| 03:18 | <Hixie> | hm, rfc4329 doesn't actually define text/javascript1.2, it just says people use it |
| 03:18 | <Hixie> | guess i'll have to add rfc4329's list to html5 |
| 03:21 | <Philip`> | I only saw it on 7 out of 16K pages, and at least the first 2 of those are the same user-tracking code |
| 03:21 | <Philip`> | At least the first 3 |
| 03:22 | <Philip`> | At least the first 4 |
| 03:22 | <Philip`> | I guess they're probably all exactly the same code |
| 03:22 | <Philip`> | (See http://www.allesjagd.com/start/index.php etc) |
| 03:28 | <Philip`> | Why does rfc4329 mention text/javascript1.(4|5)? I can't find any browser that supports them |
| 03:32 | <Philip`> | Oh, Konqueror supports them |
| 03:43 | <Hixie> | supporting them the way the spec says to seems cheap and harmless |
| 03:43 | <Hixie> | (since it makes the language="" attribute a non-issue basically, a couple of lines of code) |
| 03:45 | <Hixie> | so in http://lists.w3.org/Archives/Public/www-archive/2007Sep/0019.html john asked that i reply to that e-mail i said i wouldn't reply to |
| 03:46 | <Hixie> | i sent a reply shortly after http://lists.w3.org/Archives/Public/www-archive/2007Sep/0020.html |
| 03:46 | <Hixie> | i haven't heard from them since |
| 03:49 | <Hixie> | so one thing that's weird about the argument that we should require alt text because that will force authors of sites like flickr to get the users to include alt text is that it's so obviously false... since html4 already requires alt="", and has for a decade |
| 03:49 | <Hixie> | yet flickr doesn't ask the user for alt="" |
| 03:59 | <Hixie> | incidentally, http://html4all.org/wiki/index.php/Mission_Statement is quite a good statement, except for the last paragraph |
| 04:02 | <Hixie> | (the last paragraph makes a somewhat unjustified leap from their problem statement to a solution that is more extreme than required to solve the problem they put forward) |
| 04:02 | <Hixie> | but it might be a good start for text for the design principles |
| 04:16 | <Lachy> | Hixie, the problem I have with the alt="" debate is that I'm yet to get a clear answer to the question that: if omitting alt isn't the answer to the problem of what to do when the user doesn't provide alt text, then what is? alt="" or alt="some placeholder text"? |
| 04:17 | <Lachy> | And, if the latter, then what placeholder text? |
| 04:17 | <Hixie> | yeah |
| 04:18 | <Hixie> | the spec uses an example from flickr... i spoke to the person who put up that image, telling them that all these accessibility people were saying she should have included alt text (not that flickr lets her) |
| 04:18 | <Lachy> | although, as Steve Faulkner pointed out, omitting alt and letting the screen reader read out the file name isn't the best user experience |
| 04:18 | <Hixie> | she was like "uh, i don't care, i don't know any blind people, this photo isn't going to be looked at by any blind people" |
| 04:18 | <Philip`> | alt="the best metadata that the application can provide"? |
| 04:18 | <Hixie> | and i think a lot of people agree with that |
| 04:18 | <Hixie> | (though not accessibility experts, typically) |
| 04:19 | <Hixie> | well if the alt is omitted, the AT should do something better. we're at the early stages of image analysis, but things are getting better. |
| 04:20 | <Lachy> | indeed, like in cases like this: <figure><img src=...> <legend>Caption</legned></figure>, the UA should probably not read the file name by default and just use the caption (if no alt or title attrs have been given) |
| 04:20 | <Philip`> | Web browsers don't seem like the place to put technology that is an active research topic and still in its early stages |
| 04:21 | <Lachy> | that would fix the reading out long meaningless numeric filename problem |
| 04:21 | <Hixie> | yeah |
| 04:22 | <Philip`> | (I can't actually think of anything even vaguely like modern research areas that web browsers are doing...) |
