure is funny

[01:52:00.0000] <hsivonen>
I guess it really is inconsistent to say that HTML 4 is winning while at the same time holding the line that the implemented HTML and HTML 4 are very distinct

[01:52:01.0000] <hsivonen>
so HTML as she are spoke is winning hands down

[01:52:02.0000] <hsivonen>
not worth a thread, though

[01:54:00.0000] <annevk42>
it doesn't seem worth it to have an argument about that

[01:57:00.0000] <annevk42>
oh jd/adobe's archives are fun: http://blogs.adobe.com/jd/2008/08/lets_use_microsoft_runtimes.html

[01:59:00.0000] <annevk42>
hsivonen, we disabled UTF-7 outside email I believe

[02:00:00.0000] <annevk42>
hsivonen, UTF-7 and UTF-32 should no longer work in the latest Opera

[02:00:01.0000] <hsivonen>
annevk42: excellent

[02:00:02.0000] <hsivonen>
I've lost track of what happened to the UTF-7 disablement effort in Gecko

[02:04:00.0000] <hsivonen>
annevk42: "When asked why, the most common response is something along the lines of "Because Adobe might do something bad someday." (At this point I want to ask, "What, like they did with PostScript or PDF?" ;-)" is interesting

[02:04:01.0000] <hsivonen>
annevk42: because one might argue Adobe started morphing PDF in undesirable ways after PDF 1.4

[02:05:00.0000] <annevk42>
they killed FlashPaper

[02:05:01.0000] <hsivonen>
I didn't like the effort to turn Adobe Reader into an enterprise client platform before Adobe bought Macromedia

[02:06:00.0000] <annevk42>
they charge for Flash to make it work on devices as far as I know and Flash is not available everywhere

[02:06:01.0000] <hsivonen>
Adobe Intelligent Document platform it was called, IIRC

[02:07:00.0000] <annevk42>
does Gecko have its own Unicode and character encoding converter implementation?

[02:08:00.0000] <hsivonen>
annevk42: yes

[02:08:01.0000] <hsivonen>
annevk42: Opera has, too, right?

[02:09:00.0000] <annevk42>
afaik

[02:09:01.0000] <annevk42>
I think Safari relies on Mac OS libraries

[02:09:02.0000] <annevk42>
I wonder what Chrome is doing

[02:11:00.0000] <hsivonen>
I wouldn't be surprised if Chrome shipped ICU

[02:11:01.0000] <hsivonen>
Safari uses system ICU

[02:14:00.0000] <annevk42>
seems that Chrome uses ICU

[02:15:00.0000] <annevk42>
hmm, is it normal for browsers to be bundled with FFmpeg?

[02:15:01.0000] <annevk42>
-> http://src.chromium.org/viewvc/chrome/trunk/deps/third_party/

[02:17:00.0000] <MikeSmith>
what do they use xulrunner for?

[02:20:00.0000] <hsivonen>
MikeSmith: the check-in message is interesting: http://src.chromium.org/viewvc/chrome/trunk/deps/third_party/xulrunner-sdk/?view=log

[02:20:01.0000] <othermaciej>
SafariWin ships ICU

[02:20:02.0000] <annevk42>
I'm now getting "An Exception Has Occurred"

[02:21:00.0000] <hsivonen>
annevk42: It'll be interesting to see what they do with ffmpeg and how they navigate the patent/(L)GPL situation

[02:22:00.0000] <annevk42>
http://src.chromium.org/viewvc/chrome/trunk/deps/third_party/xulrunner-sdk/win/README.google hmm

[02:22:01.0000] <annevk42>
hsivonen, yeah

[02:28:00.0000] <othermaciej>
Chrome includes the xulrunner SDK?

[02:29:00.0000] <annevk42>
I don't think everything in that directory is included in Chromium

[02:43:00.0000] <zcorpan_>
Hixie: <td><dfn title="attr-marquee-behavior-alternate">scroll</dfn> - s/scroll/alternate/

[02:49:00.0000] <zcorpan_>
Hixie: why are marquee's event handler attributes just on <marquee> while the other event handler attributes are global?

[02:50:00.0000] <zcorpan_>
Hixie: "title="dom-marquee-trueSpeed"><code>trueSpeed</code></dfn>, DOM" - drop the comma

[02:51:00.0000] <annevk42>
zcorpan_, so they onlly pollute marquee iirc

[02:54:00.0000] <zcorpan_>
annevk2: why do we want to pollute everything with <video>-specific event handler attributes, then?

[03:00:00.0000] <annevk42>
so you can listen to video events on parent elements using event handlers

[03:00:01.0000] <annevk42>
I guess

[03:03:00.0000] <zcorpan_>
annevk2: but they don't bubble

[03:04:00.0000] <annevk42>
then I don't know

[03:04:01.0000] <Dashiva>
Capture?

[03:04:02.0000] <annevk42>
maybe event handlers are typically put on HTMLElement?

[03:04:03.0000] <annevk42>
Dashiva, event handlers can't capture

[03:05:00.0000] <Dashiva>
Oh, the attribute handlers? My bad.

[03:06:00.0000] <zcorpan_>
Hixie: "(Top to Button)" - typo

[03:23:00.0000] <Dashiva>
"There is no universal rule that will always work which would allow you to differentiate CURIEs from URIs"

[03:24:00.0000] <Dashiva>
I thought that was what safe CURIEs were for? <http://lists.w3.org/Archives/Public/public-html/2009May/0426.html>

[03:25:00.0000] <hsivonen>
Dashiva: seems like an unfortunate design

[03:33:00.0000] <MikeSmith>
would be nice to have some kind of general response available to the "we already spent month/years working on this specification"

[03:33:01.0000] <MikeSmith>
for the case where the "we" didn't include actual UA implementors

[03:34:00.0000] <MikeSmith>
or at least a way that I could get a nickel for every time somebody said that

[03:34:01.0000] <Dashiva>
Or a beer?

[03:34:02.0000] <MikeSmith>
heh

[03:34:03.0000] <MikeSmith>
yeah, even better

[03:34:04.0000] <MikeSmith>
or a chu-hai

[03:34:05.0000] <Dashiva>
You might be too drunk to work most days, though

[03:35:00.0000] <MikeSmith>
we've crossed that bridge already

[03:35:01.0000] <MikeSmith>
the too-drunk-to-work bridge

[03:37:00.0000] <jgraham>
/me is surprised MikeSmith made it all the way acroos the bridge rather than haphazardly stumbling off the edge into the water

[03:37:01.0000] <annevk42>
MikeSmith, we do, "Sorry, not part of the Web. Try again."

[03:37:02.0000] <hsivonen>
annevk42: how about some less divisive rethoric?

[03:45:00.0000] <jgraham>
annevk42: I'm not sure that is an answer to the same question. I mean RDFa could be "part of the web" in some sense but that doesn't mean it's a good idea

[03:46:00.0000] <annevk42>
"Sorry, won't work on the Web. Try again."

[03:47:00.0000] <jgraham>
As hsivonen said it's nice to be more diplomatic than that when you tell people their baby is ugly. Although it often doesn't seem to help

[03:47:01.0000] <Philip`>
MikeSmith: I guess the argument is that we should make decisions based on future costs, not on past costs

[03:48:00.0000] <annevk42>
jgraham, I think hsivonen was sarcastic

[03:48:01.0000] <annevk42>
problem with RDFa is that it's made for XHTML which is not really there nor here

[03:49:00.0000] <annevk42>
the retrofitting it for HTML process is interesting though

[03:50:00.0000] <Dashiva>
Leif doesn't seem to get the whole cowpath thing...

[03:51:00.0000] <jgraham>
annevk42: Maybe. IMHO the problem with RDFa is rather that it is too hard to use in any serialization. It doesn't really meet at least the last three criterion in http://standblog.org/blog/post/2009/05/20/About-Generativity

[03:51:01.0000] <jgraham>
But meeting all those criterion is pretty much essential to be a success on the web scale

[03:52:00.0000] <annevk42>
I was just commenting on it not even working

[03:53:00.0000] <Philip`>
If it's going to fail on the web scale, it'd be nice to minimise the harm that it causes while failing

[03:54:00.0000] <jgraham>
annevk42: Right but that is a technical problem which can (in principle) be solved. Focusing on the soluble problem whilst ignoring the hard, systemic issues seems like the wrong approach

[03:55:00.0000] <Philip`>
(e.g. by trying to ensure it's defined in a way that fits sensibly within HTML-processor architectures, so that it won't cause those architectures to be changed to accommodate it)

[04:01:00.0000] <Philip`>
(...and then it can compete with other approaches to data markup, and can win or lose based on the inherent merits of its design rather than on the bugginess/non-standardisation of implementations)

[04:02:00.0000] <Dashiva>
RDFa qualifies as QNames in content, right?

[04:04:00.0000] <jgraham>
Philip`: I am not disagreeing with the reasonableness of what your are doing, however I don't think it is the most important issue. I get the impression that some people thing that RDFa-in-HTML is a series of soluble technical challenges which, when solved, can pave the way to it being anointed the One True microdata solution.

[04:04:01.0000] <Philip`>
Dashiva: You probably shouldn't call them that, because people will disagree and say they're not QNames

[04:04:02.0000] <Dashiva>
But it's the same problem?

[04:04:03.0000] <Philip`>
Dashiva: Yes

[04:05:00.0000] <Dashiva>
I was wondering if it had fallen into obscurity because of all the other noise around RDFa, or if it's not considered an issue anymore

[04:05:01.0000] <Dashiva>
E.g. @prefix would still have this issue

[04:06:00.0000] <Philip`>
That's an issue to do with the inherent merits of RDFa's design, rather than being something that will cause a particular mess in text/html, so I don't care much about that problem now

[04:06:01.0000] <Dashiva>
Fair enough

[04:07:00.0000] <Philip`>
(xmlns:* will cause ugliness in HTML-to-XHTML serialisers and namespace-aware document models etc, whereas @prefix will just make RDFa fragile and hard to use)

[04:17:00.0000] <hsivonen>
annevk42: I wasn't being sarcastic

[04:18:00.0000] <hsivonen>
jgraham: yeah, RDFa has deeper issues than CURIEs, but it's alarming that fixing even the surface issues is so controversial

[04:19:00.0000] <annevk42>
hsivonen, oh, I thought since you're using those exact words...

[04:19:01.0000] <annevk42>
I was sarcastic, though, in case that was unclear

[04:20:00.0000] <Philip`>
At least people seem to mostly agree that the surface issues need to be fixed, and don't just discard it as being invalid markup and therefore irrelevant

[04:20:01.0000] <hsivonen>
annevk42: I was referring to shepazu's criticism, though

[05:44:00.0000] <Philip`>
http://buzzword.org.uk/2009/ttldent/spec - "CURIE prefixes are hard-coded by this specification, with a fallback to lookups on prefix.cc"

[05:45:00.0000] <hsivonen>
Is there still any data on whether Google hard-codes v: ?

[05:47:00.0000] <annevk42>
"The <title> is parsed for ttldent"

[05:47:01.0000] <annevk42>
is that legal in RDFa too?

[05:50:00.0000] <Philip`>
hsivonen: I've still not seen any way to test Google's RDFa support

[06:08:00.0000] <hsivonen>
are microbloggers expected to type ttldent?

[06:09:00.0000] <hsivonen>
or are higher-layer semantic microblogging apps supposed to use twitter/identi.ca as RDF conduits?

[07:27:00.0000] <hsivonen>
whee! lots of unit test failures in the html5lib suite after making EOF inside of tag discard the tag

[07:31:00.0000] <Dashiva>
whee as in "There is non-trivial test coverage!" or "Oh boy, lots of work fixing all these tests"? :)

[07:34:00.0000] <hsivonen>
Dashiva: more like: I wish test for condition x didn't gratuitously depend on error handling in case y

[07:35:00.0000] <hsivonen>
/me goes look up blame for tests3

[07:35:01.0000] <gsnedders>
tests3 in the tokenizer?

[07:36:00.0000] <hsivonen>
gsnedders: yeah

[07:36:01.0000] <gsnedders>
They're the programmatically generated ones by Philip` testing all cases in the tok, IIRC

[07:36:02.0000] <hsivonen>
Philip`: should I manually edit tests3 or should they be regenerated?

[07:38:00.0000] <hsivonen>
Philip`: specifically, all the tests testing transitions inside tags should end in '>' instead of ending in EOF

[08:11:00.0000] <Philip`>
hsivonen: Regeneration is non-trivial, so I suggest just editing them manually for now

[08:11:01.0000] <Philip`>
("non-trivial" means I have to implement the new spec in OCaml)

[08:12:00.0000] <Philip`>
(and probably modify the way it decides which test cases to generate)

[08:12:01.0000] <jgraham>
Philip`: But functional programming is supposed to be so simple :)

[08:13:00.0000] <Philip`>
jgraham: Sure, I wouldn't want to write a Java program to automatically generate tokeniser tests :-)

[08:13:01.0000] <Philip`>
but still I'd have to read the spec and work out what's changed

[08:14:00.0000] <jgraham>
Philip`: Maybe you should hve ritten it in lisp. It is well knon that lisp solves all problems for you

[08:25:00.0000] <hsivonen>
Philip`: ok

[08:46:00.0000] <hsivonen>
http://blog.digitalbackcountry.com/2009/05/html5-versus-flash-versions/

[08:48:00.0000] <annevk42>
seems a bit ill-informed to base it on outdated Wikipedia information

[08:48:01.0000] <Rik|work>
so creating new features in flash is contributing to the open web ?

[08:49:00.0000] <annevk42>
if they take of they certainly create incentives

[08:53:00.0000] <jgraham>
/me is intrigued at how actionscript can cover <nav>

[08:55:00.0000] <hsivonen>
jgraham: maybe they added an accessibility API in that rev?

[08:57:00.0000] <jgraham>
hsivonen: Could be. In general that table doesn't give enough information to tell what the corresponding feature is assumed to be

[08:57:01.0000] <jgraham>
e.g. I have never seen flash integrate with my browser urlbar which a good implementation of <input type="url"> would

[09:00:00.0000] <hsivonen>
http://www.webkitchen.be/2009/05/27/adobe-versus-the-open-web/ seems to miss that the Open Web is about the 'open' part--not about the existence of features

[09:05:00.0000] <annevk2>
could be that he's just trying to mislead

[09:07:00.0000] <Dashiva>
If he really is Asa, that would be expected

[09:07:01.0000] <Dashiva>
Wait, wrong blog

[09:09:00.0000] <jgraham>
His job title is "professional fanboy" so...

[09:09:01.0000] <jgraham>
Oh wait they call it "Adobe platform evangelist" but the effect is the same

[09:10:00.0000] <Dashiva>
"Remember the old days where you had to choose between Windows Media Player, Real Player and Quicktime?"

[09:10:01.0000] <Dashiva>
Remember the new days when you have to choose between Flash and Silverlight and ...

[09:10:02.0000] <hsivonen>
are you guys trying to make it to Last Week? 

[09:11:00.0000] <Dashiva>
I don't have to try to make it to Last Week

[09:11:01.0000] <Dashiva>
It just happens on its own

[09:17:00.0000] <jgraham>
I gave up trying to make MLW when it became obvious how easy that was

[09:18:00.0000] <Dashiva>
Although the last few times I'm just collateral damage for saying things inbetween the real juicy lines

[09:18:01.0000] <ezyang>
Good morning all! 

[09:19:00.0000] <Dashiva>
How open are the various adobe things he lists as open?

[09:20:00.0000] <jgraham>
ezyang: The i18n people will have you for comments like that

[09:20:01.0000] <Dashiva>
I heard something about only the RTMP spec being released, but in a way so it wasn't possible to actually use it

[09:23:00.0000] <ezyang>
jgraham: I come from MIT: "good morning" is an appropriate salutation for whenever you've woken up, whether it be 10AM, 2PM, 6PM, or 1AM 

[09:25:00.0000] <Dashiva>
But what if we haven't just woken up?

[09:25:01.0000] <Dashiva>
Why are you pushing your personal state onto us?

[09:25:02.0000] <ezyang>
/me laughs maniacally 

[09:25:03.0000] <hsivonen>
http://twitter.com/psd/statuses/1935782044

[09:25:04.0000] <Dashiva>
What if some of us are easily manipulated, and end up thinking it's morning

[09:39:00.0000] <ezyang>
gsnedders: Did you get a chance to try running the new test suite? 

[09:47:00.0000] <ezyang>
lock on php-tb TreeConstructer.php 

[10:05:00.0000] <ezyang>
Ugh... the whatwg spec keeps reflowing for no apparent reason. 

[10:05:01.0000] <ezyang>
Maybe I should get a dual monitor setup 

[10:10:00.0000] <gsnedders>
ezyang: No

[10:10:01.0000] <gsnedders>
ezyang: I read Anna Karenina for far too long last night and only got up at 1pm :P

[10:11:00.0000] <gsnedders>
ezyang: Also: there's more to life than html5lib :P

[10:11:01.0000] <gsnedders>
Or, at least, I think there is.

[10:13:00.0000] <ezyang>
Heh 

[10:13:01.0000] <ezyang>
No life for me. Not at least until Saturday 

[10:13:02.0000] <gsnedders>
I guess I'm not meant to have any seeming I'm in study leave still.

[10:14:00.0000] <jgraham>
gsnedders: You are probably suppoed to study :)

[10:14:01.0000] <gsnedders>
jgraham: But the AH English class is going out tonight!

[10:14:02.0000] <gsnedders>
jgraham: The head of English arranged it! :P

[10:14:03.0000] <Philip`>
/me ought to have no life until 2pm tomorrow, but will probably end up playing games instead of finishing his paper properly

[10:15:00.0000] <gsnedders>
Basically, we're a channel that's good at procrastinating.

[10:15:01.0000] <gsnedders>
And in the case of several of us good at procrastinating into doing things that are useful, but don't actually need to be done now.

[10:16:00.0000] <gsnedders>
jgraham: Also, I've apparently signed up in a moment of stupidity to go to a Jesus open day.

[10:16:01.0000] <gsnedders>
(College, that is)

[10:16:02.0000] <jgraham>
Oh, I was confused for a moment there

[10:16:03.0000] <jgraham>
Which is ironic

[10:16:04.0000] <gsnedders>
:D

[10:16:05.0000] <jgraham>
Anyway, good college

[10:17:00.0000] <gsnedders>
jgraham: What? Don't you think of it? Haven't you had a thing or two to do with it? :P

[10:18:00.0000] <jgraham>
/me notes that gsnedders is no longer making sense

[10:18:01.0000] <gsnedders>
"no longer"?

[10:18:02.0000] <gsnedders>
I stopped making sense a long time ago!

[10:18:03.0000] <gsnedders>
Making sense is _so boring_.

[10:18:04.0000] <gsnedders>
I mean, people understand you!

[10:19:00.0000] <Philip`>
If they're going to name colleges after people from books, they ought to pick good ones, like Saruman College

[10:20:00.0000] <gsnedders>
Philip`: Peh. There are better colleges, like Hixie College. :P

[10:21:00.0000] <jgraham>
/me is struck by the overwhelming tempation to rename all the Cambridge colleges after Harry Potter characters

[10:24:00.0000] <Philip`>
jgraham: The university's 800th anniversary seems like a great opportunity to modernise in such a way

[10:26:00.0000] <jgraham>
It would make us more accessible to people from a wide range of backgrounds

[10:30:00.0000] <gsnedders>
Philip`: Oh, and don't worry. I won't bully you into having supper with me for once. :D

[11:00:00.0000] <ezyang>
Whoo, currently at "After head" 

[11:20:00.0000] <Philip`>
http://radar.oreilly.com/upload/2009/05/html5.png - yay, geolocation

[11:29:00.0000] <ezyang>
/me rolls his sleeves up, and starts auditing the "in body" mode 

[11:33:00.0000] <Philip`>
Watch out for intestines

[11:33:01.0000] <ezyang>
Ugh, I just hit something squishy... 

[11:46:00.0000] <ezyang>
Hmm... I wonder why <h1><b><h1>foo</h1></b></h1> stopped being a parse error 

[11:55:00.0000] <Hixie>
btw has everyone seen http://www.youtube.com/html5 ?

[11:55:01.0000] <Hixie>
hover over the thumbnails to see something that couldn't be done with flash :-)

[11:55:02.0000] <gavin_>
doesn't work in Firefox

[11:56:00.0000] <Rik|work>
Hixie: crashing a webkit nightly ? :)

[11:56:01.0000] <Hixie>
gavin_: yeah, it uses h.264 sadly

[11:57:00.0000] <Hixie>
Rik|work: works on mine

[11:58:00.0000] <Rik|work>
Hixie: dailymotion already did a demo like

[11:58:01.0000] <ezyang>
Not running nightlies, unfortunately :-( 

[11:58:02.0000] <atwilson>
ap: quick question about how best to dispatch pending events in WorkerContext.close() - you around?

[11:59:00.0000] <Rik|work>
http://openvideo.dailymotion.com/

[11:59:01.0000] <ap>
atwilson: sure, let's discuss it on #webkit though

[11:59:02.0000] <atwilson>
Sigh, I'm an idiot. I'm using new IRC software and clearly can't figure out how to select the right channel :( sorry folks.

[11:59:03.0000] <Hixie>
Rik|work: nice

[12:00:00.0000] <Rik|work>
sad part is they require firefox 

[12:00:01.0000] <Rik|work>
but that's just a demo

[12:00:02.0000] <Hixie>
yeah well until we figure out a standard codec everyone can use and implement, that's always gonna be a problem

[12:01:00.0000] <Rik|work>
no, they require firefox, not ogg

[12:02:00.0000] <Hixie>
Rik|work: oh

[12:02:01.0000] <Hixie>
Rik|work: weird

[12:03:00.0000] <Rik|work>
maybe they were working with mozilla marketing, i don't know

[12:05:00.0000] <annevk42>
sort of glad I called on that email from masinter

[12:05:01.0000] <annevk42>
seems he wasn't quite in the know how we were using user agent

[12:06:00.0000] <hober>
seems like the spec uses it the way ordinary web developers use it...

[12:07:00.0000] <annevk42>
hmm, the YouTube HTML5 demo doesn't work in any of my browsers

[12:09:00.0000] <annevk42>
hober, although the spec is quite hard to read, the terms it introduces and uses are at least somewhat sane most of the time

[12:12:00.0000] <zcorpan_>
Hixie: "<dl class="intro">" should be domintro

[12:13:00.0000] <Hixie>
fixed

[12:19:00.0000] <hsivonen>
what does http://www.youtube.com/html5 require to work?

[12:19:01.0000] <ezyang>
"An HTML5 capable browser", of course! 

[12:20:00.0000] <annevk42>
WebKit nightlies I guess

[12:20:01.0000] <annevk42>
+ Safari, doesn't seem like Chromium does <video>

[12:20:02.0000] <hsivonen>
:-(

[12:21:00.0000] <Hixie>
a <video> impl with h.264 decoder, i think

[12:21:01.0000] <Hixie>
latest safari 4 beta with latest webkit works

[12:22:00.0000] <inimino>
it doesn't require O3D?

[12:23:00.0000] <inimino>
oh, I guess that's talking about the demo in the video, not the demo the video is in...

[12:24:00.0000] <annevk42>
since just before dinner: "246 more results since you started searching. Refresh to see them." (twitter.com)

[12:26:00.0000] <annevk42>
the publicity will give us more review (well, maybe), but Google did turn HTML5 into a term like DHTML, Ajax, etc.

[12:27:00.0000] <Hixie>
i said that would happen about 6 months ago :-)

[12:27:01.0000] <Hixie>
when we starting talking about splitting stuff out i said we'd need a new term

[12:27:02.0000] <Hixie>
like web 5.0 or something

[12:28:00.0000] <annevk42>
I guess HTML5 will now have 3 meanings

[12:28:01.0000] <annevk42>
language, serialization, catch-all for cool new stuff

[12:29:00.0000] <annevk42>
(and a historical fourth, XHTML 1.0)

[12:29:01.0000] <zcorpan_>
/me notes that the youtube html5 page has ugly link borders in firefox

[12:30:00.0000] <zcorpan_>
when will firefox have no border by default?

[12:30:01.0000] <annevk42>
"but also suggested that we need new extensions to HTML 5, for example, to support events from the accelerometer in the phone"

[12:32:00.0000] <ezyang>
Huh. "Parse error. Change the token's tag name to "img" and reprocess it. (Don't ask.)" 

[12:33:00.0000] <annevk42>
trouble reading?

[12:34:00.0000] <annevk42>
"Don't ask."

[12:34:01.0000] <Hixie>
he didn't technically ask :-)

[12:34:02.0000] <annevk42>
:p

[12:36:00.0000] <zcorpan_>
i wonder where <image> was first implemented

[13:02:00.0000] <Hixie>
what happened to all the interesting discussion on public-html?

[13:02:01.0000] <Hixie>
there hasn't been anything controverial in a day or so

[13:34:00.0000] <ezyang>
I've hit the small intestines. Auditing "end tag of formatting element" 

[13:34:01.0000] <ezyang>
How much should I expect this algorithm to have changed? 

[13:35:00.0000] <Philip`>
/me doesn't remember that having been changed much ever

[13:36:00.0000] <Philip`>
(The adoption agency definitely changed a bit, though)

[13:36:01.0000] <Hixie>
ezyang: i honestly can't remember

[13:37:00.0000] <ezyang>
Ok, I'm not at adoption yet :-) 

[13:38:00.0000] <ezyang>
Ok, looks like step 5+ changed 

[13:50:00.0000] <hsivonen>
I'm amused how the mailing list archive demonstrates the badness of UTF-7

[14:01:00.0000] <ezyang>
Hoorah, adoption agency algorithm vetted 

[14:23:00.0000] <gsnedders>
ezyang: Having fun while I've been out with my English class? :P

[14:50:00.0000] <zcorpan>
"“5 > 2″ declared a slide from Vic Gundotra, VP of Engineering at Google, referring to new capabilities of HTML 5 that let us move beyond Web 2.0 apps." - http://blogs.zdnet.com/Burnette/?p=1123

[14:51:00.0000] <gsnedders>
Hah.

[14:51:01.0000] <hober>
whut. http://www.reddit.com/r/programming/comments/8nomt/google_bets_big_on_html_5/c09uzvf

[14:53:00.0000] <gsnedders>
hober: Adverts only exist on the web.

[14:53:01.0000] <gsnedders>
It's simple.

[14:58:00.0000] <Hixie>
haha, i was hoping he'd keep the 5 > 2 slide

[14:58:01.0000] <Hixie>
awesome

[14:58:02.0000] <ezyang>
That is pretty good. 

[15:00:00.0000] <ezyang>
Why does the behavior for "table" end tag in "in table" insertion mode not generate implied end tags? 

[15:01:00.0000] <gsnedders>
ezyang: "Please leave your sense of logic at the door."

[15:01:01.0000] <Dashiva>
ezyang: Which implied end tags would that be?

[15:01:02.0000] <ezyang>
Hmmmm... 

[15:01:03.0000] <ezyang>
Good point. 

[15:02:00.0000] <Dashiva>
I'm not entirely sure, but intuitively I'd guess that "in table" means there is nothing open inside the table tag

[15:02:01.0000] <ezyang>
This is not strictly true (great example is "in body"), but I could convince myself it would always be true for "in table" 

[15:03:00.0000] <gsnedders>
/me is listening to Meeting Paris Hilton by Cansei De Ser Sexy from Cansei De Ser Sexy

[15:03:01.0000] <gsnedders>
/me is blatantly cool.

[15:03:02.0000] <Hixie>
ezyang: it doesn't need to because there can never be any open elements there that aren't parse errors

[15:04:00.0000] <ezyang>
savvy 

[15:04:01.0000] <gsnedders>
/me throws a parse error

[15:04:02.0000] <olliej>
/me wonders if google actually enjoys insulting the people who write their browser

[15:05:00.0000] <ezyang>
it's super effective! 

[15:05:01.0000] <Dashiva>
ezyang: Yeah, "in body" is different because body descendents don't all get their own parse states

[15:06:00.0000] <gsnedders>
/me does backflip

[15:10:00.0000] <Dashiva>
olliej: link?

[15:11:00.0000] <olliej>
Dashiva: http://radar.oreilly.com/2009/05/google-bets-big-on-html-5.html

[15:11:01.0000] <Hixie>
olliej: who are we insulting now? i thought we just gave big props to mozilla, apple, and opera today

[15:12:00.0000] <Philip`>
http://radar.oreilly.com/upload/2009/05/browser_innovation.png - apparently the Apple products are very low on the User Experience scale

[15:12:01.0000] <Hixie>
actually i think that's saying they were early

[15:12:02.0000] <olliej>
i assume because we only write the browser

[15:13:00.0000] <olliej>
rather than being awesome

[15:13:01.0000] <Philip`>
Hixie: That's what the x-axis says, not the y-axis

[15:13:02.0000] <olliej>
Hixie: ooh

[15:13:03.0000] <ezyang>
The reddit commentors have said this already, but I'll say it again: wtf images 

[15:13:04.0000] <olliej>
maybe it's because of geolocation

[15:14:00.0000] <olliej>
Hixie: which we implemented

[15:14:01.0000] <Hixie>
Philip`: sure, but if you imagine this line graph as being the cumulative result, it just means that apple contributed first

[15:14:02.0000] <olliej>
but it's good to see the arrogance and misrepresentation that defines google has not gone away

[15:14:03.0000] <Hixie>
olliej: i really have no idea what you're talking about :-)

[15:16:00.0000] <Hixie>
Philip`: (think of it as analogous to a chart that has time vs total market share, with each product release being a dot on the line)

[15:16:01.0000] <Hixie>
Philip`: (the products don't correspond to absolute values on the y-axis)

[15:17:00.0000] <sayrer>
it's just a chart at a hype event

[15:17:01.0000] <sayrer>
nothing to get worked up about

[15:18:00.0000] <sayrer>
remember when steve jobs had the stainless steel and wooden pie chart that showed Firefox disappearing?

[15:21:00.0000] <Dashiva>
Everything except Safari and IE, wasn't it?

[15:21:01.0000] <Dashiva>
Poor Chrome, killed off before it even appeared :)

[15:21:02.0000] <annevk42>
sayrer, is there a video of that? sounds like fun

[15:21:03.0000] <ezyang>
Will problems happen if I ignore "if that token wasn't ignored"? 

[15:22:00.0000] <gsnedders>
ezyang: Yes.

[15:22:01.0000] <gavin_>
annevk42: http://john.jubjubs.net/2007/06/14/a-pictures-worth-100m-users/

[15:22:02.0000] <ezyang>
ruh oh 

[15:23:00.0000] <ezyang>
/me rooby dooby doo 

[15:24:00.0000] <annevk42>
gavin_, wow, lol

[15:24:01.0000] <gavin_>
I didn't even realize it was a stainless steel and wooden pie chart

[15:24:02.0000] <gavin_>
that offends me deeply

[15:25:00.0000] <Dashiva>
What is firefox anyway? Cheese?

[15:25:01.0000] <gavin_>
(my father was killed by a wooden pie chart in the Stainless Steel War)

[15:25:02.0000] <sayrer>
yes, the chart showed shocklingly tacky keynote defaults

[15:25:03.0000] <olliej>
gavin_: hehe

[15:25:04.0000] <sayrer>
like, the yuppie kitchen of powerpoint

[15:26:00.0000] <olliej>
sayrer: i at least apologised to those moz people i knew, and more importantly we don't have a track record of acting like we are single handedly responsible for work done by everyone else

[15:27:00.0000] <annevk42>
“HTML 5 gives you a <video> tag that’s as simple to use as the <image> tag.” -- http://www.webmonkey.com/blog/Google_Throws_Its_Weight_Behind_HTML_5

[15:27:01.0000] <Hixie>
so apparently quite hard to use :-/

[15:28:00.0000] <olliej>
annevk42: it would be nice if chrome implemented it, given webkit has had it for more than a year and it took the gtk people 3 days to do their backend

[15:28:01.0000] <gsnedders>
And non-obvious in terms of what it gets parsed to :P

[15:28:02.0000] <Hixie>
olliej: seriously, what did google do today to offend you? As far as I'm aware we've done nothing but praise apple, opera, and mozilla today.

[15:28:03.0000] <Dashiva>
olliej: Why should anyone have to apologize? It's just marketing making a mess of things as usual?

[15:29:00.0000] <annevk42>
/me agrees it would be cool if Google came up with something nice for <video>

[15:29:01.0000] <olliej>
Hixie: i only pay attention to the press sound bites, because they're what will be repeated as truth from here on out

[15:29:02.0000] <olliej>
anyhoo

[15:29:03.0000] <olliej>
/me should not vent here

[15:29:04.0000] <olliej>
as it's so far from topic

[15:29:05.0000] <gsnedders>
There's a topic in here? :\

[15:30:00.0000] <annevk42>
olliej, we hardly ever stay on topic :)

[15:30:01.0000] <gsnedders>
olliej: Now, go fix #13519 :P

[15:30:02.0000] <ezyang>
On another note, shouldn't "clear the stack back to a table body context" get a global definition? 

[15:30:03.0000] <sayrer>
olliej, you just need a bunch of unsold ad inventory to promote Safari/Windows

[15:31:00.0000] <annevk42>
olliej, it's not enough that everyone who cares enough to know knows Chromium is based on WebKit?

[15:31:01.0000] <Hixie>
olliej: ok, what has the press said to offend you? I've not seen anything bad so far.

[15:31:02.0000] <olliej>
Hixie: i was referring to the slide in which chrome apparently is more awesome than safari despite just using webkit

[15:32:00.0000] <olliej>
anyhoo

[15:32:01.0000] <Hixie>
olliej: if you are reading that line chart as saying that chrome is more awesome than safari then you are completely misreading the chart

[15:32:02.0000] <olliej>
like i said, offtopic

[15:32:03.0000] <olliej>
Hixie: y-axis is user experience

[15:32:04.0000] <Hixie>
olliej: if anything, it's saying chrome was last and building on the others

[15:32:05.0000] <olliej>
Hixie: x axis is time

[15:32:06.0000] <olliej>
hmmm

[15:33:00.0000] <Hixie>
olliej: it's a single line showing the cumulative experience (whatever that means)

[15:33:01.0000] <Hixie>
olliej: if it was a bar chart, then sure

[15:33:02.0000] <Hixie>
olliej: but it's showing how each of those browsers contributed to a single good (near-native) experience

[15:33:03.0000] <Hixie>
olliej: it's not even remotely trying to insult anyone, and like i said, the most it could say is safari was the first to contribute and chrome the last

[15:34:00.0000] <Hixie>
olliej: it is, if anything, complimenting apple more than google!

[15:34:01.0000] <olliej>
/me wonders what chrome has contributed given it just uses webkit -- not a single listed feature of chrome isn't from webkit

[15:34:02.0000] <olliej>
no matter

[15:34:03.0000] <Hixie>
it _does_ matter

[15:34:04.0000] <Hixie>
you can't just go around saying that we're insulting you when we're actively trying to praise you

[15:36:00.0000] <olliej>
i read that chart as saying chrome is more awesome than safari, with no indication of what makes it more awesome -- the only difference is gelocation which is part of webkit

[15:37:00.0000] <Hixie>
then you read the chart completely wrong

[15:37:01.0000] <olliej>
apparently so

[15:37:02.0000] <olliej>
but i assume google is evil

[15:37:03.0000] <Hixie>
maybe you should stop making such assumptions

[15:37:04.0000] <Hixie>
it makes you come off as highly paranoid

[15:38:00.0000] <Dashiva>
Well, we could also agree that both Apple and Google are evil in their own ways

[15:38:01.0000] <olliej>
yeah probably

[15:38:02.0000] <gsnedders>
jgraham: Here's a challenge for this summer: get a photograph of me that I like.

[15:39:00.0000] <Dashiva>
But this doesn't really seem like a case of it...

[15:39:01.0000] <Hixie>
oh i'm not saying google is perfect, by any means

[15:39:02.0000] <Hixie>
but it's hard to take criticism seriously when one even criticises the praise

[15:40:00.0000] <olliej>
Hixie: meh

[15:40:01.0000] <olliej>
Hixie: is there a way to block the context menu?

[15:40:02.0000] <Hixie>
"meh" is hardly an apology, but i'll take it, since i assume it's the best i'll get

[15:40:03.0000] <olliej>
Hixie: eg. i right click (or whatever) to bring up the context menu

[15:41:00.0000] <olliej>
Hixie: i feel that slide isn't praise

[15:41:01.0000] <Hixie>
html5 lets you add stuff to the context menu, but there's no way to prevent it altogether

[15:41:02.0000] <Hixie>
(intentionally)

[15:41:03.0000] <olliej>
Hixie: but i can agree it's ambiguous

[15:41:04.0000] <olliej>
Hixie: ooh there is?

[15:41:05.0000] <gsnedders>
http://www.youtube.com/html5 doesn't have full-screen mode :(

[15:41:06.0000] <olliej>
/me dreams of a day when html context menus actually extend beyond the window

[15:41:07.0000] <Hixie>
olliej: yeah, look up the <menu> element and the contextmenu="" attribute

[15:41:08.0000] <gsnedders>
=> HTML 5 sux.

[15:42:00.0000] <Hixie>
gsnedders: hit f11

[15:42:01.0000] <gsnedders>
Hixie: Yes, Exposé works :P

[15:42:02.0000] <Hixie>
well if you're on mac... speak to olliej :-)

[15:42:03.0000] <olliej>
Hixie: yeah i just hit the same thing as gsnedders :D

[15:42:04.0000] <atwilson>
Hixie: quick question about worker handling of exceptions. The spec says: "For dedicated workers, if the error is still not handled afterwards, or if the error occured while handling a previous script error, the user agent must further queue a task to fire a worker error event at the Worker object associated with the worker."

[15:43:00.0000] <atwilson>
What does "still not handled" mean - does this mean "if the error event is not cancelled"?

[15:43:01.0000] <gsnedders>
/me probably ought to forsake olliej and change browser

[15:43:02.0000] <olliej>
/me cries

[15:43:03.0000] <Hixie>
atwilson: see html5, 6.5.6.5 Runtime script errors

[15:43:04.0000] <gsnedders>
olliej: Well, if Apple had hird me…

[15:43:05.0000] <atwilson>
thx

[15:43:06.0000] <gsnedders>
*hired

[15:43:07.0000] <Hixie>
atwilson: (it defines what it means for an error to be not handled)

[15:44:00.0000] <Hixie>
atwilson: let me know if it's still vague after that :-)

[15:44:01.0000] <atwilson>
All human knowledge is contained within the vastness that is www.w3.org/TR/html5

[15:45:00.0000] <Hixie>
heh

[15:45:01.0000] <Hixie>
unlikely :-P

[15:46:00.0000] <atwilson>
sweet, 5.5.6.5 is crystal clear.

[15:46:01.0000] <gsnedders>
olliej: See, it's Apple's own fault :P

[15:46:02.0000] <Hixie>
atwilson: sweet

[15:47:00.0000] <Hixie>
atwilson: sorry about the cross-spec references being hard to follow

[15:47:01.0000] <Hixie>
atwilson: blame gsnedders, he hasn't figured out a way to automatically do them yet :-)

[15:47:02.0000] <Hixie>
/me hides

[15:47:03.0000] <gsnedders>
Hixie: No, I have.

[15:47:04.0000] <gsnedders>
Hixie: I might not have implemented it yet, but I've figured it out.

[15:48:00.0000] <Hixie>
oh, good

[15:48:01.0000] <Hixie>
anything i need to do?

[15:48:02.0000] <gsnedders>
Hixie: I don't think so

[15:49:00.0000] <annevk42>
https://twitter.com/adambarth/status/1938713488 :p

[15:55:00.0000] <annevk42>
/me wonders what https://bugzilla.mozilla.org/show_bug.cgi?id=495112 is about -- looks bad

[15:55:01.0000] <gavin_>
bug 455070 caused a regression

[15:55:02.0000] <gavin_>
https://bugzilla.mozilla.org/show_bug.cgi?id=494543

[15:56:00.0000] <gavin_>
that led to https://bugzilla.mozilla.org/show_bug.cgi?id=495112 and https://bugzilla.mozilla.org/show_bug.cgi?id=494799

[15:56:01.0000] <annevk42>
still, changing from origin to domain makes it incompatible with other impl

[15:56:02.0000] <gavin_>
/me isn't really familiar with the details

[15:59:00.0000] <annevk42>
"The first test of that will be IE’s implementation of the canvas tag, which has been promised but not yet delivered." -- http://blogs.zdnet.com/Burnette/?p=1123 Somehow I doubt that's been promised.

[15:59:01.0000] <gsnedders>
/me wonders what his Yahoo id is

[16:17:00.0000] <annevk42>
to follow up on that Mozilla bug above, apparently the plan is to fix it in a later release...

[16:19:00.0000] <gsnedders>
How do you customize the toolbar that only appears while pages are loading in Opera!?

[16:20:00.0000] <Philip`>
gsnedders: Select 'customise' and 'show hidden toolbars while customising'?

[16:21:00.0000] <gsnedders>
Philip`: Oh, there.

[16:21:01.0000] <gsnedders>
Philip`: That's well hidden.

[16:21:02.0000] <Philip`>
Indeed, it's hidden as a labelled checkbox on the customisation screen - nobody would ever think to look there

[16:22:00.0000] <gsnedders>
Well, in Op10 it's in the Appearence window under the Toolbars tab, while I'm looking at Buttons wanting to drag stuff!

[16:25:00.0000] <annevk42>
http://www.reddit.com/r/programming/comments/8nomt/google_bets_big_on_html_5/c09v0fy -- "The internet isn't and has never been "trapped inside web browsers". The web, on the other hand, is. BECAUSE THAT'S THEIR FUCKING ROLE. \n Also, all of Air, Silverlight, JavaFX blow donkey balls." 

[16:25:01.0000] <ezyang>
hear hear 

[16:26:00.0000] <gsnedders>
Philip`: Huh. It appeared once, and now it's gone.

[16:27:00.0000] <ezyang>
/me has two more states to convert 

[16:27:01.0000] <ezyang>
The previous author had a similar sentiment: "/* OMG DONE!! */" 

[16:30:00.0000] <Hixie>
lol, i got five upvotes for saying that i upvoted the "HTML5 sucks! All hail the semantic web! FOR THE GLORY OF XHTML2!" comment

[16:30:01.0000] <Hixie>
i'll never understand redit

[16:30:02.0000] <Hixie>
reddit

[16:31:00.0000] <ezyang>
moooo 

[16:32:00.0000] <jacobolus>
Hixie: I hear you can cash out your reddit upvotes for plush animals in the afterlife.

