| 00:08 | <Philip`> | Hixie: <html class="split (index|chapter)"> is there now |
| 00:09 | <annevk2> | fwiw, this will cause pubrules issues down the road though I suppose the W3C edition can simply omit the classes altogether |
| 00:12 | <Philip`> | They don't like classes on <html>? |
| 00:37 | <AryehGregor> | http://www.w3.org/2005/07/pubrules?uimode=filter&uri= |
| 00:37 | <AryehGregor> | "All normative representations must either: |
| 00:37 | <AryehGregor> | validate as either HTML 4.x or as some version of XHTML that is a W3C Recommendation, or |
| 00:37 | <AryehGregor> | (for non-Recommendations) validate as XHTML+RDFa; see RDFa in XHTML: Syntax and Processing (Team Contacts please see the Communications Team to propose additional exceptions)." |
| 00:37 | <AryehGregor> | Guess that one's gonna have to be scrapped, huh? |
| 00:38 | <AryehGregor> | The front matter must appear at the beginning of the body of the document, within <div class="head">. |
| 00:39 | <AryehGregor> | <p><em>This section describes the status of this document at the time of its publication. Other documents may supersede this document. A list of current W3C publications and the latest revision of this technical report can be found in the <a href="http://www.w3.org/TR/">W3C technical reports index</a> at http://www.w3.org/TR/.</em></p>; |
| 00:40 | <AryehGregor> | That's not conforming HTML 5, it's almost certainly misuse of <em> by HTML 5 standards. |
| 00:40 | <AryehGregor> | . . . overall, I think these publication rules are going to have to be hacked a bunch anyway for HTML 5. :P |
| 00:47 | <Hixie> | Philip`: class="" is invalid in HTML4 |
| 00:47 | <Hixie> | (on <html>) |
| 00:51 | <AryehGregor> | You could put it on the <body> instead. |
| 00:51 | <AryehGregor> | Doesn't make much difference. |
| 00:51 | <AryehGregor> | I know that's valid, because MediaWiki puts a new class on the <body> every time someone's bored. |
| 00:51 | <AryehGregor> | Also, MediaWiki contributors are frequently bored. |
| 00:52 | <AryehGregor> | Wait, is the W3C version of the HTML 5 spec seriously going to be published as an HTML 4 document? |
| 00:53 | <Hixie> | the w3c isn't convinced that html5 is compatible enough, or something |
| 00:54 | <Hixie> | i dunno, i stopped listening around the time they said that it wasn't obvious that we should publish html5 as html5 |
| 00:54 | <Rik`> | that's a big vote of confidence for a standard |
| 00:55 | <Hixie> | it's more just an indication of how out of touch the w3c management is |
| 00:55 | <Hixie> | (with a couple of exceptions, who argued in favour of using html5) |
| 00:55 | <Hixie> | the whatwg version has used html5 for years |
| 00:58 | <Rik`> | eat your own dog food, they say |
| 00:59 | <AryehGregor> | In what way is it less compatible than HTML 4? |
| 00:59 | <Hixie> | beats me |
| 00:59 | <Hixie> | html4 used html4 since the earliest drafts |
| 01:00 | <Hixie> | it just seems like the w3c management is anti-html5 to me |
| 01:00 | <jcranmer> | NIH syndrome? |
| 01:02 | <Hixie> | could be |
| 01:02 | <Hixie> | dunno |
| 01:02 | <Hixie> | or just upset about the history |
| 01:02 | <Hixie> | or maybe they're just really that far out of touch |
| 01:02 | <Hixie> | who knows |
| 01:02 | <AryehGregor> | Well, they kind of had it forced on them against their will, I guess. Not likely that they'll completely change their attitude toward the direction HTML should head just because the browser vendors force their hand. |
| 01:02 | <Hixie> | and frankly who cares :-) |
| 01:03 | <AryehGregor> | I get the impression that W3C involvement here is basically political and nobody important in the WHATWG really cares too much what they think. |
| 01:03 | <Hixie> | i think the situation is a lot more subtle than that |
| 01:04 | <Hixie> | at least speaking for me, it's more that i don't care _who_ has feedback, so it makes no difference to me whether it's w3c staff who think something or some random web author |
| 01:04 | <Hixie> | since only the technical merits of the feedback matter |
| 01:07 | <karlcow> | it seems the first WD version of XHTML 1 has been published in the "in-development" xhtml 1 format http://www.w3.org/TR/1998/WD-html-in-xml-19981205/ |
| 01:08 | <AryehGregor> | But XHTML 2 and so on are all published in XHTML <= 1.1, right? |
| 01:08 | <AryehGregor> | XHTML 2, 1.2, whatever. The ones nobody ever used. |
| 01:09 | <Hixie> | XHTML 2 couldn't have been published in XHTML 2 because it wasn't backwards compatible |
| 01:09 | <karlcow> | XHTML 2 working drafts were published in xhtml 1.0 http://www.w3.org/TR/2002/WD-xhtml2-20020805/ |
| 01:09 | <Hixie> | (maybe they should have realised that it was doomed at that point, but oh well) |
| 01:10 | <karlcow> | and the last XHTML 2 WD was published using xhtml 1.0 too |
| 01:13 | <othermaciej> | AryehGregor: I care about helping W3C management make good decisions |
| 01:14 | <AryehGregor> | Okay, okay, I shouldn't have been so inflammatory, my bad. |
| 01:14 | <fupp> | ah, it's so nice to have full access to the dom tree from python with evaluated javascript and all |
| 01:14 | <AryehGregor> | HTML5 will proceed with or without the W3C, but probably better with than without. |
| 01:14 | <othermaciej> | I'm not sure the publication format of HTML5 is very important, but it does seem like it would make a better impression to publish HTML5 in HTML5 |
| 01:14 | <AryehGregor> | Although I have to wonder if maybe the W3C has just gotten too bureaucratic and closed to be a good web standards body anymore. |
| 01:15 | <AryehGregor> | The WHATWG procedure seems obviously better to me. Although I'm probably biased. |
| 01:15 | <AryehGregor> | Both because I'm not part of any W3C member organization, and because I'm one of those crazy wiki/open-source hippies. :) |
| 01:15 | <othermaciej> | creating the HTML WG, adopting HTML5, and ending the XHTML2 WG all seem like steps in the right direction |
| 01:16 | <Dashiva> | Having w3c on board helps with the patent situation, doesn't it? |
| 01:16 | <othermaciej> | in fact you can join the HTML WG freely without being part of any Member organization |
| 01:16 | <AryehGregor> | Yeah, but we're still talking about an organization where to subscribe to the bloody mailing list I had to find and fill out multiple forms and wait two weeks. |
| 01:16 | <AryehGregor> | (www-style allows open subscription, why not public-html?) |
| 01:17 | <AryehGregor> | Dashiva, is anyone particularly concerned about HTML-related patents? |
| 01:17 | <Hixie> | AryehGregor: www-style is not the csswg's list |
| 01:17 | <AryehGregor> | Sigh. |
| 01:17 | <AryehGregor> | Right. |
| 01:17 | <AryehGregor> | Hidden stuff. |
| 01:17 | <othermaciej> | the main advantages of working with the W3C are: (1) patent policy; (2) Microsoft is willing to participate; (3) the W3C still has a lot of prestige in the eyes of the general public |
| 01:17 | <Hixie> | (though they are using it a lot, which is good) |
| 01:17 | <AryehGregor> | Hmm, yeah, (2) is certainly a good point. |
| 01:17 | <AryehGregor> | Haven't seen any Microsoft employees posting to whatwg. |
| 01:17 | Hixie | isn't really convinced about (3), and (2) is because of (1), so it's really just (1), imho |
| 01:18 | <othermaciej> | AryehGregor: Apple had a bunch of HTML-related patents which we disclosed and agreed to license RF as part of the HTML WG publication and patent review process |
| 01:18 | <AryehGregor> | othermaciej, interesting. |
| 01:18 | <othermaciej> | including a patent on <canvas> |
| 01:18 | <AryehGregor> | Ah. |
| 01:18 | <AryehGregor> | That's valuable, then. |
| 01:18 | <AryehGregor> | (3) isn't important, if HTML5 is the future then people will have to accept it whether they like it or not. |
| 01:18 | <AryehGregor> | It's good that XHTML was shut down, though. |
| 01:19 | <AryehGregor> | So if only we abolished software patents, we wouldn't need the W3C anymore? :) |
| 01:19 | <othermaciej> | I'm not saying W3C is a perfect organization, but I believe it is moving in the right direction, and working with them seems worthwhile if it doesn't impose an unreasonable cost |
| 01:19 | <AryehGregor> | Sure. |
| 01:20 | <Hixie> | it's imposed a pretty high cost |
| 01:20 | <Hixie> | i don't know how much cost is considered reasonable |
| 01:20 | <Rik`> | I believe 3 is important for public organisations |
| 01:21 | <othermaciej> | I would also add that many specs that are very important but less visible than HTML5 are proceeding quite happily in the W3C without giant political fights |
| 01:21 | <AryehGregor> | Well, (3) doesn't matter in the long term, because HTML 5 will win in the long term and everyone will have to recognize it sooner or later. |
| 01:21 | <Rik`> | they might forbidden using HTML5 on their websites if it wasn't a W3C spec |
| 01:21 | <jcranmer> | AryehGregor: not all patents impacting software are software patents |
| 01:21 | <jcranmer> | H.264 is protected by DSP patents |
| 01:21 | <Hixie> | Rik`: is this the same organisations who use <embed> when youtube says that's how you embed video? |
| 01:21 | <Hixie> | Rik`: or heck, is this the same organisations who use window.open()? |
| 01:22 | <AryehGregor> | If a patent prohibits the distribution of some type of software, then as far as I'm concerned it's a software patent. I'm not sure what else "software patent" would mean, in fact. |
| 01:22 | <Hixie> | Rik`: or element.innerHTML? |
| 01:22 | <Hixie> | or showModalDialog()? |
| 01:22 | <Hixie> | or all the other things W3C never specced? |
| 01:22 | <othermaciej> | AryehGregor: I don't think win or lose is the right framing - clearly HTML5 is being implemented by browsers and adopted by content authors, and I don't think anyone wants or expects to reverse that |
| 01:22 | <jcranmer> | AryehGregor: `software patent' = the patent classification, according to the USPTO |
| 01:22 | <AryehGregor> | I think a lot of people want to reverse HTML5's success. |
