| 00:46 | <Yuhong> | <annevk> apparently http://tools.ietf.org/html/rfc6648 was not read |
| 00:46 | <Yuhong> | X-XSS-Protection dates back to IE8. |
| 00:51 | <Yuhong> | <Hixie> http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1905 |
| 00:51 | <Yuhong> | <Hixie> (also, anyone with IE around? what does IE put in the log for that test?) |
| 00:51 | <Yuhong> | In IE9 mode: false, true |
| 00:52 | <Yuhong> | IE9 in IE8 mode: true |
| 00:52 | <Yuhong> | IE9 in IE7 mode: true |
| 06:52 | <hsivonen> | the attempts to spin the claim that polyglot leads to better structure and quality into something non-bogus are embarrassing. this really is a déjà vu all over again. |
| 06:54 | <Hixie> | any idea what the real reason people want it is yet? just misguided, or is there some ulterior reason? |
| 06:57 | <hsivonen> | Hixie: I think the reason TimBL/TAG wanted it was that they were distressed about XML toolchains getting cast aside and didn’t realize there were better solutions to the problem than getting everyone to produce polyglot markup |
| 06:57 | <hsivonen> | (dunno what’s TimBL’s current thinking after the HTML—XML Task Force report) |
| 06:58 | <hsivonen> | as for why Sam wants it, Sam’s intentions are always mysterious |
| 06:59 | <hsivonen> | I also can’t tell why Leif wants it |
| 06:59 | <hsivonen> | polyglot does have certain party trick hack value which makes it appealing |
| 06:59 | <matjas> | jgraham: re: http://wiki.whatwg.org/index.php?title=Web_ECMAScript&diff=3938&oldid=3798 could you chime in here please? https://github.com/whatwg/javascript/issues/1 |
| 07:00 | <hsivonen> | Hixie: FWIW, looking at the old archives, the TAG didn’t develop its wish for polyglot on its own. Someone told them about Sam using polyglot markup on his blog. |
| 07:01 | <MikeSmith> | interesting |
| 07:02 | <hsivonen> | the email that started it all: http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2006-November/008035.html |
| 07:02 | <hsivonen> | aka. “the white pebble” |
| 07:02 | <hsivonen> | see http://www.intertwingly.net/blog/2006/12/01/The-White-Pebble |
| 07:14 | <hsivonen> | fun fact: the first occurrence of the word “polyglot” on public-html is in an email where Hixie calls it a ridiculous idea: http://lists.w3.org/Archives/Public/public-html/2008Aug/0301.html |
| 07:16 | <hsivonen> | the first mention on www-tag is in an email written by me. oops. |
| 07:18 | <hsivonen> | who coined the term polyglot? did Hixie do so in passing in the above email or was the term already used on IRC or blogs? |
| 07:19 | <hsivonen> | first mention on the whatwg list seems to be http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2009-August/021725.html |
| 07:27 | <hsivonen> | So TAG’s request for the HTML WG to create the polyglot spec emerged from a TAG meeting that Sam attended (or called into?) |
| 07:28 | <hsivonen> | are TAG f2f meetings minuted? |
| 07:29 | <hsivonen> | ah. here http://lists.w3.org/Archives/Public/www-tag/2010Apr/0060.html |
| 07:29 | <hsivonen> | oh. MikeSmith was there, too. |
| 07:29 | <SimonSapin> | what wrong with polyglot? pointlessly taking editing resources? |
| 07:33 | <hsivonen> | SimonSapin: pointlessly taking the resources of Web authors (if they are made believe they should do polyglot) or software developers (if authors who’ve been made believe they want polyglot demand polyglot features) |
| 07:33 | <hsivonen> | SimonSapin: also taking review resources at the WG |
| 07:37 | <zcorpan> | do we have a canned response in the form of a whatwg wiki page to why appendix c/polyglot is a bad idea? |
| 07:39 | <hsivonen> | unfortunately, it looks like my 386 limit is going to be exceeded at some point and I will end up writing a canned response |
| 07:49 | <hsivonen> | MikeSmith: I just learned that you saved HTML5 from having an Appendix C in the sense of having the polyglot stuff be an actual appendix. |
| 07:49 | <hsivonen> | (from reading TAG minutes) |
| 07:50 | <MikeSmith> | oh |
| 07:51 | <hsivonen> | MikeSmith: no recollection? |
| 07:52 | <MikeSmith> | yeah vaguely |
| 07:52 | <MikeSmith> | f2f meeting in Boston? |
| 07:52 | <hsivonen> | yeah |
| 07:56 | <MikeSmith> | the main thing I can remember for those discussions is that I didn't think that document had much chance of actually being written |
