00:51
<zewt>
sort of amusing to see people talking about one of the first things i raised on webapps a year and a half ago
00:52
<zewt>
The Web Platform: Software Development Patience
02:51
<cabanier>
zewt: the canvas conversation?
02:53
<zewt>
sync messages
08:11
<annevk>
whoa
08:11
<annevk>
2119 had errata?
08:12
<annevk>
maybe they should have another one to remove RECOMMEND, NOT RECOMMEND, and OPTIONAL
08:12
<annevk>
*ED
08:12
<annevk>
oh, and SHALL, SHALL NOT
08:13
<annevk>
being able to recommend rather than encourage without it meaning SHOULD is nice
09:23
<deano>
annevk: Welcome back.
10:21
<jgraham>
http://www.hackermusings.com/2012/08/using-git-to-push-to-mozillas-hg-repositories/ seems quite interesting
10:21
<jgraham>
MikeSmith: ^
10:21
<annevk>
thanks deane
10:43
<deane>
annevk: Hey there. Not sure if you read the logs or not, so not sure if you saw this => http://krijnhoetmer.nl/irc-logs/whatwg/20120901#l-384
10:45
<annevk>
deane: thanks, matjas, not sure what you want me to say
10:45
<annevk>
matjas: I suggest convincing abarth
11:14
<Smylers>
annevk: Re http://krijnhoetmer.nl/irc-logs/whatwg/20120903#l-8 onwards, is Hixie's thought that you are working on it right?
11:14
<Smylers>
Is there a bug tracking this specific point, or should I file one?
11:15
<annevk>
Smylers: I have done some work on parsing URLs; the idea is that you will end up with the / once you serialize such an object
11:15
<Smylers>
Thanks.
11:15
<annevk>
Smylers: but I'm not sure when I'll get around to doing the remainder of the research
11:16
<Smylers>
annevk: Sure.
11:17
<Smylers>
annevk: In terms of what to tell somebody who's encountered this, would “The non-IE behaviour will be specified, but that part hasn't been written yet.” be accurate?
11:18
<annevk>
yeah I suppose
11:19
<annevk>
any eventual URL specification will closely resemble Gecko/Safari/Chrome (all subtly different unfortunately)
11:19
<annevk>
IE's URL parsing has many weird quirks
11:20
<jgraham>
Smylers: I think you tell them "You expected the most fundamental parts of the web platform to be well specified? Are you new here?"
11:22
<jgraham>
Smylers: You should also tell them that it could be worse; the TAG still haven't decided if http://example.org/ refers to a website or a car
11:23
<jgraham>
Or maybe they have now. Who knows
11:56
<Smylers>
jgraham: The person in question isn't “here”, so whatever I tell him may well affect his view of the web standards process. I was trying not to alarm him too much.
11:59
<darobin>
jgraham: actually, the question of what a URL refers to has in fact been resolved: http://faketag.org/a-101-on-the-range-of-the-http-dereference-function/
12:05
<Smylers>
jgraham: They claim to've decided it here: http://lists.w3.org/Archives/Public/www-tag/2005Jun/0039.html
12:07
<Smylers>
jgraham: Though I note that ‘solution’ describes itself as removing ambiguity. Of the three cases they list, (b) is “could be any resource” and (c) is “unknown” — which some people may still interpret as leaving a little ambiguity there.
12:08
<darobin>
Smylers: don't listen to www-tag — it's an usurpation of the real TAG! only read the faketag.org resolution, it's the only one that breaks the Web in all the right ways
12:09
Ms2ger
raises an eyebrow
12:25
<jgraham>
Smylers: http://www.w3.org/2001/tag/group/track/issues/14 suggests that the issue was discussed at a F2F a few months ago, and there was a TAG-Member only thread (I guess one should hide one's shame) questioning whether the issue is in fact still open
12:25
<jgraham>
Still it does give the impression that they aren't sure if they are dereferencing their arse or their elbow
12:27
<MikeSmith>
jgraham: thanks for the link about pushing to hg from git
12:28
<jgraham>
MikeSmith: Yeah, it seems like all the actual code might be somewhat hidden
12:28
<jgraham>
But maybe we could ask them to reuse it
12:31
<jgraham>
MikeSmith: BTW people on web-notifications are asking about publication status
12:32
<jgraham>
I imagine that is blocked by the lack of annevk
12:32
<jgraham>
Or, less specifically, the lack of a chir
12:32
<jgraham>
*chair
12:32
<jgraham>
But maybe you should say that since you are staff contact
12:32
<jgraham>
(I think)
12:32
<Ms2ger>
Did jgraham just volunteer to chair?
