| 00:19 | <Hixie> | MikeSmith: so i met with lisa |
| 00:20 | <Hixie> | MikeSmith: she's open to us splitting off the protocol part of websocket, the content-sniffing section, the uri section, and a brief definition of the Origin header, and submitting them as four tentative IDs |
| 00:20 | <Hixie> | (the api part of web socket would go into a separate w3c draft) |
| 00:21 | <Hixie> | MikeSmith: assuming there's no objections, i'll start looking at doing that in early january |
| 00:27 | <takkaria> | Hixie: have you any suggestions for interesting board games to play? |
| 00:27 | <Hixie> | how long? how many people? how geeky are the people? |
| 00:27 | <Lachy> | Hixie, who's Lisa? Is she someone from the HTMLWG? |
| 00:27 | <takkaria> | any length of time, three to six, and really quite geeky |
| 00:28 | <Lachy> | takkaria, megamonopoly |
| 00:28 | <Hixie> | lachy: Lisa Dusseault = ietf apps area director |
| 00:28 | <Hixie> | takkaria: i would recommend Twilight Imperium 3 |
| 00:28 | <Lachy> | that's monopoly with 2 boards joined together |
| 00:28 | <Hixie> | takkaria: takes about 12 hours to play the first time, though |
| 00:29 | <takkaria> | christ, that looks awesome. :) ta |
| 00:30 | <takkaria> | Lachy: how does that work? |
| 00:32 | <Lachy> | join the 2 boards diagonally together with the GOs overlapping and do a figure 8 around both |
| 00:32 | <takkaria> | hmm |
| 00:33 | <rillianbis> | Lachy: does it come with highrise towers as a hotel upgrade? |
| 00:33 | <Lachy> | just watch out for the extra long expensive strip since the 2 with Mayfair, etc. end up line up |
| 00:33 | <neatnik> | Mayfair? |
| 00:33 | <Lachy> | rillianbis, you're free to make up any variation |
| 00:33 | <rillianbis> | :) |
| 00:33 | <Lachy> | neatnik, that's the most expensive property on a standard monopoly board |
| 00:33 | <neatnik> | Lachy: oh. I thought it was Boardwalk |
| 00:34 | <Lachy> | what? |
| 00:34 | <Lachy> | in which version? |
| 00:34 | <neatnik> | I've never seen a monopoly board with a Mayfair on it |
| 00:34 | <neatnik> | dunno, I grew up playing the standard Parker Bros version |
| 00:34 | <neatnik> | American version, I guess |
| 00:34 | <rillianbis> | neatnik: Mayfair is what the UK edition called Boardwalk |
| 00:34 | <neatnik> | did they change it? :( |
| 00:34 | <neatnik> | Oooh. |
| 00:34 | <neatnik> | cultural misunderstanding :D |
| 00:34 | <neatnik> | Mayfair sounds cooler than Boardwalk, for the record |
| 00:35 | <Lachy> | neatnik, in the original version based on London place names, it's Mayfair |
| 00:35 | <Lachy> | http://en.wikipedia.org/wiki/Monopoly_(game)#Board |
| 00:36 | <Lachy> | I can't remember what it's called in the first Australian edition that I have at home in Aus |
| 00:36 | <neatnik> | oh wow |
| 00:36 | <neatnik> | I had no idea. |
| 00:36 | <neatnik> | Growing up in America, you're more or less programmed to believe that everything here is the most significant and important version of everything |
| 00:36 | <neatnik> | And that nothing else matters |
| 00:37 | <neatnik> | So it never occurred to me that there'd even be another version of Monopoly, let alone that it would be the *original* version :/ |
| 00:37 | <rillianbis> | with effort and experience, this assumption can be overcome :) |
| 00:37 | <Lachy> | Growing up out of America, you're more or less programmed to believe that everything there is the least significant and unimportant version of everything :-) |
| 00:37 | <neatnik> | rillianbis: hmm, I've always assumed that it would require a time machine |
| 00:38 | <neatnik> | rillianbis: and that I'd have to go back in time and convince my parents to move to, say, Sweden when I was 1 or 2 years old |
| 00:38 | <MikeSmith> | Hixie: thanks (update about meeting with Lisa Dusseault) |
| 00:38 | <Lachy> | hmm, apparently both the US and UK versions have Park Place in the same location |
| 00:40 | <neatnik> | oh, cool |
| 00:40 | <MikeSmith> | Hixie: seems to me to be a good way to go forward and I thin (hope) it will be more palatable (for lack of a better word) to some who have had issues with those parts |
| 00:40 | <neatnik> | My sister and I would play these insanely long Monopoly games when we were younger |
| 00:40 | <roc> | it's Park Lane in the UK version |
| 00:41 | <neatnik> | we both used similar tactics, and so it was hard to beat one another |
| 00:41 | <neatnik> | we'd start in the evening, play forever, and then have to give up and go to sleep before resuming the next morning |
| 00:41 | <neatnik> | there was always a 50% chance that a cat would ruin the board during the night :( |
| 00:41 | <Hixie> | MikeSmith: cool |
| 00:42 | <Lachy> | roc, oh right. I was looking at the 2 different american layouts in wikipedia without realising. I thought one was the UK version |
| 00:42 | <MikeSmith> | Hixie: I suspect there are still going to be some who object to publishing, e.g., the content-sniffing part at all |
| 00:42 | <MikeSmith> | but I guess we can cross that bridge when we come to it |
| 00:42 | <rillianbis> | roc: did you have any more thought's about Dave's suggestion that media wrappers like SMIL need to be supported in <video>? |
| 00:43 | <neatnik> | heh, Old Kent Road |
| 00:43 | <neatnik> | that sounds like a fun place |
| 00:43 | <roc> | rillianbis: I think that's a reasonable enhancement, but it will add considerable complexity |
| 00:43 | <rillianbis> | I guess Silvia's proposal is relavant to that in that it is proposing a dom method for discovering component tracks |
| 00:43 | <neatnik> | oh wow, you don't have a "Just Visiting" area outside of your jail? |
