| 01:18 | <Hixie> | who was i discussing the websocket protocol with recently? |
| 01:18 | <Hixie> | something about how servers should handle errors |
| 01:18 | <Hixie> | was that jwalden? |
| 01:33 | <franksalim> | Hixie, you were recently discussing the use of CONNECT with jwalden |
| 01:34 | <Hixie> | ok |
| 01:34 | <Hixie> | thanks |
| 01:34 | <Hixie> | i'll have to catch him when he gets back |
| 01:54 | <Hixie> | i wish i understood the IETF better |
| 01:54 | <Hixie> | e.g. why does http://tools.ietf.org/html/draft-hixie-thewebsocketprotocol have versions 12, 13, and 15 but not 14? |
| 01:58 | <franksalim> | I don't know, and now I cannot easily see the diff for the new version |
| 01:59 | <Hixie> | the diff is at http://html5.org/tools/web-apps-tracker |
| 01:59 | <Hixie> | http://html5.org/tools/web-apps-tracker?from=3191&to=3192 specifically is today's update |
| 02:01 | <franksalim> | thanks. it is unfortunate that the missing versions broke the diff links on the ietf site, though |
| 04:52 | <jwalden> | Hixie: it'd be nice if you made it easier to download acid3 for local testing and modifications somehow, maybe a zip of the directory and the Apache config files inside it (I'm guessing at your current setup) |
| 04:53 | <Hixie> | i can provide a tarball if you want |
| 04:53 | <Hixie> | btw i made the websocket protocol spec clearer about server-side error handling rules |
| 04:53 | <jwalden> | that'd have been useful a few hours ago, but I hacked one together |
| 04:53 | <jwalden> | spat of wgets and the occasional -S to see what I was missing |
| 04:53 | jwalden | looks at ws |
| 04:55 | <Hixie> | well let me know if you ever want one, i do make the on demand |
| 04:55 | <Hixie> | i don't really want to encourage people to download it though |
| 04:58 | <jwalden> | why not? |
| 04:58 | <jwalden> | fears that people are likely to screw up the configuration process? |
| 04:59 | <Hixie> | more or less |
| 04:59 | <jwalden> | hm |
| 04:59 | <Hixie> | i don't want to have to field support questions from non-browser developers |
| 04:59 | <Hixie> | non-browser-developers i should say |
| 04:59 | <jwalden> | could put a URL in the acid3 source, might be sufficiently obscure for people not to use it unless they're interested |
| 04:59 | <Hixie> | *shrug* browser people can just ask me :-P |
| 05:06 | <ezyang> | Do we have updated tests for th latest dash-dash-bang-angle comment terminator? |
| 06:52 | <hsivonen> | Hixie: you asked if I was here |
| 06:52 | <hsivonen> | I just read the log from yesterday |
| 06:52 | <Hixie> | i wanted to ask you your input on the comment stuff, iirc |
| 06:52 | <hsivonen> | I think what Philip` said http://krijnhoetmer.nl/irc-logs/whatwg/20090604#l-1058 is the best way to discover if @summary is useful |
| 06:53 | <hsivonen> | Hixie: context? |
| 06:54 | <hsivonen> | Hixie: I want to avoid reparsing if we can avoid it. I don't have proper data, so tentatively I trust zcorpan's judgment on --!> |
| 06:54 | <Hixie> | http://html5.org/tools/web-apps-tracker?from=3190&to=3191 |
| 06:56 | <hsivonen> | Hixie: seems reasonable |
| 06:56 | <Hixie> | k |
| 06:56 | <hsivonen> | I'm already over the 8000 byte limit anyway :-( |
| 06:58 | <hsivonen> | (aside on the 8000-byte limit: not unsurprisingly, on Stack Overflow, I got more people doubting the premise on my question than contributing towards answering the actual question) |
| 06:59 | <Hixie> | heh |
| 06:59 | <hsivonen> | here http://stackoverflow.com/questions/944492/is-there-a-java-bytecode-optimizer-that-removes-useless-gotos |
| 07:03 | <Hixie> | hsivonen: have you considered just writing one manually? it seems you know what optimisation you want. |
| 07:04 | <Hixie> | (possibly by extending an existing optimiser) |
| 07:14 | <hsivonen> | Hixie: I've even considered writing a custom compiler for the method, but I'd prefer not to spend the time |
| 07:14 | <Hixie> | heh |
| 07:14 | <Hixie> | you've already written a Java to C++ convertor |
| 07:14 | <Hixie> | how much hard could a Java to bytecode convertor be :-) |
| 07:15 | <hsivonen> | I've also considered a per hack for splitting the doctype states out just ahead of compilation without spoiling the mail source |
| 07:15 | <hsivonen> | s/pel/perl/ |
| 07:39 | <zcorpan> | "I would like to kill Hixie" - http://hg.gsnedders.com/anolis/raw-file/tip/example.src.html |
| 07:41 | <pesla> | Morning |
| 07:42 | <hsivonen> | zcorpan: fails at nice and happy :-( |
| 07:55 | <jgraham> | Hixie: It you have something like <!-- abc -- > efg the efg will appear outside the comment in gecko |
| 07:55 | <jgraham> | afaict |
| 07:56 | <Hixie> | yes, because of comment reparsing |
| 07:56 | <Hixie> | try it with <!-- abc -- > efg --> |
| 07:57 | <jgraham> | Hixie: I see efg --> outside the comment |
| 07:57 | <jgraham> | (WebKit is different) |
| 08:00 | <Hixie> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/133 |
| 08:00 | <Hixie> | what do you see? |
| 08:00 | <Hixie> | i'm using Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.2a1pre) Gecko/20090604 Minefield/3.6a1pre |
| 08:07 | <zcorpan> | Hixie: why is "anything else" a parse error in the bang state? |
| 08:09 | <zcorpan> | Hixie: <!-- --!--> has one parse error but <!-- --!x--> has two |
| 08:09 | Hixie | denies everything while checking in a fix |
| 08:10 | <jgraham> | Hixie: What I see depends on standards mode v quirks mode |
| 08:10 | <jgraham> | In standards mode dash-dash-space terminates a comment |
| 08:11 | <Hixie> | i've only been checking quirks mode |
| 08:12 | <zcorpan> | in gecko |
