00:00
<Lachy>
virtuelv, I think the spec means a range in which both the upper and lower values are known and exact values
00:00
<virtuelv>
Lachy: and I'm arguing that that is completely wrong
00:01
<virtuelv>
in the Wikipedia article, there is a perfectly valid usage example of the upper bound being a random number exceeding some value
00:01
<Lachy>
how would you possibly implement a range in which, e.g., the upper value is somewhere between 9000 and infinity?
00:02
<Lachy>
which example, specifically?
00:02
<virtuelv>
the tsunami example
00:02
<virtuelv>
it's something entirely representable and acceptable within a meter
00:03
<virtuelv>
disregard the presentation given there
00:03
<Lachy>
I can't see tsunami mentioned on either the VU_meter or Logarithm_scale pages.
00:04
<virtuelv>
the map in «Graphic representation»
00:05
<virtuelv>
an alternative representation here would typically be a bar chart of values
00:05
<virtuelv>
see also the example in http://en.wikipedia.org/wiki/PH
00:05
<virtuelv>
"Applications"
00:07
<Lachy>
in that map, the scale is still physically limited to using a single colour for values of 100,000+. To represent such values with a <meter>, you would have to set a fixed upper limit to what can be represented
00:08
<virtuelv>
Lachy: and the spec claims that meter has a finite upper bound
00:09
<virtuelv>
""The meter element represents a scalar measurement within a known range""
00:09
<virtuelv>
the point that "over 9000" is not within a known range
00:10
<Lachy>
in that specific map example, you would only need to represent 6 values <meter low=1 high=6>#</meter>, where # is the value
00:11
<virtuelv>
Lachy: no
00:11
<virtuelv>
the ph scale is infinite in either direction
00:11
<Lachy>
well, actually, you might do <meter low=1 high=6 value=3>100-999</meter>
00:12
<virtuelv>
Lachy: which, again, is not included in the very first sentence in the very first paragraph of the definition of the <meter>
00:12
<Lachy>
there is no possible way to represent an infinite scale. You have to set a physical upper limit and live with it
00:13
<virtuelv>
(there is a point here about logarithmic scales and values and ranges
00:13
<virtuelv>
trying to argue against this is arguing against a couple of hundred years of science where logarithmic scales have been used
00:14
<Lachy>
I'm not arguing against logarithmic scales. I'm saying that to represent them, something, somewhere, needs to set an upper limit in order to be able to render the scale within a finite size.
00:15
<virtuelv>
no, you *can't*
00:15
<virtuelv>
the point is that +inf is a perfectly valid upper bound as well
00:16
<Hixie>
syp: works fine for me in ff3 on windows
00:17
<syp>
Hixie: ff3 should be fine, it's only 3.1 beta2 that had issues.
00:17
<Lachy>
sure, mathematically, it is. But I'm talking about how to represent that physically where it is absolutely not possible to represent truly infinite scale in a finite space. There are physical limits that have to be dealt with, and that map example is a perfect illustration, since it sets the physical limit at 100,000+
00:17
<Hixie>
syp: oh, a beta, ok
00:17
<Hixie>
syp: sorry, missed the beta part :-)
00:18
<Hixie>
virtuelv: <meter> isn't yet intended for all those cases
00:18
<virtuelv>
Hixie: what is, then?
00:18
<Hixie>
in html5? nothing, just like html4
00:18
<Hixie>
ue svg :-)
00:18
<virtuelv>
right now it seems to mostly cater to google pagerank display and search result relevance
00:18
<Hixie>
use even
00:19
<syp>
Hixie: and it now works with Firefox 3.1 post beta2, so that shouldn't be a big deal.
00:19
<virtuelv>
in the matter of max value and upper bound, sure it's ok to impose a "practical" limit
00:19
<Hixie>
syp: k
00:20
<Hixie>
virtuelv: the main use case now is to discourage people from using <progress>
00:20
<virtuelv>
huh?
00:21
<Hixie>
the main use case for <meter> is to give people who would otherwise abuse <progress> to show static data an alternative
00:22
<Hixie>
so that people don't abuse <progress>
00:23
<virtuelv>
yes, I understand that's what you meant, I'm just puzzled by why you're creating one element with limited use to discourage use of other elements
00:24
<virtuelv>
that's like reinstating <strike> into the spec because otherwise people would abuse <del>
00:28
<virtuelv>
<strike> would've been better named as <justkidding>
00:28
<Hixie>
virtuelv: well for <del> here's CSS
00:28
<Hixie>
for <progress> there isn't really a good alternative right now for non-progress meters
00:28
<Hixie>
i expect <meter> will be extended in html6, like everything else
00:28
<Hixie>
we have to start small
00:29
<Hixie>
holy crap the HTML5 spec doesn't work well in IE8 compat mode
00:29
<Hixie>
about 25% down the spec disappears
00:29
<Hixie>
wtf
00:32
<virtuelv>
Hixie: for then, could you also make a note of having a mechanism for discretization of values for meter (and possibly progress, even)
00:33
<virtuelv>
I mean, you can be 2/3rds done, but you want to represent it as one dot, two dots, three dots
00:33
<Hixie>
?
00:33
<Hixie>
for when?
00:34
<Hixie>
the presentational aspects are separate from html5 really
00:34
<Hixie>
i mean there's a presentation section that gives some constraints
00:34
<Hixie>
but generally it's up to the browser or the author with xbl
00:41
<virtuelv>
Hixie: my point is that, let's say I want to represent a large range of values with <meter> in a table
00:42
<virtuelv>
I want the range 0-100 to be represented by a value of 1 indicator step in your default presentation
00:42
<virtuelv>
the range 100-1000 by another
00:42
<virtuelv>
and 1001-9000 by a third level
00:42
<virtuelv>
and "over 9000" to be represented by a fourth
00:43
<virtuelv>
(Sorry about the 4chan-trolling-Oprah example)
00:44
<Hixie>
then just use a unicode block character or something
00:44
<Hixie>
no need for <meter>
00:45
<Hixie>
yeah the problem is the idn thing
00:46
<Hixie>
er, wrong channel again!
01:37
<Hixie>
huh, this turned out to be a pretty good article
01:37
<Hixie>
http://www.computerworld.com/action/article.do?command=viewArticleBasic&taxonomyName=protocols_and_standards&articleId=9128096&taxonomyId=141&intsrc=kc_feat
01:59
<Lachy>
compared with previous computer world articles I've read, that's certainly one of the better ones
02:19
<Hixie>
so is there a way to use these transition thingies to make an element flash yellow temporarily when it's the :target?
