| 00:00 | <jgraham> | I don't understand why html 4 suggests cells that are both data and headings should be marked as data |
| 00:00 | <jgraham> | It seems much more useful to mark them as headings |
| 00:00 | <annevk> | would you say they're headings? |
| 00:00 | <annevk> | the location column |
| 00:00 | <jgraham> | Arguably, yes. |
| 00:01 | <jgraham> | Although I think the table makes sense if you don't mark them as such |
| 00:02 | <jgraham> | Anyway, bedtime |
| 00:03 | <annevk> | prolly a good idea |
| 00:43 | <Philip`> | Hmm, Opera 9.5 has fixed the BGRA vs RGBA ordering in getImageData, but it looks like it's still the wrong way round in putImageData |
| 00:45 | <annevk> | ouch |
| 00:45 | <annevk> | I still haven't gotten around testing those methods yet (thoroughly at least); also because I don't really like how they're currently defined |
| 00:46 | <annevk> | If you could file a bug that'd be nice |
| 00:46 | annevk | -> bed |
| 00:46 | <Philip`> | http://canvex.lazyilluminati.com/misc/putimgdata_rgb.html |
| 00:47 | Philip` | will do |
| 01:05 | <Hixie> | heh |
| 01:05 | <Hixie> | i just discovered something |
| 01:05 | <Hixie> | the first mention of smell-o-vision in the htmlwg and whatwg irc archives that i can find is actually from DanC |
| 01:05 | <Hixie> | http://krijnhoetmer.nl/irc-logs/html-wg/20070426#l-517 |
| 01:06 | <Lachy> | ha! |
| 01:08 | <hober> | Yeah, but he's not a cabal member, so it's not offensive when he does it. :/ |
| 01:08 | <Lachy> | who keeps the membership lists for the cabal? |
| 01:10 | <Hixie> | i wonder if S. Mike Dierken and Mike Dierken are the same person |
| 01:11 | <kingryan> | Lachy: we developed a microformat for cabal lists, so the list can be distributed |
| 01:14 | <Lachy> | oh right, so to join, I just markup my membership details using hCard and rel=cabal? |
| 01:14 | <Lachy> | :-) |
| 01:14 | <kingryan> | yeah |
| 01:26 | <Philip`> | http://canvex.lazyilluminati.com/misc/radial.html is an interestingly crazy effect in O9.5 |
| 02:08 | <Hixie> | i have got to stop responding to john's e-mails |
| 02:08 | <Hixie> | (responses in www-archive for the morbidly curious) |
| 02:09 | <Hixie> | (his e-mails aren't publicly archived as far as i can tell -- he only cc'ed the secret html4all cabal -- membership list at http://html4all.org/wiki/index.php/Special:Listusers ) |
| 02:14 | <othermaciej> | omg, he has an *actual* secret cabal! |
| 02:15 | <Philip`> | (Isn't just the list of people who've bothered to register on the wiki (which doesn't appear to be restricted)?) |
| 02:15 | <Philip`> | s/ / that / |
| 02:24 | <Philip`> | It seems a bit of a waste of time to write <acronym title="information and communication technology">ICT</acronym> when JAWS apparently reads it like "icked" |
| 02:26 | <Philip`> | (At least that suggests that people are happy to write things that are unhelpful today, in the hope that they will be helpful in the future when the tools catch up) |
| 02:30 | <karlUshi> | interesting |
| 03:07 | Philip` | guesses http://html4all.org/pipermail/list_html4all.org/ is the relevant list |
| 03:20 | <Lachy> | Philip`, how did you find that URL? |
| 03:21 | <Lachy> | did you just guess it? I was looking for it earlier |
| 03:35 | <Philip`> | Lachy: http://html4all.org/wiki/index.php/New_Logos_Archive_and_discussion_page points to list⊙ho, so it's just the public archives from that |
| 03:36 | <Lachy> | yeah, I knew about the mailing list. Didn't know how to find the archives |
| 03:36 | <Philip`> | (Note to whoever's reading the logs: if you don't want to work in the open, I think Mailman's "Archiving Options" -> archive_private is the relevant setting) |
| 03:36 | <Lachy> | you shouldn't tell them that! |
| 03:37 | <Lachy> | if they do, we won't be able to follow their discussions |
| 03:38 | <Philip`> | We shouldn't follow their discussions if they don't want us to |