| 08:12 | <Hixie> | cos that's what the e-mail i responded to earlier today said to check :-) |
| 08:13 | <zcorpan> | from the pages that were analyzed, it was reasonably clear that pages would render better if -- > closed comments |
| 08:14 | <zcorpan> | certainly there was no page in the set that would render worse |
| 08:15 | <jgraham> | Hixie: It was opera quirks mode that does reparsing. I didn't realise that gecko behaviour depended on mode |
| 08:15 | <Hixie> | what exactly is the proposal? Same as for bang, but if you hit a space, stay in the state? How do you know how many spaces to emit when you hit a bogus character afterwards? |
| 08:15 | <Hixie> | seems like this would involve tracking more state than we'd like |
| 08:16 | <jgraham> | Hixie: Yes and I doubt it matters |
| 08:16 | <jgraham> | Oh maybe it does matter |
| 08:17 | <Hixie> | collapsing all the spaces everytime you hit two --s in a row in a comment would be mighty weird |
| 08:17 | <Hixie> | especially if one of the spaces is a newline |
| 08:17 | <Hixie> | and the comment is something like: |
| 08:17 | <Hixie> | <!------------- |
| 08:17 | <Hixie> | hello |
| 08:17 | <Hixie> | -----------> |
| 08:18 | <zcorpan> | i guess you could just append the two dashes to the comments data upon seeing whitespace so that <!-- -- > would have the data " -- " |
| 08:21 | <zcorpan> | i think it doesn't matter so much what the comment's data is, it's more important to get right where it ends |
| 08:37 | <hsivonen> | Hixie: was it deliberate that the CDATA escape flag doesn't have a bang state? |
| 08:37 | <Hixie> | yes |
| 08:38 | <hsivonen> | maybe I should defer implementation for a few hours until you guys converge on what the spec should be |
| 08:40 | <zcorpan> | Hixie: due to reparsing, http://software.hixie.ch/utilities/js/live-dom-viewer/?%3C!--%20--%20%3EHello shows "Hello" in both webkit and gecko |
| 08:40 | <zcorpan> | Hixie: note the comment's data |
| 08:41 | <Hixie> | yeah, i was convinced by your comment above |
| 08:42 | <Hixie> | sorry, didn't respond, too busy actually implementing your suggestion :-) |
| 08:42 | <zcorpan> | no worries :) |
| 08:44 | <Hixie> | done |
| 08:47 | <zcorpan> | Hixie: whitespace in the space state should just say "Stay in the space state" |
| 08:48 | <zcorpan> | er it should probably append the character too |
| 08:49 | <Hixie> | oops |
| 08:49 | <Hixie> | fied |
| 08:49 | <Hixie> | fixed |
| 08:51 | <Hixie> | http://www.w3.org/mid/49F9F176.8020108⊙me makes me sad |
| 08:52 | <zcorpan> | Hixie: thanks for fixing comment parsing |
| 08:52 | <Hixie> | np |
| 08:53 | <zcorpan> | hsivonen: i think the spec is good now |
| 08:53 | <hsivonen> | zcorpan: ok. thanks |
| 08:54 | <hsivonen> | does Bing have a #1 rank for my doctype article in general or do they just move .fi results higher when the user is in .fi? http://www.bing.com/search?q=doctype |
| 08:54 | <hsivonen> | anyway, it seems that Bing beats Google here ;-) |
| 08:55 | <Hixie> | http://www.w3.org/QA/2002/04/valid-dtd-list.html is the top hit for me |
| 08:55 | <annevk42> | in France you're not on the page at all |
| 08:55 | <Hixie> | yeah here neither |
| 08:55 | <hsivonen> | :-( |
| 08:55 | <annevk42> | first page is http://pompage.net/pompe/doctype/ |
| 08:55 | <zcorpan> | hsivonen: http://labb.seoserver.se/att-anvanda-ratt-doctype/ is #1 for me. you're #2 |
| 08:56 | <hsivonen> | thanks. It seems that Bing is very heavily locale-reshuffled |
| 08:56 | <annevk42> | wow, not on the first three pages |
| 08:56 | <virtuelv> | hsivonen: "heavily" is an understatement |
| 08:57 | <virtuelv> | for some searches I've tried, there is next to nothing in common between the NO and EN_US result pages |
| 08:57 | <hsivonen> | It seems to me that Google's idea of ranking English results high and offering machine translation is a much better solution that parochialism |
| 08:58 | <zcorpan> | it's very annoying to read machine translated text, though |
| 08:59 | <hsivonen> | one day I was searching for something in Finnish, Google offered to translate my search terms into English, found a better result at the BBC and offered to translate it back but also let me read the original |
| 09:00 | <hsivonen> | zcorpan: well, yeah, teaching kids to read English at school works better |
| 09:01 | <Hixie> | hsivonen: nice |
| 09:01 | <Hixie> | hsivonen: i've heard of that feature but never heard of anyone who actually had it trigger for them |
| 09:02 | <Hixie> | hsivonen: glad you had a good experience :-) |
| 09:03 | <zcorpan> | does google detect the language from the search query? |
| 09:04 | <zcorpan> | or does it assume finnish for google.fi? |
| 09:07 | <hsivonen> | I can't remember how I managed to trigger the feature, so I can't test |
| 09:08 | <zcorpan> | does any browser other than opera implement media=projection? |
| 09:09 | <annevk42> | no |
| 09:09 | <annevk42> | there's an extension for Firefox I think |
| 09:10 | <annevk42> | oh, hsivonen's namespace patch got superreview; I wonder how that works out when deployed |
| 09:11 | zcorpan | wonders which patch |
| 09:25 | virtuelv | thinks it's a shame no other vendor has picked up on media=projection yet |
| 09:26 | <annevk42> | it's not really an important feature |
| 09:26 | <annevk42> | though it's fun to have |
| 09:26 | <hsivonen> | annevk42: I'm very happy it got sr. however, commit access even to the trunk is restricted to Firefox 3.5 blockers, so I can't just go ahead and check in |
| 09:27 | <annevk42> | that sucks :/ |
| 09:27 | <virtuelv> | annevk42: indeed |