| 04:24 | <Lachy> | http://www.mail-archive.com/wsg⊙wo/msg30517.html |
| 04:25 | <Lachy> | that points to http://www.w3.org/TR/WAI-AUTOOLS/#check-no-default-alt |
| 04:25 | <Hixie> | Philip`: sure, today, but by the time html5 is finished... |
| 04:25 | <Lachy> | .. which says what not to do, without saying what should be done |
| 04:25 | <Hixie> | and before html6... |
| 04:28 | <Hixie> | (the other funny thing is that actually, lack or inclusion of alt text is the least of flickr's accessibility problems) |
| 04:29 | <Lachy> | this response seem to be the bury-your-head-in-the-sand-and-hope-the-problem-goes-away approach http://www.mail-archive.com/wsg⊙wo/msg30511.html |
| 04:30 | <Philip`> | Around ten years from now? If useful image analysis libraries begin to become practical then, it's still ten years of people writing content based on the HTML5 draft/CR which will work badly in UAs during that time |
| 04:33 | <othermaciej> | flickr uses layout tables! |
| 04:34 | <othermaciej> | in a way that splits related content in different table cells! |
| 04:47 | <Hixie> | more examples of poor alt text: http://www.w3.org/2006/03/TP-minutes.html |
| 11:09 | <zcorpan> | btw, i've seen "download this video" links alongside the video on several sites |
| 11:12 | <zcorpan> | e.g. http://catpages.nwmissouri.edu/m/lgmf/video-clips/slide1.html |
| 11:13 | <zcorpan> | (that page could make use of <video><source>) |
| 11:20 | <jgraham> | Alt text from an email I got recently: Oxford Playhouse Logo |
| 11:20 | <jgraham> | Floating Kate Rusby Jeremy Hardy Floating Kate Rusby Jeremy Hardy Floating Kate Rusby Jeremy Hardy |
| 11:20 | <jgraham> | Floating |
| 11:29 | <zcorpan> | jgraham: Brillant!! |
| 11:33 | <annevk> | For some reason I always get two commit e-mails |
| 11:36 | <annevk> | (from the html5 project) |
| 11:37 | <annevk> | jgraham, you should probably also update the comment that says that only "axis" makes it a heading |
| 11:39 | <jgraham> | Done |
| 11:42 | <annevk> | I think I might be on the commit list twice or something |
| 11:43 | <annevk> | jgraham, how about ensuring that a given header cell spans the same or more columns than the data cell? |
| 11:44 | <annevk> | this would help with Henri's thesis and my tmb table |
| 11:44 | <annevk> | and I can't directly think of any data cell that would have two headings and not be empty |
| 11:47 | <jgraham> | annevk: I'll try that |
| 11:47 | jgraham | is going out now |
| 11:48 | <jgraham> | I also really need to write some proper tests but #algorithms * #options =a large number |
| 11:52 | <annevk> | HTML5 also doesn't really take rowgroup into account |
| 11:52 | <annevk> | HTML5 algorithm seems a bit simplistic although at least it seems accurate |
| 11:58 | <zcorpan> | as far as associating header cells with header cells goes, it would make sense to associate a cell that spans fewer columns but not more columns, with a header cell higher up in the table |
| 11:59 | <annevk> | I think that's my suggestion, although I'd say fewer or equal amount of columns |
| 12:00 | <zcorpan> | yeah |
| 12:00 | <annevk> | probably analogously for rows |
| 12:00 | <zcorpan> | yes |
| 12:01 | <zcorpan> | ignoring <thead>, it should be possible to flip or rotate a table and still have the same associations |
| 12:02 | <annevk> | well, cells can span columns, but can't span rowgroups |
| 12:02 | <annevk> | iirc |
| 12:04 | <zcorpan> | right |
| 12:05 | <zcorpan> | ok, ignoring <thead>, multiple <tbody> and <tfoot> |
| 12:06 | <webben> | zcorpan: Why would it make sense to associate...? |
| 12:06 | <webben> | aren't related header cells likely to span the same number of columns? |
| 12:07 | <zcorpan> | webben: yes (what anne said) |
| 12:07 | <zcorpan> | though only if they are adjacent |
| 12:08 | <zcorpan> | http://html5.lachy.id.au/output?data=%3Ctable+border%3E%0D%0A%3Ctr%3E%3Cth%3Efoo%3Cth%3Ebar%0D%0A%3Ctr%3E%3Ctd%3E1%3Ctd%3E2%0D%0A%3Ctr%3E%3Cth%3Ebaz%3Cth%3Equux%0D%0A%3Ctr%3E%3Ctd%3E3%3Ctd%3E4%0D%0A%3C%2Ftable%3E&type=text%2Fhtml%3B+charset%3DUTF-8 |