| 07:57 | <MikeSmith> | because there was lots of talk about what a great idea it would be if somebody would write it |
| 07:57 | <MikeSmith> | but as usual nobody showing any sign of actually planning to do that |
| 08:04 | <hsivonen> | gotta love how TimBL’s requirements didn’t get minuted: “TBL: my requirement is [everybody talks at once]” |
| 08:05 | <MikeSmith> | which minutes? |
| 08:06 | <hsivonen> | MikeSmith: the same as above |
| 08:07 | <MikeSmith> | oh god |
| 08:08 | <MikeSmith> | I think I tried to block some of this out of my memory |
| 08:08 | <MikeSmith> | was just reading the decentralized extensibility part |
| 08:11 | <MikeSmith> | I remember being back home after that meeting -- either that one or an earlier one -- and I was speaking at a meeting of the W3C Japan members |
| 08:12 | <MikeSmith> | and someone asked me about use of namespaces in text/html |
| 08:12 | <MikeSmith> | and I said, We'll never have namespaces in text/html |
| 08:12 | <MikeSmith> | but I forgot that Henry Thompson was in the room |
| 08:12 | <MikeSmith> | he was visiting at the time for something |
| 08:13 | <MikeSmith> | and he spoke up right away to say that No, there is still ongoing discussion about that |
| 08:14 | <Hixie> | hsivonen: the term polyglot is just the term used for any file that is in multiple languages at once |
| 08:14 | <zcorpan> | annevk: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1909 |
| 08:15 | <Hixie> | hsivonen: btw, it taking the htmlwg's time up isn't a big deal if the people whose time we care about aren't in htmlwg |
| 08:15 | <zcorpan> | annevk: btw, don't we want the white background in the logo? |
| 08:15 | Hixie | gives hsivonen a look :-P |
| 08:16 | <hsivonen> | MikeSmith: regarding the HTML parser bugs you mentioned the other day, http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1910 WFM in Firefox |
| 08:17 | <Hixie> | nn |
| 08:17 | <MikeSmith> | it can take time up when people start (re)lobbying to have polyglot-checking supporte added to the validator -- their logic being, there's a spec for it, so we should therefore provide some support for it |
| 08:18 | <MikeSmith> | hsivonen: yeah, I noticed that too |
| 08:18 | <MikeSmith> | that's why I didn't file a bug for it |
| 08:18 | <MikeSmith> | I think they were testing with code from February |
| 08:18 | <zcorpan> | so polyglot could instead be Guidelines for Spec Writers Using Anolis With Libxml2 Instead Of Html5lib |
| 08:18 | <MikeSmith> | heh |
| 08:18 | <MikeSmith> | yeah |
| 08:18 | <zcorpan> | 1. If you're Hixie, don't omit </td> tags |
| 08:19 | <hsivonen> | zcorpan: yes |
| 08:19 | <annevk> | zcorpan: heh nice |
| 08:19 | <zcorpan> | 2. Your spec is small enough to not take ages to generate, use html5lib instead. |
| 08:19 | <annevk> | zcorpan: feel free to commit and update the twitter account (that's all that's needed) |
| 08:20 | <hsivonen> | just when I thought I could finally get to making a libxml2-compatible version of the V.nu parser, I get this Rust thing on my plate |
| 08:20 | <MikeSmith> | hsivonen: in that paper they found that WebKit has the same bug, but when I tested in WebKit I couldn't reproduce it there either. So I guess you must have fixed it between February and now, and also it was fixed in WebKit too |
| 08:20 | <hsivonen> | MikeSmith: I can’t recall fixing anything of that nature in February |
| 08:21 | <hsivonen> | or since February |
| 08:21 | <MikeSmith> | hmm ok |
| 08:21 | <MikeSmith> | weird then |
| 08:21 | <hsivonen> | Hixie: do you recall why you have foreign-namespace stuff on the "special" list of the parsing algorithm? |
| 08:22 | <hsivonen> | Hixie: when can an "is special?" check end up being performed on an SVG or MathML stack node? |
| 08:23 | <MikeSmith> | I vote for you work on the libxml2-compatible v.nu parser before the Rust one |
| 08:24 | <annevk> | zcorpan: as for the white background, dunno, maybe that should be added back? seems kinda nice if the logo just works on photos and such, but that's largely theoretical :) |
| 08:24 | <MikeSmith> | btw it's possible to plug a different parser into libxml2? |
| 08:24 | <hsivonen> | MikeSmith: not that I know of |
| 08:24 | <hsivonen> | MikeSmith: I was thinking of having a new entry point for parsing but returning a libxml2 tree |
| 08:24 | <MikeSmith> | ah OK |