| 00:44 | <rillianbis> | roc: I guess you said as much before |
| 00:45 | <rillianbis> | what about getting text data out of the video stream? do you see that in the same category? |
| 00:45 | <roc> | not at all |
| 00:45 | <Lachy> | neatnik, yes, Just Visiting is supposed to be there |
| 00:46 | <Lachy> | Wikipedia is just wrong |
| 00:46 | <neatnik> | phew, good |
| 00:46 | <roc> | loading multiple resources is the problem here |
| 00:46 | <roc> | dealing with what can happen if one resource loads but the other doesn't |
| 00:46 | <roc> | or they load at different rates |
| 00:47 | <roc> | SMIL is especially difficult because you have to load this intermediate XML file, validate it, and then go off and load more resources |
| 00:47 | <rillianbis> | ok, but your object applies to the <video><text></video> suggestion as well |
| 00:47 | <MikeSmith> | takkaria: Uncle Wiggily |
| 00:48 | <MikeSmith> | but I think the rules require that you play it while you're drunk |
| 00:48 | <rillianbis> | roc: your objection does, rather |
| 00:48 | <roc> | not so much |
| 00:49 | <roc> | if the text doesn't load we can just play the video |
| 00:49 | <roc> | if the video doesn't load, the text is useless |
| 00:49 | <roc> | if the text loads too slowly we can just ignore it |
| 00:49 | <roc> | it's ancillary |
| 00:50 | <nessy> | with the roe file (in contrast to smil), the roe would also be ancilliary to a video and allow basically server-side what the <text> tags do client-side |
| 00:50 | <nessy> | I guess that falls under the same concerns then |
| 00:51 | <rillianbis> | roc: ok good. that's how I originally read your email. thanks for clarifying |
| 00:51 | <rillianbis> | so what would the api for getting text data out of a video element's src stream look like? |
| 00:52 | <roc> | dunno |
| 00:52 | <roc> | I don't know what the data model for the text is |
| 00:53 | <rillianbis> | yeah, we don't either |
| 00:53 | <roc> | ah :-) |
| 00:53 | <rillianbis> | so right now that's a 'it depends' |
| 00:53 | <rillianbis> | if it's xml, being able to slup it is nice |
| 00:53 | <roc> | I imagine one option would be a set of text data objects, each with an associated time interval |
| 00:53 | <rillianbis> | but if it's muxed, would events be better? |
| 00:53 | <rillianbis> | *nod* |
| 00:54 | <roc> | if that's your data model, then I think a convenient API would be "fire an event whenever the current time enters the time interval for a text object" |
| 00:54 | <rillianbis> | my concern was that Dave implied there should be a unified method of accessing this stuff that also works with wrappers like SMIL |
| 00:54 | <roc> | or leaves |
| 00:54 | <roc> | and the event would pass in an object with the text and the interval |
| 00:54 | <rillianbis> | that would work for both, of course, but it's specific to text |
| 00:54 | <rillianbis> | (or chapter/annotation tracks) |
| 00:55 | <rillianbis> | and not so useful for a/v data, controlling conditionals, etc. |
| 00:55 | <nessy> | for srt that works nicely |
| 00:55 | <nessy> | other formats bear more information |
| 00:55 | <roc> | that API is much like cue ranges |
| 00:55 | <rillianbis> | roc: right, that's what I was thinking of |
| 00:55 | <nessy> | such as karaoke-style formatting |
| 00:56 | <rillianbis> | nessy: couldn't that just be more-complete data inside the text objects? |
| 00:56 | <roc> | some users might want to be able to grab all available text objects and their intervals |
| 00:56 | <roc> | like if you want to be able to display text ahead and behind the current point |
| 00:57 | <nessy> | rillianbis: if it's xml marked-up, it should be parsed, right? |
| 00:57 | <roc> | designing for extensibility is fine but we have to be careful to create something that can actually be used |
| 00:57 | nessy | nods |
| 00:58 | <rillianbis> | nessy: I don't know enough about dom and javascript to have an opinion there :) |
| 00:58 | <nessy> | which is why trying srt first and getting it right for that simple format is good |
| 00:58 | <nessy> | then we can extend for more complex formats such as dfxp |
| 00:58 | <roc> | it's tempting to pass back a big blob and say "this could be anything!" ... infinitely extensible but people can't use it |
| 00:58 | <nessy> | that's my concern, too |
| 00:59 | <nessy> | the dom or api should be as simple as possible |
| 00:59 | rillianbis | nods |
| 00:59 | <nessy> | yet, some things should be enabled that go beyond just text and time markers |
| 00:59 | <rillianbis> | roc: so I think the cue ranges make a lot of sense |
| 01:00 | <nessy> | for example, creating a speech bubble on a person with some text that has a hyperlink |
| 01:00 | <roc> | one option is that the "text object" passed to the event is a DOM node that's created (but not inserted) in the current HTML document |
| 01:00 | <rillianbis> | they just need another attribute we can hang text on |
| 01:00 | <Hixie> | MikeSmith: for sure. but i don't see any way to satisfy that position, so... |
| 01:00 | <roc> | so that could be a text node, but it could be an <A> element |
| 01:01 | <MikeSmith> | Hixie: right, me neither |
| 01:02 | <roc> | whatever semantics the browser can extract from the metadata, it can translate into HTML |
| 01:02 | <roc> | that translation would probably be in an auxiliary spec somewhere |
| 01:02 | <roc> | whatever can't be translated to sensible HTML could be returned as an XML blob |
| 01:03 | <nessy> | excellent, that was what I hoped would be possible |
| 01:03 | <nessy> | a bit like the iFrame - create a dom that is essentially html, but applied to the video |
