00:00
<othermaciej>
Mozilla also has some nonstandard extensions right in the mouse event - caretRangeFromPoint() lets you do the hit testing at any time though, you don't need an event
00:04
<TabAtkins>
Thanks, othermaciej. I've passed the info along.
00:12
<othermaciej>
so I hear Web Fonts is going to define not one but two new font formats
00:12
othermaciej
is not sure what to think of this
00:12
<Rik`>
WOFF and ?
00:13
<TabAtkins>
CWT.
00:13
<TabAtkins>
Which is a fancy name for version 1 EOT.
00:13
<TabAtkins>
With some of the header cruft ignored.
00:13
<othermaciej>
and then it will also recommend regular OTF and SVG fonts, and require browsers to implement at least 2
00:13
<othermaciej>
call me crazy, but isn't this even worse for interop than the status quo?
00:14
<TabAtkins>
I don't see how it can possibly be worse.
00:14
<TabAtkins>
But it's likely that everyone will end up supporting WOFF.
00:14
<Rik`>
othermaciej: my understanding is that the goal is not interop but font makers pleasure
00:14
<othermaciej>
I'm trying to figure out if I should ask our AC Rep to oppose this or just close my eyes and think of england
00:16
<TabAtkins>
What would you propose if you opposed it?
00:17
<othermaciej>
not having a Fonts working group at all
00:17
<TabAtkins>
And then do what about font interop? Hope it all works out?
00:19
<othermaciej>
wait for one side or the other to capitulate
00:20
<othermaciej>
another alternative would be to pursue only one new font format and get everybody on board with implementing it, instead of the weird "pick 2 of 4" thing
00:20
<TabAtkins>
Sylvain appears to be hinting that WOFF is his expected consensus solution.
00:21
<TabAtkins>
(Though I'm still gunning for CWT.)
00:21
<othermaciej>
that being said, WOFF seems DRM-free
00:21
<TabAtkins>
None of the 4 formats are DRMed in any way, shape, or form.
00:21
<othermaciej>
it just seems like implementing it is a giant waste of effort
00:22
<TabAtkins>
Yeah, but MS hasn't given up and implemented TTF yet, and hasn't given any indications that they'd like to, either.
00:23
<Rik`>
TabAtkins: MS hasn't implemented addEventListener but no one has invented yet another way to add handlers because of that
00:24
<TabAtkins>
Yeah, but has MS consistently openly refused to implement it, and had strong statements from people in the organization that they don't expect to?
00:25
<Rik`>
that's not the point
00:25
<TabAtkins>
It is. There's a difference between "hasn't done it yet" and "has refused to do it".
00:25
<Rik`>
developers know how to use different code paths for stupid browsers
00:27
<Rik`>
if MS doesn't want to play ball and developers have a cross browser solution, why would you invent something else ?
00:27
<TabAtkins>
But JS differences can be papered over trivially in libraries so you never have to think about it again. This isn't the same thing.
00:27
<TabAtkins>
...we *don't* have a cross-browser solution. We have two separate solutions that we have to weld together manually every single time we want to use them.
00:28
<Rik`>
they are ttf2eot tools, it's a one line fix
00:28
<TabAtkins>
And you have to store both and link both every time.
00:29
<Rik`>
"store" a < 1mo file ?
00:30
<Rik`>
people already have a lot of CSS path for IE
00:30
<TabAtkins>
It's still a hassle. Imagine if you had to duplicate every image you used on your site and link both of them.
00:30
<Rik`>
isn't that what people do for PNG alpha transparency ?
00:30
<TabAtkins>
No. They either link a non-PNG, or they use one of the pngFix libraries, or they just ignore ie6.
00:31
<Rik`>
and fwiw, people can just treat IE as one of the non supporting browsers
00:32
<TabAtkins>
Yeah, not actually possible when the marketing department wants a special headline font.
00:33
<Rik`>
or you can create a JS library that reads your font-face rules and create a new rule for IE
00:34
<TabAtkins>
That then ties to you js (which is okay if you're doing purely visual tweaks, but not if you're linking in a font for your uncommon language), and still requires two font files to be created (not much of a hassle by itself, but still more than should be necessary).
00:35
<Rik`>
what about the hassle of waiting three years before seeing a version of IE supporting those new formats
00:36
<TabAtkins>
That's why I support CWT.
00:36
<TabAtkins>
Then IE6+ will all work, and I only have to wait for the other browsers to update, which is much faster.