| 12:08 | <zcorpan> | there, foo and baz probably shouldn't be associated together |
| 12:09 | <zcorpan> | but should be with http://html5.lachy.id.au/output?data=%3Ctable+border%3E%0D%0A%3Ctr%3E%3Cth%3Efoo%3Cth%3Ebar%0D%0A%3Ctr%3E%3Cth%3Ebaz%3Cth%3Equux%0D%0A%3Ctr%3E%3Ctd%3E1%3Ctd%3E2%0D%0A%3Ctr%3E%3Ctd%3E3%3Ctd%3E4%0D%0A%3C%2Ftable%3E&type=text%2Fhtml%3B+charset%3DUTF-8 |
| 12:09 | <Lachy> | zcorpan, note the new tinyURL button http://html5.lachy.id.au/ |
| 12:10 | <peepo> | ~:" just about to send something similar.... |
| 12:10 | <webben> | zcorpan: When you say associated, what do you mean in this case? |
| 12:10 | <Lachy> | it generates that link and then gets a tinyurl.com/ address for it |
| 12:10 | <webben> | zcorpan: If you mean foo should be a header for baz, I'm not sure. |
| 12:11 | <zcorpan> | webben: yes |
| 12:11 | <webben> | zcorpan: If foo had two or more sub-th's that would certainly make sense. |
| 12:11 | <zcorpan> | why not? |
| 12:11 | <zcorpan> | yeah |
| 12:11 | <webben> | zcorpan: Or at least if at least one other th at the same level as foo had two or more sub-th's |
| 12:12 | <webben> | zcorpan: Because then it's likely that the first row of th's is being used to carve up the second row. |
| 12:13 | <zcorpan> | hmm |
| 12:13 | <webben> | if each top th has only one "child" th, I'd have thought something else is going on |
| 12:13 | annevk | wonders what http://html5.lachy.id.au/ does |
| 12:13 | <annevk> | Lachy, was today the day you were going to put that redirect in place? |
| 12:14 | <webben> | (I'm sure there could be instances where the same thing is going on; perhaps the subheadings are time related or something, and the table is automatically generated and there will be more subheadings as time progresses.) |
| 12:14 | <Lachy> | annevk, yes, I'm working on that. I just need to port the /faq/pl/ page across to the wiki |
| 12:14 | <annevk> | oh, missed that |
| 12:14 | <webben> | But it would be nice to see some live examples of th pairs to see if they do often differ. |
| 12:14 | <Lachy> | annevk, http://html5.lachy.id.au/ is what http://html5.lachy.id.au/output |
| 12:14 | <Lachy> | ... used to be |
| 12:14 | <annevk> | but the purpose is? |
| 12:14 | <webben> | Do we have any? |
| 12:15 | <annevk> | well, my table would benefit from that additional rule |
| 12:15 | <Lachy> | for generating simple test pages, without having to create and upload a file |
| 12:15 | <annevk> | Henri's thesis would |
| 12:15 | <Lachy> | annevk, just like zcorpan did above |
| 12:16 | <webben> | annevk: Your table? |
| 12:16 | <zcorpan> | annevk: we're discussing what should be done in this case specifically http://tinyurl.com/2gtf6a |
| 12:16 | <annevk> | webben, http://annevankesteren.nl/2007/09/tmb-overview |
| 12:16 | <zcorpan> | if foo a heading for baz or not? |
| 12:17 | <annevk> | it is |
| 12:17 | <zcorpan> | why? |
| 12:17 | <annevk> | http://lists.w3.org/Archives/Public/www-archive/2007Aug/att-0003/offset-mess.htm |
| 12:18 | <zcorpan> | that's not the same case (they don't span the same number of columns) |
| 12:18 | <annevk> | oh, hmm |
| 12:19 | <annevk> | i guess it doesn't really matter |
| 12:19 | <webben> | it might be equally reasonable (for example) to assume the th pairs are equivalents. |
| 12:19 | <webben> | I don't think we can make any conclusions about whether it matters without finding some examples that are actually like that. |
| 12:20 | <webben> | I wonder if Philip could find us some tables with those specific characteristics. |
| 12:23 | <webben> | Alternatively one could just follow the html4 algorithm and assume people have assumed that. |
| 12:23 | <webben> | (I /guess/ that would make foo a heading for baz.) |
| 12:24 | <annevk> | yeah, I believe HTML4 does that |
| 12:24 | <annevk> | I don't think many people have read HTML4 table association though |