| 09:27 | <virtuelv> | I very much prefer using Operashow over other technologies for presentations |
| 09:27 | <virtuelv> | (except for the fact that I did roll my own cross-browser workalike) |
| 09:29 | <zcorpan> | operashow is nice |
| 09:30 | <zcorpan> | just needs a proper editor to create them |
| 09:30 | <hsivonen> | paged screen media would make sense for hand-held refreshable paper device |
| 09:31 | <hsivonen> | I remember reading about refreshable black&white paper-like foldable display 10 years ago |
| 09:31 | <hsivonen> | I wonder what happened. Patents? actual technical flaws |
| 09:32 | <hsivonen> | it had tiny spheres embedded to it that were half white and half black and could be flipped electrically |
| 09:34 | <Philip`> | Sounds similar to http://www.eink.com/technology/howitworks.html |
| 09:34 | <Philip`> | (which is used for Kindle-like devices) |
| 09:35 | hsivonen | wants kindle-like device with a real browser engine and Web access |
| 09:35 | <jgraham> | Does the kindle make books continuous or paged? |
| 09:38 | Philip` | assumes that foldability doesn't have the right combination of usefulness and cheapness and robustness to be marketable yet, but maybe they'll solve some of those problems in the next ten years |
| 09:42 | <Hixie> | ok well i see why the ID number went up non-consecutively |
| 09:42 | <hsivonen> | content/html/document/src/nsHTMLContentSink.cpp |
| 09:42 | <hsivonen> | oops. sorry |
| 10:05 | <MikeSmith> | what would be a good description of the "challenge" attribute on the <keygen> element? |
| 10:06 | <Hixie> | i thought "challenge" was pretty descriptive of everything to do with <keygen>, myself |
| 10:06 | <Hixie> | as i understand it it's just a string that gets put in the key |
| 10:07 | <Hixie> | (to prevent replay attacks maybe?) |
| 10:08 | <MikeSmith> | ok, I guess I'll just go with "a challenge string that is submitted along with the public key" |
| 10:08 | hsivonen | is surprised to learn about the Shift_JIS to Win31J mapping |
| 10:09 | <Hixie> | i didn't verify it |
| 10:09 | <Hixie> | so if it's wrong let me know |
| 10:09 | <Hixie> | i'll be very happy to remove it |
| 10:09 | <hsivonen> | I guess it's logical that the stardard names are bunk and compat requires MS extensions |
| 10:09 | <hsivonen> | Hixie: I'm not suggesting it's wrong. just that I'm suprised. |
| 10:10 | <Hixie> | i wouldn't be surprised if it was wrong, anne pointed out the guy said conflicting things about shift-jis in the e-mail |
| 10:12 | <zcorpan> | what was conflicting? wasn't it suggested that the mappings should be shift-jis -> something and shift_jis -> something else |
| 10:15 | <Hixie> | right, but that seems unlikely in the extreme |
| 10:15 | <Hixie> | well |
| 10:15 | <Hixie> | unlikely may be the wrong word |
| 10:15 | <Hixie> | :-) |
| 10:18 | <Hixie> | ok i really should sleep |
| 10:19 | <annevk42> | zcorpan, the conflict is that Unicode matching rules assume Shift-JIS and Shift_JIS are identical |
| 10:21 | <annevk42> | always when encoding comes up as subject I'd really like to solve it, but then I realize I don't really have time :/ |
| 10:21 | <annevk42> | it also seems a lot of tedious work to figure out what maps to what |
| 10:30 | <Mrmil> | Hixie: what time is it? :) |
| 10:30 | <zcorpan> | Mrmil: bed time |
| 10:31 | <Mrmil> | zcorpan: O'Reilly |
| 10:34 | <zcorpan> | yay, "we now have one major vendor supporting [HTML5] (Google)" - http://opengardensblog.futuretext.com/archives/2009/06/why_do_competit.html |
| 10:34 | <zcorpan> | now we just wait for the other vendors to support HTML5, too |
| 10:35 | <hsivonen> | I want a greasemonkey script that redirects faqs.org URLs to tools.ietf.org URLs |
| 10:36 | <zcorpan> | oh it says a few paragraphs later that there are other vendors supporting it, too |
| 10:37 | <hsivonen> | I need to write down that the IETF rejects kings, presidents and voting |
| 10:38 | <zcorpan> | i wonder if the non-Internet world will move to the IETF model |
| 10:40 | <hsivonen> | Hixie: how would UA-driven video full screen work with site-branded JS-based controls? |
| 10:40 | <zcorpan> | hsivonen: you use native controls in full screen |
| 10:41 | hsivonen | notes that YouTube shows branded controls in Flash full screen |
| 10:41 | <annevk42> | does Flash even have native controls? |
| 10:41 | <annevk42> | admittedly for advertising it might be bad |
| 10:42 | <zcorpan> | JS controls for full screen would have to use F11 and media=projection |
| 10:42 | <hsivonen> | I'm all for native controls if the brand manager of YouTube can accept it |
| 10:44 | <hsivonen> | media=projection full screen would defeat the point of moving to fully-accelerated hardware video on mobile without a CSS compositor |
| 10:47 | <annevk42> | hsivonen, they sort of do on the iPhone |
| 10:47 | <hsivonen> | ok. I haven't tried <video> on the iPhone |
| 10:48 | <hsivonen> | (I want a phone like that but without the carrier deal and the restrictions on app installation) |
| 10:58 | <zcorpan> | http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2009-June/020177.html will be valuable when people complain that html5 chages the semantics of certain elements |
| 10:59 | <zcorpan> | people seem to forget or ignore or not be aware of the fact that the semantics of elements have changed between html2 and html4 |
| 11:00 | <zcorpan> | maybe it would be worth to document how each element is defined in the history of html |
