02:05
<Hixie>
MikeSmith: done
02:06
<MikeSmith>
thanks
02:08
<MikeSmith>
I'll get the author-edition published in TR space along with the multipage
02:08
<MikeSmith>
for next round of WDs
04:58
<karlcow>
looking at "Proposing New Features" at the bottom of this page http://www.w3.org/MarkUp/HTML-WG/
05:19
<MikeSmithX>
"We value your opinion, really we do, but we're going to put a stiff tax on its expression, so that we won't have to hear it very often."
05:23
<MikeSmith>
+1
07:01
<hsivonen>
gsnedders: OK. I was hoping to get advice on which seats I should get...
07:15
<hsivonen>
AryehGregor: It's a bit of a myth that Hixie decides everything alone. Here's an example of influencing the decision process: http://krijnhoetmer.nl/irc-logs/whatwg/20090212#l-27
07:18
<hsivonen>
MikeSmith: maybe the Author Edition should be called something like an Edition for Web Authors
07:18
<hsivonen>
MikeSmith: otherwise it looks like the Editor's Cut
07:21
<hsivonen>
I still have trouble deciding if I should get the Avenue Q tickets from the Stalls or the Royal Circle (the first balcony level)...
07:22
<MikeSmith>
hsivonen: good point
07:44
<MikeSmith>
hsivonen: changed now to "Edition for Web Authors"
07:44
<MikeSmith>
http://dev.w3.org/html5/spec-author-view/
07:45
<hsivonen>
MikeSmith: thanks
07:45
<MikeSmith>
hsivonen: btw, did you see my question yesterday about how to make the v.nu HTML5 parser usable by other Java apps from the command line?
07:45
MikeSmith
goes to get pointer to logs
07:46
<MikeSmith>
http://krijnhoetmer.nl/irc-logs/whatwg/20100602#l-181
08:54
<hsivonen>
http://html4all.org/mailman/archives/list_html4all.org/2010-June/001066.html
09:01
<zcorpan_>
hsivonen: surely the reason is that Hixie wrote the text instead of the Experts
09:04
<Dashiva>
zcorpan_: A mutation of NIH?
09:07
<jgraham>
hsivonen: You are going to London?
09:08
<annevk>
bleh
09:09
<hsivonen>
jgraham: yes. In July.
09:09
<jgraham>
hsivonen: Nice. Holiday?
09:10
<hsivonen>
jgraham: yeah, a short one, since I'd have a layover at LHR anyway
09:10
<annevk>
http://blog.chromium.org/2010/06/websocket-protocol-updated.html
09:11
<hsivonen>
I booked tickets for Avenue Q. gsnedders is effective at this social media marketing thing.
09:11
<jgraham>
What you mean he convinced you to go without ever actually having gone himself or having any idea of product quality?
09:12
<hsivonen>
jgraham: right. very effective indeed.
09:13
<annevk>
haha
09:14
<hsivonen>
great to see that Chromium is tracking the spec changes for Web Socket
09:19
<hsivonen>
working with hg on a hard disk feels very slow after getting used to an SSD
09:26
<annevk>
http://twitter.com/mikeplate/statuses/15247455331
09:26
<annevk>
'"The most unreadable specification I've read in 20 years of standardization" Daniel Glazman on WhatWG #html5 doc spec #swdc'
09:27
hsivonen
wonders if the other specs that Daniel has read have been as precise
09:27
Dashiva
wonders if he has read any other specs at all
09:28
jgraham
wonders if the other specs will lead to such interoperable behaviour in such a complex area
09:30
annevk
wonders how carefully he read CSS 2.1 :-P
09:31
<workmad3>
annevk: surely you're not suggesting that the CSS 2.1 spec was unreadable! :)
09:31
<annevk>
I'm suggesting it is pretty hard to read
09:32
<annevk>
And incomplete unfortunately on certain important topics...
