| 01:35 | <paul_irish> | if an image is served with CORS, shouldn't canvas be able to get its imageData? |
| 01:39 | <Philip`> | I think the specs currently say you shouldn't, but it would be a sensible feature to have |
| 01:40 | <paul_irish> | indeed. if i'm greenlighting use of my images crossdomain, then there shouldnt be a problem with people absorbing the imagedata |
| 01:40 | <paul_irish> | currently all browsers still throw on the getImageData, regardless of cors. :( |
| 02:36 | <abarth> | paul_irish: yeah, we want to support that in WebKit |
| 02:37 | <abarth> | paul_irish: it requires some architectural changes to how we load images off the network |
| 02:37 | <abarth> | paul_irish: but we've got some folks working on fixing that |
| 02:37 | <paul_irish> | cool! should i file a ticket to track it? |
| 02:37 | <abarth> | (me included) |
| 02:37 | <abarth> | you can if you like |
| 02:38 | <abarth> | webkit isn't very good at using its bug database to prioritize work |
| 02:38 | <abarth> | because the project is very decentralized |
| 02:38 | <abarth> | and prioritization is somewhat of a centralized concept |
| 02:38 | <paul_irish> | aye. |
| 02:38 | <paul_irish> | while i've got you.. i just came across a unique case where the html5 parser is a bit cranky |
| 02:39 | <paul_irish> | http://rexcasteel.com/ |
| 02:39 | <paul_irish> | it has an </center tag with no closing > |
| 02:39 | <abarth> | hum |
| 02:39 | <abarth> | ah |
| 02:39 | <abarth> | </center<foo> |
| 02:39 | <paul_irish> | which makes the <table> which... |
| 02:39 | <paul_irish> | yes |
| 02:39 | <paul_irish> | the table dies. |
| 02:39 | <abarth> | that's one of the top problems we have with the HTML5 algorithm |
| 02:40 | <abarth> | however, that site is probably equally busted in IE |
| 02:40 | <abarth> | shall we email the author and ask him to fix his page? |
| 02:41 | <paul_irish> | yeah i can do that no problem. |
| 02:41 | <abarth> | thanks |
| 02:41 | <abarth> | <!doctype html public "-//w3c//dtd xhtml 1.0 strict//en" "dtd/xhtml1-strict.dtd"> |
| 02:41 | <abarth> | :) |
| 02:41 | <aho> | strict |
| 02:41 | <abarth> | yeah |
| 02:41 | <aho> | of course :> |
| 02:41 | <paul_irish> | yeah.. commented out prolog+doctype, too |
| 02:42 | <paul_irish> | (yes its busted in ie6, at least.) |
| 02:42 | <abarth> | its a tokenization issue |
| 02:42 | <abarth> | both tokenization choices there break web sites |
| 02:42 | <paul_irish> | aiight.. well no problem.. just makig sure i didnt hit an edge case that wasnt captured before. |
| 02:42 | <paul_irish> | figured it wasnt likely. :) |
| 02:42 | <abarth> | its a matter of picking your poison |
| 02:42 | <abarth> | we've picked the IE poison :) |
| 02:43 | <aho> | you say that like it's a good thing |
| 02:44 | <aho> | ie6 is down to 1.23% in germany, by the way |
| 02:44 | <aho> | woo |
| 02:44 | <aho> | <: |
| 02:44 | <aho> | which isn't even half of what ff 3.0 got |
| 02:45 | <aho> | and we surely dont test for that one :) |
| 02:47 | abarth | <3 germany's browser market |
| 02:49 | <aho> | ie7 will most likely drop below 5% next month |
| 02:49 | <aho> | good times :) |
| 05:35 | <Hixie> | oh lordy, the _summary_ to http://lists.w3.org/Archives/Public/public-html/2010Oct/0268.html is longer than some of my CPs in their entirety |
| 05:35 | <Hixie> | some people really don't know how to be succint |
| 07:35 | <mnot> | hixie: one thing re: http://lists.w3.org/Archives/Public/public-html/2010Oct/0268.html |
| 07:35 | <mnot> | I haven't read the whole thing, but one of the stated motivations is that "It would avoid clashes with the Nottingham draft." |
| 07:36 | <mnot> | I don't see how that is the case, since the draft only sets up a framework for link relations, and re-registers what existed in the IETF registry at the time |
| 07:37 | <mnot> | the relations themselves — and their semantics — are effectively owned by the community; in this case, the HTML community (since HTML registered the ones he's talking about, IIRC) |