| 03:40 | <Philip`> | Incidentally, I like some of their logos - I think the WHATWG's "(?)" is rather too plain and boring |
| 03:44 | <Lachy> | the current whatwg logo was an improvement from the original one that Hixie had made |
| 03:45 | <Philip`> | Was the original one a cat? |
| 03:45 | <Lachy> | no |
| 03:47 | <Philip`> | Oh :-( |
| 03:49 | <Lachy> | http://web.archive.org/web/20040529190138/http://www.whatwg.org/ |
| 03:49 | <Lachy> | http://web.archive.org/web/20040714070149/http://www.whatwg.org/ |
| 03:51 | <Philip`> | The red circled question mark isn't bad, but the tiny writing and oddly sloped "Working Group" beside it look a bit suspect :-) |
| 04:18 | <gavin> | http://html4all.org/pipermail/list_html4all.org/2007-September/000197.html |
| 04:18 | <gavin> | haha |
| 04:20 | <gavin> | it's like a game of hide and seek with mailing lists and IRC |
| 04:21 | gavin | goes off to start a secret list⊙wo mailing list |
| 04:29 | <Philip`> | It's great how modern technology lets you have a conversation where each message goes from IRC to web-based log to mailing list to web archive then round the circle again, probably with some diversions through other mailing lists every so often |
| 04:30 | <Philip`> | We just need to integrate it with Twitter somehow |
| 04:30 | <Lachy> | quick, someone twitter it as well! |
| 04:30 | <Hixie> | what was the e-mail in which we were accused of being a cabal? |
| 04:31 | <gavin> | http://lists.w3.org/Archives/Public/public-html/2007Aug/1006.html |
| 04:32 | <Hixie> | thx |
| 04:32 | <Philip`> | http://html4all.org/pipermail/list_html4all.org/2007-August/000074.html has various cabal-related bits |
| 04:34 | <Philip`> | (http://html4all.org/pipermail/list_html4all.org/2007-August/000004.html too - they're not denying its existence) |
| 04:36 | <Lachy> | Hixie, why did you take the twitter form off whatwg.org? |
| 04:36 | <Lachy> | oh, no, it's still there |
| 04:50 | <Hixie> | http://html4all.org/pipermail/list_html4all.org/2007-September/000164.html |
| 04:53 | <Hixie> | (http://wilbur.bytowninternet.com/mailman/listinfo/list_html4all.org to subscribe, apparently) |
| 04:54 | <Hixie> | eep, i'm on the archives of that mailing list! |
| 05:00 | <Lachy> | if we subscribed, would they kick us off? |
| 05:00 | <Hixie> | dunno :-) |
| 05:00 | <Hixie> | i have no intention of subscribing |
| 05:00 | <Hixie> | ooo |
| 05:00 | <Hixie> | an e-mail from freedom scientifir |
| 05:00 | <Hixie> | freedom scientific |
| 05:00 | Hixie | reads |
| 05:00 | <Lachy> | nor do I, I follow enough mailing lists :-) |
| 05:01 | <Hixie> | ooo, they forwarded my e-mail on to someone |
| 05:01 | <Lachy> | Hixie, was it a private email or archived somewhere? |
| 05:01 | <Hixie> | private |
| 05:01 | <Lachy> | ok |
| 05:01 | <Hixie> | i e-mailed them my set of tests earlier |
| 05:03 | <Lachy> | Hixie, did you see this? http://www.isolani.co.uk/blog/access/ScreenReadersAWebDeveloperFailure |
| 05:10 | <Hixie> | i'm replying to it now |
| 07:21 | <Lachy> | Hey Hixie, to create a persistent connection, e.g. for <event-source), is the best way to do that with a loop like this: while (true) { sleep(10); /* check for new events and send them ... */ } |
| 07:24 | <Hixie> | no |
| 07:24 | <Lachy> | ok, how would I do it then? |
| 07:24 | <Hixie> | you'd want some sort of select() loop or other os-supported mechanism like waiting on a semaphor |
| 07:24 | <Hixie> | or other signalling mechanism |
| 07:25 | <Hixie> | a loop would be very cpu intensive |
| 07:25 | <marcosc> | you really should not need a loop |
| 07:25 | <marcosc> | as hixie said |
| 07:26 | <Lachy> | ok, how do you do it with your whatwg.org/status/ app? |
| 07:26 | <Hixie> | you mean issues/listen ? |
| 07:26 | <Hixie> | or? |
| 07:26 | <Lachy> | yes |
| 07:27 | <othermaciej> | Lachy: how/where are you trying to implement it? |
| 07:27 | <Lachy> | I was just wanting to do a few experements |