09:32
<Dashiva>
I can never remember where to find half the layout properties
09:33
<jgraham>
I think my serious response would be http://dbaron.org/log/20100531-specs
10:14
<annevk>
someone emailed me again the other day about href on any element... gave an explanation plus pointer to the FAQ entry
10:18
<aho>
easy way: you should have just pointed out that it's now fine to put whatever you want into a-elements :>
10:22
<annevk>
he was actually aware of that
10:25
<mikekelly>
I'm interested in understand why it isn't possible for javascript to kick off a page transition and have granular control over the request in the same as you do with XHR - anyone know?
10:27
<zcorpan_>
mikekelly: because there's only one person in the world who wants to do that
10:35
<gsnedders>
hsivonen: And Hixie was good at convincing me I should go (though mainly F2F, so no IRC logs of this)
10:38
<zcorpan_>
jgraham: who's marlin? did you mean malin?
10:39
<jgraham>
zcorpan_: Probably
10:39
<Dashiva>
He's implying she's fishy
10:40
<jgraham>
I consider it a good day if I manage to spell my own name right
10:51
<annevk>
jgraham, low standards much?
10:52
<gsnedders>
jgraham: Your spelling it better than mine. :P
10:52
<annevk>
though I guess it's a good thing it doesn't depend on externalities, like someone else spelling your name right for a change...
10:52
<gsnedders>
(There again, that is a really low standard.)
10:52
<jgraham>
annevk: I like to use the word "realistic"
12:30
<hsivonen>
Does anyone happen to have an URL for a style sheet that's guarenteed to load artificially slowly?
12:30
<Rik`>
hsivonen: steve souders should have that with cuzillion
12:33
<jgraham>
hsivonen: I guess OPera internal urls aren't that useful to you ;)
12:33
<Rik`>
hsivonen: http://stevesouders.com/cuzillion/
12:34
<jgraham>
Rik`: Neat
12:36
<hsivonen>
Rik`: thanks
12:36
<Rik`>
yeah, pretty useful
13:09
<hsivonen>
what's a cross-browser (incl. IE) way to ask an element for its color?
13:09
<Dashiva>
None that I know of, you have to combine getComputedStyle and currentStyle
13:10
<Rik`>
and don't forget the please argument
13:11
<Dashiva>
brb, implementing intercal support in webkit
13:12
<hsivonen>
ok. maybe I'll try to see if Opera and WebKit support offsetHeight
13:12
<Dashiva>
You can't just detect getComputedStyle and fallback to currentStyle?
13:13
<hsivonen>
Dashiva: I just want to check if a test style sheet is applied at a given point
13:13
<hsivonen>
I don't need to check arbitrary styles.
13:30
<zcorpan_>
offsetHeight should work
13:34
<hsivonen>
as I expected, none of the top 4 engines behaves like the spec says when a document.written internal script occurs right after an external style sheet in the same document.write
13:35
<hsivonen>
http://hsivonen.iki.fi/test/moz/sheet-blocking-script.html
13:35
<hsivonen>
time to file a spec bug
13:36
<hsivonen>
it took a lot of trying to come up with a test that explained the behavior of all 4
13:54
<Dashiva>
And there are _still_ people who object to the design principles, wonderful
14:02
<hsivonen>
whew. http://www.w3.org/Bugs/Public/show_bug.cgi?id=9843 filed.
14:23
<hsivonen>
is there a service that lets me search Hixie's commit messages?
14:23
<hsivonen>
or browse blame for the spec?
14:24
<Dashiva>
If there isn't, you could search the archives for commit-watchers
14:32
<jgraham>
hsivonen: I assume you don't count the avaliability of svn for checkout as "a service"
14:33
<hsivonen>
jgraham: isn't it slow? and no Web UI?
