00:03
<Philip`>
jwalden: Like XML's <![CDATA[... stuff? That makes it hard to write an example of XML code that uses CDATA, so it doesn't solve any problems :-p
00:03
<gsnedders>
That's SGML! :P
00:03
<gsnedders>
Perfectly valid (and useless) HTML 4.01!
00:03
<Philip`>
I prefer to believe that SGML doesn't exist
00:04
<Philip`>
My life is much simpler that way
00:04
<jwalden>
being able to nest is a false goal, in my book
00:04
<Philip`>
There's just HTML and XML
00:16
<gsnedders>
Philip`: And gsneddersML.
00:17
<gsnedders>
With elements such as "lust" and "rant".
04:19
<Hixie>
Philip`: validator.nu is market leading in the sense that it's changing the game, too
13:04
<Hixie>
boy am i glad i don't own XHR anymore
13:24
<Lachy>
Hey Hixie, did you remember to do that study on the markup generator strings?
14:33
<Philip`>
hsivonen: It's really quite confusing how Validator.nu sometimes stops half way through a document
14:33
<Philip`>
e.g. http://validator.nu/?doc=http%3A%2F%2Fwww.millermicro.com%2Fnatick.html&showimagereport=yes&showsource=yes#imagereport made me wonder why it was missing a load of images that were on the page
15:13
<annevk>
I don't understand the opting into cookies proposal...
15:14
<annevk>
Does it mean user agents can decide to do cookies and if they do they include some headers?
15:15
<annevk>
oh wait, the server includes that header, not the user agent
15:15
<annevk>
hmm
16:00
<Lachy>
oh dear, I thought this thread was over :-( http://lists.w3.org/Archives/Public/public-html/2008May/0002.html
16:09
<Dashiva>
Lachy: Various slashes movies come to mind
16:09
<Dashiva>
*slasher
16:14
<Lachy>
Dashiva, which movies?
16:17
<Dashiva>
The sequels :) "They thought he was dead, but they thought wrong!"
16:26
<Philip`>
The only way to really kill those characters is to not watch the movies, since they'll stop making sequels when there's no audience
16:29
<Philip`>
But maybe it's more like the end-of-game bosses in Final Fantasy style games, where you have to defeat them in combat seven times and then use a special sequence of attacks that you found on GameFAQs to finally see them off
16:38
Philip`
wishes Gmail interacted correctly with Thunderbird's automatic removal of " (was: ...)" from subject lines, instead of thinking the subject change meant it was a whole new thread
17:39
Philip`
finds yet another Firefox canvas crash
17:39
<Philip`>
They're worryingly common :-(
17:41
<Dashiva>
Maybe they're hydra-based. Every time you find one, two more are created.
17:47
<Philip`>
Hmm, looks like just a null pointer thingy, so not particularly exploitable :-(
17:47
<hsivonen>
Philip`: I need to do something about image report upon fatal error, yes. thanks
17:54
<Philip`>
Oh, it's already reported as https://bugzilla.mozilla.org/show_bug.cgi?id=421715 - I guess nobody cares enough to fix that issue in FF2
17:55
Philip`
adds UA-sniffing to his canvas test suite, to avoid that crash
18:05
<gavin_>
Philip`: we can take that for a dot release
18:05
<gavin_>
I'll ask for approval
18:07
<Philip`>
gavin_: Ah, thanks, that would make me happy :-)
18:10
Dashiva
reads "alt and authoring practices"
18:11
<Dashiva>
That's a nice way to make a gallery that never reveals it does, in fact, contain images
18:31
<Philip`>
Hmph - WebKit doesn't support colours rgb(x,x,x) when x >= 2^31, and Firefox 2 doesn't support them when x >= about 10^39, and Opera 9.2 doesn't support them when x >= 2^31 - 64
18:32
<Philip`>
Silly browsers :-(
18:32
<Philip`>
(FF3 and O9.5 seem to get it right, though)
18:33
Philip`
wonders if he's testing things that don't have great real-world relevance
18:36
<Dashiva>
Well, if you manage to squeeze 10^39 into one byte, that's pretty significant :)
18:39
<Philip`>
Well, I don't expect the browser to make that pixel appear three hundred orders of magnitude brighter than normal bright white, it should just be clamped to 255 :-p
18:39
<Philip`>
Oops, I didn't mean 10^39 for FF2, I meant 10^310
18:39
<gsnedders>
So, both browsers and standards suck then.
18:40
<Philip`>
gsnedders: Monitors too - we wouldn't need this clamping if it wasn't for their physical limitations
18:40
<gsnedders>
Philip`: Then we'd reach the limits of what the human brain can tell apart
18:42
<Philip`>
gsnedders: We're a huge way away from those limits - look at bright white on a computer monitor, then look at the sun with one eye, then look at a class 4 laser with the other eye, and you'll see there's still a long way for monitors to grow
18:42
<gsnedders>
Philip`: Am I meant to look at all three at once, or just the latter two?
18:43
<Philip`>
gsnedders: I recommend doing them separately and in that sequence
18:43
<gsnedders>
Philip`: That's true, it may be a bit hard if you look at the laser first
18:46
<Dashiva>
Philip`: So basically, you're seeing int and double max value limitations?
18:47
<Philip`>
Dashiva: Yes, except for Opera being crazy and doing signed int minus 64
19:00
<Dashiva>
Geez... if we were following the standard set in the early days of htmlwg, Hixie would've filed half a dozen formal complaints by now :)
19:15
<Philip`>
So, Opera's gone from 81% down to 52% on my ImageData tests since I last checked
19:15
hsivonen
wonders if KompoZer's alt UI is on mpt's radar for Ubuntu usability
19:16
<Dashiva>
Philip`: More tests, changed spec, or worse support?
19:16
<Philip`>
while WebKit has gone from 0% to 79%
19:16
<Philip`>
Dashiva: More tests and changed spec
19:17
<Philip`>
(though only a few of the extra tests are not the result of spec changes, so mostly it's just spec changes)
19:19
<Philip`>
I'm being nasty and e.g. testing that there isn't an ImageData constructor, in order to make Opera fail
19:21
<Dashiva>
Is that a requirement, though?
19:23
<Philip`>
There isn't an explicit statement in the spec saying "There must not be an ImageData constructor"
19:25
<Philip`>
I don't know if the Web IDL spec says that interfaces that aren't explicitly declared to have a constructor mustn't have a constructor
19:25
<Philip`>
but in any case it matters for interoperability, since it's not nice if people write new ImageData() in Opera and find their code breaks for Firefox users, so that's more important than what the specs explicitly say
19:28
<Dashiva>
Hoo boy
19:48
<hsivonen>
alt sure generates a lot of email
19:51
<Philip`>
hsivonen: It's used a hundred times more than abbr/acronym, so it ought to be worth a reasonable amount of discussion
19:51
<Philip`>
but I won't attempt to define what "a reasonable amount" is :-)
19:58
<Dashiva>
Well, it's not like much of it is about alt anymore
19:59
<Dashiva>
(and if it is, we've heard it before)