| 07:42 | <annevk> | mnot, Hixie didn't write that |
| 07:42 | <mnot> | I know - just making sure that feedback gets in |
| 07:42 | mnot | would never dream that hixie could write that |
| 07:43 | <Hixie> | mnot: the place to put such feedback is the poll, whenever that happens |
| 07:43 | <Hixie> | heh |
| 07:43 | <mnot> | hixie: I'm not a member of the WG |
| 07:43 | Hixie | couldn't really work out what leif was proposing anyway |
| 07:43 | <Hixie> | mnot: then you should join |
| 07:43 | <mnot> | the w3c's rules don't allow me to allow myself to join |
| 07:43 | <mnot> | it's complex |
| 07:44 | <Hixie> | oh, right, i remember now |
| 07:44 | <Hixie> | well, tell sam to put your feedback in on your behalf or something |
| 07:44 | <mnot> | k |
| 07:44 | <Hixie> | i can't believe the w3c is actively turning someone like you away |
| 07:44 | <Hixie> | so ridiculous |
| 07:45 | <Hixie> | especially given how open the html work is |
| 07:45 | <Hixie> | fwiw you are always welcome in the whatwg |
| 07:45 | <mnot> | will keep that in mind; just over-committed already. |
| 07:45 | <Hixie> | yeah, i hear ya |
| 07:45 | <mnot> | (and REALLY glad I cancelled going to Lyon) |
| 07:56 | <nessy> | I think to become part of public-html you just have to fill in some forms and stuff to become an invited expert |
| 07:57 | <MikeSmith> | mnot: why you glad you cancelled going to Lyon? |
| 07:58 | <mnot> | because its f'ing far away, and I've already been travelling a lot this year, with more to come. |
| 07:58 | <annevk> | nessy, not if you work for an organization that can become a Member |
| 07:58 | <othermaciej> | Lyon is quite nice |
| 07:58 | <mnot> | I miss the interaction, of course, but have to prioritise other stuff. |
| 07:59 | mnot | wonders if it's as good as Maastricht — I loved Maastricht. |
| 07:59 | <nessy> | annevk: http://www.w3.org/2004/01/pp-impl/40318/instructions says otherwise |
| 07:59 | <othermaciej> | it is rather far away |
| 07:59 | <annevk> | nessy, yes it is, see "If you do not work for a W3C Member organization, first consider whether your employer can join W3C and get the benefits of Membership." |
| 07:59 | <othermaciej> | from California anyway |
| 07:59 | <nessy> | I joined as invited expert without company affiliation |
| 07:59 | <annevk> | mnot, nope, and not just saying that because I'm Dutch |
| 07:59 | <mnot> | Everywhere is far from Australia. |
| 08:00 | <nessy> | annevk: and then it says "If this is not an option and you think that you have the expertise and availability to participate, you may request to participate as an Invited Expert as follows." |
| 08:00 | <annevk> | nessy, but for the company mnot works for it is an option |
| 08:00 | <nessy> | that sentence is only marketing talk |
| 08:01 | <nessy> | if the company doesn't decide to become a member, then it's not an option for you personally |
| 08:01 | <annevk> | clearly it is not |
| 08:01 | <annevk> | but I am not sure why I am debating this |
| 08:01 | <nessy> | :) |
| 08:15 | <nessy> | say, the MediaStreamTransceiver on https://labs.ericsson.com/developer-community/blog/beyond-html5-conversational-voice-and-video-implemented-webkit-gtk was made up by the Ericsson guys - it's not something that's specced anywhere - or am I missing something |
| 08:16 | <nessy> | we have ConnectionPeer for that, right? http://www.whatwg.org/specs/web-apps/current-work/complete/commands.html#peer-to-peer-connections |
| 08:17 | <Hixie> | well, we have a skeleton |
| 08:17 | <Hixie> | i don't think it makes much sense to do video over websockets |
| 08:17 | <Hixie> | you'd want to do a direct connection to the other peer |
| 08:17 | <Hixie> | which is what the ConnectionPeer stuff is about |
| 08:18 | <Hixie> | i tried making a skeleton for someone to define ICE/STUN for this: http://hixie.ch/specs/rtc-skeleton/ |
| 08:18 | <Hixie> | no takers yet though |
| 08:18 | <Hixie> | (i know nowhere near enough to fill in that spec) |