[16:32:01.0000] <Hixie>
sweet

[16:33:00.0000] <annevk42>
I upvoted that one just because I thought it was funny

[16:33:01.0000] <annevk42>
though prolly not worth much since I've a karma of 1

[16:34:00.0000] <annevk42>
"I hope the spec writers aren't presuming to tell the rest of the world what is intuitive." -- I'm not writing the spec, but I believe I just did...

[16:34:01.0000] <ezyang>
DONE! WHOOO! 

[16:34:02.0000] <annevk42>
ezyang, does it work? :)

[16:34:03.0000] <annevk42>
/me might use a PHP HTML5 parser

[16:34:04.0000] <ezyang>
I haven't run the test suite yet. 

[16:35:00.0000] <ezyang>
For all I know, it crashes and burns with a fatal error :-) 

[16:35:01.0000] <annevk42>
ah, I guess it will :)

[16:35:02.0000] <ezyang>
Also, there must be at least 40 XXX 

[16:35:03.0000] <Hixie>
annevk42: does karma affect comment voting?

[16:37:00.0000] <annevk42>
Hixie, to be honest I haven't studied the algorithms

[16:37:01.0000] <annevk42>
Hixie, maybe it just counts and doesn't take my karma into account when calculating yours

[16:37:02.0000] <Hixie>
as far as i can tell, comment voting has no effect on anything except the troll comment cut off

[16:37:03.0000] <Hixie>
and the karma seems to have no effect at all

[16:38:00.0000] <annevk42>
probably right, would be kind of expensive too now I think about it :)


2009-05-28
[17:05:00.0000] <ezyang>
+6 passes, to 90 passes! 

[17:06:00.0000] <gsnedders>
And how many failures?

[17:08:00.0000] <ezyang>
279 failures, although I know one fix that will get rid of a lot of them. I'm currently squishing exceptions 

[17:15:00.0000] <gsnedders>
Is all your work in Parser.php then?

[17:15:01.0000] <ezyang>
TreeConstructer.php 

[17:15:02.0000] <gsnedders>
Ah, that's what it's called.

[17:15:03.0000] <ezyang>
in the php-tb branch. I'll be merging it back in before pushing 

[17:15:04.0000] <Hixie>
/me writes a sentence of garbled english that he can't work out how to make real english

[17:15:05.0000] <gsnedders>
Can I change the tokenizer to an iterator?

[17:16:00.0000] <Hixie>
this doesn't bode well

[17:16:01.0000] <gsnedders>
Hixie: en-gb-x-hixie isn't real English anyway! :P

[17:16:02.0000] <Hixie>
this sentence is not en-gb-x-hixie either

[17:16:03.0000] <ezyang>
What would the output of the tokenizer be? 

[17:16:04.0000] <ezyang>
(as an iterator) 

[17:16:05.0000] <gsnedders>
ezyang: Tokens :P

[17:16:06.0000] <ezyang>
Uhh, ok 

[17:16:07.0000] <ezyang>
So, problem 

[17:17:00.0000] <gsnedders>
So, solution.

[17:17:01.0000] <ezyang>
Tree construction needs to be able to pass info back to the tokenizer 

[17:17:02.0000] <ezyang>
coroutines would be ideal, but we don't have those in PHP 

[17:17:03.0000] <Hixie>
can someone translate this to english?:

[17:17:04.0000] <Hixie>
Whenever a Document d is added to the worker's Documents, the user agent must, for each port in the list of the worker's ports that is entangled with a second port p whose owner is a WorkerGlobalScope object whose list of the worker's Documents does not contain d, add d to q's WorkerGlobalScope owner's list of the worker's Documents.

[17:18:00.0000] <gsnedders>
foreach($tokinzer as $token) { if ($token['type'] === HTML5_Tokenizer::STARTTAG) $tokenizer->content_model=HTML5_Tokenizer::PCDATA; }

[17:18:01.0000] <gsnedders>
ezyang: ^^

[17:18:02.0000] <gsnedders>
Hixie: Oh dear…

[17:19:00.0000] <ezyang>
That's not sufficient, but sure, if you want to move the twiddling there 

[17:19:01.0000] <gsnedders>
/me is too tired to do such things, so time for me to go read Anna Karenina for five hours, get up at 1pm tomorrow, do no more work…

[17:19:02.0000] <ezyang>
I think it's pseudocode time 

[17:19:03.0000] <gsnedders>
I <3 study leave.

[17:19:04.0000] <ezyang>
That would be ok. 

[17:19:05.0000] <gsnedders>
ezyang: I'll speak with you when I'm sanely awake.

[17:20:00.0000] <ezyang>
Savvy 

[17:25:00.0000] <ezyang>
WHOO! zero exceptions 

[17:26:00.0000] <ezyang>
And fixing those bumped up passes to 103 

[17:26:01.0000] <ezyang>
I think its pushin' time 

[17:27:00.0000] <ezyang>
Time to fix the text blobbing problem 

[17:31:00.0000] <ezyang>
Bumped up to 117 passes 

[17:36:00.0000] <Hixie>
public-html has been strangely silent today

[17:36:01.0000] <Hixie>
did i address all of larry's concerns? surely not.

[17:36:02.0000] <Hixie>
and leif never responded to my last e-mail to him, i wonder if i actually managed to explain what he was asking about

[17:38:00.0000] <annevk2>
Hixie, that worker Document stuff, is that about my question?

[17:38:01.0000] <Hixie>
no

[17:38:02.0000] <Hixie>
though i should fix that at the same tim,e

[17:38:03.0000] <Hixie>
time

[17:46:00.0000] <ezyang>
Ok, that's probably enough html5lib hacking for today 

[17:48:00.0000] <annevk2>
supercool that you're doing this btw

[17:51:00.0000] <Philip`>
No it isn't - it'll remove a reason for PHP programmers to migrate to a proper language, like Python :-p

[17:52:00.0000] <ezyang>
annevk2: Thanks! 

[17:53:00.0000] <ezyang>
Python is still a bitch to get set up for web applications. I do most of my scripting in it these days though 

[18:11:00.0000] <Philip`>
http://dev.opera.com/articles/view/opera-ua-string-changes/

[18:12:00.0000] <Philip`>
The web would be much simpler if there weren't User-Agents strings

[18:12:01.0000] <ezyang>
Sometimes you have to use the user-agent string, since the browser lies about what it supports via feature detection 

[18:43:00.0000] <ezyang>
mmmmmmm 

[18:43:01.0000] <ezyang>
(got tugged back into debugging the brushfires) 

[19:15:00.0000] <ezyang>
So what does the specification *mean* when it says "if it wasn't ignored" 

[19:16:00.0000] <ezyang>
Oh, it's explicitly spelled out. savvy 

[19:17:00.0000] <jcranmer>
ezyang: "if the moon is currently waxing, and Saturn is in conjunction with Jupiter"

[19:43:00.0000] <ezyang>
Quirks mode go boom 

[20:21:00.0000] <othermaciej>
Hixie: well Google totally loves HTML5 it seems

[20:34:00.0000] <ezyang>
"Pop all the nodes from the bottom of the stack of open elements, from the current node up to the root html element." Does this mean that the stack is empty afterwards? 

[20:34:01.0000] <ezyang>
Or does it just contain the html node? 

[21:45:00.0000] <Hixie>
ezyang: wow, that text is confusing. send mail in and i'll fix it. it should say "up to and excluding" the html element.

[21:46:00.0000] <othermaciej>
Hixie: I think my demo of HTML5 video was sexier than Vic's, too bad I probably can't post it publicly

[21:46:01.0000] <othermaciej>
but I have to show you sometime

[21:47:00.0000] <ezyang>
Awesome, thanks! 

[21:48:00.0000] <ezyang>
Umm... does html5 take unsubscribed emails? I unsubscribed because there was too much stuff I didn't care about and I hadn't gotten Sup to filter mailing lists 

[21:48:01.0000] <Hixie>
othermaciej: i am shocked, shocked i tell you, that an apple demo was slicker than a google demo. :-P

[21:48:02.0000] <Hixie>
ezyang: you can just mail me directly (ian⊙hc)

[21:48:03.0000] <othermaciej>
my slides were very pretty too, but not many people showed up to my session

[21:49:00.0000] <othermaciej>
Hixie: I did post my offline demo though: http://webkit.org/demos/calendar/Calendar.html

[21:49:01.0000] <othermaciej>
(uses AppCache, SQLStorate, LocalStorage, online/offline events)

[21:50:00.0000] <Hixie>
othermaciej: that demo will be even more compelling once browsers do the vEvent drag-and-drop stuff

[21:50:01.0000] <Hixie>
othermaciej: and you can just drag from one calendar to another and have the event carry through

[21:50:02.0000] <Hixie>
wouldn't even be that hard to do, on the js side

[21:50:03.0000] <othermaciej>
that would be cool

[21:50:04.0000] <ezyang>
Sent! 

[21:50:05.0000] <Hixie>
it's technically a requirement of html5 now :-)

[21:50:06.0000] <Hixie>
ezyang: thanks!

[21:51:00.0000] <othermaciej>
I did like the YouTube video demo, though really only cause of the domain where it is hosted

[21:51:01.0000] <othermaciej>
not so much intrinsic slickness

[21:51:02.0000] <ezyang>
Hmmm... why am I getting two <html> nodes in my documents... 

[21:53:00.0000] <ezyang>
Ouch, the DOM is coming preloaded with an HTML element 

[21:55:00.0000] <ezyang>
Holy crap, that caused 66 cases to start passing 

[21:55:01.0000] <Hixie>
othermaciej: yeah, people from apple and mozilla have been asking for months now for me to convince the youtube guys to do that, and it's been killing me that i couldn't just say "they've already done it, just wait"

[21:56:00.0000] <othermaciej>
I didn't know Apple was pushing for it, I thought we were gonna make such a demo ourselves

[21:57:00.0000] <Hixie>
i don't think it was a corporate position, but i have received private requests to that effect

[21:57:01.0000] <othermaciej>
it's pretty cool

[21:57:02.0000] <Hixie>
we really need to figure out the codec issue

[21:57:03.0000] <othermaciej>
it's funny that I chose to highlight similar browser features in my WebKit talk to what apparently was highlighted in the keynote

[21:58:00.0000] <othermaciej>
(I demoed video, canvas, offline stuff)

[21:58:01.0000] <Hixie>
i'm surprised offline was demoed anywhere

[21:58:02.0000] <othermaciej>
only thing I showed that Google didn't highlight was CSS animation

[21:58:03.0000] <Hixie>
it's so hard to demo in a compelling fashion

[21:58:04.0000] <othermaciej>
Hixie: that's why I have the network indicator in my demo

[21:58:05.0000] <Hixie>
oh they took out the animation demo?

[21:58:06.0000] <othermaciej>
I pulled the network cable, quit and relaunched

[21:59:00.0000] <othermaciej>
I don't think there was one - if there was I hadn't heard

[21:59:01.0000] <Hixie>
one of the early versions of the keynote i saw had the video being rotated by css animation stuff

[21:59:02.0000] <Hixie>
maybe it wasn't reliably enough or something

[21:59:03.0000] <othermaciej>
I would have demo'd workers too but olliej ran out of time to multithread my fluid dynamics demo

[21:59:04.0000] <othermaciej>
(solves navier-stokes equations in realtime and draws results to canvas)

[22:00:00.0000] <Hixie>
nice

[22:06:00.0000] <othermaciej>
you know, one convenient thing about everything being in one spec prior to the recent splitting was that you could use HTML5 as a shorthand  for "next generation of Web standards"

[22:06:01.0000] <othermaciej>
now you have to be mildly ignorant to do so

[22:06:02.0000] <othermaciej>
http://www.webdesignerhelp.co.uk/index.php/2009/05/5-fantastic-features-of-html-5/

[22:07:00.0000] <othermaciej>
Hixie: anyway I dunno what to do about codecs

[22:07:01.0000] <othermaciej>
I should look up when the patents on a plausible H.264 baseline expire

[22:10:00.0000] <Hixie>
i say we co-opt the term Web 5.0

[22:10:01.0000] <othermaciej>
ES 5 conveniently renumbered themselves!

[22:11:00.0000] <ezyang>
Hmm... when I get around to throwing out ~1/2 of my HTML Purifier codebase to use the html5lib parser, I'll be at version 5.0! 

[22:11:01.0000] <ezyang>
All of the stars are conveniently aligned. 

[22:14:00.0000] <ezyang>
"Remove node from the stack of open elements. " If node isn't the current node, do I excise it from the stack? 

[22:16:00.0000] <ezyang>
That wouldn't make any sense at all. 

[22:16:01.0000] <ezyang>
So I guess I'm supposed to pop the stack until I get to the node 

[22:29:00.0000] <ezyang>
Houston, I think I see an error in the test-case 

[22:29:01.0000] <mpilgrim>
there was a bit in the keynote where vic showed the rotated video demo

[22:30:00.0000] <mpilgrim>
he specifically pointed out that once video was in the DOM, you could do fun stuff with it, including CSS animations and styling

[22:30:01.0000] <othermaciej>
that's cool

[22:30:02.0000] <mpilgrim>
("in the DOM" instead of "in plugin prison")

[22:31:00.0000] <othermaciej>
I guess it must have looked like I planned my stuff based on Google's conference agenda but I seriously had no idea

[22:31:01.0000] <mpilgrim>
but yeah, you had to look really closely to notice that he was switching between two different computers and three different browsers in order to demo this stuff

[22:31:02.0000] <mpilgrim>
the rotating video was a firefox demo (with ogg theora video)

[22:32:00.0000] <mpilgrim>
the youtube html5 demo only works in safari 4 and (apparently) a nightly build of chrome/chromium that i don't have

[22:32:01.0000] <mpilgrim>
and on and on

[22:32:02.0000] <othermaciej>
video is in chromium nightlies now?

[22:32:03.0000] <othermaciej>
like actually working?

[22:32:04.0000] <othermaciej>
that would be sweet if so, I hadn't heard

[22:32:05.0000] <mpilgrim>
http://codereview.chromium.org/115625

[22:33:00.0000] <mpilgrim>
i could be misremembering, he may have shown the youtube demo in safari 4

[22:33:01.0000] <mpilgrim>
he did NOT make any mention of the fact that he kept switching browsers (and OSes)

[22:33:02.0000] <Hixie>
ezyang: it means excise

[22:34:00.0000] <othermaciej>
that checkin would add codecs but presumably you would need something more to actually use them

[22:34:01.0000] <mpilgrim>
that's all i know :(

[22:34:02.0000] <othermaciej>
yeah, the bleeding edge of the Open Web platform can be sharp

[22:34:03.0000] <Hixie>
/me runs his recent chromium build to find out

[22:34:04.0000] <ezyang>
Oh, really! 

[22:34:05.0000] <othermaciej>
I did all my demos in Safari cause I am lazy

[22:34:06.0000] <Hixie>
doesn't work on the mac chromium builds

[22:35:00.0000] <Hixie>
dunno about windows

[22:35:01.0000] <ezyang>
So, that means that <form><div></form><div> results in two nested divs inside of <form>, and nothing outside? 

[22:35:02.0000] <mpilgrim>
i have windows right now

[22:35:03.0000] <mpilgrim>
where can i get nightly builds?

[22:35:04.0000] <Hixie>
i build them from svn

[22:35:05.0000] <Hixie>
ezyang: nested ones, yeah

[22:35:06.0000] <ezyang>
Wow. That's weird. 

[22:35:07.0000] <mpilgrim>
nm, i found it

[22:35:08.0000] <Hixie>
/me points to the topic

[22:36:00.0000] <ezyang>
:-) 

[22:36:01.0000] <mpilgrim>
wow, my hotel wireless doesn't completely suck

[22:37:00.0000] <ezyang>
mpilgrim: Congratulations! A million travelers would love to be in your hotel right now :-P 

[22:37:01.0000] <mpilgrim>
getting sustained 115 kB/s down

[22:37:02.0000] <ezyang>
Maybe that should get a note. 

[22:42:00.0000] <ezyang>
Huh. Google Code's authorization will randomly fail. 

[22:45:00.0000] <mpilgrim>
it's official: i'm on crack

[22:45:01.0000] <mpilgrim>
chromium nightly does not play youtube html5 video

[22:45:02.0000] <mpilgrim>
nor wikimedia commons videos (theora)

[22:46:00.0000] <mpilgrim>
still, someone is obviously working on it

[22:46:01.0000] <mpilgrim>
if it's really built on top of ffmpeg, that would be exciting

[22:46:02.0000] <mpilgrim>
(to me anyway)

[22:46:03.0000] <othermaciej>
I've definitely heard that folks are working on it

[22:46:04.0000] <othermaciej>
what would be specifically exciting about ffmpeg being involved?

[22:46:05.0000] <mpilgrim>
broad codec support

[22:46:06.0000] <mpilgrim>
on all platforms

[22:47:00.0000] <mpilgrim>
including linux, which i run on my primary home workstation

[22:48:00.0000] <mpilgrim>
for posterity, i tested chromium 3.0.183.0 (r17063)

[22:49:00.0000] <mpilgrim>
on windows

[22:49:01.0000] <othermaciej>
Chromium is up to 3 already?

[22:49:02.0000] <mpilgrim>
from http://build.chromium.org/buildbot/snapshots/chromium-rel-xp/17063/

[22:49:03.0000] <mpilgrim>
that's what it says in the about box

[22:49:04.0000] <othermaciej>
at this rate it will have a higher version number than Opera by next year

[22:49:05.0000] <mpilgrim>
lol

[22:56:00.0000] <Hixie>
othermaciej: let's not get crazy here, don't forget opera is also releasing browsers!

[22:57:00.0000] <othermaciej>
Hixie: Opera only consumes 2/3 of a major version number a year though

[22:58:00.0000] <othermaciej>
Chrome is on track to 3 major versions per year or more

[22:58:01.0000] <Hixie>
true

[22:58:02.0000] <Hixie>
i wonder how long they'll keep it up

[22:58:03.0000] <othermaciej>
probably until their number is big enough to sound comparable to other browsers

[22:59:00.0000] <Hixie>
firefox and safari are at 3

[22:59:01.0000] <Hixie>
so that shouldn't take leng

[22:59:02.0000] <Hixie>
lon7g

[22:59:03.0000] <Hixie>
long

[23:00:00.0000] <othermaciej>
I mean, I dunno if they really want to hit IE / Opera level numbers

[23:00:01.0000] <Hixie>
oh, well, catching up to IE wouldn't take long at this rate

[23:00:02.0000] <Hixie>
given the average time between IE releases

[23:01:00.0000] <othermaciej>
heh

[23:01:01.0000] <mpilgrim>
everybody gives us shit for keeping products in perpetual beta

[23:01:02.0000] <mpilgrim>
then we release 2 stable versions in < 1 year, and people give us shit for that instead

[23:02:00.0000] <mpilgrim>
i think i can see the common factor here

[23:02:01.0000] <mpilgrim>
;)

[23:02:02.0000] <Hixie>
hah

[23:04:00.0000] <othermaciej>
I would argue GMail has more right to claim to be a non-beta than Chrome 2

[23:05:00.0000] <ezyang>
Foster parenting is so convoluted 

[23:09:00.0000] <Hixie>
foster parenting has nothing on the AAA

[23:10:00.0000] <ezyang>
AAA? 

[23:11:00.0000] <mpilgrim>
"HTML 5 is an appluadable initiative, however it is doomed to fail. One only needs to look at the features in Silverlight (particularly silverlight 3) to see that the level of features offered by HTML 5 is not enought to meet the needs of future RICH applications."

[23:11:01.0000] <mpilgrim>
http://radar.oreilly.com/2009/05/google-bets-big-on-html-5.html#comment-2060516

[23:13:00.0000] <othermaciej>
Stop HTML5! Keep Web applications poor!

[23:13:01.0000] <Hixie>
i saw that comment earlier

[23:13:02.0000] <othermaciej>
why you gotta be so jealous of *Rich* internet applications, HTML?

[23:13:03.0000] <Hixie>
it's an accurate comment if you assume (as the author does) that the goal is to replace native platforms

[23:14:00.0000] <Hixie>
but the spec explicitly says that that is not the goal

[23:14:01.0000] <Hixie>
mozilla and opera explicitly said, as did microsoft and sun and redhat, at the famous workshop, that doing that a non-starter with a standards-based approach

[23:14:02.0000] <Hixie>
specs don't move fast enough

[23:15:00.0000] <Hixie>
interoperable platforms are doomed to always be inferior to the cutting edge proprietary ones

[23:15:01.0000] <Hixie>
however that misses the entire point

[23:15:02.0000] <Hixie>
which is that they _can_ be good enough

[23:15:03.0000] <Hixie>
and we can make the web much better

[23:15:04.0000] <othermaciej>
If the commentor thinks "do more stuff than Silverlight" is the goal then he's probably right that HTML5 will fail on those terms

[23:16:00.0000] <othermaciej>
but I'm happy with how we're doing on "enable Web developers to do more cool stuff"

[23:16:01.0000] <Hixie>
yup

[23:16:02.0000] <Hixie>
html5 is also doomed to fail if the goal is to cure cancer

[23:19:00.0000] <weinig>
Hixie: with that attitude it is!

[23:20:00.0000] <othermaciej>
hmm validator.nu doesn't like the sizes attribute on this:

[23:20:01.0000] <othermaciej>
<link rel="icon" href="surfin-safari.icns" sizes="32x32 128x128 512x512">

[23:20:02.0000] <othermaciej>
I can't figure out what if anything is wrong

[23:21:00.0000] <othermaciej>
is that actually a bad value or is it a validator bug?

[23:22:00.0000] <othermaciej>
it looks conforming to me....

[23:23:00.0000] <roc>
I wonder how Google is going to address its patent licensing obligations if they distribute ffmpeg under the LGPL

[23:27:00.0000] <mpilgrim>
i know i've beat on rubys in the (recent) past

[23:28:00.0000] <othermaciej>
beat on him about what?

[23:28:01.0000] <mpilgrim>
but i must give him credit for his stewardship of the rdfa discussion

[23:28:02.0000] <othermaciej>
he's been fine on that

[23:28:03.0000] <othermaciej>
I am no fan of his stewardship of the title discussion, in that he started a title discussion

[23:29:00.0000] <mpilgrim>
are these people for real? http://lists.w3.org/Archives/Public/public-html/2009May/0436.html

[23:29:01.0000] <othermaciej>
though it has been uphill from there, now that he and Larry have expressed their actual issues with the spec

[23:29:02.0000] <mpilgrim>
actual_issues++

[23:29:03.0000] <othermaciej>
sideshow_proxy_discussions--

[23:30:00.0000] <mpilgrim>
"if I have a document that is HTML5 with RDFa and parse it, and a similar document that is XHTML + RDFa and parse it, would you expect *exactly* the same triples to be emitted?"

[23:30:01.0000] <mpilgrim>
words fail me

[23:31:00.0000] <othermaciej>
you would kind of expect people interested in semantic markup to want it to work robustly

[23:31:01.0000] <mpilgrim>
also, http://lists.w3.org/Archives/Public/public-html/2009May/0269.html

[23:31:02.0000] <mpilgrim>
i think shane is laboring under the misapprehension that he has done the hard part and is about 99% done

[23:32:00.0000] <othermaciej>
well, he's writing a spec to tack onto HTML4

[23:32:01.0000] <mpilgrim>
and all he has to do with RDFa is "dumb it down" for HTML, drop the "X" from the title, and declare victory

[23:32:02.0000] <mpilgrim>
dude, you're about 1% done

[23:32:03.0000] <othermaciej>
so HTML4-level quality is not surprising as an outcome

[23:32:04.0000] <mpilgrim>
and it was the easy 1%

[23:32:05.0000] <othermaciej>
I would say he's 99% done, and now he has to do the other 99%

[23:33:00.0000] <mpilgrim>
RDFa-in-XHTML is an interesting theoretical exercise

[23:34:00.0000] <mpilgrim>
a toy language in an academic's sandbox

[23:34:01.0000] <billyjackass>
othermaciej: the cause of the sizes thing is a bug in validator.nu, from code I added. I'll fix it.

[23:34:02.0000] <othermaciej>
MikeSmith: thanks!

[23:34:03.0000] <MikeSmith>
thanks for catching it

[23:35:00.0000] <othermaciej>
my only other validator issue is use of proprietary <input type="search"> attributes

[23:36:00.0000] <mpilgrim>
anyway, rubys has been good at knocking some sense into them so far

[23:36:01.0000] <othermaciej>
I wonder if we implement enough of Web Forms 2 to replace that all with standards ones

[23:36:02.0000] <othermaciej>
mpilgrim: he's knocking, but I suspect there is nobody home

[23:36:03.0000] <mpilgrim>
i hope he can manage to get them to understand the magnitude of the task ahead of them

[23:37:00.0000] <othermaciej>
it would have been far easier to consider these issues when designing RDFa in the first place...

[23:37:01.0000] <othermaciej>
but of course, They Weren't Chartered to do That(tm)

[23:37:02.0000] <mpilgrim>
i can't fault them for that, really

[23:37:03.0000] <mpilgrim>
they were told XHTML was the future

[23:37:04.0000] <mpilgrim>
been there, done that

[23:38:00.0000] <mpilgrim>
not everybody got the memo in 2003

[23:38:01.0000] <othermaciej>
I thought flying cars were the future

[23:40:00.0000] <mpilgrim>
http://www.threadless.com/product/63/Damn_Scientists !

[23:41:00.0000] <mpilgrim>
where is my jetpack, dammit!

[23:42:00.0000] <mpilgrim>
man, this is awesome in retrospect: http://ln.hixie.ch/?start=1037595583&count=1

[23:42:01.0000] <mpilgrim>
and http://diveintomark.org/archives/2002/11/21/a_warning_to_others

[23:44:00.0000] <mpilgrim>
"New XHTML version 1.1! Now more hostile than ever!"

[23:44:01.0000] <mpilgrim>
good times

[23:45:00.0000] <hsivonen>
othermaciej: validator bug most likely

[23:45:01.0000] <ezyang>
XHTML 1.1 was nice enough to give us the modularization though... 

[23:45:02.0000] <othermaciej>
hsivonen: MikeSmith said it was and said he'll fix it

[23:46:00.0000] <Hixie>
i wonder if the rdfa guys have noticed that the issues they're battling with are non-issues in the microdata proposal...

[23:47:00.0000] <othermaciej>
I don't know that they are battling those issues so much as trying to sweep them under the carpet

[23:47:01.0000] <othermaciej>
I think the goal is more to show that RDFa is already perfect than to fix it

[23:47:02.0000] <mpilgrim>
MICRODATA IS TEH SUCK!

[23:48:00.0000] <mpilgrim>
U STOLE OUR ATTRIBUTEZ!

[23:48:01.0000] <othermaciej>
whoah, Hixie used the phrase "in the pre-XHTML2 world"

[23:48:02.0000] <Hixie>
it's ok, i gave them back

[23:48:03.0000] <othermaciej>
and it doesn't look like he was being at all sarcastic

[23:48:04.0000] <Hixie>
othermaciej: you think that's bad. see http://ln.hixie.ch/?start=1042630901&count=1

[23:49:00.0000] <hsivonen>
mpilgrim: http://lists.w3.org/Archives/Public/public-html/2009May/0301.html is interesting

[23:49:01.0000] <mpilgrim>
BUT STILL! WE MUST BRING IT UP AT EVERY OPPORTUNITY! IN CASE IT HAPPENS AGAIN!

[23:49:02.0000] <othermaciej>
"Personally, I think XHTML2 is shaping up to be a great specification." - Ian Hickson

[23:49:03.0000] <mpilgrim>
REMEMBER THE @PROPERTY!

[23:49:04.0000] <othermaciej>
I should twitter that

[23:49:05.0000] <Hixie>
i started work on html5 8 months later

[23:50:00.0000] <Hixie>
it's amazing when you realise that in 6 years since then xhtml2 has hardly changed at all

[23:50:01.0000] <mpilgrim>
not true, it grew roles and states

[23:50:02.0000] <othermaciej>
what made you grow disenchanted so quickly?

[23:51:00.0000] <Hixie>
othermaciej: xforms

[23:51:01.0000] <roc>
and can we bottle it and sell it?

[23:51:02.0000] <Hixie>
as you will recall :-)

[23:51:03.0000] <othermaciej>
was it really not obvious before that, that starting with a clean slate of purity was a bad idea?

[23:51:04.0000] <Hixie>
i guess not

[23:51:05.0000] <othermaciej>
still, it is good to know you can change your mind even on major questions of strategy when presented with the evidence

[23:52:00.0000] <othermaciej>
in the nascent days of Safari I initially thought we should base it on Gecko

[23:52:01.0000] <othermaciej>
as long as we are bringing up regrettable past beliefs

[23:53:00.0000] <mpilgrim>
on the subject of embarassing quotes, i like this one: http://lists.w3.org/Archives/Public/www-html/2003Jan/0277.html

[23:54:00.0000] <Hixie>
it's interesting to note that the end of the blog post does acknowledge that a boil-the-oceans strategy will take a long time

[23:55:00.0000] <othermaciej>
so long, it's infinite!

[23:56:00.0000] <Hixie>
nah, if the technology is good enough, you can do it

[23:56:01.0000] <Hixie>
e.g. mac os 9 to mac os x

[23:56:02.0000] <Hixie>
or the iphone

[23:56:03.0000] <Hixie>
or windows to linux

[23:56:04.0000] <Hixie>
etc

[23:57:00.0000] <Hixie>
mpilgrim: i still don't think we should have the style="" attribute.

[23:57:01.0000] <othermaciej>
OS X had to build a lot of bridges

[23:57:02.0000] <othermaciej>
from the original total-ocean-boiling strategy

[23:57:03.0000] <Hixie>
yes, you need to build a lot of bridges to do it

[23:57:04.0000] <othermaciej>
(Classic and Carbon were not part of the original plan)

[23:57:05.0000] <Hixie>
it's far more work than the xhtml2 wg are even remotely considering

[23:57:06.0000] <Hixie>
also xhtml2 isn't enough of an improvement (any more)

[23:57:07.0000] <Hixie>
(if it ever was)

[23:58:00.0000] <mpilgrim>
http://www.zeldman.com/daily/0103c.shtml#inatwist "Nevertheless, we get the feeling that XHTML 2’s most ardent supporters think ordinary designers and developers are bad and stupid and backward and intractable, and that only brute force can deliver the semantic web."

[23:58:01.0000] <othermaciej>
Apple also had the advantage of having a totally vendor-controlled ecosystem

[23:58:02.0000] <othermaciej>
zeldman beats Hixie!

[23:58:03.0000] <mpilgrim>
funny, i still get that feeling

[23:59:00.0000] <mpilgrim>
zeldman wrote that over 6 years ago

[00:00:00.0000] <Hixie>
that was written around the same time as my blog post

[00:00:01.0000] <Hixie>
probably as part of the same "blogosphere meme"

[00:01:00.0000] <hsivonen>
might be interesting to look up my ancient XHTML2 technical feedback to see if it has been addressed

[00:01:01.0000] <hsivonen>
nah

[00:01:02.0000] <hsivonen>
I'll spend my day doing perf profiling instead

[00:02:00.0000] <Hixie>
hsivonen: the wg finally got back to me today on some feedback i sent on xhtml1 in 2003

[00:03:00.0000] <ezyang>
Question: Why doesn't the whitespace in ><div><table><a>foo</a> <tr><td>bar</td> </tr></table></div> get foster parented? 

[00:03:01.0000] <ezyang>
(erm, ignore that stray >) 

[00:03:02.0000] <ezyang>
Doesn't <a>foo</a> trigger the taint? 

[00:03:03.0000] <Hixie>
in the spec, you mean?

[00:04:00.0000] <ezyang>
This example comes from html5lib test suite 

[00:04:01.0000] <ezyang>
I'm trying to decide if I've done something boneheaded before changing the test 

[00:05:00.0000] <hsivonen>
Hixie: cool. better late than never

[00:05:01.0000] <hsivonen>
Hixie: had you already fixed it in XHTML5?

[00:05:02.0000] <Hixie>
ezyang: looks like a bug in the test to me

[00:05:03.0000] <ezyang>
awesome 

[00:06:00.0000] <Hixie>
hsivonen: the feedback was basically "don't allow text/html to be used for xhtml", so yes, we fixed that years ago

[00:06:01.0000] <hsivonen>
ezyang: the test was updated when I thought Hixie would do WebKit-style taintless tables

[00:06:02.0000] <ezyang>
Oh? 

[00:06:03.0000] <Hixie>
hsivonen: their response was "we split xhtml appendix c into a separate note", and my further response was to ask for the editor's draft so i could check their new wording

[00:06:04.0000] <ezyang>
Well, is Hixie going to do taintless tables? 

[00:06:05.0000] <hsivonen>
ezyang: not at the moment. I need to convince him some more

[00:07:00.0000] <Hixie>
iirc i came up with some reason why it wasn't a good idea

[00:07:01.0000] <ezyang>
Heh, ok. 

[00:08:00.0000] <hsivonen>
it's certainly possible to construct cases where it's a bad idea not to have taint

[00:09:00.0000] <hsivonen>
but WebKit seems to get away with not having taint, so...

[00:09:01.0000] <ezyang>
I see; you want to keep things simple. 

[00:10:00.0000] <ezyang>
Well, it's nice to know that it's intentional. I hate it when I implement parts of the spec and the number of passing tests goes down :-) 

[00:10:01.0000] <hsivonen>
Hixie: so far, I haven't seen real-world issues

[00:10:02.0000] <ezyang>
Hmmm... I suppose I should check how many tests the python implementation has passing and check that 

[00:11:00.0000] <ezyang>
This means the lower bound to fails is at least 1... possibly more 

[00:11:01.0000] <hsivonen>
Hixie: I've seen one reftest fail bus so mildly that users wouldn't have noticed if it had been in the wild

[00:11:02.0000] <hsivonen>
s7bus/but/

[00:12:00.0000] <othermaciej>
I love this old Hixie blog post

[00:12:01.0000] <othermaciej>
every line is a treasure

[00:14:00.0000] <hsivonen>
mpilgrim: it seems that HTML5 is hostile in some ways like XHTML 1.1

[00:14:01.0000] <othermaciej>
actually, some of these things would be great to fix if they were practically fixable

[00:14:02.0000] <othermaciej>
like lists in paragraphs

[00:14:03.0000] <hsivonen>
mpilgrim: e.g. name=foo being non-conforming

[00:15:00.0000] <hsivonen>
mpilgrim: whoa! you've had a FOAF profile!

[00:16:00.0000] <hsivonen>
mpilgrim: what made you not have one any longer?

[00:16:01.0000] <mpilgrim>
i did, when they were all the rage

[00:17:00.0000] <mpilgrim>
i had a blogroll and an rss feed once too

[00:17:01.0000] <mpilgrim>
i believe if you go back far enough, i had a scriptingNews feed

[00:17:02.0000] <mpilgrim>
(betcha don't even know what that is)

[00:17:03.0000] <hsivonen>
mpilgrim: I think I've seen something about that

[00:18:00.0000] <ezyang>
/me lost another four test-cases... argh 

[00:18:01.0000] <hsivonen>
/me remembers seeing a YSoD on diveintomark around the time of that post

[00:19:00.0000] <ezyang>
Ok, the test-case is almost definitely wrong 

[00:20:00.0000] <ezyang>
Oh wait, my diff is being wonky 

[00:22:00.0000] <ezyang>
No, this is most certainly bogus 

[00:22:01.0000] <ezyang>
The test case wants me to have <input> be a child of <table> 

[00:23:00.0000] <hsivonen>
ezyang: not "hidden"?

[00:23:01.0000] <ezyang>
Ooh, good point. 

[00:25:00.0000] <ezyang>
Whoo, converted into passes 

[00:29:00.0000] <Hixie>
othermaciej_: pretty much everything in that list is either something we've done (e.g. cite="" is now on <section> and <article>, we have <section>/<h1>, we allow <a href=""> around blocks, etc) or things we want to do but can't (e.g. replace <br> with <l>, have href="" everywhere, have <h> instead of <h1>, etc)

[00:29:01.0000] <ezyang>
hsivonen: How good is our <frameset> coverage? 

[00:32:00.0000] <ezyang>
'A start tag whose tag name is "frameset"' in mode "in body" seems to be a dead zone 

[00:32:01.0000] <mpilgrim>
hixie: you say xhtml 2 hasn't changed, but it has an <img> element now!

[00:32:02.0000] <Hixie>
it's true, it has regressed in some ways

[00:33:00.0000] <Hixie>
:-P

[00:33:01.0000] <mpilgrim>
http://microformats.org/discuss/mail/microformats-discuss/2005-October/001783.html "<img> was reintroduced to XHTML 2, no doubt after a vigorous and healthy debate in which all parties treated each other with mutual respect.  But now it has a different content model than <img> in XHTML 1 and HTML 4, which just goes to show that mutual respect is for chumps."

[00:34:00.0000] <mpilgrim>
the bit about the content model is still true, AFAICT: http://www.w3.org/TR/xhtml2/mod-image.html

[00:36:00.0000] <Hixie>
yes

[00:37:00.0000] <mpilgrim>
ironically, i actually like the idea of using the content inside <img></img> as fallback

[00:38:00.0000] <mpilgrim>
bummer about the whole legacy content thing

[00:40:00.0000] <zcorpan>
how annoying - opera's UA string *was* pretty good

[00:40:01.0000] <Hixie>
mpilgrim: oh there are lots of things that suck in html that i'd do differently if we could

[00:49:00.0000] <Mrmil>
Hi, can <time> be used this way? <time>May 28 2009 08:48:44</time> I remember there was some fuss about this. I know it's useless without the datetime attribute but I don't have to use everywhere, do I? :)

[00:51:00.0000] <Hixie>
Mrmil: either you have to have the attribute or the contents have to be YYYY-MM-DDTHH:MM.ssZ

[00:53:00.0000] <Mrmil>
Mmm, okey. The spec says something else though: http://www.whatwg.org/specs/web-apps/current-work/multipage/text-level-semantics.html#the-time-element - The time element is not necessary for encoding dates or times. In the following snippet, the time is encoded using time, so that it can be restyled (e.g. using XBL2) to match local conventions, while the year is not marked up at all, since marking it up would not be particularly useful.

[00:53:01.0000] <Mrmil>
<p>I usually have a snack at <time>16:00</time>.</p>

[00:58:00.0000] <Hixie>
Mrmil: oh, i misspoke. it's actually either YYYY-MM-DDTHH:MM:SS.ssZ or YYYY-MM-DD or HH:MM:SS.ss

[00:58:01.0000] <Hixie>
and the :SS.ss part is optional

[01:00:00.0000] <Mrmil>
Hixie: Oh okey. Does it mean the example with <time>16:00</time> is wrong?

[01:01:00.0000] <Hixie>
no, HH:MM is a valid syntax (since the :SS.ss part is optional)

[01:01:01.0000] <Mrmil>
right, now I mislooked. Ok, thanks

[01:04:00.0000] <ezyang>
"<body><frameset><frame></frameset>" This wouldn't result in <body> getting replaced with <frameset>, would it? 

[01:06:00.0000] <ezyang>
because <body> sets the frame-ok flag to false, and then frameset doesn't get to do the wibbly-wobbly timey wimey stuff 

[01:07:00.0000] <hsivonen>
ezyang: frameset test coverage is very, very bad

[01:08:00.0000] <ezyang>
Well, I've had about a 50/50 chance of detecting bad tests tonight, so it could be just me 

[01:09:00.0000] <ezyang>
hsivonen: Should I go ahead and "fix" the test, or send mail to html5lib-discuss and wait for an ack? 

[01:09:01.0000] <hsivonen>
ezyang: Also, it would be great to have tests for trying to enter 'in foreign' when the tree builder is already in 'in foreign'

[01:10:00.0000] <ezyang>
oh ho ho, I haven't even implemented that yet. 

[01:10:01.0000] <hsivonen>
ezyang: you mean the tain thing?

[01:10:02.0000] <ezyang>
no, the frame-ok one I just mentioned 

[01:10:03.0000] <hsivonen>
ezyang: yes, please fix

[01:12:00.0000] <ezyang>
Hmm... I'm probably going to get the errors wrong 

[01:13:00.0000] <ezyang>
Can I add a comment to .dat files? 

[01:14:00.0000] <zcorpan>
Hixie: "xHTML2" - http://ln.hixie.ch/?start=1042630901&count=1 - typo or intentional lowercase x?

[01:14:01.0000] <Hixie>
typo

[01:14:02.0000] <Hixie>
happens in several posts from back then

[01:14:03.0000] <Hixie>
i assume it was a typo, anyway

[01:14:04.0000] <Hixie>
maybe i was making some crazy subtle point that i now forget

[01:14:05.0000] <Hixie>
there's a lot of that in my blog entries

[01:15:00.0000] <zcorpan>
i always get confused about what people mean when people write "xHTML"

[01:17:00.0000] <Mrmil>
That means they don't know whether they use HTML or XHTML :)

[01:21:00.0000] <hsivonen>
Usually "xHTML" is signal for "I don't want to upset the XHTML2 WG by calling the XML serialization of HTML5 XHTML."

[01:21:01.0000] <hsivonen>
ezyang: I think there aren't comments

[01:22:00.0000] <hsivonen>
ezyang: also, the format is very brittle in my harness, so please avoid reordering fields

[01:22:01.0000] <hsivonen>
(I have a streaming impl.)

[01:23:00.0000] <hsivonen>
How often should I tell Shark to sample if I leave it running for an hour?

[01:23:01.0000] <hsivonen>
so that I don't get too much data

[01:33:00.0000] <zcorpan>
hsivonen: maybe in the contexts you see "xHTML", but i'm sure lots of instances of "xHTML" i've seen the author has no idea about HTML5 or the XHTML2 WG

[01:40:00.0000] <hsivonen>
zcorpan: that's possibel

[01:40:01.0000] <Hixie>
crap

[01:40:02.0000] <hsivonen>
zcorpan: did Opera change its UA string again?