00:36
<Rik`>
and the user base ?
00:36
<TabAtkins>
The userbase of non-IE browsers updates *much* faster.
00:37
<TabAtkins>
I imagine mostly because the browsers are actually willing to auto-update.
00:37
<Rik`>
still need around one year after release
00:37
<TabAtkins>
Yup, that's not so bad. It is the fastest path to interop among all the options.
00:37
<TabAtkins>
I mean, even if IE promised TTF support in IE9, it'd still be 5+ years before it's usable.
00:38
<Rik`>
or use ttf + eot right now
00:38
<Rik`>
(or in one year for a solid userbase)
00:38
<TabAtkins>
Yeah, that's still a crappy option though.
00:40
<AryehGregor>
How does CWT handle origin control?
00:41
<AryehGregor>
I recall that being a sticking point for an EOT-based solution before I stopped following www-font.
00:41
<TabAtkins>
AryehGregor: The same way TTF does - through same-origin controls.
00:41
<Rik`>
anyway, off to bed
00:42
<TabAtkins>
Later, Rik`.
00:42
<AryehGregor>
If the rootstring is nonempty, is it ignored, or do non-IE browsers refuse to process the file?
00:43
<TabAtkins>
CWT has no rootstring - it's current based on eot version 1.
00:43
<AryehGregor>
So IE <= 8 (i.e., large majority of users' browsers) won't implement origin restrictions at all, then?
00:43
<TabAtkins>
Even when it was based on version 2, cwt itself had no rootstring, it just had meaningless padding. Whether or not legacy nonconforming clients interpreted that padding as a rootstring was up to them.
00:43
<TabAtkins>
(Which I preferred, but people bitched about it.)
00:43
<AryehGregor>
Are font foundries actually willing to license anything in this format?
00:43
<TabAtkins>
And yes.
00:44
<TabAtkins>
Anyone who's licensing in TTF would obviously license in CWT.
00:44
<AryehGregor>
Right, but the point is a lot of foundries don't want to license in TTF, I thought?
00:44
<AryehGregor>
Or have a lot of them cracked by now?
00:45
<TabAtkins>
There's still plenty that don't.
00:45
<TabAtkins>
;_;
00:46
<TabAtkins>
But that's why I thought the whole uproar over basing CWT on EOTv2 was stupid. It would let us make an end-run around that crap without bringing rootstring baggage into future browsers.
00:46
<AryehGregor>
Has any foundry said they'll license EOT but not WOFF, or is everyone as okay with WOFF as they're going to get?
00:46
<TabAtkins>
I dunno. I know Ascender's cool with CWT, but I dunno if anyone's outright said that they're cool with WOFF.
00:47
TabAtkins
is heading to dinner now.
00:47
<AryehGregor>
Oh well. We'll see.
01:28
<roc>
TabAtkins: CWT really isn't appealing as currently specified, see my email. You would have to use Referer checking. Easier to just produce and serve two (or more) font files.
01:28
<roc>
othermaciej: WOFF is trivial. At worst, implementing it is a *tiny* waste of effort
01:29
<othermaciej>
the font guys at Apple were not at all enthusiastic about supporting a new font format
01:29
<othermaciej>
if WOFF can easily and efficiently be converted to OTF or TTF then I guess it's somewhat less of a waste of time
01:31
<roc>
it really is the thinnest possible wrapper around OTF+TTF, plus per-table zlib compression
01:31
<othermaciej>
I'm wary that the "plus" is what will make it nontrivial to implement efficiently (at least given only a blackbox OTF/TTF implementation)
01:32
<roc>
you have to decompress the tables
01:32
<roc>
but you already have to mangle tables
01:32
<othermaciej>
I do?