| 01:22 | <othermaciej> | the question is a matter of how positive and peaceful the rollout can be |
| 01:23 | <Rik`> | Hixie: the only process to check that the policy is respected is the HTML validator |
| 01:23 | <othermaciej> | if there is an active resistance movement, that would be a bad thing for the Web platform, even if there is no hope of truly scuttling the effort |
| 01:23 | <AryehGregor> | jcranmer, hmm, didn't know there was such a thing. Well, then "software-related patent" or whatever term you'd like me to use. At least they shouldn't exist insofar as they prohibit the creation of computing standards, file formats, etc. |
| 01:23 | <Hixie> | Rik`: that says <embed> is bad, but people still use <embed> |
| 01:23 | <AryehGregor> | othermaciej, yes, I agree. I'm just being overly confrontational (not that there's anyone here I'm confronting). |
| 01:24 | <Rik`> | I haven't seen a lot of Youtube videos on such sites |
| 01:24 | <AryehGregor> | Anyway, I don't think most of this contradicts my original point, which is that getting HTML5 approved as a W3C standard is mostly just a political move. |
| 01:24 | <AryehGregor> | Hixie, even public organizations? Wasn't there a big breakthrough recently when the US government allowed its websites to use cookies? |
| 01:25 | <AryehGregor> | Just because everyone uses it doesn't mean some crazy government bureaucrats will allow the government to use it. :) |
| 01:25 | <Hixie> | AryehGregor: gov'ts have been embedding flash and videos and the like for plenty of time |
| 01:25 | <othermaciej> | standards are political |
| 01:25 | <AryehGregor> | I guess they don't really have any choice, if they want to post videos . . . |
| 01:25 | <Hixie> | AryehGregor: and i doubt that if the world moved forward, they would continue to stay in 1999 until such time as w3c stamped the spec |
| 01:26 | <othermaciej> | actually, standards are more like international relations than politics |
| 01:26 | <jcranmer> | from what I can tell, gov'ts are rather technologically incompetent |
| 01:26 | <jcranmer> | they just use <insert some standard utility here> |
| 01:27 | <othermaciej> | you can be a realist or an idealist, but ultimately there's no direct enforcement for defecting, so it has to be in everyone's interest to cooperate |
| 01:27 | <othermaciej> | or the effort fails |
| 01:27 | <othermaciej> | so you can compare the W3C to the UN |
| 01:28 | <Rik`> | othermaciej: do you have a blue helmet ? |
| 01:28 | <othermaciej> | it starts as a collection of its constituents, but eventually develops its own institutional interests which at times also need to be considered |
| 01:29 | <othermaciej> | Rik`: not a bad analogy - responsible for keeping the peace, but little power to enforce it and not much respect from the locals |
| 01:29 | <Dashiva> | So what's the w3c equivalent of the human rights' council coup? |
| 01:30 | <othermaciej> | the TAG |
| 01:30 | <othermaciej> | maybe WAI? I could make a case for either, but this is getting to be a silly metaphor |
| 01:31 | <karlcow> | I18N + TAG + WAI + QA (when it was still existing) |
| 01:32 | <AryehGregor> | Are we going to come up with an analogy to the fact that China and Russia are members of the HRC? :) |
| 01:33 | <karlcow> | http://beta.w3.org/Consortium/mission |
| 01:39 | <cardona507> | why oh why must they plug twitter at the bottom of that page?! |
| 02:04 | <takkaria> | what about the security council? |
| 02:05 | <AryehGregor> | For HTML5, the W3C is the General Assembly and the Steering Committee is the Security Council. :) |
| 06:28 | <othermaciej> | does anyone have a link handy to Mike Smith's draft or Lachy's authoring guide? |
| 06:39 | <hsivonen> | hmm. there's a CC logo on the ODRL front page. I wonder how ODRL relates to CC/REL |
| 06:43 | <Lachy> | othermaciej, http://dev.w3.org/html5/html-author/ |
| 06:48 | <hsivonen> | Lachy: s/contain/contains/ in "and the body, which contain all of the page’s content" |
| 06:49 | <hsivonen> | Lachy: also, when you mention RDF, you could also mention DocBook and say that HTML falls in between |
| 06:49 | <hsivonen> | after all, HTML is pretty light on semantics and, therefore, not exemplary of a semantic markup language |
| 06:50 | <hsivonen> | s/RDF/RTF/ |
| 08:44 | <lazni> | I can't checkout v.nu code, svn: Could not open the requested SVN filesystem |
| 08:46 | <hsivonen> | lazni: which svn URL? |
| 08:46 | <lazni> | svn co http://svn.versiondude.net/whattf/build/trunk/ build |
| 08:46 | <hsivonen> | lazni: that URL is no more |
| 08:46 | <hsivonen> | lazni: do I still have documentation out there that gives that URL? |
| 08:46 | <hsivonen> | lazni: see http://about.validator.nu/#src |
| 08:47 | <lazni> | the last time I go to http://about.validator.nu/#src it was there |
| 08:47 | <lazni> | sorry |
| 08:47 | <hsivonen> | lazni: CVSDude changed their system |
| 09:05 | <lazni> | line 281 of html5lib/python3/src/html5lib/treebuilders/etree_lxml.py uses python2 syntax print docStr |
| 09:13 | <annevk2> | are we porting html5lib to Python 3? |
| 09:15 | <Philip`> | Someone seemingly started |
| 09:16 | <Philip`> | Is there a better approach than maintaining two parallel versions of all the code? |
| 09:16 | <hsivonen> | Philip`: writing a translator that converts one into the other |
| 09:16 | <hsivonen> | language to language compilation FTW! |
| 09:17 | <annevk2> | I'm not sure that mantra applies to everyone :) |
| 09:17 | <Philip`> | Python already has a tool to do that, but I don't know if it's perfect or if some manual fixup is needed on the output |
| 09:19 | <hsivonen> | does Python 3 have a switch statement yet? |
| 09:19 | <annevk2> | Philip`, it's not perfect at all, but does quite a bit of work; see diveintopython3 |
| 09:20 | <Philip`> | hsivonen: It has if/elif - what more do you want? |
| 09:21 | <hsivonen> | Philip`: what are the perf characteristics of an if/elif chain compared to switch in C or Java? |
| 09:22 | <annevk2> | hsivonen, I don't think so: http://www.python.org/dev/peps/pep-3103/ |
| 09:22 | <hsivonen> | at least when the switch has a contiguous range of integer cases, switch compiler more nicely |
| 09:22 | <Philip`> | Depends on the compiler, since surely it could optimise "if x == 1: ... elif x == 2: ..." without requiring the programmer to use special syntax |
| 09:22 | <Philip`> | (I'm sure it doesn't, but it could) |
| 09:23 | hsivonen | wonders if non-contiguous switch compiler to anything nicer than if-else in GCC |
| 09:23 | <jgraham> | We have a python3 branch of html5lib because I hacked one together one weekend |
| 09:23 | <hsivonen> | s/compiler/compiles/ |
| 09:23 | <Philip`> | You can just use an array of functions in Python if you want contiguous indexed switching |
| 09:23 | <jgraham> | I guess if there is a bug in it like that it is because a) that is a file I didn't get to yet or b) something got checked in on the wrong file |
| 09:24 | <jgraham> | OTOH the Python3 branch is not really maintained yet |
| 09:24 | <lazni> | you didn't use 2to3? |
| 09:24 | <hsivonen> | if you have functions, you can't use the fall through tricks you can with switch cases |
| 09:24 | <jgraham> | I /did/ use 2to3 but it's not magic |
| 09:24 | lazni | thought 2to3 knows well about print() |
| 09:25 | <jgraham> | It will replace the print statements but it can't guess where you need byte arrays and where you need strings, for example |
| 09:26 | <Hixie> | gah, there's tons of interference on my local network |
| 09:26 | <jgraham> | (like said that file may never have been converted, or that print statement may have slipped in by accident later on) |
| 09:26 | <lazni> | oh ok |
| 09:26 | <Hixie> | i should not be getting 35ms pings to a box i can throw rocks at in less time than 35ms |
| 09:27 | jgraham | suspects that isn't true |
| 09:27 | <Ox032F> | hello, is there an easy way to determine (serverside wise would be better, but client side also ok) whether the client supports the html5 audio-tag? |
| 09:28 | <Hixie> | you shouldn't need to, generally speaking -- just put whatever your fallback is inside the <audio> element |
| 09:28 | <Ox032F> | ah cool |
| 09:29 | <Philip`> | hsivonen: In the non-contiguous case, it looks like GCC sometimes does a normal table jump (if you've just got a few holes, which it makes jump to code that does nothing) and sometimes does a kind of binary-tree lookup (though not really binary, but anyway it's not just a linear search) |
| 09:29 | <hsivonen> | I get tons of interference with the proprietary wireless dongle of my home mouse. I'm very annoyed at Logitech's failure to use bluetooth |
| 09:30 | <Philip`> | and I guess sometimes it will just do a linear search if it thinks that's fastest |
| 09:30 | <hsivonen> | at least Macs have decent bluetooth antennas |
| 09:30 | <hsivonen> | Philip`: cool |
| 09:30 | <hsivonen> | Philip`: thanks |
| 09:30 | <Hixie> | i wonder what this interference is coming from |
| 09:30 | <Philip`> | (I mean "looks like" in terms of the assembly output of a simple test program, and "guess" in terms of having no evidence) |
| 09:30 | <hsivonen> | good to know my use of switch isn't all in vein |
| 09:30 | <Hixie> | it's been going on longer than a microwave would, so it's not the neighbours microwaving something |
| 09:31 | <Philip`> | Intravenous switch? |
| 09:31 | <Hixie> | there are about 25 wifi networks in range, which probably doesn't help |
| 09:31 | <hsivonen> | s/vein/vain/ |
| 09:32 | <Philip`> | Hixie: Try logging in to the others' router configuration interfaces (with a compatible browser) and then switch them all off |
| 09:32 | <lazni> | or juggle the channels |
| 09:32 | <Philip`> | or change their channel assignments to leave a nice big gap for you |
| 09:33 | <lazni> | WRT54G password is linksys:admin |
| 09:34 | <boblet> | Hixie: consistency note; 4.6.25 Usage summary ruby example doesn’t include closing </rt> and </rp>, and #the-rp-element example does (no big deal, but just in case) |