14:34
hsivonen
needs to find the SVG URL
14:34
hsivonen
guesses correctly
14:38
<jgraham>
hsivonen: Yes
14:38
<hsivonen>
OK, now I need to find the list discussion that motivated http://html5.org/tools/web-apps-tracker?from=13&to=14
14:47
<hsivonen>
I was unable to find a public discussion leading to that change.
14:48
<gsnedders>
r14!? Man, that's old sk00l.
16:30
<crankharder>
this manifest should hit /offline/update_all_events regardless of cache/network status right?
16:30
<crankharder>
http://pastie.org/990656
17:34
<AryehGregor>
Hmm. <input required> in Chrome dev channel seemingly doesn't submit the form, but gives no error.
17:34
<AryehGregor>
It just focuses the password field.
17:35
<AryehGregor>
I mean, the required field.
17:35
<remysharp>
AryehGregor: it's flagged as invalid on the dom node - you have to go all JS on the element
17:35
<AryehGregor>
That seems like a really bad feature to deploy. If you don't have UI, you shouldn't enforce the requirements either.
17:36
<remysharp>
AryehGregor: completely agree
17:36
<AryehGregor>
The form validation API is not meant to be JS-only. It's meant to work even if the author provides no JS.
17:37
<AryehGregor>
Are there minimal UI requirements in the spec here?
17:37
AryehGregor
is looking
17:37
<gsnedders>
No
17:37
gsnedders
would hope there aren't any UI requirements
17:37
<AryehGregor>
Not even "you have to tell the user and not just silently fail"?
17:37
<AryehGregor>
That seems like a pretty reasonable UI requirement to me.
17:37
<gsnedders>
What if it's Googlebot?
17:38
<AryehGregor>
I'm talking about browsers here. Googlebot is not a browser.
17:39
<AryehGregor>
A Googlebot does not fall into the "Web browsers and other interactive user agents" category.
17:39
<AryehGregor>
It's "Data mining tools", I suppose.
17:39
<AryehGregor>
So it only has to follow the semantic requirements, not UI requirements.
17:43
<AryehGregor>
http://www.whatwg.org/specs/web-apps/current-work/multipage/association-of-controls-and-forms.html#constraint-validation "Report the problems with the constraints of at least one of the elements given in unhandled invalid controls to the user."
17:43
<AryehGregor>
It explicitly suggests focusing the element, but that's just not enough. There needs to be some text somehow.
17:45
<AryehGregor>
https://bugs.webkit.org/show_bug.cgi?id=34945
18:36
<zcorpan_>
hmm, so <link rel=next> is being spread on twitter as being 'html5 prefetch'
18:37
<zcorpan_>
http://keyboardy.com/programming/html5-link-prefetching/
18:38
<AryehGregor>
Didn't it way predate -- oh, never mind, buzzword.
18:45
zcorpan_
comments
18:48
<TabAtkins>
Haha, I love <canvas>+<video> *so* much - http://betaweb.csug.rochester.edu/~jparish/public/canvas/video.html
18:49
<miketaylr>
that's awesome
18:50
<TabAtkins>
I think it would be even more awesome if it didn't draw into the canvas, but just used it for calculation, and instead did the display in real text.
18:50
<miketaylr>
dhtml 4 life
18:55
<miketaylr>
somewhat unrelated, but did you see kangax's svg -> canvas parser thingy? http://yura.thinkweb2.com/canvas_demo/
19:00
<TabAtkins>
No, I had not previously seen that!
19:00
<TabAtkins>
Random note - anyone know who the dude in that image is?
19:01
<miketaylr>
the tongue sticking out guy?
19:01
<TabAtkins>
Yeah.
19:01
<miketaylr>
ha. no clue. i can ping kangax on IM and ask though
19:02
<TabAtkins>
kk. I'm just wondering because it looks sorta like jimmy urine.
19:04
<miketaylr>
quoth the kangax, "just some svg i found on the web"
19:04
<TabAtkins>
Kk.