[01:40:03.0000] <Hixie>
/me realises a flaw in his recent checkin

[01:40:04.0000] <Dashiva>
hsivonen: Issues with two-digit version numbers

[01:41:00.0000] <hsivonen>
Dashiva: yay

[01:41:01.0000] <zcorpan>
hsivonen: Opera/9.80 ... Version/10.00

[01:41:02.0000] <hsivonen>
UA sniffing FTW!

[01:41:03.0000] <Dashiva>
I suggested just converging towards 10 with an ever-inreasing number of 9.9999...

[01:42:00.0000] <hsivonen>
zcorpan: does it also contain substrings "Gecko" and "WebKit" now that you aren't hard-line about version?

[01:42:01.0000] <zcorpan>
hsivonen: no

[01:43:00.0000] <zcorpan>
hsivonen: http://dev.opera.com/articles/view/opera-ua-string-changes/

[01:45:00.0000] <zcorpan>
maybe we should change name and start over from version 1

[01:46:00.0000] <Dashiva>
Ballett/1.0

[01:46:01.0000] <Dashiva>
-t

[01:46:02.0000] <zcorpan>
(or version 4 or 5 or whatever the web sniffers require to work)

[01:47:00.0000] <hsivonen>
my recent UA string annoyances have been with major sites sniffing for Firefox and blocking or semiblocking Minefield

[01:47:01.0000] <Philip`>
What if you changed the version number to 19.00, so that people who use /(\d)\.\d\d/ to find the version number will not think it's an old version?

[01:48:00.0000] <olliej>
Philip`: i believe safari 2.0.4 was detected as netscape 4 by a number of sites

[01:48:01.0000] <Dashiva>
Philip`: It's usually the first character after Opera/

[01:48:02.0000] <zcorpan>
Philip`: i think some pages use /\/(\d)/

[01:48:03.0000] <Dashiva>
So we could go with 90, 91, etc

[01:49:00.0000] <Philip`>
It'd be much easier if the UA string was structured data, like JSON perhaps, so people didn't have to use regexps

[01:49:01.0000] <mpilgrim>
/me recalls that windows 95 identified itself to applications as "major version 3, minor version 95", for compatibility with stupid windows 3.1 apps

[01:50:00.0000] <hsivonen>
mpilgrim: I recall reading Windows 7 doing something similar

[01:50:01.0000] <mpilgrim>
see, we're just replicating the native application experience!

[01:51:00.0000] <zcorpan>
Maybe we should use Mozilla/5.0 (actually Opera/10.00 and not Gecko/Firefox/WebKit/Safari) Presto/2.2.15

[01:51:01.0000] <Philip`>
http://blogs.msdn.com/oldnewthing/archive/2004/02/13/72476.aspx

[01:52:00.0000] <mpilgrim>
oldnewthing++

[01:53:00.0000] <mpilgrim>
must sleep now. "big surprise" coming in tomorrow's keynote

[01:53:01.0000] <mpilgrim>
i hear all google i/o attendees are getting their own personal robot

[01:54:00.0000] <Hixie>
dude

[01:54:01.0000] <Hixie>
nda!

[01:55:00.0000] <Philip`>
Is it a 30-foot dinosaur robot that eats cars?

[01:55:01.0000] <mpilgrim>
no, it's one of these: http://www.ftrain.com/robot_exclusion_protocol.html

[01:55:02.0000] <Dashiva>
Does it destroy humanity?

[01:55:03.0000] <Philip`>
If not, it hardly seems worth bothering

[01:59:00.0000] <hsivonen>
what was the right OS X incantation to do a resource fork-safe copy of a directory in such a way that symlinks get replaced with concrete copies?

[02:04:00.0000] <Mrmil>
Is it normal to release a new version of a software when the development hasn't been finished yet? I just love to fix things 3 times rather than once.

[02:09:00.0000] <hsivonen>
Mrmil: software development is never finished

[02:11:00.0000] <Mrmil>
hsivonen: talk about fighting with entropy

[02:12:00.0000] <Dashiva>
It's not even entropy, it's a malicious agent

[02:29:00.0000] <MikeSmith>
hsivonen: so <link rel="icon" href="surfin-safari.icns" sizes="32x32"> validates, but <link rel="icon" href="surfin-safari.icns" sizes="32x32 64x64"> does not

[02:29:01.0000] <MikeSmith>
it's not actually allowing a list, only a single token 

[02:31:00.0000] <MikeSmith>
the schema defines the value of sizes as 'list { xsd:string { pattern = "[1-9][0-9]*x[1-9][0-9]*" }'

[02:34:00.0000] <MikeSmith>
maybe a jing bug?

[02:38:00.0000] <roc>
the best way to do version checking is to only provide an API that does IsVersionGreaterThanOrEqualTo(version)

[02:41:00.0000] <annevk42>
actually, I'm speccing sync XHR to not have any side effects either

[02:41:01.0000] <annevk42>
I mean, that scripts cannot run

[02:42:00.0000] <Dashiva>
roc: What's the datatype of version?

[02:43:00.0000] <roc>
I don't think it matters too much

[02:44:00.0000] <roc>
a single int into which you pack major/minor is probably OK

[02:45:00.0000] <Hixie>
annevk42: if sync xhr has no network-based side-effects, it's a pretty useless networking api :-P

[02:46:00.0000] <annevk42>
meh, I mean that when send() is invoked the event loop is taken out of the loop

[02:46:01.0000] <annevk42>
until all is done

[02:46:02.0000] <annevk42>
(or exception is thrown, blah)

[02:46:03.0000] <annevk42>
I'm working on defining this in terms of the event loop

[02:47:00.0000] <annevk42>
It's been very annoying so far though also interesting

[02:49:00.0000] <Hixie>
annevk42: sounds exactly like what implementors have said about it :-)

[02:55:00.0000] <roc>
no, I think it's only annoying

[02:58:00.0000] <annevk42>
I wonder if we could've done it a bit more abstractly

[02:58:01.0000] <annevk42>
Like "run the following steps synchronously" and "queue a task for these steps and run them whenever what is running runs to completion"

[03:02:00.0000] <ap>
annevk42: defining abstract expectations would be more desirable than describing algorithms in pseudo-code, in my opinion

[03:03:00.0000] <ap>
annevk42: one often cannot even compare the pseudo-code with actual implementation - e.g. if the actual implementation of event loop is closed source

[03:04:00.0000] <annevk42>
it also seems mightily hard to write tests against

[03:04:01.0000] <jgraham>
My rather biased view is that making it easy to write tests is more important than having the spec map closely onto the implementation

[03:05:00.0000] <othermaciej>
/me fails at sleeping again

[03:05:01.0000] <annevk42>
othermaciej, I have a similar issue

[03:05:02.0000] <annevk42>
(though today it was because I was woken up early by someone coming by to fix something)

[03:06:00.0000] <othermaciej>
annevk42: are you in a time zone where you are supposed to be asleep?

[03:06:01.0000] <roc>
I don't think anyone's arguing that writing psuedo-code algorithms is good because it maps closely to implementations

[03:06:02.0000] <annevk42>
othermaciej, not anymore :)

[03:06:03.0000] <othermaciej>
sometimes pseudo-code can define things more unambiguously

[03:06:04.0000] <othermaciej>
but it can also be hard to figure out what the consequences are for a practical implementation

[03:06:05.0000] <othermaciej>
Workers is an example of where this goes bad IMO

[03:06:06.0000] <annevk42>
<input iprof>: http://bulknews.typepad.com/blog/2009/05/another-way-to-autofill-your-profile-in-html-forms.html

[03:07:00.0000] <othermaciej>
it's very breezy about concurrency isues

[03:07:01.0000] <othermaciej>
and as a result ends up requiring crazy things for practical implementations

[03:07:02.0000] <othermaciej>
(of course, that's not in the HTML5 spec as such)

[03:08:00.0000] <zcorpan>
/me finds http://www.w3.org/2009/04/video-player.xhtml

[03:08:01.0000] <roc>
to be fair, writing specs for things involving concurrency is very hard no matter how you do it

[03:08:02.0000] <othermaciej>
that is true

[03:09:00.0000] <othermaciej>
it's very easy to require things that are unimplementable or unbelievably hard to implement

[03:09:01.0000] <othermaciej>
it's also easy to state requirements that are unsound and end up having race conditions

[03:09:02.0000] <zcorpan>
i wonder who wrote the video-player.xhtml demo

[03:09:03.0000] <roc>
it's pretty cool

[03:10:00.0000] <Philip`>
http://lists.w3.org/Archives/Public/public-pfwg-comments/2009AprJun/0085.html

[03:10:01.0000] <zcorpan>
thanks Philip`

[03:10:02.0000] <Philip`>
/me thanks Google

[03:11:00.0000] <roc>
I'm glad they're enjoying the video element, I just hope they don't ask us to implement DFXP

[03:13:00.0000] <othermaciej>
video-player.xhtml is indeed neat

[03:13:01.0000] <othermaciej>
what is DFXP?

[03:13:02.0000] <Hixie>
annevk42: "whenever what is running runs to completion" doesn't work because it doesn't define ordering of async events

[03:14:00.0000] <Hixie>
annevk42: and we can't just say that that things fire in the order the fire, because certain things get priority (hence the multiple task sources)

[03:14:01.0000] <MikeSmith>
roc: btw, do you know if there's been any new news about the 'Accelerated 3D on the Web' work at Khronos?

[03:15:00.0000] <Hixie>
othermaciej: btw i just made a number of changes to workers to hopefully reduce that (though that wasn't the main goal of the changes, the main goal was reducing cross-thread reachability tests)

[03:17:00.0000] <zcorpan>
"This will never work on Internet Explorer" - http://www.dailymotion.com/openvideodemo

[03:17:01.0000] <roc>
othermaciej: W3C Timed Text, some giant bloated thing IIRC

[03:17:02.0000] <roc>
MikeSmith: there may be, but I'm not involved

[03:22:00.0000] <annevk42>
Hixie, but the priority is up to the UA

[03:23:00.0000] <annevk42>
Hixie, so we could just say they go in a bucket or some such

[03:23:01.0000] <Hixie>
annevk42: no, because some things have to happen in order

[03:23:02.0000] <Hixie>
relative to each other

[03:23:03.0000] <Hixie>
but not necessarily relative to others

[03:24:00.0000] <annevk42>
hmm yeah

[03:24:01.0000] <Hixie>
also, we have to define that inactive documents don't run their tasks

[03:24:02.0000] <Hixie>
and we have to define when the storage mutex is released

[03:24:03.0000] <Hixie>
and we have to define when rendering happens relative to task processing

[03:25:00.0000] <Hixie>
i just don't know how to do all this without using algorithms

[03:25:01.0000] <annevk42>
we should write this list down somewhere

[03:25:02.0000] <Hixie>
it's all already specified :-)

[03:25:03.0000] <annevk42>
I'll make a wiki page

[03:25:04.0000] <Hixie>
i have just been reading straight out of the html5 event loop description

[03:25:05.0000] <annevk42>
a definition is not a list of requirements

[03:26:00.0000] <Hixie>
the spec already meets these requirements as far as i can tell

[03:26:01.0000] <annevk42>
sure, but this is easier to understand than the spec :)

[03:29:00.0000] <annevk42>
hmm nm

[03:29:01.0000] <annevk42>
writing it down doesn't make it look like anything useful

[03:34:00.0000] <Hixie>
annevk42: just saying "we must define when bla happens" isn't useful, indeed :-)

[03:34:01.0000] <Hixie>
(at least, not once we've defined it)

[03:34:02.0000] <Hixie>
i mean, you probably could come up with some complicated non-algorithmic description

[03:34:03.0000] <Hixie>
but it'd almost certainly be far less clear

[03:34:04.0000] <Hixie>
i haven't been able to come up with clear non-algorithmic descriptions for these things

[03:35:00.0000] <annevk42>
"Google Chrome now supports the video tag."

[03:36:00.0000] <annevk42>
I was hoping for more information than that :)

[03:37:00.0000] <othermaciej_>
last I heard, it did not work when markpilgrim tried it

[03:37:01.0000] <zcorpan>
/me adds the mentioned demos to the spec's annotation box for <video>

[03:38:00.0000] <othermaciej_>
and I don't see any code in the public WebKit tree for a chromium back end for <video>

[03:41:00.0000] <annevk42>
othermaciej_, http://googlechromereleases.blogspot.com/2009/05/dev-channel-release-301822.html

[03:41:01.0000] <annevk42>
othermaciej_, seems they use FFmpeg

[03:49:00.0000] <othermaciej>
I wonder if their video stuff is just not submitted yet

[03:49:01.0000] <annevk42>
It's also not in the actual release notes: http://sites.google.com/a/chromium.org/dev/getting-involved/dev-channel/release-notes/301822

[03:50:00.0000] <annevk42>
oh, but something is in: http://sites.google.com/a/chromium.org/dev/getting-involved/dev-channel/release-notes/201800

[03:52:00.0000] <annevk42>
Hixie, do you know anything about license story and if there's more info somewhere about <video> and Chomium/Chrome?

[03:54:00.0000] <othermaciej>
hmm I see only the faintest trace of a Chromium media engine in the WebKit tree

[03:55:00.0000] <othermaciej>
will have to ask some Chromium folks when they are around

[04:00:00.0000] <hsivonen>
MikeSmith: on the surface, the sizes thing sure looks like a Jing bug

[04:01:00.0000] <hsivonen>
MikeSmith: oops. I think you need to have a quantifier for the datatype

[04:01:01.0000] <hsivonen>
MikeSmith: so not a Jing bug

[04:06:00.0000] <annevk42>
http://www.ffmpeg.org/legal.html makes it look very dodgy

[04:06:01.0000] <annevk42>
guess the next question is whether MPEG-LA is in contact with Google

[04:08:00.0000] <roc>
this is something I am very curious about

[04:09:00.0000] <hsivonen>
MikeSmith: fix being deployed

[04:09:01.0000] <hsivonen>
annevk42: MPEG-LA deal would explain use on server without ever distributing it, but the case of distributing it is interesting

[04:10:00.0000] <hsivonen>
annevk42: (L)GPL considered

[04:10:01.0000] <hsivonen>
annevk42: I can see how they could wiggle around Apache License 2 for Android

[04:11:00.0000] <annevk42>
so what exactly is the deal with the licenses? they're not compatible?

[04:13:00.0000] <jgraham>
Isn't the problem just that the LGPL bits have to remain open source yet there will be a patent licensing fee associated with distributing them. Which causes interesting issues

[04:14:00.0000] <hsivonen>
annevk42: see section 11 of http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html

[04:16:00.0000] <roc>
if Google has arranged H.264 patent licensing for all users of ffmpeg, that would be a neat solution to the codec problem

[04:17:00.0000] <annevk42>
hsivonen, wow

[04:19:00.0000] <jgraham>
roc: I'm not sure how that would work without making the ffmpeg patents worthless

[04:19:01.0000] <roc>
you mean the H.264 patents

[04:19:02.0000] <roc>
but yes

[04:19:03.0000] <Mrmil>
Doesn anybody know how to make HTML 5 work in IE? Bruce Lawson pointed out this script http://remysharp.com/2009/01/07/html5-enabling-script/ but I thought I'd ask here to make sure.

[04:19:04.0000] <roc>
it's hard to see the MPEG LA agreeing to that

[04:19:05.0000] <jgraham>
yeah I did mean that

[04:21:00.0000] <roc>
there are possible alternatives, like Google isn't actually using ffmpeg's H.264 implementation, but it's useless to speculate. We should just wait and see

[04:21:01.0000] <annevk42>
so this basically means all browsers can ship FFmpeg if Google is shipping a complete version

[04:21:02.0000] <annevk42>
fascinating

[04:22:00.0000] <annevk42>
(quite brutal license terms too)

[04:22:01.0000] <jgraham>
It it fun to speculate wildly. Maybe we can start a rumour that Google bought all the patents

[04:22:02.0000] <zcorpan>
then google will sue microsoft when they ship <video> support in IE11

[04:23:00.0000] <othermaciej>
Hixie: is this one of the changes you made for Worker lifecycle?

[04:23:01.0000] <othermaciej>
"User agents must either act as if MessagePort objects have a strong reference to their entangled MessagePort object or as if each MessagePort object's owner has a strong reference to the MessagePort object."

[04:23:02.0000] <roc>
Maybe they replaced the MPEG LA board with Cyclons

[04:23:03.0000] <othermaciej>
(I failed to notice the second clause initially and was writing a reply.)

[04:24:00.0000] <jgraham>
If someone would only twwitter one or more of these things as the "big Keynote announcement"...

[04:28:00.0000] <annevk42>
zcorpan, how could Google do that?

[04:31:00.0000] <zcorpan>
annevk42: no idea, just wild speculation

[04:39:00.0000] <MikeSmith>
hsivonen: thanks

[04:39:01.0000] <zcorpan>
/me wonders whether gsnedders' outliner should be added to http://validator.whatwg.org/

[04:40:00.0000] <MikeSmith>
I didn't realize list{} could even have a quantifier

[04:40:01.0000] <jgraham>
Having an outliner as part of validator.nu would rock

[04:40:02.0000] <MikeSmith>
oh

[04:41:00.0000] <MikeSmith>
hsivonen: sorry, I was confused

[04:41:01.0000] <MikeSmith>
I see the quantifier is on the pattern

[04:41:02.0000] <zcorpan>
jgraham: yes

[04:42:00.0000] <MikeSmith>
hsivonen: I thought list{} by itself meant "one or more"

[04:43:00.0000] <zcorpan>
maybe RELAX NG should be changed to match expectations better

[04:44:00.0000] <zcorpan>
send feedback to OASIS

[04:44:01.0000] <MikeSmith>
heh

[04:45:00.0000] <MikeSmith>
zcorpan: (assuming you meant that sarcastically)

[04:45:01.0000] <annevk42>
hmm, http://src.chromium.org/viewvc/chrome/trunk/deps/third_party/ffmpeg/README.chromium

[04:46:00.0000] <annevk42>
"Example configure flags suitable for Chromium (Ogg+Theora+Vorbis)"

[04:46:01.0000] <zcorpan>
MikeSmith: i wasn't sarcastic

[04:46:02.0000] <MikeSmith>
oh

[04:46:03.0000] <zcorpan>
MikeSmith: relax ng is intended to be free of surprises like these

[04:47:00.0000] <zcorpan>
see http://www.imc.org/ietf-xml-use/mail-archive/msg00217.html

[04:48:00.0000] <MikeSmith>
I see. well, OASIS has not had anything to do with RELAX NG for a long time. And had next to nothing to do with it to begin with. James just picked OASIS for convenience, because he needed somewhere

[04:48:01.0000] <MikeSmith>
and taking it to W3C did not seem to be an option

[04:48:02.0000] <zcorpan>
MikeSmith: ok, send feedback to James :)

[04:48:03.0000] <MikeSmith>
I was actually on the RELAXNG TC when it was formed

[04:49:00.0000] <MikeSmith>
zcorpan: its home is at ISO now, fwiw

[04:49:01.0000] <zcorpan>
MikeSmith: as well as OASIS, i think?

[04:49:02.0000] <MikeSmith>
nominally

[04:49:03.0000] <MikeSmith>
only

[04:51:00.0000] <zcorpan>
http://www.imc.org/ietf-xml-use/mail-archive/msg00247.html

[05:02:00.0000] <hsivonen>
haha. the ISO version is more than "a little" harder to read

[05:02:01.0000] <hsivonen>
the ISO version is only suitable as a reference for people who already know what the spec says

[05:03:00.0000] <myakura>
should the W3C version of HTML5 be a little harder, too?

[05:03:01.0000] <itpastorn>
This discussion might be of interest: http://www.sitepoint.com/blogs/2009/05/26/html-5-now-or-never/

[05:05:00.0000] <hsivonen>
myakura: it's not impossible that that's what's going to happen :-/

[05:05:01.0000] <myakura>
hsivonen: yeah :(

[05:05:02.0000] <hsivonen>
myakura: oh, and it already lacks the scripted readability features

[05:05:03.0000] <hsivonen>
myakura: so we are already there :-(

[05:06:00.0000] <zcorpan>
would be nice with a sitepoint book about canvas (and svg, too, probably, so that people don't use canvas for cases where svg is more appropriate)

[05:08:00.0000] <zcorpan>
"SVG still doesn’t feel like it has arrived on today’s web, especially in mobile browsers; on the other hand, canvas is already available in a handful of mobile and most desktop browsers"

[05:08:01.0000] <zcorpan>
seems not quite correct - i think a number of mobiles gained support for svg before canvas and desktop browsers were implementing svg at the same time as canvas

[05:09:00.0000] <hsivonen>
Aren't both SVG and canvas in Fennec, Mobile Safari and Opera Mini?

[05:09:01.0000] <hsivonen>
zcorpan: which mobile browser that is suitable for browsing the Web had SVG before <canvas>?

[05:09:02.0000] <annevk42>
itpastorn, lots of FUD there

[05:10:00.0000] <zcorpan>
hsivonen: not sure if there was any that was suitable for browsing the web

[05:10:01.0000] <annevk42>
e.g. "Anything can happen before HTML5 becomes a stable candidate recommendation." -- clearly that's true, but for a lot of things it's highly unlikely

[05:11:00.0000] <zcorpan>
hsivonen: maybe at one point opera had a partial svg implementation but no canvas :)

[05:11:01.0000] <annevk42>
when did we add <canvas>?

[05:11:02.0000] <annevk42>
we had SVG in Opera 8

[05:12:00.0000] <hsivonen>
annevk42: I'm pretty sure you didn't on mobile

[05:12:01.0000] <hsivonen>
/me launches Opera 8.65 on S60

[05:12:02.0000] <zcorpan>
annevk2: i think <canvas> was opera 9

[05:13:00.0000] <roc>
/me wonders why on earth someone from Nokia is building Firefox for the Playstation 3

[05:13:01.0000] <annevk42>
prolly because the default browser is crap

[05:13:02.0000] <hsivonen>
roc: why not? the default browser isn't nice

[05:14:00.0000] <hsivonen>
annevk42: no SVG in Opera 8.65 on S60r3.1 on intertwingly.net

[05:14:01.0000] <roc>
what is the default browser?

[05:14:02.0000] <hsivonen>
annevk42: I do see the SVG in Mini

[05:14:03.0000] <hsivonen>
roc: NetFront, I assume

[05:14:04.0000] <jgraham>
Hehe those sitepoint comments are funny

[05:14:05.0000] <jgraham>
Now is a _good_ time for a HTML5 book

[05:14:06.0000] <annevk42>
mkay, might be that it was not good enough for mobiles yet

[05:15:00.0000] <jgraham>
It is no more subject to change than, say, Rails and there is a whole book _industry_ built around that

[05:16:00.0000] <jgraham>
Plus whoever wites  a good book now gets to influence how people think about HTML 5 in the future

[05:16:01.0000] <hsivonen>
annevk42: it could also be that Opera 8.65 sends headers that lead to me getting the wrong mime type

[05:16:02.0000] <othermaciej>
the parts of HTML 5 that are already widely implemented probably won't change very much

[05:16:03.0000] <zcorpan>
"Microformats have a low barrier to entry, build on existing semantics and offer far more useful opportunities today, in 2009, than HTML 5." - i wonder what those useful opportunities are

[05:16:04.0000] <othermaciej>
although it's not until they are widely used that it moves from "unlikely" to "near impossible"

[05:17:00.0000] <othermaciej>
I think only <canvas> is past that threshold of use

[05:17:01.0000] <roc>
I think localStorage is further ahead actually

[05:17:02.0000] <roc>
at least it has IE8 support

[05:17:03.0000] <jgraham>
Doing a book about the features like <canvas>, databases, video, storage, etc. seems quite reasonable

[05:18:00.0000] <hsivonen>
annevk42: SVG not supported even by navigating directly to the tiger

[05:18:01.0000] <jgraham>
/me likes Shelly's comment "It exists whether you’re using HTML5 or not.", without mentioning that the same is true of every other HTML5 feature

[05:18:02.0000] <annevk42>
hsivonen, thanks for testing

[05:19:00.0000] <hsivonen>
/me wonders if Opera for S60 is a discontinued product and whether it's worthwhile to file bugs

[05:19:01.0000] <jgraham>
(it was in relation to <canvas> which I guess she covers in "Paining the Web")

[05:19:02.0000] <othermaciej>
good point

[05:19:03.0000] <gsnedders>
/me still isn't used to seeing the Opera icon in his dock where he's used to seeing Safari

[05:21:00.0000] <jgraham>
gsnedders: Put it somewhere else in the dock then?

[05:21:01.0000] <gsnedders>
Then it isn't where I instinctively click for a web browser :P

[05:21:02.0000] <hsivonen>
annevk42: see SRS60-3775 about request header issues

[05:22:00.0000] <zcorpan>
/me has his browsers side-by-side in the dock

[05:22:01.0000] <gsnedders>
/me is very selective about what goes in his dock

[05:23:00.0000] <jgraham>
/me needs all his browsers in the dock

[05:23:01.0000] <zcorpan>
what do you have? Finder, Opera, Trash?

[05:23:02.0000] <annevk42>
/me almost never uses the dock

[05:23:03.0000] <jgraham>
Although Firefox removes itself for some reason

[05:23:04.0000] <annevk42>
/me always searches for applications and hits enter

[05:23:05.0000] <jgraham>
annevk42: Quicksilver?

[05:23:06.0000] <gsnedders>
zcorpan: Finder, System Preferences, Mail, Adium, Opera, Colloquy, Transmit, SubEthaEdit, iTunes, Terminal, Trash

[05:23:07.0000] <zcorpan>
jgraham: have you checked "Keep in Dock" in the context menu?

[05:23:08.0000] <annevk42>
the search bar in the top right

[05:23:09.0000] <jgraham>
zcorpan: yes

[05:24:00.0000] <zcorpan>
jgraham: weird

[05:24:01.0000] <jgraham>
annevk42: Oh, I'm glad I'm not the only one

[05:24:02.0000] <jgraham>
zcorpan: (I think it is after it updates)

[05:24:03.0000] <zcorpan>
ah

[05:24:04.0000] <gsnedders>
Spotlight is quick enough to be used like that in 10.5

[05:24:05.0000] <hsivonen>
gsnedders: it a kind of FAIL for OS X pref design that users need to keep System Prefs in the Dock

[05:24:06.0000] <zcorpan>
minefield stays in dock for me despite updates

[05:24:07.0000] <hsivonen>
(I have it there, too.)

[05:24:08.0000] <gsnedders>
hsivonen: They don't really at all

[05:25:00.0000] <jgraham>
(In unrelated non-news the applications folder is a really really difficult way to find applications)

[05:25:01.0000] <gsnedders>
hsivonen: It's in the Apple menu :P

[05:25:02.0000] <gsnedders>
jgraham: Um, just start typing the application name to go to that point in the list of files?

[05:25:03.0000] <gsnedders>
I guess that doesn't work if you use icons

[05:25:04.0000] <annevk42>
usually you only need like two or three letters

[05:25:05.0000] <annevk42>
very nice

[05:26:00.0000] <jgraham>
gsnedders: It doesn't organise itself in any nice way. Also I never worked out how to execute things using the keyboard

[05:26:01.0000] <jgraham>
And it is hard to find the folder

[05:26:02.0000] <gsnedders>
jgraham: Cmd+O?

[05:26:03.0000] <jgraham>
(like you have to find a finder window and then navigate to the applications folder)

[05:27:00.0000] <gsnedders>
jgraham: Keep it in the dock?

[05:27:01.0000] <jgraham>
gsnedders: Oh OK. It ill never replace enter in my muscle memory though

[05:28:00.0000] <jgraham>
gsnedders: I could do. But it feels like I am trying to fight the system (it is much easier to just use spotlight for everything)

[05:28:01.0000] <gsnedders>
Indeed.

[05:28:02.0000] <zcorpan>
i wonder why os x uses cmd+o instead of enter

[05:28:03.0000] <gsnedders>
zcorpan: Legacy

[05:29:00.0000] <gsnedders>
(Classic did)

[05:29:01.0000] <gsnedders>
/me wonders what NeXTSTEP did

[05:29:02.0000] <jgraham>
gsnedders: That should so have been changed. It is just insane to have that one part of the UI inconsistent with everything else on the system

[05:30:00.0000] <zcorpan>
i don't see why they can't change it now

[05:30:01.0000] <gsnedders>
jgraham: It is consistent. Cmd+O always opens stuff!

[05:30:02.0000] <zcorpan>
they could have cmd+o and enter both open, and mint a new shortcut for rename

[05:30:03.0000] <jgraham>
gsnedders: Enter always executes actions (except when it renames files)

[05:31:00.0000] <jgraham>
Plus renaming files is a really rare activity and so it makes no sense whatsoever to bind it to the most prominent key on the keyboard

[05:31:01.0000] <zcorpan>
e.g. cmd+enter for rename

[05:31:02.0000] <jgraham>
And Apple have UI people who understand that

[05:32:00.0000] <gsnedders>
Heck, until OS X, cmd+n was new folder

[05:34:00.0000] <othermaciej>
I find "enter" for rename to be kind of weird

[05:34:01.0000] <zcorpan>
another thing that annoys me is that os x is hard to use with just keyboard. e.g. if i get a dialog and want something else than the default button

[05:34:02.0000] <othermaciej>
even after using Mac for 8 years now...

[05:35:00.0000] <annevk42>
yeah, that sucks

[05:35:01.0000] <othermaciej>
zcorpan: if you turn on "full keyboard access" it's easier to use with just the keyboard

[05:35:02.0000] <hsivonen>
IIRC, the enter thing came along after System 6.0.7 and in 7.5 at the latest

[05:35:03.0000] <hsivonen>
othermaciej: easier but still harder than Windows or Gnome

[05:35:04.0000] <zcorpan>
othermaciej: interesting

[05:36:00.0000] <hsivonen>
at one point, I considered swithing to Ubuntu for development in order to get better keyboardability in Eclipse

[05:36:01.0000] <othermaciej>
that makes tab cycle through every control, even ones you don't normally type in

[05:36:02.0000] <othermaciej>
(OS X default is that tab only cycles through controls that are interesting typing targets)

[05:36:03.0000] <jgraham>
othermaciej: It doesn't consider dropdowns to be interesting typing targets, right?

[05:36:04.0000] <gsnedders>
hsivonen: It's in 7.0, I'm pretty sure

[05:37:00.0000] <zcorpan>
othermaciej: thanks

[05:37:01.0000] <hsivonen>
there's something apple got right about keyboardability: in Cancel, OK dialogs, Cancel is focused by default and OK is the default button, so both have keyboard equivs without moving focus

[05:37:02.0000] <othermaciej>
we try to avoid dialogs that literally say [Cancel] [OK]

[05:38:00.0000] <hsivonen>
othermaciej: my keyboard-related point stands, though

[05:38:01.0000] <othermaciej>
but anyway, Esc is always an equivalent for cancel in dialogs on Mac

[05:38:02.0000] <zcorpan>
it's still hard to discover which keys to press if you don't want to tab to where you want

[05:38:03.0000] <othermaciej>
regardless of full keyboard focus

[05:38:04.0000] <Rik|work>
is there a way to go to the last item of a directory in the finder ?

[05:38:05.0000] <zcorpan>
in windows, when i press alt, the shortcut keys get underlined

[05:39:00.0000] <othermaciej>
zcorpan: most apps don't have keyboard shortcuts to focus specific controls

[05:39:01.0000] <othermaciej>
(Safari is the main exception I can think of)

[05:40:00.0000] <hsivonen>
I wonder if I should bother to analyze what pages that are slower with the HTML5 parser have in common...

[05:40:01.0000] <hsivonen>
even if things average out in favor of HTML5

[05:40:02.0000] <hsivonen>
perhaps I should...

[05:40:03.0000] <jgraham>
hsivonen: That sounds very interesting

[05:40:04.0000] <hsivonen>
more immediate benchmarking issues to address, though

[05:41:00.0000] <hsivonen>
for example, www.sun.com is worse with HTML5 but www.spiegel.de is better

[05:41:01.0000] <zcorpan>
othermaciej: some things have shortcuts (like buttons in a dialog) but no indicators as to what they are. also, i have no idea how to go to the menu bar at the top with the keyboard

[05:42:00.0000] <othermaciej>
there is a way, though I am not sure what it is

[05:42:01.0000] <othermaciej>
hsivonen: might be interesting to find if any error handling is kicking in

[05:43:00.0000] <hsivonen>
othermaciej: indeed.

[05:43:01.0000] <othermaciej>
zcorpan: this may help: http://support.apple.com/kb/HT1343

[05:44:00.0000] <hsivonen>
looks like the HTML5 spec is unusually biased towards spending time in selector matching compared to Alexa top sites

[05:44:01.0000] <othermaciej>
it mentions that you can use Ctrl-F2 to focus the menu bar

[05:44:02.0000] <othermaciej>
Ctrl-F3 to focus the dock, which I did not even know was possible

[05:45:00.0000] <othermaciej>
the HTML5 spec uses some horrible selectors that are painful to compute

[05:45:01.0000] <othermaciej>
I recall Hyatt ranting about it a lot

[05:47:00.0000] <zcorpan>
othermaciej: thanks. (i have to press fn too)

[05:47:01.0000] <zcorpan>
wonder if there's a way to flip the fn modifier

[05:48:00.0000] <webben>
Hixie: http://www.whatwg.org/specs/web-apps/current-work/#refsRFC3986 ... that reference seems broke

[05:48:01.0000] <annevk42>
zcorpan, how is Web DOM Core going?

[05:48:02.0000] <othermaciej>
zcorpan: there is

[05:48:03.0000] <webben>
oh i see, missing section, nm :)

[05:48:04.0000] <annevk42>
zcorpan, would be nice to have a more realistic DOM spec to reference in specs

[05:49:00.0000] <zcorpan>
othermaciej: found it

[05:49:01.0000] <othermaciej>
zcorpan: in the "Keyboard" tab in the "Keyboard and Mouse" pane of System Preferences

[05:49:02.0000] <othermaciej>
there you go

[05:49:03.0000] <zcorpan>
thanks :)

[05:50:00.0000] <zcorpan>
annevk2: haven't done much on it lately

[05:53:00.0000] <gsnedders>
hsivonen: Where are up-to-date builds of HTML5 Fx

[05:53:01.0000] <annevk42>
zcorpan, any chance I can convince you to finish it?

[05:54:00.0000] <hsivonen>
gsnedders: not available. sorry.

[05:54:01.0000] <othermaciej>
Web DOM Core is definitely a good idea

[05:54:02.0000] <hsivonen>
gsnedders: I've been stuck with this Tp issue.

[05:54:03.0000] <othermaciej>
wonder what the odds are of Web API WG being willing to publish it

[05:54:04.0000] <othermaciej>
Web DOM Events would also be good

[05:55:00.0000] <gsnedders>
hsivonen: k

[05:55:01.0000] <zcorpan>
annevk42: i'm intending to finish it, it's just priorities

[05:56:00.0000] <zcorpan>
i should also be doing it in some working group, but i blame chaals for that

[05:56:01.0000] <annevk42>
yeah, Web DOM Events would be neat

[05:57:00.0000] <annevk42>
DOM 3 Events seems to be going nowhere :/

[06:01:00.0000] <Dashiva>
Did lastweek change subtitle?

[06:20:00.0000] <gsnedders>
Is it bad my compulsive response to finding a broken website is to create a minimal test-case?

[06:20:01.0000] <annevk42>
it's good if you wanna get employed as QA person :p

[06:20:02.0000] <gsnedders>
:P

[06:22:00.0000] <gsnedders>
/me returns to reporting bug on Opera :P

[06:22:01.0000] <annevk42>
/me finds http://code.google.com/p/chromium/issues/detail?id=4363

[06:26:00.0000] <askara>
hi

[06:34:00.0000] <zcorpan>
itpastorn: "Stuff in the spec that is seeing wide implementation can be seen as very unlikely to be changed in any fundamental way, as is stuff that has been part of the spec for a long time. I.e. <aside> is robust, <hgroup> is not." - i think any feature is not robust until it's implemented

[06:35:00.0000] <zcorpan>
itpastorn: so <aside> is not robust and might well be dropped or changed in response to implementation experience

[06:35:01.0000] <zcorpan>
itpastorn: despite having been unchanged in the spec for years

[06:37:00.0000] <zcorpan>
itpastorn: and actually, even after it has been implemented, it can still change in incompatible ways, if it's just implemented in one browser or not widely deployed yet

[06:37:01.0000] <zcorpan>
itpastorn: e.g. <event-source>

[06:39:00.0000] <itpastorn>
zcorpan: I hear you, but can't respond in detail (In a meeting with Skolverket!)

[06:41:00.0000] <annevk42>
stable ≈ two implementations + reasonable deployment

[07:02:00.0000] <gsnedders>
Hixie: Can I send you a suggestion for one thing I want in Acid4?

[07:02:01.0000] <gsnedders>
Or is doing so futile?

[07:05:00.0000] <zcorpan>
xml core wg process http://www.w3.org/mid/CF83BAA719FD2C439D25CBB1C9D1D3020FC70A20⊙Hppc

[07:13:00.0000] <annevk42>
so bikesheds is actually from http://en.wikipedia.org/wiki/Parkinson's_Law_of_Triviality

[07:20:00.0000] <Mrmil>
html5 is like jquery - write less & do more *bloody happy face*

[07:27:00.0000] <hsivonen>
lack of interop: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/120

[07:28:00.0000] <hsivonen>
Hixie: could you please add the IE-specific header to disable the XSS protection on Live DOM Viewer?

[07:34:00.0000] <krijnh>
Mrmil: totally agreed! :)

[07:35:00.0000] <Mrmil>
krijnh: hehe :) I really look forward to using the new form elements

[07:36:00.0000] <krijnh>
Why don't you already?

[07:36:01.0000] <krijnh>
All those 0.5% Opera users will benefit :)

[07:39:00.0000] <Mrmil>
krijnh: Well, my own blog is very simple and written in ruby on rails, it doesn't have comments and the back-end is also very simple so there was no need for them yet. I'll do it when I get to implement more features :)

[07:46:00.0000] <Lachy>
hsivonen, what negative effects, if any, does sniffing the first 1024 bytes instead of 512 for encoding detection have? Does it affect performance in any significant way?

[07:47:00.0000] <Lachy>
I'm curious, at what point do the negative effects of sniffing more characters outweigh the benefits of doing so?

[07:47:01.0000] <Philip`>
It affects performance significantly if that's two packets rather than one

[07:48:00.0000] <jgraham>
I'm not sure that Philip`'s graph indicates that 512 bytes is past the point of diminishing returns

[07:48:01.0000] <hsivonen>
Lachy: I haven't measured the effect, but intuitively, what Philip` says makes sense

[07:48:02.0000] <Philip`>
Someone should measure the average HTTP response header size

[07:48:03.0000] <jgraham>
It indicates a knee in the curve before that but it's a totally different thing

[07:48:04.0000] <Lachy>
sure, but if the algorithm can stop as soon as it's found the encoding, then in ~80% of the cases it would stop before 512 bytes anyway, wouldn't it?

[07:48:05.0000] <Philip`>
and subtract from the average size of a TCP packet

[07:49:00.0000] <jgraham>
because the point of diminishing returns only comes when adding more bytes has a disadvantage that outweighs the advantage

[07:50:00.0000] <jgraham>
I would guess that having to switch the character encoding for 1 page in 5 would be rather expensive

[07:50:01.0000] <jgraham>
1 page in 10 (512 bytes) might give better average performance

[07:50:02.0000] <jgraham>
s/512/1024/

[07:52:00.0000] <ap>
/me hopes that WebKit will never have to implement that evil encoding switching thing

[07:52:01.0000] <Lachy>
I guess those stats from phillip are a little misleading anyway, cause they say nothing about the popularity of the sites affected.  I mean, if the extra 10% of pages the get sniffed properly, are pages that virtually no-one ever looks at, then it's not really much benefit in supporting them better

[07:53:00.0000] <ap>
pages that only people with matching default encoding look at are also unimportant for this comparison

[07:54:00.0000] <jgraham>
Lachy: Actually the biggest problem with them is that they are not from a sufficiently wide range of locales. The curve could look entirely different for .jp sites wwhich are far more likely to have a non-default encoding

[07:54:01.0000] <Lachy>
yeah, that's also a problem

[07:54:02.0000] <hsivonen>
ap: whoa! what do you do instead of the evil late-switch?

[07:54:03.0000] <hsivonen>
ap: You'll have to implement it in order to conform unless you convince Hixie to take it out of the spec

[07:55:00.0000] <ap>
hsivonen: nothing - if there is no charset in http headers, <head> or 1024 bytes, we just use the default encoding

[07:55:01.0000] <hsivonen>
ap: that's daring

[07:55:02.0000] <jgraham>
/me thinks the "evil late switch" is rather good

[07:56:00.0000] <ap>
hsivonen: users don't complain

[07:56:01.0000] <Philip`>
Hmm, the most common HTTP header size seems to be around 240 bytes, and a lot are 120 to ~700 bytes, and not many are above 1000 bytes

[07:56:02.0000] <hsivonen>
ap: very interesting

[07:56:03.0000] <annevk42>
and cool

[07:56:04.0000] <hsivonen>
...now that I've already toiled away implementing the evil switch

[07:56:05.0000] <jgraham>
priority of constituencies: it makes no sense to authors or users if moving the encoding decleration around a bit changes the behaviour

[07:56:06.0000] <jgraham>
Even if it is harder to implement

[07:56:07.0000] <ap>
hsivonen: well, I have one or two examples of sites that would benefit from this, but over the years, that amounts to "don't complain"

[07:57:00.0000] <hsivonen>
jgraham: with the evil switch, it makes things different if you have scripts with side effects

[07:57:01.0000] <annevk42>
jgraham, pages reloading is bad

[07:57:02.0000] <Lachy>
Philip`, is there anyway you could divide your stats up into separate graphs based on the detected encodings and/or page languages so we get a slightly better representation of different locales?

[07:57:03.0000] <annevk42>
jgraham, and authors can figure out what went wrong easily enough

[07:57:04.0000] <jgraham>
ap: Are you sure that users aren't just so used to seeing wrong character encoding that they don't complain

[07:58:00.0000] <jgraham>
I would never report a bug with character encoding becuase I'm so used to seeing errors

[07:58:01.0000] <Philip`>
Lachy: I think I don't have enough data for that to be meaningful

[07:58:02.0000] <ap>
jgraham: I do get bug reports about other encoding problems a plenty (less so, now that our code is more compliant)

[07:58:03.0000] <Lachy>
ok

[07:58:04.0000] <Philip`>
Lachy: and it wouldn't be trivial to do

[07:59:00.0000] <Philip`>
Lachy: so it's unlikely that I'll bother to do it

[07:59:01.0000] <hsivonen>
does IE 5.5 mode have the evil late switch?