| 07:27 | <Lachy> | on my own server |
| 07:28 | <Hixie> | it uses blocking I/O |
| 07:28 | <marcosc> | seems that the server is not closing the connection... |
| 07:28 | <Hixie> | i can put the code up somewhere hold on |
| 07:29 | <Lachy> | is it in perl? |
| 07:29 | <Hixie> | yeah |
| 07:30 | <Lachy> | perl is very cryptic, but ok :-( |
| 07:30 | <othermaciej> | Lachy: you want to make a server-side script that's meant to connect to an event-source? |
| 07:30 | <marcosc> | my mum codes in perl :) |
| 07:30 | <Hixie> | http://damowmow.com/temp/listen and http://damowmow.com/temp/common.pm |
| 07:31 | <Lachy> | othermaciej, yeah, just wanted to test out <event-source> in opera |
| 07:31 | <othermaciej> | a loop that sleeps won't hog the CPU |
| 07:31 | <othermaciej> | but it won't be that responsive, if it is polling for events somewhere |
| 07:31 | <Hixie> | a loop that sleeps will not scale |
| 07:32 | <Lachy> | yeah, I didn't think it would |
| 07:32 | <Hixie> | it will quickly hog the cpu if you have many |
| 07:32 | <Hixie> | which is what i meant |
| 07:32 | <Hixie> | even one alone is bad practice, since it prevents the cpu from stepping down, etc |
| 07:32 | <othermaciej> | how to do it depends on where your events are coming from |
| 07:32 | <Hixie> | yes |
| 07:33 | <Hixie> | Lachy: did the code help? |
| 07:33 | <othermaciej> | sleep() doesn't prevent the CPU from stepping down afaik |
| 07:33 | <othermaciej> | sleep yields the CPU until the interval has passed |
| 07:33 | <Lachy> | Hixie, I'm just trying to decrypt it :-) |
| 07:33 | <Hixie> | hehe |
| 07:33 | <Hixie> | othermaciej: it doesn't prevent it from stepping down while sleeping, no |
| 07:34 | <othermaciej> | but yeah, if events are infrequent, you don't want your process to wake all the time |
| 07:34 | <Hixie> | othermaciej: but it forces the cpu to run every Nms even if nothing has happened |
| 07:34 | <Hixie> | right |
| 07:36 | <othermaciej> | if your event-source script is ultimately getting events from a network connection, you can just read off of that network connection |
| 07:36 | <othermaciej> | if it is a single connection, just blocking I/O should do |
| 07:36 | <othermaciej> | if you are multiplexing, you'd want to use select(), or multithreading |
| 07:38 | <othermaciej> | if it's just for testing purposes, you could just send a fixed set of events with predetermined intervals in between |
| 07:39 | <Lachy> | othermaciej, that's what I was thinking about doing, but I was curious how else it could be done |
| 07:40 | <othermaciej> | well if it's just a test case, using sleep is fine |
| 11:17 | <zcorpan> | Hixie: yt? |
| 11:54 | <Philip`> | I like Twisted for asynchronous network IO (in Python) |
| 11:54 | <Philip`> | (though I've not tried using it for event-source in particular) |
| 11:55 | <Philip`> | (and it does take quite a while to work out how Twisted works) |
| 11:57 | <virtuelv> | Philip`: twisted works great for event-source |
| 11:58 | <MikeSmith> | new SVG viewer for MSIE released here today in Japan |
| 11:58 | <MikeSmith> | http://blog.svg-map.com/2007/09/svg_map_toolkit.html |
| 12:04 | <Philip`> | Is there meant to be a decent two-way communication system somewhere (e.g. for real-time multiplayer games)? event-source only seems to cover one direction, and I couldn't find a reliable way to go the other way |
| 12:10 | <annevk> | I think the idea was to extend TCPConnection to P2P at some point (and also let it cover bluetooth and such) but that's just an idea |
| 12:11 | <othermaciej> | TCPConnection could be used as a two-way protocol to talk to a server |
| 12:11 | <Philip`> | I wouldn't expect P2P to be too useful in most cases, since firewalls tend to get in the way |
| 12:11 | <othermaciej> | now that I think about it, TCPConnection and event-source do awfully similar things but in almost completely different ways |