19:19
<AryehGregor>
Awesome, Ubuntu versions are now tracking latest Firefox instead of freezing on a stable branch forever.
19:23
<gratz|home>
hm
19:23
<gratz|home>
AryehGregor, link to info?
19:24
<AryehGregor>
gratz|home, http://fridge.ubuntu.com/node/2051
19:26
<gratz|home>
cool
19:26
<gratz|home>
was a bit surprised since they freeze everything for standardisation / support
19:26
<gratz|home>
good to see though :D
19:28
<gratz|home>
brb
19:43
<TabAtkins>
AryehGregor: You around? Question about Hebrew numbering.
19:43
<AryehGregor>
TabAtkins, I'm here.
19:43
<TabAtkins>
Do you know how accurate the current css3-lists description of hebrew numbering is? It looks like it's pretty easy to describe with a simple greedy additive algorithm.
19:44
<AryehGregor>
I commented about that some time ago, didn't I?
19:44
<TabAtkins>
Yeah, but I can't follow that discussion well enough to be sure.
19:44
<AryehGregor>
The conclusion was that it should just stop at 999,999.
19:44
<TabAtkins>
Okay, that's fine.
19:44
<AryehGregor>
If you do that, then it's very simple.
19:44
<TabAtkins>
And the appears if and only if the number is 1k or greater?
19:45
<TabAtkins>
the geresh
19:45
<AryehGregor>
Well, there are still a couple weird things, actually.
19:45
<TabAtkins>
Well, plus the 15/16 exceptions.
19:45
<AryehGregor>
I'd ditch all the places where it uses a Hebrew word, like 0, 1000, 2000, etc.
19:45
<TabAtkins>
The current algo doesn't use a word anywhere.
19:46
<AryehGregor>
Okay, what URL are you looking at?
19:46
<AryehGregor>
I'm looking at a WD, I think.
19:46
<TabAtkins>
http://dev.w3.org/csswg/css3-lists/#algorithmic
19:46
<TabAtkins>
It's just "choose a value from this list for the 100k digit, 10k digit, 1k digit, add a geresh if necessary, 100 digit, 10 digit, 1 digit".
19:47
<AryehGregor>
This looks fine to me, yes.
19:47
<TabAtkins>
Kk, cool.
19:47
<AryehGregor>
I don't know how standard it is for stuff above 1000. I've seen that convention used, but people also sometimes just continue on without a geresh.
19:47
<AryehGregor>
E.g., תתר for 1000.
19:47
<TabAtkins>
I'm developing a syntax to allow authors to specify list-styles themselves. All the non-algorithmics are trivial, and I'm trying to solve as many algorithmics as I can do simply.
19:48
<AryehGregor>
That gets unwieldy above about 1200, though.
19:48
<TabAtkins>
Most of them fall into either "cjk" or "additive" camps.
19:48
<AryehGregor>
The Hebrew one is quite simple. The only oddity is the 15/16 thing, I think.
19:49
<TabAtkins>
Yeah, but that's easy enough to do with a greedy additive algorithm. Just specify 19/18/17/16/15 as possible tokens.
19:49
<TabAtkins>
Then it'll take those in preference to building 15/16 the "normal" way with a 10 and a 5/6.
19:49
<AryehGregor>
Ah, I see.
19:49
<AryehGregor>
Yeah, that should work fine.
19:50
<TabAtkins>
Good to know.
19:51
<TabAtkins>
Stupid ethiopic and tamil numbering systems being all complicated.
19:52
<zcorpan_>
does css support roman numbers?
19:52
<TabAtkins>
I can handle tamil with an addition to my additive system, but ethiopic is impossible without creating something that might as well be general purpose.
19:52
<TabAtkins>
zcorpan_: CSS3 Lists describes it (under the algorithmic types).
19:53
<TabAtkins>
My method of defining list-styles also supports it.