[07:59:02.0000] <jgraham>
annevk42: I agree page reloading isn't that nice

[07:59:03.0000] <Philip`>
Lachy: (See http://philip.html5.org/data/charsets.html#usage-frequencies - there's very few that aren't utf-8 or iso-8859-1)

[08:00:00.0000] <hsivonen>
Philip`: your sample may be biased

[08:00:01.0000] <Philip`>
hsivonen: "may"? :-)

[08:00:02.0000] <jgraham>
hsivonen: No, his sample is biased. It is just a case of knowing what the biases are and how they are significant :)

[08:00:03.0000] <jgraham>
(in this case I think they are very significant)

[08:00:04.0000] <Philip`>
/me tried to explicitly mention the bias when mentioning the data on public-html

[08:01:00.0000] <Philip`>
jgraham: Why would the chosen encoding significantly affect where the author puts the encoding declaration in their page?

[08:02:00.0000] <Philip`>
(...assuming that's the bias you think is significant)

[08:03:00.0000] <ap>
Philip`: I would be interested to know not just the number of bytes necessary to read to find charset declaration, but the number of bytes outside <head>

[08:04:00.0000] <ap>
Philip`: WebKit encoding sniffer always looks for meta charset in <head>, even if that's huge

[08:04:01.0000] <jgraham>
Philip`: It is possible that languages with more bytes per character will, on avaerge, have more bytes before the encoding compared to one byte per character languages. Or less if the langauge is more compact

[08:04:02.0000] <hsivonen>
ap: I'm not convinced that having to track <head> is better than the evil switch

[08:04:03.0000] <jgraham>
Assuming that relative to the source tree, the enmcoding decleration comes in roughly the same place

[08:04:04.0000] <Philip`>
ap: Hmm, I don't have any data like that - I was just using the HTML5 pre-parse encoding sniffing algorithm, which ignores tags entirely (except when they're <meta>)

[08:05:00.0000] <Philip`>
ap: and it seems complex to try to record the location of </head>

[08:05:01.0000] <jgraham>
Philip`: I would be more interested in knowing how often a charset decleration comes after a <script>

[08:05:02.0000] <Philip`>
ap: and I'm lazy :-p

[08:05:03.0000] <jgraham>
But not in the first N bytes

[08:06:00.0000] <jgraham>
(since script side effects seem to be the major objection to the reparse)

[08:06:01.0000] <ap>
hsivonen, Philip`: any place in <head> used to be valid for <meta> in HTML4

[08:06:02.0000] <hsivonen>
ap: HTML5 looks everywhere

[08:07:00.0000] <Philip`>
jgraham: Hmm, I guess the effect of multi-byte encodings would be mitigated by all the ASCII for doctypes and tag names that dominates the starts of documents, but I have no idea how much effect it would still have

[08:07:01.0000] <gsnedders>
So, anything interesting been happening on the lists?

[08:09:00.0000] <Philip`>
gsnedders: Yes

[08:10:00.0000] <ap>
hsivonen: as far as I can tell, it's only valid in head, no?

[08:10:01.0000] <hsivonen>
ap: yes

[08:10:02.0000] <gsnedders>
ap: yes

[08:11:00.0000] <annevk42>
since WebKit doesn't restrict itself to the first 1024 bytes I guess it doesn't really matter

[08:23:00.0000] <gsnedders>
/me is amazed at the lack of flame wars

[08:23:01.0000] <gsnedders>
Bikesheds are indeed being painted, but no flames.

[08:26:00.0000] <hsivonen>
flamelessness is good

[08:26:01.0000] <gsnedders>
Indeed.

[08:37:00.0000] <Philip`>
If the bikeshed burns down then nobody will care what colour it was

[08:40:00.0000] <annevk2>
gsnedders, http://krijnhoetmer.nl/irc-logs/whatwg/20070829#l-229 is at least one technorati reference

[08:40:01.0000] <gsnedders>
I was guessing it would be kingryan and I was trying to find stuff on the lists from him

[08:41:00.0000] <gsnedders>
annevk2: Should I send that to the list?

[08:44:00.0000] <gsnedders>
/me sends it anyway

[08:56:00.0000] <theMadness>
annevk2, nice blog post, you got me excited about video again.

[08:58:00.0000] <theMadness>
The irony, http://www.youtube.com/html5 doesn't work on Chrome. :)

[09:09:00.0000] <annevk42>
theMadness, yeah, it's a bit of a mess still

[09:09:01.0000] <annevk42>
theMadness, we'll get there :)

[09:14:00.0000] <theMadness>
I'm sure you will, still the irony is there :P :D

[09:18:00.0000] <hsivonen>
gsnedders: new HTML5 builds starting to appear: https://build.mozilla.org/tryserver-builds/hsivonen⊙mc/

[09:19:00.0000] <gsnedders>
hsivonen: thx

[09:39:00.0000] <jgraham>
Does anyone have any idea what http://blog.typekit.com/ actualy does?

[09:42:00.0000] <Philip`>
No

[09:45:00.0000] <gsnedders>
jgraham: " Just add a line of JavaScript to your markup, tell us what fonts you want to use"

[09:46:00.0000] <jgraham>
gsnedders: I can already add a line of Javascript to my markup to tell you want fonts I want to use. Nothing will happen of course.

[09:48:00.0000] <Philip`>
http://www.flickr.com/photos/veen/3572372312/sizes/o/ - according to step 3, it involves string instruments

[10:14:00.0000] <krijnh>
/me starts turning evil on /irc-logs/

[10:16:00.0000] <hsivonen>
what's the deal with these JS-requiring font systems when you can do it in CSS only?

[10:17:00.0000] <Rik|work>
hsivonen: this might be a way to DRMised them

[10:17:01.0000] <Philip`>
It saves people from having to copy-and-paste giant chunks of CSS that they're likely to get wrong and that will prevent the service provider from updating them

[10:17:02.0000] <hsivonen>
Philip`: isn't that what @import is for?

[10:18:00.0000] <Philip`>
hsivonen: Hmm, good point

[10:18:01.0000] <hsivonen>
@import has nicer parallelism properties than <script>

[10:19:00.0000] <Philip`>
though maybe they want the flexibility to do more client-side work (e.g. testing the UA string, verifying document.domain, etc)

[10:19:01.0000] <Rik|work>
hsivonen: not really

[10:20:00.0000] <hsivonen>
Rik|work: how do you mean? @import blocks style computation and layout but not HTML parsing.

[10:21:00.0000] <Rik|work>
scripts are not blocking html parsing in some browsers now

[10:21:01.0000] <hsivonen>
Rik|work: what browsers now do proper speculative parsing instead of just prescanning for other downloadables?

[10:22:00.0000] <Rik|work>
really don't know that level of details

[10:28:00.0000] <beowulf>
it may be a shorter way to get the weight and font you want

[10:29:00.0000] <beowulf>
it's a lot of css to get a font in the two formats and then set the weight

[10:29:01.0000] <Philip`>
It might automatically scan your page to find what characters you use that font for, so it can generate an efficient subset

[11:51:00.0000] <ezyang>
Sighhh... the Python implementation has 80 fails on the test-suite. As I get closer and close to full implementation, bad test cases are really going to waste my time :-( 

[12:04:00.0000] <ezyang>
Why doesn't this work? python test_parser.py TestCase.test_tests7_7_simpletree 

[12:11:00.0000] <Philip`>
ezyang: Feel free to fix bugs in the Python implementation ;-)

[12:13:00.0000] <ezyang>
Yeah, but I have to triage each failure and figure out exactly what's going on before making a call 

[12:14:00.0000] <hsivonen>
ezyang: the tests should be mostly correct per spec except the table taint thing

[12:14:01.0000] <ezyang>
Ok. 

[12:14:02.0000] <ezyang>
Is there any way to make test_parser.py run a single test? 

[12:14:03.0000] <ezyang>
The obvious command is failing for me 

[12:14:04.0000] <Philip`>
You could delete all the other tests, and then run it

[12:14:05.0000] <Philip`>
/me doesn't know of any other mechanism for that

[12:15:00.0000] <ezyang>
That's... not sustainable 

[12:15:01.0000] <hsivonen>
I nkwo nothing about the Python harness

[12:15:02.0000] <ezyang>
kk 

[12:15:03.0000] <hsivonen>
I make my Java harness run a single-test file first and I copy and paste the test of interest there

[12:15:04.0000] <ezyang>
So... what is the implementation that's closest to being correct? 

[12:16:00.0000] <Philip`>
We did have excellent interoperability between HTML5 parsers some time last year

[12:16:01.0000] <hsivonen>
ezyang: I haven't evaluated wakaba's perl impl., but I believe the V.nu impl. is closer to spec than html5lib at the moment

[12:16:02.0000] <Philip`>
so I blame Hixie for making them all break

[12:16:03.0000] <ezyang>
Ok, so I should probably grab a copy of that. 

[12:16:04.0000] <hsivonen>
ezyang: the main failure in the V.nu parser isn't visible in non-browser environments

[12:17:00.0000] <hsivonen>
I've been postponing fixes until I can figure out this Tp thing.

[12:17:01.0000] <ezyang>
not visible in non-browser environments? 

[12:18:00.0000] <hsivonen>
ezyang: not visible in environment where scripts can't run interleaved with the parser

[12:18:01.0000] <ezyang>
Sorry, it's unclear to me what "visibility" means 

[12:19:00.0000] <ezyang>
Hmm... what should JAVA_HOME be? 

[12:19:01.0000] <hsivonen>
ezyang: the spec violation isn't detectable without a script running during the parse

[12:20:00.0000] <hsivonen>
ezyang: on Mac, /Library/Java/Home

[12:20:01.0000] <ezyang>
Ouch. 

[12:20:02.0000] <ezyang>
(I'm on Intrepid) 

[12:20:03.0000] <ezyang>
So... you need a JavaScript parser 

[12:20:04.0000] <hsivonen>
ezyang: locate javac

[12:20:05.0000] <ezyang>
That's batty 

[12:20:06.0000] <hsivonen>
then remove the bin/javac part

[12:20:07.0000] <hsivonen>
and put the prefix in JAVA_HOME

[12:21:00.0000] <ezyang>
Looks like I don't have javac installed 

[12:21:01.0000] <ezyang>
Do you recommend openjdk or Sun's jdk? 

[12:21:02.0000] <hsivonen>
openjdk

[12:21:03.0000] <ezyang>
So... does Java have any embedded JavaScript engines? 

[12:22:00.0000] <ezyang>
And how are you going to deal with malicious input? 

[12:22:01.0000] <hsivonen>
ezyang: Java does, but I haven't integrated the parser with it

[12:22:02.0000] <hsivonen>
ezyang: so it doesn't count as a bug in the Java version currently but it counts as a bug in the C++ version

[12:23:00.0000] <hsivonen>
(the bug being that the parser still clones formatting element nodes instead of creating new ones)

[12:23:01.0000] <hsivonen>
ezyang: /usr/lib/jvm/java-6-openjdk on Ubuntu

[12:24:00.0000] <hsivonen>
ezyang: if you only want the parser instead of the whole validator, it should build if you stick the dependency jars and the sources to Eclipse's default builder

[12:24:01.0000] <hsivonen>
(or to vanilla javac)

[12:25:00.0000] <ezyang>
I need the tree-building and test-suite, but I don't mind if I build the whole thing 

[12:25:01.0000] <hsivonen>
ok

[12:25:02.0000] <ezyang>
In the meaantime, what's the 2nd closest to compliant implementation? 

[12:26:00.0000] <hsivonen>
I don't know

[12:27:00.0000] <hsivonen>
/htmlparser/test-src/nu/validator/htmlparser/test/TokenizerTester.java and /htmlparser/test-src/nu/validator/htmlparser/test/TreeTester.java are the test harnesses

[12:27:01.0000] <hsivonen>
they take file paths to html5lib-format test data files as arguments

[12:27:02.0000] <ezyang>
Great 

[12:27:03.0000] <ezyang>
Oh no! build fail 

[12:28:00.0000] <ezyang>
"java.lang.ClassCastException: org.xml.sax.InputSource cannot be cast to nu.validator.xml.TypedInputSource" 

[12:28:01.0000] <hsivonen>
ezyang: that's not a build fail. that's a run fail

[12:28:02.0000] <hsivonen>
ezyang: something went wrong with the schema setup

[12:28:03.0000] <ezyang>
well, it happened while I was running build.py! 

[12:28:04.0000] <ezyang>
Shall I blow away my copy and try it again? 

[12:28:05.0000] <hsivonen>
ezyang: you can ignore that and look at the parser.

[12:28:06.0000] <hsivonen>
it has been built by the time that error can happen

[12:29:00.0000] <hsivonen>
ezyang: build.py is also used for running the validator

[12:29:01.0000] <hsivonen>
'all' builds it and then starts the whole thing

[12:31:00.0000] <ezyang>
I fail at running Java 

[12:31:01.0000] <ezyang>
Do I just `java htmlparser/test-src/nu/validator/htmlparser/test/TokenizerTester.java`? 

[12:31:02.0000] <hsivonen>
ezyang: no. I'll look up the incantations

[12:31:03.0000] <ezyang>
heh 

[12:32:00.0000] <hsivonen>
argh. build.py doesn't build the test harnesses

[12:33:00.0000] <hsivonen>
hmm. I don't even remember how to do this without Eclipse with two java source dirs...

[12:33:01.0000] <ezyang>
I'm trying python build.py test 

[12:33:02.0000] <hsivonen>
different 'test'

[12:33:03.0000] <ezyang>
Ok. 

[12:34:00.0000] <hsivonen>
I guess it's easiest to build the tests and the parser separately

[12:34:01.0000] <hsivonen>
so under your checkout dir you now have jars/

[12:34:02.0000] <hsivonen>
and there you have htmlparser.jar

[12:35:00.0000] <hsivonen>
you need to cd to htmlparser/test-src/

[12:35:01.0000] <hsivonen>
and run

[12:38:00.0000] <hsivonen>
scratch run

[12:39:00.0000] <hsivonen>
you need to create a target dir, say classes as sibling to test-src

[12:39:01.0000] <hsivonen>
then, in test-src, run

[12:39:02.0000] <hsivonen>
javac -cp .:../../jars/htmlparser.jar:../../dependencies/jsontools-core-1.5.jar:../../dependencies/antlr.jar -d ../classes/ nu/validator/htmlparser/test/TreeTester.java nu/validator/htmlparser/test/TokenizerTester.java

[12:39:03.0000] <hsivonen>
javac is such a pain

[12:39:04.0000] <hsivonen>
after which you can do

[12:40:00.0000] <hsivonen>
java -cp .:../../jars/htmlparser.jar:../../dependencies/jsontools-core-1.5.jar:../../dependencies/antlr.jar:../classes/ nu.validator.htmlparser.test.TreeTester html5lib source files

[12:40:01.0000] <hsivonen>
s/source files/test files/

[12:44:00.0000] <ezyang>
../classes/nu/validator/htmlparser/test doesn't seem to exist? 

[12:44:01.0000] <ezyang>
Also, classes already exists as a sibling to test-src 

[12:45:00.0000] <hsivonen>
ezyang: ../classes/nu/validator/htmlparser/test should have emerged as a result of the javac... command above

[12:46:00.0000] <hsivonen>
ezyang: did you copy and paste from above or retype?

[12:47:00.0000] <ezyang>
With the first command I get: "javac: directory not found: ../classes/nu/validator/htmlparser/test/TreeTester.java" 

[12:47:01.0000] <ezyang>
Copypasta 

[12:48:00.0000] <hsivonen>
ezyang: there should be a space after ../classes/

[12:48:01.0000] <hsivonen>
I'm guessing your terminal line wrapped just there?

[12:48:02.0000] <ezyang>
Yep 

[12:49:00.0000] <ezyang>
Now I get a number of "cannot find symbol" errors 

[12:49:01.0000] <ezyang>
Which implies that the dependencies weren't setup properly 

[12:50:00.0000] <ezyang>
Namely, these packages don't exist: "com.ibm.icu.text.UnicodeSet" etc 

[12:50:01.0000] <hsivonen>
hmm. I wonder why I didn't get that

[12:50:02.0000] <hsivonen>
to remedy

[12:51:00.0000] <hsivonen>
add ../../dependencies/icu4j-4_0.jar to the -cp stuff separated with colon from the other paths

[12:53:00.0000] <ezyang>
Second command, I get: FileNotFoundException: html5lib (No such file or directory) 

[12:54:00.0000] <hsivonen>
ezyang: replace "html5lib  source files

[12:54:01.0000] <hsivonen>
" with the paths of the actual html5lib tree test files

[12:54:02.0000] <ezyang>
haha, ok 

[12:55:00.0000] <ezyang>
Awesome 

[12:55:01.0000] <ezyang>
We have lift-off 

[12:55:02.0000] <hsivonen>
great

[12:59:00.0000] <ezyang>
Ok, one fail in tests1.dat, which is from the fact that I just changed that 

[12:59:01.0000] <ezyang>
Which makes me even more skeptical that my change was actually correct 

[13:00:00.0000] <hsivonen>
was it on the topic of frameset-ok or something else?

[13:01:00.0000] <ezyang>
Yep 

[13:01:01.0000] <hsivonen>
ezyang: could be bug in V.nu parser

[13:02:00.0000] <hsivonen>
ezyang: I didn't implement frameset-ok as a flag but as a mode

[13:02:01.0000] <ezyang>
Hmm... yeah, that'd probalby do it 

[13:02:02.0000] <hsivonen>
I may have goofed in my tranformation of the algorithm

[13:02:03.0000] <ezyang>
*probably 

[13:02:04.0000] <ezyang>
Because frameset-ok doesn't really make sense as a mode 

[13:03:00.0000] <hsivonen>
oh?

[13:03:01.0000] <ezyang>
I mean, there are a lot of places it can get set 

[13:04:00.0000] <ezyang>
And you can very easily go back to a previous mode, but have the frameset ok be the only thing that's different 

[13:04:01.0000] <ezyang>
It's not obvious to me, without recopying all of the modes, how a mode frameset-ok would work 

[13:05:00.0000] <hsivonen>
ezyang: except for frameset itself, it either just forwards to 'in body' or sets the mode to 'in body' and then moves on to 'in body'

[13:06:00.0000] <hsivonen>
It's past 22:00 here. I should head home.

[13:08:00.0000] <ezyang>
cya 

[13:08:01.0000] <hsivonen>
later

[13:36:00.0000] <ezyang>
/me is awesome (fixed Python bug by changing func.__name__) 

[13:56:00.0000] <Hixie>
http://www.grabup.com/uploads/240ccede5360b093dbf298f8946025a5.png is fantastic

[13:56:01.0000] <Hixie>
(notice the lower third text)

[13:57:00.0000] <ezyang>
Hahahahha 

[13:58:00.0000] <gsnedders>
Uh, OK.

[13:58:01.0000] <gsnedders>
Hixie: Can I email you suggestion for Acid4?

[13:59:00.0000] <gsnedders>
Or is there no point?

[14:00:00.0000] <mgrdcm>
FOX Business Channel:  Most popular newspaper after The Tribune Company and Amazon's Kindle.

[14:00:01.0000] <Hixie>
gsnedders: no point yet

[14:18:00.0000] <jgraham>
ezyang: BTW the python svgmathml branch is likely more spec compliant than the trunk

[14:19:00.0000] <ezyang>
jgraham: Oh really? 

[14:19:01.0000] <jgraham>
Although I think it doen't have the latest AAA changes

[14:19:02.0000] <ezyang>
Is this, beyond just the foreign content tests? 

[14:19:03.0000] <jgraham>
ezyang: Yeah, I did a bunch of work on that branch bu I haven't finished the tedious task of converting all the treewalker stuff to deal with namespaces

[14:20:00.0000] <jgraham>
So I haven't merged

[14:20:01.0000] <jgraham>
(actually I haven't merged because the merge was non-trivial)

[14:20:02.0000] <jgraham>
(so I should make some time and stop being such a crappy maintainer)

[14:29:00.0000] <annevk42>
anything known yet about Google and video?

[14:33:00.0000] <annevk42>
http://twitter.com/jdowdell/statuses/1950818992 -- "Reporters who say "the HTML 5 standard" instead of "the WhatWG's RIA proposals for hypertext markup" or such are misleading their readers."

[14:34:00.0000] <gsnedders>
RIA?

[14:35:00.0000] <annevk42>
Did people see: http://blog.typekit.com/2009/05/27/introducing-typekit/ (Especially interesting for those who talked about business models for Web fonts.)

[14:36:00.0000] <jgraham>
I love that Adobe are doing their best o promote HTML5

[14:36:01.0000] <jgraham>
annevk42: See my comment from 5 hours ago

[14:39:00.0000] <Philip`>
gsnedders: Rich internet applications?

[14:39:01.0000] <jgraham>
annevk42: You should maybe point out to Roy that Opera shipd to a bunch of non-desktop devices

[14:40:00.0000] <jgraham>
And AFAIK we have the same requirements for all of them

[14:41:00.0000] <annevk42>
Which email is that?

[14:41:01.0000] <annevk42>
I might have deleted it already...

[14:43:00.0000] <jgraham>
annevk42: http://lists.w3.org/Archives/Public/public-html/2009May/0547.html

[14:44:00.0000] <annevk42>
I do have that one, but I don't see where he's suggesting we don't

[14:44:01.0000] <annevk42>
and the words desktop or device are not in it

[14:44:02.0000] <jgraham>
annevk42: He suggests that different devices require different error handling

[14:45:00.0000] <ezyang>
Oh, it's Roy T. Fielding 

[14:45:01.0000] <ezyang>
He rejected my patch to Apache :-( 

[14:45:02.0000] <annevk42>
jgraham, seems to be mostly about different types of applications

[14:45:03.0000] <jgraham>
annevk42: I think he is vastly overstating the number of different conformance classes

[14:46:00.0000] <jgraham>
annevk42: If we can sell Opera for fridges I guess we would want the same error handling requirements

[14:46:01.0000] <jgraham>
(similarly if it was Webkit or Gecko on the fridge of course)

[14:47:00.0000] <annevk42>
He's being very vague, I would say. It would help me much more if he explained more about what kind of HTML he expects these fridges/switches/etc. to consume/output and why it needs to be different from Web browsers.

[14:47:01.0000] <hsivonen>
Hixie: well, over here, the Web censor policeman said that Google is a browser...

[14:49:00.0000] <jgraham>
/me wonders if he should reply to Shelley

[14:49:01.0000] <hsivonen>
Hixie: when asked why he doesn't poison the DNS record for google.

[14:51:00.0000] <annevk42>
reminds me of http://www.youtube.com/watch?v=MunMCO3uNdA

[14:52:00.0000] <gsnedders>
/me stabs CSS

[14:53:00.0000] <Philip`>
jgraham: Depends on whether your reply would cause progress in productive areas of work, I guess :-)

[14:54:00.0000] <gsnedders>
There has _got_ to be a nicer way to do this than CSS tables

[14:55:00.0000] <jgraham>
Philip`: Yeah. I'm not sure what the best way to cause progress is

[14:57:00.0000] <Philip`>
gsnedders: <table>

[14:58:00.0000] <gsnedders>
Philip`: It isn't a table

[14:58:01.0000] <Philip`>
gsnedders: Presumably you're doing something layout-related, and therefore <table> is a very effective solution

[14:59:00.0000] <beowulf>
/me burns Philip` with fire for saying such a thing

[15:00:00.0000] <jgraham>
As opposed to burning him with acid I guess

[15:01:00.0000] <Hixie>
christ almighty. 46 messages from backtype and google alerts about "html5"

[15:01:01.0000] <beowulf>
well, that would be taking things to far

[15:02:00.0000] <annevk2>
everytime I look at twitter after a few hours there's like 400+ html5/html 5 messages

[15:02:01.0000] <annevk2>
not really following that anymore :)

[15:02:02.0000] <ezyang>
You need a "State of the HTML 5" report :-) 

[15:03:00.0000] <Hixie>
that's mpilgrim's job :-P

[15:03:01.0000] <hsivonen>
annevk2: still not a "trending topic", though

[15:03:02.0000] <hsivonen>
(it seems that twitter is confusing 'trend' and 'fad')

[15:04:00.0000] <Hixie>
dear lord, i just got my first google alert for a 4chan post

[15:07:00.0000] <gsnedders>
Can I table row not have padding!?

[15:08:00.0000] <ezyang>
cell-spacing:0? 

[15:08:01.0000] <ezyang>
erm, border-spacing 

[15:08:02.0000] <Hixie>
gsnedders: what would it mean for a row to have padding?

[15:08:03.0000] <gsnedders>
Hixie: The same as anything else? What's so hard about it?

[15:09:00.0000] <ezyang>
Anyone know what version of simplejson I need in order to make the Python tokenizer not die? 

[15:09:01.0000] <Hixie>
gsnedders: the row doesn't exist as a rendered construct.

[15:09:02.0000] <ezyang>
*tokenizer test 

[15:09:03.0000] <gsnedders>
Hixie: Ah

[15:09:04.0000] <Hixie>
gsnedders: what would you do if you had three rows with different padding? what would it look like?

[15:09:05.0000] <gsnedders>
Hixie: That does make things more complex :)

[15:10:00.0000] <gsnedders>
CSS 2.1 references a "column element". What's that?

[15:11:00.0000] <gsnedders>
display: table-column?

[15:11:01.0000] <Hixie>
gsnedders: dunno, ask csswg :-)

[15:12:00.0000] <gsnedders>
Hixie: Hey, you're listed as an editor, for all little that's worth! :P

[15:12:01.0000] <Hixie>
i was hoping you wouldn't notice :-P

[15:13:00.0000] <ezyang>
Same goes for the ruby tokenizer tests. ARRGH 

[15:14:00.0000] <gsnedders>
/me realizes what it all means

[15:14:01.0000] <gsnedders>
It just omits the table- everywhere there

[15:15:00.0000] <jgraham>
ezyang: I noticed the simplejson problem too :(

[15:15:01.0000] <jgraham>
I think it is just one set of tests that you can disable though

[15:15:02.0000] <ezyang>
Ok, so maybe there was an incompatible "fix" that upstream made 

[15:16:00.0000] <jgraham>
ezyang: Yeah, they made it much more strict with character encodings

[15:16:01.0000] <ezyang>
JSON is such a bad format, really, since it only supports Unicode 

[15:16:02.0000] <jgraham>
It's a bad choice for testing lone surrogates and similar things

[15:17:00.0000] <ezyang>
Well, I suppose I can get rid of the offending tests, pending a new test format 

[15:19:00.0000] <ezyang>
Aweomse, only unicode characters test. 

[15:19:01.0000] <ezyang>
... needs to be removed 

[15:20:00.0000] <Hixie>
wow, i just read larry's latest e-mail. he actually just totally backed down from his earlier opinions and now agrees with the position that html5 should apply to all html user agents and authors! awesome!

[15:20:01.0000] <jgraham>
ezyang: It would be awesome if you converted it to the same format as the tree construction tests

[15:20:02.0000] <jgraham>
(roughly)

[15:20:03.0000] <jgraham>
Like

[15:20:04.0000] <jgraham>
#input

[15:21:00.0000] <ezyang>
That might not be too difficult 

[15:21:01.0000] <jgraham>
some data

[15:21:02.0000] <jgraham>
#output

[15:21:03.0000] <jgraham>
some tokens

[15:21:04.0000] <ezyang>
But in interest of not needing a binary editor, we probably want to keep them escaped 

[15:21:05.0000] <roc>
do we know any more about Google's video story yet?

[15:21:06.0000] <ezyang>
(PHP's JSON parser is really lenient) 

[15:22:00.0000] <ezyang>
We'd also need a token format 

[15:24:00.0000] <ezyang>
gsnedders: what bugs in PHP's tokenizers were you dealing with? 

[15:24:01.0000] <gsnedders>
Is http://svenellingen.com/stream a table?

[15:24:02.0000] <gsnedders>
ezyang: The failing test cases

[15:24:03.0000] <ezyang>
All of them? 

[15:24:04.0000] <gsnedders>
ezyang: There are only two bugs :P

[15:24:05.0000] <gsnedders>
(Actually, three)

[15:24:06.0000] <ezyang>
Huh. I have 15 failures 

[15:24:07.0000] <gsnedders>
I have 10 when I last ran

[15:24:08.0000] <jgraham>
ezyang: The json-like format is good enough

[15:25:00.0000] <jgraham>
for the output format

[15:25:01.0000] <jgraham>
unless yu want something easier to parse

[15:25:02.0000] <gsnedders>
ezyang: I get 10.

[15:26:00.0000] <ezyang>
Let me push 

[15:26:01.0000] <ezyang>
(although that shouldn't make a difference) 

[15:26:02.0000] <ezyang>
Ok, you should pull 

[15:26:03.0000] <ezyang>
It's very possible that you've fixed a bunch 

[15:26:04.0000] <ezyang>
And forgot to push your changes 

[15:26:05.0000] <gsnedders>
There are ten failures on trunk.

[15:27:00.0000] <gsnedders>
I haven't touched it in days, everything is pushed.

[15:27:01.0000] <ezyang>
Hmmm... 

[15:27:02.0000] <ezyang>
What version of PHP are you using? 

[15:27:03.0000] <gsnedders>
Pastebin them

[15:27:04.0000] <gsnedders>
5.2.9

[15:27:05.0000] <ezyang>
Ok. Old version of PHP = more errors. 

[15:28:00.0000] <ezyang>
With 5.2.9 I get 10 

[15:28:01.0000] <ezyang>
This is a bug, btw 

[15:28:02.0000] <gsnedders>
Indeed.

[15:28:03.0000] <gsnedders>
Right now I'm more concerned with getting it working at least locally :P

[15:29:00.0000] <ezyang>
Sure, but it's easier to deal with cross-version bugs as quickly as possible 

[15:29:01.0000] <ezyang>
I've had to deal with it in HTML Purifier, and they're nasty little buggers 

[15:29:02.0000] <ezyang>
(anyway, that's 11 worth of more passing test-cases for TreeBuilder.php) 

[15:29:03.0000] <gsnedders>
/me keeps on hitting bugs that crash PHP

[15:29:04.0000] <ezyang>
Oh? 

[15:30:00.0000] <ezyang>
As in, segfault? 

[15:30:01.0000] <gsnedders>
Not with html5lib, though

[15:30:02.0000] <gsnedders>
yeah

[15:30:03.0000] <ezyang>
Get a core dump and complain to the devs 

[15:30:04.0000] <gsnedders>
That doesn't help when you need to support current releases of PHP

[15:31:00.0000] <ezyang>
Usually, doing some unset()s or superstitious function calls makes the crash go away :-) 

[15:31:01.0000] <gsnedders>
Oh, here's a fun crash:

[15:31:02.0000] <olliej>
Philip`: weeee - http://nerget.com/pressure/pressure.html

[15:31:03.0000] <gsnedders>
curl_version() in some setups crashes PHP.

[15:31:04.0000] <gsnedders>
Just calling it at all crashes PHP.

[15:31:05.0000] <ezyang>
Oh. 

[15:31:06.0000] <gsnedders>
It's great fun. :\

[15:32:00.0000] <ezyang>
Well, that's no good 

[15:32:01.0000] <gsnedders>
IIRC it's fixed now.

[15:32:02.0000] <olliej>
Philip`: its complete absence of html5 validation will make Hixie cry

[15:32:03.0000] <gsnedders>
But that is when I got to the point of just saying to users, "If the PHP devs can't create an interpreter that doesn't crash, it's not my problem."

[15:33:00.0000] <ezyang>
I disagree. 

[15:33:01.0000] <ezyang>
But it depends on how much you care about your users. 

[15:33:02.0000] <ezyang>
If you want the best experience possible, it is very much *your* problem. With open-source, you don't have to care to that level. 

[15:33:03.0000] <gsnedders>
It had got to the point that the majority of bug fixes were working around interpreter bugs.

[15:34:00.0000] <gsnedders>
And that's when I just give up on the interpreter :P

[15:34:01.0000] <ezyang>
What the heck were you doing? 

[15:34:02.0000] <gsnedders>
Nothing much.

[15:34:03.0000] <ezyang>
I've only needed to do about two dozen interpreter fixes 

[15:35:00.0000] <Philip`>
olliej: It's a bit slow in my browsers :-(

[15:35:01.0000] <gsnedders>
/me goes back to his Python wunderland where he's never had to work around interpreter bugs :P

[15:36:00.0000] <ezyang>
Oh, I've had to work around python bugs too. 

[15:36:01.0000] <ezyang>
They're pretty spectacular 

[15:36:02.0000] <ezyang>
One would go away when I added a print statement 

[15:36:03.0000] <Philip`>
olliej: but kind of neat :-)

[15:36:04.0000] <olliej>
Philip`: really?

[15:36:05.0000] <olliej>
what machine are you on?

[15:36:06.0000] <ezyang>
One would only happen of there was a pyc file, but not when it was missing 

[15:37:00.0000] <Hixie>
so is www.w3.org down for anyone else?

[15:37:01.0000] <ezyang>
Works for me 

[15:37:02.0000] <jgraham>
/me only remembers one python interpreter bug

[15:37:03.0000] <Philip`>
Hixie: http://downforeveryoneorjustme.com/www.w3.org

[15:37:04.0000] <jgraham>
(.iteritems() and .items() produced different results)

[15:37:05.0000] <Hixie>
well bummer

[15:38:00.0000] <ezyang>
jgraham: Haha. That sounds excellent 

[15:38:01.0000] <gsnedders>
jgraham: They're meant to.

[15:38:02.0000] <ezyang>
Anyway, afk, for a bit 

[15:38:03.0000] <gsnedders>
jgraham: One produces an iterator and the other a list :P

[15:38:04.0000] <jgraham>
gsnedders: I was waiting for you to say that

[15:39:00.0000] <Philip`>
olliej: A not particularly slow one, but one that's running old (i.e. released) versions of Opera and Firefox rather than a proper browser :-p

[15:39:01.0000] <olliej>
Philip`: try a firefox nightly

[15:39:02.0000] <jgraham>
olliej: That's cool

[15:39:03.0000] <olliej>
Philip`: they do much much better

[15:39:04.0000] <gsnedders>
I'm like Romeo and Juliet: everything after the prologue is inevitable, fate will have its course, no matter what I try.

[15:39:05.0000] <olliej>
Philip`: or a recent webkit build

[15:40:00.0000] <jgraham>
/me was getting about 1500ms/frame with three sources and quite a complex velocity field

[15:41:00.0000] <olliej>
jgraham: the performance is entirely dependent on overall resolution

[15:41:01.0000] <gsnedders>
olliej: You've changed it!

[15:41:02.0000] <gsnedders>
It had a white background before!

[15:41:03.0000] <olliej>
jgraham: number of sources is actually irrelevant 

[15:42:00.0000] <olliej>
gsnedders: i've also improved performance a bit by assuming that there's always 0 diffusion

[15:42:01.0000] <jgraham>
olliej: That makes sense I guess since it presumably does the same number of calculations in any case

[15:42:02.0000] <olliej>
jgraham: yup

[15:42:03.0000] <olliej>
jgraham: the only reason webkit trunk and firefox start off slightly faster is because initially there are no gc allocated numbers

[15:43:00.0000] <jgraham>
(although fewer source terms would cut down a bit and you could theoretically see some difference if he canvas doesn't have to change colour I assume)

[15:43:01.0000] <Philip`>
/me wonders if it'd be possible to optimise it by doing the big calculations via <canvas> compositing

[15:43:02.0000] <olliej>
jgraham: and the number of gc numbers (and hence gc pressure) gradually increases until there are no longer any 0s

[15:43:03.0000] <Philip`>
olliej: Seems a bit better in some kind of Firefox 3.5, but it's still ~200ms per frame

[15:44:00.0000] <olliej>
Philip`: depending on resolution my measurements show either webkit-proper or chrome to be faster

[15:44:01.0000] <gsnedders>
olliej: So WebKit-proper or WebKit-fake? :P

[15:44:02.0000] <olliej>
Philip`: largely because jsc still has a stupid gc

[15:45:00.0000] <olliej>
gsnedders: webkit-proper and webkit-with-an-unnecessarily-replaced-js-engine

[15:45:01.0000] <Philip`>
olliej: I'd be more able to test that if there was a decent WebKit-based browser on Linux

[15:45:02.0000] <olliej>
Philip`: epiphany?

[15:45:03.0000] <olliej>
Philip`: apparently it works

[15:49:00.0000] <gsnedders>
Is onclick fired on an element and all its parents?

[15:49:01.0000] <gsnedders>
In what order?

[15:50:00.0000] <ezyang>
ooh, there's a really good quirks mode article on that 

[15:50:01.0000] <roc>
onclick is fired on one element

[15:50:02.0000] <roc>
it bubbles up through its ancestors

[15:50:03.0000] <ezyang>
http://www.quirksmode.org/js/events_order.html 

[15:50:04.0000] <gsnedders>
/me needs to learn about DOM Events, badly

[15:50:05.0000] <jgraham>
Philip`: chromium nightlies or the webkittestgtk thing?

[15:51:00.0000] <olliej>
Philip`: epiphany-webkit i mean

[15:52:00.0000] <roc>
video question... The spec says "When the current playback position reaches the end of the media resource when the direction of playback is forwards, then the user agent must follow these steps:"

[15:53:00.0000] <roc>
does "reaches the end of the media resource" include reaching the end due to seeking?

[15:55:00.0000] <Hixie>
yes

[15:56:00.0000] <Philip`>
Hmm, Chromium seems to be the only thing that's trivially available on Gentoo, but it complains "version `GLIBCXX_3.4.9' not found (required by ./chrome)" :-(

[15:56:01.0000] <gsnedders>
Hmm, I need to disable an onclick event when a click on an interactive element triggers it

[15:57:00.0000] <Hixie>
it's so funny seeing jd defend flash everywhere

[15:58:00.0000] <Philip`>
Oh, right, it works fine if I change LD_LIBRARY_PATH to point to gcc-4.3.2, so that's okay

[15:58:01.0000] <Hixie>
i think the fact that so many people are so happy to hear they might be able to get rid of flash pretty much sums up how frustrating his job must be

[15:58:02.0000] <gsnedders>
/me wonders how to do this

[15:59:00.0000] <Philip`>
olliej: Ooh, it's very fast in Chromium, except I don't know how fast because it doesn't render any of the text saying how fast it is

[16:00:00.0000] <gsnedders>
Is there any way to tell if something has a default action?

[16:00:01.0000] <Hixie>
i'm also amused by the people trying to get Microsoft to admit to working on <canvas>

[16:00:02.0000] <Hixie>
how about getting them to admit to working on DOM2 Events or something more fundamental like that first? :-)

[16:01:00.0000] <roc>
Hixie: thanks

[16:01:01.0000] <gsnedders>
/me has a strong disdain for DOM Events…

[16:02:00.0000] <roc>
where are people trying to get Microsoft to admit to stuff?

[16:02:01.0000] <olliej>
Philip`: oddly it's faster in webkit trunk, which shouldn't happen -- i assume mac chrome has some perf hit i'm unaware of

[16:02:02.0000] <olliej>
thy used to be on par

[16:05:00.0000] <olliej>
Philip`: hehehe

[16:05:01.0000] <olliej>
Philip`: performance characteristics of our gc

[16:06:00.0000] <olliej>
Philip`: starts at 30ms a frame, becomes 100ms once all the zeros go away

[16:13:00.0000] <gsnedders>
http://codingforums.com/showthread.php?p=821898#post821898

[16:41:00.0000] <Hixie>
roc: e.g. http://processingjs.org/blog/?p=77

[16:44:00.0000] <roc>
ta

[16:45:00.0000] <Philip`>
Seems to be quite a bit of wishful thinking in how they interpret the Microsoft people's responses

[16:51:00.0000] <jwalden>
/me concurs

[16:53:00.0000] <hober>
http://www.w3.org/QA/2009/05/_watching_the_google_io.html


2009-05-29
[17:08:00.0000] <othermaciej_>
hober: PLH has some good factual points, but I think it's the first time I have seen the W3C object to even a WD-level spec being promoted...

[17:10:00.0000] <rubys>
I don't see the word "object" in PLH's post

[17:12:00.0000] <othermaciej>
most objections are not phrased in the form "I object"

[17:13:00.0000] <othermaciej>
let me put it this way, he spent more words on aspects he didn't like of Google's promotion of HTML5, than on aspects he liked

[17:34:00.0000] <Hixie>
hsivonen: yt?

[17:40:00.0000] <Hixie>
well i'll make this change and see if hsivonen complains, i guess

[17:49:00.0000] <sayrer>
/me is reminded of

[17:49:01.0000] <sayrer>
http://tomayko.com/writings/that-dilbert-cartoon

[17:59:00.0000] <Hixie>
othermaciej: your e-mail cuts off after "if Larry feels there is a resemblance between"

[18:01:00.0000] <othermaciej>
Hixie: thanks

[18:12:00.0000] <Philip`>
/me sees that Hixie is single-handedly making tens of percents of web pages less invalid than before

[18:12:01.0000] <Hixie>
i'm trying

[18:57:00.0000] <Hixie>
re http://www.w3.org/QA/2009/05/_watching_the_google_io.html -- maybe i should change the title on the whatwg version to "HTML5 Draft Standard" instead of "HTML5 Draft Recommendation" to remove any confusion... :-P

[19:31:00.0000] <theMadness>
Hixie, yes please.

[19:31:01.0000] <theMadness>
Change it HARD.

[20:22:00.0000] <roc>
hehe

[20:24:00.0000] <roc>
the latest Roy email is priceless

[20:26:00.0000] <ezyang>
linky? 

[20:28:00.0000] <othermaciej>
roc: I find his "I'm the expert, stfu" attitude tiresom

[20:29:00.0000] <othermaciej>
roc: and I will reply to that effect on the list so I can be Shelley-compatible in my behavior

[20:30:00.0000] <roc>
yes

[20:30:01.0000] <roc>
plus the assertion about how browsers are only 1% of HTML clients

[20:30:02.0000] <ezyang>
gsnedders... you didn't unit test InputStream... 