12:32
<jgraham>
Hahahaha
12:35
<odinho>
Ms2ger: Lots of voulenteering happening in this channel lately.
12:35
<Ms2ger>
odinho, oh yes, you were going to edit something, I just can't remember what...
12:45
<Ms2ger>
jsbell--
12:48
<jgraham>
?
12:49
<Ms2ger>
Writing tests with some obscure test harness
12:53
<jgraham>
Ugh. Link?
12:54
<Ms2ger>
https://code.google.com/p/stringencoding/
12:54
<Ms2ger>
On another note
12:54
<Ms2ger>
> we ran into site compatibility issues
12:54
<Ms2ger>
Is there a compat problem?
12:59
<jgraham>
Oh, qunit, yuck
13:00
<jgraham>
Doesn't look too hard to convert though
13:34
<crocket>
How can I make a web browser display consecutive whitespace characters?
13:34
<crocket>
They all collapse into one whitespace character.
13:38
<Lachy>
crocket, depends on what you need to do that for. There are multiple methods
13:40
<Lachy>
crocket, 1. <pre> </pre>. 2. Use CSS 'white-space' property, or 3. Use non-breaking spaces &nbsp;
13:40
<crocket>
Lachy, How do I apply a monospace font-family?
13:41
<Lachy>
CSS 'font-family' property, or the 'font' shorthand.
13:41
<Lachy>
crocket, if you're trying to markup code and want it displayed in a monospace font with whitespace preserved, then just use <pre>
13:42
<Lachy>
that's the default rendering
13:45
<crocket>
Lachy, What if I want a monospace serif font?
13:48
<Lachy>
crocket, lots of monospace fonts have serifs on the glyphs.
13:49
<Lachy>
but you can't declare both monospace and serif generic font families together
13:50
<crocket>
Lachy, That's unfortunate.
13:51
<Lachy>
crocket, you can name a specific font if you have one in mind
13:51
<Lachy>
But even Courier New has serifs, so I don't see the problem.
14:43
<Ms2ger>
annevk, I guess you haven't set up the post-push hook for DOM yet?
15:15
<annevk>
Ms2ger: no, I should
15:15
<annevk>
Ms2ger: that is, my plan is to update dom.spec.whatwg.org on push
15:15
<Ms2ger>
Good :)
15:16
<annevk>
I wanted to wait a bit for the W3C to reply to my inquiry
15:16
<annevk>
and then move on
15:17
<annevk>
with a timeout of about one more week, as I'm somewhat eager to do some spec updates
15:17
<Ms2ger>
I did a few
15:17
<annevk>
I noticed you pushed to github already, nice :)
15:18
<Ms2ger>
hg-git works well enough
15:22
<jgraham>
Hixie: I assume you don't object to your document.open-in-unload tests being submitted to W3C under the normal license for the testsuite? I adapted them to testharness.js
15:26
<Ms2ger>
jgraham++
15:37
<Hixie>
jgraham: sure
15:37
<Hixie>
jgraham: (what's the license? mit?)
15:39
<jgraham>
Hixie: Dual W3C Testsuite / 3 clause BSD
15:39
<Hixie>
yeah, that's fine
15:40
<jgraham>
http://www.w3.org/Consortium/Legal/2008/04-testsuite-copyright.html
15:44
<jgraham>
http://w3c-test.org/html/tests/submission/Opera/loading_web_pages/browsing_the_web/unloading_documents/document_open/
15:45
<jgraham>
I may have broken the tests of course, please let me know if I did
15:46
<jgraham>
(also, WTF, I just tried one of the tests in WebKit and got a QUOTA_EXCEEDED_ERR)
15:46
<jgraham>
OK, that's just 005
15:51
<Hixie>
how can you get a storage exceeded error
15:52
<Hixie>
sweet lord, the tests now import tons of scripts
15:52
<Hixie>
i make no guarantees that they are valid
15:52
<Hixie>
(and i still don't understand why we have to do this rather than just exposing an API and using iframes)
15:53
<jgraham>
I'm not sure what you mean by "exposing an API and using iframes"
15:53
<Hixie>
oh i see, that's in fact what you do
15:53
<Hixie>
nevermind
15:53
<Hixie>
i didn't realise you renamed it too :-)
15:53
<jgraham>
Oh you mean 001.html -> 001-1.html
15:54
<Hixie>
yeah
15:54
<Hixie>
i also don't understand the quota error in 005
15:54
<jgraham>
Yeah, your tests are running in their own browsing context. New windows are slightly less evil than iframes for things that poke at the history
15:54
<jgraham>
Or document loading
15:55
<Hixie>
right
16:02
<zewt>
jgraham: more likely to trigger unrelated issues, though (popup blockers and not being allowed to open new windows--two issues I just hit trying to run that one)
16:02
<zewt>
(the first is easy enough to undo, the second i'd have to go play with obscure about:config stuff so I didn't bother)
16:03
<jgraham>
zewt: Right, alllowing popups from test sites does become necessary
16:03
<Hixie>
I really don't understand this quota problem
16:04
<jgraham>
But it can be a tab rather than a window
16:04
<Hixie>
is there some infinite loop going on or something?