| 12:11 | <othermaciej> | seems kinda random |
| 12:12 | <othermaciej> | P2P in the browser == instant botnet |
| 14:29 | <gsnedders> | Hixie: ping |
| 14:40 | <MikeSmith> | annevk - when is XHR spec expected to go to CR? |
| 14:41 | <hsivonen> | MikeSmith: btw, I'm implementing text output now |
| 14:42 | <MikeSmith> | hsivonen - cool. very happy to hear that |
| 14:43 | <zcorpan> | text output? |
| 14:44 | <hsivonen> | zcorpan: validator output in text/plain as opposed to text/html |
| 14:50 | <annevk> | MikeSmith, needs to go to LC first and there are some outstanding issues |
| 14:52 | <zcorpan> | hsivonen: aha |
| 15:01 | <MikeSmith> | annevk - so any idea how soon it might go to LC? |
| 15:04 | <annevk> | "when it's ready" |
| 15:04 | <annevk> | this likely involves someone creating lots of additional testcases to make sure we're ready although maybe we just try to go to LC again like last time... |
| 15:04 | <annevk> | dunno |
| 15:08 | zcorpan | should do that some time |
| 15:09 | <Philip`> | Hmm, I can (in Opera) do <canvas width=0> and it stays 0, but if I do setAttribute('width', 0) then it goes to 300 |
| 15:41 | <hsivonen> | MikeSmith: http://html5.validator.nu/?doc=http%3A%2F%2Fabout.validator.nu%2F&out=text |
| 15:42 | MikeSmith | trying now |
| 15:42 | <MikeSmith> | sweet |
| 15:42 | <hsivonen> | whew. that was a slow deployment. still fallout of the outage from the 21st of last month to fix |
| 15:42 | <MikeSmith> | hsivonen - hmm, any way to put the error message and line numbers on one line? |
| 15:44 | <hsivonen> | MikeSmith: I guess I could do that, but this format is supposed to be human-readable as opposed to machine-parseable |
| 15:46 | <hsivonen> | MikeSmith: the service will requild itself in a moment |
| 15:46 | <hsivonen> | (I need to figure out how to make these updates less disruptive by using mod_jk load balancing) |
| 15:46 | <Lachy_> | hsivonen, that looks like it's quite well designed for machine parsing anyway |
| 15:48 | <Lachy_> | shouldn't it output the erroneous code for reference? e.g. for the bad attribute value, so users can easily see what the mistake is before going and looking at the markup |
| 15:49 | <hsivonen> | Lachy_: yes, once I first implement support for that feature in any output format at all |
| 15:49 | <hsivonen> | Lachy_: on my near-future todo list |
| 15:49 | <Lachy_> | oh, ok |
| 15:56 | <hsivonen> | MikeSmith: fixed |
| 16:04 | <zcorpan> | hsivonen: when the document passes, it might be useful to know if it was html or xhtml |
| 16:04 | <hsivonen> | zcorpan: yeah |
| 16:05 | <hsivonen> | zcorpan: umm. it alread tells it |
| 16:06 | <zcorpan> | ah. indeed |
| 16:07 | <hsivonen> | only on the (X)HTML5 side, though |
| 16:07 | <hsivonen> | the generic side should emit Info: blah blah |
| 16:08 | <zcorpan> | yeah, looks good |
| 16:18 | <hsivonen> | annevk: POSTing doc in entity body now supported by validator.nu |
| 16:18 | <hsivonen> | annevk: curl -d 'uoeua' -H 'Content-Type: text/html' 'http://html5.validator.nu/?out=text'; |
| 16:19 | <hsivonen> | annevk: compare with: curl -d '<!DOCTYPE html><title></title>' -H 'Content-Type: text/html; charset=utf-8' 'http://html5.validator.nu/?out=text'; |
| 16:33 | <MikeSmith> | hsivonen - now I'm getting the error message on first line, line number on next, column number on next |
| 16:33 | <MikeSmith> | 3 lines instead of 1 ... |
| 16:33 | <MikeSmith> | that is by design? |
| 17:02 | <hsivonen> | MikeSmith: sorry. I misread your request |
| 17:03 | <hsivonen> | I'd rather not put the line and col on the same line as the message |
| 17:03 | <hsivonen> | if I put them after, the message may be so long that the numbers are hard to glance at |
| 17:04 | <hsivonen> | if I put them before, the text before the message will vary |
| 17:04 | <hsivonen> | as it is possible to have no number, only line or both line and col |