| 08:19 | <annevk> | Hixie, even stuff like Skype often carries it via an intermediate server because of firewall issues |
| 08:20 | <Hixie> | yeah, as a last resort |
| 08:20 | <Hixie> | but the browser would do that, you wouldn't want to tunnel it over websockets |
| 08:20 | <annevk> | from what I hear from our Unite stuff P2P fails often |
| 08:21 | <annevk> | the browser would do that? you mean browsers need to set up intermediate servers and such? |
| 08:21 | <annevk> | and start paying the traffic bill? |
| 08:21 | <annevk> | that seems somewhat unlikely |
| 08:24 | <othermaciej> | my favorite part of every TPAC is when Henri gets up to the mic for a question |
| 08:25 | <nessy> | Hixie: I think that whole RTC group wants to have a go at that spec by the looks of it |
| 08:26 | <Hixie> | annevk: no, i mean the browser would fall back to a given routing server, you wouldn't have to create a WebSocket connection, etc |
| 08:26 | <Hixie> | nessy: have i been missing mail? i didn't see any rtc mail after i sent that proposal in |
| 08:26 | <nessy> | no, I don't think so either |
| 08:27 | <Hixie> | looked to me more like the whole TRC group wanted someone else to have a go at it :-) |
| 08:27 | <nessy> | but doesn't that whole meeting indicate that they want to attack the issue? |
| 08:27 | <annevk> | Hixie, oh, so you give a peer URL and a routing URL? |
| 08:27 | <weinig> | good morning annevk! |
| 08:27 | <annevk> | hey there weinig |
| 08:27 | <Hixie> | annevk: you give whatever ICE/STUN want, i'm not sure what that is exactly and am hoping to make it NMP |
| 08:27 | <Hixie> | or rather, SEP |
| 08:27 | <Hixie> | (Not My Problem / Someone Else's Problem) |
| 08:28 | <annevk> | fair enough |
| 08:28 | <annevk> | I do sort of think we want to have video -> websocket too |
| 08:28 | <annevk> | for live broadcast |
| 08:28 | <Hixie> | sure, eventually |
| 08:28 | <annevk> | and it seems easier than video -> p2p, but I could be missing things |
| 08:28 | <Hixie> | maybe |
| 08:29 | <othermaciej> | what kind of information do peers need to have previously exchanged to use ICE/STUN to connect? |
| 08:29 | <annevk> | weinig, #tpac-chat is the backchannel basically |
| 08:29 | othermaciej | has not much idea how this stuff works |
| 08:29 | <othermaciej> | for a live broadcast with the web client doing the broadcasting? |
| 08:29 | <othermaciej> | i.e. streaming upload of video? |
| 08:29 | <Hixie> | othermaciej: they need a low-bandwidth JS-supported communication channel, as i understand it, it's not just here's some info, go ahead. |
| 08:29 | <weinig> | annevk: ah, cute |
| 08:30 | <Hixie> | anyway, i should head to bed |
| 08:31 | <othermaciej> | I wish I knew how FaceTime does the initial exchange of info |
| 08:31 | eighty4 | hates minified js files |
| 08:31 | <MikeSmith> | the first time I ever saw hsivonen was when he came up to the mic at sessions at XTech 2005 or whatever year it was |
| 08:31 | <othermaciej> | good night! |
| 08:32 | <othermaciej> | heh |
| 08:32 | <othermaciej> | my favorite thing about his questions/comments is that 95% of the time it's clear the speaker did not actually understand what hsivonen was talking about, even though it makes perfect sense |
| 08:32 | <MikeSmith> | lol |
| 08:32 | <MikeSmith> | yeah |
| 08:33 | <MikeSmith> | or most of the people in the room |
| 08:33 | <annevk> | also sad |
| 08:33 | <MikeSmith> | I was expecting some reaction to his comment today |
| 08:33 | <MikeSmith> | about the parser |
| 08:33 | <MikeSmith> | but I guess I was expecting too much |
| 08:34 | <othermaciej> | I was hoping after glazou suggested "XML2", that someone would suggest to call it XML5 |
| 08:35 | <annevk> | at one point I wrote a mostly functioning XML5 parser |
| 08:35 | <MikeSmith> | maybe it would be more effective to just go up to the mic, say, "This problem has already been solved.", and then just walk away from the mic |
| 08:35 | <annevk> | MikeSmith, that's great |