| 08:25 | <hsivonen> | So my isScoping() and isSpecial() checks assume it’s impossible for those questions to be posed about foreign nodes |
| 08:25 | <hsivonen> | great optimization or horrible bug? |
| 08:27 | <zcorpan> | annevk: the original has the white background. i guess me being used to it makes me think it looks better with the white. |
| 08:27 | <zcorpan> | annevk: maybe i should ask divya, she likes bikesheds i heard |
| 08:28 | <MikeSmith> | wtf why does live dom viewer add a <style type="text/css"></style> node in that example? |
| 08:28 | <MikeSmith> | I never noticed it doing that before |
| 08:28 | <MikeSmith> | hsivonen: ↑ |
| 08:29 | <hsivonen> | MikeSmith: what example? I see no style with 1910 |
| 08:29 | <MikeSmith> | weird |
| 08:29 | <MikeSmith> | I do in Chrome |
| 08:30 | <MikeSmith> | <!DOCTYPE HTML><html><head><style type="text/css"></style></head><body><ruby><button><rp></rp></button></ruby></body></html> |
| 08:31 | <hsivonen> | ooh. not a horrible bug. just a more clever optimization elsewhere |
| 08:31 | <MikeSmith> | chrome dev-version bug I guess |
| 08:31 | <hsivonen> | I’m so old I forget what code I have written |
| 08:32 | <zcorpan> | MikeSmith: can't reproduce in 24.0.1312.5 dev |
| 08:32 | <MikeSmith> | zcorpan: crazy |
| 08:32 | <MikeSmith> | that's what I'm running too |
| 08:32 | <zcorpan> | MikeSmith: any addons? |
| 08:32 | <MikeSmith> | ah yeah |
| 08:32 | <MikeSmith> | several |
| 08:32 | <MikeSmith> | lemme try turning some off |
| 08:33 | <MikeSmith> | ah |
| 08:33 | <MikeSmith> | Adblock |
| 08:34 | <hsivonen> | MikeSmith: so all in all, I have just the <dd><optgroup><dd> bug to fix from the paywalled paper? |
| 08:34 | <zcorpan> | so adblock advertises its presence to the page |
| 08:35 | <MikeSmith> | hsivonen: yes, just that one bug |
| 08:36 | <MikeSmith> | zcorpan: I guess so |
| 08:36 | <MikeSmith> | I just starting using Adblock today |
| 08:37 | <zcorpan> | annevk: how are the pngs created? |
| 08:42 | <annevk> | http://www.fileformat.info/convert/image/svg2raster.htm with 500x500 |
| 08:42 | <annevk> | I guess I should add a README.md for that |
| 08:44 | <SimonSapin> | annevk: how about using http://cairosvg.org/ ? :) |
| 08:44 | <annevk> | zcorpan: btw, (hope you read the logs), you can omit "," in a path if the next character is a "-" |
| 08:44 | <annevk> | SimonSapin: why download something if I can use the web? |
| 08:45 | <hsivonen> | annevk: what do you need the bitmaps for? |
| 08:45 | <SimonSapin> | annevk: why depend on a service if you can run stuff locally? |
| 08:46 | <annevk> | hsivonen: twitter |
| 08:46 | <annevk> | SimonSapin: because they're easier |
| 08:46 | <SimonSapin> | ok |
| 08:47 | <SimonSapin> | whatever works for you :) |
| 08:48 | <SimonSapin> | MikeSmith: "btw it's possible to plug a different parser into libxml2?" maybe not in libxml2, but in lxml yes. See http://lxml.de/html5parser.html |
| 08:48 | <MikeSmith> | yeah that I knew |
| 08:49 | <annevk> | zcorpan: see logs |
| 08:49 | <zcorpan> | yep |
| 08:49 | <annevk> | biab |
| 08:49 | <SimonSapin> | zcorpan: "so polyglot could instead be Guidelines for Spec Writers Using Anolis With Libxml2 Instead Of Html5lib" is that libxml2’s XML parser, not HTML? I know the latter is not quite html5 but is it that bad? |
| 08:50 | <SimonSapin> | (it’s what I use in WeasyPrint) |
| 08:50 | <zcorpan> | SimonSapin: html |
| 08:51 | <SimonSapin> | so how is polyglot relevant? |
| 08:51 | <hsivonen> | SimonSapin: apparently having explicit end tags helps the HTML parser in libxml2 |
| 08:51 | <zcorpan> | SimonSapin: http://intertwingly.net/blog/2012/11/09/In-defence-of-Polyglot |
| 08:53 | <zcorpan> | however i think libxml2 also screws up hex character references (so the non-normative column in the entities table in the spec is wrong) |
| 08:53 | <zcorpan> | but the polyglot spec doesn't discuss any solutions to that problem |
| 08:54 | <zcorpan> | the html5 editors seem to have chosen "use html5lib" as the solution; i'm not sure if they care about it being slow |
| 08:59 | <zcorpan> | annevk: svg2raster is too slow for me on the train |
| 09:01 | <hsivonen> | annevk: boo for twitter not accepting SVG avatars |
| 09:04 | <MikeSmith> | "I'd like to propose as a constructive strategy not to flame/offend everybody right off the bat." |