19:53
<zcorpan_>
nice
19:54
<TabAtkins>
I currently have an example UA stylesheet that defines every non-glyph and non-algorithmic list type. I'll be adding the algorithmics that I can handle soon.
19:55
<TabAtkins>
I wonder where I could find someone who knows the tamil numbering system?
19:55
<TabAtkins>
I want to confirm something before I give up on it.
19:55
<TabAtkins>
Random fun fact - in my syntax, it's pretty easy to describe a "one, two, three" list-style.
19:56
<AryehGregor>
You should support Roman numerals the way the romans actually used them most of the time, with IIII instead of IV and so on. Then it's really trivial.
19:57
<AryehGregor>
(fun fact: watches with Roman numerals typically have IIII instead of IV, but IX instead of VIIII)
19:57
<TabAtkins>
Indeed, both of those systems are trivial.
19:58
<TabAtkins>
The subtractive roman numerals just require a handful more entries in the token list.
19:58
<TabAtkins>
(So that it'll take "XC" for 90 if it can, frex.)
19:58
<TabAtkins>
The only thing I can't do in my system is support overbars, but honestly who cares.
19:59
<TabAtkins>
I'm fine with the range just being 1-39999.
19:59
zcorpan_
didn't know about overbars in roman numbers
20:00
<AryehGregor>
There's no Roman-numeral-with-overbar Unicode character?
20:00
<AryehGregor>
characters.
20:00
<TabAtkins>
I dunno. Maybe?
20:02
<AryehGregor>
I don't see it.
20:02
<AryehGregor>
There are Roman numerals around U+2160 ff., but none with overbars.
20:03
<AryehGregor>
At least not in my font.
20:03
<TabAtkins>
Yeah, though apparently unicode recommends against using them. Apparently they're mainly for cjk usage.
20:03
<TabAtkins>
They recommend just using latin letters normally.
20:03
<AryehGregor>
Right, I remember that.
20:09
<TabAtkins>
Hmm, does it make sense to include the 5k and 10k characters? They're marked as "archaic" roman numerals in the unicode listing.
20:14
<Philip`>
I thought most Roman numerals were pretty archaic
20:15
<TabAtkins>
They prefer the term "classical".
20:34
<AryehGregor>
Only Roman letters are usually used for Roman numerals.
20:34
<AryehGregor>
Up to M for 1000.
20:34
<AryehGregor>
Today, I mean.
20:38
<zcorpan_>
chrome 6.0.422.0 crashes on my websocket testsuite
20:39
<zcorpan_>
hmm, one test not only killed the tab but killed all of chrome
20:41
<zcorpan_>
now it seems to be any test - maybe chrome got fed up and didn't like websockets anymore
20:41
<AryehGregor>
You know Chrome changed the WebSocket implementation incompatibly in 6.0, right?
20:41
AryehGregor
assumes so, but is just checking
20:42
<zcorpan_>
yes, that's why my tests work at all
20:42
<zcorpan_>
i should probably look into making my tests public
20:44
<zcorpan_>
seems it crashes if the server sends a binary frame with length 2^32
21:05
<zcorpan_>
ok out of 158 tests, chrome passes 98 and crashes on 2, and sometimes seemingly crashes hard on any test
21:05
<gsnedders>
zcorpan_: On OS X?
21:05
<zcorpan_>
yes
21:14
<zcorpan_>
seems chrome happily opens the connection without checking the upgrade and connection fields
22:21
<TabAtkins>
Argh, damn you hebrew. I'm just going to skip providing the hebrew list-type in actual characters and only do it with escapes, because bidi mixing is super-frustrating. >_<
22:30
<AryehGregor>
It's super-frustrating even if you can actually read it.
22:31
<AryehGregor>
Especially when it keeps switching direction as you type.
22:36
<TabAtkins>
For real. I was about to start cussing when it was swapping the direction of backspace and delete based on which direction I approached the comma from.
22:38
<AryehGregor>
Yep.