01:32
<roc>
to work around Windows bugs etc
01:32
<roc>
yep
01:32
<roc>
you do
01:32
<roc>
and the zlib decompression is easier than that
01:34
<roc>
OTF/TTF implementations are more of a black box to us than to you, remember
01:34
<roc>
at least until we ship Harfbuzz
01:34
<othermaciej>
I don't think we do any mangling of tables
01:35
<othermaciej>
(that's either due to my ignorance, or due to the fact that we use Apple's font implementation on Windows)
01:35
<othermaciej>
still seems like a waste of time to me
01:35
<roc>
I'm pretty sure you do
01:36
<othermaciej>
pursuing both WOFF and CWT (but then effectively not mandating either) seems even more wasteful
01:36
<roc>
yeah, we're not interested in that
01:38
<othermaciej>
if everyone else implements WOFF then we'll probably have to, but I can't say the prospect fills me with any enthusiasm
01:39
<roc>
you implemented SVG fonts, which are considerably more useless IMHO
01:39
<othermaciej>
incidentally - am I wrong to be concerned about the idea of a Microsoft-edited Canvas spec? Can't tell if I'm being too paranoid or not paranoid enough
01:39
<roc>
I thought that was weird too
01:39
<jcranmer>
roc: I think Hixie disagrees... it's in Acid3, after all
01:40
<othermaciej>
SVG fonts are kind of useless for real content, but at least they can achieve special effects that other font formats can't do
01:40
<roc>
othermaciej: yeah, but you didn't implement that part
01:40
<roc>
jcranmer: I think Hixie may regret putting in Acid3, you should ask him
01:40
<othermaciej>
what do you mean? colored glyphs work fine afaik
01:41
<roc>
not in my tests
01:41
<othermaciej>
as does crazy stuff like per-character rotation and positioning on a path (though the latter I think you can do with SVG text and non-SVG fonts)
01:41
<roc>
yes, you can
01:42
<jcranmer>
hmm, why are the animation tests failing on FF 3.6?
01:42
<roc>
othermaciej: see http://lists.w3.org/Archives/Public/www-svg/2009Sep/0013.html
01:42
<othermaciej>
text on a path should really be doable without using SVG at all IMO
01:42
<roc>
maybe my testcase is broken but no-one contradicted me
01:43
<othermaciej>
our implementation doesn't rasterize the glyphs, it just draws them anew every time, so I'd be surprised if it had that limitation
01:43
<roc>
what limitation?
01:43
<roc>
SVG glyphs really have two parts
01:44
<roc>
a path in the "d" attribute of the <glyph> element
01:44
<roc>
and arbitrary SVG content in the subtree rooted at <glyph>
01:44
<roc>
as far as I could tell, Opera and Webkit support the former but not the latter
01:44
<roc>
again, maybe I'm wrong
01:44
<othermaciej>
I see
01:45
<roc>
if you read that thread, I have some arguments that the way arbitrary-content glyphs are handled is actually quite broken
01:46
<roc>
oh, in a slightly later thread: http://lists.w3.org/Archives/Public/www-svg/2009Sep/0043.html
01:47
<othermaciej>
I'm sure you're right - <foreignObject> in a glyph could be quite crazy
01:47
<othermaciej>
(though not that much worse than <use>)
01:48
<roc>
somewhat worse, since you have to support associating the anonymous content with HTML text, you can't just confine the madness to SVG
01:50
<roc>
the quandry we have over SVG fonts is whether to implement the subset you guys did --- basically to just pass Acid3, since it offers nothing that is not more conveniently expressed with normal downloadable fonts --- or to implement the whole thing, which actually might be useful but has serious implementation difficulties that no-one else has seen fit to tackle yet
01:50
<roc>
and actually kinda sucks for authors anyway
02:00
<othermaciej>
the original reason we have SVG fonts at all is due to enthusiasm of some of the contributors who wrote our original SVG implementation
02:01
<othermaciej>
(they added fonts later, but it was the same guys basically)
02:01
<othermaciej>
although we did fix a bunch of bugs with metrics and other arcane details for Acid3
02:02
<othermaciej>
I think having SVG font and SVG animation tests in Acid3 was kind of lame; at the time only Opera had these things implemented and I think it was an Opera developer who submitted those particular tests
02:02
<roc>
that is true
02:04
<roc>
Acid3 marketing is likely to just force us to implement at least that subset of SVG fonts
02:05
<roc>
but if the arbitrary-SVG feature doesn't get implemented, or does get implemented according to the current spec and sucks as much as I think it will, then Acid3 will be responsible for a stupid wart in the Web platform
02:08
<roc>
oh well, few participants are completely guiltless on that score :-)
02:18
<othermaciej>
indeed
02:18
<othermaciej>
there's no shortage of warts
02:23
<roc>
they still irritate me
02:23
<othermaciej>
I try to focus my irritation on proposals to add new warts
03:08
<TabAtkins>
roc, othermaciej: If I recall correctly, CWT is similarly a super-thin wrapper over TTF. Isn't it nothing more than a header of a couple of bytes? It doesn't even employ any manipulation of the font-data (EOT can do compression on fonts, but the CWT proposal currently specifies uncompressed fonts, to avoid patent issues).