02:19
<Hixie>
or does that require script somehow
02:21
<Lachy>
Hixie, I think that would need a script or CSS Animations http://webkit.org/specs/CSSVisualEffects/CSSAnimation.html
02:22
<Hixie>
ooh, css animations
02:22
<Hixie>
forgot about those
02:22
<Lachy>
I think webkit supports that, though I haven't played with them yet. Opera doesn't yet have support for them, even internally
02:23
<roc_>
I'm not sold on CSS animations
02:24
<roc>
but I think you can do Hixie's thing using CSS transitions
02:24
<Lachy>
how?
02:24
<Lachy>
you would need a script to control it
02:24
<roc>
set up a transition on, say, background-color
02:25
<roc>
add a CSS rule blahblah:target { background-color:yellow; }
02:25
<Lachy>
sure, but that only allows it to transition from one value to another. It won't flash temporarily as he wants.
02:25
<Hixie>
yeah i'm just looking to flash it temporarily
02:25
<roc>
just make the transition slow
02:26
<Hixie>
i want the element to quickly go to yellow, then fade to white
02:26
<Hixie>
when it is :target
02:26
<Hixie>
the rule above would make it fade to yellow and stay yellow
02:26
<roc>
you can control the shape of the transition
02:26
<Hixie>
can you make it go backwards?
02:26
<Lachy>
no, you can't make it go backwards
02:27
<Hixie>
then it doesn't really help
02:27
<Hixie>
unless i'm missing something
02:27
<roc>
I guess you're right
02:28
<Lachy>
maybe you could do it with generated content
02:29
<Hixie>
oh?
02:29
<Lachy>
hmm, thinking...
02:31
<Lachy>
foo::before { -o-transition: background-color 2s; background: yellow; visibility: hidden; position: ...; /* make it sit behind the element*/ } foo:target::before { visibility: visible; background: transparent; }
02:32
<roc>
you could even use display:none instead of visiblity:hidden to make that less expensive
02:33
<Lachy>
roc, sure. I didn't realise there was a performance difference between the two
02:33
<Hixie>
the "..." part is the tough one
02:34
<Lachy>
hence why I did the "..." :-)
02:34
<Hixie>
since for various reasons i don't want to set position:relative on everything else
02:35
<Lachy>
if you want it to work for everything, setting position: relative; would be relatively inexpensive. But doing it for everything certainly would be painful.
02:35
<Lachy>
I meant, if you want to do it for *headings only*
02:37
<roc>
you could hack it with borders
02:37
<Lachy>
try the old fashioned way: :target { background-image: url(animated-yellow-fadout.gif); }
02:38
<Lachy>
how?
02:38
<Lachy>
roc, how would you do it with borders?
02:38
<roc>
foo { transition: border-color 2s; border-style:none; border-color:yellow; } foo:target { border-style:solid; border-color:transparent; }
02:39
<roc>
well I guess that doesn't do exactly what you want
02:39
<Lachy>
but that doesn't have the border behind the element
02:39
<roc>
maybe you want border-style:hidden so the layout doesn't change
02:39
<roc>
or do it with 'outline' instead
02:40
<Lachy>
an outline with an outline-offset might work.
02:40
<roc>
Hixie didn't say it had to be behind the element
02:40
<Lachy>
well, he said he wanted it for the background, which is behind the element.
02:41
<Lachy>
well, at least, he implied it
02:41
<roc>
not to me he didn't :-)
02:41
<Lachy>
how else do you make "the element flash yellow"
02:42
<Hixie>
i found another solution to bz's problem
02:42
<Hixie>
body { margin-bottom: 25% }
02:42
<Lachy>
what was his problem?
02:43
<Hixie>
going to a small bottom section in the multipage version didn't clearly show which section was targetted
02:44
<Lachy>
oh, ok. I was wondering why the spec wasn't long enough as it is :-)
02:44
<Hixie>
hah
02:46
<Lachy>
Hixie, there's a bug in the status box script, which shows the floating status box for one section in subsequent sections if they don't have one of their own
02:47
<Hixie>
that's intentional
02:47
<Hixie>
feel free to add new section boxes if it's going from like an h3 section to an h2 section though
02:47
<Hixie>
(the section boxes are actually just targetting any element child of <body> with an ID, iirc)
02:48
<Lachy>
see, e.g., http://www.whatwg.org/specs/web-apps/current-work/#the-marquee-element-0 and subsequent sections. It didn't make sense that it said Yet to be specified, even though the sections I'd scrolled to had lots of content
02:48
<Hixie>
(i.e. they're not really "section" boxes)
02:48
<Lachy>
ok
02:50
<Hixie>
i added a box
02:50
<Hixie>
ok dinner bbiab
08:32
Hixie
kills <eventsource>
08:32
<Hixie>
poor element
08:32
<Hixie>
it had a rough life
08:32
Hixie
welcomes the new EventSource('...') object!
08:33
<hsivonen>
Is new EventSource() compatible with what Opera shipped? That is, does it create an <event-source> DOM node in Opera? like Image()
08:33
<Hixie>
no, not even remotely
08:33
<Hixie>
but what the spec said yesterday wasn't either
08:34
<hsivonen>
ok. sucks to be the one gaining impl. experience
08:34
<Hixie>
yup
08:34
<Hixie>
we've changed the wire format and its mime type since opera shipped event-sourec
08:34
<Hixie>
event-source
08:34
<Hixie>
and now the api
08:34
<Hixie>
there's basically nothing left except the idea :-)
08:35
<hsivonen>
Hixie: so what's the remaining selling point over Web Sockets? that you can feed EventSource from a regular HTTP server?
08:35
<Hixie>
the only selling point over websocket that it's ever had is that it is compatible with CGI scripts
08:36
<Hixie>
i.e. yes
08:36
<hsivonen>
ok
08:36
<Hixie>
but apparently that's enough to keep people eager to have it
08:42
<annevk2>
I'm sort of glad the API is much simpler now
08:43
<Hixie>
it's actually technically more complicated :-)
08:43
<Hixie>
at least, the idl is bigger
08:43
<Hixie>
but yeah, it's simpler
08:43
<annevk2>
dude
08:43
<Hixie>
yes?
08:43
<annevk2>
it was even more complicated at the start, when you could create all kinds of event objects
08:44
<annevk2>
Hixie, I object to technically more complicated on the basis that the IDL has more lines!