| 12:25 | <zcorpan> | since <th> is used in layout tables... :) |
| 12:25 | <webben> | zcorpan: How often is that? |
| 12:26 | <webben> | (used in layout tables, I mean) |
| 12:26 | <zcorpan> | not very common but i've seen it |
| 12:32 | <Philip`> | I can parse lots of pages now (whereas before I was just tokenising), so I should be able to find things with certain characteristics (assuming those characteristics aren't too hard to determine) if it'd be helpful |
| 12:32 | <webben> | zcorpan: When you have seen it, was th still being used for headers? |
| 12:34 | <webben> | Philip`: would looking for <tr><th>+</tr> followed immediately by another <tr> with an equal number of <th> be easy? |
| 12:34 | <annevk> | Philip`, you finished your tree builder? |
| 12:35 | <Philip`> | annevk: No, I used hsivonen's one since that's much easier and probably better :-) |
| 12:36 | <zcorpan> | webben: just recently when debugging a parsing problem in opera... can't find the bug again though |
| 12:36 | <zcorpan> | webben: the page didn't have data tables |
| 12:37 | <webben> | zcorpan: No I meant structural headers (where say <hX> or <dt> or something might have been vaguely appropriate), rather than for data headers. |
| 12:37 | <webben> | Or whether people just get th and td confused. |
| 12:37 | <webben> | (they're not the most intuitive of names!) |
| 12:38 | <zcorpan> | it was for the page logo at least |
| 12:38 | <zcorpan> | but that table didn't cover the whole page, just the top part |
| 12:39 | <Philip`> | webben: That sounds feasible, though I'm not sure exactly how to write the code to detect it (or whether it'd be easier with the DOM parser rather than the SAX one) |
| 12:39 | <annevk> | Philip`, HTML5 has an algorithm on how to construct a table |
| 12:40 | <annevk> | although maybe implementing that is overkill, dunno |
| 12:42 | <Philip`> | annevk: That sounds like it might be hard work, particularly since writing Java is no fun |
| 12:42 | <zcorpan> | webben: found it, it only had one <th> for the page logo http://www.itquestionbank.com/ |
| 12:43 | <webben> | Well at least it makes a certain amount of logical sense. |
| 12:44 | <zcorpan> | yeah, but it's still a layout table... :) |
| 12:44 | <webben> | okay, so twisted logic ;) |
| 12:45 | <Philip`> | I saw some pages a while ago that had only <th> and no <td>s |
| 12:46 | <Philip`> | http://krijnhoetmer.nl/irc-logs/html-wg/20070717#l-11 |
| 12:54 | <Philip`> | http://www.gbk-unternehmensberatung.de/ | 8 |
| 12:54 | <Philip`> | http://www.new-video.de/film-lolita/ | 173 |
| 12:54 | <Philip`> | http://www.chalupa24.cz/ | 2 |
| 12:54 | <Philip`> | http://ccel.org/cceh/ | 1 |
| 12:54 | <Philip`> | http://www.akatalent.com | 1 |
| 12:54 | <Philip`> | http://www.spiderrobinson.com/ | 1 |
| 12:54 | <Philip`> | http://www.soundmedical.com/home.html | 8 |
| 12:54 | <Philip`> | are pages with the given number of <th>s, and no <td>s |
| 12:55 | <Philip`> | (Fun game: guess how many are layout tables!) |
| 12:55 | <zcorpan> | 7? :) |
| 12:58 | <annevk> | you're in on the next round in Markup Stats! |
| 12:58 | Philip` | guesses that those people are just pressing the wrong button in their WYSIWYG editors |
| 13:10 | <Philip`> | krijnh: I think you might need something like white-space:pre in the IRC logs else it messes up my aligned columns :-) |
| 13:21 | <zcorpan> | pre-wrap ? |
| 13:31 | <Lachy> | is there a function in python that reads a file and outputs it, sort of like the include() function in PHP? |
| 13:32 | <Philip`> | print(open("filename").read()) ? |
| 13:33 | <Lachy> | yeah, something like that, but won't I need to close the file afterwards? Or does it do that automatically? |
| 13:34 | <Philip`> | I think it closes when the file object gets garbage-collected, which is at some unknown point in the future but before the program terminates |
| 13:34 | <Philip`> | so it only matters if you've got a long-running process and don't want to run out of file handles |