03:09
<othermaciej>
doesn't EOT have XOR "encryption"?
03:09
<TabAtkins>
I don't believe that the EOT version used for CWT does.
03:09
<TabAtkins>
(It's been a few months since the font discussion, though.)
03:10
TabAtkins
is pulling up the archives real quick to see if he can answer definitively.
03:14
<TabAtkins>
othermaciej: I just confirmed - CWT uses *no* scrambling, XOR, MTX, or otherwise.
03:14
<othermaciej>
does it alter the font file less than WOFF?
03:16
<TabAtkins>
*All* that it does is prepend the font with some header bytes. That's literally it. Chop off the header and you've got a perfectly valid ttf font.
03:17
<TabAtkins>
So, yes.
03:17
<TabAtkins>
(Since WOFF does the per-table compression.)
03:17
<othermaciej>
I guess that is an advantage - the downside being that people who only test in IE may accidentally serve EOT fonts using more than the CWT subset
03:19
<TabAtkins>
True, but the common options used when generating EOT will make it immediately fail in a normal CWT-supporting browser. So you'd really have to test *only* in IE.
03:20
<TabAtkins>
Here's Daggett's 1.1 proposal. I think it's the latest 'official' specced version: http://lists.w3.org/Archives/Public/www-font/2009JulSep/0969.html
03:21
<othermaciej>
"the common options used when generating EOT will make it immediately fail in a normal CWT-supporting browser" -- that's not a feature!
03:21
<TabAtkins>
Sure it is. It keeps you from accidentally serving EOT. ^_^
03:22
<othermaciej>
it would be much better if the common options used when generating EOT made it Just Work(tm) in a normal CWT-supporting browser
03:22
<othermaciej>
though I'm sure the tradeoff there is implementation complexity
03:23
<TabAtkins>
But the common option is one of the compression options.
03:23
<TabAtkins>
Yeah.
09:41
gsnedders
is trying to fix the only test failure in html5lib
10:05
<MikeSmith>
hsivonen, Hixie: I just checked in a v.nu change to fix the <caption> error
10:05
<MikeSmith>
so the spec now validates once again
10:05
<MikeSmith>
http://qa-dev.w3.org:8888/?doc=http://www.whatwg.org/specs/web-apps/current-work/
10:06
<Hixie>
i bet the complete.html file still doesn't validate :-)
10:06
<MikeSmith>
ah
10:06
<MikeSmith>
OK, I'll check that next
10:06
<MikeSmith>
Hixie: that was due to the microdata thing, right?
10:06
<Hixie>
yeah
10:06
<MikeSmith>
OK
10:06
<MikeSmith>
I'll look at that next
10:08
<MikeSmith>
hsivonen: it seems that maxFileSize needs to be increased once again
10:08
<MikeSmith>
HTML5 is now 4304 KB
10:09
MikeSmith
hopes complete.html is not more than 5MB
10:34
<Philip`>
MikeSmith: Content-Length: 4831259
10:35
<Philip`>
But I bet it'll grow :-p
10:35
<MikeSmith>
yeah
10:35
<MikeSmith>
I upped it to 5120 on qa-dev for now
10:36
<MikeSmith>
Hixie: OK, patched it enough for now to get complete.html to validate -
10:36
<MikeSmith>
http://qa-dev.w3.org:8888/?doc=http://www.whatwg.org/specs/web-apps/current-work/complete.html
10:36
<Hixie>
i'm gonna have to start writing test cases for validators :-)
10:36
<MikeSmith>
heh
10:36
<MikeSmith>
<snort>
10:38
<MikeSmith>
with that, I gots to drop off and catch a train back to civilization.. back later
10:58
<mikekelly>
I hope Roy doesn't bite my head off :(
11:00
<mikekelly>
we RESTafarians have a vengeful god
11:02
<Dashiva>
Don't worry, Roy bites everyone's head off
11:03
<Philip`>
How is this "other applicable specifications" thing meant to work when applying multiple specs that conflict?
11:03
<Philip`>
and you'd get a different interpretation depending on which order you apply them in
11:04
<Dashiva>
Philip`: If they conflict they aren't all applicable?