| 11:01 | <Philip`> | If OS X is in a state where it refuses to open any applications, and refuses to shut down because it thinks some application I tried to open is still open and won't shut down, what am I meant to do? :-( |
| 11:03 | <jgraham> | Philip`: Hold down the power button? |
| 11:04 | <Philip`> | Hmm, "force quit" lets me kill the offending applications |
| 11:04 | <Philip`> | but now I've quit everything and there's no menu bar at the top of the screen any more, so I can't select "shutdown" from the menu |
| 11:05 | <Philip`> | Power button isn't a very elegant solution, but seems to work |
| 11:05 | hsivonen | wishes OS X had a working panic button for "kill all processes, flush filesystems caches, reboot" |
| 11:05 | <hsivonen> | AFAICT, power button fails the flush filesystem caches point |
| 11:07 | <zcorpan> | i guess the power button should be that panic button |
| 11:11 | <Philip`> | zcorpan: If the power button was that button, what would you do when the filesystem cache flusher had hung? |
| 11:13 | <Lachy> | hsivonen, re "I want a phone like that but without the carrier deal and the restrictions on app installation" -- a jailbroken iPhone bought without a plan in Australia by paying a small unlocking fee would meet your needs |
| 11:13 | <Lachy> | (unlike many other countries, Australia doesn't seem to permit exclusivity like in the uS) |
| 11:13 | <jgraham> | An Android G! dev phone would meet those needs |
| 11:13 | <jgraham> | G1 |
| 11:14 | <jgraham> | (without requiring jailbreaking, I think) |
| 11:14 | <hsivonen> | Lachy: I don't want to pay for a jailbroken phone, since my money would be a vote for the DRM-to-install-code model |
| 11:15 | <zcorpan> | Philip`: i guess if you hold the button for long enough it will force turn off |
| 11:15 | <Lachy> | well, you'd have to jailbreak it yourself. But my point was that you can buy an iphone that isn't locked to a specific carrier |
| 11:18 | zcorpan | has just discovered a whole new way to fold tortillas |
| 11:19 | <zcorpan> | wonder why i haven't thought of it before |
| 11:20 | <zcorpan> | cut it in half, roll it along the straight line, fold the bottom, then fill it |
| 11:21 | <Lachy> | zcorpan, how does cutting a tortilla in half help with folding it? |
| 11:22 | <Lachy> | I don't understand how you can roll it and fold the bottom before filling it either |
| 11:22 | <Lachy> | since you need to fill it when it's flat, then roll it |
| 11:22 | <zcorpan> | Lachy: it doesn't directly but it makes it easier to fill |
| 11:22 | <zcorpan> | why do you need to fill it first and not last? |
| 11:22 | <Lachy> | I really don't understand how |
| 11:23 | <Lachy> | because if you roll it first, then you need to try and fill it through a small hole in one end |
| 11:24 | <Lachy> | that's bound to be a messy experience |
| 11:24 | <zcorpan> | the whole doesn't need to be so small |
| 11:24 | <zcorpan> | and since i cut it in half it was actually not messy at all |
| 11:25 | <zcorpan> | s/w// |
| 11:25 | <Lachy> | maybe I need to see it to understand it, cause I'm trying to imagine a non-messy way to fill it |
| 11:26 | <Lachy> | and I can't |
| 11:26 | <zcorpan> | i just use a fork |
| 11:26 | <zcorpan> | it might help in understanding that i fill it with fajitas |
| 11:26 | <Lachy> | so you take half a tortialla, make it a cone shape and fill it? |
| 11:27 | <zcorpan> | yup |
| 11:27 | <Lachy> | but then you only get a half-size meal |
| 11:27 | <zcorpan> | but you can eat twice the amount :) |
| 11:28 | <Lachy> | it's so much easy to place the tortialla in your hand, fill it in the middle and then wrap the whole tortilla around the filling |
| 11:29 | <zcorpan> | maybe i suck at doing it (even though people around me usually have more problem with it than me) but i found this to be less messy |
| 11:30 | <Lachy> | sure, I've found many other people are clueless when it comes to filling them too |
| 11:32 | <Lachy> | but I wonder, do the Swedish like to put real hot sauce on their mexican food? The Norwegian variety of hot sauce is, unfortunately, very mild |
| 11:32 | <zcorpan> | in sweden there's usually three grades - mind, medium and hot |
| 11:32 | <Lachy> | and even in many restaurants here, if you ask for hot, it comes back medium at best |
| 11:32 | <zcorpan> | i usually use the "hot" and it's not very hot |
| 11:33 | <zcorpan> | so i guess the answer to the question is "no" |
| 11:33 | <Lachy> | sure, they sell 3 grades of hot sauce here in norway too. But even the jar labelled Hot, is mild by comparison with what I'm used to |
| 11:34 | <Lachy> | I guess they just sell the same stuff in both countries. IIRC, there's both swedish and norwegian instructions on the various mexican kits |
| 11:36 | <jgraham> | Tune in next week for another episode of "Lachy does Macho"! |
| 11:36 | <jgraham> | (I joke, I joke) |
| 12:38 | <Lachy> | oops, I sent that webstorage mail to public-html instead of public-webapps. |
| 12:38 | <Lachy> | oh well, hopefully that won't matter |
| 12:57 | <hsivonen> | my sarcasm detector doesn't work on this comment: http://www.molly.com/2009/06/02/the-real-why-xhtml-discussion/comment-page-1/#comment-2446810 |
| 12:58 | <hsivonen> | "I am just glad that Steve says that XHTML as text/html is fine. Admittedly I hadn’t lost much sleep over this before but now I can officially ignore some of the more dogmatic viewpoints on things like that." |
| 12:58 | <annevk42> | no sarcasm there I think |
| 12:59 | <hsivonen> | ok |
| 12:59 | <annevk42> | his site uses XHTML and his name doesn't ring any bells with respect to this "debate" |