| 13:35 | <annevk> | there's .close() iirc |
| 13:35 | <annevk> | but i never needed it so far |
| 13:37 | <Philip`> | You can also do |
| 13:38 | <Philip`> | from __future__ import with_statement; |
| 13:38 | <Philip`> | with open("filename") as f: print(f.read()) |
| 13:38 | <Philip`> | Uh, without the semicolon |
| 13:38 | <Philip`> | (but only in Python 2.5) |
| 13:39 | <Philip`> | which will close it immediately |
| 13:40 | <Philip`> | (The explicit close seems quite inelegant compared to Perl) |
| 13:54 | <webben> | Hixie: Ignoring the Hofstader-Clark antagonism, you may find the recent posts at http://blindconfidential.blogspot.com/ (about your blogpost re Jaws and Isolani's response) interesting. Chris Hofstader used to run the JAWS development team. |
| 14:10 | <Philip`> | '"For the best results using a screen reader we suggest you read this site with <some screen reader or another>."' - hmm, that sounds familiar, and I thought most people now agreed it's a bad idea (hence http://www.anybrowser.org/campaign/ ) |
| 14:30 | <webben> | Philip`: I don't think most people agree with the Any Browser principle. Most people paid to do webdev work (from what I've seen) seem to care only about IE + Firefox + maybe Safari. |
| 14:30 | <webben> | Philip`: It's quite hard to persuade people they need to worry about ELinks or Konqueror. |
| 14:32 | <Philip`> | webben: At least that's somewhat better than "Best viewed in Netscape 4, 800x600 resolution, 16-bit color" |
| 14:34 | <webben> | Philip`: Hmm. I dunno. I see a lot of notices saying things only work in Firefox. |
| 14:34 | <webben> | Once Netscape 4 was the biggest and best. |
| 14:34 | <webben> | what really dooms such notices is the inexorable march of time. |
| 14:41 | <webben> | Philip`: See also things like: http://minghong.f2g.net/products/infobar/ |
| 14:46 | <hsivonen> | hmm. contrary to what anne assumed, the cabal acknowledgment has been in the spec since at least May 24 2006 (likely longer, but I have a snapshot from May 24 2006). |
| 14:57 | <Philip`> | Hmm, looking at my old sites, the only browser suggestion is that users should use one with Java such as Netscape Navigator |
| 14:59 | <hsivonen> | Philip`: what did you use applets for? |
| 14:59 | <Philip`> | Wobbly text and marquee-like scrolling banners |
| 14:59 | <Philip`> | It was great! |
| 15:00 | <hsivonen> | ok. my previous statements about applet use cases hold, then :-) |
| 15:03 | <Lachy> | I just migrated this from the blog http://wiki.whatwg.org/wiki/FAQ/pl |
| 15:03 | <Lachy> | now I will set up the redirects |
| 15:03 | Philip` | finds his ancient Theme Hospital site, and uses it to solve a mystery on a Wikipedia talk page about who first patched the "pregnancy" disease into the game |
| 15:04 | <Lachy> | Theme Hospital was an awesome game! :-) |
| 15:05 | <annevk> | indeed |
| 15:06 | hsivonen | has no idea what Theme Hospital is |
| 15:07 | <Philip`> | http://web.archive.org/web/19991008150009/http://www.geocities.com/SiliconValley/Park/7676/pregnant.html - hooray, layout tables |
| 15:08 | <annevk> | so much for cool URIs don't change |
| 15:10 | <Lachy> | annevk, http://blog.whatwg.org/faq/ and http://blog.whatwg.org/faq/pl/ now redirect to the wiki |
| 15:11 | <annevk> | cool, now hopefully people will update :) |
| 15:12 | <Lachy> | I wish I knew of an HTML to wiki markup script. I had to write a simple one myself, but didn't cover everything |
| 15:13 | <Lachy> | would have made migrating the pl version easier |
| 15:13 | <Philip`> | http://www.uni-bonn.de/~manfear/html2wiki-tables.php ? |
| 15:14 | <Lachy> | that looks like it only handles tables, not headings, links, etc. |
| 15:14 | annevk | wonders why people don't ask the question: "why would it be non-conforming to include those characters" |
| 15:14 | <Philip`> | http://diberri.dyndns.org/wikipedia/html2wiki/ ? |
| 15:15 | Lachy | should have googled for one! |