| 01:03 | <roc> | the idea would be that a script could expect to just insert that HTML somewhere and something good would happen |
| 01:03 | <nessy> | would that work with a html excerpt or does it need a full html page ? |
| 01:04 | <roc> | or the browser could just display the HTML itself inside or next to the video |
| 01:04 | <roc> | a full HTML page is too heavyweight |
| 01:04 | <roc> | IMHO |
| 01:04 | <nessy> | good, I prefer that, to |
| 01:04 | <nessy> | too |
| 01:04 | <nessy> | since a time-aligned text excerpt tends to be something like a <div>, it would be better as an excerpt |
| 01:05 | <roc> | so an interesting thought experiment would be for you to look at the existing formats and try to define how they would be mapped into HTML nodes |
| 01:05 | <rillianbis> | should this be a separate callback just for text events? |
| 01:06 | <roc> | I don't know |
| 01:06 | <rillianbis> | or is it reasonable for all cue ranges to do something like this |
| 01:06 | <roc> | I forget what the current status of cue ranges is |
| 01:06 | <nessy> | yes, that was what I intended to do |
| 01:06 | <rillianbis> | and define (formally or by convention) some magic classes? |
| 01:06 | <rillianbis> | roc: there's not much in the spec |
| 01:06 | <rillianbis> | you add them, and they call your callbacks |
| 01:06 | <adu> | can i has getContext('3d')? |
| 01:06 | <roc> | rillianbis: classes? You mean like "class" attributes? |
| 01:07 | <rillianbis> | yes |
| 01:07 | <rillianbis> | that's how cue ranges are sorted in the current spec |
| 01:07 | <rillianbis> | class attr for groups |
| 01:07 | <rillianbis> | id attr for individual ranges |
| 01:07 | <Philip`> | adu: Not yet :-p |
| 01:07 | <roc> | I have to reread the spec I guess |
| 01:07 | <nessy> | subtitles are almost the same as cure ranges, except for an additional <div> tag |
| 01:07 | <rillianbis> | there's currently no api for getting the available classes/ranges |
| 01:08 | <roc> | yeah, cue ranges with data |
| 01:08 | <roc> | would be nice to unify them I guess, if it doesn't burden cue range users unnecessarily |
| 01:08 | <rillianbis> | so there's no way to get the intrinsic ranges the player might know about from the stream |
| 01:08 | <adu> | Philip`: ok, 2d will do then :) |
| 01:08 | <adu> | Philip`: are you familiar with x3d? |
| 01:08 | <rillianbis> | roc: we did that in CMML and I always thought it was a nice idea |
| 01:08 | <roc> | yeah an API to get available data would be nice |
| 01:09 | <rillianbis> | cue ranges, chapter markers, subtitles, outgoing hyperlinks, anchors...they're all the same kind of data functionally |
| 01:09 | <roc> | the API would probably take a time interval and return a list of ranges (including associated data, if any) that intersect the interval |
| 01:10 | <roc> | of course it would only return the ranges that are available. scripts would have to poll for updated results, maybe in a progress listener |
| 01:10 | <rillianbis> | yes |
| 01:10 | <rillianbis> | that also means they have to remember the last time they checked |
| 01:10 | <roc> | hmm, but that would be a problem for streams with caption data ... the browser would have to keep all captions in memory forever in case anyone asks |
| 01:10 | <Philip`> | adu: Somewhat - I made a very incomplete prototype X3D renderer using Mozilla's experimental 3d canvas a while ago :-) |
| 01:11 | <rillianbis> | roc: that's why I was thinking of magic classes |
| 01:11 | <adu> | Philip`: nice! |
| 01:11 | <roc> | so maybe a better idea would be to have an event that fires whenever range data is received from the server, and scripts can store whatever they need themselves |
| 01:11 | <nessy> | also, you only really want one subtitle track to display at one time - but be able to select between different languages |
| 01:11 | <rillianbis> | so you could add a callback for the 'subtitle_de' class, and just get called when a new one goes in/out of playback |
| 01:11 | <adu> | Philip`: that reminds me, have you heard of CrystalZilla? |
| 01:11 | <rillianbis> | but you still need some way to find out what's actually available in the stream |
| 01:12 | <rillianbis> | which is another getter, or polling anyway |
| 01:12 | <Philip`> | adu: No, but I have now |
| 01:12 | <adu> | Philip`: its kind of the opposite: http://interreality.org/wiki/crystalzilla |
| 01:12 | <nessy> | when used inside the <video> tag with <text> elements, it is easy to see the alternatives |
| 01:12 | <rillianbis> | hmm, but polling for available lanuages and updating the ui in the progress callback isn't as bad as general query for a range |
| 01:13 | <rillianbis> | nessy: so should the player create and insert text elements to represent embedded text tracks? |
| 01:13 | <Philip`> | adu: I've been kind of interested in using WebKit in an OpenGL game, but nobody seems to have quite made a good enough port of it yet :-( |
| 01:13 | <nessy> | maybe just the availability information could go into the dom |
| 01:13 | <adu> | Philip`: have you heard of glitz? |
| 01:13 | <nessy> | but we access the data through callbacks |
| 01:13 | <roc> | glitz is toast |
| 01:14 | <roc> | Chris Lord is working on a Gecko port for use with Clutter |
| 01:14 | <roc> | it's supposed to be working pretty well |
| 01:14 | <rillianbis> | nessy: does it help to split it into two use cases? (a) I want notification when there's a new subtitle, and (b) I want all the subtitles to dump in a transcript |
| 01:14 | <Philip`> | adu: For Cairo? |
| 01:15 | <adu> | Philip`: yeah, doesn't Gecko use Cairo since ff3? |