| 09:34 | <Hixie> | boblet: intentional |
| 09:34 | <Hixie> | boblet: the examples aren't supposed to be consistent |
| 09:36 | <boblet> | Hixie: actually I was happy to see the usage summary example, b/c all those elements in a small space was a little much last time I marked some up (XHTML-goggles) |
| 09:36 | <Hixie> | ok 1.3s to the router, which is literaly within kicking distance, is ridiculous |
| 09:36 | hsivonen | recommends gigabit ethernet |
| 09:37 | <Hixie> | i am indeed tempted to get out the ethernet cable |
| 09:37 | <Hixie> | (i'm using a laptop, the proximity to the router is coincidental) |
| 09:37 | <Philip`> | (By the way, by "a kind of binary tree lookup" I don't mean there's an actual tree data structure - it's just code, kind of like "if (x < 10) { if (x < 5) ... else ... } else if (x < 20) { ... }", maybe with a branch for each contiguous section or something) |
| 09:38 | <hsivonen> | speaking of routers, the UI for setting up a VPN between two Linksys boxes is ridiculously complex |
| 09:38 | hsivonen | wonders if the UI is designed for the people who spec the protocols at the IETF |
| 09:38 | annevk2 | is glad he hardly has issues |
| 09:39 | <annevk2> | my six or so devices can all happily connect |
| 09:39 | <boblet> | Lachy: if you’re there I’d like to ask you about HTML5Doctors section article; “typically with a heading” is quite different to the “With very few exceptions, section should not be used if there is no natural heading for it” |
| 09:41 | <boblet> | that defn wouldn’t cover 4.6.26 Footnotes “longer annotation” example (section containing only <p>) |
| 09:42 | Hixie | accidentally DNS spoofs himself and gets a MITM message from ssh |
| 09:42 | Hixie | registers his ethernet address in DNS... |
| 09:47 | <Lachy> | boblet, how are they very different? |
| 09:48 | <boblet> | Lachy: level of strictness |
| 09:49 | <Lachy> | anyway, the point of the article was to try and discourage the common misuse of section as merely a wrapper |
| 09:49 | <boblet> | Lachy: eg I’ve got previous/next page links in a section (related content), but it has no heading. This seems correct for “typically”, but probably not for “should not” |
| 09:50 | <boblet> | Lachy: yeah I wondered if the language was stronger (well, to me) than the spec for that reason |
| 09:50 | <Lachy> | <footer> might be more appropriate for the prev/next links |
| 09:51 | <boblet> | Lachy: Since the article doesn’t list any, could you give an example or two of ok exceptions? |
| 09:52 | <boblet> | Lachy: it’s a <section> in the page <footer> :) |
| 09:53 | <Lachy> | web applications generally have more use for headingless sections |
| 09:53 | <Lachy> | the gmail inbox was mentioned in here as one such example recently |
| 09:55 | <Lachy> | or perhaps the center column of facebook where it lists all your friends statuses |
| 09:55 | <boblet> | Lachy: so would eg the typical <div id="main"> and <div id="sidebar"> be appropriate as <section>s? (assuming sidebar contains disparate content so <aside> isn’t appropriate) |
| 09:56 | <boblet> | on one hand they’re layout hooks, but otoh they are logical chunks of related content |
| 09:56 | <jgraham> | I'm not sure that gmail is a good example |
| 09:56 | <Lachy> | <div id="main"> shoud generally keep using <div> if it's just used as a wrapper to separate it from the header and footer |
| 09:56 | <jgraham> | I think if I were using AT I would want headings on each section so that I could quickly navigate between different parts of the UI |
| 09:57 | <boblet> | wow, ok |
| 09:57 | <jgraham> | (the could be hidden using CSS ofc) |
| 09:57 | <Lachy> | <aside> is for sidebar, unless we introduce <sidebar> as was discussed recently |
| 09:58 | <boblet> | jgraham: ok, that’s kinda the rule of thumb I’m working to after reading the HTML5Doctors article |
| 09:59 | <boblet> | Lachy: <aside> couldn’t include non-related content tho (list of site-wide popular links etc), right? |
| 10:00 | <Lachy> | why not? |
| 10:00 | <Hixie> | wouldn't side-wide popular links be <nav>? |
| 10:00 | <Ox032F> | Hixie: you said I should just put the fallback code inside the audio-tag, but <audio id='sound'><object id='sound'></object></audio> won't give me access by id if the audio-tag is not avaiable, so I guess I have to give different ids and test via javascript which is usable!? If the audio-tag is avaible would the object be avaiable by id? |
| 10:01 | <Hixie> | Ox032F: if you're using javascript, you'll have to test to see if the element is supported anyway, since it's unlikely that your plugin will have the same API as the <audio> element |
| 10:01 | <Lachy> | boblet, what do you mean by "site-wide popular links"? |
| 10:01 | <Ox032F> | thats not a problem, I'm testing for Play(), play() and DoPlay() already |
| 10:01 | <Lachy> | I thought maybe it was a blog roll or something |
| 10:03 | <boblet> | Lachy: I perceive <aside> should only contain tangental info that’s relevant to the <article>/<section>. Weblogs often have a grabbag of random stuff on every page’s sidebar; imported twitter posts, site search, most popular articles, recent site-wide comments etc |
| 10:04 | <boblet> | Lachy: I would have thought these aren’t relevant to the <article>, so shouldn’t be in an <aside> …? |
| 10:05 | <Lachy> | seems like it would go in the body's aside. <body><header/><div id=main/> <aside>[all that stuff in here]</aside></body> |
| 10:08 | <boblet> | Lachy: aah indeed. d’oh |
| 10:11 | <boblet> | I was gonna comment on the woork HTML5 cheatsheet mistakes: http://woork.blogspot.com/2009/09/html-5-visual-cheat-sheet-by-woork.html maybe I shouldn’t :x |
| 10:11 | <boblet> | angry? how about :| |
| 10:11 | <annevk2> | Ox032F, you should not use duplicate IDs |
| 10:11 | <annevk2> | Ox032F, 'cause both end up in the DOM |
| 10:11 | <Ox032F> | ok |
| 10:12 | <annevk2> | Ox032F, also, if the <object> plug-in automatically starts playing you have a problem as it will do so in new browsers too |
| 10:12 | <Hixie> | only if they don't implement the spec properly |
| 10:12 | <Hixie> | the spec says <object> inside <video>/<audio> doesn't instantiate |
| 10:12 | <Ox032F> | shouldn't <parameter autoplay='false'> prevent that? |
| 10:16 | <annevk2> | Hixie, ah, I forgot it changed |
| 10:18 | <hsivonen> | apparently all blog commenters aren't OK with the new no-space spec titling :-( |
| 10:18 | <Hixie> | i cannot _believe_ how much time has been wasted on the presence or existence of that space |
| 10:18 | <Hixie> | if i knew zeldman's request was going to cause this much pain, i'd have said no |
| 10:22 | <virtuelv> | Hixie: that's the danger of announcing that a bikeshed is being built in the back yard |
| 10:22 | <hsivonen> | virtuelv: I tried to announce it had been built and painted already. |
| 10:23 | <Hixie> | i don't really understand why anyone cares |
| 10:24 | <hsivonen> | Hixie: trying to make "HTML 5" and "HTML5" mean different things was too obscure |
| 10:24 | <Philip`> | People who care about specifications are likely to care about accuracy and preciseness and getting things exactly right, and caring about all the little details |
| 10:24 | <hsivonen> | Hixie: also, people want to know what's "right" |
| 10:24 | <Hixie> | hsivonen: when i said they meant different things i figured people would take it as a ridiculous joke |
| 10:24 | <Philip`> | Try a different community if you want people who don't care about this stuff :-) |
| 10:25 | <Philip`> | (Please remove some redundant uses of "care" in the previous sentences) |
| 10:25 | <hsivonen> | Hixie: you are in the business of writing *precise* specs, so that kind of things don't work as jokes |
| 10:25 | <Philip`> | Hixie: It doesn't seem any more ridiculous than e.g. Perl and perl meaning different things |
| 10:25 | <Hixie> | "Perl" and "perl" mean different things? |
| 10:25 | <Hixie> | hsivonen: apparently |
| 10:26 | <hsivonen> | Hixie: Perl means the language and perl means the implementation |
| 10:26 | <Philip`> | Perl is the language, perl is the standard implementation |
| 10:26 | <Hixie> | :-/ |
| 10:26 | <Philip`> | and PERL means you're stupid and can be ignored and/or ridiculed because you can't even capitalise the name correctly |
| 10:27 | <hober> | and perl is the only thing that can parse Perl, IIRC. |
| 10:28 | <Lachy> | boblet, that HTML5 visual cheat sheet includes elements that were dropped a long time ago |
| 10:28 | hsivonen | guesses Perl is undecidable like C++ |
| 10:28 | <Lachy> | like <datatemplate>, <rule> and <nest> |
| 10:28 | <boblet> | Lachy: apart from adactio’s I haven’t seen a cheatsheet that wasn’t riddled with errors :/ |
| 10:28 | <Philip`> | hober: http://search.cpan.org/~adamk/PPI-1.206/lib/PPI.pm apparently does quite a good job at parsing Perl independently of perl |
| 10:29 | <boblet> | Lachy: writing a comment now—will add those |
| 10:29 | <boblet> | (to the list of mistakes) |
| 10:29 | <Philip`> | hsivonen: What do you mean by "undecidable" in this context? |
| 10:29 | <Philip`> | For C++, compiling source into an executable is undecidable |
| 10:29 | <Philip`> | but Perl doesn't really compile like that |
| 10:30 | <Lachy> | boblet, I'm guessing he's working from some old draft or somethign |
| 10:30 | <annevk2> | his source is stated |
| 10:30 | <Lachy> | otherwise, I can't figure out how he would have ended up with those long gone elements |
| 10:30 | <annevk2> | it is w3schools |
| 10:30 | <hsivonen> | Philip`: that you can't build an AST from a source file without auxiliary information |
| 10:30 | <boblet> | Lachy: maybe the W3Schools cheatsheet? :D |
| 10:30 | <boblet> | hah, kabutta |
| 10:30 | <lazni> | http://ovh.dl.sourceforge.net/sourceforge/junit/junit-4.4.jar times out, http://ncu.dl.sourceforge.net/project/junit/junit/4.4/junit-4.4.jar works |
| 10:31 | <hsivonen> | lazni: could be a temporary problem |