16:04
<zewt>
jgraham: i have firefox set to never allow new windows or tabs; links always open in the same tab
16:04
<zewt>
(too many abusive sites trying to open new windows for every single link)
16:04
<jgraham>
zewt: Oh, well yeah I guess if you have weird non-standard configurations it is harder
16:05
<zewt>
one of those things i've never understood browsers allowing, since the abuse is so widespread and annoying
16:06
<jgraham>
Hixie: So I note that 001 seems to give 012389 rather consistently in existing browsers
16:06
<jgraham>
Ignoring Opera which is clearly broken and IE which I would have to change computers to test
16:06
<Hixie>
yeah, several of the test cases give reliable results that differ from what i ended up requiring, but it turns out that if you examine what they're doing it is crazy and doesn't really work in all cases
16:07
<Hixie>
(gecko and webkit in particular actually do quite different things)
16:07
<Hixie>
(yet end up with similar results on some of the tests out of basically coincidence)
16:08
<jgraham>
OK
16:08
<jgraham>
I am always nervous when the spec doesn't match implementations
16:08
<jgraham>
Especially if they match each other
16:08
<Hixie>
none of the implementations actually match here
16:08
<Hixie>
see my rants on irc (i linked to them from the checkin comment) for more details
16:08
<jgraham>
But obviously this is an insane thing to rely on
16:09
<Hixie>
(though note that i didn't do exactly what i suggested i should do in those notes, in the end)
16:10
<jgraham>
Which probably means that only facebook, google and twitter depend on it ;)
16:58
<annevk>
hmm
16:58
<annevk>
github is annoyingly inconsistent with URLs
16:59
<annevk>
/whatwg and /WHATWG both work
16:59
<annevk>
as well as /whatWG and other crap
17:01
<zewt>
okay, that's annoying
17:11
<ubitux>
i don't understand what this http://dev.w3.org/html5/webvtt/#webvtt-cue-ruby-span is
17:14
<ubitux>
anyone has a simple explanation about what to do with this?
17:15
<zewt>
http://www.whatwg.org/specs/web-apps/current-work/#the-ruby-element
17:15
<webben>
ubitux: It's a way to put Ruby annotations in captions.
17:16
<webben>
ubitux: basically it's a type of oriental gloss: https://en.wikipedia.org/wiki/Ruby_character
17:18
<ubitux>
oh ok
17:18
<ubitux>
thanks
17:32
<ubitux>
that will be kind of painful to support such format :(
17:33
<zewt>
(how? who? where? context :)
17:34
<ubitux>
outside a browser
17:34
<ubitux>
C app
17:34
<ubitux>
subtitles transcoding
17:34
<ubitux>
rendering
17:34
<zewt>
transcoding is easy enough (if the format you're converting to doesn't support it, you remove the unsupported feature)
17:35
<ubitux>
transcoding is one of the most difficult
17:35
<ubitux>
the worst being the rendering, closely related
17:35
<ubitux>
the problem is, for subtitles rendering, the only library we can use is libass
17:35
<ubitux>
so we basically need to convert webvtt markup to ASS one
17:36
<zewt>
not due to this feature--you already have to parse the cue text, and you remove nodes for unsupported features
17:36
<zewt>
(converting to another format for rendering is probably not the best approach, though it may be the quickest short-term)
17:36
<ubitux>
that's the only solution
17:36
<ubitux>
or maybe you want to write a subtitles rendering library?