| 01:15 | <adu> | gecko -> cairo -> glitz -> opengl |
| 01:16 | <adu> | maybe I'm missing something |
| 01:16 | <Philip`> | adu: It seems to be much harder to make that kind of thing work in practice, sadly :-) |
| 01:16 | <roc> | you're just missing that glitz is unmaintained and broken |
| 01:16 | <adu> | Philip`: I'll do it |
| 01:16 | <adu> | roc: sounds like so many beautiful projects |
| 01:17 | <rillianbis> | rock, nessy: anyway, thanks for the brainstorm. |
| 01:17 | rillianbis | is off to dinner |
| 01:17 | <nessy> | rillianbis: not sure, really |
| 01:17 | <nessy> | rillianbis: enjoy :) |
| 01:17 | <roc> | a good quality GL backend for cairo is something everyone wants but no-one has gotten around to doing yet. Once we have one, getting Gecko working on it should be easy |
| 01:17 | <nessy> | yeah - I will go map existing formats to html, I think that's the first step |
| 01:17 | <nessy> | we will see the next :) |
| 01:18 | <Philip`> | I guess there are all sorts of problems with trying to integrate input events and timers and stuff with the rest of the application when you're trying to embed it, and it should all be cross-platform and ideally add zero size to the application |
| 01:19 | <roc> | and popup windows |
| 01:19 | <roc> | and plugins |
| 01:19 | <roc> | there are lots of issues :-) |
| 01:19 | <Philip`> | http://ajeanius.wordpress.com/2008/09/08/awesomium-google-chrome-hacked/ seems like what I want, except it's Windows-only |
| 01:21 | Philip` | doesn't care about plugins :-) |
| 01:55 | <adu> | i like the sound of Clutter |
| 01:55 | <adu> | i wonder why I never heard of this before |
| 02:04 | <fakeolliej> | Philip`: sigh, i find blogs like that depressing -- a lot of the "kick-ass" API has always been in WebKit, but people seem to love claiming the google somehow invented all of it |
| 02:08 | <roc> | what I currently find unfair is the current crop of news articles comparing browser JS performance that compare Chrome to Safari 3.2 |
| 02:09 | <Philip`> | olliej: I guess it'd be nice if people could actually use that WebKit API directly, without getting stuck by the lack of a decent open source Win32 port :-) |
| 02:14 | <adu> | roc: whats wrong with that comparison? |
| 02:15 | <roc> | Chrome is in beta (and so is FF3.1, which they also compared against), so they should have compared against a beta Safari or ToT Webkit |
| 02:17 | <roc> | it's especially tickly because when you compare Chrome against Safari 3.2 you're comparing new-Webkit vs old-Webkit, so to some extent you're using Apple's hard work to beat up their own browser |
| 02:21 | <MikeSmith> | Hixie: I reckon Adam Barth might be amenable to editing the Origin header ID, if you wanted to spread the work out a bit. |
| 02:21 | <MikeSmith> | though I do realize it's not going to be a big draft |
| 02:24 | <Hixie> | MikeSmith: yeah, he said he's also willing to do the sniffing stuff |
| 02:24 | <MikeSmith> | cool |
| 02:24 | <Hixie> | roc: as i understand it chrome is using a really old webkit, btw, so actually it probably cuts both ways :-) |
| 02:25 | <roc> | not that old |
| 02:25 | <MikeSmith> | Hixie: who knows, maybe he'd like it enough to volunteer to edit some other spec. Though why anyone would actually like spec editing is beyond me... :) |
| 02:25 | <Hixie> | MikeSmith: the girls |
| 02:26 | <MikeSmith> | heh |
| 02:26 | <MikeSmith> | yeah, I forgot about that part |
| 02:27 | <roc> | the POWER |
| 02:28 | <Hixie> | it's the implementors who have the power, not the spec writers :-( |
| 02:33 | <roc> | no, we are slaves to your will |
| 02:35 | <adu> | roc: well, Chrome isn't all WebKit, its also got some XPI stuff |
| 02:35 | <roc> | sure |
| 02:35 | <roc> | I inserted some appropriate weasel words |
| 02:35 | <adu> | roc: from what I hear its a mozilla-webkit cyber-child |
| 02:36 | <roc> | not really |
| 02:39 | Philip` | thought it was more along the lines of being WebKit plus some anti-phishing code that Google originally contributed to Mozilla, but he has almost exactly zero real knowledge and is probably totally wrong |
| 02:41 | <roc> | that's close |
| 02:41 | <Philip`> | It's close to totally wrong? :-) |
| 02:42 | <MikeSmith> | spec writers are "the unacknowledged legislators of the world".. |
| 02:42 | <MikeSmith> | also, "I fall upon the thorns of life! I bleed!" |
| 02:44 | <Hixie> | hsivonen: btw if you have opinions on http://www.whatwg.org/specs/web-apps/current-work/multipage/obsolete-features.html#conformance-checkers-0 it would be good to hear them :-) |
| 02:46 | <Philip`> | MikeSmith: I like how search engines allow me to understand literary references despite being an uneducated oaf who's never read any of those things |
| 02:49 | <MikeSmith> | Philip`: the Web has really taken a lot of the fun out of things. no longer possible to stump anybody with references to literary works, even relatively obscure stuff |
| 02:49 | <MikeSmith> | I remember years ago taking this quiz where the questions were all the final lines of poems |
| 02:49 | <MikeSmith> | and you had to figure out what poems they were from |
| 02:49 | <Philip`> | MikeSmith: You'll have to start paraphrasing, so people can't just do the obvious searches |
| 02:50 | <MikeSmith> | and if you didn't know, you had to figure out how to find them. in books |
| 02:50 | <MikeSmith> | Philip`: yeah |
| 02:50 | <MikeSmith> | not sure how to paraphrase "I fall upon the thorns of life! I bleed!" |
| 02:51 | <MikeSmith> | Hixie: dunno you if saw my note here earlier, but David Carlisle updated the "XML Entity definitions for Characters" draft to include the set of "html5-uppercase" names |