| 17:05 | <MikeSmith> | hsivonen - I see the issues with that. Wondering if in spite of that, if there's any way you could you provide single-line output as an option. |
| 17:05 | <MikeSmith> | I understand the need to make the output human-readable |
| 17:06 | <hsivonen> | MikeSmith: would xml or JSON be overkill for your use case? |
| 17:06 | <MikeSmith> | hsivonen - yeah, for the use case I had in mind, it would be |
| 17:07 | <MikeSmith> | The use case behind my original request was to make the output easily consumable by editing apps |
| 17:07 | <hsivonen> | MikeSmith: ok. I'll change the format |
| 17:08 | <hsivonen> | MikeSmith: for that use case, XML or JSON should be used |
| 17:08 | <hsivonen> | this format is too ambiguous and unstable |
| 17:09 | <MikeSmith> | hsivonen - ideally, yeah. But there is a certain standard format that many *nix apps at least already have support for parsing out |
| 17:09 | <MikeSmith> | Emacs-style or something |
| 17:09 | <hsivonen> | MikeSmith: do you have a spec for the desired output? |
| 17:10 | <MikeSmith> | hsivonen - looking now |
| 17:11 | <MikeSmith> | hsivonen - couldn't find anything right away now, but will look again later |
| 17:12 | <MikeSmith> | I'm headed off to sleep (1:19am here now) |
| 17:14 | <hsivonen> | MikeSmith: ok |
| 18:04 | <hsivonen> | isn't there a way to make the python httplib handle redirects? |
| 18:06 | <annevk> | there should be... |
| 18:06 | <hober> | httplib2 does it... |
| 18:06 | <Lachy_> | hsivonen, you can just do redirects like this in python: |
| 18:06 | <Lachy_> | print "Status: 302 Found" |
| 18:06 | <Lachy_> | print "Location: ..." |
| 18:06 | <annevk> | it's not about making, it's about handling |
| 18:06 | <hsivonen> | Lachy_: on the client |
| 18:09 | <hsivonen> | hober: I'm trying to make a simple script that works with the standard lib |
| 18:11 | <hsivonen> | http://about.validator.nu/html5check.py |
| 18:14 | <hsivonen> | annevk: well, now there is a way to upload files. still not through the browser, though |
| 18:15 | <annevk> | you should probably add .htm and .xht |
| 18:15 | <Lachy> | wow, this is a very nice response from John Foliot! http://www.w3.org/mid/004301c7f0a7$bdc00070$643d42ab@Piglet |
| 18:16 | <hsivonen> | annevk: they are *so* last century :-) |
| 18:16 | <hsivonen> | annevk: but yeah |
| 18:18 | <annevk> | and .xml |
| 18:18 | <annevk> | and maybe others... |
| 18:32 | <Lachy> | hsivonen, you should output the validation results to stdout instead of stderr. stderr should be for errors in the program itself |
| 18:42 | annevk | is afraid his pictures will remain inaccessible |
| 18:42 | <annevk> | adding alt= is just too much work |
| 18:44 | <takkaria> | anyone else find it amusing that the first thing in the list archives for html4all is making a logo? |
| 18:54 | <Lachy> | hsivonen, have you added Content-Access-Control: allow <*> to the validator response so it can be done from XHR? |
| 18:55 | <annevk> | that should be Access-Control:allow <*> |
| 18:55 | <annevk> | please code to the standards! |
| 18:55 | <annevk> | (well, editor drafts) |
| 18:55 | <Lachy> | oh, why did it change? |
| 18:55 | <Lachy> | I was looking at the /TR/ page |
| 18:56 | <annevk> | because people wanted it to be shorter and I agreed |
| 18:57 | <annevk> | http://lists.w3.org/Archives/Public/public-appformats/2007Aug/0010.html |
| 18:57 | <Philip`> | takkaria: That sounds like an unusual definition of "first" - the earliest messages are about it being a new list and wondering what to do about the old one |
| 18:58 | <Lachy> | ok. I've updated my own script that uses it now :-) |
| 18:59 | <takkaria> | Philip`: logo idea is 13th post, on the second day of the list's archived posts |
| 19:00 | <takkaria> | that's close enough to first for me |
| 19:01 | <Philip`> | It'd be great if someone could compete in the Olympics and get a gold medal for coming 13th since it's close enough to first ;-) |