| 13:05 | <Lachy> | this mail refers to some html4all discussions http://lists.w3.org/Archives/Public/public-html/2009Jun/0201.html - but the html4all mailing list archives show no activity since February. http://wilbur.bytowninternet.com/mailman/private/list_html4all.org/ Do they have some other secret list now? |
| 13:06 | <Lachy> | unless Leif just meant previous discussions from several months ago |
| 13:11 | <Lachy> | anyway, using <figure><summary>...</summary> <table>...</table></figure> would be an interesting solution if there's a legitimate problem for it to solve (I'm staying neutral for now on the debate about whether or not there is a problem though) |
| 13:34 | <hsivonen> | Does Chrome on Mac use Quartz instead of Skia? |
| 13:35 | <hsivonen> | no <video> in Mac Chrome yet |
| 13:36 | zcorpan | fails to see how <figure><summary>...</summary> <table>...</table></figure> is better than <table><caption>...</caption>...</table> |
| 13:37 | <zcorpan> | certainly the latter works better in existing AT |
| 13:39 | <Lachy> | zcorpan, the latter would be better if, as I said, there were a legitimate problem being solved with the current approach of using <caption>. |
| 13:39 | <Lachy> | s/the latter/the fomer/ |
| 13:41 | <jgraham> | I fail to see how either is better than <table><caption><details>...<legend>...</legend></caption> |
| 13:42 | <Lachy> | jgraham, using details like that seems unintuitive |
| 13:42 | <jgraham> | Lachy: Almost no one is going to do it anyway |
| 13:42 | <jgraham> | and for those people that do it doesn't seem that hard |
| 13:43 | <jgraham> | (since they will presumably have been trained/advocated into putting more information for AT users into their table) |
| 13:43 | <Lachy> | personally, I prefer just writing a summary in the prose before the table using an ordinary <p>. I fail to see why it needs to be directly associated with the table by markup |
| 13:44 | <Lachy> | I don't particularly like the idea of writing the summary in the caption, since captions are supposed to be short |
| 13:47 | <Philip`> | That's easy to solve - just stop supposing that captions should be short |
| 13:48 | <jgraham> | Lachy: Captions, in general are not short. That is a weird HTMLism |
| 13:48 | <jgraham> | or rather HTML4ism |
| 13:49 | <zcorpan> | i put the summary in <p> in http://simon.html5.org/articles/mobile-results |
| 13:49 | <zcorpan> | maybe i should update it to put it in the caption |
| 13:49 | <Lachy> | they should be 1 or 2 sentences, just enough to describe the content of the figure/table |
| 13:56 | <Philip`> | Why should they? |
| 13:56 | <beowulf> | Lachy: you suppose 1 or 2 sentences is enough to describe the content |
| 13:57 | <Philip`> | You could always make it two very long sentences with lots of semicolons |
| 13:57 | <Lachy> | because the content should be sufficiently self describing to not need an excessively long caption. Also, in most books and newspapers where I've seen captions, they're usually quite short |
| 14:07 | <jgraham> | Lachy: Try writing a scientific paper with the major results in a table and fitting the caption into 1/2 sentences. It often isn't possible |
| 14:08 | <jgraham> | (I guess you could use <figure> for that case but it's not really clear why there should be two ways to do it with some arbitary cutoff based on length) |
| 14:09 | <Lachy> | jgraham, can you give an example of such a long caption? |
| 14:11 | <zcorpan> | Philip`: could you grep for pages that have non-whitespace and "-->" on the last non-whitespace line of scripts? |
| 14:12 | <zcorpan> | ie and opera have interesting behavior where the last line is stripped if it ends with --> |
| 14:13 | <zcorpan> | in opera the line doesn't get stripped if it also contains // |
| 14:13 | <Lachy> | which versions of IE have you tested? |
| 14:13 | <zcorpan> | Lachy: ie8 |
| 14:13 | <jgraham> | Lachy: Consider http://arxiv.org/pdf/astro-ph/0603449v2 Table 2 has a rather long caption. Many other tables do not. Insisting that they should be different elements doesn't make much sense |
| 14:13 | <zcorpan> | in ie it doesn't get stripped if the line contains // or <!-- (i think) |
| 14:13 | <Lachy> | ah, then maybe they're complaining about a syntax error if the the --> isn't commented out with // |
| 14:14 | <Philip`> | zcorpan: Do you mean something like "^\S+-->\s*</script>"? |
| 14:15 | <zcorpan> | Philip`: there could be whitespace on the line, just not only whitespace followed --> |
| 14:15 | <zcorpan> | foo = 'bar'; --> |
| 14:15 | <zcorpan> | </script> |
| 14:16 | <Lachy> | jgraham, the captions in that contain a lot of content that I would have thought more appropriate for the main content of the paper where referring to the relevant figures |
| 14:16 | <Philip`> | zcorpan: Oops, yes |
| 14:17 | <jgraham> | Lachy: That is a typical style for those papers. They are rahter optimised for people using figures and tables as references without also having to reread the whole paper |
| 14:17 | <Philip`> | zcorpan: Do you want to ignore lines that use //--> (or similar)? |
| 14:17 | <zcorpan> | Philip`: yes, ignore lines that have // |
| 14:17 | <Lachy> | well, I'd consider that bad form, but ok |
| 14:19 | <Philip`> | zcorpan: What if they start with <!-- on that line? |
| 14:20 | <Philip`> | (Lots of people write <script><!-- comment --></script>) |
| 14:22 | <Lachy> | Philip`, I've seen that done in order to prevent some broken backend systems from mistakenly outputting empty scripts as <script src="..." /> |
| 14:22 | <Lachy> | so they put in a comment to make it appear non-empty |