| 02:51 | <MikeSmith> | http://www.w3.org/2003/entities/2007doc/html5-uppercase.html |
| 02:51 | <Hixie> | yeah i saw |
| 02:51 | <MikeSmith> | http://www.w3.org/2003/entities/2007doc/byalpha.html |
| 02:51 | <Hixie> | wasn't sure whether that affected us or not |
| 02:51 | <MikeSmith> | OK |
| 02:52 | <Hixie> | it doesn't change the spec, right? |
| 02:52 | <MikeSmith> | Hixie: no |
| 02:52 | <Hixie> | i guess i'll find out when i next run my regen script |
| 02:52 | <Hixie> | since it pulls the file automatically each time :-) |
| 02:52 | <Philip`> | MikeSmith: You could adapt the message to the medium, and say "/me falls upon the thorns of life, and bleeds!" |
| 02:52 | <Hixie> | ok. time to go eat. bbl. |
| 02:52 | <MikeSmith> | Philip`: :) |
| 02:54 | <MikeSmith> | Lachy: fwiw, I notice that http://www.w3.org/2003/entities/2007doc/byalpha.html links to image files for all the characters |
| 02:54 | <MikeSmith> | that seems like a better way than actually including the images in the doc itself |
| 02:56 | <Lachy> | not if I do it the way I was intending to have the images load dynamically within the page as when needed, rather than all at once |
| 02:56 | <Lachy> | s/as when/when/ |
| 03:07 | <MikeSmith> | Lachy: anyway, you had mentioned need to have images available under a license that allowed you to use them. so those would seem to be |
| 03:08 | <MikeSmith> | though again, it seems to me the existence of http://www.w3.org/2003/entities/2007doc/byalpha.html possibly obviates the need to include the list in other specs, since that can just be referenced |
| 03:09 | <MikeSmith> | but that said, I'd definitely be interested in seeing what you put together |
| 03:09 | <Lachy> | they seem to be quite small though |
| 03:09 | <MikeSmith> | yeah |
| 03:09 | <MikeSmith> | true |
| 03:09 | <MikeSmith> | dunno why they don't make em bigger |
| 03:13 | <MikeSmith> | Lachy: http://www.eki.ee/letter/chardata.cgi?search=z+with+caron has some bigger ones |
| 03:13 | <MikeSmith> | but it is missing many |
| 03:13 | <MikeSmith> | plus they are GIFs |
| 03:15 | <Lachy> | MikeSmith, decodeunicode.org also has images available, but I'm not sure of the licences |
| 05:55 | <Hixie> | i just use the fonts |
| 05:56 | <Hixie> | after all if the user doesn't have an appropriate font installed, he probably won't use the character |
| 08:08 | <MikeSmith> | Hixie: are there issues with referencing RFC 5322 instead of RFC 2822? |
| 08:09 | <Hixie> | do we reference either? |
| 08:09 | <Hixie> | what are they? |
| 08:09 | <Hixie> | e-mail something or other? |
| 08:09 | <MikeSmith> | yeah, e-mail |
| 08:09 | <MikeSmith> | RFC 5322 is intended to obsolete 2822 |
| 08:10 | <MikeSmith> | http://tools.ietf.org/html/rfc5322#section-3.2.3 |
| 08:10 | <MikeSmith> | they seem to define "dot-atom" identically |
| 08:12 | <MikeSmith> | http://www.whatwg.org/specs/web-apps/current-work/#valid-e-mail-address |
| 08:12 | <MikeSmith> | you have: |
| 08:12 | <MikeSmith> | ""A valid e-mail address is a string that matches the production dot-atom "@" dot-atom where dot-atom is defined in RFC 2822 section 3.2.4, excluding the CFWS production everywhere. [RFC2822]" |
| 08:13 | <MikeSmith> | Hixie: do you recall why you didn't just spec it as needing to match the "addr-spec" production? |
| 08:14 | <MikeSmith> | ..with qualifications about CWFS and FWS |
| 08:14 | <MikeSmith> | that's what http://wiki.whatwg.org/wiki/MicrosyntaxDescriptions#email has |
| 08:15 | <MikeSmith> | and so I'd suspect that's what validator.nu is checking at least |
| 08:19 | <MikeSmith> | Hixie: production for "dtext" in RFC 5322 is more liberal than in RFC 2822 |
| 08:19 | <MikeSmith> | [[ |
| 08:20 | <MikeSmith> | dtext = %d33-90 / ; Printable US-ASCII |
| 08:20 | <MikeSmith> | %d94-126 / ; characters not including |
| 08:20 | <MikeSmith> | obs-dtext ; "[", "]", or "\" |
| 08:20 | <MikeSmith> | ]] |
| 08:20 | <Hixie> | addr-spec doesn't match user expectations |
| 08:21 | <MikeSmith> | OK |
| 08:21 | <Hixie> | e.g. iirc something like this matches addr-spec: "foo" (bar) @ foo.com |
| 08:21 | <Hixie> | meaning foo⊙fc |
| 08:21 | <MikeSmith> | I see |
| 08:22 | <Hixie> | and then you can start introducing escapes and all kinds of stuff |
| 08:22 | <MikeSmith> | yeah |
| 08:22 | <MikeSmith> | would be good to find out from hsivonen what precisely validator.nu checks for |
| 08:23 | <MikeSmith> | seems like it might need to be more strict, if it's to match the spec |
| 08:23 | <Hixie> | the spec recently changed relative to wf2 |
| 08:23 | <Hixie> | so he might not have updated to that |
| 08:23 | <MikeSmith> | ah, OK |
| 08:24 | <MikeSmith> | yeah, I think he's not updated a number of things that changed since the WF2 integration |
| 08:27 | <MikeSmith> | Hixie: anyway, fwiw, I think if you're just directly referencing dot-atom only there, I think you could safely reference RFC 5322 instead if you cared to |
| 08:28 | <Hixie> | k, send mail or file a bug or something |
| 08:28 | <Hixie> | but i don't plan to update references until CR |
| 08:28 | <Hixie> | or at least LC |
| 08:28 | <MikeSmith> | OK |
| 08:28 | <MikeSmith> | will just raise it in bugzilla for now |
| 10:17 | <MikeSmith> | Is there a document of some kind that mandates American spelling in international standards? Or is it just understood? |
| 10:20 | jgraham | hopes no such document exists |
| 10:20 | <hallvors> | (won't new standards be en-GB-hixie from now? :-p) |
| 10:22 | <MikeSmith> | hmm, for Internet standards at least -- RFCs and W3C Recs -- the observed practice is to use American spellings |