| 15:16 | <Philip`> | Do we have use cases for making the character "q" conforming? Could they be satisfied by only making "q" conforming if it is followed by a "u"? |
| 15:16 | <gsnedders> | Philip`: there are some exceptions to that |
| 15:17 | <Philip`> | gsnedders: Sure, but there's exceptions to everything, and we only need to handle 80% of cases |
| 15:17 | <annevk> | :) |
| 15:18 | <gsnedders> | Philip`: that's true. |
| 15:19 | <annevk> | however, apart from how to construct a DOM with them characters are more or less out of scope |
| 15:20 | <Lachy> | Philip`, there are plenty of cases for a Q not followed by a u. e.g. QANTAS, FAQ. Q&A, etc. |
| 15:21 | <Philip`> | Lachy: Those are all "Q", not "q" |
| 15:21 | <Lachy> | :-) |
| 15:22 | <Lachy> | http://en.wikipedia.org/wiki/List_of_English_words_containing_Q_not_followed_by_U |
| 15:25 | <hsivonen> | http://lists.w3.org/Archives/Public/public-html/2007Aug/1255.html |
| 15:25 | <hsivonen> | hmm. I understood T.V. Raman's point as an indictment of designing specs for current JAWS. |
| 15:26 | <hsivonen> | I wonder if I understood his points right. |
| 15:27 | <Philip`> | Lachy: A few dozen examples, out of half a million words in the English language? The statistics aren't very convincing |
| 15:29 | <Lachy> | Philip`, sure, but qwerty is a very common word these days, and so it makes it into the 80% mark |
| 15:37 | <gsnedders> | Lachy: you look very qwerty today |
| 15:38 | <Lachy> | are you saying I look like a keyboard? |
| 15:38 | <Lachy> | :-) |
| 15:38 | <gsnedders> | I dunno. I'm just wondering what the usage is :) |
| 15:38 | <gsnedders> | Philip`: why don't we only handle 0.0014% of web content (i.e., XML) |
| 15:46 | <Philip`> | http://www.google.com/search?q=qwerty suggests that "QWERTY" and "Qwerty" are much more common than "qwerty" |
| 15:47 | <Philip`> | so I've still not seen any great arguments for why characters should be conforming :-) |
| 15:48 | <Philip`> | (Switching the question from "why should it be conforming?" to "why should it be non-conforming?" isn't very helpful since it's not giving answers to either question) |
| 15:49 | <annevk> | there's a good reason for some characters, such as NULL, to be non-conforming, there's not really a good reason for the other characters to be non-conforming afaict; they're not causing interoperability concerns and it's out of scope of HTML to say what a conforming string of text is |
| 15:50 | <Philip`> | It would be easy if there was a design principle that said "things that don't cause interoperability problems should be conforming" |
| 15:53 | <Philip`> | The null character seems special since it's always(?) converted into U+FFFD, which is not what you'd ever want to happen, so it indicates an authoring error and conformance checkers should complain about it |
| 15:54 | <Philip`> | whereas any(?) other character is passed straight through, regardless of whether it's conforming, unless people want to convert control characters into spaces or something |
| 15:54 | Philip` | isn't sure what happens in HTML5 if you try adding U+0000 through the DOM |
| 15:56 | <annevk> | I think U+0000 doesn't reach the DOM APIs in some browsers and others remove it... |
| 15:56 | <annevk> | I mentioned U+0000 because it's special |
| 16:00 | <Philip`> | It seems IE7/Opera9.5 cut off the string after the \0, Safari3/Firefox3 keep the \0 (and Safari doesn't render it, while FF renders a little [00/00] box) |
| 16:21 | <annevk> | what is the Unicode character that lowercases to "i" again? |
| 16:22 | annevk | needs it to make sure comparisons are done based on ASCII |
| 16:25 | <zcorpan> | İ |
| 16:28 | <zcorpan> | ı might be uppercased to I |
| 16:30 | <annevk> | seems that IE7 does ASCII case-insensitive here: http://tc.labs.opera.com/html/global-attributes/contenteditable/014.htm |
| 16:32 | annevk | adds a test for ı |
| 16:32 | <zcorpan> | annevk: ie also says PASS if you change the value to "blurb" |