08:44
<Hixie>
:-)
08:44
<Hixie>
it is indeed far simpler now
08:44
<Hixie>
for example you can no longer hook an EventSource to an XMLHttpRequest object
08:44
<Hixie>
(cue "awwwww")
08:44
<hsivonen>
not AWWW? :-)
08:44
<Hixie>
but i did add a few more features while i was at it
08:44
<Hixie>
hsivonen: hah
08:47
<gsnedders>
awwwww
08:47
gsnedders
misses that feature
08:49
<annevk2>
Hixie, the URL attribute should probably return the absolute URL
08:51
<Hixie>
remind me in ten minutes
09:17
<hsivonen>
Hixie: http://lists.whatwg.org/pipermail/whatwg-whatwg.org/ is 403
09:17
<Hixie>
yikes
09:19
<Hixie>
http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/
09:23
<annevk2>
I have to say, it really sucks that the archives keep moving around
09:24
<annevk2>
maybe we should ask the W3C for some mail server space o_O
09:24
<Hixie>
heh
09:26
<Hixie>
did i miss some e-mails on the spellcheck issue? i have some still in my whatwg folder from before my last e-mail on the subject
09:26
<Hixie>
how odd
09:29
<hsivonen>
Hixie: thanks
09:30
<hsivonen>
I gather the URI structure instability is due to Dreamhost managing the whole thing?
09:30
<Hixie>
yeah
09:30
<Hixie>
the utter disaster of an archive system is blamed on the same reason
09:30
<annevk2>
Hixie, do you remember the reason #123 does not work in CSS?
09:33
<hsivonen>
http://stackoverflow.com/questions/582352/how-can-i-ignore-dtd-validation-but-keep-the-doctype-when-writing-an-xml-file
09:34
<Hixie>
annevk2: because css1 says it doesn't? dunno
09:35
<hsivonen>
Is this yet another case of an arbitrary decision that indents can't start with digits?
09:36
<hsivonen>
Is that some kind of old programming language thing that parsers should be able to decide if they are parsing an ident or a number literal from the first character?
09:36
<annevk2>
well, that has a reason, to distinguish them from numbers + unit
09:36
<annevk2>
Hixie, I thought there was some other reason, ta
09:37
<hsivonen>
oh, right CSS has units. not arbitrary then. sorry
09:38
<hsivonen>
I'm just knee-jerk annoyed by stuff like XML Name
09:38
<annevk2>
no shit
09:38
<annevk2>
XML5 baby
09:42
<Hixie>
annevk2: my guess is originally hakon and bert were trying to match html
09:43
<Hixie>
annevk2: hakon might know
09:43
<annevk2>
kk
09:46
<annevk2>
crap, web-apps-tracker fails with a 500
09:46
<annevk2>
did DreamHost change anything?
09:51
Hixie
looks in his filter folder
09:51
<Hixie>
holy crap
09:51
<olliej>
Hixie: that sky has fallen
09:51
<Hixie>
so that's where all the public-html activity has gone
09:51
<olliej>
s/that/the
09:51
<Hixie>
olliej: which one?
09:51
<olliej>
Hixie: can you test html5.org ?
09:52
<Hixie>
test?
09:52
Hixie
confoosed
09:52
<olliej>
it's gone for me
09:52
<olliej>
server no workie
09:52
<olliej>
i'm trying to find out whether the server is down, or comcast hates freedom
09:52
<Hixie>
yeah doesn't work for me either, but i'm on comcast
09:52
<Hixie>
anne runs that server though
09:53
<olliej>
bad annevk2
09:53
<annevk2>
you joined just after I complained about DreamHost
09:54
<olliej>
ah :D
09:54
<annevk2>
because I didn't do shit
09:54
<olliej>
annevk2: nerget.com is doing fine :D
09:54
<Hixie>
so's whatwg.org :-)
09:55
<didymos>
http://downforeveryoneorjustme.com/html5.org heh
09:55
<didymos>
Hixie, but whatwg.org works fine
09:56
<annevk2>
so everything went down?
09:56
<annevk2>
geez
09:56
<annevk2>
initially it was just the tracker
09:56
<Hixie>
everything on your account, i guess
09:57
<annevk2>
yeah
09:58
<gsnedders>
And gsnedders.com isn't on dreamhost :P
09:59
<jgraham>
gsnedders: What is it on?
10:00
<gsnedders>
jgraham: mediatemple, for now
10:00
<Hixie>
probably the same stuff gsnedders himself is on
10:00
<gsnedders>
Touché.
10:00
<Hixie>
:-D
10:00
<gsnedders>
Is it bad that I realized I haven't taken any (medical) drugs today because of that?
10:01
<Hixie>
as someone who rarely if ever takes medical drugs of any kind, i am not in a position to answer that question
10:02
Philip`
can't connect to hobgoblin at all
10:02
<gsnedders>
annevk2: What's a spec you edit that uses biblio?
10:03
<gsnedders>
(or anyone else edits, but you probably know what you edit better)
10:03
<Philip`>
annevk2: I hope you'll refund a substantial portion of what I'm paying you for hosting philip.html5.org
10:04
<jgraham>
gsnedders: It depends if it reminded you to take some regular medical drugs or if it was just a passing observation that you haven't taken any irregular drugs tody
10:04
<gsnedders>
jgraham: Regular medical drugs
10:04
<annevk2>
gsnedders, I don't use biblio, I do stuff by hand
10:04
<gsnedders>
Who does? P
10:04
<gsnedders>
* :P
10:04
<annevk2>
gsnedders, actually, css3-mediaqueries might use biblio
10:05
<annevk2>
Philip`, EUR -1000 coming right at ye
10:05
<jgraham>
gsnedders: Oh well in that case it is probably good that it reminded you but bad that you had forgotten :)
10:05
<Philip`>
annevk2: Excellent!
10:05
gsnedders
remembers seeing something that claims annevk2 edits using it
10:05
<gsnedders>
annevk2: css-namespace does
10:06
<annevk2>
well, that's two specs then
10:06
<gsnedders>
But that was mainly fantasai judging from the commit log
10:06
<gsnedders>
jgraham: Also, it's reminded the reason why I haven't: I need food to have some of the pills, and I haven't had breakfast yet.
10:07
jgraham
sympathises
10:08
jgraham
suffers for incredible medication related paranoia as it is hard to conclusively prove to yourself that you took the right number of tablets at the right time
10:08
gsnedders
has probably taken numerous overdoses of paracetamol because of that, but has seemingly lived
10:11
<jgraham>
I guess paracetamol is not so bad
10:11
gsnedders
is also nowadays reguarly on Diclofenac
10:14
<gsnedders>
(without taking either I can scarcely move for pain)
10:14
jgraham
looks that up, decides that gsnedders is unlikely to be suffering from menstrual pain
10:14
<gsnedders>
:P
10:14
<gsnedders>
Despite most of my friends at school concluding I am a girl, that is indeed correct.