17:36
<ubitux>
at the moment libass is the most mature one
17:37
<zewt>
how could it possibly be the only solution? :)
17:37
<ubitux>
well i don't really want to dedicate 5 yr of my life to rewrite a rendering engine
17:38
<zewt>
rendering subtitles isn't a 5-year project unless a codebase is catastrophically broken
17:38
<Ms2ger>
Who's to say it isn't so?
17:38
<ubitux>
libass still has font glyph issues
17:38
<ubitux>
font stuff is a complex matter
17:39
<ubitux>
mixed with blur, rotation and misc. other transformations
17:39
<zewt>
that's why you don't write a font renderer, you use a font rendering engine (like freetype)
17:39
<ubitux>
and now ruby and stuff..
17:39
<ubitux>
that's what libass does
17:40
<ubitux>
libass is a "small" project of 10k lines
17:40
<ubitux>
the best move would be to fork it and make it supports the new vtt markup and stuff
17:41
<ubitux>
but that also means supporting CSS stuff
17:41
<ubitux>
and making available the access to the AST etc
17:42
<ubitux>
anyway, that's far from trivial
17:42
<ubitux>
i guess all of this is not really a problem for browsers
17:42
<ubitux>
but for anything else it is
17:43
<zewt>
(nothing is "trivial", and no matter what you do you need to support vtt's subset of CSS at some level)
17:43
<ubitux>
that's assuming browser will actually force the limit to these properties
17:44
<ubitux>
which won't happen and users will start abusing it quite quickl
17:44
<ubitux>
+y
17:44
<gsnedders>
ubitux: Do you have any evidence for that?
17:44
<ubitux>
well, that's what always happen
17:45
<zewt>
(i'm not assuming anything--no matter what you do, you need to be parsing CSS somewhere if you intend to support vtt's styling features at all)
17:45
<ubitux>
if even only one browser allows all the CSS stuff (because that's likely simpler), users WILL abuse it
17:45
<zewt>
whether that happens as part of a transcoding layer or closer to the renderer, it's got to be somewhere
17:47
<ubitux>
well anyway, requiring a CSS parser/matcher is kind of a problem
17:47
<ubitux>
of course, we can just ignore all that part
17:50
<ubitux>
now if we assume we actually write that CSS stuff (which is unlikely because of the complexity) the rendering will require something at least as big as libass
17:51
<ubitux>
so well, this is why i said it will be kind of painful to support such format
17:51
<ubitux>
(outside the browser)
17:51
<ubitux>
(so any players or subtitles editor/converter)
17:53
<zewt>
... that's something any serious captioning format is going to need (though comparing to ass isn't accurate--it supports lots of stuff other formats, including vtt, don't, like karaoke effects)
17:54
<ubitux>
ass supports karaoke
17:54
<ubitux>
jacosub as well
17:54
<zewt>
what? i just said ass does and vtt doesn't :)
17:55
<ubitux>
i thought vtt did
17:55
<zewt>
(vtt has very basic "change color at this time", but not all the goofy effects people do with ass)
17:55
<ubitux>
vtt has features not supported by ASS
17:55
<ubitux>
so you can't really fallback on the now-not-fully-featured-enough-ASS-system
17:56
<ubitux>
OTOH, if a full featured (parsing/styles-and-events-AST-publicly-accessible/rendering/portable) libsubtitles/libwebvtt was available, all of that wouldn't be a problem
17:56
<ubitux>
but that's a long term project.