| 16:33 | <annevk> | correct |
| 16:33 | <annevk> | which is why it's a good test :) |
| 16:33 | <zcorpan> | aha |
| 16:34 | zcorpan | didn't read carefully |
| 16:34 | <annevk> | although they would fail again if I'm going to test for the specific error... |
| 16:35 | <annevk> | guess I'll do that in another test for now |
| 16:35 | <annevk> | tests are cheap |
| 16:38 | <zcorpan> | <!html html>? :) |
| 16:43 | <annevk> | way ahead of you |
| 17:02 | <annevk> | btw, I wonder if the RFC for script types needs to be updated, after all, server MIME types are ignored and we could just say that the type= attribute takes certain magical values |
| 17:02 | <annevk> | such as text/javascript1.1, etc. |
| 17:11 | <Philip`> | Is the server type ignored if you send text/javascript vs text/vbscript to IE? |
| 17:14 | <annevk> | dunno |
| 17:14 | <annevk> | (read as: please tell me if you find out :) ) |
| 17:16 | <Philip`> | http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!DOCTYPE%20html%3E%0D%0A%3Cscript%20src%3Dhttp%3A%2F%2Fhtml5.lachy.id.au%2Foutput%3Fdata%3D%2527%2B%2527%253B%2Balert%2528%2522JavaScript%2522%2529%253B%2Bif%2B%25280%2529%250D%250Aalert%2528%2522VBScript%2522%2529%26type%3Dtext%252Fvbscript%3E%3C%2Fscript%3E%20 |
| 17:16 | <Philip`> | It is ignored |
| 17:17 | <Philip`> | (IE assumes JavaScript regardless of the Content-Type) |
| 17:17 | <Philip`> | (unless you do <script type=text/vbscript> etc) |
| 17:17 | <Philip`> | (IE7, at least) |
| 17:18 | <Philip`> | It's odd how few people use VBScript |
| 17:23 | <annevk> | the spec doesn't seem to consider script elements without either type or language attributes... |
| 17:44 | <annevk> | made some script tests: http://tc.labs.opera.com/html/script/ |
| 18:01 | <Philip`> | http://canvex.lazyilluminati.com/misc/stats/scripttypes.html |
| 18:02 | <Philip`> | More text/html than text/javascript? That's quite odd |
| 18:04 | <Philip`> | http://www.artesea.co.uk/cnps/blogs/artesea.js - I wish there was a programming language where you could put images in the source code like that |
| 18:08 | <PHPechowiec> | Is there someone from poland? |
| 18:12 | <annevk> | Philip`, not really, application/x-javascript is the default |
| 18:13 | <annevk> | although I suppose it's weird :) |
| 18:13 | <annevk> | I updated my two tests btw to make them slightly more useful |
| 18:15 | <annevk> | seems that IE also executes: "javascript 1.3" and "text/javascript " etc. |
| 18:15 | <Philip`> | http://www.assoc-amazon.co.uk/s/ads.js - that's odd - it sends a 404 with Content-Type: image/gif if you do a HEAD request, but a 200 text/javascript if you do GET |
| 18:17 | <gsnedders> | Philip`: oh, that's nothing that odd |
| 18:18 | Philip` | wonders if he should just do GET and try to abort the connection before it's downloaded the body |
| 18:19 | <annevk> | that's prolly better if you want accurate results |
| 18:20 | <annevk> | btw, it seems for type= and language= parsing some browsers simply use startswith("text/javascript") ... |
| 18:20 | <PHPechowiec> | where I can find about HTML5 inpolish |
| 18:20 | <PHPechowiec> | **in polish |
| 18:21 | <Philip`> | annevk: I didn't see that behaviour for type= in any browser |
| 18:22 | <annevk> | I only see it when I construct my tests wrong |
| 18:22 | <annevk> | I made a small python script to generate the tests and then I managed to turn some stuff around |
| 18:22 | <annevk> | so, oops |
| 18:24 | gsnedders | is going to fail Computing having missed the last two lessons — not knowing what the (wrong) answers are sucks :P |
| 18:26 | <Philip`> | http://canvex.lazyilluminati.com/misc/stats/scripttypes2.html - hmm, it's a bit better when I use GET and when I correctly stop processing 404s |
| 18:27 | <Philip`> | http://ads49.bpath.com/gif.cfm?ID=266432&Page=0&Ver=40 - that one really is an image |
| 18:28 | <Philip`> | but http://alissathechihuahua.freeservers.com/index.html includes it via <script src> |