10:31
<Philip`>
"Building Services has just installed anti-bacterial soap dispensers in each of the kitchens to help to reduce the risk of the vomiting viral infection."
10:32
<Philip`>
Isn't anti-bacterial soap a bit unsuitable for stopping viruses?
10:33
<Dashiva>
Not if it's anti-your-own-bacteria and leads to your hands falling off?
10:34
<Philip`>
Good point
10:34
<jgraham>
Maybe the soap smells really bad so it causes people to stay further away from each other reducing the risk of virus transmission?
10:41
Dashiva
wonders why rubys is plotting to overthrow the W3C when they've already given him all the power he wants
10:41
<jgraham>
What makes you think he's not gone power crazy?
10:42
<Dashiva>
But if he overthrows them, he will lose the power he has because the legitimacy disappears
10:43
<Philip`>
The power to overthrow an organisation is greater than the power you can attain by operating within that organisation
10:43
<jgraham>
And in nay case power-crazed evil geniuses are not noted for their logical reasoning skills
10:44
<Dashiva>
Oh, so now he's a genius too?
10:45
<jgraham>
For example givne the choice beteen killing batman using a) a simple gun and b) placing him in the bottom of an eggtimer and slowly letting the sand run through, hich would you pick?
10:46
<Dashiva>
He's got bulletproof armor, so b) would at least have a non-zero chance of success
10:46
<Philip`>
Does he have bulletproof armour on his face?
10:46
<jgraham>
Can the bulletproof armour be removed once you have captured him to the extent that getting him into a giant eggtimer is a possibility?
10:47
<Philip`>
The giant eggtimer seems easier than capturing him and undressing him
10:48
<Dashiva>
Now you're assuming capture
10:48
<Philip`>
e.g. you could drop it from a helicopter
10:48
<Dashiva>
Or put a hostage inside, so he had to go in voluntarily
10:48
<Hixie>
Philip`: i hate to interrupt this fascinating if strange conversation, but do you have a uri to that page with summary values?
10:49
<jgraham>
You could do both of those things with a simple cage, remove any armour, and then finally use a traditional, reliable method of killing
10:50
<Philip`>
Hixie: http://canvex.lazyilluminati.com/misc/summary.html ? (But it's not formatted in a way that makes it easy to tell if most uses are bad, and it's not from anything even attempting to approach a sensible sample of pages)
10:50
<Dashiva>
Then you'd have to go inside the cage
10:51
<jgraham>
Dashiva: Expendable henchmen
10:51
<Hixie>
Philip`: thanks
10:51
<Philip`>
(But at least it does demonstrate of some pages which would get worse if the summary was rendered in graphical UAs)
10:51
<Philip`>
s/of//
10:51
<Dashiva>
jgraham: I think you should provide some justification for why the egg timer isn't good enough
10:52
<Philip`>
s/s\/of\/\///
10:52
<Philip`>
s/of/the existence of/
10:54
<jgraham>
Dashiva: It p[rovides an unacceptable risk. Egg timers are unproven technology in the killing market. Moreover it introduces a long delay between initiation of death-process and actual death. This provides an ample oppertunity to cleverly escape. Clever escapes are known to be a strong point of batman-like superheros
10:55
<Dashiva>
We didn't become supervillains by following the rules and playing safe
10:55
<Philip`>
On the other hand, slow killing processes mean you can film it and sell the distribution rights for a lot of money
10:56
<roc>
I would like to make a series of very short films. Each one would be based on the opening sequence of an action movie, but showing what happen in real life.
10:57
<roc>
For example: hero runs from enemies, hero ducks behind cover, hero breaks cover, hero gets shot in the head
10:58
<Philip`>
These so-called "entertainment" shows are part of a plot by an evil genius to set the public's expectations in such situations, and when he captures the real superhero he will shock the world by demonstrating that his absurd plans actually work in practice when nobody believed they ever would
10:58
<roc>
Hero notices bomb in room, hero leaps out of window, hero breaks neck
10:59
<jgraham>
roc: If they were shot-by-shot remakes of well knowwn action movies I would watch that for sure
10:59
<jgraham>
Oh and assuming they were sufficiently funny
10:59
<jgraham>
But I don't see how they could fail to be
11:00
<Dashiva>
You'd need a good sound track
11:01
<Philip`>
Why do movies often have the hero run across a busy road and all the cars honk and swerve, and the hero never gets run over?
11:02
<Dashiva>
Because otherwise the hero would have to be a slow runner to let the enemy escape
11:03
<Hixie>
it would be pretty funny to remake action movies from the start, stopping at the first point where the hero dies or is incapacitated
11:03
<Hixie>
many wouldn't last 2 seconds
11:03
<Hixie>
from what i hear the latest indiana jones is like that -- movie starts, character gets nuked. story ends.
11:03
<Hixie>
along with its franchise.
11:04
<roc>
actually he would have died several times before reaching the nuke bit
11:04
<Philip`>
You could still tell the story in flashbacks, so it wouldn't necessarily be a bad movie
11:04
<Hixie>
roc: ah. haven't seen it. good to know.
11:06
<Philip`>
(like that Monkey Island game where you start hanging on a rope over the edge of a giant hole, and someone comes along and you spend most of the game explaining how you ended up in that predicament, and then the rope snaps)
11:07
<Dashiva>
Then you spend 10 years arguing about whether what happens after you fall down is part of canon or not
11:08
zcorpan
adds table:before { content:attr(summary); display:table-caption } to his user style sheet
11:13
Philip`
likes Hot Fuzz as kind of an action film parody, though it doesn't bother too much with realism
11:13
<didymos>
Philip`, I'd say it decidedly doesn't bother with realism :)
11:13
<roc>
LAST ACTION HERO actually attempted this
11:13
<roc>
I appear to be the only person in the world who likes that movie
11:13
<Dashiva>
I liked the part where the villain shoots a guy and is surprised there are no cops around after five minutes
11:13
<jgraham>
"If you wanna be a big cop in a small town fuck off to the model village"
11:13
jgraham
liked Hot Fuzz too
11:21
<Philip`>
I don't remember model villages being a feature of any other movie I've seen
11:21
<Philip`>
particularly not for the climactic fight scene
11:41
<jgraham>
Right but Stevef seemed to be implying that it was a bad dataset. But mere non-randomness is not a useful criterion
11:43
<Philip`>
Datasets are only good or bad to the extent that they support conclusions you might draw from them
11:43
<jgraham>
Depends what you mean "support" and "want"
11:44
<Philip`>
and we don't have any datasets that can support precise conclusions about "average sites" or "most sites" etc (since we don't even know what those terms mean)
11:44
<Philip`>
so we just have to avoid concluding such things
11:44
<Philip`>
but it's easy to support conclusions like "there exists at least this number of sites with this property"
11:45
<Philip`>
jgraham: How does it depend on what I mean by "want", when I never used that word? :-)
11:46
<jgraham>
Ah, I misread "might" as "want to" somehow ;)
11:48
<Hixie>
zcorpan: the heuristic for layout tables that i thought of the other day is "table has no borders"
11:49
<jgraham>
Hixie: Have you seen the Mozilla code for this?