| 10:31 | <hsivonen> | unless ovh has stopped being a sf.net mirror |
| 10:31 | <lazni> | I think you can use the general link instead of hardcoding the mirror |
| 10:32 | <lazni> | I tried nchc and it directed me to http://downloads.sourceforge.net/sourceforge/junit/junit-4.4.jar?download&failedmirror=nchc.dl.sourceforge.net |
| 10:32 | <Lachy> | ah, well, that sucks. No-one should use W3schools as a source for anything :-( |
| 10:32 | <hsivonen> | lazni: do they use sane redirects instead of serving an HTML file form the general link? |
| 10:32 | <zcorpan_> | Ox032F: set the id on the <audio>, test support with if ('play' in audio) |
| 10:32 | <Philip`> | hsivonen: In Perl you can write 'BEGIN { undecidable_function() } syntax error' so it seems boringly true that you can't parse the whole file decidably |
| 10:32 | <Hixie> | new <aside> text is up |
| 10:32 | <Hixie> | with a new example |
| 10:32 | <zcorpan_> | Ox032F: else use audio.firstChild (the object) |
| 10:32 | <lazni> | it looks sane, I used wget nchc.dl... |
| 10:32 | <Philip`> | At least C++ forces you to do interesting crazy template stuff before it's impossible to compile |
| 10:32 | <boblet> | actually, is there a definitive list of HTML5 elements, something like http://www.w3schools.com/tags/html5.asp but actually correct? |
| 10:33 | <annevk2> | boblet, http://simon.html5.org/html5-elements is maintained |
| 10:33 | <hsivonen> | lazni: OK. thanks. I guess I need to recheck the sf.net mirroring stuff |
| 10:33 | <hsivonen> | is it a *cheat* sheet if it's maintained? :-) |
| 10:33 | <boblet> | annevk2: great—that’s actually what I’d been using. nice to know |
| 10:33 | <Hixie> | zcorpan_: btw, any chance you could make your script generate http://www.whatwg.org/specs/web-apps/current-work/#index ? |
| 10:33 | <othermaciej> | it's an honesty sheet |
| 10:33 | <jgraham> | Maybe the time people are spending debating the " " could be better used launching a DOS attack on w3schools |
| 10:34 | <boblet> | lol |
| 10:34 | <zcorpan_> | Hixie: i don't have a script for it |
| 10:34 | <Hixie> | ah, bummer |
| 10:35 | <zcorpan_> | keeping it up-to-date manually helps me find mistakes in the spec :) |
| 10:35 | <Lachy> | Hixie, I have some scripts that could be used for those |
| 10:35 | <Hixie> | zcorpan_: :-) |
| 10:35 | <Ox032F> | zcorpan_: yes, I'm doing something like that now, but I set a seperate id for the object, to have less correlation between html and js, thx |
| 10:35 | <Hixie> | Lachy: that would rock |
| 10:35 | <Lachy> | well, some that are incomplete, but I'd written some for my HTML5 Reference |
| 10:36 | <Hixie> | Lachy: i'm happy to adjust the markup if it would help |
| 10:36 | <Lachy> | I doubt it would take much to modify them to output what you need |
| 10:36 | <annevk2> | Hixie, the current requirements, especially the asterisk ones, are hard to automate |
| 10:36 | <Hixie> | yes |
| 10:36 | <Hixie> | i expect there would be some manual input |
| 10:37 | <Hixie> | e.g. a configuration file, or some annotations in the html5 spec |
| 10:37 | <annevk2> | Hixie, it would also be nice to have a really short description for each element, similar to what Mike has |
| 10:37 | <Lachy> | Hixie, I will prrioritise those scripts, since I need to have the "List of elements" and "List of attributes" tables for the HTML5 reference too |
| 10:38 | <Hixie> | Lachy: cool |
| 10:38 | <Hixie> | annevk2: good point, let me add a column for that |
| 10:38 | <Lachy> | but they're all here if you want to do it yourself http://dev.w3.org/cvsweb/html5/html-author/utils/ |
| 10:38 | <Hixie> | i'm going to prioritise getting feedback to zero |
| 10:39 | zcorpan_ | is going to prioritize keeping feedback above zero |
| 10:39 | <Hixie> | :-/ |
| 10:39 | <zcorpan_> | j/k :-) |
| 10:40 | jgraham | wonders if anyone else had the cartoon "Captain Planet" when they were small |
| 10:40 | <Hixie> | :-) |
| 10:40 | <hsivonen> | http://www.whatwg.org/issues/data.html is looking awesome |
| 10:40 | <Lachy> | jgraham, I remember that :-) |
| 10:40 | <jgraham> | IIRC the theme tune had something like "Captain Planet, he's a hero, gonna get pollution down to zero" |
| 10:40 | <jgraham> | So now I imagine Hixie in a kind of superhero role... |
| 10:40 | <Hixie> | :-/ |
| 10:41 | <Hixie> | that's the superfriends |
| 10:41 | <Hixie> | flying around in capes |
| 10:41 | <Philip`> | Wikipedia says "Captain Planet he's our hero" |
| 10:41 | <jgraham> | They could be your comedy sidekicks or something |
| 10:41 | <Philip`> | but I remember hearing it as more like "a" |
| 10:41 | <hsivonen> | jgraham: Hixie doesn't do capes |
| 10:41 | <Hixie> | (apparently their super abilities don't include sending feedback, though) |
| 10:42 | <Philip`> | "Earth, Fire, Wind, Water, Heart... Go Planet" - that doesn't seem to quite match the original Aristotle |
| 10:42 | <jgraham> | hsivonen: I thought he did |
| 10:43 | <Hixie> | no capes! |
| 10:43 | <Hixie> | have you not seen the incredibles! |
| 10:43 | <jgraham> | (specifically I thought he did LARP which seems like it could involve capes if the need arose) |
| 10:43 | <MikeSmith> | Hixie: feel free to use verbatim any of the short descriptions from my draft that you want to |
| 10:43 | <Hixie> | MikeSmith: thanks |
| 10:44 | <Hixie> | MikeSmith: so far i've only got one line per table, i haven't started generating the tables automatically yet |
| 10:44 | <Hixie> | one day |
| 10:44 | <MikeSmith> | OK |
| 10:45 | MikeSmith | would rather have a pimp cane than a cape |
| 10:46 | <Philip`> | What would your super ability be? |
| 10:46 | <MikeSmith> | Philip`: my ability would be super macking |
| 10:47 | <MikeSmith> | combined with an exceptional ability to be able to make a bong out of any available materials |
| 10:48 | <MikeSmith> | wait, I already have both of those abilities |
| 10:49 | MikeSmith | tries to think of other useful abilities |
| 10:50 | <MikeSmith> | Philip`: X-ray vision |
| 10:50 | <MikeSmith> | or better yet, XXX-ray vision |
| 10:50 | <Hixie> | MikeSmith: btw, i was informed that the description of the <ruby> example we came up with was wrong. |
| 10:50 | <Hixie> | MikeSmith: something about it not being kanji after all, but being hiragana |
| 10:51 | <Hixie> | i was also told it was furigana |
| 10:52 | <Hixie> | MikeSmith: http://www.whatwg.org/specs/web-apps/current-work/#the-ruby-element |
| 10:52 | <MikeSmith> | Hixie: yeah, I saw that. I guess "furigana reading" is probably the most precise way to describe it. |
| 10:52 | <Hixie> | MikeSmith: not hiragana? |
| 10:52 | <hsivonen> | can the reading be furigana? |
| 10:53 | hsivonen | thought the annotation itself was typographically furigana |
| 10:53 | <hsivonen> | but "reading"? |
| 10:53 | <MikeSmith> | well, I think I originally suggested "kanji reading" (meaning the reading of the kanji, not that the reading itself was kanji) |
| 10:54 | <MikeSmith> | but people pointed out that was ambiguous |
| 10:54 | <Hixie> | what is the reading itself in? |
| 10:54 | <MikeSmith> | I guess "reading in furigana" would be the least ambiguous |
| 10:54 | <MikeSmith> | the reading you have in there now is hiragana |
| 10:54 | <Hixie> | so wouldn't "reading in hiragana" be better? |
| 10:54 | <Hixie> | furigana is the name of the typographical tool, right? not the letters? |
| 10:54 | <MikeSmith> | it's more generally furigana |
| 10:55 | <MikeSmith> | because you might have another case where it's in katakana |
| 10:55 | <Hixie> | i'm talking specifically about the case in the example here |
| 10:55 | <Hixie> | not anything generic |
| 10:55 | <MikeSmith> | specifically, it's hiragana |
| 10:55 | <Hixie> | ok |
| 10:57 | <MikeSmith> | Hixie: I think to be as unambiguous as possible, it should be "annotated with its reading in hiragana" |
| 10:57 | <Hixie> | k |
| 11:02 | <zcorpan_> | Hixie: http://annevankesteren.nl/2006/01/fallback for the fallback discussion |
| 11:03 | <Hixie> | what about it? |
| 11:03 | <zcorpan_> | you could define two different terms for fallback |
| 11:03 | <zcorpan_> | one for the kind of fallback that img and object use |
| 11:03 | <zcorpan_> | and one for the kind that iframe and audio use |
| 11:04 | <zcorpan_> | i think <applet> actually supports both kinds |
| 11:05 | <Hixie> | i'm trying to get rid of the term altogether actually |
| 11:05 | <zcorpan_> | ok |
| 11:05 | <Hixie> | not trying very hard |
| 11:05 | <Hixie> | but hard enough not to add anything new along those lines :-) |
| 11:16 | <annevk2> | reading that again the terms content fallback and element fallback seems nice |
| 11:16 | <annevk2> | s/seems/seem/ |
| 11:26 | <hsivonen> | http://stackoverflow.com/questions/1417683/add-html-5-doctype-to-xdocument-net |
| 11:40 | <hsivonen> | Hixie: when the UA supports no crypto for <keygen>, wouldn't it make sense to require the element to be an HTMLUnknownElement in that case? |
| 11:41 | <zcorpan_> | how are authors supposed to check for support for keygen? |
| 11:41 | <zcorpan_> | or are they expected to test support for ActiveX first? |
| 11:42 | <erlehmann_> | lol |
| 11:42 | gsnedders | would prefer we didn't have HTMLUnknownElement |
| 11:42 | <zcorpan_> | what do they do today? |
| 11:42 | <Hixie> | hsivonen: see my e-mail |
| 11:43 | <Ox032F> | doesn't firefox recognize the 'loop' and 'autoplay' parameters in audio-tag? |
| 11:44 | <hsivonen> | Hixie: is it actually desirable to make it act like a form control when it has no crypto? |
| 11:44 | <jgraham_> | gsnedders: Do you have any good reason for that wish? |
| 11:44 | <Hixie> | it's desierable to not have random other parts of scripts fail, yes |
| 11:44 | <gsnedders> | jgraham_: It just seems so ugly |
| 11:44 | <jgraham_> | gsnedders: Why? |
| 11:45 | <hsivonen> | Hixie: ok. let's see if IE9 implements those parts |
| 11:45 | <annevk2> | jgraham_, it ought to be just HTMLElement |
| 11:45 | <jgraham_> | annevk2: It's not obvious to me why it makes a big difference |