| 08:37 | <Rik`> | ah ah, annevk is a women again in a french blog |
| 08:37 | <Rik`> | http://www.macgeneration.com/news/voir/175132/html5-le-tableau-d-honneur-du-w3c-conteste |
| 08:37 | <Rik`> | "La représentante d'Opera" |
| 08:37 | <weinig> | Rik`: one of the top women in Web Standards? |
| 08:38 | <othermaciej> | annevk should get a sex change just to stop confusing people |
| 08:38 | <annevk> | also sexy |
| 08:39 | <Rik`> | othermaciej: don't laugh, you are said to be "annevk's equivalent at Apple" |
| 08:40 | <othermaciej> | lol |
| 08:40 | <zcorpan> | why, do people think othermaciej is a girl too? |
| 08:41 | <weinig> | zcorpan: you mean Macie J? |
| 08:41 | <othermaciej> | sometimes people think my name is "Macie J. Stachowiak" and Macie might be a woman's name in the US, but then again, could be a man's name too |
| 08:41 | <Rik`> | zcorpan: no no, it says "His counterpart at Apple" |
| 08:41 | <Rik`> | the comments of this article are so wrong… |
| 08:43 | <Rik`> | oh wait no, othermaciej is considered a woman in this article |
| 08:44 | <annevk> | I do more background check than these journalists on a day-to-day basis |
| 08:44 | <annevk> | sad |
| 08:45 | <annevk> | but also funny :) |
| 08:46 | <Rik`> | I wouldn't consider them journalists |
| 08:47 | <MikeSmith> | they are more like "journalists light" |
| 08:47 | <MikeSmith> | like lite beer |
| 08:48 | <annevk> | that makes it sound quite bad |
| 08:50 | <MikeSmith> | we should start an "annevk's equivalent at…" meme |
| 08:54 | <nessy> | I'm about to start checking if IE really does 100% on video - I really doubt it |
| 08:54 | <zcorpan> | MikeSmith: so apparently you have some competition in generating the best html version of es5 |
| 08:54 | <MikeSmith> | zcorpan: may a thousand flowers grow |
| 08:55 | <zcorpan> | MikeSmith: http://people.mozilla.org/~jorendorff/es5.html |
| 08:55 | <MikeSmith> | yep |
| 08:55 | <MikeSmith> | looks better than what I put together |
| 08:55 | <MikeSmith> | more hyperlinks |
| 09:01 | <zcorpan> | but no ToC |
| 09:02 | <MikeSmith> | oh |
| 09:02 | <MikeSmith> | well, that's easy enough to fix |
| 09:02 | <MikeSmith> | he could just use the online HTML5 outliner |
| 09:02 | <MikeSmith> | and generate one |
| 09:03 | <MikeSmith> | and then just copy that generated TOC in |
| 09:03 | <MikeSmith> | that's how I made the one I included |
| 09:03 | <MikeSmith> | I think |
| 09:49 | <zcorpan> | i wonder if i should suggest that we move away from http-like headers and instead use JSON or something simpler for websockets |
| 09:49 | <zcorpan> | if we're going to XOR it, there's no reason to use http-like headers |
| 09:52 | <mnot> | FWIW, I think that's worth considering; the HTTP-ish-ness of the current proposal misleads people. |
| 09:55 | <annevk> | XML! |
| 09:56 | <annevk> | no wait, RDF/XML! |
| 11:00 | <slinkcoding> | re |
| 11:13 | <micheil> | abarth: ping? |
| 11:13 | <micheil> | hmm, maybe not, it's kinda late state-side |
| 13:43 | <AryehGregor> | Hixie, didn't I already say at least twice what the difference is? http://www.w3.org/Bugs/Public/show_bug.cgi?id=10808#c17 |
| 13:44 | <AryehGregor> | Hixie, go to data:text/html,<textarea dir=rtl></textarea> and type a few sentences in English. That's what you get when you try to type in Hebrew on any LTR site, i.e., practically any site. One use-case of bug 10808 is avoiding that. |
| 13:45 | <AryehGregor> | Another use-case is inserting blocks of text where you might know the direction, but don't want to bother manually specifying direction on each paragraph, so a Hebrew paragraph inserted in a mostly-English document should magically be interpreted as RTL instead of LTR (so, e.g., a trailing period would be put in the correct place). |
| 13:48 | <AryehGregor> | bdi has nothing to do with any use-case where you're inserting entire paragraphs, since paragraphs are already isolated. |
| 13:48 | <AryehGregor> | Is that a good enough explanation, or should I explain it a fourth time? :) |