11:49
<Hixie>
code for what?
11:49
<jgraham>
layout-tables-heuristics
11:49
<Hixie>
there is code for it?
11:49
<Hixie>
what does it do?
11:50
<Philip`>
http://mxr.mozilla.org/mozilla-central/source/accessible/src/html/nsHTMLTableAccessible.cpp#1029
11:50
<jgraham>
IIRC it thinks that anything with <th> is a data table, anything with 1 column is a layout table, things with nested tables are (probably?) layout tables, things wwith no borders are probably layout tables and things with > 5 columns are data tables
11:50
<jgraham>
It's kinda complex
11:51
<jgraham>
And I would be against specifying anything to distinguish layout tables and data tables at this stage
11:51
<Hixie>
yes me too
11:52
<Hixie>
i'd be interested in seeing what the comparison was of that algorithm vs just checking for no borders anywhere in the table
11:54
<hsivonen>
Hixie: what about tables like the doctype table in http://hsivonen.iki.fi/new-design/ ?
11:54
<hsivonen>
zcorpan: speaking of the doctype table, updates regarding various opera versions are on my mental todo list. sorry about the delay.
11:55
<Lachy>
In that whole <canvas> accessibility thread on public-html, were there any actual solid proposals for how to address the problem? So far I've only seen arguments about what the problem is and I'm just over half way through the thread.
11:55
<hsivonen>
Lachy: perhaps I should post one, but I've been trying to avoid the thread
11:56
<jgraham>
Lachy: I think there is some vauge suggestion of providing a DOM-accessability api interface
11:56
<hsivonen>
jgraham: don't we have one: a tree of divs with ARIA stuff on them
11:56
<Philip`>
Lachy: If I remember correctly, the only specific proposals were to expose the drawText strings to AT somehow, and to add a pushAnnotation() method that does something to indicate to AT what's being drawn
11:56
<jgraham>
hsivonen: Doesn't meet the bespin usecase afaict
11:57
<hsivonen>
jgraham: the reason against that I can see is that the DOM may be slower than a novel tree
11:57
<Philip`>
but those were pretty vague
11:57
<Philip`>
and I pointed out various problems that make extracting drawText strings very difficult in typical canvas uses
11:57
<hsivonen>
jgraham: for perf or for other reason?
11:57
<jgraham>
For performace with large files
11:58
<Lachy>
ah, yeah, I remember those vague suggestions of DOM Accessibility APIs
11:58
<Lachy>
but I think I will ignore the rest of that thread for now and move onto something more productive.
12:00
<jgraham>
(AFAICT the constraint on bespin is "can't build a DOM tree for all text in the buffer", so maybe it would be possible to do better by dynamically building a DOM just for the visible text or something)
12:00
<roc>
Hixie: I've seen a fair few data tables that just used backgrounds for styling
12:01
hsivonen
guesses annevankesteren.nl and html5.org are on the same failing Dreamhost VM
12:02
<hsivonen>
jgraham: does the concept of not having an accessible tree for the whole document work for VoiceOver and Orca?
12:02
<Hixie>
hsivonen: funky
12:02
<jgraham>
hsivonen: No idea
12:02
<Hixie>
roc: oh well
12:02
<hsivonen>
Hixie: is "funky" good or bad here?
12:02
<Hixie>
neither :-)
12:03
<hsivonen>
ok.
12:03
<Hixie>
unconventional and striking
12:04
<Philip`>
jgraham: It's not clear that they can't build a DOM tree; the problem may be more to do with layout/rendering of the DOM tree
12:05
<jgraham>
Philip`: That's true, I guess.
12:07
<roc>
it seems like it should be possible to do what they want with the DOM, with enough engine work
12:07
<roc>
but there are closely related problems that aren't really possible
12:07
<roc>
like "virtual" tree rows
12:08
<roc>
where you have some underlying data model and you want to render complex content for items in the model
12:09
<roc>
creating the complex content for all items in the model is arbitrarily expensive
12:10
<roc>
so you end up creating it on the fly in clumsy, error-prone ways
12:10
<roc>
or building some kind of custom renderer
12:11
<roc>
I'm not aware of anyone having developed a really great solution to this problem
12:12
<Hixie>
http://www.nytimes.com/library/national/science/health/011800hth-behavior-incompetents.html
12:12
<Hixie>
well that explains a lot
12:13
<roc>
there are lots of "tree control" widgets that provide a limited set of renderings for arbitrary models with great performance
12:13
<roc>
and there are template solutions that provide flexible rendering with crappy performance
12:14
<hsivonen>
Hixie: the paper is http://www.apa.org/journals/features/psp7761121.pdf
12:14
<hsivonen>
Hixie: I thought it had already been mentioned here
12:15
<Hixie>
possible, i don't always read everything y'all say :-)
12:15
<roc>
seems like if we just tried a bit harder we could get some kind of template system with great performance
12:15
jgraham
also remembers hearing about that before
12:17
<Hixie>
ok well i guess is should go to bed
12:17
<Hixie>
nn
12:17
<roc>
me too
12:17
<Hixie>
you're on the other side of the planet!
12:17
<Hixie>
we should NOT be in the same circadion cycle
12:18
<Hixie>
circadian even
12:18
<jgraham>
Hixie: I think you're the one with the weird sleep patterns :)
12:18
<roc>
oh, you're in Europe?
12:18
<Hixie>
jgraham: can't argue that one
12:18
<Hixie>
roc: nah, california. we're not really that far away.
12:18
<roc>
yeah
12:18
<roc>
just over the water
12:18
<Hixie>
but you're on the other side of typical maps
12:18
<Hixie>
so as far as i'm concerned that's the Other Side of the Planet!
12:19
<Hixie>
anyway
12:19
<Hixie>
nn
12:21
<zcorpan>
ok i've now looked at about 100 pages from http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/pages/tagattr/table/summary and not found a single data table with summary (let alone a useful summary)
12:39
<Lachy>
Hixie, in the latest Firefox 3.1 beta, the script that shows the floating status box seems to be causing an infinite loop. I haven't tried the latest trunk yet.