| 11:45 | <gsnedders> | jgraham_: Adding new HTML elements it will lead to compat issues |
| 11:45 | <annevk2> | jgraham_, HTMLUnknownElement is some Firefox invention that somehow started affecting other browsers much like crap IE once introduced |
| 11:46 | <jgraham_> | annevk2: I know |
| 11:46 | <annevk2> | it doesn't make much of a difference, it's just annoying |
| 11:46 | <gsnedders> | jgraham_: Things change the prototype of HTMLUnknownElement and rely upon that working, so if that element becomes known, then we have compat issues |
| 11:46 | <zcorpan_> | Ox032F: autoplay seems to work for me, but not loop |
| 11:46 | <annevk2> | wihch reminds me, we still haven't dealt with XMLDocument vs Document vs etc. |
| 11:46 | <annevk2> | at the spec-level |
| 11:47 | <Hixie> | hsivonen: i imagine if we test suite it, some code monkey will be assigned the job of fixing the minor things and they'll just end up coding it without knowing why, and it'll be done |
| 11:47 | <jgraham_> | gsnedders: Things that change the prototype of DOM objects in general and rather underspecified... |
| 11:47 | <hsivonen> | Hixie: fair enough |
| 11:47 | <jgraham_> | s/and/are/ |
| 11:47 | <jgraham_> | gsnedders: But presumably we would have had compat issues anyway |
| 11:48 | <Ox032F> | zcorpan_: well, the problem is I can't disable autoplay, and when I start the sound with javascript it seems to loop, but maybe thats even a problem with the object tag inside the audio, I can't tell so far |
| 11:48 | <jgraham_> | e.g. if I was using a made-up <video> tag to signify a place where a script should insert a youtube video, that would break when browsers introduced native <video> |
| 11:51 | <jgraham_> | (in fact having HTMLUnknownElement might somehow be better because you could use it to detect unexpected native support for a tag you were abusing) |
| 11:53 | <zcorpan_> | Ox032F: by default, it doesn't autoplay and it doesn't loop |
| 11:53 | <zcorpan_> | Ox032F: they are boolean attributes, so trying to set autoplay=false will actually cause it to autoplay (attribute present means "true") |
| 11:54 | <Ox032F> | zcorpan_: ok, but thats not true for parameter-tags in object, is it? |
| 11:54 | <zcorpan_> | Ox032F: oh, i thought you were talking about <audio> |
| 11:55 | <Ox032F> | I was |
| 11:55 | <zcorpan_> | ok |
| 11:55 | <zcorpan_> | no, <param>s are different |
| 11:55 | <zcorpan_> | and depend on the plugin |
| 11:56 | <Ox032F> | ok, I'll do another test |
| 11:59 | <karlcow> | http://blog.isotoma.com/2009/09/textual-log-analysis-using-python/ |
| 12:00 | <Ox032F> | ah, do you know what the stop-command is for the audio-tag? |
| 12:00 | <karlcow> | > text statistics produced by this channel, who has what reading age, and how much they’ve talked in comparison to other people. |
| 12:01 | <zcorpan_> | Ox032F: pause(); currentTime = startTime; |
| 12:01 | hsivonen | wonders if Opera did user testing with real Web content to determine if this helps users: http://www.iheni.com/londesc-support-opera-1010/ |
| 12:08 | <Hixie> | what's up with mozilla bugzilla and moznet |
| 12:09 | <annevk2> | hsivonen, I don't think so, but you can probably ask her in a comment |
| 12:10 | <annevk2> | I'm not quite sure why we implemented it, but I wasn't really involved in the process either |
| 12:11 | <hsivonen> | I'm a bit surprised that an implementation is announced after the WAI consensus said it's OK to obsolete longdesc. |
| 12:16 | <Ox032F> | could it be that setting currentTime starts play? |
| 12:17 | <Hixie> | shouldn't do |
| 12:17 | <Hixie> | might though |
| 12:17 | <Hixie> | if there's a bug |
| 12:18 | <Dashiva> | hsivonen: Maybe it's to let people see how much bogus there is? |
| 12:19 | <Ox032F> | *grml* can't access bugzilla.mozilla.org |
| 12:20 | <roc> | me neither |
| 12:21 | <hsivonen> | tinderboxen are up, though, so it's not a complete outage |
| 12:21 | <Hixie> | moznet and bugzilla seem to be down |
| 12:22 | <Hixie> | moznet as in the irc network |
| 12:23 | <hsivonen> | I've received email from bugzilla 30 minutes ago |
| 12:23 | jgraham_ | grumbles at jQuery.ui for having dozens of broken tests in each part of the testsuite |
| 12:23 | <Ox032F> | seems like its a firefox bug, leaving out the currentTime=0; makes it much better, now theres only the problem of a sound not playing sometimes (randomly) |
| 12:24 | <Ox032F> | but I'll first upgrade to 3.5.3 from 3.5.2, just to be safe |
| 12:29 | <Hixie> | irc's back |
| 12:29 | <Hixie> | bugzilla isn't |
| 12:32 | <hsivonen> | DevMo is gone, too |
| 12:42 | <hsivonen> | I guess I'll have to start a new chunk of code without a bug number then |
| 12:43 | <roc> | Ox032F: currentTime doesn't start playing for me |
| 12:43 | <roc> | well, this is a latest-trunk build, but nothing much has changed in that are |
| 12:43 | <roc> | a |
| 12:43 | <Ox032F> | roc: which version do you have? |
| 12:43 | <Ox032F> | ok |
| 12:43 | <Ox032F> | I just upgraded to 3.5.3 |
| 12:43 | <Ox032F> | lets see |
| 12:43 | <roc> | that won't change anything |
| 12:45 | <roc> | woohoo, I have now taught our media cache manager how to let multiple streams share the same underlying cache data |
| 12:48 | <Ox032F> | arg, thats really shitty, I now have controls enabled, so I can test manually |
| 12:48 | <Ox032F> | and what do I get? all these errors are firefox inherent and not javascript based |
| 12:48 | <Ox032F> | crap |
| 12:48 | <Ox032F> | maybe the media is just to short |
| 12:52 | <Ox032F> | or maybe its a problem with gecko-mediaplayer again |
| 12:52 | <Ox032F> | but on the other hand, I'm not the only one having this problem |
| 12:53 | hsivonen | wonders if anyone has experimented with decoding Vorbis or Theora in OpenCL |
| 12:53 | <Ox032F> | mmh, but firefox uses its own player for the audio-tag, doesn't it? |
| 12:59 | <jgraham_> | annevk2: It seems like your argument about <details> applies equally to any markup feature that adds new functionaility e.g. people could start using <input type="color"> but not get the format right so, once a browser implements it, they reject input from that browser |
| 12:59 | <zcorpan_> | indeed |
| 12:59 | <annevk2> | what's your point? |
| 13:00 | <jgraham_> | That it seems we don't consider that a strong enough argument to override graceful degradation in other circumstances |
| 13:00 | <annevk2> | I'm less worried about <input type=color> because that is actually being implemented in e.g. WebKit |
| 13:00 | <jgraham_> | iirc WebKit want to implement <details> |
| 13:01 | <jgraham_> | (in the short term) |
| 13:01 | <annevk2> | I thought Maciej's email said they were not in a hurry |
| 13:01 | <annevk2> | maybe I missed something |
| 13:01 | <jgraham_> | I could well be wrong |
| 13:02 | <annevk2> | but yes my argument applies to other cases too and I think there's a risk there too if authors move ahead of implementors |
| 13:06 | <zcorpan_> | maybe the validator should warn about features that aren't implemented in browsers yet (except sectioning elements) |
| 13:13 | <Philip`> | We don't want the purity of our new language features to be disrupted by having people actually use them |
| 13:13 | <Philip`> | The existence of users just makes everything unnecessarily complex |
| 13:13 | <hsivonen> | whoa: http://www.w3.org/mid/640dd5060909130402u24d0c2c0j2e5b66e7d39c7f86⊙mgc |
| 13:14 | <jgraham_> | All fear The Power of RDF |
| 13:19 | <annevk2> | jgraham_, I don't think <legend> breaks degrade gracefully |
| 13:20 | <annevk2> | though it's certainly not ideal |
| 13:20 | <jgraham_> | annevk2: Really? It seems like a poster child for how to break that principle to me... |
| 13:20 | <hsivonen> | I wonder if Mark Birbeck has noticed that ARIA got rid of the RDF-based extensibility mechanism |
| 13:21 | <jgraham_> | annevk2: (I am really curious to know why you think it doesn't break the principle) |
| 13:21 | <Lachy> | annevk2, the lack of graceful degradation is the whole problem with using legend |
| 13:21 | <Lachy> | s/lack of/poor/ |
| 13:21 | <jgraham_> | (because to me it is like using <color> rather than <input type="color"> and saying "just don't use it until all browsers support it") |
| 13:22 | <jgraham_> | s/using/specifying/ |
| 13:25 | <zcorpan_> | <div><figure><img><legend><div>foo</div></legend></figure></div> - figure>span, legend>span { background:yellow } |
| 13:25 | <zcorpan_> | works fine for me |
| 13:25 | <zcorpan_> | s/span/div/ |
| 13:26 | <zcorpan_> | or just figure div {} |
| 13:28 | <Philip`> | (Is T.V Raman's argument for @role basically that since @class got abused, we should make an identical new attribute with a different name and then magically it won't be abused in exactly the same way?) |
| 13:29 | <jgraham_> | zcorpan_: Since <figure> can have a <div> child that is not in the <legend> that doesn't work in general |
| 13:29 | <Lachy> | zcorpan_, that only tests a very limited set of styles. <legend> creates an extra fieldset element in some browsers, and has some styles that simply cannot be altered in current browsers |
| 13:30 | <jgraham_> | (you could add an extra classname or something but then you get the problems Lachy mentioned) |
| 13:30 | <zcorpan_> | jgraham_: true, set a class |
| 13:30 | <zcorpan_> | Lachy: i don't get a fieldset for the above in firefox |
| 13:31 | <Lachy> | zcorpan_, turn off the HTML5 parser |
| 13:31 | <zcorpan_> | i have |
| 13:31 | <zcorpan_> | i get a fieldset if i remove the outer div |
| 13:31 | <jgraham_> | And you still have to deal with the fact that the DOM is confusing and unintuitively different from the markup |
| 13:32 | <jgraham_> | Which makes it more likely that the problem Anne mentioned will occur when browsesr are changed to unbreak <legend> |
| 13:33 | <jgraham_> | i.e. people will do legend div {something} and it won't do anything in old bropwsers but will in new ones |