| 19:01 | <takkaria> | :) |
| 19:02 | <takkaria> | I've just been in groups of various sorts before where one of the first things people have talked about is getting a good logo |
| 19:02 | <takkaria> | I mean really, what kind of cabal needs a logo? it goes against the grain |
| 19:05 | <Philip`> | A cabal with a wiki with a logo like http://html4all.org/wiki/skins/common/images/wiki.png needs one |
| 19:07 | <kingryan> | a cabal needs a secret handshake before a logo |
| 19:09 | <Philip`> | What's up with http://html4all.org/wiki/index.php?title=MediaWiki:Common.css&action=raw&ctype=text/css&smaxage=18000 ? That doesn't really look like sensible CSS |
| 19:09 | annevk | needs history.pushState() |
| 19:09 | <annevk> | bookmarks like http://anne.is.weggeweest.nl/image-viewer#2006,madrid-waf,1 are dubious |
| 19:19 | <Hixie> | sweet, they actually are going to try to fix the bugs i reported, in jaws9 |
| 19:20 | <annevk> | nice |
| 19:29 | <Hixie> | people need to stop seeing statements of fact and assuming that i forgot to make them requiremenets |
| 19:31 | <annevk> | that would probably require pointers to the place where the requirement is made |
| 19:31 | <Philip`> | People should assume you never make mistakes and trust that everything is perfect without needing to ask questions? :-) |
| 19:32 | <Hixie> | yes! :-P |
| 19:33 | <Dashiva> | Philip`: Not much point in being sinister overlord of the cabal otherwise, is there? |
| 19:33 | <kingryan> | Hixie: I think you need to be more sinister in order to get people to stop asking questions. ;) |
| 19:37 | <Hixie> | annevk: ping |
| 19:42 | <annevk> | Hixie, pong |
| 19:43 | <Hixie> | hey |
| 19:43 | <Hixie> | so i'm looking at xhr2 |
| 19:43 | <Hixie> | regarding the constructor |
| 19:44 | <Hixie> | would it make snese to have var x = new XMLHttpRequest(url, method, data, onreadystatechange); ? |
| 19:45 | <annevk> | sounds interesting |
| 20:02 | <virtuelv_> | Hixie: I'd like to see that as well |
| 20:05 | <annevk> | would you still be able to do var x = new XHR(...); x.onload = ...; as in, would the request be done after the current script block is executed? |
| 20:08 | <jgraham> | Am I right in thinking that screenreaders don't read display:none content? |
| 20:08 | <billmason> | That is correct. |
| 20:09 | <virtuelv_> | annevk: possibly? |
| 20:09 | <jgraham> | Thanks |
| 20:10 | <virtuelv_> | I'd still like to be able to actually specify and send it in one operation |
| 20:10 | <annevk> | virtuelv_, I believe that's the case with img.src = ...; img.onload = ...; for instance... |
| 20:10 | <annevk> | although not necessarily so in all implementations... |
| 20:10 | <annevk> | well yeah, Hixie's constructor allows that already |
| 20:12 | <Dashiva> | It looks quite a similar to the one I suggested, so I'm all for it |
| 20:15 | <takkaria> | 6 |
| 20:15 | <takkaria> | er, sorry, cat on keyboard |
| 20:17 | <Hixie> | annevk: well, that's why i included the handler in the constructor |
| 20:18 | <hsivonen> | annevk: extension fixed in my local copy |
| 20:18 | <hsivonen> | Lachy: stderr fixed in my local copy |
| 20:19 | <hsivonen> | Lachy: what are the implications of Access-Control? |
| 20:19 | <hsivonen> | Lachy: should I allow XHR only to XML and JSON? |
| 20:20 | <hsivonen> | hmm. RFC 2616 tells me I MUST NOT silently follow redirects on POST |
| 20:20 | <hsivonen> | should I care? |
| 20:20 | annevk | wonders what the scenario is |
| 20:21 | <hsivonen> | annevk: the html5check.py script |
| 20:22 | <hsivonen> | annevk: if html5.validator.nu stops being the preferred location of the service |
| 20:22 | <annevk> | ah, I see |
| 20:23 | <annevk> | you probably should, so they don't send their data to some weird place instead if html5.validator.nu is hijacked |
| 20:23 | <hsivonen> | ok |
| 20:23 | <hsivonen> | (or on a bad WLAN setup) |