11:06
Philip`
tries in vain to think of a concrete example
11:09
<Philip`>
I don't mean things that conflict incompatibly, just things that define non-commutative changes to parts of HTML5
11:10
<mikekelly>
Is that point the Smylers guy made true - that html implements the features of related specs/protocols that it deems appropriate?
11:10
<Philip`>
No
11:10
<mikekelly>
so interop is an objective?
11:11
<Dashiva>
Philip`: Could you make up an unrealistic example?
11:11
<mikekelly>
I really think that http conneg needs to be accepted how its defined right now (as I'm reading it) or re-written to suit the purposes of everyon else's interpretation
11:12
<Hixie>
Philip`: you have to pick the order in which they apply
11:12
<mikekelly>
because everyone else's interpretation (regardless of whether it's "practical or not") makes assumptions which are not specified in the spec
11:12
<Philip`>
Dashiva: Maybe I'm thinking of something like if HTML5 didn't include SVG itself, and a separate spec said "parse everything as normal except when you get an xmlns:xlink attribute do this magic thing etc", and another spec says "parse everything as normal except when you get an xmlns:* attribute do Namespaces-like processing"
11:13
<Dashiva>
Right
11:13
<Dashiva>
That's the motivation for internalizing everything requiring parser/tokenizer changes
11:13
<Dashiva>
Maybe that should be clarified
11:14
<Philip`>
and so an HTML5+SVG+NS implementation will give different output to an HTML5+NS+SVG implementation
11:14
<Hixie>
always a risk
11:14
<Hixie>
but that's the case regardless of what html5 says
11:14
<Hixie>
since other specs can always say they override it
11:14
<Hixie>
and you can always chose to honour or not honour spces
11:14
<Hixie>
specs
11:16
<mikekelly>
you can assuming you are comfortable with the social costs of doing that
11:17
<Philip`>
The (hypothetical) issue is that you choose to perfectly honour precisely the same set of specs as somebody else, yet get a different output because you're interpreting them in a different order
11:17
<mikekelly>
of course that is not of massive concern if you live with your head up your anus
11:17
<Philip`>
so the order is just as important as the choice of honouring or not
11:18
<Hixie>
Philip`: yes
11:19
<Philip`>
mikekelly: Rewriting specs to match reality sounds like a good idea - try asking the HTTP people to do that, if you think they're wrong
11:19
<Dashiva>
HTTP being wrong? Heresy
11:21
<Philip`>
mikekelly: Oh, when I said "No" to you earlier I had misread your question
11:21
<Philip`>
I thought you said "Is that the point ..." and it wasn't that point, but that wasn't what you said
11:21
<Philip`>
so ignore my response
11:25
<mikekelly>
ok..
11:25
<mikekelly>
what of http isn't written to match reality?
11:26
<mikekelly>
i'm not challenging you I'm asking for your opinion
11:26
<Philip`>
mikekelly: Presumably the bits where you say everybody (except you) disagrees with it
11:27
<mikekelly>
right - so the conneg definitions which clearly state they are tied to a *request* and yet everyone insist they are somehow tied to client default preferences
11:27
<mikekelly>
I agree.
11:28
<mikekelly>
I think you're all missing an opportunity for efficiency gains
11:28
<mikekelly>
but that's fine - it should be cleared up in the http spec if everyone insists on being wrong :P
11:29
<mikekelly>
the world of caching shall forever be flat.
11:30
<mikekelly>
aside from that, I don't think http gets much wrong
11:30
<mikekelly>
what else did you have in mind?
11:31
<mikekelly>
(bearing in mind that HTML's deficiencies will play a large role in what does or doesn't work "in practice")
11:31
<mikekelly>
presumably why PUT/DELETE are being added
11:32
<mikekelly>
even though "in practice" PUT/DELETE "don't work"
11:32
<mikekelly>
because they "aren't used"
11:32
<gsnedders>
mikekelly: See AryehGregor's email for an example where HTTP cannot help caching, due to the complexity of a lot of websites
11:32
<gsnedders>
mikekelly: e.g., on my blog, posting a new page via PUT would only invalid one page's cache, not the entire website's as would be needed
11:32
<mikekelly>
gsnedders: I deliberately didn't repsond to that because the response would be long winded and not a lot to do with html
11:32
<mikekelly>
it's to do with how you identify resources
11:32
<mikekelly>
and how granular you make them
11:33
<mikekelly>
and/or how much you avoid composite resources
11:33
<mikekelly>
which derive state from other resources
11:33
<mikekelly>
that is a design decision and your application can address those issues depending on the requirements
11:33
<mikekelly>
there's also absolutely no requirement to implement that kind of caching mechanisms across the board
11:33
<mikekelly>
it can be applied where and when it is most valuable
11:35
<mikekelly>
feel free to contest that btw..