| 13:49 | <AryehGregor> | (Although yes, the use-cases do overlap a bit -- as Aharon points out, dir=auto would also be useful if you're inserting inline stuff of unknown directionality. But they're mostly disjoint.) |
| 13:56 | <AryehGregor> | (I need to get around to reviewing what Aharon last said on bug 10808. I'm quite sure his solution is overcomplicated and too cautious and generally wrong, but I doubt my proposal is particularly good either, and I need to review his objections to it in detail.) |
| 18:08 | <Hixie> | AryehGregor: yeah i ended up understanding it better when i reread the bug, but thanks |
| 19:11 | <AryehGregor> | Grr, validator.nu uses POST instead of GET now? No copy-pasteable URLs. |
| 19:11 | <AryehGregor> | Doesn't seem to even accept GETs. |
| 19:12 | <AryehGregor> | Or is this some sneaky anti-XSS thing interfering or whatnot? |
| 19:12 | <AryehGregor> | How obnoxious. |
| 19:18 | <AryehGregor> | Oh, I can still use data URLs. |
| 22:13 | <Hixie> | anyone write arabic or hebrew? |
| 22:15 | <annevk> | your colleague Aharon that is at TPAC prolly does :) |
| 22:16 | <Hixie> | good point |
| 22:16 | <Hixie> | thanks |
| 22:18 | <hsivonen> | oh, great. the host of hsivonen.iki.fi seems to be down... |
| 22:19 | <Hixie> | on another note, does anyone have a page that, given text, gives back a list of unicode codepoints? my tool for doing that is buggy with non-ascii sadly |
| 22:19 | <hsivonen> | I really should take the time to relocate it to a more reliable host |
| 22:19 | <Hixie> | if you pick dreamhost, dibs on being your reference! :-P |
| 22:20 | <drclue> | Just out of curiosity , does anyone know if there would be an issue with a regular "ws:" websocket trying to connect from a page delivered with a red https in chrome? |
| 22:20 | <Hixie> | (but i don't think dreamhost would be first choice for reliability) |
| 22:21 | <annevk> | you can do it in JavaScript quite easily for everything that fits in 16 bits |
| 22:23 | <hsivonen> | Hixie: I'm considering the VM that now has about.validator.nu |
| 22:23 | <hsivonen> | it lacks PHP and Jython and GCC |
| 22:23 | <hsivonen> | I have crazy legacy stuff running in cron |
| 22:25 | <drclue> | I've got a websocket server that has been running fine, but I'm trying to get it happy in a particular VPS situation and the browser stalls at the handshake. |
| 22:26 | <Hixie> | annevk: true, i'll do that |
| 22:28 | <Aleoss> | HTML5 REQUEST: Is it possible you guys could add a maxlength attribute to the <textarea> element? |
| 22:29 | <Hixie> | Aleoss: done |
| 22:29 | <Aleoss> | Oh? -checks the spec again- |
| 22:29 | <Hixie> | :-) |
| 22:31 | <Aleoss> | So it is. Missed that while looking over the spec. |
| 22:35 | <Hixie> | is hebrew "he"? |
| 22:35 | <Hixie> | it appears so |
| 22:39 | <hober> | Hixie: I could make an Emacs command to replace chars with codepoints if you'd like |
| 22:46 | <hober> | Hixie: (defun hixie-codepointize-region (start end) "Get the Unicode codepoints of the chars between START and END." (interactive "r") (goto-char start) (while (< (point) end) (message "U+%04X" (encode-char (char-after) 'ucs)) (forward-char))) |
| 22:46 | <hober> | only lightly tested |
| 22:47 | <hober> | select a region, M-x hixie-codepointize-region RET, and you'll have all the codepoints in *Messages* |
| 22:48 | <Hixie> | actually the reason i need it is so that i can send just ASCII to my emacs :-) |
| 22:48 | <Hixie> | my editing environment is not 8-bit clean, sadly |
| 22:48 | <david_carlisle> | hixie: http://people.w3.org/rishida/tools/conversion/ |
| 22:48 | <Hixie> | it's ok, i did it in js when anne suggested it :-) |
| 22:53 | <hober> | ahh |
| 23:45 | <nessy> | Hixie: I recently saw some SRT files that had metadata in a first cue with the cue length 0, i.e. going from 0 —> 0; I can see that as the only way to introduce file-wide metadata without breaking backwards compatibility. Do you have a better idea? |