| 20:24 | <annevk> | actually, if html5.validator.nu is hijacked they're screwed anyway |
| 20:24 | <hsivonen> | yes |
| 20:29 | <annevk> | prolly most practical is just to distribute a new script and give some sort of stable link where they can find updates for the script |
| 20:57 | <hsivonen> | http://about.validator.nu/html5check.py now does redirects but the code is ugly |
| 20:59 | <annevk> | shouldn't it handle 303 and 307 too? |
| 20:59 | <Hixie> | if anyone wants to reply to http://html4all.org/pipermail/list_html4all.org/2007-September/000198.html (also at http://lists.w3.org/Archives/Public/www-archive/2007Sep/0014.html) then feel free to, i can't see what else i can say to explain why his approach isn't going to have the best accessibility impact. |
| 21:00 | <kingryan> | Hixie: I don't see what more could be said. |
| 21:00 | <Hixie> | jgraham: http://james.html5.org/cgi-bin/tables/table_inspector.py?uri=http%3A%2F%2Fgoogle.com%2F&algorithm=experimental&scope=1&headers=1&b_headings=1&strong_headings=1 crashes |
| 21:01 | <Hixie> | kingryan: yeah me either, and my replying is taking time away from editing the spec |
| 21:01 | <kingryan> | Hixie: yeah, it's time to give up |
| 21:02 | <hsivonen> | annevk: 303, no |
| 21:02 | <annevk> | Hixie, he does seem to claim that some alt= text (even really bad alt text) is better than none |
| 21:03 | <annevk> | I'm not sure what that's based on, but then I'm not sure what the current text in the HTML5 spec is based on either |
| 21:03 | <hsivonen> | annevk: added 307 |
| 21:03 | <Hixie> | he seemes to be suggesting that duplicating content elsewhere on the page is good alt text, which seems clearly untrue to me |
| 21:03 | <kingryan> | Hixie: here's the part that gets me: "...but making it too easy to invoke the exception dilutes the rule. ". As if stronger rules are more likely to be followed, just because they're stronger, more stringent. |
| 21:03 | <kingryan> | its a very legalistic view of conformance |
| 21:03 | <Hixie> | kingryan: yeah, i didn't understand that either |
| 21:04 | <hsivonen> | kingryan: many people have a legalistic view of conformance |
| 21:04 | <kingryan> | to follow his analogy, some people just won't wear seatbelts, no matter how strong the laws |
| 21:04 | <hober> | I liked the part about free beer. |
| 21:04 | <kingryan> | hsivonen: I agree. Unfortunately that's counter productive. |
| 21:05 | <Dashiva> | I want free candy instead |
| 21:05 | <Dashiva> | Stop the beer monopoly on analogies |
| 21:05 | <kingryan> | I don't remember when, but I at one point, finally, came to realize that conformance is just like lint-checking. |
| 21:05 | annevk | missed the free beer |
| 21:05 | <kingryan> | it's just a tool for building better tools |
| 21:07 | <hsivonen> | kingryan: validator.nu is two-point-ohey in a participatory way: it lets you load in your own rules (if they are describable as RNG or Schematron) and it lets you turn off any Java-based checker |
| 21:08 | <kingryan> | hsivonen: that's nice |
| 21:08 | <kingryan> | ironically, /me just finished porting html5lib's python validator to ruby |
| 21:08 | <hsivonen> | does html5lib now have a validator? what does it do? |
| 21:09 | <kingryan> | yes, markp wrote it: http://html5lib.googlecode.com/svn/trunk/python/src/html5lib/filters/validator.py |
| 21:09 | <annevk> | it checks content models |
| 21:09 | <annevk> | maybe more |
| 21:09 | <kingryan> | tags, attributes (required and forbidden), attribute formats |
| 21:10 | <kingryan> | by tags I mean allowed and forbidden elements |
| 21:10 | <kingryan> | enumerated attribute values |
| 21:11 | <hsivonen> | hmm. that looks more compact than the RNC schema |
| 21:12 | <kingryan> | lunchtime, bbiab |
| 21:13 | <Hixie> | at some point i'll need to write a validator |
| 21:16 | <hsivonen> | Hixie: why? |
| 21:16 | <Hixie> | i want to try an approach that doesn't use data tables |