[20:31:00.0000] <ezyang>
This is an offense punishable... by death. 

[20:31:01.0000] <othermaciej>
roc: I didn't even bother to argue with that

[20:31:02.0000] <othermaciej>
roc: but perhaps he is counting by distinct pieces of software as opposed to by number of users or volume of use

[20:31:03.0000] <roc>
yes, I think he is

[20:32:00.0000] <othermaciej>
which is not completely irrational, if he is using that to support an argument that the spec should be more general

[20:32:01.0000] <roc>
IIRC from threads long ago

[20:32:02.0000] <othermaciej>
but he still hasn't given any concrete examples

[20:32:03.0000] <othermaciej>
I'd *love* to hear how it's impossible for libwww-perl to comply with HTML5

[20:32:04.0000] <roc>
it's still irrational in at least two ways

[20:33:00.0000] <roc>
you've got to weight by how much an application is used, or you'll be swamped by never-used edge cases

[20:33:01.0000] <roc>
also, authors test in browsers, not any of these other hypothetical applications

[20:33:02.0000] <roc>
(maybe search engines, sort of)

[20:34:00.0000] <othermaciej>
as I see it, software that consumes HTML falls into one of two categories:

[20:34:01.0000] <othermaciej>
(a) it is intended to consume arbitrary HTML content from the Web

[20:34:02.0000] <roc>
so applications that want to process HTML need to process it the way authors do, if they want to capture the author's intent

[20:34:03.0000] <othermaciej>
(b) it is intended to consume only a restricted subset of the language or a specific set of documents in a controlled environment

[20:35:00.0000] <othermaciej>
it seems to me that category (a) must interoperate with browsers

[20:35:01.0000] <roc>
yeah

[20:35:02.0000] <othermaciej>
and for category (b), the spec is irrelevant

[20:35:03.0000] <roc>
right

[20:35:04.0000] <othermaciej>
if they want a relevant spec, they have to spec their subset

[20:35:05.0000] <othermaciej>
or they can just work ad-hoc without a spec

[20:36:00.0000] <othermaciej>
though I will add, that even for category (b) it is often desirable to use standard tools to create content and ordinary browsers to test

[20:36:01.0000] <othermaciej>
so even then, it may be highly desirable to interoperate with browsers

[20:47:00.0000] <ezyang>
Sweet. Extra five fails came from change in substr() behavior 

[20:47:01.0000] <ezyang>
Now fixed 

[20:50:00.0000] <ezyang>
Sigh, Google Code authorization is acting up again. 

[20:50:01.0000] <ezyang>
Seriously dudes. You're google. 

[20:50:02.0000] <ezyang>
You eat your own dogfood. (I hope) 

[20:51:00.0000] <ezyang>
Strangely enough, the problem seems to fix itself when I navigate to my code.google.com/hosting/settings page :-/ 

[20:59:00.0000] <othermaciej>
Shelley quit again?

[21:04:00.0000] <ezyang>
Oh, awesome, the rest of the tokenizer fails are parse-error fails 

[21:08:00.0000] <othermaciej>
roc: I guess if I were more inclined to be indignant I would be scandalized by Roy's claim that I am "clueless" about interpreters and "don't know anything about the field" of HTML

[21:09:00.0000] <roc>
you should be

[21:09:01.0000] <othermaciej>
I think it just makes him look like an idiot to say things like that

[21:48:00.0000] <olliej>
othermaciej: are you claiming you know stuff about interpreters again?

[21:48:01.0000] <ezyang>
Oh hey, whoops. Parse errors are not tokens 

[21:49:00.0000] <othermaciej>
olliej: yeah, I like to pretend

[21:50:00.0000] <olliej>
othermaciej: :D

[22:20:00.0000] <ezyang>
Argh foster parenting 

[22:33:00.0000] <mpilgrim>
hixie: html5/reddit alert: http://www.reddit.com/r/technology/comments/8nzok/google_waves_goodbye_to_email/c09w4u0

[22:33:01.0000] <mpilgrim>
that may overstate the situation somewhat

[22:34:00.0000] <mpilgrim>
would you settle for "html5 reference on reddit"?

[22:38:00.0000] <mpilgrim>
discussion of html5 video and codecs: http://www.reddit.com/r/programming/comments/8np9i/youtube_experimenting_with_html5_video/

[22:38:01.0000] <mpilgrim>
more video: http://www.reddit.com/r/programming/comments/8nr98/daily_motion_uses_html_5_video_with_ogg_theora/

[22:40:00.0000] <mpilgrim>
some meta-discussion: http://www.reddit.com/r/programming/comments/8ns2l/dear_proggit_html5_is_not_programming/

[22:46:00.0000] <othermaciej>
wow, lotsa discussion

[22:46:01.0000] <othermaciej>
(I believe Hixie has seen at least some of those threads)

[22:47:00.0000] <othermaciej>
mpilgrim: I

[22:47:01.0000] <othermaciej>
I'm waiting to see how your linking of those threads can be spun as malevolent

[22:48:00.0000] <jwalden>
1.  knowledge is power.  2. power corrupts.   3. ...   4. GOOGLE PROFIT!

[22:53:00.0000] <othermaciej>
Google Wave looks a lot more visually busy than the typical Google web app

[22:59:00.0000] <roc>
I wonder why these codec discussions fail to mention that the moratorium on H.264 Internet "broadcasting" fees runs out next year

[23:44:00.0000] <othermaciej>
"HTML5 Could be the OS Killer"

[23:44:01.0000] <othermaciej>
apparently I'm suicidal

[23:46:00.0000] <othermaciej>
at the same time: "Google shows Native Client built into HTML 5"

[23:51:00.0000] <ezyang>
/me deals the death blow to the bug and emerges victorious 

[23:51:01.0000] <ezyang>
Now, time for sleep 

[01:02:00.0000] <Hixie>
wait... did shelley leave the wg again?

[01:02:01.0000] <Hixie>
man, she makes me dizzy

[01:03:00.0000] <othermaciej>
it seems that she is not presently  a member of the HTML WG

[01:04:00.0000] <othermaciej>
according to the member list on the Web

[01:06:00.0000] <Hixie>
i guess it must be thursday

[01:08:00.0000] <othermaciej>
I assume she was offended by Sam pointing out that while she complains about outside-the-list snark/hostility, she also produces it

[01:23:00.0000] <Hixie>
othermaciej: no, i didn't intend the asymmetric case, i'll fix that in due course

[01:24:00.0000] <othermaciej>
Hixie: are you still going to allow UAs to switch between the two options at any time?

[01:24:01.0000] <Hixie>
othermaciej: yeah

[01:24:02.0000] <Hixie>
the requirement should be that for ports A and B

[01:24:03.0000] <othermaciej>
seems like a pretty low-value option if they would have to do so synchronously for both endpoints

[01:24:04.0000] <Hixie>
it doesn't have to be symmetric

[01:24:05.0000] <Hixie>
for port A, either A.owner->A or B->A

[01:24:06.0000] <othermaciej>
if you want to make it less of a loophole the requirement should be that A is either owned by B or by A's owner

[01:24:07.0000] <othermaciej>
right

[01:25:00.0000] <Hixie>
right now the text says either A.owner-> or A->B

[01:25:01.0000] <Hixie>
which is bogus

[01:25:02.0000] <Hixie>
the idea is still that you don't have to keep a reference to the object for it to survive long, since otherwise we expose GC

[01:26:00.0000] <Hixie>
"why do the first five messages get through?" "because after that it got around to GCing your port away" isn't a conversation i ever want to have :-)

[01:26:01.0000] <othermaciej>
so long as .close() gives you an out to clean up your resources, that's not so bad

[01:26:02.0000] <othermaciej>
but

[01:26:03.0000] <Hixie>
yeah

[01:26:04.0000] <othermaciej>
I don't see why exposing the potential unpredictability of GC is so bad

[01:26:05.0000] <othermaciej>
in an API to be used with concurrency

[01:26:06.0000] <othermaciej>
because concurrency means you can't really have deterministic guarantees of order of operations anyway

[01:27:00.0000] <Hixie>
you think this conversation is one that authors will be ok with? "why do the first five messages get through?" "because after that it got around to GCing your port away"

[01:27:01.0000] <othermaciej>
I wouldn't answer that way - I'd say "hold a reference to your MessagePort while you are using it"

[01:27:02.0000] <Hixie>
i think debugging such a bug would be a nightmare

[01:27:03.0000] <othermaciej>
debugging concurrency bugs is often a nightmare

[01:28:00.0000] <othermaciej>
imagine if your code subtly depends on whether worker A receives a message from worker B or window W first

[01:28:01.0000] <othermaciej>
you might never see the wrong order on "your" test system

[01:28:02.0000] <Hixie>
i agree that it has intrinsic pain

[01:28:03.0000] <Hixie>
adding more seems bad

[01:28:04.0000] <othermaciej>
this is why I think removing nondeterminism from an API for a concurrent system is wasted effort

[01:29:00.0000] <othermaciej>
well, by always either holding a reference or using .close() you can easily remove the risk

[01:29:01.0000] <othermaciej>
the only difference at this point is what happens to authors who are not careful in this way

[01:30:00.0000] <othermaciej>
do they get random memory leaks (result of the strong reference rule) or a bit of extra nondeterminism (result of not having such a rule at all)

[01:30:01.0000] <zcorpan_>
MikeSmith: "If the area element has no href attribute, then the area represented by the element cannot be selected, and the alt attribute must be omitted."

[01:31:00.0000] <othermaciej>
I don't think it's so important which happens, so I am not very concerned about the choice, as long as there is a reasonable way for content authors to do the right thing

[01:32:00.0000] <Hixie>
othermaciej: i think the memory leaks are the better option since if they are found to be common, browsers will just do the more complicated alternative.

[01:32:01.0000] <Hixie>
othermaciej: whereas they can't do anything about the other problem.

[01:32:02.0000] <othermaciej>
I don't think you understand how complicated the alternative is

[01:32:03.0000] <othermaciej>
if the leaks become a problem, I'd just violate the spec

[01:34:00.0000] <othermaciej>
(but hopefully giving authors the tools to DTRT will be sufficient)

[01:35:00.0000] <Hixie>
i don't think teh alternative is as bad as you make out. you just need to keep track of when port gets to zero js references, and when that happens, tell the other side the ID of the last message you received and the ID of the last message you sent, and say that you're ready to go away. when both sides do that and they both sent the same IDs, then they both know they can go away.

[01:35:01.0000] <Hixie>
consider the opposite problem -- one browser with 90% market share does GC every 20 seconds, the other does GC aggressively. a page that always sends two messages only always works on the former browser, always fails on the other.

[01:36:00.0000] <othermaciej>
JS isn't reference counted

[01:36:01.0000] <othermaciej>
there isn't a simple concept of "get to 0 references"

[01:36:02.0000] <othermaciej>
but there are also much more complex cases than you describe

[01:36:03.0000] <Hixie>
s/zero js references/would otherwise get sweeped/

[01:36:04.0000] <othermaciej>
consider a triangle of MessageChannels

[01:36:05.0000] <othermaciej>
where the MessagePorts in each window indirectly reference each other

[01:37:00.0000] <Hixie>
yeah, that's a tough one

[01:37:01.0000] <othermaciej>
(keep in mind also that linkage between three heaps is still a far simpler case for distributed GC than the fully general case)

[01:39:00.0000] <othermaciej>
(btw it's easy to get 3-way linkage like that through event listener closures capturing multiple message ports in scope)

[01:39:01.0000] <othermaciej>
(so it's not some bizzarre theoretical construct)

[01:41:00.0000] <Hixie>
agreed

[01:42:00.0000] <othermaciej>
so the upshot is that the opposite-endpoint ownership model is not really practical if you have separate heaps for workers, thus MessagePorts are immortal until you close() one of the endpoints or the other, or leave the Document

[01:43:00.0000] <othermaciej>
(er, for the last bit, I guess I should say, "until you navigate away")

[01:44:00.0000] <Hixie>
well exposing the GC just doesn't seem like an option to me

[01:44:01.0000] <Hixie>
i mean, we're going to extreme lengths to avoid it elsewhere

[01:45:00.0000] <othermaciej>
that just means that explicit resource management is the alternative

[01:45:01.0000] <othermaciej>
which is not necessarily that terrible

[01:45:02.0000] <Hixie>
yeah, i guess

[01:45:03.0000] <Hixie>
though people will screw that up too

[01:45:04.0000] <othermaciej>
just saying, it's not super helpful for the spec to make it seem like you can count on automatic resource management of these things

[01:48:00.0000] <Hixie>
hm, yes, we could make it more explicit that you should call .close() on ports

[01:55:00.0000] <Hixie>
http://beta.w3.org/standards/about.html

[01:56:00.0000] <Hixie>
after 15 years of telling people the w3c wasn't in the business of writing standards but was in fact in the business of writing recommendations, they're planning on confusing everyone by changing their mind

[01:56:01.0000] <Hixie>
good work w3c.

[01:59:00.0000] <annevk2>
seems like a positive change to me

[02:01:00.0000] <Philip`>
othermaciej: libwww-perl uses the HTML::HeadParser module which uses HTML::Parser and then looks for things like <meta http-equiv> and <base> and <link> and <isindex>

[02:01:01.0000] <othermaciej>
most people think of them as "Standards"

[02:01:02.0000] <Philip`>
(so they can be exposed via the HTTP header API)

[02:01:03.0000] <othermaciej>
calling them "Recommendations" and not standards was just confusing

[02:01:04.0000] <othermaciej>
Philip`: I did not know that

[02:03:00.0000] <othermaciej>
Philip`: is HTML::HeadParser // HTML::Parser unable to conform to HTML5?

[02:03:01.0000] <othermaciej>
(I assume not, since people have written HTML5 parsers in perl...)

[02:04:00.0000] <Philip`>
othermaciej: I see no reason that it couldn't use an HTML5 parser

[02:04:01.0000] <Philip`>
(though I don't know whether that's enough to technically "conform to HTML5")

[02:05:00.0000] <othermaciej>
I would guess it falls under the 'Data mining tools' conformance class

[02:05:01.0000] <othermaciej>
though perhaps there should be a conformance class for reusable parser libraries

[02:05:02.0000] <othermaciej>
since such a library is testable by itself usually, but can't predict what kind of software will use it

[02:06:00.0000] <Philip`>
Seems hard to specify since the libraries will all have different output formats

[02:06:01.0000] <othermaciej>
Philip`: anyway - thanks for pointing that out, though overall I still think Roy is wrong in his assertions, and incredibly rude

[02:06:02.0000] <othermaciej>
well, XML has rules for XML processors and people are able to determine whether XML parser libraries conform

[02:07:00.0000] <Philip`>
I guess it would have to specify equivalence between the conceptual DOM created by the spec's algorithm, and the concrete DOM/SAX/ElementTree/etc created by implementations

[02:07:01.0000] <Philip`>
or just leave it undefined and tell people to use common sense

[02:10:00.0000] <Philip`>
http://translate.google.com/translate?hl=en&sl=ru&tl=en&u=http://robotstxt.org.ru/rurobots/yandex

[02:10:01.0000] <Philip`>
"Yandex Robot supports tag noindex, which prohibits job Yandex index queries (office) of the text. At the beginning of a service fragment is <noindex>, but in the end - </ noindex>, Yandex, and will not index the site text."

[02:11:00.0000] <zcorpan_>
apache indexes would benefit from using an html5 parser here http://simon.html5.org/test/html/parsing/fragment/content-model-flag/

[02:12:00.0000] <zcorpan_>
(it shows just "setting innerHTML" instead of the correct "setting innerHTML on <title>")

[02:12:01.0000] <zcorpan_>
"setting innerHTML on"

[02:13:00.0000] <zcorpan_>
nowadays i try to remember escaping < as &lt; in titles to cater for the apache bug but it would be nice if it worked correctly

[02:21:00.0000] <zcorpan_>
Hixie: how was http://www.w3.org/Bugs/Public/show_bug.cgi?id=6586 fixed?

[02:22:00.0000] <Hixie>
it was fixed lng ago

[02:22:01.0000] <Hixie>
i just forgot to mark the bug as fixed

[02:22:02.0000] <Hixie>
spec now says:

[02:22:03.0000] <Hixie>
# When a Document is in quirks mode, margins on HTML elements at the top or bottom of the initial containing block, or the top of bottom of td or th elements, are expected to be collapsed to zero.

[02:23:00.0000] <Hixie>
othermaciej: your e-mail to roy had an incomplete sentence again

[02:23:01.0000] <Hixie>
"If they can only handle a defined subset inside their walled garden, then they are not conforming implementations. If they are"

[02:23:02.0000] <othermaciej>
Hixie: in case you coudn't tell, I edit out of order a lot

[02:23:03.0000] <zcorpan_>
Hixie: the spec quotes that exact text and says it's wrong

[02:24:00.0000] <zcorpan_>
er

[02:24:01.0000] <othermaciej>
*couldn't

[02:24:02.0000] <zcorpan_>
the bug

[02:24:03.0000] <Hixie>
othermaciej: so do i, but then i proof-read :-P

[02:24:04.0000] <othermaciej>
I do too - had a real bad headache today though

[02:24:05.0000] <othermaciej>
surprised I was able to be coherent at all

[02:24:06.0000] <othermaciej>
I should do a TL;DR reply to myself

[02:24:07.0000] <Hixie>
zcorpan_: no, the text in the bug is different

[02:24:08.0000] <othermaciej>
because in all the arguing, I do have an interesting main point

[02:24:09.0000] <Hixie>
othermaciej: ah, sorry to hear that

[02:24:10.0000] <Hixie>
about the headache, not the point

[02:25:00.0000] <othermaciej>
which is that HTML processors either (a) are expected to work with general public Web content, in which case they must interoperate with browsers, or (b) work with a restricted subset of content (either conforming to some subset rule or in a walled garden) in which case the spec is irrelevant to them

[02:27:00.0000] <zcorpan_>
Hixie: ok it's different, but i don't see how it collapses when the body has a border

[02:28:00.0000] <Hixie>
zcorpan_: why would the border affect it?

[02:28:01.0000] <Hixie>
the margins just collapse to zero whatever else is there

[02:28:02.0000] <Hixie>
with whatever else, i mean

[02:29:00.0000] <zcorpan_>
Hixie: the body is not the initial containing block, and margin collapsing doesn't happen when there's a border, so as i read it it wouldn't collapse when there's a border

[02:30:00.0000] <zcorpan_>
Hixie: instead the body's margins would collapse, which they shouldn't

[02:30:01.0000] <Hixie>
oh it should say top or bottom of the <body>, not the ICB?

[02:30:02.0000] <zcorpan_>
yeah

[02:33:00.0000] <Mrmil>
Dear WHATWG, I'm yet-another-I-want-to-help person and I would like to ask you kindly for your feedback.

[02:33:01.0000] <Mrmil>
I'm working on a case-study, a common web presentation homepage written in HTML 5, which (when finished) can get chopped up into smaller parts and form a tutorial.

[02:33:02.0000] <Mrmil>
Please note it's not finished at all, many things need to be done and I need to educate myself, too. :) If you are interested, you can find it here: http://server.ebrana.cz/olda/_apps/html5/ .

[02:33:03.0000] <Mrmil>
It's a temporary URL until I/we decide what to do with it next.

[02:33:04.0000] <Mrmil>
I'm planning on adding more feature to this page, so I'll post here again when done. If you want to provide feedback, please email me at vetesnik⊙mc And yeh, I'll buy you a cup of coffee. :) Thanks.

[02:34:00.0000] <zcorpan_>
mmm coffee

[02:35:00.0000] <zcorpan_>
Mrmil: http://gsnedders.html5.org/outliner/

[02:36:00.0000] <Mrmil>
zcorpan_: Ooo, cool, thanks!

[02:36:01.0000] <zcorpan_>
Mrmil: <html lang="cs"> - shouldn't it be "en"?

[02:37:00.0000] <Mrmil>
zcorpan_: Yes, I'm czech so I forgot to change it.

[02:37:01.0000] <Mrmil>
zcorpan_: done

[02:37:02.0000] <hsivonen>
Philip`: I thought the correspondence of DOM/SAX/ElementTree etc. to infoset is already understood

[02:37:03.0000] <zcorpan_>
<section id="header"> - without checking the outline or further in the source, knee-jerk reaction is that this should be <header>

[02:38:00.0000] <hsivonen>
Philip`: and HTML5 defines how to coerce the output of the parsing algorithm to infoste

[02:39:00.0000] <zcorpan_>
hsivonen: still people say "HTML5 talks about DOM, there are tools that don't use a DOM, hence HTML5 is useless for those tools"

[02:39:01.0000] <hsivonen>
zcorpan_: It's a sign of people not having read the spec carefully

[02:40:00.0000] <zcorpan_>
Mrmil: the <h1>Navigation</h1> should probably go inside the <nav>

[02:40:01.0000] <jgraham>
Mrmil: <p class="skipLinks"> should be unnecessary (the whole paragraph)

[02:41:00.0000] <zcorpan_>
<div id="wrapper"> is unnecessary too (style <body> instead)

[02:41:01.0000] <jgraham>
<section id="content"> -> <article> maybe

[02:41:02.0000] <jgraham>
<section id="sidebar"> -> <aside> aiui

[02:41:03.0000] <Philip`>
hsivonen: Coercion to infosets doesn't seem an ideal way to define equivalence, since the coercion can lose information

[02:42:00.0000] <jgraham>
Philip`: I'm still not sure I understand what information is lost

[02:42:01.0000] <zcorpan_>
Mrmil: i would probably have a single <nav> and have subheadings for quicknav and language

[02:42:02.0000] <Philip`>
jgraham: It can e.g. drop attributes whose name starts with "xmlns"

[02:42:03.0000] <zcorpan_>
s/subheadings/subsections/

[02:42:04.0000] <jgraham>
Philip`: Oh yeah, I had forgotten that. 

[02:43:00.0000] <hsivonen>
Philip`: I think common sense should be enough to fill in the gaps given the stream to DOM spec and the coercion spec, but I guess it could be more explicit

[02:43:01.0000] <zcorpan_>
Mrmil: <hr class="displayNone"> - clearly presentational abuse of class

[02:43:02.0000] <hsivonen>
Philip`: is there any implementor who hasn't been able to figure out how to map the non-infoset parts of the DOM to an arbitrary non-infoset-enforcing API?

[02:43:03.0000] <Philip`>
Maybe it should be explicit that you should use common sense

[02:44:00.0000] <hsivonen>
common sense is rare, though :-(

[02:44:01.0000] <Philip`>
e.g. add a conformance class for "HTML parser library" and require that it has a common-sense equivalence to the DOM produced by the specified algorithm

[02:44:02.0000] <jgraham>
zcorpan_: Presenational "abuse" of class isn't forbidden is it?

[02:44:03.0000] <zcorpan_>
How to read this specification: use common sense

[02:44:04.0000] <jgraham>
(but I think that <hr> should go)

[02:44:05.0000] <Mrmil>
zcorpan_: I was wondering about that, I'll switch it to the <nav>

[02:45:00.0000] <hsivonen>
Philip`: common-sense equivalance or where prohibited, using the coercion section's rules

[02:45:01.0000] <zcorpan_>
jgraham: no it's not forbidden but it's not good practice

[02:45:02.0000] <Philip`>
hsivonen: That sounds sensible

[02:45:03.0000] <Mrmil>
jgraham: what would you suggest then?

[02:45:04.0000] <hsivonen>
Philip`: I think I could formulate general rules for any API using the infoset as an aid

[02:45:05.0000] <Mrmil>
jgraham: removing whole skiplinks or removing the <p>?

[02:45:06.0000] <jgraham>
Mrmil: Why do you need an <hr>?

[02:45:07.0000] <hsivonen>
because the problems are mostly arbitrary restrictions on what characters are allowed where

[02:46:00.0000] <zcorpan_>
Mrmil: the arvhices list in the sidebar probably doesn't need <nav>

[02:47:00.0000] <hsivonen>
Philip`: 1) figure out how the infoset maps to API X

[02:47:01.0000] <zcorpan_>
Mrmil: <section id="footer"> - <footer>

[02:47:02.0000] <Philip`>
I suppose I just think it'd be nice to be able to say "this is a conforming HTML5 parser library according to the spec", instead of having to say "this is an HTML5 parser library which could be used as part of a data mining tool and would not by itself cause the data mining tool to be non-conforming"

[02:47:03.0000] <hsivonen>
Philip`: 2) Apply the DOM to infoset coercion from DOM Level 3

[02:47:04.0000] <Hixie>
nn

[02:47:05.0000] <jgraham>
Mrmil: Remove the skiplink. AIUI they don't work that well in practice an in theory the browser can guess rather well anyway (e.g. look for the first <article>)

[02:47:06.0000] <hsivonen>
Philip`: in step #2, pretend that restrictions on what characters are allowed where don't apply

[02:48:00.0000] <Mrmil>
jgraham: hr's removed, OK, will remove skiplinks too :)

[02:48:01.0000] <jgraham>
Mrmil: <section class="todo"> no need for the wrapper <div>

[02:48:02.0000] <Mrmil>
jgraham: done

[02:48:03.0000] <hsivonen>
Philip`: 3) Map infoset to API X per step 1

[02:48:04.0000] <Mrmil>
jgraham: I use the div only for styling purpose

[02:48:05.0000] <zcorpan_>
Mrmil: "© Company Name 2009, All rights reserved. Company Product" doesn't make sense as a paragraph

[02:48:06.0000] <hsivonen>
Philip`: 4) If this causes API X to fail, remove failures by applying the coercion section

[02:49:00.0000] <hsivonen>
Philip`: is that comprehensive enough?

[02:50:00.0000] <jgraham>
Mrmil: I bet you don't really need it

[02:50:01.0000] <hsivonen>
Hixie: you should make a reference to DOM Level 3 for the baseline DOM to infoset mapping that the coercion rules modify when needed

[02:50:02.0000] <Philip`>
hsivonen: I guess that could work

[02:51:00.0000] <hsivonen>
Philip`: works for everything but RDFa

[02:51:01.0000] <Mrmil>
jgraham: you bet right, but what if I got a design with rounded corners?

[02:51:02.0000] <zcorpan_>
/me ponders about <address> in the sidebar

[02:51:03.0000] <jgraham>
Mrmil: border-radius?

[02:52:00.0000] <Mrmil>
jgraham: doesn't work in ie, not implemented enough, cannot use it for production use, my superiors would kill me

[02:52:01.0000] <Mrmil>
jgraham: will have to wait for css3 a little while

[02:53:00.0000] <jgraham>
Mrmil: I thought this was a case study not a production site for clients who demand precidely the same look in all browsers

[02:53:01.0000] <jgraham>
*precisely

[02:53:02.0000] <Mrmil>
jgraham: Well it's a case study based on real demands, I don't want to make it a SCI-FI, people need real tutorials for real needs, I should've said that I guess

[02:54:00.0000] <zcorpan_>
Mrmil: the link farm at the bottom should probably go in the <footer>, too

[02:56:00.0000] <Mrmil>
zcorpan_: Are all the things you said alright now?

[02:56:01.0000] <hsivonen>
zcorpan_: are you planning on having a mapping from Web DOM to Infoset?

[02:57:00.0000] <Mrmil>
zcorpan_: "? Company Name 2009, All rights reserved. Company Product" doesn't make sense as a paragraph -> what would you suggest then? I don't like the idea of <div>ing everthing

[02:58:00.0000] <hsivonen>
Mrmil: it's a "paragraph" in the HTML5 sense of "paragraph"

[02:58:01.0000] <zcorpan_>
Mrmil: <p><small>copyright</small></p><p><a><img>...

[03:00:00.0000] <zcorpan_>
hsivonen: dunno

[03:01:00.0000] <zcorpan_>
hsivonen: i guess it would be good

[03:01:01.0000] <Mrmil>
zcorpan_: better now? I'm a bit confused :)

[03:01:02.0000] <zcorpan_>
Mrmil: yes, now at least the copyright paragraph makes sense. :)

[03:02:00.0000] <Mrmil>
zcorpan_: ok :)

[03:03:00.0000] <zcorpan_>
now where's my coffee?

[03:03:01.0000] <Mrmil>
zcorpan_: let me know if there are any other problems, I'll have some lunch now. Thanks for the feedback. Mmm, how do you like your coffee?

[03:03:02.0000] <zcorpan_>
Mrmil: black

[03:04:00.0000] <Mrmil>
zcorpan_: there you go http://www.nothinggeek.com/wp-content/uploads/2009/01/black-coffee.jpg

[03:04:01.0000] <zcorpan_>
thanks!

[03:04:02.0000] <Mrmil>
Cheers. I'll have some after lunch. Love it.

[03:05:00.0000] <Mrmil>
I have a question: should there be only one <nav> on a page?

[03:05:01.0000] <zcorpan_>
there's no such rule

[03:05:02.0000] <zcorpan_>
but i guess it makes the page easier to navigate

[03:06:00.0000] <Philip`>
zcorpan_: Maybe the TAG could give you some advice on the difference between coffee and a representation of coffee

[03:07:00.0000] <Philip`>
(Seems a similar problem to http://www.w3.org/2001/tag/issues#httpRange-14)

[03:07:01.0000] <Mrmil>
zcorpan_: On some of our presentations, we have a header-menu which goes throughout whole site, and then we have a column menu which goes into subpages for the particular page. Does that mean that both of them should be in a nav? I guess so.

[03:11:00.0000] <Mrmil>
jgraham: <section id="content"> -> <article> -> I was thinking about that. Then if I have real news up there, I'll nest <article> right?

[03:12:00.0000] <Mrmil>
jgraham: <section id="sidebar"> -> <aside> aiui -> I was thinking about that too, but wasn't sure

[03:12:01.0000] <jgraham>
Mrmil: Well it should only be <article> if it really is an article

[03:12:02.0000] <jgraham>
If it is a collection of articles <section> is probably fine

[03:13:00.0000] <Mrmil>
jgraham: it's not always an article. The contents of the #content part varies - depending whether you are on homepage, detail product, checkout, etc.

[03:13:01.0000] <jgraham>
Mrmil: If it is a collection of things then <section> probably works better

[03:13:02.0000] <Mrmil>
jgraham: Ok.

[03:13:03.0000] <jgraham>
(a collection on the same page)

[03:15:00.0000] <Mrmil>
jgraham: is my #sidebar had a product catalogue, bestselling products or banners, should it still be an <aside>?

[03:15:01.0000] <jgraham>
Like <section><h1>Blog posts</h1><article><h1>My first Blog Post</h1></article><article><h1>My second blog post</h1></article>

[03:15:02.0000] <jgraham>
</section>

[03:16:00.0000] <jgraham>
Is better than it would be with s/section/article/

[03:16:01.0000] <jgraham>
But <article><h1>Product details</h1></article> is fine

[03:17:00.0000] <jgraham>
Mrmil: Per spec, yes

[03:17:01.0000] <jgraham>
(re: <aside>)

[03:18:00.0000] <annevk2>
Hixie, reason for quitting seems other work: http://twitter.com/burningbird/status/1953228067

[03:19:00.0000] <zcorpan_>
Mrmil: i usually have a sub list for that case i.e. <nav><ul><li><a>home</a><li><a>products</a><ul><li>product 1<li><a>product 2</a></ul><li><a>etc

[03:22:00.0000] <zcorpan_>
Philip`: i can always print the representation of coffe to obtain a physical form

[03:23:00.0000] <zcorpan_>
coffee

[03:36:00.0000] <annevk2>
"Hurray for the tracking view!" nice to know it's appreciated :)

[03:37:00.0000] <annevk2>
it's unfortunate that making it better means a significant increase in complexity (afaict)

[03:40:00.0000] <jgraham>
annevk2: Where are you quoting?

[03:42:00.0000] <Philip`>
jgraham: http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2009-May/019982.html

[04:01:00.0000] <Mrmil>
Ok, any other suggestions for http://server.ebrana.cz/olda/_apps/html5/ ?

[04:03:00.0000] <jgraham>
<section id="header"> -> <header>

[04:04:00.0000] <jgraham>
You could probably remove <section id="content"> altogether

[04:04:01.0000] <jgraham>
(just keep the articles)

[04:04:02.0000] <jgraham>
Since you don't have another <h1> that applies to the <body>

[04:05:00.0000] <jgraham>
The empty <span> elements are kind of ugly

[04:06:00.0000] <jgraham>
Having both <section id="footer"> and <footer> seems odd

[04:11:00.0000] <Mrmil>
jgraham: removing <section id="content"> might be a problem due to 2-col layout and 3-col layout problems, but will try it occasionally.

[04:12:00.0000] <Mrmil>
jgraham: empty span elements are ugly but they convey meaning so I have to keep the text behind it so it's accessible with images disabled. And I don't want to use <img>'s for that, that's even more ugly. :)

[04:15:00.0000] <Mrmil>
jgraham: I changed section id="header" to header id="header", I need the id so I don't style every header this way. What would you suggest do with the footers? I'd change the section id="footer" into footer element but then there would be 2 footer's. The link farm is ugly but our IM dept. adds it everywhere so I have to deal with it somehow.

[04:19:00.0000] <annevk2>
http://twitter.com/gazcoop/statuses/1957781728 :)

[04:21:00.0000] <annevk2>
http://twitter.com/martin_probst/statuses/1958288180 -- "HTML5 Spec is a weird document. Offline Applications section is almost just C code, nothing about intent, expected outcomes or behaviour." hard to argue with that; hopefully the introduction section addresses these concerns adequately

[04:36:00.0000] <Mrmil>
jgraham: One more qustion, if I want to keep the #content section just for styling purposes then, should I make it <div> instead?

[04:36:01.0000] <hsivonen>
Whew. The check-in message for  r3148 is wrong. 

[04:36:02.0000] <hsivonen>
Spec looking good.

[04:42:00.0000] <annevk2>
Google Wave looks pretty neat

[04:42:01.0000] <annevk2>
little bit scared of all the probable data lock-in though

[04:43:00.0000] <zcorpan>
http://www.techcrunch.com/2009/05/28/google-wave-the-full-video-from-google-io/ - drag and drop several images from desktop to web app in one go

[04:44:00.0000] <jgraham>
annevk2: Isn't it supposed to be an open protocol somehow?

[04:45:00.0000] <annevk2>
I was not assuming open protocol means that really

[04:45:01.0000] <annevk2>
but I don't know the details of that yet

[04:48:00.0000] <jgraham>
Mrmil: Re: #content it probably doesn't matter much, but per spec, if the <body> doesn't have a <h1> then it will look like an empty section in an outline

[04:50:00.0000] <jgraham>
But I'm trying to get Hixie to change that

[04:51:00.0000] <jgraham>
(because I expect a bunch of people to do essentially the same thing that you have done)

[05:05:00.0000] <gsnedders>
ezyang: But all the Tokenizer test cases test it :P

[05:06:00.0000] <annevk2>
I guess since they open source it it's ok

[05:07:00.0000] <annevk2>
Hopefully the network effects are still good enough if you run your own...

[05:07:01.0000] <gsnedders>
""\"\"\"\"geoffers\"\"\"@gmail com — that's certainly an interesting email address to receive from

[05:08:00.0000] <Philip`>
annevk2: http://www.waveprotocol.org/ - sounds like the idea is that you can be part of the global network without being locked into Google's servers

[05:09:00.0000] <annevk2>
yeah

[05:09:01.0000] <Philip`>
Seems like it's basically XMPP (Jabber) with some extensions

[05:09:02.0000] <annevk2>
TAG might enjoy speculating about that versioning issue :)

[05:09:03.0000] <Philip`>
so it should have the same level of openness as XMPP, with everyone able to link to each other's servers

[05:11:00.0000] <Philip`>
and the extensions seem to be a way of encoding deltas for XML documents

[05:11:01.0000] <Philip`>
(in a composable way)

[06:00:00.0000] <annevk2>
I wonder how HTML WG discussions would look in Wave. I have the feeling it might be pretty overwhelming.

[06:04:00.0000] <jgraham>
annevk2: I was thinking the same thing

[06:05:00.0000] <Philip`>
HTML WG discussions are pretty overwhelming regardless of the medium

[06:06:00.0000] <annevk2>
I wonder how URL addressing works with Wave. I guess you'd have a "bot" that is invited in Wave discussions that pushes content out now and then...

[06:26:00.0000] <Lachy>
wow, I just noticed that pave the cowpaths is getting discussed again. Haven't read the whole thread yet, guessing it's going to be a lot of nonsense and misunderstanding again

[06:55:00.0000] <gsnedders>
"How do you work this [computer]? You go click! Oh fuck it, why doesn't it work, I thought if you put the thing [cursor] there [over text field] it would go there!"

[06:58:00.0000] <gsnedders>
/me sighs

[06:58:01.0000] <gsnedders>
being around my mother using a computer is stressful.

[06:59:00.0000] <hsivonen>
I want to complain about HotSpot's 8000-byte limit, but you've heard it already

[06:59:01.0000] <hsivonen>
seems crazy to have to tweak code around a magic number like that

[07:02:00.0000] <jgraham>
hsivonen: I promise to say nothing about proposals about charset declerations only working in the first 512/1024/some other fixed number of bytes

[07:03:00.0000] <jgraham>
;)

[07:03:01.0000] <hsivonen>
jgraham: that's totally different!

[07:04:00.0000] <Philip`>
Is there no command-line option to change HotSpot's behaviour?

[07:05:00.0000] <hsivonen>
Philip`: -XX:-DontCompileHugeMethods

[07:06:00.0000] <hsivonen>
Philip`: but requiring library users to know about a flag like that would not be cool

[07:06:01.0000] <hsivonen>
I wonder if AppEngine has a limitation like that

[07:07:00.0000] <hsivonen>
/me starts refactoring the tokenizer loop for better perf, line numbers in C++ and better localizability

[07:07:01.0000] <gsnedders>
http://codingforums.com/showthread.php?t=167485 — anyone help?

[07:08:00.0000] <hsivonen>
oh, and CRLF and \0 suck too

[07:11:00.0000] <hsivonen>
maybe I can start an HTML optimization meme about how much faster HTML parser if it has no CR characters and has LFs instead

[07:12:00.0000] <hsivonen>
s/parser/parses/

[07:12:01.0000] <jgraham>
gsnedders: Don't know but you are describing a partial solution rather than a problem. If you describe a problem you may get better help

[07:12:02.0000] <gsnedders>
jgraham: The problem is calling a function on click except when the click falls on an element with a default action!

[07:13:00.0000] <jgraham>
gsnedders: I doubt it. That's part of a solution to some other problem

[07:13:01.0000] <gsnedders>
jgraham: I want to be able to click on a block to hide and show content.

[07:38:00.0000] <hsivonen>
w00t! refactoring error messages saves hundreds of byte codes

[07:38:01.0000] <hsivonen>
(innocent string appends compile into a lot of byte codes)

[07:39:00.0000] <ezyang>
gsnedders: Nonsense! 

[07:39:01.0000] <gsnedders>
ezyang: Yes, I'm a bad little boy.

[07:39:02.0000] <ezyang>
gsnedders: More seriously, if there ever is a failing test-case in Tokenizer, it's nice to know that it actually is Tokenizer's fault, and not InputStream's. A test-suite goes a way to do this 

[07:39:03.0000] <gsnedders>
ezyang: Indeed

[07:40:00.0000] <gsnedders>
ezyang: Some of it was really not very fun to debug while splitting it out with obscure bugs in the input stream

[07:40:01.0000] <ezyang>
Yeah. Kudos on getting Parse Errors to mostly work 

[07:41:00.0000] <ezyang>
Like, that's some pretty in-depth work you did 

[07:41:01.0000] <ezyang>
(also, I'm ignoring parse errors completely for TreeConstructer, so you get to do it again :-) 

[07:41:02.0000] <gsnedders>
ezyang: But I couldn't decide how to do a test suite for something where we weren't using JSON, etc. for input. We'd probably end up disagreeing how to do it. :P

[07:42:00.0000] <gsnedders>
ezyang: Have you merged back into trunk yet?

[07:42:01.0000] <ezyang>
Yeah, it's all in the trunk 

[07:42:02.0000] <gsnedders>
/me ought to do revision for his computing exam

[07:42:03.0000] <ezyang>
Re test suite: I dunno; I think that the class is simple enough that pure PHP works 

[07:42:04.0000] <ezyang>
I thought you finished your exams? 

[07:42:05.0000] <gsnedders>
Actually, learning for the first time some of it.

[07:43:00.0000] <gsnedders>
No, computing next Thursday.

[07:50:00.0000] <hsivonen>
form feeds are almost as bad as CRs

[07:57:00.0000] <hsivonen>
http://www.builderau.com.au/news/soa/Google-Chrome-gets-HTML-video-support/0,339028227,339296704,00.htm

[08:01:00.0000] <Lachy>
hsivonen, I wouldn't object to the validator issuing warnings about the use of CR or CRLF.

[08:01:01.0000] <Lachy>
it's unfortunate, though, that nothing you do could ever lead to CR being abolished entirely :-(

[08:02:00.0000] <jgraham>
Lachy: Windows users wouldn't like that

[08:03:00.0000] <Lachy>
jgraham, only Notepad users would have serious problems with it. But they have bigger issues to worry about anyway.

[08:03:01.0000] <hsivonen>
Lachy: yeah, it's unfortunate :-(

[08:04:00.0000] <hsivonen>
to make speculative parsing work sanely, the only solution I've come up with makes CR slower than LF

[08:04:01.0000] <jgraham>
Also, "Consider use cases" doesn't convey the spirit of "pave the cowpaths" at all

[08:04:02.0000] <Lachy>
I wonder what it would take to get Microsoft to start migrating to the use of LF only

[08:04:03.0000] <Lachy>
jgraham, yes it does, since pave the cowpaths is entirely about use cases

[08:05:00.0000] <Lachy>
well, at least, it's meant to be

[08:05:01.0000] <jgraham>
Lachy: No it's about solutions

[08:05:02.0000] <Lachy>
no it's not!

[08:05:03.0000] <jgraham>
It is. 