| 10:22 | <annevk5> | yeah, HTML5 is en-US-x-Hixie |
| 10:22 | <annevk5> | mostly because the W3C style guide or some such requires it |
| 10:23 | <MikeSmith> | ah |
| 10:23 | <MikeSmith> | guess I should actually read that |
| 10:23 | <MikeSmith> | I wonder if the IETF has a style guide |
| 10:29 | <MikeSmith> | http://www.w3.org/2001/06/manual/#Spelling |
| 10:35 | <annevk5> | zcorpan, free tonight? |
| 10:35 | annevk5 | will be in Linköping around 10:30PM or so |
| 10:36 | <zcorpan> | annevk5: maybe |
| 10:41 | <MikeSmith> | annevk5, zcorpan : I think David Storey may be there. If he run into him, tell him I said hi |
| 10:41 | <MikeSmith> | I know he's in Sweden, anyway |
| 10:49 | <annevk5> | MikeSmith, he's there already? interesting |
| 10:49 | <annevk5> | MikeSmith, I know I've a meeting with him tomorrow :) |
| 10:56 | jgraham | notes that he is also in Linköping :p |
| 10:59 | <annevk5> | jgraham, it's not like you have time, right? :p |
| 11:03 | <zcorpan> | hmm, http://simon.html5.org/html5-elements shows up blank in ie8 unless in compat view |
| 11:03 | <zcorpan> | i wonder why |
| 11:12 | <zcorpan> | double-colon syntax for pseudo-elements still doesn't work in ie8 |
| 11:13 | <zcorpan> | "From the 43 selectors 22 have passed, 1 are buggy and 20 are unsupported (Passed 349 out of 578 tests)" |
| 11:14 | <annevk5> | jgraham, your phone number is not on the intranet |
| 11:17 | <zcorpan> | most of the html5 spec is blank in compat view |
| 11:18 | <roc> | in compat view? really? |
| 11:19 | <zcorpan> | yeah |
| 11:19 | <roc> | that doesn't sound very compat |
| 11:19 | <annevk5> | http://twitter.com/diveintomark/status/1049747857 :) |
| 11:19 | <jgraham> | annevk5: |
| 11:19 | <jgraham> | Je n'ai pas de téléphone |
| 11:21 | <annevk5> | you realize you're learning the wrong language for Sweden right? :) |
| 11:21 | <zcorpan> | oops, acid3 made ie8 crash |
| 11:22 | <annevk5> | jgraham, I'm pretty sure that at some point (midnight prolly) we go to the pub close to the office |
| 11:22 | <jgraham> | annevk5: That sounds pretty late. |
| 11:22 | jgraham | likes sleeping |
| 11:22 | <annevk5> | guess we'll see you tomorrow then :) |
| 11:22 | <zcorpan> | depending on whether i click "yes" or "no" on all the various promts that i get on acid3, i get somewhere between 12/100, a crash, or 21/100 |
| 11:22 | annevk5 | will sleep on the train |
| 11:23 | <annevk5> | we arrive at 10:30PM and need to get some food at that point... so it would be relatively early |
| 11:24 | <Philip`> | zcorpan: html5-elements seems to work perfectly well for me in IE8 |
| 11:24 | <zcorpan> | oh actually, i just either get a crash or 21/100 |
| 11:24 | <zcorpan> | Philip`: in compat view? |
| 11:24 | jgraham | has to go grapple with the Swedish banking system now :( |
| 11:25 | <Philip`> | zcorpan: No, in normal view |
| 11:25 | <zcorpan> | Philip`: i get that too. try compat view |
| 11:25 | <zcorpan> | oh wait |
| 11:25 | <zcorpan> | that's weird |
| 11:26 | <zcorpan> | well it's blank for me in normal view |
| 11:26 | <zcorpan> | after the script has run |
| 11:26 | <Philip`> | Actually, it doesn't quite work - clicking the fragment links in the frame on the left doesn't scroll the frame on the right |
| 11:28 | <Philip`> | Oh, but clicking the fragment links in the frame on the right doesn't work either |
| 11:29 | <zcorpan> | Philip`: do you have the "automatically recover from page layout errors with compat view" setting checked? |
| 11:30 | <Philip`> | zcorpan: It looks like I do |
| 11:30 | <Philip`> | (so that must be the default) |
| 11:31 | <zcorpan> | apparently it renders html5-elements in compat view if that setting is checked |
| 11:31 | <zcorpan> | i wonder how it works |
| 11:33 | zcorpan | notices that 'border:thin solid' is 1px in normal view but 2px in compat view |
| 11:35 | <zcorpan> | typing javascript:'asdf' in the address field brings up ie6 with a dialog "Internet Explorer cannot download. Unidentified error. [ OK ]" |
| 11:37 | <zcorpan> | live dom viewer works again |
| 11:38 | <Lachy_> | zcorpan, Chris Wilson told me when I asked him about the double-colon pseudo-elements at TPAC that it was too late for support to be added to IE8 |
| 11:38 | <zcorpan> | i filed a bug about it during ie7b2 iirc |
| 11:39 | <Philip`> | zcorpan: Oh, that compat view thing sounds not confusing at all |
| 11:40 | <zcorpan> | Philip`: ? |
| 11:41 | <Philip`> | (I mean the automatically recovering from page layout errors with compat view thing) |
| 11:41 | <Philip`> | (And by "not confusing at all", I mean the exact opposite) |
| 11:41 | <zcorpan> | ok |
| 11:42 | <zcorpan> | if their goal was to make pages render in random ways depending on a bazillion factors (including user settings), they've succeeded |
| 11:44 | <Philip`> | Then they can simplify it all by just rendering every page in compat view |
| 11:44 | <zcorpan> | even when ie7 isn't used anymore, authors will have to test their pages against both normal view and compat view |
| 12:18 | <hallvors> | anyone here knows Gecko's ecma engine well? It seems to include elided properties when enumerating an array with for..in, I wonder if that is a known bug? |
| 12:28 | <hallvors> | I've looked stuff up in the spec, it seems to be a clear bug and IE/Opera/Safari all disagree with Firefox and do what the spec says. (by my reading of it anyway). Can't find it in bugzilla though. |
| 12:29 | <annevk5> | you might have better luck with this question at 10PM or so |
| 12:29 | <annevk5> | or on irc.mozilla.org :) |
| 12:32 | <hallvors> | good point about the time <:-) |