| 14:22 | <zcorpan> | Philip`: if they are overwhelming then yeah it would be good to remove those that have <!-- on the same line, too |
| 14:22 | <jgraham> | I think it is the cases where there is a --> but no <!-- or // that are the most interesting |
| 14:25 | Philip` | searches for (?i)^\s*(?!//|<!--)\S((?!//|<!--).)*-->\s*</script> |
| 14:25 | <Philip`> | (but only searching individual lines, so it won't find anything if the </script> is on the next line after the content) |
| 14:26 | <Lachy> | if you did the search using just (?i)^\s*(?!//|<!--)\S((?!//|<!--).)*-->\s* would you get too many false positives? |
| 14:27 | <jgraham> | Philip`: If possible it would be nice to get cases where the </script> is on the next line too |
| 14:27 | <jgraham> | Although I guess that would be slower |
| 14:27 | <Philip`> | Lachy: I expect that would find a bazillion HTML comments that have nothing to do with scripts |
| 14:28 | <Lachy> | oh, right, so you need a way to limit the search to within script elements |
| 14:35 | Philip` | tries running it across multiple lines, but it'll take forever (since the single-line one hasn't even finished yet) |
| 14:36 | <Philip`> | I really ought to parallelise my grep :-/ |
| 14:38 | <zcorpan> | Philip`: i'm not in a hurry. :) sorry for using your cpu |
| 14:39 | <jgraham> | "Aplologies for increasing the entropy of the universe" |
| 14:39 | <jgraham> | *Apologies |
| 14:40 | <Philip`> | Ooh, the first one finished |
| 14:40 | <Philip`> | zcorpan: http://philip.html5.org/data/comment-close-at-end-of-script.txt |
| 14:41 | <zcorpan> | Philip`: thanks |
| 14:51 | <zcorpan> | Philip`: the first few pages of those use //--> but your .txt says /--> |
| 14:53 | <zcorpan> | though www.mp3music.co.il/forums/topic/?id=6&guid=MNRQFBZGBU12252008235139 does use a single slash |
| 14:54 | <zcorpan> | so maybe www.oclc.org/services/preservation/ has changed |
| 14:54 | <Philip`> | zcorpan: The pages like http://www.oclc.org/services/preservation/http:/www.oclc.org/programsandresearch/kilgouraward/http:/www.oclc.org/research/http:/www.oclc.org/support/systemalerts/default.asp say "/--></script>" twice when I look at them |
| 14:54 | <Philip`> | s/twice/thrice/ |
| 14:54 | <Philip`> | at the bottom |
| 14:55 | <zcorpan> | ah |
| 14:55 | <zcorpan> | i thought your script emitted several urls |
| 14:55 | <Philip`> | Only one per line |
| 14:56 | <Philip`> | That is a pretty weird URL, though |
| 14:56 | <zcorpan> | i guess it's not too bad if a 404 page doesn't work correctly :) |
| 14:58 | <Philip`> | Hmm, my other grep has reached chunk 6 out of f |
| 14:58 | <Philip`> | so it might take a long time |
| 14:58 | <zcorpan> | http://www.onpointevents.com/PPF/Page/1/resources.asp has a script that sets the today variable but it's not used anywhere, so it wouldn't break |
| 14:59 | <Philip`> | zcorpan: http://philip.html5.org/data/comment-close-at-end-of-script-across-multiple-lines.txt is some of the multi-line grep's output |
| 15:00 | <Philip`> | (Apologies for the barely-readable output format) |
| 15:00 | <zcorpan> | Philip`: thanks. it's readable enough :) |
| 15:01 | <zcorpan> | the script at mp3music.co.il/lyrics2/text.aspx?id=26601 would stop working |
| 15:02 | <zcorpan> | but it doesn't seem like a vital script |
| 15:10 | <zcorpan> | it breaks a script at http://www.marika.com/productinfo_item_43-dc_614-md_614-pd_564-cd_on-sn_5.htm but i don't know what the script does - the page seems to work in firefox |
| 15:12 | <zcorpan> | if(lt){d.write('<'+'!-- ');}//--></script><noscript> |
| 15:12 | <zcorpan> | <img src="http://adlik.akavita.com/bin/lik?id=28254"; |
| 15:12 | <zcorpan> | border=0 height=31 width=88 alt="Akavita: каталог, рейтинг, счетчик для сайтов Беларуси"> |
| 15:12 | <zcorpan> | </noscript><script language="JavaScript"><!-- |
| 15:12 | <zcorpan> | if(lt){d.write('--'+'>');} --></script> |
| 15:12 | <zcorpan> | http://mogbiz.by/modules.php?name=pages&pa=list_pag&cid=32 |
| 15:13 | <zcorpan> | won't break since the second <script> tag will be part of an html comment |
| 15:13 | <Philip`> | Ouch |
| 15:17 | <zcorpan> | yay i've gone through http://philip.html5.org/data/comment-close-at-end-of-script.txt - none of them actually breaks in functionality afaict, a few of them would cause a script error |
| 15:17 | <zcorpan> | on to multiline |
| 15:18 | <zcorpan> | <script> |
| 15:18 | <zcorpan> | <!--- |
| 15:18 | <zcorpan> | var g_date="2005.01.12 00:00:00" |
| 15:18 | <zcorpan> | document.write(g_date.substring(0,10)); |
| 15:18 | <zcorpan> | ---> |
| 15:18 | <zcorpan> | </script> |
| 15:18 | <zcorpan> | sigh |
| 15:19 | <zcorpan> | http://www.ctax.org.cn/was40/search?channelid=39843 does break |
| 15:22 | <Philip`> | Oops, X crashed, which killed my SSH session, which killed my grep |
| 15:22 | <Philip`> | I hope that partial list is adequate for now |
| 15:22 | <zcorpan> | yeah it's good enough |
| 15:22 | <zcorpan> | thanks again |
| 15:23 | <zcorpan> | http://www.modelenium.co.uk/models/TACA-A319--GJTAI149.aspx just tries to disable right click so not too serious if the script doesn't run :) |
| 15:24 | <zcorpan> | http://www.stadtguide24.de/index.php?myGeoId=14556&site=edit&lang=de has an ie-specific script with document.all check so doesn't matter |
| 15:27 | <gsnedders> | http://s.ytimg.com/yt/js/base_all_with_bidi-vfl101077.js is being really slow taking YouTube down here :\ |
| 15:27 | <gsnedders> | s.ytimg.com is slow in general :\ |
| 15:28 | <zcorpan> | http://www.meganime.org/foro/viewtopic.php?p=316494 PreloadImages() will break, but only breaks the *pre*load so not too serious |