[08:05:04.0000] <hsivonen>
hmm. I think I'm not going to support speculative parsing and coercion into XML infoset in the same parser instance

[08:06:00.0000] <jgraham>
Lachy: There should probably be a principle like "Work from use cases" but it doesn't mean anything like "pave the cowpaths"

[08:06:01.0000] <othermaciej>
I agree with jgraham 

[08:06:02.0000] <othermaciej>
"Work from use cases" should be a principle

[08:07:00.0000] <othermaciej>
"solve real problems" is meant to convey that idea but its title is needlessly confrontational

[08:07:01.0000] <Lachy>
jgraham, it's about looking at what authors are trying to do and providing solutions, possibly based on the existing solution, or providing a better solution

[08:07:02.0000] <othermaciej>
"pave the cowpaths" is all about "if authors do X a lot in markup, then it's probably a good idea to enable that, or something close to it, as a feature"

[08:07:03.0000] <Lachy>
othermaciej, that's not what it's meant to be.

[08:07:04.0000] <othermaciej>
"as opposed to making up a wildly different solution to the same thing"

[08:08:00.0000] <othermaciej>
well, I put it in the Design Principles document, back when it was a wiki page

[08:08:01.0000] <jgraham>
Lachy: That is part of it but "pave the cowpaths" implies that common practice should be given more consideration than it would if it was not in common practice

[08:08:02.0000] <othermaciej>
I would like to think I knew what I was saying

[08:08:03.0000] <othermaciej>
it's inspired by the microformats principle of the same name

[08:08:04.0000] <jgraham>
e.g. we would never invent <br/> if it wasn't already common practice but since authors want to do it, we allow it

[08:09:00.0000] <Lachy>
yeah, and I remember tantek coming in here once and insisting that cowpaths was about use cases, and that that's how it's applied to microformats

[08:10:00.0000] <othermaciej>
microformats wiki says:

[08:10:01.0000] <othermaciej>
"Remember, we're paving the cowpaths- before you do that you have to find the cowpaths. Your examples should be a collection of real world sites and pages which are publishing the kind of data you wish to structure with a microformat. From those pages and sites, you should extract markup examples and the schemas implied therein, and provide analysis."

[08:10:02.0000] <jgraham>
Lachy: It is about use cases in the sense that a cow path must be something that people want to do

[08:10:03.0000] <jgraham>
and already do do

[08:10:04.0000] <Lachy>
http://krijnhoetmer.nl/irc-logs/whatwg/20070812#l-86

[08:10:05.0000] <othermaciej>
HTML5 also considers use cases that aren't anything anyone does

[08:10:06.0000] <Lachy>
jgraham, yes

[08:10:07.0000] <othermaciej>
those would not be a "cow path"

[08:11:00.0000] <annevk2>
http://microformats.org/wiki/process#Document_Current_Behavior

[08:11:01.0000] <othermaciej>
(or rather, aren't anything anyone does, because you can't yet)

[08:11:02.0000] <jgraham>
Lachy: But it explicitly favours adopting solutions that are close to the de-facto solutions even if they are not the ones that you would design with a clean-slate approach

[08:12:00.0000] <othermaciej>
anyway, I would be happy to remove the exact phrase "pave the cowpaths" if only so I never have to hear accessibility enthusiasts argue that some markup feature is or isn't a cowpath

[08:12:01.0000] <Lachy>
jgraham, to the extent that the existing solutions are not significantly problematic, yes.

[08:13:00.0000] <othermaciej>
but I think the idea should be captured

[08:13:01.0000] <othermaciej>
and it's not the same as "work from use cases"

[08:13:02.0000] <jgraham>
Lachy: Of course. "Favours" is not absolute. It is, as always, a trade off

[08:14:00.0000] <jgraham>
othermaciej: I agree that e should remove the phrase "pave the cowpaths". Whilst it seems intuitive to me what it implied by such a principle it seems like it creates a great deal of confusion

[08:14:01.0000] <jgraham>
Maybe it should be captured by a principle like "Support common practice" or something

[08:15:00.0000] <Lachy>
How about "Analyse Existing Practices"

[08:15:01.0000] <Lachy>
or what jgraham suggested

[08:15:02.0000] <jgraham>
I was just about to suggest "investigate existing practice"

[08:16:00.0000] <jgraham>
:)

[08:16:01.0000] <Lachy>
either of those work

[08:18:00.0000] <jgraham>
Maybe with the principle saying something like "[...] where the existing solution does not have significant drawbacks consider adopting it rather than forbidding it or creating something new"

[08:19:00.0000] <Lachy>
that's close to Don't Reinvent the Wheel

[08:20:00.0000] <jgraham>
Lachy: You would need more text at the start to explain that you have to investigate the ways that things are already being done

[08:21:00.0000] <jgraham>
(my main point was that an explicit disclaimer that solutions with big problems should not be adopted wholesale)

[08:21:01.0000] <Lachy>
ok

[08:29:00.0000] <Lachy>
Something like this might work as part of the description "Investigate existing, de-facto solutions to problems and evaluate them in regards to the use cases being addressed. Consider either adopting or developing solutions based on the existing practices."

[08:30:00.0000] <Lachy>
(in addition to something like what jgraham suggested)

[08:33:00.0000] <annevk2>
Don't Reinvent the Wheel is more about impl

[08:36:00.0000] <annevk2>
so Google does both Theora and H.264

[08:36:01.0000] <hsivonen>
annevk2: if the story is correct

[08:37:00.0000] <hsivonen>
annevk2: it doesn't make sense to me, though.

[08:37:01.0000] <jgraham>
annevk2: pointer?

[08:37:02.0000] <annevk2>
see inbox

[08:37:03.0000] <jgraham>
Oh, interesting

[08:38:00.0000] <jgraham>
It makes no sense to me either

[08:38:01.0000] <annevk2>
http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2009-May/019992.html

[08:38:02.0000] <annevk2>
no sense how?

[08:39:00.0000] <Lachy>
annevk2, don't reinvent the wheel isn't about reusing implementations. It's about reusing existing solutions if they work

[08:39:01.0000] <hsivonen>
annevk2: if they don't think Theora is the kind of risk that Apple and Nokia claim, why not ship Theora only and switch YouTube to Theora?

[08:40:00.0000] <annevk2>
Lachy, I didn't say that

[08:40:01.0000] <Lachy>
then I don't understand what you meant by "Don't Reinvent the Wheel is more about impl"

[08:41:00.0000] <jgraham>
hsivonen: Possibly because H.264 would have better quality/bandwidth so they would save bandwidth costs exceeding the cost of licensing H.264

[08:41:01.0000] <Philip`>
hsivonen: Because it's lower quality and rarely used and doesn't have hardware support, perhaps?

[08:41:02.0000] <annevk2>
and then they'd need three versions of each video

[08:41:03.0000] <annevk2>
well, two, I guess

[08:41:04.0000] <jgraham>
hsivonen: Although it still requires that they ship a H.264 implementation with chrome

[08:41:05.0000] <hsivonen>
Philip`: rarely used would not apply if YouTube switched :-)

[08:41:06.0000] <jgraham>
annevk2: That seems like much less of a problem

[08:42:00.0000] <Philip`>
hsivonen: It would if you count the number of people who are familiar with the process of encoding to Theora :-)

[08:42:01.0000] <annevk2>
jgraham, hah, they get 20 hours of content every minute

[08:42:02.0000] <hsivonen>
annevk2: YouTube already has 3 per video: flv, H.264 and .3gp

[08:42:03.0000] <annevk2>
hsivonen, 3gp? isn't flv just streaming the H.264?

[08:42:04.0000] <jgraham>
I would be somewhat unsurprised if Youtube offered Theora to theora-only browsers

[08:43:00.0000] <hsivonen>
annevk2: there's traditional YouTube (flv), there's "HD" (h.264) and there's non-iPhone mobile (3gp)

[08:43:01.0000] <hsivonen>
3gp is MPEG-4 Simple Profile in small size

[08:43:02.0000] <Philip`>
The BBC iPlayer seems to have ten versions of some videos

[08:43:03.0000] <annevk2>
i thought they were in the process of converting the traditional stuff to h264

[08:43:04.0000] <annevk2>
but okay, I guess we'll see :)

[08:44:00.0000] <hsivonen>
annevk2: oh. I don't know about that.

[08:44:01.0000] <Philip`>
(http://beebhack.wikia.com/wiki/IPlayer_TV#Comparison_Table)

[08:45:00.0000] <Lachy>
hsivonen, there's actually standard quality, high quality, HD, plus the mobile phone versions

[08:45:01.0000] <Philip`>
(They probably don't get more than about a minute of video per minute, though)

[08:45:02.0000] <hsivonen>
it's amazing that there's enough compute power in the world to encode all those videos

[08:46:00.0000] <annevk2>
http://twitter.com/circa1977/statuses/1960304218 -- "... HTML will always be XML subset." look, someone is wrong on the internet!

[08:46:01.0000] <Lachy>
I assume YouTube also keeps the original uploaded videos around too, for any future conversions

[08:46:02.0000] <hsivonen>
I wonder if they are shipping both in order to demonstrate their ability to jettison h.264 to MPEG-LA

[08:47:00.0000] <hsivonen>
anyway, very cool that they'll ship theora

[08:47:01.0000] <Philip`>
hsivonen: It sounds like you seriously underestimate the amount of compute power in the world :-)

[08:48:00.0000] <hsivonen>
/me wonders what the carbon footprint of YouTube is

[08:49:00.0000] <Philip`>
hsivonen: If it's 20 hours of video per minute, you only need 1200 machines doing real-time conversion (and I guess you can do better than real-time on modern CPUs), multiplied by however many versions of videos you've got

[08:49:01.0000] <annevk2>
I guess H.264 won't be done via FFmpeg

[08:49:02.0000] <hsivonen>
Philip`: re-encoding the back catalog is a lot of minutes

[08:51:00.0000] <Philip`>
hsivonen: I wonder if they only bother converting the relatively popular videos to newer formats

[08:51:01.0000] <Philip`>
(I guess there are lots of videos that had 12 views a year ago and have never been looked at since)

[08:53:00.0000] <jgraham>
There are a rather large number of videos that are audio + a static picture (e.g. some copyright-infringing music "videos")

[08:53:01.0000] <Philip`>
/me would kind of guess that Google has lots of spare CPU capacity since most of its work will be limited by network I/O instead

[08:53:02.0000] <Philip`>
though I could be totally wrong

[08:54:00.0000] <Philip`>
jgraham: It's weird that a video sharing site is the most popular way to share music

[08:54:01.0000] <hsivonen>
I wonder if it's possible to recode lazily on demand

[08:55:00.0000] <Lachy>
Philip`, it's probably because there are no popular social networking sites built around the idea of publishing and listening to audio

[08:57:00.0000] <Philip`>
hsivonen: As in real-time encoding and streaming? That sounds like it'd be entirely incompatible with their usual architecture of serving videos as plain files from caching HTTP servers, so I guess it'd be quite a pain to do

[09:00:00.0000] <Lachy>
I doubt on-demand encoding would work given the number of simultaneous views they get across all their videos.

[09:00:01.0000] <Lachy>
but they would probably prioritise re-encoding of the back catalogue based on popularity or some other metrics

[09:00:02.0000] <Philip`>
Lachy: I assume the idea is they could recode the video on demand when somebody first views it, and then save the recoded output to send to any other viewers

[09:01:00.0000] <Philip`>
which would mean everybody would be able to see the recoded version of every video with no delay, without requiring a giant batch conversion before switching to the new format

[09:05:00.0000] <Philip`>
Streaming video seem to be the kind of thing that telecom companies care a lot about, so they want to invest in high-bandwidth high-reliability low-latency links and Quality of Service and multicast and all sorts of proprietary protocols and everything, but then consumers don't care about any of that and just download video files over HTTP and wait a few seconds until it's buffered enough to start playing smoothly

[09:07:00.0000] <hsivonen>
telcos care about QoS a lot more than justified by their customers' willingness to pay for it

[09:15:00.0000] <Philip`>
/me is not yet sure to what extent QoS is a real issue for the internet, and how much is just outdated Bellhead vs Nethead ideas

[09:21:00.0000] <mgrdcm>
hsivonen:  one reason telcos care about it is to give their own VoIP traffic priority.  they care about it within their own networks at least.

[09:22:00.0000] <mgrdcm>
/me used to work on big telco traffic quality monitoring software

[09:23:00.0000] <hsivonen>
mgrdcm: my point is that telcos fret about Erland formulas and circuit switching when the casual customer is happy with the Skype level of uncertainty

[09:26:00.0000] <mgrdcm>
hsivonen:  for voice, though, i think customers' expectations of quality are higher when provided by a "phone company", even if it is over IP just like they'd be getting from vonage.

[09:29:00.0000] <hsivonen>
what does Vonage do for 911 calls?

[09:30:00.0000] <ezyang>
gsnedders: Houston, we have a problem. 

[09:30:01.0000] <Philip`>
/me 's university switched to VOIP last year, and every few months there are data network outages that cause the phone system to fail for several hours

[09:31:00.0000] <ezyang>
/me 's university is switching to VOIP, but none of the students use the phones anyway... 

[09:32:00.0000] <jgraham>
Philip`: Very locally that worked nicely for me since you couldn't hear anyone on the analouge phone that was in our office (it was too quiet)

[09:32:01.0000] <jgraham>
(and had a many-metre extension cable)

[09:32:02.0000] <Philip`>
(The phone system has actually been *less* reliable than the data network, since my building has a backup link for data but the phone is on a different VLAN that apparently won't work over the backup link)

[09:33:00.0000] <jgraham>
(so being able to hear people sometimes was overall a worthwhile gain)

[09:34:00.0000] <Philip`>
Analogue phones have the advantage that you don't have to wait a minute for them to reboot and acquire an IP address whenever there's a problem or whenever they get unplugged

[09:35:00.0000] <Philip`>
But when the VOIP phones work, they seem to work fine and they have new features and stuff, so I guess that'd good

[09:35:01.0000] <Philip`>
s/'d/'s/

[09:37:00.0000] <ezyang>
Do you have the feature where you can have voicemail emailed to you as an MP3? 

[09:39:00.0000] <gsnedders>
/me notes that would mean paying the fees to encode an MP3

[09:39:01.0000] <Philip`>
ezyang: Not sure, but there is a web page where you can listen the voicemail as an .au file or something

[09:40:00.0000] <ezyang>
Basically the same 

[09:40:01.0000] <Philip`>
(That's actually the only way I know how to listen to voicemail - I'm sure there must be some feature on the phone itself to do that, but I haven't figured it out)

[09:40:02.0000] <ezyang>
gsnedders: So, you know how we're optimizing tokenizer by globbing as many characters as we can? Well, it's actually kind of important to keep whitespace separated 

[09:40:03.0000] <Philip`>
(But I've only had one voicemail message in the past year, so I haven't cared a lot)

[09:40:04.0000] <gsnedders>
ezyang: I know

[09:41:00.0000] <gsnedders>
ezyang: But only when we move in or out of the data state, IIRC

[09:41:01.0000] <Philip`>
I imagine you don't want to split on whitespace between words inside an element, because that'd hurt performance a lot

[09:42:00.0000] <ezyang>
Yeah... I'm trying to find which one is causing the test case to fail 

[09:42:01.0000] <Philip`>
/me isn't sure how Python html5lib handles this (if it handles it at all)

[09:42:02.0000] <ezyang>
But if you have: <!DOCTYPE html><script> <!-- </script> --> </script> EOF 

[09:42:03.0000] <ezyang>
(note space between </script> and EOF) 

[09:42:04.0000] <gsnedders>
Philip`: It only gives WhitespaceCharacters when moving into the data state, IIRC

[09:42:05.0000] <ezyang>
The whitespace gets placed in <head>, but EOF gets placed in <body> 

[09:43:00.0000] <gsnedders>
/me notes it is quite likely his memory is wrong

[09:44:00.0000] <ezyang>
Ok, this is definitely data' fault; my fix was just wrong 

[09:46:00.0000] <ezyang>
What does HTML5 define as whitespace? 

[09:46:01.0000] <gsnedders>
We should probably have a constant for that

[09:46:02.0000] <gsnedders>
but from memory: U+0009, U+000A, U+000C, U+000D, U+0020

[09:47:00.0000] <hsivonen>
gsnedders: yes

[09:47:01.0000] <ezyang>
Hmm. U+000B isn't one of them? 

[09:47:02.0000] <gsnedders>
Nope.

[09:47:03.0000] <ezyang>
Ok, so, like, all of our code is wrong :-) 

[09:47:04.0000] <hsivonen>
ezyang: not anymore

[09:47:05.0000] <gsnedders>
ezyang: Tok doesn't allow it

[09:47:06.0000] <ezyang>
Ah, that makes sense 

[09:48:00.0000] <ezyang>
So, in TreeConstructer, we've got loads and loads of  preg_match('/^[\t\n\x0b\x0c ]+$/' 

[09:48:01.0000] <ezyang>
I should to a global search replace at some point 

[09:48:02.0000] <ezyang>
In favor of a constant 

[09:48:03.0000] <gsnedders>
Ah, in TreeConstructor

[09:48:04.0000] <ezyang>
Oh yeah, that's bugging me to 

[09:48:05.0000] <ezyang>
We should rename the class 

[09:48:06.0000] <ezyang>
*too 

[09:48:07.0000] <gsnedders>
Also: remove pcre dependancy.

[09:49:00.0000] <ezyang>
I'll let you do that, since you're the resident expert :-) 

[09:49:01.0000] <gsnedders>
Me!?

[09:49:02.0000] <hsivonen>
seems like you are using much higher-level constructs in the tokenizer than I am

[09:49:03.0000] <gsnedders>
hsivonen: This isn't tokenizer there.

[09:49:04.0000] <gsnedders>
hsivonen: We're about the same level in the tokenizer as Python

[09:51:00.0000] <ezyang>
gsnedders: InputStream takes care of \r, so I don't need to check for it, right? 

[09:51:01.0000] <gsnedders>
ezyang: What about &#x0d; ?

[09:52:00.0000] <ezyang>
Oh, yeah, good point 

[09:53:00.0000] <Philip`>
Doesn't &#x0d; get replaced in the tokeniser?

[09:53:01.0000] <ezyang>
Huh. Fixing that introduced three more fails 

[09:53:02.0000] <gsnedders>
Philip`: dunno

[09:54:00.0000] <Philip`>
http://www.whatwg.org/specs/web-apps/current-work/multipage/syntax.html#tokenizing-character-references says replace 0x0D with U+000A

[09:54:01.0000] <gsnedders>
Oh well, OK

[09:54:02.0000] <gsnedders>
ezyang: You don't need to check for it.

[10:00:00.0000] <gsnedders>
ezyang: Where about are you working atm?

[10:00:01.0000] <ezyang>
I'm smashing tree builder bugs 

[10:00:02.0000] <ezyang>
As long as TreeBuilderTest.php keeps working, you can do whatever you want to Tokenizer/InputStream 

[10:00:03.0000] <ezyang>
Let me just commit and push my recent changes 

[10:00:04.0000] <gsnedders>
Uh, yeah, I may end up touching TreeBuilderTest :)

[10:01:00.0000] <hsivonen>
http://en.wikipedia.org/w/index.php?title=HTML_5&diff=293104533&oldid=prev

[10:01:01.0000] <ezyang>
As long as it keeps working 

[10:01:02.0000] <ezyang>
I'm not really editing that file per se. Just using it. 

[10:02:00.0000] <ezyang>
Question: Why does the </cite> in <b>A<cite>B<div>C</cite>D *not* close the <div>? 

[10:03:00.0000] <ezyang>
<cite> is not mentioned anywhere in the spec, so it should close the div... 

[10:03:01.0000] <gsnedders>
"Please leave your sense of logic at the door, thanks!"

[10:04:00.0000] <ezyang>
No, this is w.r.t. the spec 

[10:04:01.0000] <ezyang>
AFAICT, the spec mandates that <div> get closed 

[10:04:02.0000] <ezyang>
But the test-case asserts differently 

[10:04:03.0000] <ezyang>
Oh, my algo is wrong. Savvy 

[10:05:00.0000] <Philip`>
"TreeConstructer.php"?!

[10:05:01.0000] <gsnedders>
Philip`: We didn't decide to call it that!

[10:06:00.0000] <ezyang>
Yeah, that's what the original was named 

[10:06:01.0000] <ezyang>
and I didn't feel like renaming it without a bunch of tests first 

[10:06:02.0000] <gsnedders>
Where has Jeroen gone? I haven't heard from him in a while

[10:07:00.0000] <Philip`>
You should rename it before anyone starts relying on it :-)

[10:07:01.0000] <Philip`>
/me is reminded of http://blogs.msdn.com/oldnewthing/archive/2008/05/19/8518565.aspx

[10:07:02.0000] <gsnedders>
Philip`: There are uglier things

[10:07:03.0000] <ezyang>
Philip`: aye-aye, sir 

[10:07:04.0000] <Philip`>
gsnedders: Like the use of PHP?

[10:07:05.0000] <gsnedders>
Philip`: For example

[10:08:00.0000] <Philip`>
You really should have written it in Haskell instead

[10:08:01.0000] <ezyang>
Hahaha 

[10:08:02.0000] <ezyang>
I mean, I'm totaly learning Haskell right now 

[10:08:03.0000] <ezyang>
*totally 

[10:08:04.0000] <gsnedders>
Philip`: Or Referer in HTTP

[10:08:05.0000] <gsnedders>
/me had learning Haskell on his to-do list for last summer

[10:08:06.0000] <gsnedders>
I never got to the end of the first item over the summer

[10:09:00.0000] <ezyang>
Haskell is totally worth skipping the rest of your list for. 

[10:09:01.0000] <ezyang>
Catamorphisms yum! 

[10:10:00.0000] <gsnedders>
So, InputStream tests…

[10:11:00.0000] <gsnedders>
What do I want to extend for the class?

[10:11:01.0000] <Philip`>
Python html5lib has a few you could steal, but I'm not sure how useful or appropriate or extensive they are

[10:11:02.0000] <ezyang>
Hmm... so it looks like foster parented elements don't get active formatting elements applied to them. 

[10:11:03.0000] <ezyang>
Oh wait they do. 

[10:11:04.0000] <gsnedders>
ezyang: I guess not HTML5_TestDataHarness…

[10:12:00.0000] <gsnedders>
ezyang: UnitTestCase?

[10:12:01.0000] <ezyang>
Hmm? 

[10:12:02.0000] <ezyang>
Oh yeah, UnitTestCase is what you want 

[10:12:03.0000] <ezyang>
that and $this->assertIdentical() are probably all you need 

[10:12:04.0000] <ezyang>
maybe a setUp() or two 

[10:12:05.0000] <ezyang>
/me does sit ups 

[10:12:06.0000] <gsnedders>
/me hasn't used SimpleTest before

[10:13:00.0000] <gsnedders>
/me wishes it did code coverage…

[10:14:00.0000] <ezyang>
Does PHPUnit do code coverage these days? 

[10:14:01.0000] <gsnedders>
Has for ages.

[10:15:00.0000] <gsnedders>
/me wishes there was a generic code-coverage tool for PHP that worked with any script

[10:15:01.0000] <ezyang>
YEah, it's called XDebug 

[10:15:02.0000] <gsnedders>
That doesn't create anything nice to look at, just arrays

[10:16:00.0000] <ezyang>
Well, it's like a really simple PHP script to get some nice output 

[10:16:01.0000] <ezyang>
(just someone has to write it) 

[10:16:02.0000] <gsnedders>
/me doesn't remember it being overly simple

[10:16:03.0000] <gsnedders>
I started to write something, then realized it was more than I could be bothered to write

[10:17:00.0000] <ezyang>
Let me bang something out 

[10:19:00.0000] <gsnedders>
Do we want one test per method?

[10:20:00.0000] <ezyang>
Yep. 

[10:20:01.0000] <ezyang>
Pick expressive method names 

[10:26:00.0000] <ezyang>
Done 

[10:26:01.0000] <gsnedders>
Done what?

[10:26:02.0000] <ezyang>
With the highlight script 

[10:26:03.0000] <gsnedders>
ah

[10:26:04.0000] <ezyang>
Let me stick in GitHub or something 

[10:26:05.0000] <ezyang>
So you can try it out. 

[10:27:00.0000] <Philip`>
Hmm, excellent, new versions of MySQL think width-140 with unsigned smallint width=80 should result in 18446744073709551556

[10:27:01.0000] <Philip`>
(compared to a much older version that thought it was some crazy like -60)

[10:29:00.0000] <ezyang>
Use case is to do the code coverage, and then save it as a .ser file 

[10:29:01.0000] <ezyang>
http://github.com/ezyang/xdebug-highlight/tree/master 

[10:29:02.0000] <ezyang>
I haven't tested it myself 

[10:31:00.0000] <gsnedders>
ezyang: That doesn't break down into coverage % for functions, etc. :P

[10:31:01.0000] <hsivonen>
Hixie: how did http://dev.w3.org/html5/spec/Overview.html#declarative-3d-scenes end up in the spec?

[10:32:00.0000] <ezyang>
Oh, you want that info 

[10:32:01.0000] <ezyang>
Yeah, then I officially don't care. 

[10:32:02.0000] <ezyang>
Port our test-cases to PHPUnit or something 

[10:32:03.0000] <gsnedders>
ezyang: Uh, that implies effort.

[10:33:00.0000] <ezyang>
You're working on html5lib. Come on, mate. 

[10:33:01.0000] <gsnedders>
:P

[10:33:02.0000] <gsnedders>
ezyang: "Please leave your sense of logic at the door, thanks!"

[10:33:03.0000] <ezyang>
Anyway, I'm a SimpleTest dev, so I can fix problems and stuff. 

[10:33:04.0000] <ezyang>
Also, SimpleTest has an experimental code coverage branch 

[10:34:00.0000] <ezyang>
I've never used it though. 

[10:36:00.0000] <Philip`>
hsivonen: There was some discussion about 3d <canvas> years and years ago, and someone mentioned X3D in there, so I guess that's what resulted in it being mentioned in the spec

[10:37:00.0000] <Philip`>
(Some of the X3D people seem to be misinterpreting it as stating that X3D is the official solution for embedding 3D in HTML5)

[10:37:01.0000] <hsivonen>
has any browser vendor shown interest in X3D?

[10:38:00.0000] <Philip`>
No, as far as I'm aware

[10:38:01.0000] <hsivonen>
Mozilla, Google and Opera have all tried something else

[10:38:02.0000] <Philip`>
All the current X3D support is just in plugins (or standalone applications), I believe

[10:39:00.0000] <Philip`>
but various X3D people want more integration with the browser

[10:39:01.0000] <Philip`>
e.g. linking it directly to the DOM inside an XHTML page, like with SVG

[10:43:00.0000] <Philip`>
(Maybe it could be useful in the same way that having both SVG and 2d <canvas> is useful, since they're appropriate for different situations)

[10:44:00.0000] <Philip`>
(Alternatively, maybe it could be a waste of effort in the same that having both SVG and 2d <canvas> is a waste of effort, since there's significant overlap)

[10:48:00.0000] <ezyang>
Subtle: if we reconstruct an active formatting element, *that* becomes the foster parent. I wonder where the python implementation special cases this. 

[10:52:00.0000] <Dashiva>
So if you build a paved road, and later a few lonely cows start walking on it... does it even make sense to call it a cowpath?

[10:55:00.0000] <Philip`>
What if you can't see any cows at all, but the road is covered in cow pats?

[10:56:00.0000] <ezyang>
Is it a cow if no one sees it? 

[10:56:01.0000] <Philip`>
ezyang: Yes

[10:58:00.0000] <Dashiva>
Well, I'm just wondering because there seem to be a few people saying "<existing spec feature> is a cowpath because some people use it"

[11:04:00.0000] <gsnedders>
Dashiva: But cows are cute!

[11:04:01.0000] <gsnedders>
We must keep them!

[11:08:00.0000] <Philip`>
gsnedders: That's the main reason I use Gentoo

[11:08:01.0000] <Philip`>
"emerge moo" prints a nice picture of the mascot, Larry the Cow

[11:10:00.0000] <Dashiva>
Philip`: You linked that bytes-to-encoding image again, which reminded me that you were going to make one with a more sensible y axis :)

[11:10:01.0000] <Philip`>
Was I?

[11:11:00.0000] <Dashiva>
Logarithmic in the percentage of non-working sites, I think

[11:11:01.0000] <Philip`>
You mean like http://philip.html5.org/data/encoding-detection-loglog.svg or something?

[11:14:00.0000] <Dashiva>
Yeah, that

[11:15:00.0000] <Dashiva>
I guess I just forgot you had made it :)

[11:16:00.0000] <Philip`>
How could you forget? It wasn't even 15 months ago

[11:16:01.0000] <Philip`>
(I suppose I do have the distinct advantage of being able to see the directory listing...)

[11:17:00.0000] <Dashiva>
Maybe I assumed you would've linked that one if it existed, even though it's just my personal taste saying it's better

[11:17:01.0000] <Philip`>
The other graph does a better job of misrepresenting the data so that 512 bytes looks like a sensible figure to choose

[11:18:00.0000] <Dashiva>
One thing I've wondered is, does it really make a difference to wait until 1024 bytes?

[11:18:01.0000] <Dashiva>
Isn't that usually within the first packet anyhow?

[11:18:02.0000] <Philip`>
Usually, but not close to always

[11:19:00.0000] <Philip`>
http://krijnhoetmer.nl/irc-logs/whatwg/20090528#l-1107

[11:20:00.0000] <Philip`>
(Packets are usually a bit under 1500 bytes, I think)

[11:21:00.0000] <Dashiva>
Hmm, so if the header is 400+ you might run past the first packet

[11:21:01.0000] <gsnedders>
ezyang: Right, so I think we basically need to have our own UTF-8 decoder, but we can skip it if we have an up-to-date copy of PCRE with Unicode or a sane iconv, _and_ we have PCRE.

[11:22:00.0000] <ezyang>
Sounds like it. hsivonen has one 

[11:22:01.0000] <gsnedders>
I have one too.

[11:22:02.0000] <gsnedders>
I have an MIT licensed one, which is more useful for html5lib than hsivonen's

[11:23:00.0000] <Dashiva>
My internet connection is too fast for me to be able to care about the difference in one and two packets, I suppose :)

[11:25:00.0000] <gsnedders>
ezyang: The only reason we need PCRE always for when we don't need to use our own UTF-8 decoder is what is currently line 130 of inputstream

[11:25:01.0000] <ezyang>
I've seen. It's quite an impressive regex you have there. 

[11:26:00.0000] <gsnedders>
Yeah. It was fun to write. </sarcasm>

[11:26:01.0000] <Philip`>
Dashiva: Latency matters more than bandwidth, and I assume your internet connection still suffers from latency when accessing data on the other side of the world

[11:26:02.0000] <Philip`>
although I don't how much it actually matters in reality

[11:26:03.0000] <Philip`>
because I don't really have any idea how TCP works

[11:26:04.0000] <Philip`>
(Will it send lots of response packets at once, rather than doing the first and waiting for an ACK?)

[11:26:05.0000] <gsnedders>
ezyang: I used to require PCRE with UTF-8 support there, but that says any string containing low surrogates is invalid, which doesn't work when we want to detect them.

[11:27:00.0000] <gsnedders>
(It's also cheaper not having to decode it)

[11:27:01.0000] <ezyang>
Awesome. 

[11:27:02.0000] <Dashiva>
Philip`: It will send several, yes

[11:28:00.0000] <Philip`>
I guess this has something to do with the default window size

[11:28:01.0000] <sr0unet>
Hello.

[11:28:02.0000] <gsnedders>
ezyang: (Most of the test case failures I have so far come from using iconv!)

[11:28:03.0000] <Dashiva>
Yeah

[11:28:04.0000] <sr0unet>
Is this a WPF channel ?

[11:28:05.0000] <gsnedders>
s/most/all/

[11:28:06.0000] <Dashiva>
That's how far ahead it's willing to go, as I recall

[11:28:07.0000] <Dashiva>
sr0unet: What's WPF? Wikipedia foundation?

[11:28:08.0000] <sr0unet>
^^

[11:28:09.0000] <Dashiva>
Then no

[11:28:10.0000] <ezyang>
As in, we need to emit parse errors and iconv doesn't do that? 

[11:28:11.0000] <Philip`>
Walrus protection fund?

[11:28:12.0000] <sr0unet>
WPF C#

[11:28:13.0000] <gsnedders>
ezyang: No

[11:29:00.0000] <gsnedders>
ezyang: Replace invalid sequences with U+FFFD

[11:29:01.0000] <Dashiva>
This is where HTML happens

[11:29:02.0000] <Philip`>
I don't think anyone using an HTML5 parser library is actually going to care about parse errors, so it'd be much easier to not implement them :-)

[11:29:03.0000] <gsnedders>
Philip`: But we already have :P

[11:30:00.0000] <Philip`>
/me should add a no-error mode to Python html5lib so it doesn't have to go through all the bother of tracking line numbers and suchlike

[11:33:00.0000] <ezyang>
aha 

[11:33:01.0000] <ezyang>
Philip`: Actually, people surprisingly do 

[11:43:00.0000] <gsnedders>
Wait, what…

[11:43:01.0000] <gsnedders>
DOM Level 2 Events uses "thru"

[11:45:00.0000] <Dashiva>
To the Errata-mobile

[11:48:00.0000] <gsnedders>
How do you get each member of a NodeList in ECMAScript?

[11:48:01.0000] <gsnedders>
My out of practiceness of JS shows

[11:51:00.0000] <gsnedders>
Can you not set an event handler on DOMActivate?

[11:52:00.0000] <ezyang>
Hey, question for y'all: when the spec says 'Process the token using the rules for the "in head" insertion mode.', do they implicitly want me to temporarily add <head> to the top of the stack? 

[11:52:01.0000] <Dashiva>
gsnedders: .length and iteration?

[11:52:02.0000] <ezyang>
This is section 9.2.5.10 

[11:52:03.0000] <gsnedders>
http://hixie.ch/tests/adhoc/dom/events/DOMActivate/001.html makes it seem you can, but Op 10 fails

[11:52:04.0000] <Dashiva>
DOMActivate is part of mutation, isn't it?

[11:54:00.0000] <gsnedders>
Oh, wait. I realize what I'm doing wrong.

[11:56:00.0000] <gsnedders>
Is it bad I more or less permanently have HTML 5 open?

[11:56:01.0000] <ezyang>
Hmm, it looks like the spec doesn't actually want that 

[11:56:02.0000] <Dashiva>
Philip`: What's your comment to being labeled as not part of the whatwg crowd?

[11:57:00.0000] <hsivonen>
http://lists.w3.org/Archives/Member/tag/2009May/0084.html

[11:59:00.0000] <Dashiva>
/me misses being member of a Member so he could read Member mails

[11:59:01.0000] <Philip`>
Dashiva: Depends on whether that comment came from a cool person i.e. a WHATWG member, or someone else

[12:01:00.0000] <Dashiva>
It was our friend Shelley

[12:01:01.0000] <gsnedders>
Does 'a, img[usemap], video[controls], audio[controls], label, input:not([type=hidden]), button, select, textarea, keygen, details, datagrid, bb, menu[type=toolbar]' match all interactive elements?

[12:04:00.0000] <ezyang>
Ok, I think I see a bug in the test-cases 

[12:04:01.0000] <Philip`>
Dashiva: Hmm, where was that?

[12:04:02.0000] <ezyang>
<head></html><meta><p> should result in <meta> being placed in <body>, since </html> pops us to AFTER_AFTER_BODY 

[12:05:00.0000] <ezyang>
Oh shoot, I'm in the wrong mode 

[12:05:01.0000] <Dashiva>
http://lists.w3.org/Archives/Public/public-html/2009May/0537.html

[12:05:02.0000] <ezyang>
Oh no, it's the same behavior 

[12:05:03.0000] <ezyang>
Awesome. I think this is legit. Anyone else want to verify? 

[12:06:00.0000] <hsivonen>
http://lists.w3.org/Archives/Public/www-tag/2009May/0123.html has interesting bits

[12:07:00.0000] <ezyang>
The question is does <head></html> result in AFTER_AFTER_BODY or IN_HEAD mode 

[12:07:01.0000] <Philip`>
Dashiva: Oh, right

[12:08:00.0000] <hsivonen>
ezyang: V.nu may have a bug in that case

[12:08:01.0000] <Dashiva>
ezyang: What's the non-legit behavior seen elsewhere?

[12:09:00.0000] <ezyang>
The python, V.nu, and test-case think that tihs should result in IN_HEAD 

[12:09:01.0000] <ezyang>
I think it should be AFTER_AFTER_BODY 

[12:09:02.0000] <Dashiva>
Okay

[12:12:00.0000] <Dashiva>
So the </html> is "in head", becomes "after head", becomes "in body", becomes "after body", becomes "after after body". Then the <meta> is sent to "in body", and then sent to "in head"

[12:13:00.0000] <ezyang>
Dashiva: yep. 

[12:14:00.0000] <ezyang>
But the "in body" case doesn't push the head_pointer onto the stack, so meta ends up in body, not html 

[12:14:01.0000] <Dashiva>
But the <meta> ends up in the head

[12:14:02.0000] <ezyang>
s/html/head/ 

[12:14:03.0000] <ezyang>
If meta should end up in head, I think we have a spec bug. 

[12:14:04.0000] <Dashiva>
Oh, you mean in the code?

[12:14:05.0000] <ezyang>
The Python/V.nu code puts it in the head. I think spec puts it in body. 

[12:14:06.0000] <Dashiva>
Not that I can see

[12:15:00.0000] <ezyang>
It's subtle. When you're "in body", it passes it on to "in head" 

[12:15:01.0000] <Dashiva>
Here's how I see it in the spec: You start in "after after body". You follow the step "anything else" which sends you to "in body"

[12:15:02.0000] <Dashiva>
There you follow A start tag token whose tag name is one of: "base", "command", "link", "meta" ... which takes you to "in head"

[12:15:03.0000] <ezyang>
But the top of the stack is still <body>, not <head> 

[12:15:04.0000] <ezyang>
yep and yep 

[12:15:05.0000] <Dashiva>
Ah, I see what you mean

[12:15:06.0000] <ezyang>
We are in agreement. 

[12:16:00.0000] <hsivonen>
ezyang: at some point, the spec said that </body> and </html> were ignored in head

[12:16:01.0000] <ezyang>
Yep. 

[12:17:00.0000] <ezyang>
So, assuming that the spec change is correct, this test-case needs to change 

[12:17:01.0000] <hsivonen>
yes

[12:18:00.0000] <ezyang>
Ok, will do. 

[12:18:01.0000] <ezyang>
I'll also patch the Python implementation for free :-) 

[12:18:02.0000] <Dashiva>
But the intent is still that <meta> should always end up in head, isn't it?

[12:19:00.0000] <Dashiva>
That is, it's a spec bug too

[12:19:01.0000] <ezyang>
Well, we have test-cases that very specifically say <meta> should end up in <body> 

[12:19:02.0000] <Dashiva>
Oh

[12:19:03.0000] <Dashiva>
Microdata, of course

[12:19:04.0000] <Dashiva>
I had things mixed up

[12:21:00.0000] <hsivonen>
Dashiva: it predates microdata. IE and Opera compat

[12:21:01.0000] <hsivonen>
the parsing spec was *not* changed for microdata

[12:21:02.0000] <hsivonen>
/me waves to log readers

[12:21:03.0000] <Dashiva>
But remembering microdata told me which way is the intended one :)

[12:22:00.0000] <ezyang>
Browsers are... special. 

[12:22:01.0000] <ezyang>
Yup yup 

[12:22:02.0000] <Dashiva>
hsivonen: It's a bit worrisome if anyone were to take anything I say as reasonable

[12:22:03.0000] <jgraham>
/me wonders if Philip` has a clever way of making parseerrorless html5lib significantly faster whilst still allowing the possibility of reporting parse errors

[12:23:00.0000] <gsnedders>
/me notes the perf. hit was fairly low even in PHP

[12:23:01.0000] <Dashiva>
jgraham: Would it be parseerrorless if it had that capability?

[12:23:02.0000] <Philip`>
jgraham: Depends on whether 5% counts as significant

[12:23:03.0000] <hsivonen>
jgalvez_: does python compile away empty methods that could be filled in by a subclass but aren't at the time of execution?

[12:24:00.0000] <gsnedders>
Does 'a, img[usemap], video[controls], audio[controls], label, input:not([type=hidden]), button, select, textarea, keygen, details, datagrid, bb, menu[type=toolbar]' match all interactive elements?

[12:24:01.0000] <ezyang>
"two-heads-are-not-better-than-one" HAHAHA 

[12:24:02.0000] <hsivonen>
s/jgalvez_/jgraham/

[12:24:03.0000] <jgraham>
hsivonen: Assuming you meant me, no idea but I doubt it

[12:25:00.0000] <jgraham>
Philip`: How?

[12:26:00.0000] <hsivonen>
jgraham: I'm refactoring code assuming that HotSpot does that

[12:26:01.0000] <Dashiva>
hsivonen: So tag is getting involved in the RDFa side of things?

[12:26:02.0000] <jgraham>
ezyang: Ah, the delicate touch of mpilgrim

[12:26:03.0000] <Philip`>
jgraham: By not bothering to keep track of line numbers

[12:26:04.0000] <hsivonen>
Dashiva: looks like it might

[12:26:05.0000] <Philip`>
jgraham: (I think it saved roughly that much when I last checked)

[12:27:00.0000] <jgraham>
Philip`: So just an if (parse_errors): calculate_line_numbers type thing

[12:27:01.0000] <Philip`>
jgraham: Something like that

[12:27:02.0000] <jgraham>
It would be nice to cut all the method calls to parseError()

[12:27:03.0000] <jgraham>
But that is harder I guess

[12:28:00.0000] <Dashiva>
At least the talk seemed rather calm and reasonable

[12:28:01.0000] <ezyang>
jgraham: Ah, so this is mpilgrim's baby :-) 

[12:28:02.0000] <Dashiva>
No flaming rhetoric from the get-go :)

[12:28:03.0000] <jgraham>
ezyang: mpilgrim tried making a html5lib based validator

[12:28:04.0000] <ezyang>
/me .oO{ghetto} 

[12:28:05.0000] <jgraham>
Which never really went anywhere

[12:28:06.0000] <ezyang>
Ah. 

[12:29:00.0000] <ezyang>
So, sorta like V.nu, except vapourware? 