| 12:37 | zcorpan | is on a bug hunt for validator.w3.org's html5 integration |
| 12:38 | <zcorpan> | hsivonen: http://www.w3.org/Bugs/Public/show_bug.cgi?id=6301 |
| 16:19 | Philip` | tries the performance tests from http://blog.ianbicking.org/2008/03/30/python-html-parser-performance/ and gets very similar results except that modern html5lib is maybe 15-20% faster now |
| 16:47 | <Philip`> | Amazon S3 seems to have a peculiar approach to the namespacing of storage buckets |
| 16:47 | <Philip`> | in that it doesn't do any namespacing at all - you just have to try to pick a name that nobody else has used yet |
| 16:53 | <gsnedders> | /etc/passwd taken? |
| 16:57 | <Philip`> | gsnedders: No - it's limited to [a-z0-9][a-z0-9._-]{2,254} |
| 16:58 | <Philip`> | (Actually it's apparently more like (?!\d+\.\d+\.\d+\.\d+)[a-z0-9][a-z0-9._-]{2,254} but that detail doesn't seem to be explicitly specified) |
| 17:37 | <hsivonen> | one nice thing about SRT is that it's super-simple to map to a calback API, since the data maps to unstructured JS strings |
| 17:43 | <jgraham> | gsnedders, Hixie: http://pimpmyspec.net/ |
| 17:44 | <gsnedders> | jgraham: Bug reports and feature requests about the web ui shouldn't be! |
| 17:44 | <jgraham> | gsnedders: Yes they should :) |
| 17:44 | <jgraham> | :-p |
| 17:44 | <gsnedders> | jgraham: But I can't fix them! |
| 17:45 | <jgraham> | You can decide if they are important enough to bother me with :) |
| 17:47 | <jgraham> | gsnedders: I changed it since I guess that was a bug report :-p |
| 17:47 | <gsnedders> | jgraham: Just give my email there |
| 17:48 | <jgraham> | that means looking for your email address |
| 17:48 | <gsnedders> | geoffers⊙gc |
| 17:49 | <jgraham> | Better? |
| 17:49 | <gsnedders> | Yes |
| 17:50 | <gsnedders> | http://pimpmyspec.net/?url=http%3A%2F%2Fpimpmyspec.net%2F :P |
| 17:50 | jgraham | is leaving now |
| 17:53 | <Dashiva> | http://pimpmyspec.net/?url=http%3A%2F%2Fwhatwg.org%2Fhtml5 |
| 17:53 | <Dashiva> | :( |
| 17:56 | <gsnedders> | Dashiva: http://pimpmyspec.net/?url=http%3A%2F%2Fwww.whatwg.org%2Fspecs%2Fweb-apps%2Fcurrent-work%2Fsource-whatwg :P |
| 17:59 | <Dashiva> | gsnedders: pimp my bug report too |
| 18:00 | gsnedders | has no idea of the cause |
| 18:00 | <Dashiva> | Maybe it doesn't support redirects |
| 18:02 | gsnedders | has no idea of the cause of the bug |
| 18:04 | <gsnedders> | It works here, so bug is with jgraham and HTTP fetching |
| 19:10 | <hsivonen> | Do blackberries come with a RIM-developed browser? |
| 19:10 | hsivonen | is a bit surprised to learn about the -rim- CSS prefix |
| 19:12 | <gavin> | I think they do have their own browser, yeah |
| 19:12 | <gavin> | I've heard it's quite crappy |
| 19:17 | <smedero> | http://na.blackberry.com/eng/support/docs/subcategories/?userType=21&category=BlackBerry+Browser |
| 19:19 | <smedero> | judging from the Opera Mobile numbers they release every month, a lot of BlackBerry users are installing Opera Mini. |
| 20:15 | <Philip`> | Wow, Google stopped calling something "beta" |
| 20:15 | <Philip`> | http://googleblog.blogspot.com/2008/12/google-chrome-beta.html |
| 20:16 | <gsnedders> | no wai. |
| 20:47 | <Hixie> | what's with the low amount of mail these days |
| 20:48 | <takkaria> | you complaining? |
| 20:49 | <Hixie> | yes! |
| 20:49 | <Hixie> | where are the entertaining flames? |
| 20:49 | <Hixie> | where are the e-mails of people saying things i disagree with that spur me into replying? |
| 20:50 | <Hixie> | whenever i go more than 5 minutes without e-mail, i assume my mail delivery is broken |
| 20:53 | <zcorpan> | Hixie: what do you think about my thoughts on handling foreign content? |
| 20:53 | <Hixie> | which thoughts were yours again? |
| 20:53 | <Hixie> | (i briefly scanned the various e-mails on that thread and saved them for later) |
| 20:53 | <zcorpan> | making svg <script> a cdata element |
| 20:54 | <zcorpan> | http://lists.w3.org/Archives/Public/public-html/2008Dec/0134.html |
| 20:54 | <Hixie> | oh well |
| 20:55 | <Hixie> | like i told mike? hsivonen? i don't recall. but either way. like i said earlier: we can do many things that make svg-in-text/html simpler |
| 20:55 | <Hixie> | the problem is getting it past the svgwg |
| 20:55 | <Hixie> | their goals aren't compatible with making things simpler than they are now |
| 20:55 | <Hixie> | (assuming i've understood their goals correctly) |
| 20:56 | <Hixie> | (and ignoring the goals they say they have but that they haven't met in their own proposals) |
| 20:56 | <zcorpan> | i guess i should go back to ignore the issue and see how it folds |
| 20:57 | <Hixie> | that's my plan |
| 20:57 | <Hixie> | at least until i hear back from svg |
| 21:18 | <Hixie> | i think i will invent a new exception code for calling a script in a frozen script execution environment or a script freed after document.open() |
| 22:15 | <gsnedders> | Hixie: Should I try sending an email asking for the XML serialization to be removed? |
| 22:16 | <gsnedders> | Hixie: I'm sure that would get some traffic on the list. |
| 22:20 | <Hixie> | -_- |
| 22:20 | Hixie | finds himself in a hole he dug for himself without noticing |
| 22:21 | <Hixie> | well crap. |
| 22:21 | <jcranmer> | that you're in a hole |
| 22:21 | <jcranmer> | that you dug it |
| 22:21 | <jcranmer> | or that you didn't notice? |
| 22:21 | <jcranmer> | er, s/(?<!?)$/,/g |
| 22:22 | <jcranmer> | pretend that ? was a \?... :-( |
| 22:22 | <takkaria> | which one? :) |
| 22:22 | <jcranmer> | in the regex |
| 22:23 | <Hixie> | that i was digging it |
| 22:23 | <Hixie> | i was happily building some scaffolding |