| 15:30 | <zcorpan> | http://www.fise.org/associazioni/assioma/index.php?we_objectID=9643&we_objectTID=425 - breaks the image swap in the menu. not too serious, either |
| 15:31 | <zcorpan> | http://www.wholesalemegamall.com/index.php?c=office&n=1084128&i=B000LYAX1G&a=buy&m=All&p=1&x=Panasonic_Dect_60_Series_3_Handset_Cordless_Phone_System_with_Answering_System_KX_TG1033S has ie-specific script |
| 15:33 | <zcorpan> | http://www.diadora.de/planetnewsplus/misc/tell_a_friend/tell.php?id=20070301120209 also breaks preload / image swap (but i don't know which images) |
| 15:35 | <zcorpan> | http://www.foodireland.com/Merchant2/merchant.mvc?Screen=PROD&Product_Code=w580420&Category_Code=caseflour&Product_Count=4 - breaks the script that prints out today's date |
| 15:38 | <zcorpan> | http://www.emumax.com/emumax_tvnews_detail.asp?id=188 - seems to break a tracker script or something |
| 15:42 | <zcorpan> | seems to possibly break the comment form in http://www.langya.cn/travel/lydx/200801/t137487.htm |
| 15:45 | <zcorpan> | the gay.sexe-photos site uses <script language="vbscript"> |
| 15:45 | zcorpan | realises that he just watched porn at work |
| 15:46 | gsnedders | facepalms |
| 15:48 | <zcorpan> | community.bravo.de/boards/viewtopic.php?t=134678: /* ]]> */--> |
| 15:49 | <zcorpan> | that works in spidermonkey but breaks in nitro and v8 |
| 15:52 | <zcorpan> | http://www.rapido.nl/care-plus-click-away-bite-relieve-travel-accessoire-artikelnummer-91654.html probably breaks some progress bar script |
| 15:55 | <zcorpan> | well the rest just seem to be more of the same stuff |
| 15:56 | <virtuelv> | well, Opera summer party coming up in 9 seconds. |
| 15:58 | <zcorpan> | crap, http://www.bbsone.com/html.php?file=bbs/business/gb/messages/908.htm would break a link |
| 16:00 | <zcorpan> | http://www.hjsflower.com/flower/show.asp?id=A3004&bookid=961&anid=3&nid=34 might break the form |
| 16:01 | <zcorpan> | http://www.yeongnam.com/yeongnam/html/edu/university/article.shtml?id=20090112.010330807390001 breaks form validation (not too serious) |
| 16:04 | <zcorpan> | i'm through the list! |
| 16:04 | <zcorpan> | so let's see |
| 16:06 | <Philip`> | Let me know if you want a longer list ;-) |
| 16:07 | <zcorpan> | 2 with missing content, 1 with broken link, 2 something with forms, 2 with a progress bar something, a few with preload/imageswap and the rest didn't really break anything |
| 16:08 | <zcorpan> | so that's about 5 pages breaking in functionality |
| 16:12 | <zcorpan> | (since i didn't see any progress bar and the imageswap isn't really functionality) |
| 16:13 | <zcorpan> | wonder if the other vendors want to implement this |
| 16:19 | zcorpan | updates http://wiki.whatwg.org/wiki/Web_ECMAScript#HTML_comments |
| 19:09 | gsnedders | grumbles at the W3C's own inability to follow its own schemas |
| 19:11 | <jgraham> | gsnedders: are you surprised? |
| 19:24 | <gsnedders> | jgraham: Knowing how many bugs I've filed on this already, no :P |
| 19:24 | <gsnedders> | Actually, in some ways, I'm surprised I haven't found all the issues yet. |
| 19:24 | gsnedders | wonders if he's doing a Philip` and reporting issues on his own suggestions |
| 20:59 | <hober> | http://radar.oreilly.com/2009/06/fbml-yml-osml-oh-my.html |
| 22:09 | <gsnedders> | ezyang: We need to deal with stuff at a Unicode level in the TreeBuilder so we can coerce what we have into an XML infoset, which means we will either need to decode the data _again_, or we just use an array of codepoints, which eats memory… |
| 22:09 | gsnedders | sighs |
| 22:11 | <ezyang> | Oh, we want XML infoset support? |
| 22:11 | <gsnedders> | ezyang: We need to, if we are to use DOM… |
| 22:11 | <gsnedders> | ezyang: It's the cause of all 10 current exceptions thrown |
| 22:11 | <ezyang> | Well, I have no intentions of staying on DOM |
| 22:12 | <gsnedders> | ezyang: How are you going to write anything in userland with anywhere near decent perf? :) |
| 22:12 | <ezyang> | Write it in C and PHP |
| 22:12 | <gsnedders> | ezyang: Not re-distributable. |
| 22:12 | <gsnedders> | (Well, sure, it is, but if you want to ship it in normal software…) |
| 22:12 | <ezyang> | Performance is "not re-distributable" |
| 22:12 | <ezyang> | We'll obviously have a PHP drop-in replacement |
| 22:13 | <ezyang> | A userland tree implementation can have decent performance, given some concessions on the tree-builder and API side |
| 22:13 | <ezyang> | Decent enough, at least, for anyone who wants to casually use the library |
| 22:13 | <gsnedders> | What would be nice would be to get DOM to do no checks when strictErrorHandling = false |
| 22:14 | <ezyang> | What's more likely to happen is that userspace SimpleTree will blow up memory usage |
| 22:14 | <ezyang> | (not that DOM doesn't do that already, but you can't see that if you profile) |
| 22:14 | <ezyang> | gsnedders: We filed a bug on that a long time. Didn't work. |
| 22:14 | <ezyang> | *a long time ago |
| 22:15 | <gsnedders> | Didn't work in what way? |
| 22:15 | <bdodson> | Hi, can anyone tell me how websockets will handle/support binary data? Will you be able to download the bytes of a jpeg through it and load it in the DOM for example? |
| 22:15 | <Hixie> | eventually, yes |
| 22:15 | <Hixie> | not in the first version though |
| 22:16 | <ezyang> | As in, they were like, WONTFIX, and proceeded to add the check for the place they missed |
| 22:16 | <ezyang> | Let me dig up the bug |
| 22:16 | <ezyang> | (we had this conversation before) |