[12:29:01.0000] <jgraham>
ezyang: sort of

[12:30:00.0000] <Philip`>
ezyang: Not really, since there wasn't any vapour

[12:30:01.0000] <Philip`>
Just some code that checked a few conformance requirements

[12:30:02.0000] <ezyang>
Aha. 

[12:31:00.0000] <ezyang>
(in other news, test-case fix pushed. Rubyistas and Javaistas, fix yer code!) 

[12:31:01.0000] <ezyang>
Maybe I should fix the ruby impl too. 

[12:33:00.0000] <gsnedders>
When was DOM 3 XPath support added to browsers?

[12:42:00.0000] <hsivonen>
gsnedders: March 2002

[12:43:00.0000] <hsivonen>
gsnedders: http://bonsai.mozilla.org/cvslog.cgi?file=mozilla/dom/public/idl/xpath/nsIDOMXPathEvaluator.idl&rev=HEAD&mark=1.3

[12:43:01.0000] <hsivonen>
dunno about other code bases

[12:43:02.0000] <ezyang>
Ugh, fixing the ruby impl increased the fail count by two 

[12:52:00.0000] <hsivonen>
http://twitter.com/jdowdell/status/1961148437

[12:53:00.0000] <ezyang>
Is there any particular reason there's a smattering of test-cases that have error messages like "Unexpected end tag (). Ignored."? 

[12:54:00.0000] <ezyang>
(four, specifically) 

[12:58:00.0000] <jwalden>
judging by this channel, jd is an extremely effective troll

[13:01:00.0000] <Hixie>
don't confuse people being entertained with people being enraged :-)

[13:02:00.0000] <jwalden>
lots of talk for not that much entertainment, to me :-)

[13:02:01.0000] <ezyang>
I'm not poking the ruby implementation any more (although I managed to introduce 10 more failing cases.) >:-( 

[13:24:00.0000] <Wolfman2000>
Morning/afternoon.  www.pumpproedits.com  <-- I present, the newest member of the HTML5 family.

[13:29:00.0000] <ezyang>
tests1.dat -> full passes with PHP implementation 

[13:29:01.0000] <ezyang>
moving on to tests2.dat 

[13:30:00.0000] <Wolfman2000>
...testing?  What type of testing?

[13:31:00.0000] <ezyang>
Woflman2000: the PHP port of html5lib 

[13:33:00.0000] <Wolfman2000>
Will a python library be required?

[13:34:00.0000] <ezyang>
Nope. That's the port of a port. 

[13:34:01.0000] <Wolfman2000>
...I'll get clarification on that later

[13:34:02.0000] <ezyang>
erm, *point of a port 

[13:35:00.0000] <ezyang>
(port of a point? port of a sherry? The mystery deepens) 

[13:35:01.0000] <gsnedders>
What about the side of a port, then?

[13:35:02.0000] <Wolfman2000>
I'll rephrase the question

[13:36:00.0000] <Wolfman2000>
will a Python port be required?

[13:36:01.0000] <Wolfman2000>
to be built

[13:36:02.0000] <ezyang>
We already have a Python port 

[13:37:00.0000] <Wolfman2000>
*nods*

[13:37:01.0000] <ezyang>
(I suppose, if the Python impl was first, it's not technically a port...) 

[13:37:02.0000] <beowulf>
/me has an image of snakes weighing anchor

[13:38:00.0000] <ezyang>
We can call it starboard, or something 

[13:40:00.0000] <Philip`>
We can just call it html5lib

[13:40:01.0000] <Philip`>
(All the others are pretenders)

[13:41:00.0000] <ezyang>
Ok, another test-case bug (I think) 

[13:42:00.0000] <ezyang>
nvm. 

[13:43:00.0000] <mpilgrim>
http://www.youtube.com/html5 works in chrome 3.0.182.3

[13:43:01.0000] <Wolfman2000>
...and my copy of Firefox won't work on that

[13:44:00.0000] <Wolfman2000>
3.0.10 for the record

[13:44:01.0000] <mpilgrim>
http://commons.wikimedia.org/wiki/Category:Ogg_video works in chrome 3.0.182.3 too

[13:48:00.0000] <mpilgrim>
mark@atlantis:~% mp4creator -list google_main.mp4

[13:48:01.0000] <mpilgrim>
Track   Type    Info

[13:48:02.0000] <mpilgrim>
1       audio   MPEG-4 AAC LC, 58.049 secs, 125 kbps, 44100 Hz

[13:48:03.0000] <mpilgrim>
2       video   H264 Baseline⊙2, 57.599 secs, 499 kbps, 480x270 @ 29.983159 fps

[13:48:04.0000] <mpilgrim>
(google_main.mp4 is the video on http://www.youtube.com/html5 )

[13:48:05.0000] <mpilgrim>
wikimedia commons has theora video with vorbis audio in an OGG container

[13:50:00.0000] <Wolfman2000>
mpilgrim: the youtube.com/html5 thing is NOT working on Firefox 3.5 beta 4 on Mac OS X

[13:50:01.0000] <Wolfman2000>
the video is not playing

[13:51:00.0000] <scherkus>
the youtube site uses h264/aac, which as of right now is only supported in chrome 3.0.182.3 and safari 4 beta

[13:51:01.0000] <mpilgrim>
i am aware of that

[13:51:02.0000] <Wolfman2000>
scherkus: thank you

[13:51:03.0000] <mpilgrim>
i'm testing google chrome right now

[13:52:00.0000] <mpilgrim>
http://wearehugh.com/public/2006/12/20061225-large.mp4 plays in google chrome 3.0.182.3

[13:52:01.0000] <mpilgrim>
stats on that:

[13:52:02.0000] <mpilgrim>
Track   Type    Info

[13:52:03.0000] <mpilgrim>
1       video   H264 Main⊙5, 142.842 secs, 751 kbps, 640x480 @ 29.970177 fps

[13:52:04.0000] <mpilgrim>
2       audio   MPEG-4 AAC LC, 142.762 secs, 0 kbps, 48000 Hz

[13:52:05.0000] <mpilgrim>
so that's H.264 Main Profile video, AAC-LC audio, in an MP4 container

[13:53:00.0000] <mpilgrim>
i don't have any H.264 High Profile video handy

[13:53:01.0000] <Wolfman2000>
...wonder when Firefox will support h264/aac then

[13:54:00.0000] <hsivonen>
mpilgrim: is the h.264 part open source?

[13:54:01.0000] <Rik|work>
btw, openvideo on dailymotion now works in safari 4

[13:54:02.0000] <mpilgrim>
i'm assuming it's ffmpeg

[13:54:03.0000] <mpilgrim>
as discussed a few days ago

[13:55:00.0000] <hsivonen>
Rik|work: with XiphQT

[13:55:01.0000] <hsivonen>
mpilgrim: interesting considering licensing

[13:55:02.0000] <mpilgrim>
/me wonders if google chrome for mac could bundle XiphQT

[13:55:03.0000] <hsivonen>
Rik|work: with XiphQT?

[13:55:04.0000] <Rik|work>
hsivonen: no, with mp4

[13:55:05.0000] <mpilgrim>
bbiab

[13:55:06.0000] <hsivonen>
Rik|work: also interesting

[13:56:00.0000] <Rik|work>
hsivonen: but they're not using <source>, they do browser sniffing

[13:56:01.0000] <ezyang>
Ok, this algorithm is confusing me 

[13:56:02.0000] <hsivonen>
:-(

[13:56:03.0000] <Rik|work>
I'm talking with the devs (I'm french) and they answered they have many reasons for not using that

[13:56:04.0000] <ezyang>
Otherwise, set node to the previous entry in the stack of open elements and return to step 2." and then "Initialize node to be the current node (the bottommost node of the stack). " 

[13:56:05.0000] <Rik|work>
one of them was lack of support in a previous firefox beta

[13:57:00.0000] <ezyang>
Does the second step mean that we pop nodes node is the current node, or what? 

[13:57:01.0000] <ezyang>
Or does the spec actually mean I should go to step 3? 

[13:58:00.0000] <ezyang>
Hmm, looks like I somehow fixed it. Nevermind 

[14:01:00.0000] <Groovy>
you noticed google.com stopped being <!doctype html> ?

[14:01:01.0000] <Wolfman2000>
I never noticed them using it to begin with.

[14:01:02.0000] <ezyang>
Huh. We don't have PLAINTEXT support in our tokenizer. wtf 

[14:03:00.0000] <ezyang>
Huh. The spec doesn't handle PLAINTEXT 

[14:03:01.0000] <Groovy>
Wolfman2000: yea, about from beginning of may to now they went <!doctype html>

[14:03:02.0000] <ezyang>
Uhh... Hixie? Any comments? 

[14:03:03.0000] <Wolfman2000>
...they must be secretly in bed with Microsoft

[14:04:00.0000] <Wolfman2000>
HTML5 doesn't work for IE unless you use javascript to add the elements to the DOM

[14:07:00.0000] <gsnedders>
ezyang: It's an optimization.

[14:07:01.0000] <ezyang>
Huh? 

[14:07:02.0000] <ezyang>
Sorry, I don't follow. 

[14:08:00.0000] <gsnedders>
ezyang: PLAINTEXT state means everything stays in the data state

[14:08:01.0000] <gsnedders>
ezyang: We handle it by special casing other content models

[14:08:02.0000] <ezyang>
Uhm, no we don't 

[14:08:03.0000] <gsnedders>
ezyang: Yes we do.

[14:08:04.0000] <gsnedders>
ezyang: We do nothing for it. In it, you will never move out of the data state.

[14:08:05.0000] <ezyang>
If we did, this test would pass: <table><plaintext><td> 

[14:09:00.0000] <ezyang>
the tokenizer is happly continuing parsing even after being put in PLAINTEXT content model 

[14:09:01.0000] <ezyang>
*happily 

[14:09:02.0000] <gsnedders>
We should be in the data state after it

[14:10:00.0000] <ezyang>
Right. And we never check for mode PLAINTEXT 

[14:10:01.0000] <gsnedders>
We don't need to.

[14:10:02.0000] <ezyang>
REally? 

[14:10:03.0000] <gsnedders>
Hence we always stay in the data state.

[14:10:04.0000] <gsnedders>
ezyang: Look at the spec.

[14:10:05.0000] <gsnedders>
ezyang: We only move out of it after checking state is PCDATA, RCDATA or CDATA

[14:11:00.0000] <gsnedders>
The tokenizer looks fine to mee.

[14:11:01.0000] <gsnedders>
*me

[14:12:00.0000] <ezyang>
Oh, well, that's because PLAINTEXT content model is not getting set. 

[14:12:01.0000] <gsnedders>
:D

[14:12:02.0000] <gsnedders>
See, the code I've been working on works. :P

[14:13:00.0000] <ezyang>
Yep. 

[14:13:01.0000] <ezyang>
You're right 

[14:13:02.0000] <gsnedders>
Of course.

[14:15:00.0000] <ezyang>
This return the content model business is obnoxious 

[14:15:01.0000] <ezyang>
I think I'm going to refactor it out. 

[14:42:00.0000] <ezyang>
Ok, so if I have <p><b></p> <p> does the space get wrapped by <b>? 

[14:43:00.0000] <ezyang>
I'm leaning towards yes, althought the test case claims it doesn't. 

[14:43:01.0000] <ezyang>
Also, the Python implementation agrees with me 

[14:43:02.0000] <ezyang>
So does the Ruby implementation 

[14:44:00.0000] <ezyang>
(could it be? could ezyang have actually found a real test-case bug? 

[14:46:00.0000] <ezyang>
It could be a spec bug, since what then happens is we stick the <p> inside the <b><i><u> tags, which is not what active formatting elements is supposed to do. 

[14:47:00.0000] <ezyang>
Any comments? 

[14:48:00.0000] <jgraham>
ezyang: WDVND?

[14:48:01.0000] <ezyang>
come again? 

[14:49:00.0000] <jgraham>
(What does Validator.nu do?)

[14:49:01.0000] <ezyang>
Oh, yeah 

[14:50:00.0000] <ezyang>
V.nu does the test-case behavior 

[14:51:00.0000] <ezyang>
I guess I could figure out what they did different. 

[14:53:00.0000] <jgraham>
ezyang: Henri is generally more up to date than html5lib. Although there is a case with whitespace that he preemptively changed anticipating a spec change that hasn't happened

[14:54:00.0000] <ezyang>
Yep, that seems like it. 

[14:54:01.0000] <ezyang>
hsivonen is not reconstructing active formatting elements when there's whitespace 

[14:55:00.0000] <ezyang>
The spec hasn't changed w.r.t. yet, though 

[14:56:00.0000] <ezyang>
I thought tokenizing/treebuilding was pretty stable now? 

[15:00:00.0000] <ezyang>
"Update the tree building tests with frameset-ok, new AAA and (not yet in spec) WebKit-style foster-parenting" 

[15:00:01.0000] <hsivonen>
if something depends on whitespace in 'in body', it's likely a bug

[15:00:02.0000] <ezyang>
So it's the webkit thing again :-) 

[15:00:03.0000] <hsivonen>
oh, you meant foster parenting

[15:00:04.0000] <hsivonen>
that's deliberate

[15:01:00.0000] <ezyang>
So, I understand that the spec may not be at the point that V.nu is 

[15:01:01.0000] <ezyang>
But having tests that the spec deliberately fails really doesn't help me out, if I'm trying for 0 fails 

[15:01:02.0000] <hsivonen>
ezyang: when I checked that in, Hixie's IRC statements hinted at spec going that way

[15:02:00.0000] <ezyang>
Anyway, it's whitespace "in body" 

[15:03:00.0000] <ezyang>
Specifically, the difference is V.nu checks if a character token is whitespace, and if it is, doesn't reconstruct active formatting elements 

[15:03:01.0000] <hsivonen>
ezyang: not while foster parenting?

[15:03:02.0000] <ezyang>
Nope. 

[15:03:03.0000] <ezyang>
No foster parenting at all. 

[15:03:04.0000] <ezyang>
You changed the test case, however, in commit 535a1040e2df 

[15:03:05.0000] <hsivonen>
weird

[15:04:00.0000] <ezyang>
for which I just posted the summary 

[15:04:01.0000] <hsivonen>
wait, have the test cases moved to hg?

[15:04:02.0000] <hsivonen>
I've been checking in to svn still

[15:04:03.0000] <ezyang>
Yes. 

[15:05:00.0000] <hsivonen>
ouch

[15:05:01.0000] <ezyang>
Does Google Code magically transfer it? 

[15:06:00.0000] <ezyang>
I mean, there's no way to tell if there was an svn repository from Google Code website anymore 

[15:06:01.0000] <jgraham>
ezyang: Dunno. I had to do the initial migration manually though

[15:06:02.0000] <ezyang>
Oh hey, that's the last commit you made, according to mercurial 

[15:07:00.0000] <ezyang>
(re hsivonen) 

[15:07:01.0000] <ezyang>
jgraham: aha 

[15:07:02.0000] <ezyang>
When did the migration occur? 

[15:07:03.0000] <hsivonen>
ezyang: you'll save a lot of EOF grief if you get my changes from this week from svn

[15:08:00.0000] <ezyang>
But I don't even remember where the svn repository *is* 

[15:08:01.0000] <hsivonen>
I don't either. I'm not on the computer that has my html5lib sandbox

[15:09:00.0000] <ezyang>
Anyway, not turning off access to the subversion repos seems... poor. 

[15:09:01.0000] <ezyang>
I'm not sure how we can easily merge your changes back in. 

[15:09:02.0000] <jgraham>
ezyang: https://html5lib.googlecode.com/svn/trunk/

[15:09:03.0000] <ezyang>
Awesome 

[15:09:04.0000] <jgraham>
It seems like svn should become read-only

[15:11:00.0000] <jgraham>
(I mean ideally, not that it does)

[15:13:00.0000] <ezyang>
AWesome, we only lost one commit 

[15:13:01.0000] <ezyang>
(maybe two, if one of those was in a branch) 

[15:14:00.0000] <ezyang>
hsivonen: you can probably take its diff and apply to the new hg repository 

[15:14:01.0000] <hsivonen>
I haven't committed to branches

[15:14:02.0000] <hsivonen>
ezyang: I'll do that when I get to my work computer

[15:15:00.0000] <ezyang>
Odd. 1303 has nothing in it 

[15:15:01.0000] <ezyang>
Great. 

[15:16:00.0000] <ezyang>
Holy crap that's a big change 

[15:17:00.0000] <ezyang>
Oh, these are all tokenizer updates 

[15:17:01.0000] <ezyang>
Hmm... we already pass all of the tests... uh oh 

[15:17:02.0000] <ezyang>
Anyway, hsivonen, you maintain that Hixie is going to change the spec to follow V.nu behavior? 

[15:20:00.0000] <hsivonen>
ezyang: I intend to keep pushing Hixie that way :-)

[15:20:01.0000] <Wolfman2000>
...huh.  The <header> tag has changed functionality?

[15:20:02.0000] <Wolfman2000>
/me catches up on the blog reading

[15:20:03.0000] <Wolfman2000>
...am I really only allowed to use <header> with <h#>?

[15:20:04.0000] <hsivonen>
ezyang: the tokenizer changed to discard tag token if EOF happens inside tag

[15:20:05.0000] <ezyang>
Ok. Do you mind if I separate out that test to its own file? 

[15:20:06.0000] <ezyang>
Aha. 

[15:21:00.0000] <hsivonen>
ezyang: that would be good

[15:25:00.0000] <jgraham>
Wolfman2000: You can only use <hgroup> with <hx> <header> you can use with anything

[15:25:01.0000] <ezyang>
Done. 

[15:25:02.0000] <ezyang>
I should document this somewhere 

[15:26:00.0000] <Wolfman2000>
ok

[15:31:00.0000] <ezyang>
hsivonen: I assume test 44 is another one of those? 

[15:31:01.0000] <ezyang>
(in tests2.dat) 

[15:32:00.0000] <ezyang>
This one is <p><b><i><u></p>\n<p>X 

[15:33:00.0000] <hsivonen>
ezyang: if that one deviates from spec, it's a bug

[15:35:00.0000] <ezyang>
It's the same thing (note that in my post \n was meant to be interpreted as an scape) 

[15:35:01.0000] <jgraham>
/me wonders how to stop hg merge from merging the python3 directory into the python directory

[15:35:02.0000] <ezyang>
V.nu doesn't reconstruct active formatting elements on \n; everyone else does 

[15:35:03.0000] <hsivonen>
ezyang: sounds like a bug in V.nu

[15:35:04.0000] <ezyang>
You could ask #mercurial about it 

[15:35:05.0000] <ezyang>
But... \n is whitespace 

[15:36:00.0000] <hsivonen>
ezyang: I need to study what code I have written

[15:36:01.0000] <ezyang>
ok 

[15:37:00.0000] <ezyang>
/me thinks "Wow, it's not often I get three reference implementations :-)" 

[15:39:00.0000] <ezyang>
THIRTY-SIX! 

[15:39:01.0000] <ezyang>
(failures, that is) 

[15:41:00.0000] <ezyang>
Ok, now I need to implement the fragment algorithm 

[15:50:00.0000] <ezyang>
I think that's enough html5lib hacking for today 

[15:50:01.0000] <ezyang>
Test suites 1-3 are now passing 


2009-05-30
[18:37:00.0000] <jwalden>
/me finds an error in the web socket protocol draft

[19:08:00.0000] <jwalden>
or have I

[19:37:00.0000] <jwalden>
/me curses the mess that is RFC 2817 and its offhand definition of the CONNECT method

[19:46:00.0000] <jwalden>
no, seems to be an error

[20:22:00.0000] <jwalden>
email sent to that list, seems a bit of delay in its being visible in the web interface

[20:30:00.0000] <jwalden>
http://www.ietf.org/mail-archive/web/hybi/current/msg00312.html

[20:32:00.0000] <jwalden>
...and fixed already :-)

[21:44:00.0000] <Hixie>
hey, we crossed the 1000 subscriber barrier!

[21:44:01.0000] <Hixie>
sweet

[21:50:00.0000] <takkaria>
Hixie: what big of the spec are you working on now?

[21:50:01.0000] <Hixie>
none at this very instant

[21:50:02.0000] <Hixie>
probably going to go through htmlwg bugs and try to get them down

[21:50:03.0000] <takkaria>
cool

[21:51:00.0000] <Hixie>
(i'll work on whatever people want me to work on, so if you have something important, let me know)

[21:51:01.0000] <ezyang>
Nice! 

[21:52:00.0000] <Hixie>
this week has been hell in terms of getting anything done

[21:52:01.0000] <Hixie>
what with everyone and their dog talking about html5 because of Google I/O

[21:55:00.0000] <inimino>
I will admit to being mildly disappointed in the (ahem) wave of recent interest in HTML5 but I'm not quite sure why

[21:56:00.0000] <Hixie>
disappointed ho?

[21:56:01.0000] <Hixie>
how, rather

[21:56:02.0000] <othermaciej>
disappointed?

[21:56:03.0000] <othermaciej>
why?

[21:56:04.0000] <Hixie>
he said he wasn't sure why :-)

[21:56:05.0000] <Hixie>
that's why i asked how instead :-)

[21:56:06.0000] <othermaciej>
(I have been mildly irked by the garbling of the technical message.)

[21:56:07.0000] <othermaciej>
(But I like that the overall response is positive.)

[21:57:00.0000] <inimino>
yes most of it is probably just seeing "HTML5" become the next "Ajax"

[21:57:01.0000] <ezyang>
HTML5 is huge. (I mean that in a scope way) 

[21:57:02.0000] <othermaciej>
HTML5 is huge, and people use it to refer to things that are not, technically, part of HTML5

[21:57:03.0000] <othermaciej>
one thing that surprised me in the coverage was how in sync Apple and Google especially are with respect to what is most important in Web technology

[21:57:04.0000] <ezyang>
And it's building up the "eye-candy" factor that will sell it to the public 

[21:57:05.0000] <othermaciej>
if I didn't know better, I would think it was a conspiracy

[21:58:00.0000] <Hixie>
othermaciej: if you think that message got garbled, just check out the lower-third text on http://cache.gawker.com/assets/images/gizmodo/2009/05/240ccede5360b093dbf298f8946025a5.png

[21:58:01.0000] <othermaciej>
Hixie: lol

[21:58:02.0000] <othermaciej>
(I actually laughed out loud)

[21:58:03.0000] <othermaciej>
so maybe it's just me

[21:58:04.0000] <inimino>
hehe

[21:59:00.0000] <jruderman>
i'm thinking of making a game

[21:59:01.0000] <othermaciej>
but I feel like back in the 80s and early 90s, technical reporters seemed to have more of a clue

[21:59:02.0000] <jruderman>
and if i do, i'll probably use javascript and <canvas>

[21:59:03.0000] <othermaciej>
I remember reading PC MAgazine back in the day

[21:59:04.0000] <othermaciej>
and it was full of technical facts and stuff

[21:59:05.0000] <jwalden>
othermaciej: did you ever see the coverage of the Morris worm?  they didn't have a clue :-D

[21:59:06.0000] <othermaciej>
nowadays I look at CNet and I cry

[21:59:07.0000] <inimino>
well I think screen grabs of ridiculous Fox News tickers are practically a genre of their own by now...

[22:00:00.0000] <inimino>
I just don't look at CNet

[22:00:01.0000] <othermaciej>
ArsTechnica is one of the few beacons of relative technical clue, as far as tech news sites, and even they publish total claptrap at times

[22:00:02.0000] <inimino>
I think the technical audience has moved on and they are all reading each other's blogs instead of magazines

[22:01:00.0000] <jwalden>
demise of ddj and all that

[22:01:01.0000] <othermaciej>
with blogs it's hard to tell up front who is full of shit

[22:01:02.0000] <othermaciej>
unless you are an expert in the particular subject are yourself, or know the person by reputation

[22:01:03.0000] <jruderman>
i'll help you out. my blog -- http://www.squarefree.com/ -- full of shit.

[22:01:04.0000] <othermaciej>
like Gizmodo is super popular and clearly appeals to tech enthusiasts but it is at best 50% true

[22:01:05.0000] <inimino>
true

[22:02:00.0000] <othermaciej>
jruderman: I've rarely known you to be full of shit

[22:02:01.0000] <othermaciej>
and...

[22:02:02.0000] <othermaciej>
depending on the nature of the game, JS + <canvas> is probably a decent choice

[22:02:03.0000] <othermaciej>
have you seen olliej's fluid dynamics demo?

[22:02:04.0000] <jruderman>
yep

[22:03:00.0000] <jruderman>
then i left it open in a background tab and wondered why my computer was hot

[22:03:01.0000] <othermaciej>
I don't know why but I find it endlessly fun to play with

[22:03:02.0000] <othermaciej>
it does manage to replicate that feature of Flash games

[22:03:03.0000] <jruderman>
hehe

[22:04:00.0000] <othermaciej>
whoah he managed to boost the frame rate

[22:04:01.0000] <inimino>
where is it?

[22:04:02.0000] <ezyang>
/me imagines the new cookie cutter industry: "Free, Addictive HTML5 Games!" 

[22:04:03.0000] <othermaciej>
I'm getting nearly 60 fps here

[22:04:04.0000] <othermaciej>
http://nerget.com/pressure/pressure.html

[22:04:05.0000] <Hixie>
i just get a black screen :-(

[22:04:06.0000] <othermaciej>
it numerically approximates the solution to the Navier-Stokes equations of fluid dynamics

[22:04:07.0000] <Hixie>
oh i see

[22:04:08.0000] <Hixie>
you have to click

[22:04:09.0000] <othermaciej>
Hixie: you have to click or drag to make things happen

[22:05:00.0000] <othermaciej>
and if you toggle the view, it flips from pressure to velocity

[22:05:01.0000] <othermaciej>
I wanna put this up on webkit.org but I have to get rid of the layout table first

[22:05:02.0000] <othermaciej>
or people will laff at us

[22:05:03.0000] <jruderman>
lol

[22:05:04.0000] <inimino>
ooh, fun

[22:06:00.0000] <othermaciej>
Dear CSS Layout, Please learn how to be easy like tables. Love, The Internet.

[22:06:01.0000] <inimino>
now it just needs the audio equivalent of <canvas> and you can add splooshy sounds

[22:06:02.0000] <othermaciej>
realtime audio generation is very hard

[22:06:03.0000] <othermaciej>
actually

[22:06:04.0000] <othermaciej>
the latency requirements to get smooth realtime sound generation are ridiculous

[22:07:00.0000] <othermaciej>
so it's a much harder problem than canvas :-/

[22:07:01.0000] <ezyang>
/me is mesmerized 

[22:07:02.0000] <jruderman>
othermaciej: latency requirements?

[22:07:03.0000] <jruderman>
what makes it different from graphics?

[22:08:00.0000] <othermaciej>
jruderman: for audio to feel reactive, you need a fairly small buffer - audio delay is much more noticeable than video

[22:08:01.0000] <othermaciej>
jruderman: but on the flip side, if you don't fill the buffer in time, you get horrible sounding artifacts

[22:09:00.0000] <othermaciej>
this is why things like CoreAudio and OpenAL are so complicated

[22:10:00.0000] <jruderman>
games do fine choosing prerecorded sound effects and timing them to go with the video

[22:10:01.0000] <doublec>
that's what js8080 does for space invaders, etc, using <audio>

[22:11:00.0000] <jruderman>
what changes when you're creating the sound effects on the fly?

[22:11:01.0000] <doublec>
I would like a openal type api for audio however (in the browser)

[22:11:02.0000] <doublec>
you can generate wav data and use data url's to <audio>

[22:12:00.0000] <othermaciej>
with prerecorded sounds, you don't have the problem of the audio output buffer underflowing cause you couldn't fill it fast enough

[22:13:00.0000] <othermaciej>
having an ImageData like approach where you provided a large-ish buffer that gets mixed in might work

[22:13:01.0000] <othermaciej>
but doesn't have much of an obvious advantage over using <audio>

[22:15:00.0000] <inimino>
/me switches back to the fluid dynamics window and discovers The Blob

[22:19:00.0000] <othermaciej>
heh

[22:20:00.0000] <othermaciej>
I wish olliej let you control the viscosity

[22:43:00.0000] <jwalden>
um... http://twitter.com/WHATWG/status/1953544919

[22:43:01.0000] <jwalden>
way to go twitter

[22:44:00.0000] <ezyang>
Wow. Twitter doesn't escape ampersands 

[22:44:01.0000] <jwalden>
gotta be an XSS waiting to happen in that somehow, I'd expect

[22:45:00.0000] <ezyang>
Mm, not really 

[22:46:00.0000] <jwalden>
/me notes that account has 2**10 followers

[22:50:00.0000] <mgrdcm>
/me +1

[23:32:00.0000] <jwalden>
/me wonders what specification would have the most claim to being responsible for describing what must happen when DOM APIs are provided with non-UTF-16 strings, like "foo\ud863\u000abar" -- webidl?

[23:32:01.0000] <jwalden>
/me is looking again at acid3 test 68, for a little context

[23:33:00.0000] <heycam`>
webidl doesn't say anything about non-UTF-16 strings in DOMStrings, at the moment

[23:33:01.0000] <heycam`>
is acid test 68 the svg string methods?

[23:34:00.0000] <heycam`>
ah no, ecmascript i see

[23:35:00.0000] <jwalden>
no, it's |document.createElement("input").value = highSurrogate + asciiData| and whether that throws an exception, sets the identical value, or sets a value with the high surrogate replaced with U+FFFD

[23:35:01.0000] <jwalden>
or rather utf16Data, I guess

[23:35:02.0000] <jwalden>
/me thinks he enumerated all the permitted behaviors

[23:36:00.0000] <heycam`>
yeah i'm not sure what layer that should be handled on

[23:36:01.0000] <jwalden>
ah, missed one, could set to utf16Data as well

[23:36:02.0000] <jwalden>
DOM 3 Core seems to not even entertain the prospect

[23:36:03.0000] <heycam`>
heh

[23:37:00.0000] <jwalden>
webidl says "(to be interpreted as UTF-16 encoded strings [RFC2781])"

[23:37:01.0000] <heycam`>
taboo, please don't mention it in mixed company

[23:37:02.0000] <jwalden>
rfc2781 of course says "Error recovery is not specified by this document."

[23:38:00.0000] <heycam`>
if there's consistent behaviour across attributes / operation arguments, then specific behaviour could be mandated in webidl

[23:38:01.0000] <heycam`>
i haven't tested that though

[23:43:00.0000] <jwalden>
gecko sets a value with the high surrogate replaced with U+FFFD; webkit preserves the code units exactly; opera does the same; ie does the same

[23:44:00.0000] <jwalden>
this is rather unfortunate, as gecko's behavior is rather deeply embedded, as best as I recall

[23:44:01.0000] <heycam`>
the latter seems cleaner to me

[23:45:00.0000] <jwalden>
testing with javascript: var e = document.createElement("input"); e.value = "\ud863test"; alert(e.value.split("").map(function(v) { return v.charCodeAt(0); })) in non-IE and javascript: var e = document.createElement("input"); e.value = "\ud863test"; alert(e.value.split("")) in IE

[23:47:00.0000] <jwalden>
although looking slightly, this may end up being specific to input.value, not sure yet

[23:48:00.0000] <heycam`>
k. i'll put a note in webidl and see if anyone has an opinion on it being dealt with in that spec.

[23:50:00.0000] <jwalden>
cool

[23:53:00.0000] <jwalden>
hm, looks like it turns into a ghastly mess partly because attribute values are optimized into particular encodings, if I'm reading right

[23:53:01.0000] <jwalden>
and even beyond that things like numbers are stored as such

[00:01:00.0000] <jwalden>
hm, or maybe no, but now I'm diving into template metaprogramming of string manipulation

[00:03:00.0000] <jwalden>
stupid traits antipattern

[00:05:00.0000] <jwalden>
heycam`: another fun one: javascript: var e = document.createElement("input"); e.value = "\ud863\0\0test"; alert(e.value.split(""))

[00:06:00.0000] <jwalden>
seems all the browsers I can test truncate that

[00:06:01.0000] <jwalden>
gah, what a mess

[00:08:00.0000] <heycam`>
ouch

[00:10:00.0000] <jwalden>
based on what I know of gecko, I bet we truncate to U+FFFD while the others truncate to a lone high surrogate

[00:11:00.0000] <heycam`>
differing behaviour between javascript string classes and dom string classes doesn't seem good

[00:16:00.0000] <jwalden>
none of this is any good

[00:18:00.0000] <jwalden>
particularly since encoding mistakes have a knack for becoming security problems if you're unlucky

[02:46:00.0000] <Hixie>
i'm starting to need to keep track of e-mails for which i'm awaiting replies

[02:47:00.0000] <hsivonen>
Hixie: it seems to me that the comments at http://www.w3.org/QA/2009/05/_watching_the_google_io.html#comments don't deal with the main "it's not done yet" point of the post

[02:49:00.0000] <Hixie>
as i just mentioned on the owf list, html5 is more done than html4 ever was

[02:49:01.0000] <Hixie>
so that seems like a meaningless point :-)

[02:50:00.0000] <hsivonen>
Hixie: maybe that point would have been more useful to make in the Q&A/QA blog

[02:51:00.0000] <Hixie>
be my guest

[02:56:00.0000] <hsivonen>
I wonder when we'll see a log post warning about premature deployment of RDFa in text/html

[02:56:01.0000] <hsivonen>
*blog

[02:59:00.0000] <theMadness>
I think Crockford accidentally made the point for html5 a while ago.

[02:59:01.0000] <hsivonen>
theMadness: which point and where?

[02:59:02.0000] <theMadness>
http://blog.360.yahoo.com/blog-TBPekxc1dLNy5DOloPfzVvFIVOWMB0li?p=978 here

[03:00:00.0000] <theMadness>
He advocated for a new html that is not very new, just a fix of the 4.01.

[03:01:00.0000] <theMadness>
But if you take away webforms and the new elements (which are all more specific and semantic heavy versions of div), html5 is just what he calls html 4.2

[03:03:00.0000] <theMadness>
But maybe he is considering the DOM changes too.

[03:05:00.0000] <tantek>
theMadness - this is not a new idea, I advocated for a new html that is not very new, just a fix of 4.01, back in 2003: http://tantek.com/log/2003/01.html#L20030114

[03:05:01.0000] <hsivonen>
making "just a fix" *now* would be rather disappointing

[03:05:02.0000] <theMadness>
tantek, I see, but I recalled that post as when I read it in my mind it sounded "html5 is too much, let's have html5 instead".

[03:06:00.0000] <tantek>
hsivonen, I don't disagree. those were my opinions as of 6+ years ago.

[03:07:00.0000] <theMadness>
Which probably was the right time for them.

[03:08:00.0000] <theMadness>
But right now we're too high on RIA, stepping back will shoot us back to the plugin/applet/proprietary extensions era. :P

[03:08:01.0000] <tantek>
however, note that many folks here have been writing up well thought out criticisms of XHTML2 / focus-of-W3C for 6+ years. how little has changed in that regard is IMHO what is rather disappointing.

[03:08:02.0000] <Philip`>
Let me be the first to advocate for a new HTML that is not very new, just a fix of 5, so I can point back to this IRC log entry several years from now

[03:09:00.0000] <theMadness>
Philip`, you mean 4?

[03:10:00.0000] <Philip`>
theMadness: No - I'd be at least six years too late to be the first to advocate that :-)

[03:10:01.0000] <tantek>
Philip`, I think you have to wait at least until HTML5 is a REC (projected 2022?) and the first post-REC issue is discovered/acknowledged before advocating a new HTML that is just a fix of 5.

[03:11:00.0000] <theMadness>
What he said. I didn't have the courage to say it myself.

[03:11:01.0000] <Hixie>
/me mumbles that he doesn't think specs should be numbered in this way and that we should just have rolling versions the way we're doing HTML5 today and the way most websites work

[03:12:00.0000] <hsivonen>
/me I always feel uncomfortable when I see "RIA" in the context of HTML5.

[03:12:01.0000] <Hixie>
why?

[03:12:02.0000] <tantek>
Anyone seen mpilgrim recently?

[03:12:03.0000] <tantek>
This link appears to have broken: http://diveintomark.org/archives/2003/01/13.html#semantic_obsolescence

[03:12:04.0000] <tantek>
It was a criticism of XHTML2 IIRC.

[03:13:00.0000] <hsivonen>
Hixie: "RIA" seems to be a Flash/Silverlight/JavaFX not-the-bowser-platform term

[03:13:01.0000] <theMadness>
hsivonen, exactly the reason why we must "take it back". :>

[03:14:00.0000] <Philip`>
tantek: Why would post- vs pre-REC make a difference to the need to create a new version to fix issues?

[03:14:01.0000] <tantek>
because pre-REC, issues can be fixed without creating a new version.

[03:14:02.0000] <othermaciej>
I'm happy to say "Web application"

[03:14:03.0000] <othermaciej>
but WA is not a TLA

[03:14:04.0000] <othermaciej>
Standard-based Web Application?

[03:15:00.0000] <Hixie>
tantek: they can post-rec too

[03:15:01.0000] <tantek>
Hixie, indeed, that whole "2nd edition" trick

[03:15:02.0000] <Philip`>
tantek: If the spec is ever going to reach REC, at some point it will have to stop responding to issues (e.g. stop adding features) so it can stabilise and get test suites and interoperable implementations, long before it can become REC, and during that time period it won't be able to respond adequately to all issues

[03:15:03.0000] <Hixie>
anyway i'm sure html6 will have started long before html5 is in rec

[03:15:04.0000] <annevk2>
Hixie, btw, apparently you have some action item assigned to you

[03:16:00.0000] <Hixie>
annevk2: no, i don't

[03:16:01.0000] <theMadness>
By the way, I've seen complaints about versioning and the doctype, how will a browser be able to tell a html6 page?

[03:16:02.0000] <Philip`>
theMadness: It won't be able to

[03:16:03.0000] <Hixie>
theMadness: why would it need to?

[03:16:04.0000] <Philip`>
theMadness: (and it won't need to)

[03:17:00.0000] <theMadness>
To trigger legacy/edge support?

[03:17:01.0000] <tantek>
Philip`: sure, amend previous statement to pre-CR then :)

[03:17:02.0000] <annevk2>
tantek, http://diveintomark.org/archives/2003/01/13/semantic_obsolescence

[03:17:03.0000] <Hixie>
theMadness: we've never had to do that before, why would we do it now? :)

[03:18:00.0000] <annevk2>
Hixie, I guess you never accepted it, but all indications are that you have

[03:18:01.0000] <theMadness>
Wait, we did, what about quirks/standards?

[03:18:02.0000] <Hixie>
annevk2: check the tracker

[03:18:03.0000] <annevk2>
ah

[03:18:04.0000] <Hixie>
theMadness: quirks vs standards isn't done along version lines

[03:18:05.0000] <tantek>
annevk2, note that the blog post at that URL links to http://diveintomark.org/archives/2003/01/14.html#eddies_in_the_spacetime_continuum which is also 404

[03:19:00.0000] <Hixie>
theMadness: or to put it another way, why would we want to add more quirks modes?

[03:19:01.0000] <othermaciej>
I think the experience of quirks/standards has made many of us decide we never want to do that sort of thing again

[03:19:02.0000] <theMadness>
So basically html6 will have to be compatible with 5.

[03:19:03.0000] <Hixie>
(except microsoft, who decided that it was teh awesome)

[03:19:04.0000] <Philip`>
tantek: Okay, fair enough :-)

[03:19:05.0000] <Hixie>
(and added 3 or 4 new modes just in IE8)

[03:19:06.0000] <othermaciej>
it will have to be compatible with the corpus of deployed HTML content

[03:19:07.0000] <othermaciej>
which by that time will hopefully include lots of HTML5

[03:20:00.0000] <annevk2>
tantek, do some URL hacking, http://diveintomark.org/archives/2003/01/14/eddies_in_the_spacetime_continuum :)

[03:20:01.0000] <othermaciej>
HTML5 is also trying to be compatible with deployed Web content as opposed to with prior specs

[03:20:02.0000] <othermaciej>
hopefully it will get less things wrong than HTML 4.01 did

[03:20:03.0000] <annevk2>
tantek, they used to redirect though, Mark dropped the ball with regards to cool URLs

[03:20:04.0000] <Philip`>
tantek: (...and amend the statement from 2022 to 2012)

[03:20:05.0000] <tantek>
did he drop his daily archives?

[03:21:00.0000] <annevk2>
tantek, no, they're just under a different URL

[03:21:01.0000] <Philip`>
tantek: But also, the people maintaining HTML5 a decade from now will not be the same people who are working on it today, and the new people who are innovating on the web will get fed up with the stuffy old HTML5 guys and will want to change it to go in a different direction

[03:21:02.0000] <annevk2>
tantek, i.e. without the .html

[03:21:03.0000] <tantek>
Philip`, indeed, need to get ahead of the Mayans (or was it Aztecs?) ;)

[03:22:00.0000] <othermaciej>
Philip`: does that mean we can all aspire to be the Roy Fielding of the future?

[03:22:01.0000] <Philip`>
othermaciej: It's not an aspiration, it's an inevitability

[03:23:00.0000] <tantek>
Philip`: I kind of doubt it - unless another Great Web Schism occurs.

[03:23:01.0000] <tantek>
speaking of which, is anyone organizing a celebration of the 5th anniversary of the Great Web Standards Schism?

[03:23:02.0000] <tantek>
http://www.w3.org/2004/04/webapps-cdf-ws/minutes-20040601.html

[03:23:03.0000] <othermaciej>
the people I work with who have age, experience and accomplishments similar to Roy do not have his kind of attitude

[03:24:00.0000] <theMadness>
Standards: the 21st century version of protestantism.

[03:24:01.0000] <Philip`>
othermaciej: Oh, okay then

[03:24:02.0000] <othermaciej>
(for instance Darin Adler has never told me to STFU because he used to create software by toggling in the 1s and 0s with a magnet)

[03:24:03.0000] <tantek>
that meeting is what really caused the momentum to get behind HTML5, and for that matter microformats.

[03:24:04.0000] <Philip`>
othermaciej: Hopefully we can all aspire to be more like him :-)

[03:24:05.0000] <othermaciej>
yes, the Workshop of the Bastille

[03:25:00.0000] <othermaciej>
Philip`: he only tells me to STFU when I actually need to and am not aware enough to realize it

[03:25:01.0000] <othermaciej>
(which is more rarely these days)

[03:25:02.0000] <tantek>
it was quite an epic workshop, othermaciej. at the time I didn't quite realize how historic an event it was going to be.