12:50
<Philip`>
zcorpan: If http://philip.html5.org/data/table-summary-gov.txt wasn't 404 then it would probably be a list of pages slightly more likely to have decent summaries
12:52
<Philip`>
http://google.com/search?q=cache:http://philip.html5.org/data/table-summary-gov.txt
13:02
<jgraham>
It's not quite clear why p(summary is useful|summary & table is data table) should be higher for deep content pages
13:02
<jgraham>
Since front pages tend to get more attention from designers, consultants etc.
13:02
<jgraham>
But maybe that is not what you meant
13:02
<Philip`>
Designers, consultants etc are likely to suggest not putting complex data tables on your front page
13:02
<jgraham>
Right, I agree that there may be fewer (complex) tables on those pages. I'm not sure you can infer anything about the likely quality of any summary though
13:02
<Philip`>
I was intending to refer to p(page contains a useful summary | page contains a summary), and I'm assuming p(summary is useful | data table) > p(summary is useful | layout table), and assuming p(data table) is lower on front pages
13:02
<Philip`>
so looking at n front pages with a summary is likely to result in a higher proportion of layout tables, and hence a lower proportion of useful summaries, than looking at n deep content pages
13:02
<jgraham>
Philip`: That seems reasonable
13:11
Philip`
tries producing a less rubbish list of summary values
13:11
<Philip`>
s/less/marginally less/
13:13
<jgraham>
Philip`: How?
13:13
<hsivonen>
whoa! a lot of summary email while I was looking away from the email app
13:14
<Philip`>
jgraham: By not basing it on an ancient list of pages scraped from Yahoo search results
13:14
<Philip`>
and by counting domains rather than pages
13:18
<jgraham>
Where are you getting the data from?
13:19
<Philip`>
From the internets
13:20
<jgraham>
Ah, good start. Waiting for extraterrestrial beings to communicate with earth in HTML format would have been a worse strategy, for example.
13:26
Philip`
wonders if jgraham was intending a more specific question
13:28
<jgraham>
I suppose I might have meant "how are you selecting pages on the internet to use when looking for data" or something
13:29
<zcorpan>
Philip`: it seems most of the top few that i looked at of that list have summary="Table for formatting purposes"
13:29
<Philip`>
jgraham: Ah - just the same dmoz.org list I've always been using
13:30
<jgraham>
Philip`: Does dmoz.org not have roghly the same front-page bias?
13:30
<Philip`>
jgraham: It has some front-page bias, but I don't know if it's more or less
13:31
<Philip`>
(I would guess somewhat less than the other list, which mostly came from Yahoo search results for queries like "a" and "the")
13:33
<jgraham>
Would a search for a term like "table below" work any better?
13:35
<zcorpan>
from the gov list, i found 2 or 3 pages that had useful summary and the rest were either repeating a heading or were "for layout"
13:36
<jgraham>
I guess that could introduce subtle biases since it would mean that the table was being discussed in the body of the text
13:46
<Philip`>
http://canvex.lazyilluminati.com/misc/summary-20090226.html - it's a bit boring really
13:46
<Philip`>
Oh, html5.org is back, I should have uploaded it there instead
13:48
hsivonen
wonders why cocoa apps go crazy when another cocoa app being debugged dies
13:48
Philip`
is getting quite good at writing pages that turn out to be valid when he first runs a validator on them
13:49
<hsivonen>
garbage collectors are wonderful things for productivity
13:50
hsivonen
just finished a long hunt for a premature memory release in C++
13:50
<Philip`>
Except when you're trying to optimise your program's memory usage and have to reverse-engineer and fight against the GC, I guess
13:50
<Philip`>
Valgrind makes memory allocation errors much less painful :-)
13:51
Philip`
thinks the latest version even tells you what line of code freed a piece of memory that you're using after it was freed
13:51
<hsivonen>
woohoo! my innerHTML setter no longer crashes
13:54
<hsivonen>
Did Hixie just appeal to the expertise of markp on public-html?
13:58
<jgraham>
hsivonen: Yeah, that seemed a bit weak.
14:17
<zcorpan>
http://livedom.validator.nu/?%3C!DOCTYPE%20html%3E%0D%0A%3Csvg%3E%3Cdesc%3E%3Cmath%3E%3Cb%3E
14:20
<zcorpan>
is the <b> escaping too far?
14:23
<zcorpan>
Hixie: ^
14:30
<hsivonen>
interesting case. my impl. is per spec, though
15:29
<hsivonen>
Hixie: I observe that not dropping foster parented stuff on the floor when parent gone adds a third node operand to the tree operations that an HTML5 tree builder can generate
15:30
<hsivonen>
Hixie: that is, nsHtml5TreeOperation is bloated by one nsCOMPtr
15:44
<rubys>
The q element is "being considered for removal"?
15:48
gsnedders
needs some way of properly sorting people's names in Python
15:48
<jgraham>
rubys: I think that is wrong. Although I do remember arguning that <q> is mostly useless. Or was that some other element...
15:48
<jgraham>
gsnedders: Define "properly"
15:49
<gsnedders>
jgraham: Deals with people like Anne
15:50
<jgraham>
gsnedders: That's not a definition
15:50
<hsivonen>
gsnedders: please let me know if you find you how Danish names are sorted by surname
15:50
gsnedders
headdesks
15:50
<hsivonen>
gsnedders: you could be really radical and sort by the whole name like Hixie does
15:50
<hsivonen>
s/find you/find out/
15:51
<rubys>
Python even has builtin support for sorting by whole names
15:51
<didymos>
hsivonen, maybe I can be of assistance regarding the Danish names -- what are you uncertain about?
15:52
<hsivonen>
didymos: if you have an RFC author like Henrik Frystyk Nielsen, do you sort by N or F if your document is in English and sorted by surname?
15:53
<jgraham>
gsnedders: Wikipedia claims "Capturing this rule in a computer collation algorithm is difficult, and simple attempts will necessarily fail"
15:53
<gsnedders>
rubys: Where?
15:53
jgraham
guesses rubys means sort()
15:53
<didymos>
hsivonen, I'd sort it by the F -- but to get it exactly right, you'd have to know whether 'Frystyk' is a middle name or surname
15:53
<gsnedders>
Oh, for whole names
15:53
gsnedders
is being dumb
15:54
<didymos>
hsivonen, but in this case I'm whiling to take a guess and says it's part of his surname
15:54
<didymos>
wiling*
15:54
<hsivonen>
didymos: but Danes tend to have names that are between the first name and the last name but don't have the same middle name semantic that e.g. English, Finnish and Swedish middle names have, AFAIK
15:54
Philip`
assumes the need to "take a guess" is what makes it hard to capture the rules in a computer collation algorithm :-)
15:56
<didymos>
Philip`, certainly, but you can often hear on a name, whether it's considered a legal middle- or surname
15:56
<hsivonen>
didymos: i.e. a middle name that is more surnameish that second given nameish
15:56
<didymos>
but I certainly see your point
15:56
<didymos>
hsivonen, exactly
15:56
<didymos>
and that's why I'm willing to guess -- Frystyk is not a Danish given name
15:58
<hsivonen>
didymos: should be fun for gsnedders to implement in software :-)
15:58
<didymos>
hsivonen, yup -- I guess you would have to have the list
15:58
<Philip`>
gsnedders: Why implement it in software, rather than getting a human to specify the ordering?