17:56
<zewt>
anyway, if what you want is every subtitle format to be a subset of what you already have so you don't have to implement anything, that's not likely to happen :)
17:56
<ubitux>
yes, and that's a problem ;)
17:57
<ubitux>
at the moment most systems rely on the fact that almost all existing subtitles can be expressed as an ASS-subset
17:57
<zewt>
um, most systems don't support ass at all, heh
17:58
<ubitux>
and this is neat in the sense that there is a parsing/rendering lib to do all the stuff
17:58
<ubitux>
zewt: most players do
17:58
<ubitux>
they're using the libass rendering system for all the subtitles
17:58
<ubitux>
(except the simple formats)
17:58
<ubitux>
(sometimes)
17:58
<zewt>
you mean "desktop software"; set-top stuff rarely has usable support at all (wdtv claims support, and they lie--it's so incomplete it's unusable)
17:59
<ubitux>
i meant any app ≠ browsers
18:00
<zewt>
wdtv isn't a browser, heh
18:00
<ubitux>
so players like mplayer, vlc, possible the windows one (mpc?), as well as ffmpeg and such
18:00
<ubitux>
dunno what wdtv is
18:05
<ubitux>
anyway, i guess the best way would be to start a subtitles library being able to represent any kind of style structures and related crazyness, as well as the rendering part
18:05
<ubitux>
possibly a libass fork
18:05
<ubitux>
and then port all the subtitles decoders and encoders to it
18:05
<ubitux>
but this really is IMO a 5yr project
18:05
<zewt>
you're just not likely to get traction asking for all subtitle formats to be a subset of some particular library, which partially implements a particular weird ad hoc format
18:06
<ubitux>
just to support webvtt :)
18:07
<zewt>
i'd call it a project of a few months, if you can find a standalone css parsing/selector matching implementation
18:07
<ubitux>
zewt: i believe the main point of webvtt is to make a standard way of using subtitles accross the web, but i don't think it requires all that styling stuff
18:08
<zewt>
stylesheets are always technically optional
18:08
<ubitux>
yes but they are used as the basis of most styles, such as coloring and stuff
18:08
<zewt>
what? i just agreed with you, heh
18:08
<ubitux>
this means not supporting them is kind of a problem
18:08
<zewt>
i didn't say it wasn't, it seemed that you did :)
18:09
<ubitux>
(oh and i'm omitting the charset/encoding stuff)
18:09
<zewt>
stylesheets are important for presentation (especially if the current hideous black-rectangle-behind-captions default doesn't change), but no, it's not required as such
18:10
<zewt>
encoding is straightforward; webvtt is always utf-8
18:11
<zewt>
you really can't ask for anything simpler than *that* :)
18:12
<ubitux>
ASCII!
18:12
<zewt>
not so simple for japanese users :P
18:12
<ubitux>
romanji is more than enough
18:12
<ubitux>
:)
18:13
<zewt>
really, though, if you want implementation advice, feel free to open a discussion on public-texttracks
18:14
<zewt>
(for example, suggestions of a standalone CSS implementation, which I don't have off the top of my head)
18:15
<ubitux>
http://www.w3.org/Style/CSS/SAC/ ?
18:15
<zewt>
don't know (you'd need a parser and selector matching, at least)
18:16
<ubitux>
and then we will need a CSS rendering engine
18:16
<zewt>
and a robust enough API to allow iterating over :past/:future states and adding ::cue
18:17
<ubitux>
:/
18:17
<zewt>
no, you only need to be able to determine the state of properties at specific times (for example, font sizes aren't allowed to change once text is displayed, which simplifies a lot)
18:19
<zewt>
and the list of times at which properties change is finite, so i believe css rules + cues can be flattened to a simple list of states
18:21
<zewt>
anyway, i can't engineer this for you (and i need to eat), but again, mailing list :)
18:23
<ubitux>
i need to brainstorm by myself about what i really want indeed
18:23
<ubitux>
i think we will just make webvtt an equivalent of subrip with a lot of unsupported features
18:24
<ubitux>
and it will be more than enough :p
18:25
<zewt>
which implementation is this for, so I know what to avoid :P
18:25
<ubitux>
ffmpeg, and most likely will be used in the future by mplayer
18:25
<ubitux>
i believe vlc won't do anything more complicated as well
18:25
<ubitux>
and i don't think anything outside browsers will do as well
18:26
<ubitux>
at least not in the near future
18:28
<zewt>
(you're making me want to implement it for xbmc, so people have a good alternative)
18:29
<ubitux>
is xbmc linking with --enable-webkit?
18:30
<ubitux>
oh and btw, we can't actually rely on any library
18:30
<ubitux>
because what's the decoded packets for subtitles will contain the exploitable AST of the events & styles
18:31
<ubitux>
so it would require api user to link on that library as well, which is a problem
18:31
<ubitux>
so the full chain needs to be available
18:31
<ubitux>
as builtin
18:53
<zewt>
no, and you seem to have a misunderstanding of the complexity of webvtt if you think you need webkit to render it
19:15
<annevk>
so Instant Upload from Google+
19:16
<Ms2ger>
Google+ still exists?
19:16
<annevk>
is that known to fail when you attempt to upload 1500 at once?
19:16
<zewt>
is that the camera upload thing? because it's so un-instant i turned it off (useless)
19:16
<annevk>
it says "Starting upload..." but nothing happens
19:17
<annevk>
I'm not interested in paying Apple for iCloud so I figured I'd use this as backup instead, but it's rather crappy