| 22:16 | <gsnedders> | http://bugs.php.net/bug.php?id=44648&edit=1? |
| 22:17 | <gsnedders> | That's the only thing I find looking for http://bugs.php.net/bug.php?id=44648&edit=1 |
| 22:17 | <ezyang> | Ya, that's it |
| 22:17 | <gsnedders> | *strictErrorHandling |
| 22:17 | <gsnedders> | Clipboard fail. |
| 22:17 | <gsnedders> | I got told that it could be done… if I wrote the patch. |
| 22:17 | <gsnedders> | Which is all fine and well… if it isn't for the fact I don't know C. :P |
| 22:17 | <ezyang> | It's not that difficult a patch |
| 22:17 | <ezyang> | PHP insulates you sufficiently from actually C-ness |
| 22:17 | <ezyang> | The hard problem is figuring out how to compile PHP |
| 22:17 | <gsnedders> | I couldn't work out how to get the strictErrorHandling property |
| 22:18 | <ezyang> | Hm. |
| 22:18 | <gsnedders> | If I could work that out, I could probably write a patch :P |
| 22:18 | <ezyang> | That's slightly non-trivial |
| 22:18 | <ezyang> | Anyway, I still argue, that's not tenable for the sake of portability |
| 22:18 | <gsnedders> | How so? Older shipped versions? |
| 22:19 | <ezyang> | Yep |
| 22:19 | <gsnedders> | Agreed, but in the longer-term it would be nice to be able to use it. |
| 22:19 | <ezyang> | Sure. |
| 22:19 | <ezyang> | I wonder if SimpleXML does checking... |
| 22:19 | <gsnedders> | In places. |
| 22:20 | <bdodson> | does anyone have basic websocket support now? |
| 22:20 | <ezyang> | :-( |
| 22:20 | gsnedders | wishes they would fix XMLWriter so you could actually rely on it outputting XML. |
| 22:20 | <Philip`> | An XML serialiser that outputs XML? I think you're asking for too much |
| 22:21 | <gsnedders> | http://bugs.php.net/bug.php?id=41800 |
| 22:21 | <gsnedders> | See, it's not a bug! |
| 22:21 | <ezyang> | gsnedders: If you want to take a stab at it again (I probably won't have time for it this weekend), the first thing you'll want to figure out is whether or not it's something stored in the PHP layer, or the libxml layer |
| 22:22 | <gsnedders> | ezyang: I don't care enough. I have too many bad experiences to think I have any chance of it ever being committed. |
| 22:23 | <gsnedders> | http://bugs.php.net/bug.php?id=44619 |
| 22:24 | <ezyang> | They're going to fix it The Right Way! No Really! |
| 22:24 | <ezyang> | More seriously, |
| 22:25 | gsnedders | adds XMLWriter grips to his list of problems with PHP |
| 22:25 | <gsnedders> | *gripe |
| 22:25 | <ezyang> | I think the PHP dev community is not too receptive about patches unless you bug them a lot about it |
| 22:25 | <gsnedders> | *gripes |
| 22:25 | <ezyang> | You will almost certainly not get a patch that big into PHP5.2. Maybe PHP5.3 |
| 22:25 | <Philip`> | gsnedders: Are you foolishly expecting the XML serialiser to deal with the issues of encoding data so it can be serialised as XML? Clearly that should be the user's responsibility |
| 22:26 | <gsnedders> | Philip`: I am expecting an XML serializer to always output something that meets the document production in the XML spec. |
| 22:26 | <Philip`> | It's a bug that they've even got any methods other than $xml->write('<test>blabla</test>') |
| 22:26 | <Hixie> | bdodson: webkit and mozilla are working on it, i believe |
| 22:27 | <gsnedders> | ezyang: I bet you 5.3 will be in commit freeze by the time I get around to anything like writing a patch… |
| 22:27 | <Philip`> | (It's up to you to make sure the data is encoded properly) |
| 22:28 | <ezyang> | So, if PHP didn't have binary strings by default, this would make sense |
| 22:28 | <bdodson> | Hixie: great, I guess I'll wait patiently for a bit :) |
| 22:28 | <gsnedders> | ezyang: s/by default/always/ |
| 22:30 | <Hixie> | bdodson: you and me both! |
| 22:30 | <ezyang> | Well, if most of the strings whizzing around are UTF-16, and then I want to put a binary string into the XML writer |
| 22:30 | <ezyang> | I'd say it has two options: refuse to do anything with it, or base64 encode it |
| 22:30 | <bdodson> | Hixie, huh, here's a video claiming minefield has WebSocket support ... |
| 22:30 | <bdodson> | http://www.youtube.com/watch?v=CrV9jJQn_MA |
| 22:31 | <ezyang> | "Fixing it" so that it's well-formed UTF-whatever is totally not the correct behavior. |
| 22:31 | <gsnedders> | ezyang: I'd expect it to die. |
| 22:31 | <ezyang> | And then, you can't even do that the smart way in PHP, because it doesn't have any notion of types |
| 22:31 | <ezyang> | So this won't get fixed until PHP 6, when we get Unicode strings |
| 22:32 | <Hixie> | bdodson: that's a demo of the work they're doing, it's not in yet i don't think |
| 22:32 | <ezyang> | PHP is currently operating an abstraction level too low |
| 22:32 | <ezyang> | And this is a leak. |
| 22:38 | <gsnedders> | Up to 12 items on my list of problems with PHP… |
| 22:39 | <ezyang> | i,i Only twelve? |
| 22:39 | <gsnedders> | So far :) |
| 22:39 | <gsnedders> | These are mainly fairly high-level issues. |
| 23:43 | <Hixie> | wow, shelley just sent what i believe is the most honest answer as to why people want to keep summary="" despite it having demonstrably failed to improve accessibility |
| 23:48 | <annevk2> | 'They've spent, what seems to me, a lot of time promoting this attribute, and accessibility in general. And now, a few people just want to yank it because "it's not being used right".' -> http://en.wikipedia.org/wiki/Sunk_costs |
| 23:52 | <Hixie> | yup |
| 23:53 | <annevk2> | nn |
| 23:53 | <Hixie> | nn |
| 23:54 | <Dashiva> | It could be that some people believe that the bad use just doesn't count. It's just a tiny waste of time each occasion. |