15:58
<didymos>
and even then, you wouldn't be 100%
15:59
<gsnedders>
didymos: Humans are lazy
15:59
<gsnedders>
Philip`, rather
15:59
<didymos>
gsnedders, :)
15:59
<didymos>
oh well, I'm off -- good luck gsnedders and hsivonen
15:59
<gsnedders>
It's only me
16:00
<hsivonen>
I've dodged the issue by not trying to sort or transpose people's names in bibliography
16:02
hsivonen
wonders if hCard supports surnameish but optional middle names these days
16:02
<hsivonen>
IIRC, it didn't a couple of years ago
16:05
<gsnedders>
hsivonen: What do you mean?
16:05
<Philip`>
Hmm, is hobgoblin down again? It was fine a little while ago, but now it doesn't respond
16:06
<Philip`>
Anyway: http://fonts.philip.html5.org/ now has more fonts and fewer bugs, since I finally uploaded the most recent code
16:06
<gsnedders>
hsivonen: It supports them fine when you explicitly mark each first/middle/sur- name
16:07
<hsivonen>
gsnedders: my point is that e.g. Danish surnameish middle names are semantically different from English-oriented "middle names", and this semantic distinction is not captured
16:07
<hsivonen>
gsnedders: use case: sorting
16:08
<gsnedders>
It's captured as much as it is in vCard :P
16:08
<jgraham>
gsnedders: Just make authors mark which name-component they expect it to be sorted by
16:08
<jgraham>
That satisfys WWLD
16:08
<jgraham>
(or rather, in this case, WWBD)
16:08
<gsnedders>
B?
16:08
<jgraham>
Bibtex
16:08
<Philip`>
L?
16:08
<gsnedders>
Ahj
16:08
<Philip`>
Oh
16:08
<gsnedders>
*Ah
16:09
<hsivonen>
jgraham: seems like a failure for semantics :-)
16:54
<Philip`>
http://blogs.msdn.com/jensenh/archive/2006/04/07/570798.aspx - apparently people really like making text bold
16:55
<gsnedders>
Philip`: with <strong>?
16:56
<jgraham>
hsivonen: Heh Save is number 2. That says something about how much users truct Word...
16:56
<Philip`>
gsnedders: No, with the button that says "B" on it
16:56
<Philip`>
jgraham: hsivonen?
16:56
<gsnedders>
jgraham: I do that with everything though
16:56
<jgraham>
Er, I was going to say something to hsivonen earlier and it was still on my input line
16:57
<jgraham>
gsnedders: Get lightroom
16:57
<jgraham>
No save
16:57
gsnedders
needs to do that
16:57
<jgraham>
also: 100% pure awesomeness
16:57
<gsnedders>
The issue is getting it at edu cost is such a pain and a lot of effort
16:58
<jgraham>
gsnedders: It was pretty easy for me. Just had to scn+email my student ID
16:58
<gsnedders>
Actually, that isn't the issue.
16:58
<gsnedders>
The fact is I'm lazy
16:58
<gsnedders>
jgraham: I don't remember it being so easy when I got PS
16:59
<jgraham>
In fact I think there might be a eb form
16:59
<gsnedders>
eb?
16:59
jgraham
has to go now
16:59
<jgraham>
web
16:59
jgraham
is standing up and typing
16:59
<gsnedders>
adios
17:26
<rubys>
hixie: ping?
18:05
<rubys>
For hixie when he gets back: your name has been added to http://esw.w3.org/topic/IETF_HTML5_Meeting_March_2009 ; if you disagree, please either let me know or simply remove it
18:14
<annevk2>
from Opera Yngve will likely be there since he's going to the IETF stuff anyway
18:15
<annevk2>
I might want to join some IRC discussion channel if feasible to comment on CORS, but I don't expect it to matter much
18:17
<gsnedders>
Oh, I was going to email Yngve
18:17
gsnedders
is reminded
19:54
gsnedders
stabs lxml's data structure
19:56
<gsnedders>
DOM would be nice for once! :'(
22:45
<fantasai>
Is there a fragment identifier syntax defined for bitmap images?
22:45
<fantasai>
I haven't found one, but I figured someone here would be more likely to know for sure.
22:50
<roc>
what do you mean? like <a href="foo.png#abc">?
22:51
<fantasai>
roc: yeah
22:52
<fantasai>
roc: although probably for PNG you'd want something like <a href="foo.png#slice(40,50,20,100)">
22:52
<roc>
I've never heard of such a thing
22:52
<nessy>
the W3C media fragments working group is looking at specifying something like this - mostly for video and audio, but it would be applicable to images that support fragment naming
22:52
<roc>
I see
22:52
<roc>
I guess that would be nice for image tiling
22:52
<fantasai>
image sprites
22:53
<roc>
sprites I meant
22:53
<fantasai>
yeah
22:53
<roc>
although personally I'd rather just promulgate jar:
22:53
<nessy>
http://www.w3.org/2008/WebVideo/Fragments/wiki/Main_Page
22:53
<nessy>
http://www.w3.org/2008/WebVideo/Fragments/wiki/Syntax
22:54
<karlcow>
fantasai: nope. Video activity is working on something for video.
22:54
<karlcow>
for png none
22:54
<karlcow>
for webcgm I think yes
22:55
<nessy>
the proposed sprite syntax is xywh = [unit " : "] int " , " int " , " int " , " int
22:55
<fantasai>
xywh?
22:55
<nessy>
for example foo.png#xywh=40,50,20,100
22:55
<karlcow>
2.3.6 Hyperlinling http://www.w3.org/TR/REC-WebCGM/REC-02-CGM-Concepts.html#webcgm_2_3
22:55
<fantasai>
oh
22:55
<karlcow>
png is not a web format unfortunately
22:55
<nessy>
xy defines the bottom left point
22:55
<nessy>
wh width and height
22:55
<fantasai>
right, right
22:56
<fantasai>
bottom left?
22:56
<fantasai>
why not the top left?