| 13:33 | <virtuelv> | I'm reading http://hsivonen.iki.fi/doctype/ie8-mode.png |
| 13:33 | <virtuelv> | I'm incredibly confused |
| 13:33 | <jgraham_> | virtuelv: Yes |
| 13:33 | <jgraham_> | virtuelv: That is the good and natural reaction |
| 13:33 | <jgraham_> | virtuelv: It can be used as a Turing test |
| 13:34 | <virtuelv> | so, for a site considered to be in the intranet zone, it's impossible to trigger standards mode? |
| 13:34 | <zcorpan_> | no, you can use the x-ua-compatible override |
| 13:35 | <virtuelv> | if I was a web developer and had to target IE, this is the time I'd promptly quit my job and become a sheep farmer, or something |
| 13:38 | <virtuelv> | zcorpan_: there is one other way, depending on user preference |
| 13:39 | <zcorpan_> | virtuelv: yes, there's a user override too |
| 13:40 | <virtuelv> | + |
| 13:43 | <Lachy> | the benefit of that though, is that it gives us a really good argument against versioning in HTML |
| 13:44 | <Philip`> | virtuelv: It's easy, all you need to do is target IE7 and everything will work fine |
| 13:44 | <jgraham_> | (unless you want to use some fancy new IE80only features like native JSON support) |
| 13:44 | <Philip`> | (Well, sure, but you can just use Flash or Silverlight if you want a modern development environment) |
| 13:47 | <Lachy> | the flow chart seems incomplete. Some doctypes like <!DOCTYPE html> are supposed to trigger IE8 standards mode too, without the need for X-UA-Compatible |
| 13:49 | <Philip`> | Have you tested that? |
| 13:49 | Philip` | is suspicious of the use of "supposed to" |
| 13:50 | <hsivonen> | Lachy: I thought so, too, before I tested |
| 13:51 | <annevk2> | Lachy, that is covered by the chart |
| 13:51 | <annevk2> | or do you mean in intranets? |
| 13:51 | <Lachy> | no, I don't mean intranets |
| 13:52 | <Lachy> | I'm sure I read that IE changed their mind and had decided to allow (X)HTML Strict and HTML5 doctypes to trigger full standards mode by default |
| 13:52 | <hsivonen> | Lachy: did you see the "or absent" bit in the flow chart? |
| 13:52 | <Lachy> | yeah |
| 13:53 | <Lachy> | that leads down the path to IE7 standards mode |
| 13:53 | <annevk2> | no |
| 13:53 | <hsivonen> | Lachy: take absent, absent, No, No, No, No, No, No, No |
| 13:53 | <Lachy> | what? |
| 13:54 | <Lachy> | oh, I misunderstood the Display All Websites pref choice |
| 13:56 | <Philip`> | "No, No, No, No, No, No, No" sounds like a good response to the compatibility-mode graph |
| 13:57 | <Lachy> | hsivonen, it would help if that "Display All Web Sites... Pref Set?" choice said something less ambiguous like "Compatibility View Enabled?" |
| 13:58 | <hsivonen> | Lachy: that's more ambiguous |
| 13:58 | <hsivonen> | Lachy: many ways to enable it |
| 14:00 | <Lachy> | hsivonen, the point is I didn't have a clue what you meant by "Displya All Web Sites...", and with the Yes option leading to Intranet stuff, I assumed right answer should have been no |
| 14:00 | <Lachy> | and the user override button is labelled "Compatibility View", so say something that mentions that it's a user override for compat view would be good |
| 14:02 | <hsivonen> | Lachy: I'll try to fix those issues when I next revise the chart |
| 14:03 | <hsivonen> | Lachy: however, it sucks to make one of the diamonds larger to fit more text |
| 14:04 | <Lachy> | it would be less text to have "Comatibility View Override" |
| 14:05 | <hsivonen> | Lachy: the key is that I'm talking about the pref |
| 14:05 | <Lachy> | yeah, and the button for that pref is called Compatibility View. How is that not clear? |
| 14:06 | <Lachy> | Is there some pref listed in the options dialog with a longer name like "Display All Web Sites..."? |
| 14:07 | <hsivonen> | Lachy: yes. that's the point of the chart |
| 14:08 | <hsivonen> | Lachy: the button comes further toward bottom right of the chain |
| 14:08 | <Lachy> | where? I can't find it |
| 14:10 | <hsivonen> | Lachy: third diamond from the right |
| 14:11 | <Lachy> | no, not that. Where's the pref in IE? |
| 14:11 | <hsivonen> | Lachy: in compat mode settings |
| 14:12 | <hsivonen> | Lachy: different from the usual pref panes |
| 14:12 | <Lachy> | ah |
| 14:13 | <Lachy> | surely the pref and button have the same effect as far as the flow chart is concerned |
| 14:14 | <hsivonen> | Lachy: nope |
| 14:14 | <hsivonen> | Lachy: the button toggles the per-domain pref entry |
| 14:14 | <hsivonen> | Lachy: the main pref overrides |
| 14:14 | <hsivonen> | Lachy: just like the chart says |
| 14:14 | <Lachy> | dammit. There has to be some way to make the diagram less confusing than it is |
| 14:15 | <hsivonen> | Lachy: I refuse to believe the mess is simpler unless you provide testing-based steps to reproduce showing that it is simpler |
| 14:18 | <Lachy> | if either the pref is set or the button is on, then it will lead to the quirky doctype check. If they're both off, then they have no effect on the rest. Moving "Compatibility Mode Button Pressed?" up to the No side of "Display All Web Sites... Pref Set?" will not have any effect on the algorithm whatsoever |
| 14:19 | <Philip`> | I suppose for <script implements> (see public-html) one would have to do <script implements="http://www.w3.org/1999/xhtml/attribute#implements http://www.w3.org/1999/xhtml/element#details">; to make sure the implementation implements implements too |
| 14:19 | <Lachy> | and treating them as the same thing will also have no effect on the end result |
| 14:20 | <hsivonen> | Lachy: the button needs to come after the "Framed by" diamond |
| 14:21 | <Lachy> | no it doesn't |
| 14:22 | <hsivonen> | Lachy: ok. true |
| 14:22 | <hsivonen> | Lachy: I don't really see a win in shuffling the diagram around |
| 14:23 | <hsivonen> | Lachy: the purpose isn't to make it easy for you determine which mode your stuff will render in |
| 14:23 | <hsivonen> | Lachy: the purpose is to provide evidence that I did my homework and therefore you should believe my recommendations instead of trying to play the mode game |
| 14:23 | <Philip`> | The route to "Quirky or No Doctype?" is (display-all || (display-intranet && intranet) || domain-on-list || framed-by-compat || compat-mode-button) and || is commutative so you can rearrange all those things in any way |
| 14:24 | <hsivonen> | Lachy: likewise, the doctype table is there to show I did my homework and you should follow my recommendations instead of trying to do something else based on the table |
| 14:25 | <hsivonen> | Philip`: I think this order is logical, although I see the point about commutativity |
| 14:25 | <karlushi> | [08:35] <virtuelv> I'm reading http://hsivonen.iki.fi/doctype/ie8-mode.png |
| 14:25 | <karlushi> | that would be cool to do it for all browsers if it doesn't already exist |
| 14:26 | <hsivonen> | karlushi: there's http://hsivonen.iki.fi/doctype/ that has the same level of detail for other browsers |
| 14:27 | <hsivonen> | karlushi: the choice in other browsers isn't as crazy, so it doesn't need a chart |
| 14:28 | <hsivonen> | well, Gecko has interesting restrictions on how near the start of the stream the doctype needs to be |
| 14:28 | <hsivonen> | but the HTML5 parser fixes that |
| 14:30 | <karlushi> | hsivonen, except that it would give a side by side quick view |
| 14:30 | <annevk2> | non-IE: normal || IE: insane |
| 14:30 | <hsivonen> | karlushi: oh for visualizing the lesser complexity? |
| 14:30 | <karlushi> | hsivonen, yes |
| 14:30 | <hsivonen> | karlushi: I see |
| 14:31 | <Rik|work> | there's only doctype swithcing in other browsers ? |
| 14:31 | <hsivonen> | Rik|work: doctype and mime type |
| 14:31 | <hsivonen> | Rik|work: and Dashboardness |
| 14:32 | <hsivonen> | Rik|work: but the Dashboardness stuff isn't Web-exposed |
| 14:32 | <Rik|work> | mimetype ? |
| 14:32 | <annevk2> | Rik|work, text/xml triggers no quirks mode |
| 14:32 | <hsivonen> | Rik|work: XML mime types trigger the XML variant of the Standards Mode |
| 14:32 | <annevk2> | but maybe that's unfair since IE doesn't even have a non-text/html mode for this stuff |
| 14:33 | <Rik|work> | obvisouly, XML mime types triggers XML mode |
| 14:33 | <Rik|work> | but i didn't know about text/xml |
| 14:34 | <annevk2> | there's no XML mode really, they all use no quirks mode |
| 14:34 | <Lachy> | wow, the implements="" attribute idea on public-html doesn't seem very scalable |
| 14:35 | <Lachy> | nor sensible |
| 14:35 | <Philip`> | Lachy: It uses URIs, of course it's scalable |
| 14:36 | <Lachy> | no, I mean for authors. If you have to test for a lot of features, you'd have to write implements="http://.../feature1 http://.../feature2 ..." |
| 14:38 | <Philip`> | Then you could define a new feature at a new URI which lists all those features (e.g. define a "video-version-1" which includes a dozen individual video features) as RDF and then the browser can just use some OWL to deduce whether it implements the feature or not |
| 14:43 | <Philip`> | ("The optional implements attribute indicates that the script provides an implementation of the feature or features identified via this attribute. The script SHOULD only be loaded and used if the user agent does not have an implementation of the specified feature." seems to be the extent to which it's defined, in http://www.w3.org/MarkUp/2008/ED-xml-events-20081223/) |
| 14:43 | <Philip`> | (and sadly it doesn't suggest my awesome RDF idea) |
| 14:44 | hsivonen | wonders how people fall for RDF boondoggles |
| 14:45 | <hsivonen> | they don't seem attractive to implementors |
| 14:45 | <hsivonen> | they don't seem attractive to content authors either |
| 14:54 | <jgraham_> | hsivonen: Curiously neither authors nor implementors seem to be the ones interested in RDF boondoggles |
| 14:55 | <Lachy> | jgraham_, if not authors, what would you call the community of people that are claiming to use RDF? |
| 14:56 | <jgraham_> | Theoriaticians |
| 14:56 | <jgraham_> | er Theoreticians |
| 14:57 | <Philip`> | Call them 'theors' - much easier to type |