11:36
<mikekelly>
gsnedders: if you submit a new post with POST to your 'root' blog resource
11:36
<mikekelly>
POST /Blog would indicate the blog resource shouldbe invalidated
11:37
<mikekelly>
a composite resource at blog would emit the blog post information within the blog resource
11:37
<mikekelly>
the alternative is to simply provie the links to the posts
11:37
<mikekelly>
and use javascript to fetch each post and render it onto the page
11:38
<mikekelly>
which prevents the blog page becoming stale if you PUT a new version of a particular post
11:39
<mikekelly>
obviously if you create a new blog post using PUT to a new URI - then the blog page would not be updated
11:39
<mikekelly>
that makes sense..
11:39
<mikekelly>
which is why it would make more sense to use POST
11:39
annevk
is with othermaciej on the fonts crap
11:40
<Philip`>
(Becuse a dozen cached requests to generate a blog's front page with a dozen posts is going to be much more efficient than a single uncached request?)
11:40
<Philip`>
s//a/
11:40
<mikekelly>
depends on the complexity behind the URI
11:40
<mikekelly>
that is just an example
11:40
<mikekelly>
it's a trade-off
11:40
<mikekelly>
I think I've said that about 12 times now :)
11:42
<mikekelly>
disagree?
11:43
<mikekelly>
ok.. then..
11:43
<mikekelly>
:)
11:45
Philip`
is too busy eating biscuits to disagree
11:45
<mikekelly>
heh, this is a good example of why we are going round in circles
11:46
<mikekelly>
gsnedders: which parts of that do you disagree with/are skeptical about?
11:47
<annevk>
gsnedders, where are you?
11:47
<annevk>
gsnedders, and where do we need to be? I'm somewhere close to the bar, near the beachfront
11:48
<gsnedders>
annevk: Between the restaurant and the room we were in earlier
11:48
<gsnedders>
annevk: I'm sitting with jgraham hacking on html5lib
11:48
<annevk>
good stuff
11:49
<annevk>
I haven't gotten further than reading email
11:49
annevk
was planning on editing
11:51
<annevk>
with regards to the range API discussion earlier
11:51
<annevk>
that is planned
11:51
<annevk>
but note that hit testing is not actually defined yet anywhere
11:52
<annevk>
so technically everything is up in the air
11:52
<annevk>
but if a UA implements a sensible definition of hit testing it will prolly work
11:55
<gsnedders>
annevk: Is it cold there?
11:55
jgraham
would prefer nto be near the beach
11:55
<annevk>
not really
11:55
<jgraham>
Or on the beach if the wifi would extend that far
11:55
<annevk>
dunno about that
11:55
<gsnedders>
That would be cold.
11:56
<gsnedders>
This is Sweden, after all.
11:56
<annevk>
yeah, you need more than a t-shirt
11:56
<jgraham>
I could get a hat
11:56
<annevk>
it'd be a start
11:56
<jgraham>
gloves might make typing hard though
11:56
<annevk>
though sand in your keyboard sucks
11:56
<gsnedders>
I have a hoodie t-shirt on, so anything on this looks kinda stupid
11:57
<gsnedders>
annevk: How'd you get there? Just through the big hall?
12:04
<annevk>
I ate lunch here
12:04
<annevk>
there's a staircase just before you enter the dining area
12:11
<gsnedders>
annevk: Oh, that place, I thought by "near the beachfront" you meant outside. But there again, here, near the beachfront can quite easily be inside :)
12:12
<annevk>
indeed
14:44
<gsnedders>
html5lib now passes 100% on a stock Python setup!
14:44
<jgraham>
gsnedders: Now install bs
14:44
<gsnedders>
jgraham: That's bullshit
14:44
<jgraham>
Sigh
14:52
TabAtkins
giggles.
15:27
<TabAtkins>
Would anyone have any idea/way of checking if aria roles are already commonly used in HTML on the web?
15:30
<Philip`>
TabAtkins: I wouldn't
15:30
<Philip`>
(I can only look at stuff from six months ago)
15:30
<TabAtkins>
k.