| 22:23 | <Hixie> | when i discovered it had in fact been a mine's support structure |
| 22:58 | <Dorward> | Hi. I seem to recall that Google did a survey of webpages to see what constructs were commonly used as part of their HTML5 work. I can't find it though, can anyone give me a pointer? Thanks. |
| 23:00 | <hasather> | Dorward: http://code.google.com/webstats/index.html |
| 23:01 | <Dorward> | hsivonen: Fantastic. Thanks very much. |
| 23:02 | <hasather> | Dorward: there's also Opera's MAMA for similar stuff: http://dev.opera.com/articles/view/mama/ |
| 23:03 | <Dorward> | hasather: I've already checked that out - didn't have the data I was looking for :( |
| 23:06 | <Philip`> | Dorward: http://philip.html5.org/data.html has various bits of data but maybe nothing relevant |
| 23:06 | <Hixie> | woohoo! |
| 23:06 | Hixie | elevates his aforementioned mine support structure |
| 23:07 | <Hixie> | Dorward: what data are you looking for? |
| 23:08 | <Dorward> | Hixie: Proportion of documents with an XHTML Doctype that are valid/invalid and served as each of the acceptable mime types for XHTML. |
| 23:08 | <Hixie> | i don't think i published those numbers |
| 23:09 | <Dorward> | Hixie: It doesn't look like you did :) |
| 23:09 | <Hixie> | but iirc in one study i did application/xhtml was 0.0044%, xhtml as text/xml or application/xml wasn't visible at all, and xhtml was about 15% of all text/html |
| 23:09 | Philip` | only looked at a hundred thousand pages so he doesn't have enough non-text/html content to get any useful numbers |
| 23:09 | <Hixie> | (from memory, so those numbers are likely wrong, but they're in the right ballpark) |
| 23:09 | <Dorward> | Hixie: Thanks |
| 23:10 | <Dorward> | Hixie: May I quote you (with your proviso)? |
| 23:10 | <Hixie> | i used the namespace as the trigger though, not the doctype, iirc |
| 23:10 | <Hixie> | for detecting xhtml in text/html, i mean |
| 23:10 | <Dorward> | Yup |
| 23:10 | <Hixie> | sure, quote away |
| 23:10 | <Dorward> | Thanks |
| 23:10 | <Hixie> | but like i said, these numbers are likely somewhat bogus |
| 23:10 | <Hixie> | and it was a year or more ago now |
| 23:10 | <Dorward> | Indeed |
| 23:10 | <Dorward> | I doubt things will have changed a great deal since then |
| 23:10 | <Hixie> | so application/xhtml+xml is probably at least 0.0045% now! |
| 23:11 | <Hixie> | (might even be 0.2%, which i think i saw in another study i did, now that i think about it) |
| 23:11 | <Hixie> | (though i don't recall what that study was biased towards, so i'm not sure what that number was) |
| 23:11 | <Hixie> | either way application/xhtml+xml is far less common than, say, word documents |
| 23:11 | <Hixie> | at least in the google index |
| 23:12 | <Hixie> | of course it might be that the google index is biased against xhtml in some way |
| 23:12 | <Hixie> | though i'm not aware of such a bias |
| 23:12 | <Dorward> | Off the top of your head, do you know what Accept header GoogleBot outputs? |
| 23:12 | <Hixie> | (e.g. maybe sites that do content negotiation serve text/html to google, not application/xhtml+xml) |
| 23:12 | <Hixie> | not sure |
| 23:12 | <Hixie> | probably */* though |
| 23:13 | <Hixie> | Dorward: trying to win some argument somewhere? :-) |
| 23:13 | <Dorward> | That would fail to trip all those content negotiation systems that trigger on an explicit application/xhtml+xml |
| 23:14 | <Philip`> | Dorward: http://www.google.com/search?q=cache:www3.wind.ne.jp/hassii/env.cgi - looks like */* |
| 23:14 | <Dorward> | Hixie: Someone using one of the content negotiation systems with missing q support has exploded over www-validator. I'm just trying to address his points. At the moment is he demanding a system for content negotiation that is (a) easy (b) pure Apache conf (c) doesn't involve symlinks and (d) doesn't involve multiple files for each document |
| 23:14 | <Dorward> | Philip`: heh. Thanks. |
| 23:18 | <Hixie> | content negotiation was such a mistake |
| 23:18 | <Hixie> | content negotiation and the content-type header both in fact where imho design mistakes of http |
| 23:19 | <Hixie> | we should have had a 1:1 mapping of resources to representations, and we should have used well-defined magic sequences to recognise types, so there could be no content sniffing or content-type mislabelling |
| 23:19 | <Hixie> | but oh well |
| 23:19 | <Hixie> | too late for that i guess |
| 23:19 | <Hixie> | s/where/were/ |
| 23:30 | <Philip`> | Hixie: How would the use of magic sequences interact with legacy file formats that don't have any unique distinguishing sequences? |
| 23:31 | <Hixie> | like what? |
| 23:32 | <Dashiva> | plain text? |
| 23:33 | <Hixie> | plain text would be the default fallback format |
| 23:34 | <Dashiva> | So every time a new format appears, you get binary soup in your face while each web server is updated? |
| 23:34 | <Dashiva> | *web browser |
| 23:34 | <Dashiva> | Or maybe either would do |
| 23:34 | <Hixie> | web servers wouldn't ever have to change to support new types under this model |
| 23:35 | <Hixie> | web browsers could detect binary data and treat them as we treat unknown binary data now |
| 23:35 | <Hixie> | (using a well-defined mechanism for detecting binary data -- it would be the fallback detection step before falling back to plain text) |
| 23:36 | <takkaria> | instead of hooking in to support mime types, I guess plugins would hook in for certain binary sequences? |
| 23:37 | <Hixie> | yup |
| 23:37 | <Hixie> | similar to what file(1) does on unix, but with a normative specification and a defined detection order |