| 14:57 | <Philip`> | What aspects of RDF are you considering to be boondoggles? |
| 14:58 | <Philip`> | (All aspects?) |
| 15:00 | <jgraham_> | I was specifically thinking of the aspects where you use RDF to communicate information about unknown features to an implementation e.g. this implements thing or the original aria role thing |
| 15:01 | <Lachy> | I don't see where in either the latest XHTML 2 or XML Events drafts that the implements attribute appears, as Toby claimed |
| 15:02 | <annevk2> | Philip` just pointed it out |
| 15:02 | <Philip`> | There seems to be a huge disconnection between e.g. Google's real-world use of RDFa, where they buggily convert a subset into some kind of tree based on hard-coded names, and some of the discussions in the RDFa group about schemas and automatic reasoning and everything |
| 15:04 | <Philip`> | Lachy: http://www.w3.org/MarkUp/2008/ED-xml-events-20081223/ |
| 15:04 | <karlushi> | I wonder if implementation stuff is related to http://www.w3.org/2007/uwa/wiki/Main_Page |
| 15:04 | <Philip`> | Lachy: and http://www.w3.org/MarkUp/2009/ED-xhtml2-20090407/mod-scripting.html |
| 15:05 | <hsivonen> | my guess is that Web ubiquity will be enabled by Fennec, various WebKit-based browsers and Opera Mini rather than DIAL and stuff |
| 15:16 | <annevk2> | https://bugzilla.mozilla.org/show_bug.cgi?id=507970 -- new web font formats... |
| 15:17 | <Philip`> | Sounds like DIAL is about letting you write content for lots of heterogeneous devices simultaneously, which sounds like it's either unscalable (if heterogeneity increases in the future, and you have to write and test for a zillion devices) or only a short-term problem (if heterogeneity decreases, and everyone has HTML5 browsers on iPhone clones) |
| 15:18 | <hsivonen> | Philip`: it's a transitient problem that is addressed by making mobile browsers suck less |
| 15:19 | <hsivonen> | Philip`: i.e. your "iPhone clones" case |
| 15:22 | <Philip`> | Hmm, the examples in the Primer make it look like just a server-side templating language |
| 15:23 | Philip` | isn't sure why that's something that needs standardisation |
| 15:24 | <Philip`> | (and the listed use cases in the Primer don't seem to be related to any kind of need for standardisation or interoperability) |
| 15:24 | <annevk2> | so what the company that implements it can claim it's W3C-approved |
| 15:26 | <hsivonen> | Philip`: well, if you are a mobile operator and you want to be able to buy compatible systems for publishing to bad UAs from multiple vendors, it makes sense to want standardization |
| 15:27 | <Lachy> | it seems odd that Apple would open source Grand Central Dispatch, yet do so with a licence that isn't GPL2 compatible, and thus incompaible with the most commonly used licence for linux distros. |
| 15:27 | <Lachy> | http://arstechnica.com/open-source/news/2009/09/apple-opens-gcd-challenges-impede-adoption-on-linux.ars |
| 15:27 | <Philip`> | hsivonen: Wouldn't you just want standardisation of an API to find device capabilities, and then feed it into an arbitrary non-standardised server-side templating system? |
| 15:28 | <hsivonen> | Philip`: if you are the party who owns a lot of templated content and you want to be able to purchase your next template engine from a different vendor |
| 15:28 | <hsivonen> | Philip`: ...you want to be able to buy compatible template engines from multiple vendors |
| 15:30 | <Philip`> | I suppose I don't understand why you'd want to buy a template engine from a different vendor, any more than you'd want to buy any other code library from a different vendor |
| 15:30 | <Philip`> | (and most code libraries don't have any kind of standardised API) |
| 15:33 | <gsnedders> | Lachy: But does libdispatch need to be integrated with anything that makes that a problem? |
| 15:33 | <Philip`> | Lachy: What should they do instead, that wouldn't impose unwanted restrictions (like the GPL does) and wouldn't lose desired restrictions (like Apache's patent stuff) and wouldn't cause license proliferation? |
| 15:35 | <Philip`> | If you have GPLv2-only code then (I believe) you can't link with GPLv3 libraries either, so you're already suffering and it's no worse with Apache-licensed libraries |
| 15:37 | hsivonen | mumbles about zfs and cddl |
| 15:37 | <Philip`> | gsnedders: I assume the idea is that arbitrary applications link with libdispatch so they can use GCD, and so it's a problem whenever those applications are Apache-incompatible |
| 15:39 | <gsnedders> | Philip`: Ah, true |
| 15:39 | <gsnedders> | Philip`: But doesn't that issue occur on OS X too? |
| 15:39 | <Lachy> | the compatibility problem probably lies with the GPL2, which has apparently been resovled in GPL3, but there's been a lot of resistance to adopting GPL3 for other reasons |
| 15:40 | <hsivonen> | it would certainly be amusing if Apple pushed GPLv2-or-later apps to become GPLv3-or-later |
| 15:40 | Philip` | wonders why you need a whole library for GCD anyway, Euclid's algorithm is only a dozen lines |
| 15:41 | <hsivonen> | hah |
| 15:41 | <gsnedders> | unless libdispatch falls under, "However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable." |
| 15:41 | <Lachy> | would have been nice to get a GPL 2.1 or something that resolved the uncontroversial licence compat issues, but which avoids the controversial Tivo-isation issues |
| 15:44 | <jgraham_> | Umm, isn't it rather possible that Apple don't want GCD on Linux? |
| 15:44 | <Philip`> | Lachy: But it would be unethical to publish a license that doesn't protect users from Tivo-isation |
| 15:46 | Philip` | kind of likes the idea of C being based on a standard, and not having random extensions like blocks added by a single vendor |
| 15:47 | <hsivonen> | adding extensions to C is indeed unholy |
| 15:48 | <gsnedders> | Philip`: I thought their implementations of blocks matched C++0x |
| 15:49 | <takkaria> | Philip`: apple are pushing for blocks for be included in the next version of C |
| 15:49 | <hsivonen> | but GCC already has non-standard extensions to C, right? |
| 15:49 | <takkaria> | yup |
| 15:49 | <hsivonen> | and Linux even depends on them, IIRC |
| 15:49 | <takkaria> | aplenty |
| 15:49 | <Philip`> | gsnedders: It seems to be completely different syntax to C++0x's lambdas |
| 15:50 | <gsnedders> | Confessions, part 0x10000: I bought The Sims 3. |
| 15:50 | <Philip`> | "^ int (void) { return 3; }" vs "[](int x, int y) { return x + y; }" etc (copying some random examples) |
| 15:51 | <jgraham_> | Yay for readable syntax |
| 15:52 | Philip` | presumes the blocks syntax conflicts with C++/CLI's syntax |
| 15:52 | <Philip`> | (which isn't necessarily a problem; it's just a natural collision of extensions) |
| 15:53 | <gsnedders> | We need namespaces. |
| 15:53 | <Philip`> | Namespaces for syntax? |
| 15:53 | <Philip`> | "blocks::^ int (void) { return 3; }" |
| 15:53 | <gsnedders> | Yes. |
| 15:53 | <gsnedders> | I mean, namespaces solve all problems. |
| 15:54 | <Philip`> | Apart from the "Yay for readable syntax" problem |
| 15:54 | <Philip`> | hsivonen: GCC has lots of extensions, but at least you can sure they'll be portable to every platform's GCC, which is as close to universality as you're likely to want :-) |
| 15:55 | <Philip`> | (unless you care about Windows, I suppose) |
| 15:56 | <Philip`> | takkaria: Do you know when they proposed it for standardisation? |
| 15:56 | <Philip`> | (Seems a bit late if they're doing it after shipping it) |
| 15:56 | gsnedders | notes Apple's fork of gcc has more extensions to C than just that |
| 15:58 | <takkaria> | Philip`: it's mentioned in another Ars article |
| 15:58 | <Philip`> | http://developer.apple.com/mac/library/documentation/Darwin/Reference/ManPages/man1/gcc.1.html would be nice if it wasn't restricted to a single screen height in Opera so I can't scroll down to read the page |
| 15:58 | <takkaria> | C++0x closures are horrible ugly constructions, though, and Apple's aren't |
| 15:58 | <Philip`> | (unless I disable CSS) |
| 16:00 | <takkaria> | "It's Apple intention to submit blocks as an official extension to one or more of the C-based languages, though it's not yet clear which standards bodies are receptive to the proposal." [http://arstechnica.com/apple/reviews/2009/08/mac-os-x-10-6.ars/10] |
| 16:00 | <takkaria> | so maybe I was slightly off-mark |
| 16:01 | <Philip`> | (What I meant to say was: Maybe Opera has a bug on that page, but I'm too lazy to look into it) |
| 16:01 | <Philip`> | takkaria: Hmm, sounds a little bit woolly |
| 16:01 | <takkaria> | indeed |
| 16:01 | <takkaria> | though actually, two compilers have blocks: llvm and apple-gcc |
| 16:02 | <takkaria> | so they're already more portable than gccisms |
| 16:03 | <Philip`> | gsnedders: The documented differences seem to almost all be optimisations and warnings, not language changes |
| 16:04 | <Philip`> | takkaria: Two compilers have most gccisms (gcc and icc) :-p |
| 16:04 | <Philip`> | (icc at least had enough to compile the Linux kernel, some years ago) |
| 16:13 | <takkaria> | OK, you win |
| 16:15 | <Philip`> | Hooray |
| 16:15 | <Philip`> | Do I get a medal? |
| 16:18 | <takkaria> | winning doesn't entail anything here except for smugness, sadly |
| 16:18 | <takkaria> | and not even if that if you have restraint |
| 16:18 | <Philip`> | http://clang.llvm.org/docs/BlockLanguageSpec.txt - "A Block reference may be cast to a pointer of arbitrary type and vice versa." - I always forget how little C bothers with type safety |
| 16:19 | <Dashiva> | Type safety, more like type convenience |
| 16:19 | <Philip`> | (and then I get annoyed when C++ cares about type safety and complains about some third-party library casting a pointer-to-member-function to a pointer violates the strict aliasing rules or whatever) |
| 16:19 | <Philip`> | s/about/that/ |
| 16:20 | <Philip`> | s/casting/is casting/ |
| 16:20 | <Philip`> | Wait, that still doesn't make grammar |