15:30
<TabAtkins>
Well hey, was aria common the web six months ago?
15:36
<Philip`>
TabAtkins: http://philip.html5.org/data/role-attributes-raw.txt is the raw data from 425K dotbot pages
15:36
<Philip`>
(including non-text/html pages)
15:42
<Philip`>
Does the validator.nu parser have the scripting flag enabled or disabled by default?
15:45
<TabAtkins>
Philip`: Thanks, that's very useful! It appears that aria was almost completely unused.
15:52
<Philip`>
TabAtkins: Almost completely used in static markup in pages which were in the sample
15:52
<TabAtkins>
Yus. Good enough for me.
15:52
<Philip`>
The "static" is probably the main issue - it's presumably used a lot more in scripted widget libraries, not static content
15:53
<TabAtkins>
Philip`: But I don't see a single aria-related use of role which is grossly changing the semantics of the element it is applied to, which is what I was looking for.
15:54
<TabAtkins>
I expect that widget libraries tend to rely on <div> and <span>, where all roles are valid, and thus they aren't relevant to the discussion with Leif.
15:54
<Philip`>
TabAtkins: http://google.com/codesearch?q=role%3D+lang%3Ahtml
15:54
<Philip`>
That might be more useful
15:55
<Philip`>
http://google.com/codesearch?q=role%3D+lang%3Ajs too
15:56
<TabAtkins>
That is useful, thank ou!
15:56
<Philip`>
Dojo uses waiRole="..."? I thought it was meant to be role="..."...
16:03
<Philip`>
Why aren't computers infinitely fast :-(
16:44
<TabAtkins>
Philip`: Because if they were we'd be living in a simulation with probability 100%.
16:49
<mikekelly>
hello browser people
16:51
<gsnedders>
jgraham: Can I just make html5lib raise a warning with bs + ns?
16:51
<mikekelly>
is there a chance we could ever have functionality where javascript can cause new page to load with custom headers?
16:55
<Philip`>
zcorpan: Ping
16:56
gsnedders
notes he hasn't seen zcorpan with a computer all day
16:56
<Philip`>
He can read logs :-)
16:56
<gsnedders>
I am aware. I'm just letting you know you won't get a quick response
16:56
<Philip`>
Or you could ping him manually
16:56
<Philip`>
if you're near him
16:57
<gsnedders>
Philip`: I don't think he's in this room
16:57
<Philip`>
or even point him at http://philip.html5.org/data/cdata-containing-self-close.txt
16:57
<Philip`>
which hopefully isn't too incompatible with what he wanted
16:58
Philip`
is happy to wait until whenever he gets back here
17:09
<gsnedders>
Philip`: pong from zcorpan
17:09
<jgraham>
gsnedders: It already does, doesn't it?
17:09
<gsnedders>
Philip`: thanks from zcorpan
17:19
<gsnedders>
Hixie: WAKE UP!
17:27
<Philip`>
gsnedders: If you use figlet you can shout louder on IRC, which might wake him up
17:28
<RIK|WORK>
PS : IT IS CAPS LOCK DAY
17:31
<Philip`>
RIK|WORK: I have caps lock on all the time but I hold down shift too so that I irritate people less
17:52
<Philip`>
The spec comment box ought to be bigger
23:00
<TabAtkins>
roc: Which is easier for you - (1) defaulting the first/last stops if necessary, then correcting misordered stops, then defaulting any remaining stops, or (2) correcting misordered stops, then defaulting all remaining stops, with first being min(0,first specified stop) and last being max(100%,last specified stop)?
23:00
<roc>
the former, since that's what I implemented
23:01
<TabAtkins>
kk.
23:01
<roc>
it's efficient, I do it in O(N) time in the number of stops, and it's reasonably simple
23:03
<sicking>
Hixie: does @rel no longer allow URI tokens?
23:04
<sicking>
Hixie: without registring them in the wiki that is
23:04
<Hixie>
it's never allowed tokens that aren't registered as far as i recall
23:04
<Hixie>
registering them is pretty trivial though
23:09
<TabAtkins>
roc: Changes made, and are visible in the draft on my site. Thanks for the "specified position" terminology - I was struggling with how to say that. Also rearranged/reworded linear-gradient()'s treatment of the two <angle> cases to allow the behavior you've implemented.