| 21:17 | <Hixie> | and see if it is better or worse than the more common approach of data tables with exceptions |
| 21:28 | <jgraham> | Hixie: Thanks for the bug report. That's what I get for accidentally rolling untested changes into updates :) |
| 21:42 | <jgraham> | Hixie: Crash is fixed |
| 21:42 | jgraham | needs to check if the <td><b> stuff actually works |
| 21:53 | <zcorpan> | anyone have jero (ph5p) 's email address? |
| 21:55 | annevk | posted about the table inspector |
| 21:56 | <annevk> | http://jero.net/contact |
| 21:56 | <annevk> | oh, nm |
| 21:57 | <zcorpan> | or anyone know how to use ph5p? |
| 21:58 | <Dashiva> | ph5p? |
| 21:58 | <zcorpan> | http://jero.net/lab/ph5p/ |
| 22:02 | <kingryan> | Hixie: i'm a validator n00b, what other approaches are there besides data tables? |
| 22:34 | Philip` | tries reporting quite a few Opera canvas bugs, and apologises for any that are dupes |
| 22:37 | <Dashiva> | Philip`: Maybe you should run it through zcorpan or annevk? |
| 22:37 | <Dashiva> | I'm going to bed, so can't help right now :) |
| 22:37 | <zcorpan> | don't worry about reporting dups :) |
| 22:37 | <zcorpan> | better safe than sorry :) |
| 22:38 | <Dashiva> | Poor first-line QA, nobody thinks about their feelings |
| 22:39 | zcorpan | does some of that |
| 22:39 | <Philip`> | I hope it doesn't matter that most of my "Describe in 3 steps or more how to reproduce this bug" only have two steps |
| 22:39 | <zcorpan> | no |
| 22:39 | <Dashiva> | That's pure filler |
| 22:40 | <zcorpan> | 1. Look at TC |
| 22:40 | <zcorpan> | 2. See the bug!!11 |
| 22:40 | <zcorpan> | :) |
| 22:40 | <Dashiva> | I have at least one bug with description "Words fail me, see testcase" |
| 22:41 | <Philip`> | Bah, I only got consecutive bug numbers once :-( |
| 22:41 | <zcorpan> | Philip`: hmm? |
| 22:42 | <Philip`> | I got 282997 and 282998, but the next was 283001 and none of the subsequent ones were consecutive numbers |
| 22:48 | <zcorpan> | aha |
| 22:49 | <zcorpan> | you're not fast enough ;) |
| 23:00 | Philip` | needs to get back to implementing canvas shadows in Mozilla, so he can be more justified in complaining that Opera doesn't support them when every other browser does |
| 23:02 | <zcorpan> | :) |
| 23:18 | <zcorpan> | smart colspan seems to work pretty well |
| 23:18 | <zcorpan> | jgraham: you think it would be possible to implement the same algorithm for both directions? |
| 23:20 | <zcorpan> | i think we had another idea also... that headers in <thead> imply scope="col" so that they apply to the entire table regardless of other cells |
| 23:20 | <zcorpan> | because you can have scrollable tbody or repeat the thead on each page when printed, it only makes sense that they should apply to the whole table |
| 23:35 | <webben> | IIRC you can have more than one thead per table. |
| 23:36 | <webben> | zcorpan: e.g. http://www.w3.org/TR/html401/struct/tables.html#h-11.2.3: "When present, each THEAD, TFOOT, and TBODY contains a row group." |
| 23:37 | <webben> | hmm, actually maybe that's not what that means |
| 23:37 | <webben> | no looking at the DTD it can't mean that, nm :) |
| 23:42 | <zcorpan> | webben: only one thead is allowed :) |
| 23:42 | <webben> | yep |
| 23:43 | <webben> | zcorpan: how would one undo scope-="col" in that situation? |
| 23:43 | <webben> | zcorpan: for example if you had a column th of Country ... but then grouped countries in continental rowgroups |
| 23:44 | <webben> | Country wouldn't be the header for "Europe" |
| 23:45 | <webben> | but you'd still want the header at the top of each page |
| 23:50 | <zcorpan> | webben: i don't quite follow |
| 23:50 | <zcorpan> | why would you place headers in <thead> if you don't want them to apply to the whole table? |
| 23:52 | <webben> | zcorpan: Because they apply to the whole table does not mean they apply to every cell. |
| 23:53 | <zcorpan> | webben: could you create such a sample table, please? |