| 16:20 | Philip` | gives up |
| 16:20 | <takkaria> | *** No rule to make target `grammar'. Stop. |
| 16:22 | <Philip`> | Hmm, it sounds like blocks just do boring const copies of stack variables |
| 16:22 | <Philip`> | like what Java inner methods do |
| 16:23 | <Philip`> | and not like real closures at all |
| 16:23 | <Philip`> | (Actually I suppose that depends on what real closures are) |
| 16:24 | <Philip`> | (but anyway it seems you can't do the equivalent of "sub adder { my $n = 0; return sub { return ++$n } }" without doing explicit heap allocation) |
| 16:26 | <jgraham_> | You mean they can't modify varibles in the outer scope |
| 16:26 | <jgraham_> | ? |
| 16:27 | <Philip`> | (...unless you mark the local variable with __block) |
| 16:27 | <Philip`> | jgraham_: Yes (except with __block) |
| 16:28 | <Philip`> | If you wanted to modify local variables in the outer scope, you'd really need garbage-collected heap-allocated stack frames |
| 16:29 | <Philip`> | which isn't going to happen in C, so you can't modify them and it just copies their values into a per-block bit of memory |
| 16:30 | <Philip`> | so that it doesn't matter when the stack frame gets popped |
| 16:31 | <Philip`> | (At least I think that's how it works; please tell me I'm being stupid if I am) |
| 16:32 | <takkaria> | that seems to be how they work |
| 16:32 | <takkaria> | seems about the closest you get to real closures in C, really |
| 16:33 | <Philip`> | Introducing type inference on return values seems weird, since nothing else in C does anything like that, as far as I'm aware |
| 16:35 | <Philip`> | and C is not known for syntactic sugar |
| 16:47 | <Steve^> | That's how I build be closures, at least |
| 16:48 | <Steve^> | partly because it is easier and partly because it is much easier to define the behaviour |
| 16:49 | <Steve^> | sub adder { $a = shift; return sub { return $a + shift; }}; sub add10 = adder(10); |
| 16:49 | <Steve^> | being the notable example |
| 16:49 | Philip` | eventually realises he should have done his earlier example in JS, like "function adder() { var n = 0; return function () { return ++n; } }" |
| 16:49 | <Philip`> | (because I guess more people here know JS than Perl) |
| 16:51 | <Steve^> | sub double { $m = shift; return sub { $m($m(shift)); }} is awesome too |
| 16:52 | <Philip`> | It'd be better if it wasn't a syntax error :-p |
| 16:52 | <Philip`> | $m->($m->(shift)) I think |
| 16:52 | <Steve^> | it will actually work in perl? |
| 16:52 | <Steve^> | I suppose it might |
| 16:53 | <Philip`> | Why wouldn't it? |
| 16:54 | <Steve^> | I didn't consider it. I wrote a compiler for a c-like language with closures and was just translating |
| 16:54 | <Philip`> | Ah |
| 16:54 | <Pure> | Can I make it so, my application has an offline and an online version of a page and only the offline version is cached? |
| 16:57 | <Philip`> | Hmm, the discussion in http://arstechnica.com/apple/reviews/2009/08/mac-os-x-10-6.ars/13 of how awesomely trivial GCD makes it to do a 30-second document analysis in a background thread seems to kind of entirely ignore the hard bits of concurrency, like making sure you don't crash when another thread modifies the document you're looking at |
| 16:57 | <Philip`> | Also, the syntax doesn't seem that much easier than OpenMP |
| 17:34 | <JonathanNeal> | Is the html5 outliner having some issues? |
| 17:35 | <JonathanNeal> | Say I outline nasa.gov, I get this @ http://gsnedders.html5.org/outliner/process.py?url=http%3A%2F%2Fwww.nasa.gov%2F - if you get the same result, it's an XML Syntax Error. |
| 17:35 | <JonathanNeal> | Does that mean nasa's got the problem, or the html5 outliner? |
| 17:37 | <miketaylr> | eek, JonathanNeal that ain't good |
| 17:39 | annevk2 | wonders why Ubuntu still uses Firefox 3.0.x |
| 17:41 | <miketaylr> | JonathanNeal: did you try here? http://james.html5.org/outliner.html |
| 17:41 | <annevk2> | maybe it's because the xmlns:xsi at the top |
| 17:41 | <jgraham_> | JonathanNeal: That looks like a bug in html5lib |
| 17:43 | <annevk2> | why would html5lib emit an XML Syntax Error? |
| 17:43 | <jgraham_> | specifically we do some hack with lxml involving parsing a document using its built in parser to make it recognise the doctype and other useless crap. But I guess it breaks with strange input |
| 17:44 | <jgraham_> | (we could maybe use the html parser instead but there was some reason not to at the time. Possibly lxml 1.x compat.) |
| 17:45 | <jgraham_> | (which I no longer care about) |
| 17:47 | <JonathanNeal> | jgraham_, woops! |
| 17:47 | <JonathanNeal> | :D |
| 18:19 | <annevk2> | hsivonen, I meant to tell you that not just <keygen>, but also <isindex>, needs to be re-implemented in Gecko |
| 18:20 | <JonathanNeal> | What sites do you folks think are currently doing great work using html5? |
| 18:26 | <Philip`> | JonathanNeal: Clearly all sites using HTML5 are great |
| 18:29 | <annevk2> | I was going to say whatwg.org but it appears to be offline |
| 18:30 | <JonathanNeal> | annevk2, woops! |
| 18:32 | <JonathanNeal> | It's working for me though. |
| 18:32 | <JonathanNeal> | Just a wee bit slow. |
| 18:34 | <annevk2> | http://downforeveryoneorjustme.com/whatwg.org |
| 18:38 | <JonathanNeal> | annevk2, hmm, interesting. |
| 18:39 | <JonathanNeal> | http://downforeveryoneorjustme.com/downforeveryoneorjustme.whatwg.org -- hehehe |
| 18:39 | <JonathanNeal> | silly dfwojm |
| 18:39 | <JonathanNeal> | sorry, dfeojm |
| 20:08 | <hsivonen> | annevk2: do you mean the submission magic of isindex? |
| 20:15 | <annevk42> | hsivonen, could be, also, in previous versions the <isindex> element was not a macro |
| 20:16 | <annevk42> | hsivonen, I think all the <hr> and <input> stuff was done through XBL or something |
| 20:34 | <zcorpan_> | Hixie: the link "HTML5" on whatwg.org should link to the multipage version |
| 20:44 | <annevk42> | I wonder what happens when fakepath guy learns about the rest of things HTML5 defines |
| 20:44 | <annevk42> | the things* |
| 20:45 | <zcorpan_> | annevk42: you could find out |
| 20:46 | <zcorpan_> | maybe we should make a list of the top ten ugly compat hacks in html5 |
| 20:46 | <zcorpan_> | although some of them will maybe only appear as ugly for implementors |
| 20:48 | <zcorpan_> | i bet authors couldn't care less about the hacks involved with document.all |
| 20:48 | <zcorpan_> | (which, from an implementation pov, is ten times uglier than fakepath) |
| 20:50 | <annevk42> | it would have to be non-obsolete features |
| 20:50 | <annevk42> | i suppose we could obsolete .value in favor of .files |
| 20:51 | <annevk42> | that'd make it comparable :p |
| 20:51 | <zcorpan_> | but that's not really helpful :) |
| 20:54 | <Philip`> | Are there any script interfaces that are currently considered obsolete? |
| 20:54 | <Philip`> | (Seems a bit pointless in practice since validators can't warn you about them) |
| 20:55 | <zcorpan_> | document.all is obsolete |
| 20:55 | <zcorpan_> | IDL attributes that map to obsolete attributes are obsolete |
| 20:56 | <annevk42> | Philip`, error consoles could |
| 20:58 | <zcorpan_> | so does anyone know how <applet> works? |
| 21:06 | <annevk42> | zcorpan_, you could make an applet and figure it out... I once thought of doing that but then thought it'd be a waste of time |
| 21:07 | <zcorpan_> | i like that it has an alt attribute |
| 21:08 | <zcorpan_> | <applet alt="install java">upgrade your browser</applet> |
| 21:10 | <MikeSmith> | there seems to be a bug in my OSX Opera 10 that when a contenteditable part of page first receives focus, I get no visual indicator that it's editable -- i.e., on OSX, a blue outline around the element boundary. but when I switch windows to another app and then switch back to Opera, it does now show the blue outline |
| 21:11 | <annevk42> | zcorpan_, one of those is not universally supported |
| 21:12 | <zcorpan_> | MikeSmith: the blue outline gets easily confused |
| 21:13 | <zcorpan_> | MikeSmith: feel free to file a bug :) |
| 21:15 | <MikeSmith> | zcorpan_: OK |
| 21:15 | <MikeSmith> | hmm, I notice on Linux it never shows any indicator at all to signal that it's editable |
| 21:16 | <MikeSmith> | well, except for the caret |
| 21:16 | <zcorpan_> | no I-beam cursor? |
| 21:17 | <MikeSmith> | zcorpan_: yeah, that's what I meant be caret -- it does show that, blinking cursor |
| 21:17 | <zcorpan_> | the caret is the blinking thing, the cursor is the thing you move with the mouse :) |
| 21:18 | <MikeSmith> | roger that |
| 21:25 | <hsivonen> | Lachy: the author guide probably shouldn't tell authors that <svg:foo> is OK in XHTML |
| 21:26 | <hsivonen> | Lachy: it's better to encourage text/html-compatible habits |
| 21:26 | <zcorpan_> | polyglot++ |
| 21:26 | <Lachy> | hsivonen, send mail with all your comments |
| 21:27 | <hsivonen> | Lachy: to you or to a list. does bugzilla work for you? |
| 21:27 | <Lachy> | bugzilla is fine. |
| 21:27 | <hsivonen> | ok |
| 21:27 | <Lachy> | just select the authoring guide project |
| 21:31 | <Lachy> | oh, it's been renamed to 'Lachlan Hunt's "HTML5 Reference"' http://www.w3.org/Bugs/Public/enter_bug.cgi?product=HTML%20WG&component=Lachlan%20Hunt%27s%20%22HTML%205%20Reference%22 |
| 21:33 | <jgraham_> | Makes you sound like a wannabe celebrity |
| 21:34 | <Lachy> | I am |
| 21:42 | <zcorpan_> | is that to differentiate it from 'W3Schools' "HTML5 Reference"'? |
| 21:44 | <zcorpan_> | funny that keygen and isindex are both different from gecko in terms of macroness |
| 21:45 | <zcorpan_> | hsivonen: are you going to propose that isindex should be a normal void element? |
| 21:47 | <hsivonen> | zcorpan_: I'm not sure. I wasn't going to |
| 21:48 | <zcorpan_> | hsivonen: btw, http://tc.labs.opera.com/html/parsing/doctype/001.htm doesn't work with the html5 parser enabled |
| 21:49 | <zcorpan_> | says doc.body is null |
| 21:51 | <annevk42> | <isindex> does make more sense as a macro given how it's rendered |