[03:25:03.0000] <othermaciej>
so some people mellow with age, like wine, instead of becoming sharper, like cheddar

[03:26:00.0000] <othermaciej>
I know a surprising number of people on the attendees list for non-web-standards-related reasons

[03:26:01.0000] <othermaciej>
(and of course many for web standards related reasons too)

[03:27:00.0000] <tantek>
re-reading the minutes, note that namespaces in data on the web are just as dead as they were back then, 5 years ago.

[03:28:00.0000] <othermaciej>
Creative Commons is doing its best to spread them via copypasta

[03:28:01.0000] <othermaciej>
does that count?

[03:29:00.0000] <tantek>
not when typical implementation (not going to name names) simply uses regexes to scrap NS prefixes, no, that doesn't count. ;)

[03:29:01.0000] <tantek>
and yes, we've critiqued ccREL already in this channel

[03:30:00.0000] <tantek>
its very design promotes combinatorial license proliferation which is actually a stated non-goal of both Creative Commons (numerous presentations etc.) and many experienced folks in the open source communities.

[03:31:00.0000] <othermaciej>
was it obvious at the time they were invented how hard namespaces would be to use and understand?

[03:31:01.0000] <Hixie>
othermaciej: i'm sure gramps has told you to STFU :-P

[03:31:02.0000] <tantek>
othermaciej, yes, namespaces in content have been ugly since day 1

[03:31:03.0000] <othermaciej>
Hixie: he has, but (a) only when I actually needed to and (b) he did not justify it by giving a lengthy list of his dinosaur accomplishments

[03:31:04.0000] <tantek>
I think I heard the term somewhere else, but namespaces in content is an excellent example of "syntactic vinnegar"

[03:31:05.0000] <Hixie>
othermaciej: :-P

[03:32:00.0000] <othermaciej>
gramps is not even 5% as crusty as Roy

[03:32:01.0000] <Hixie>
othermaciej: i did not mean to imply gramps was like roy :-P

[03:32:02.0000] <othermaciej>
I've heard it as "syntactic salt", but yes

[03:33:00.0000] <tantek>
syntactic vinegar as in the opposite of syntactic sugar

[03:33:01.0000] <theMadness>
I'm officially hungry.

[03:33:02.0000] <othermaciej>
yeah, I get the reference

[03:33:03.0000] <tantek>
it makes the syntax uglier / less friendly without actually providing a benefit (and in many cases causing harm - e.g. silos)

[03:34:00.0000] <Philip`>
Syntactic salt as in raising one's blood pressure?

[03:34:01.0000] <theMadness>
Also, being from Modena, I'm not taking the disrespect for vinegar. :P

[03:34:02.0000] <tantek>
it's not balsamic

[03:34:03.0000] <theMadness>
Ah, someone did his homework. :)

[03:35:00.0000] <tantek>
I have some Mediterranean heritage ;)

[03:35:01.0000] <othermaciej>
"syntactic salt" is just a different joke way of saying "the opposite of syntactic sugar"

[03:36:00.0000] <tantek>
othermaciej - it's not the same thing

[03:36:01.0000] <Hixie>
http://www.flickr.com/photos/glutnix/3325234350/ lol

[03:36:02.0000] <tantek>
see http://en.wikipedia.org/wiki/Syntactic_sugar

[03:36:03.0000] <othermaciej>
that's not the way I have heard it used

[03:36:04.0000] <theMadness>
Complicated versus descriptive.

[03:36:05.0000] <othermaciej>
(the wikipedia note)

[03:37:00.0000] <theMadness>
Also head is not what he means by head.

[03:37:01.0000] <othermaciej>
I'll agree with the "a hoop the programmer must jump through" part, but not "designed to make it harder to write bad code"

[03:37:02.0000] <othermaciej>
but anyway

[03:37:03.0000] <Hixie>
right well lest i start actually editing the spec, i think i should go to bed now

[03:37:04.0000] <Hixie>
nn

[03:38:00.0000] <othermaciej>
I think we are on the same page with respect to </3 namespaces

[03:38:01.0000] <othermaciej>
g'night

[03:38:02.0000] <tantek>
right. syntactic vinegar is both a hoop the programmer/author must jump through, and is designed (perhaps unintentionally) to make it harder to write easy code.

[03:38:03.0000] <tantek>
now, who's willing to add that description of syntactic vinegar to http://en.wikipedia.org/wiki/Syntactic_sugar ?

[03:41:00.0000] <tantek>
ok, well if no one else is planning a 5 year remembrance/celebration of the Great Web Standards Schism - we can certainly do it at the microformats meetup dinner in SF on Tuesday 2009-153 at 21st Amendment: http://microformats.org/wiki/events/2009-06-02-weekly-dinner-sf

[03:41:01.0000] <Hixie>
/me is too busy looking forwards to look back :-P

[03:42:00.0000] <Philip`>
/me is too busy looking at his navel to look either backwards or forwards

[05:01:00.0000] <othermaciej>
the Schism is ancient history in internet time

[05:05:00.0000] <annevk2>
pre-historic, even :)

[05:07:00.0000] <annevk2>
I like the name "The Workshop" better though (coined by hsivonen iirc)

[05:13:00.0000] <heycam>
/me finds it interesting that he presented at The Workshop even though he didn't know of its significance at the time

[09:21:00.0000] <annevk42>
http://twitpic.com/69dic #contentsniffingfail

[09:23:00.0000] <myakura>
:(

[09:27:00.0000] <myakura>
/me regrets that he should've test that with http://tools.ietf.org/id/draft-abarth-mime-sniff-00.txt

[09:28:00.0000] <annevk42>
heh

[09:29:00.0000] <annevk42>
meanwhile I'm getting pretty close to the point where I can check in XMLHttpRequest again

[09:30:00.0000] <annevk42>
and I guess then I've to port everything to XMLHttpRequest Level 2 -- sigh

[09:47:00.0000] <gsnedders>
The sun is a great anti-productivity tool.

[09:50:00.0000] <annevk42>
weekends are too

[09:51:00.0000] <gsnedders>
/me has sand in his mouth from the beath

[09:51:01.0000] <gsnedders>
*beach

[12:44:00.0000] <gsnedders>
Is there any way to force Opera to reload from server, ignoring cache?

[12:46:00.0000] <Philip`>
gsnedders: Click the reload button

[12:46:01.0000] <Philip`>
(I think that tells it to ignore caches)

[12:47:00.0000] <Philip`>
(though I could be wrong)

[12:48:00.0000] <Dashiva>
Yes

[12:55:00.0000] <gsnedders>
/me is slowly adjusting to this alien browser :P

[12:59:00.0000] <annevk42>
http://jszen.blogspot.com/2009/05/html-5.html -- I'm guessing he's mostly talking about the parts splitted out of HTML5 but it's still nice to hear :)

[13:00:00.0000] <Hixie>
this press has certainly been very encouraging

[13:00:01.0000] <Hixie>
it's almost all positive

[13:51:00.0000] <Philip`>
Hixie: "return to the top of step labeled <i>loop</i>" - should say "*the* step"

[13:52:00.0000] <Hixie>
fixed

[13:53:00.0000] <gsnedders>
/me implemented that, yet didn't notice that.

[13:53:01.0000] <gsnedders>
Oh well, I'm obviously a brilliant reviewer :P

[13:57:00.0000] <gsnedders>
/me wonders why that section is #history-0

[13:57:01.0000] <gsnedders>
I think I do say the order in which ids are generated is undefined.

[14:00:00.0000] <theMadness>
Sigh, people is blind.

[14:01:00.0000] <theMadness>
They didn't get excited about html5 until google told them to.

[14:01:01.0000] <theMadness>
How sad is that?

[14:01:02.0000] <Hixie>
pretty much par for the course

[14:01:03.0000] <gsnedders>
/me wonders if anyone in standards land is the slightest bit surprised by that

[14:02:00.0000] <theMadness>
Heh, gsnedders, I'm allowed to, I'm a lowly developer :D

[14:02:01.0000] <gsnedders>
/me stands up on high

[14:02:02.0000] <gsnedders>
:D

[14:07:00.0000] <jruderman>
"Mozilla executives also promised HTML5 support inside the forthcoming Firefox 3.5 browser." orly

[14:07:01.0000] <theMadness>
I wonder if that includes webforms.

[14:08:00.0000] <theMadness>
or "the part of html5 formerly known as webforms".

[14:08:01.0000] <Hixie>
it's still known as webforms :-P

[14:08:02.0000] <gsnedders>
Alternatively, "HTML 5 forms"

[14:08:03.0000] <Hixie>
or "forms" for short :-P

[14:08:04.0000] <gsnedders>
Or: "Bob's your uncle."

[14:08:05.0000] <theMadness>
Formshizzle.

[14:45:00.0000] <gsnedders>
/me is sure ezyang said he had got tests2.dat working

[14:59:00.0000] <gsnedders>
/me knows he's gonna end up checking over the tree constructor

[15:30:00.0000] <annevk42>
afaik only Opera does HTML5 forms

[15:31:00.0000] <theMadness>
Most of it, yep.

[15:31:01.0000] <theMadness>
Except stuff like placeholder.

[15:31:02.0000] <annevk42>
though WebKit does some I think and I read on IRC that someone from Google was looking into implementing more

[15:32:00.0000] <jgraham>
hmm. I think the mosty recent change to tests2.dat is wrong

[15:33:00.0000] <jgraham>
<head></html><meta> should put <meta> as a child of <head> because </html> in <head> puts you in the after-head mode (not the in-body mode) and in after-head headish elements get appended to the head

[15:34:00.0000] <jgraham>
Anyone fancy confirming that for me before I revert the change?

[15:35:00.0000] <annevk42>
let me check

[15:37:00.0000] <annevk42>
you're wrong

[15:38:00.0000] <jgraham>
Oh

[15:39:00.0000] <jgraham>
Really? Why?

[15:43:00.0000] <Hixie>
you seem correct to me

[15:44:00.0000] <Hixie>
oh no

[15:44:01.0000] <Hixie>
the </html> implies <body> first

[15:44:02.0000] <Hixie>
so you fly through the after head mode, then into in body mode, then into after body, then after after body

[15:44:03.0000] <Hixie>
then you consume the </html>

[15:44:04.0000] <Hixie>
and then the <meta> goes only back as far as in body mode

[15:46:00.0000] <jgraham>
Ah, I forgot to reprocess the </html> in after-head

[15:46:01.0000] <jgraham>
Thanks :)

[15:56:00.0000] <Dashiva>
jgraham: I made the same mistake yesterday, when the fix was discussed

[15:56:01.0000] <Dashiva>
You end up in "in head", but the stack has body. Sneaky!

[16:00:00.0000] <jgraham>
/me thinks he has done his merge the wrong way around

[16:01:00.0000] <jgraham>
So I have ended up with the tip on a branch called svgmathml

[16:12:00.0000] <gsnedders>
Oh, if only HTML were logical…

[16:15:00.0000] <Hixie>
> For example, perhaps the document should advised following

[16:15:01.0000] <Hixie>
> the algorithms only when it is clearly necessary

[16:15:02.0000] <Hixie>
> (SHOULD NOT perform content sniffing EXCEPT when

[16:15:03.0000] <Hixie>
> necessary because of continued misconfiguration of

[16:15:04.0000] <Hixie>
> HTTP servers),

[16:15:05.0000] <Hixie>
/me wonders how that would work

[16:16:00.0000] <Philip`>
You could add a X-Not-Misconfigured header to servers that aren't misconfigured

[16:16:01.0000] <takkaria>
ha

[16:17:00.0000] <Hixie>
lol

[16:26:00.0000] <gsnedders>
Content-Type-But-I-Really-Mean-It-This-Time?

[16:28:00.0000] <Philip`>
Quite a few people use X-Content-Type-Options: nosniff already

[16:28:01.0000] <Dashiva>
Recreate the universe so that all http servers ship with an automatic legacy: true header? :)

[16:28:02.0000] <Philip`>
(including most Google sites)

[16:31:00.0000] <Philip`>
You don't need to recreate the universe, you just need to shift into a worldtrack in which that's already true

[16:37:00.0000] <Hixie>
hsivonen! curse you for filing bug 6766!

[16:38:00.0000] <Hixie>
ok which browser should i base document.all on

[16:38:01.0000] <Hixie>
any browser vendors here want to bribe me to make theirs the compliant one? :-P

[16:46:00.0000] <annevk42>
I'm not sure I'm in favor of Opera's one. Can I get money from you for the reverse favor?

[16:54:00.0000] <Hixie>
who should i use instead?

[16:57:00.0000] <annevk42>
It seems best to be informed by all crappy impl of it

[16:58:00.0000] <Hixie>
heycam: yt?


2009-05-31
[18:47:00.0000] <Hixie>
document.all seems to be a pretty normal collection, with all the elements in it; calling it is the same as dereferencing it

[18:47:01.0000] <Hixie>
and in IE all collections seem to have .tags() method that returns a filtered collection with just the members of that tag name

[18:48:00.0000] <Hixie>
after lowercasing

[18:52:00.0000] <heycam>
Hixie, here now

[18:53:00.0000] <Hixie>
heycam: any chance we can get magic in WebIDL that defines document.all's wacko behavior? :-)

[18:53:01.0000] <Hixie>
i may have asked this before

[18:53:02.0000] <Hixie>
interesting, IE8's namedItem() method broke compatibility with earlier versions

[18:54:00.0000] <Hixie>
i wonder if they did that because of trying to be compatible with the specs or something

[18:54:01.0000] <heycam>
what's the wacko behaviour?

[18:57:00.0000] <Hixie>
if (document.all) { throw 0; } doesn't throw

[18:57:01.0000] <gavin_>
in non-IE browsers only, right?

[18:58:00.0000] <Hixie>
typeof document.all === undefined

[18:58:01.0000] <gavin_>
Mozilla had to make it "undetectable" when we implemented it, because sites used that to check for IE

[18:58:02.0000] <Hixie>
document.all.toString() is something like "[object HTMLCollection]"

[18:58:03.0000] <Hixie>
(will probably be [object HTMLAllCollection] in HTML5)

[18:59:00.0000] <Hixie>
document.all instanceof HTMLCollection === false

[18:59:01.0000] <heycam>
such wackoness seems like it shouldn't be in webidl

[18:59:02.0000] <Hixie>
document.all.length == document.getElementsByTagName('*').length

[19:00:00.0000] <Hixie>
the object returned by document.all is callable and needs WebIDL's property indexing stuff

[19:00:01.0000] <heycam>
ok so that bit should be supported already

[19:00:02.0000] <Hixie>
the main reason i was hoping you'd be ok with adding that to webidl is i have no idea what i should be saying in html5 :-)

[19:00:03.0000] <heycam>
aha :)

[19:01:00.0000] <heycam>
there's no way in ecma-262 to legally have an object that is "undetectable" like that

[19:02:00.0000] <Hixie>
yeah well we already violate e262 for one thing, what's one more!

[19:02:01.0000] <heycam>
i don't though, my hands are clean thus far =)

[19:03:00.0000] <Hixie>
hah

[19:04:00.0000] <heycam>
do people really do   if (document.all instanceof HTMLCollection)  to test for IE?

[19:04:01.0000] <Hixie>
no

[19:05:00.0000] <Hixie>
they probably do do typeof

[19:05:01.0000] <Hixie>
but most just do if (document.all)

[19:05:02.0000] <Hixie>
so that would probably be enough

[19:05:03.0000] <heycam>
so why the need for that wacko requirement?

[19:06:00.0000] <Hixie>
trying to be as close as possible to existing implementations

[19:07:00.0000] <heycam>
just wonder if it's really necessary

[19:08:00.0000] <heycam>
anyway

[19:08:01.0000] <heycam>
i'm not sure what the best way is to describe the "undetectable" requirements is

[19:09:00.0000] <heycam>
requirements on how certain productions from the ecmascript grammar are evaluated maybe?

[19:10:00.0000] <heycam>
/me goes out for some shopping

[19:59:00.0000] <Hixie>
heycam: i have bad news. It looks like document.all is not the only feature that uses this in webkit -- style.filter does too :-)

[20:00:00.0000] <jwalden>
yeah, it's something ghastly where it converts to a boolean false

[20:00:01.0000] <jwalden>
dunno why it's necessary, gecko doesn't have the same to not trigger MS filter-property detection tests

[20:00:02.0000] <jwalden>
or at least it doesn't as far as I know

[20:00:03.0000] <Hixie>
gecko doesn't have style.filter at all does it?

[20:01:00.0000] <jwalden>
via SVG maybe?

[20:01:01.0000] <jwalden>
/me isn't sure

[20:01:02.0000] <Hixie>
document.body.filter is undefined in gecko

[20:01:03.0000] <Hixie>
er

[20:01:04.0000] <Hixie>
document.body.style.filter is undefined in gecko

[20:01:05.0000] <jwalden>
k

[21:07:00.0000] <Hixie>
i really think we should number the design principles instead of naming them

[21:07:01.0000] <Hixie>
people are only reading their names

[00:03:00.0000] <ezyang>
Anyone around to answer a question about fragment processing? 

[00:04:00.0000] <Hixie>
fragment processing?

[00:04:01.0000] <ezyang>
Parsing html fragments 

[00:04:02.0000] <ezyang>
Although... it looks like I just fixed it :-) 

[00:05:00.0000] <ezyang>
Perhaps "and set node to the context  element." is a little ambiguous (step 3 of reset insertion mode algorithm) 

[00:08:00.0000] <Hixie>
/me looks

[00:08:01.0000] <ezyang>
The algorithm works if I replace the actual entry in the stack of open elements 

[00:09:00.0000] <Hixie>
i don't understand why it's ambiguous or why you would mutate the stack

[00:10:00.0000] <ezyang>
Ah. Then I've implemented it wrong 

[00:10:01.0000] <Hixie>
how do you interpret it?

[00:11:00.0000] <ezyang>
Oh, I see where I've gone wrong. I'm assuming things actually get inserted into the context element. 

[00:11:01.0000] <ezyang>
User error! Boink 

[00:11:02.0000] <Hixie>
the reset the insertion mode thing does nothing but reset the insertion mode :-)

[00:12:00.0000] <ezyang>
Yep 

[00:14:00.0000] <ezyang>
I always forget if the "last" node on a stack is the most recently added one or the oldest one 

[00:15:00.0000] <hobo>
you'd think it would be the most recent one

[00:15:01.0000] <hobo>
as you add them it increases the stack so the last one would be the most recently added

[00:16:00.0000] <ezyang>
For step three, "If node is the first node in the stack of open elements, then set last to true and set node to the context  element.", do I set node to the context element if there is no context element? 

[00:16:01.0000] <Hixie>
yeah i apologise for the mess around the stack

[00:16:02.0000] <Hixie>
ezyang: how can there be no context element?

[00:17:00.0000] <ezyang>
Umm, when I'm not parsing a fragment? 

[00:18:00.0000] <Hixie>
when you're not passing a fragment, /node/ will never be the first node in the stack

[00:18:01.0000] <ezyang>
OK. 

[00:18:02.0000] <Hixie>
so you can assert(context is not null) in that condition, if you like -- you should never hit it unless you or the spec has a bug

[00:19:00.0000] <ezyang>
Looking at "reset insertion mode" more closely, I can see that would be the case. 

[00:23:00.0000] <ezyang>
FIFTEEN! 

[00:23:01.0000] <ezyang>
(failures) 

[00:24:00.0000] <Hixie>
heycam: yt?

[00:31:00.0000] <ezyang>
Ugh. Now I need to incorporate a list of doctypes in the library 

[01:46:00.0000] <ezyang>
Are there any tests in the test-suite for quirks mode specific behavior? 

[01:49:00.0000] <ezyang>
Also, I'm pretty sure this is a bogus test: <!doctype html></html> <head> 

[01:50:00.0000] <ezyang>
since </html> puts the mode to "After After Body", then the whitespace gets handled as if it was "In Body" and should get placed in <body> 

[01:53:00.0000] <Hixie>
</html> doesn't do anything before head iirc

[01:55:00.0000] <ezyang>
"before head" causes </html> to be reprocessed as if an implicit <head> tag was seen 

[01:55:01.0000] <ezyang>
Which further reprocesses as if an end </head> was seen 

[01:55:02.0000] <ezyang>
and so forth 

[01:56:00.0000] <Hixie>
really? i thought we changed that

[01:56:01.0000] <ezyang>
Oh I see, we're in "before html", not "before head" 

[01:56:02.0000] <Hixie>
huh go figure, i am wrong

[01:56:03.0000] <Hixie>
oh well

[01:57:00.0000] <ezyang>
Ehh, the behavior is the same 

[01:57:01.0000] <ezyang>
Yeah. 

[01:57:02.0000] <ezyang>
So what's supposed to happen? 

[01:57:03.0000] <Hixie>
what the spec says is what the rule is at the moment

[01:57:04.0000] <Hixie>
no idea if it'll change

[01:58:00.0000] <ezyang>
Ok. I guess I should hg blame the relevant test-case and yell at hsivonen if it's one of his nudge nudge things (more realistically, stick the test-case in test99) 

[02:00:00.0000] <ezyang>
Ick. hg blame doesn't show deletions. So I *can't* find out 

[02:01:00.0000] <ezyang>
Someone must have changed it to not result in whitespace at some point, though, so it will get punted to 99 

[02:05:00.0000] <ezyang>
Wow. Most of these are bogus 

[02:06:00.0000] <ezyang>
Ok. The spec needs to be changed to handle <html></html><!-- comment --> 

[02:10:00.0000] <Hixie>
send mail or file a bug if you need changes

[02:11:00.0000] <Hixie>
(i can't track irc comments so otherwise i'll just forget)

[02:12:00.0000] <ezyang>
Ok. 

[02:13:00.0000] <ezyang>
It's also kind of debatable whether or not these changes are correct or not 

[02:18:00.0000] <Hixie>
it basically boils down to "are there pages that break if we do it the other way"

[02:18:01.0000] <ezyang>
Since <title> isn't going into <head>, probably. 

[02:29:00.0000] <ezyang>
For <a><div><p></a>, is the div supposed to be a child of <a>? 

[02:31:00.0000] <ezyang>
Oh, it's a foster parented case. 

[02:47:00.0000] <heycam>
Hixie, you were after me before?

[03:08:00.0000] <ezyang>
PHP implementation of html5lib has ~100% coverage (there is one error where we can't actually express a doctype with no qualified name with libxml, but whatever) 

[03:09:00.0000] <ezyang>
Time to sleep 

[05:06:00.0000] <annevk42>
heycam, style.filter is the same as document.all due to the SVG WG not renaming it to avoid conflicts with IE

[05:13:00.0000] <othermaciej>
well, WebKit does style.filter that way

[05:13:01.0000] <othermaciej>
though it's even weirder, since the return is normally a string

[05:14:00.0000] <othermaciej>
we have to make it a special string-like thing that masquerades as undefined

[05:14:01.0000] <othermaciej>
dunno what other browsers do

[05:27:00.0000] <annevk42>
I think we do something else but it is causing issues

[05:41:00.0000] <Philip`>
/me is saved by semicolon insertion

[05:42:00.0000] <Philip`>
For some inexplicable reason I was accidentally printing a date in the middle of a <script> generated by some code, so there was a spurious line saying "2009-05-31"

[05:43:00.0000] <Philip`>
and it must have been interpreted as a statement computing 1973 with an implicit semicolon at the end, so it didn't cause a syntax error

[05:43:01.0000] <Philip`>
and it's been like that for about six months

[06:50:00.0000] <gsnedders>
So, hmm…

[06:51:00.0000] <gsnedders>
/me realizes what he was going to ask

[06:53:00.0000] <thomaslee>
hi all -- am working with the canvas element in the firefox 3.5 beta -- is this an appropriate place to ask questions about the behaviour of the API?

[07:03:00.0000] <gsnedders>
thomaslee: Yes.

[07:03:01.0000] <thomaslee>
cool

[07:03:02.0000] <gsnedders>
/me can't actually answer any questions on the canvas API himself, but can say that this is the right place to ask :P

[07:04:00.0000] <thomaslee>
haha great :)

[07:04:01.0000] <thomaslee>
that's fine

[07:04:02.0000] <thomaslee>
question is:

[07:04:03.0000] <thomaslee>
ctx.moveTo(canvas.width, 0); ctx.lineTo(canvas.width, canvas.height);

[07:06:00.0000] <thomaslee>
once you ctx.stroke(), firefox's implementation of canvas, only draws half of the line -- the other half presumably "off-canvas".

[07:06:01.0000] <thomaslee>
oops, trigger happy with the commas

[07:06:02.0000] <gsnedders>
/me grumbles about ezyang having made php html5lib really slow

[07:06:03.0000] <thomaslee>
anyway, the question is -- is this expected behaviour?

[07:07:00.0000] <annevk42>
I think so

[07:07:01.0000] <thomaslee>
the other thing that's confusing me is for a lineWidth = 1.0, I'm seeing what looks like two pixel lines -- although I'm assuming this has something to do with coordinate space and firefox's internal representation of the canvas.

[07:08:00.0000] <Philip`>
thomaslee: Integer coordinates refer to the positions *between* pixels

[07:08:01.0000] <thomaslee>
annevk42: so assuming I want a line down the right hand side, I should be drawing this line for x=canvas.width-1 ?

[07:08:02.0000] <gsnedders>
ezyang: You've made it go from taking < 6s to tokenize the spec to taking > 131s

[07:09:00.0000] <thomaslee>
Philip`: come again? :)

[07:09:01.0000] <Philip`>
thomaslee: so if you stroke a line with integer coordinates, it goes between two columns of pixels and ends up shading pixels in both columns

[07:09:02.0000] <gsnedders>
Oh, wait, I ran that with XDebug on.

[07:09:03.0000] <thomaslee>
Philip`: oh okay, wow. why's that?

[07:09:04.0000] <Philip`>
thomaslee: If you want to draw a line through just a single column of pixels, you have to shift the coordinates by 0.5

[07:10:00.0000] <Philip`>
thomaslee: i.e. ctx.moveTo(canvas.width-0.5, 0); ctx.lineTo(canvas.width-0.5, canvas.height)

[07:10:01.0000] <annevk42>
unless the line has a width of 2n, presumably?

[07:10:02.0000] <gsnedders>
ezyang: Let me try that again: you've made it go from < 6s to > 11s

[07:10:03.0000] <Philip`>
annevk42: If it has width 2n and you're trying to draw a single column of pixels, that's never going to work :-)

[07:11:00.0000] <annevk42>
meh

[07:12:00.0000] <thomaslee>
Philip`: right, so that would explain why I'm still seeing two pixel lines when reducing line width. But why on earth do integer coordinates refer to the space in between pixels?

[07:12:01.0000] <Philip`>
thomaslee: It's that way so that fills work like you would expect (with sharp edges), but it has the consequence that strokes don't quite work like you expect (so you have to shift them by 0.5 to the centers of pixels)

[07:13:00.0000] <Philip`>
thomaslee: e.g. fillRect(10, 10, 1, 1) fills between the corners of pixels, instead of filling a rectangle between the centers of four pixels

[07:13:01.0000] <Philip`>
(...assuming you imagine pixels as being squares)

[07:14:00.0000] <thomaslee>
Philip`: okay, I think I understand. Going to have to remember that one!

[07:16:00.0000] <Philip`>
thomaslee: It's not the most intuitive thing in the world :-)

[07:16:01.0000] <Philip`>
but I can't imagine any way to make it better without making other things worse

[07:16:02.0000] <thomaslee>
Philip`: my brain hurts already :)

[07:17:00.0000] <thomaslee>
fair enough ... I only started messing with canvas yesterday, maybe it'll make sense when I've got a better understanding of it all.

[07:17:01.0000] <thomaslee>
anyway, thanks very much for your help

[07:20:00.0000] <heycam>
annevk42, we did consider renaming it recently

[07:20:01.0000] <heycam>
not sure it's worth it

[07:24:00.0000] <gsnedders>
/me has a brief wish PHP did lazy evaluation

[07:25:00.0000] <thomaslee>
Philip`: just to be clear, the 0.5 coordinate offset is expected behaviour as per the spec, right? (as in, I'm assuming this isn't a firefox-specific thing)

[07:27:00.0000] <Philip`>
thomaslee: The spec doesn't really define the mapping between rendering concepts and physical pixels, e.g. it doesn't talk about antialiasing and it doesn't limit implementations to one device pixel per canvas pixel

[07:27:01.0000] <Philip`>
thomaslee: but all current implementations do the same as Firefox here

[07:27:02.0000] <Philip`>
thomaslee: so in reality it's the behaviour you should expect

[07:31:00.0000] <thomaslee>
Philip`: great, thanks.

[07:42:00.0000] <gsnedders>
Does "reconstruct the active formatting elements" step 7 need to have after it a check that it isn't a marker?

[08:17:00.0000] <annevk42>
someone is asking me to be added to html5lib team members so he write documentation? no need for vetting right?

[08:18:00.0000] <gsnedders>
annevk2: I think generally we've had at least one person agreeing to it

[08:19:00.0000] <annevk42>
so one in total or two?

[08:19:01.0000] <gsnedders>
One in total.

[08:20:00.0000] <annevk42>
i guess that'd be me then

[08:21:00.0000] <annevk42>
for reference: juguang

[09:54:00.0000] <ezyang>
gsnedders: Test cases pass first. Profiling (and not until that) and then optimization later. 

[10:08:00.0000] <gsnedders>
ezyang: Peh. I was just taking the same solution as Python for this :)

[10:08:01.0000] <gsnedders>
ezyang: Also: parsing the spec fails.

[10:08:02.0000] <ezyang>
It... errors? 

[10:08:03.0000] <ezyang>
Wow. 

[10:09:00.0000] <ezyang>
That means our test-coverage is not good enough. 

[10:09:01.0000] <gsnedders>
Fatal error: Call to a member function cloneNode() on a non-object in /Users/gsnedders/Documents/Stuff I'm Working On/html5lib/php/library/HTML5/TreeConstructer.php on line 3037

[10:09:02.0000] <gsnedders>
The non-object is int(300)

[10:09:03.0000] <ezyang>
Huh. That should never happen. 

[10:09:04.0000] <gsnedders>
Well it does. :D

[10:10:00.0000] <ezyang>
Oh, hey, that's the marker 

[10:10:01.0000] <ezyang>
Ok, so we need to figure out why that algorithm is failing 

[10:10:02.0000] <ezyang>
Do you have a minimal test-case? 

[10:10:03.0000] <gsnedders>
No.

[10:10:04.0000] <Philip`>
curl http://www.whatwg.org/specs/web-apps/current-work/ -o html5lib/testdata/treeconstruction/tests13.dat

[10:10:05.0000] <Philip`>
That'll encourage people to optimise their implementations

[10:10:06.0000] <gsnedders>
ezyang: that.

[10:10:07.0000] <ezyang>
Heh 

[10:10:08.0000] <ezyang>
Will do. 

[10:11:00.0000] <gsnedders>
/me wonders if we should actually do that…

[10:11:01.0000] <ezyang>
I wish we had better names for our test-cases 

[10:11:02.0000] <gsnedders>
What could be better than a number? :P

[10:12:00.0000] <Philip`>
I wouldn't object to regrouping and renaming the tests

[10:12:01.0000] <ezyang>
It's just that, some of the test-cases make assumptions about the naming of the tests 

[10:13:00.0000] <ezyang>
gsnedders: It would be super-uber-awesome if you could find a minimal test-case that tickles the bug. 

[10:13:01.0000] <gsnedders>
No, it would be super-über-awesome.

[10:13:02.0000] <gsnedders>
uber isn't a word, damnit!

[10:17:00.0000] <gsnedders>
Last token emitted is with stream at Line  4667, column 5

[10:18:00.0000] <jgraham>
/me would positivly welcome renaming the tests but doesn't want to actually spend time doing it

[10:18:01.0000] <gsnedders>
<dfn title="dom-uda-protocol">

[10:23:00.0000] <ezyang>
Is it conceptually clean for me to refer to elements in the SVG namespace as 'svg:foo'? I know that prefixes can change, but this is the most convenient representation 

[10:25:00.0000] <gsnedders>
ezyang: <table><tr><td><code>protocol</code> </table>

[10:26:00.0000] <ezyang>
Awesome 

[10:29:00.0000] <gsnedders>
ezyang: s/protocol//

[10:31:00.0000] <gsnedders>
/me has no idea which test file to put the test in

[10:32:00.0000] <ezyang>
There's a bunch of foster parenting tests in... erm... 7, I think 

[10:32:01.0000] <ezyang>
(this is why we need better names) 

[10:33:00.0000] <gsnedders>
Yeah, I think 7 is best.

[10:34:00.0000] <gsnedders>
/me runs it against Python to make sure something agrees with the parse tree

[10:35:00.0000] <gsnedders>
/me gets auth failed :\

[10:35:01.0000] <gsnedders>
ezyang: Pushed

[10:36:00.0000] <ezyang>
Awesome 

[10:37:00.0000] <gsnedders>
Have fun with your test suite which throws a fatal error :P

[10:37:01.0000] <ezyang>
I'm finishing XForeign support first 

[10:37:02.0000] <gsnedders>
/me notes the spec is a good test document because it is so huge and does so much

[10:38:00.0000] <Philip`>
/me notes the spec is a bad test document because it's very repetitive and is valid

[10:38:01.0000] <gsnedders>
ezyang: I take it you're fixing the fact that HTML elements should go into the HTML namespace?

[10:38:02.0000] <gsnedders>
Philip`: But it does everything that's valid, more or less :P

[10:39:00.0000] <Philip`>
gsnedders: That seems unlikely :-p

[10:39:01.0000] <ezyang>
gsnedders: I assume that's the default behavior. I suppose I could fix that trivially though 

[10:39:02.0000] <gsnedders>
Well, it's almost infinitely long :P

[10:39:03.0000] <Philip`>
gsnedders: Why is that relevant?

[10:40:00.0000] <Philip`>
gsnedders: I could make a document that's infinitely long and consists entirely of whitespace, and it wouldn't test much behaviour

[10:40:01.0000] <gsnedders>
Philip`: Indeed

[10:40:02.0000] <gsnedders>
Philip`: But you'd need an infinitely long document to try every possible valid character stream

[10:41:00.0000] <Philip`>
gsnedders: An infinitely long document could only test precisely one character stream

[10:41:01.0000] <gsnedders>
ezyang: It _should_ just mean changing to createElementNS

[10:41:02.0000] <ezyang>
Right, but because the tokenizer+parser are finite state machines 

[10:41:03.0000] <Philip`>
gsnedders: You'd need infinitely many documents if you want to test infinitely many character streams

[10:41:04.0000] <ezyang>
We don't need an infinite input stream to test all state combinations 

[10:42:00.0000] <gsnedders>
Philip`: Indeed.

[10:42:01.0000] <gsnedders>
ezyang: If we implement both as finite state machines exactly to spec, yes :P

[10:42:02.0000] <Philip`>
gsnedders: But the interesting state in an implementation is finite, so a finite set of finite test cases could test all the interesting cases

[10:43:00.0000] <jgraham>
Philip`: And your're going to autogenreate tests to cover them all, right? ;)

[10:43:01.0000] <Philip`>
(By "interesting", I mean that e.g. "<br><br>" might be interesting but "<br><br><br>" is very unlikely to be)

[10:43:02.0000] <gsnedders>
ezyang: Though it may well make <foo:bar> throw an error

[10:43:03.0000] <gsnedders>
/me sighs

[10:43:04.0000] <Philip`>
jgraham: That's what I did for the tokeniser :-)

[10:43:05.0000] <Philip`>
(or at least attempted to)

[10:43:06.0000] <Philip`>
jgraham: and if I ever got around to finishing my tree-constructor implementation, I suppose I could generate tests for that too

[10:43:07.0000] <ezyang>
I think it would be possible to programatically generate tree-constructer tests 

[10:44:00.0000] <gsnedders>
Philip`: It's OCaml, right?

[10:44:01.0000] <jgraham>
Philip`: I know and I know, respectively

[10:44:02.0000] <Philip`>
gsnedders: Yes

[11:08:00.0000] <ezyang>
gsnedders: I'm looking at your new SPACECHARACTERS token type, and I'm thinking that's not actually a good idea. 

[11:08:01.0000] <gsnedders>
ezyang: It's what Python does.

[11:08:02.0000] <ezyang>
It diverges from spec, and anywhere we matched just CHARACTERS, we have to match against both 

[11:08:03.0000] <ezyang>
Sure. 

[11:09:00.0000] <ezyang>
The way I would personally implement it, though, would be as an advisory extra flag placed in the token 

[11:09:01.0000] <gsnedders>
ezyang: Diverging from spec is not a problem.

[11:09:02.0000] <ezyang>
Adding a new token type is a fairly large divergence 

[11:12:00.0000] <Hixie>
heycam: what i was going to say is that it looks like style.filter also needs this magic (specifically it needs to return a string that masquerades as undefined)

[11:12:01.0000] <Hixie>
othermaciej: the way i specced document.all was to say it ToBoolean()s to false, but is otherwise a normal HTMLCollection

[11:13:00.0000] <Hixie>
othermaciej: is that not enough? I couldn't work out how to make it masquerade as 'undefined' for the purposes of the JS spec

[11:13:01.0000] <jgraham>
Hixie: So instanceof and typeof will work like any other HTMLCollection?

[11:13:02.0000] <jgraham>
/me has no idea if that is a problem

[11:14:00.0000] <Hixie>
yeh

[11:42:00.0000] <gsnedders>
Hmm, abath isn't around

[11:42:01.0000] <gsnedders>
*abarth

[11:58:00.0000] <Hixie>
christ i wish larry would follow the process the chairs set out and file bugs instead of sending all these e-mails where i have to carefully parse each line to see if there's a change request there

[12:13:00.0000] <jgraham>
Hixie: What would "masqurading as undefined" imply for document.all?

[12:14:00.0000] <Hixie>
doing whatever it is webkit or gecko do

[12:14:01.0000] <Hixie>
but i don't know how to express that in terms of the js spec

[12:15:00.0000] <Dashiva>
That's a tall order

[12:17:00.0000] <hsivonen>
Hixie: wouldn't another "willful violation" be less work for Gecko and WebKit developers?

[12:18:00.0000] <Hixie>
i mean i don't know how to phrase the willful violation

[12:18:01.0000] <jgraham>
Hixie: so document.all === undefined or more?

[12:18:02.0000] <Hixie>
more

[12:18:03.0000] <Hixie>
try it

[12:18:04.0000] <Hixie>
i can't describe it

[12:18:05.0000] <jgraham>
I am trying it. I want to know what to try :)

[12:18:06.0000] <Hixie>
typeof does weird things, instanceof does weird things, it matters if you're in with() or not, all kinds of weird things

[12:18:07.0000] <Hixie>
depends on webkit vs gecko vs opera, too

[12:19:00.0000] <jgraham>
typeof and instanceof seem easy to deal with, === seems hard to deal with

[12:19:01.0000] <Dashiva>
Yeah, I remember the bug discussion in Opera's BTS

[12:20:00.0000] <jgraham>
Breaking document.all in 'with' seems like it shouldn't be a big deal

[12:21:00.0000] <annevk2>
I recall discussions about it requiring fairly low-level changes to the ECMAScript engine

[12:22:00.0000] <hsivonen>
perhaps ES5 should provide a spec hook for this

[12:23:00.0000] <annevk2>
I don't think they're too interested in things like that

[12:23:01.0000] <Dashiva>
Boolean value false, typeof undefined.

[12:23:02.0000] <annevk2>
http://wiki.whatwg.org/wiki/Web_ECMAScript

[12:24:00.0000] <Hixie>
jgraham: it's actually harder to break it than not

[12:24:01.0000] <Hixie>
right now the html5 spec just says it ToBoolean()s to fdalse

[12:24:02.0000] <Hixie>
false

[12:25:00.0000] <Hixie>
anyway, i'm outta here

[12:25:01.0000] <Dashiva>
Should probably mention a typeof override too?

[12:27:00.0000] <jgraham>
Hixie: I meant that making it work or not in with seems like it shouldn't be too big a deal

[12:27:01.0000] <jgraham>
But I guess I could be wrong about that

[12:27:02.0000] <jgraham>
/me wishes for the death of "with"

[12:29:00.0000] <Dashiva>
There's that reshaping the universe to your whims thing again

[15:32:00.0000] <annevk42>
/me has the feeling Larry and Roy are both making issues appear more easily than they are; maybe even intentionally

[15:33:00.0000] <annevk42>
e.g. Larry ignores most of the comments against having versioning for character encoding sniffing and tries to convince people to look at the general idea and Roy somehow believes the URL issue is restricted to HTML src/href attributes

[15:48:00.0000] <annevk42>
takkaria, <canvas> is perfectly safe

[15:48:01.0000] <annevk42>
takkaria, even some amount of scripting can be safe

[15:49:00.0000] <annevk42>
just has to be sandboxed

[15:50:00.0000] <takkaria>
browsers don't currently do any kind of sandboxing though

[15:50:01.0000] <takkaria>
and until they all do, a preparse and whitelist will be required

[15:52:00.0000] <annevk42>
still, canvas can be on that whitelist just fine :)

[15:52:01.0000] <gsnedders>
annevk2: But whitelisting script is a lot harder :)

[15:53:00.0000] <takkaria>
annevk42: mm, but it's in effect useless unless you happen to be able to be able to whitelist the code that paints on it, which is really what I meant

[15:54:00.0000] <annevk42>
gsnedders, you wanna sandbox what it executes, not whitelist what it contains, methinks

[15:54:01.0000] <gsnedders>
annevk42: How do you sandbox in currently shipping browsers, then?

[16:17:00.0000] <jgraham>
Wow is Laura really azsking for quantitative principles?

[16:17:01.0000] <jgraham>
That's... crazy

[16:19:00.0000] <jgraham>
At the very least the demand that the world be black or white seems unreasonable and unrealistic

[16:24:00.0000] <weinig>
Hixie: I don't see a dataunavailable event being dispatched anywhere, is that from some other spec? 

[16:42:00.0000] <hober>
gave a talk today at our local BarCamp about extensibility & HTML http://edward.oconnor.cx/2009/BarCamp-San-Diego-5/

[16:43:00.0000] <hober>
(more-or-less a YAGNI rant about namespaces & distributed extensibility)