22:56
<nessy>
of the rectable to slice out
22:56
<fantasai>
yeah
22:56
<Hixie>
nessy: should i do anything with your whatwg account?
22:56
fantasai
thought image maps started at the top left
22:56
<nessy>
oh, sorry, I think you're right, it might be the top left
22:56
fantasai
hasn't looked at them in a long time, though
22:56
<nessy>
Hixie: yes, please - I just sent you an email about that :)
22:57
<Hixie>
ah, cool
22:57
<Hixie>
haven't caught up with e-mail yet
22:57
<nessy>
no worries - no rush
22:57
<nessy>
I have two threads to reply to :)
22:57
<karlcow>
for video http://www.w3.org/2008/Talks/26-video-plh/#%2814%29
22:59
<karlcow>
http://www.w3.org/2008/WebVideo/Fragments/
22:59
<fantasai>
karlcow, nessy: So there's two different formats proposed for rectangles?
22:59
<fantasai>
xywh= and rect()
23:01
<john_fallows>
Hixie: just read the spec updates for SSE, thanks for adding the onerror attribute.
23:01
<Hixie>
pleasure
23:01
<Hixie>
i figured i'd do it at the same time as all the other changes
23:01
<john_fallows>
one point of clarification, is there now a change of behavior for the error scenarion?
23:02
<nessy>
fantasai: where did you see rect() ?
23:02
<john_fallows>
for example, before it seemed to imply that network or server error would silently abort future attempts to request the event stream
23:02
<karlcow>
fantasai: I don't know what they have chosen for video yet and if it has been definitely chosen.
23:02
<nessy>
I am part of the media fragments wg
23:02
<fantasai>
nessy: in the link karlcow sent
23:02
<john_fallows>
whereas now it seems to continuously retry, giving an opportunity in the error event listener to disconnect
23:03
<nessy>
we are preparing a WD right now, which should be ready for comments in a few weeks
23:03
<fantasai>
nessy: cool
23:03
<fantasai>
nessy: looking forward to it
23:03
<nessy>
fantasai: that WebCG document is rather old and it seems not being used anywhere
23:03
<Hixie>
john_fallows: that wasn't my intent; what makes it say it continually retries?
23:03
<nessy>
correct me if I'm wrong
23:03
<karlcow>
fantasai: nope the talk was to introduce the video activity
23:03
<fantasai>
nessy: the csswg would probably be interested
23:03
<karlcow>
nessy: you are right, there is webcgm 2
23:04
<fantasai>
nessy: because designers would like to place all their little graphics pieces in one file
23:04
<nessy>
hehe
23:04
<karlcow>
ah interesting.
23:04
<nessy>
well, there are many groups that will want to comment
23:04
<karlcow>
I wish there was links from png too embedded into the format
23:04
<karlcow>
s/was/were/
23:04
<nessy>
that's why there will be a WD with input request
23:05
<nessy>
links from png ?
23:05
<john_fallows>
actually, i see it now, there is a clear distinction between "reset the connection" (with retry) and "fail the connection" (no retry), right?
23:05
<karlcow>
so from html -> to specific part of png.
23:05
<karlcow>
but also from png to somewhere in the web
23:05
<Hixie>
john_fallows: i've tried to make it even more crystal clear -- reload and tell me if you still think it retries :-)
23:05
<karlcow>
so png would really be a web format
23:05
<roc>
karlcow: imagemaps and SVG should really cover that
23:05
<nessy>
the URI scheme allows any document to point to a part of a png, including html
23:05
<fantasai>
nessy: also, the print industry would like the ability to slice images like that in XHTML
23:05
<nessy>
for outgoing links you have to change the file format :)
23:05
<karlcow>
roc: SVG is indeed a Web format
23:06
<fantasai>
nessy: for PNG and JPG
23:06
<fantasai>
nessy: so if you can specify xywh for any image/* type that'd be cool
23:06
<Hixie>
john_fallows: (change only visible on the single page copy for now)
23:06
<nessy>
image maps really are the only way
23:06
<karlcow>
nessy: yes
23:06
<roc>
we don't want to bloat a relatively simple format like PNG
23:06
<roc>
that way lies MNG
23:07
<john_fallows>
Hixie: looks good, thanks.
23:07
roc
glances around nervously
23:07
<fantasai>
heh
23:07
<nessy>
the challenge is: you can create a URL to a png or jpg or mpeg or ogg or anything that has a xywh and/or time offset it in, but you cannot guarantee that that URL will be resolved by the server or client
23:07
<Hixie>
john_fallows: no problem
23:08
<john_fallows>
Hixie: btw, in the spirit of whatwg naming conventions, perhaps call EventSource something like WebStream? :-)
23:08
<nessy>
the actual splicing will still need to be done by some software
23:08
<fantasai>
nessy: yeah, I guess that's the disadvantage there
23:08
<Hixie>
john_fallows: i was hoping for a better name... not sure WebStream is it, we have other plans for the concept of "streams"
23:08
<fantasai>
nessy: we can't use CSS's typical "fails to parse" fallback mechanism
23:08
<nessy>
all you can do is provide a scheme - then the conformant implementations can come over time
23:09
<john_fallows>
Hixie: really, what do you have in mind there?
23:09
<nessy>
fantasai: I think we're planning a scheme for making sure there is feedback on whether the URL resolved or not
23:09
<Hixie>
john_fallows: streams of video from one server plugged into a <video>, streams from a local camera plugged into an XHR to a server, or into a WebSocket to a server, that kind of thing
23:09
<nessy>
fantasai: so the fallback would be to return the full resource
23:10
<fantasai>
nessy: yeah, that wouldn't work well for CSS, since we'd want to trigger CSS's parse-time fallback system
23:10
<fantasai>
nessy: but we can think about that some more on our end :)
23:11
<fantasai>
nessy: I'm glad to hear someone's working on this
23:12
<john_fallows>
Hixie: i see, btw i didn't notice any maximum size on a WebSocket frame. When the frame is large, it might be desirable to obtain the contents as a stream instead of a single fully allocated buffer.
23:23
<Hixie>
john_fallows: indeed
23:50
<Hixie>
rubys1: i tried reading the minutes, but saw nothing for me to do; let me know if i missed anything
23:50
<Hixie>
the minutes weren't exactly clear
23:55
<Hixie>
oops, wrong channel
23:55
<Hixie>
oh, sam isn't in the right channel
23:55
<Hixie>
oh well
23:57
<Hixie>
nessy: done
23:58
<gsnedders>
karlcow: Can I ask you for help with SPARQL?
23:59
<gsnedders>
Also, where would be a better place to ask generally (like, async.)?