| 18:28 | <Philip`> | Hmm, Alissa looks rather ugly |
| 18:30 | <Philip`> | http://environmentalchemistry.com/JavaScripts/Primary.js - how does that end up as text/css? |
| 18:34 | <annevk> | seems that there are almost no parameters around looking at your stats for <script type> and <script language> |
| 18:36 | <Philip`> | The few users of language=javascript1.5 that I looked at were browser-JS-version-detection user-stats-tracking scripts, so it seems those aren't used for anything important either |
| 18:39 | <annevk> | it's quite a big mess |
| 18:39 | <annevk> | and such a tiny feature... :( |
| 18:41 | <Philip`> | Can we just say that <script> means JavaScript and UAs can ignore everything else? :-) |
| 18:42 | <Philip`> | It's not like anybody really uses any other language |
| 18:43 | <annevk> | some people use <script type=text/xml> or something |
| 18:45 | <Philip`> | That wouldn't work too terribly if you parsed it as JS into an E4X object and then, er, hacked in a compatible API somewhere |
| 18:47 | <Philip`> | http://www.americasalfalfa.com/ - I can't quite tell what they're using text/xml for |
| 18:47 | <Philip`> | http://www.kinderklinik-buch.de/Neonatologie/Richtlinien.html - they're using it for exactly the same |
| 18:48 | <Philip`> | (Those are the only two sites I saw) |
| 18:49 | <Philip`> | (Still, it'd be nasty to prevent other languages or language versions, even if they're not used very commonly) |
| 18:50 | <zcorpan> | how do you allow other langauges for onclick=""? do we need to? |
| 18:51 | <zcorpan> | html4 says content-script-type, but i think browsers ignore it |
| 20:40 | <Hixie> | anything interesting happening in the world? |
| 20:41 | <annevk> | http://news.google.com/ ? |
| 20:54 | <Hixie> | i meant in our screwed up corner of it :-) |
| 20:55 | <annevk> | oh, people played with tables and <script type language> |
| 20:56 | <annevk> | i pinged the guy on the forms TF who hasn't introduced himself yet as you suggested |
| 21:06 | <Hixie> | i guess i better do some work on datatemplates tonight |
| 21:06 | <Hixie> | bbiab |
| 22:03 | hsivonen | wonders how to force apache to respond with 503 with apache config only without having to resort to scripting |
| 22:10 | <Dashiva> | On what condition? |
| 22:12 | <Dashiva> | I know you can use mod_rewrite, but it's kinda hacky |
| 22:12 | <hsivonen> | hmm. RedirectMatch 503 ^/ might do the trick |
| 22:12 | <hsivonen> | Dashiva: on the condition that required modules did not load |
| 22:13 | <hsivonen> | e.g. mod_jk did not load |
| 22:13 | <hsivonen> | use case: admin updates Apache, forgets to reinstall mod_jk |
| 22:14 | <Dashiva> | Redirect 503 / should work too |
| 23:04 | <jgraham> | html5lib just reached rev. 1000 |
| 23:05 | <annevk> | party? |
| 23:11 | <annevk> | jgraham, ah, so you're building a table model using the HTML5 algorithm and made various heading algorithms on top of that? |
| 23:15 | <jgraham> | annevk: Yeah (to both ;) ) |
| 23:18 | <hsivonen> | after talks about an "IRC cabal", I find http://wilbur.bytowninternet.com/pipermail/list_html4all.org/ interesting |
| 23:19 | <annevk> | there's also http://html4all.org/pipermail/list_html4all.org/ ... |
| 23:19 | <annevk> | I guess they're the same |
| 23:45 | <hsivonen> | http://html4all.org/pipermail/list_html4all.org/2007-August/000032.html |
| 23:45 | <hsivonen> | wow. |
| 23:47 | <annevk> | the "html4all cabal" |
| 23:48 | <takkaria> | it's a cabal-eat-cabal world... |
| 23:54 | hsivonen | finds that his relative openness to ideas compared to annevk is assessed at http://www.w3.org/2002/09/wbs/40318/dprv/results |
| 23:55 | <hsivonen> | doh. edit in mid sentence |
| 23:55 | <hsivonen> | s/is assessed // |
| 23:57 | <annevk> | I'm open to changes, I just don't really mind the current text |
| 23:58 | <annevk> | (also taking into account that perfection is not needed for a first draft, imo) |