23:09
<roc>
yay, thanks
23:12
<TabAtkins>
No problem; thanks for implementing it. ^_^ I'm speccing the repeating-* functions now.
23:14
<roc>
BTW, Firefox builds with the implementation here: https://build.mozilla.org/tryserver-builds/zweinberg⊙mc
23:15
<TabAtkins>
There any instructions on installing ff builds? Never done one before.
23:23
<roc>
depends on the platform
23:23
<roc>
but these builds install just like any other Firefox release
23:24
<roc>
if you're on Windows, the easiest thing to do is to just download the ZIP, unzip it somewhere, and run firefox.exe
23:24
<roc>
it's that simple
23:25
<Philip`>
and make sure you're not already running a Firefox process
23:25
<roc>
you may want to run "firefox.exe -P" and create a new profile for testing, that lets you run concurrently with an already-running Firefox
23:26
<Philip`>
Don't you need to add -noremote or something?
23:26
<roc>
yeah, you probably do
23:27
<Philip`>
So it's not quite that simple ;-)
23:27
<roc>
running with a different profile can also be good because there is also a small chance that running a trunk build with your existing profile will cause problems when you go back to using your release version with that profile, although that has never happened to me, so the chances are very small indeed
23:28
<TabAtkins>
Since I do IRC using chatzilla, being able to run it concurrently seems useful.
23:28
<roc>
Philip`: perhaps Tab runs IE as his main browser? :-)
23:29
<jwalden>
TabAtkins: http://chatzilla.rdmsoft.com/xulrunner/
23:29
Philip`
shudders
23:29
<TabAtkins>
No, I run chatzilla in FF, gmail in chrome, then everything else in whichever one is handy when I ctrl+t.
23:31
<TabAtkins>
jwalden: Thanks!
23:33
<jwalden>
chatzilla as its own app is so much money it's not even funny
23:34
<jwalden>
especially back when its link-opening behavior wouldn't open a new tab but rather blow away the most recent one
23:35
<daedb>
sigh... I'm trying to learn some simple <canvas> stuff (using Dive into html5 as a guide), and I can't even get past the first bloody getContext. Javascript hates me :(
23:38
<Philip`>
daedb: Sounds like a bug in Dive into HTML5 if you can't get its examples to work
23:38
<Philip`>
daedb: Blame mpilgrim :-)
23:40
<TabAtkins>
All right, got it running. Awesome.
23:40
<TabAtkins>
That was too easy.
23:43
<daedb>
Philip`: I found a tutorial on Mozilla's dev site, and it's exactly the same with their example code...
23:45
<Philip`>
daedb: What kind of problem are you getting?
23:45
<Philip`>
daedb: Also, what browser are you running it in? (I hope it's not IE :-) )
23:46
<daedb>
Philip`: <script>var canvas = document.getElementById('tid'); var context = canvas.getContext('2d');</script> <-- all my code so far, gives a TypeError (message: Statement on line 1: Cannot convert undefined or null to Object
23:46
<daedb>
)
23:46
<daedb>
this is in Opera, btw
23:46
<Philip`>
daedb: Do you have a <canvas id=tid></canvas>, before the script?
23:48
<daedb>
The <canvas> has to be before the <script>? I had the script in the head...
23:48
<TabAtkins>
The script is trying to run before the <canvas> is loaded.
23:49
<Philip`>
Is there some way I can do something equivalent to <div style="background: black; opacity: 0.8"> (on top of a background image) that also works in IE6+?
23:49
<daedb>
Whee, it works after moving down the script. Thanks, Philip`!
23:49
<Philip`>
daedb: The <script> executes as soon as it's parsed, so the <canvas> hasn't been parsed yet, but you could solve it more nicely by using onload
23:49
<TabAtkins>
Philip: use a PNG and one of the pngFix libraries?
23:49
<Philip`>
daedb: like <script>function draw() { var canvas = ... }; window.onload = draw;</script> ... <canvas>
23:51
<TabAtkins>
Or like <script>$(function(){ var canvas=...; });</script> ^_^
23:51
<Philip`>
TabAtkins: My way requires tens of kilobytes less scripting :-p
23:51
<TabAtkins>
Bah.
23:52
<daedb>
heh, I don't do much javascript... I should probably learn it properly some day :p
23:53
<TabAtkins>
It's a good language, once you library your way past the bad DOM apis.
23:57
<TabAtkins>
Sigh... Damn you, magic fieldset styling. I hate you *so* much.