| 09:04 | <MikeSmith> | too late for that |
| 09:20 | <hsivonen> | MikeSmith: context? |
| 09:21 | <MikeSmith> | http://lists.w3.org/Archives/Public/public-webapps/2012OctDec/thread.html#msg430 |
| 09:23 | <hsivonen> | ah. |
| 09:23 | hsivonen | expected something HTML-related |
| 09:34 | <annevk> | hsivonen: well, SVG is way more complex from a security pov, but yeah, they should fix that |
| 09:38 | <annevk> | zcorpan: I guess I can do that once I add the javascript logo for matjas |
| 09:39 | <zcorpan> | annevk: ok. thanks. i'm at the office now, but rasterizing the logos one by one from a web form seems boring |
| 09:40 | <SimonSapin> | zcorpan: well, you could use a local, scriptable tool :) … if you like |
| 09:47 | <hsivonen> | yay. still not resolved: https://www.w3.org/Bugs/Public/show_bug.cgi?id=13392 a bug arising from stating Leif’s preference rather than just the conclusions arising from other specs |
| 09:56 | <sangwhan> | Noob question, apart from being a effective tormenting agent for browser engineer one has a grudge against, what upsides *do* polyglot bring? |
| 09:56 | <sangwhan> | s/engineer one/engineers one/ |
| 09:57 | <annevk> | zcorpan: just generating logo-dom.png and uploading it to twitter does not seem too bad |
| 09:57 | <zcorpan> | annevk: i changed the background on all logos except encoding |
| 09:58 | <annevk> | if it doesn't affect the twitter icon I don't think I'm going to bother either |
| 10:07 | <hsivonen> | sangwhan: no real upsides of its own. but there’s overlap between other sets of restrictions that have upsides and the polyglot set of restrictions. |
| 10:08 | <hsivonen> | sangwhan: for example, if you need explicit end tags to deal with libxml2, then polyglot has that upside |
| 10:08 | <hsivonen> | but also misses other relevant things and brings along a bunch of irrelevant things |
| 10:15 | <MikeSmith> | I thought for some people at least the original motivation was that they wanted to use XML toolchains to create their XHTML documents but IE did not handle application/xhtml+xml so they wanted their XHTML documents to work in IE when they served them as text/html |
| 10:15 | <sangwhan> | hsivonen: ok, that sort of makes sense - although that might not be a strong enough argument to convince people en masse |
| 10:22 | <MikeSmith> | hsivonen: about the validator stats, would it be worth adding stats on the input mechanisms -- address vs file upload vs text field? |
| 10:23 | <MikeSmith> | oh an also post with entity body |
| 10:23 | <MikeSmith> | if you think it's worth it I can try adding it |
| 10:29 | <annevk> | zcorpan: you could write your own web service using <canvas> that converts those images :) |
| 10:33 | <zcorpan> | annevk: that does seem less boring |
| 10:34 | <annevk> | I created http://resources.whatwg.org/logo-javascript.svg btw |
| 10:34 | <annevk> | and opened https://github.com/voodootikigod/logo.js/issues/32 for contributing it back to the original project |
| 10:34 | <annevk> | matjas: ^^ |
| 10:35 | <MikeSmith> | annevk: you should use http://es5.github.com/js-mascot.svg instead |
| 10:38 | <annevk> | MikeSmith: heh |
| 10:43 | <annevk> | IETF does session recording a lot better than W3C, funny tha |
| 10:43 | <annevk> | t |
| 10:43 | <annevk> | http://ietf85.conf.meetecho.com/index.php/Recorded_Sessions |
| 10:43 | <annevk> | video + audio + minutes |
| 10:44 | <annevk> | with W3C you never know what was said unless you were there |
| 10:46 | <annevk> | jgraham: https://github.com/whatwg/javascript/issues/1 |
| 11:16 | <annevk> | http://html5sec.org/#19 is interesting |
| 11:16 | <annevk> | didn't know supporting other Mac encodings can actually be a security risk |
| 11:42 | <hsivonen> | MikeSmith: nope, polyglot does not really help in the “I’ve got an XML serializer but want to support IE8” case |
| 11:43 | <hsivonen> | MikeSmith: if you have a use for the input stats, I don’t see any reason not to add those |
| 11:50 | <hsivonen> | hmm. why is Gecko passing zcorpan’s BOMless UTF-16 test... |
| 12:01 | <hsivonen> | the test doesn’t fail, because it doesn’t get run |
| 12:02 | <matjas> | logo suggestion for domparsing.spec.whatwg.org: <dom/> |
| 12:02 | <matjas> | and for XHR: “!Ajax” |
| 12:04 | <hsivonen> | anomalous – Ms2ger is not on IRC |
| 12:04 | <annevk> | heh |
| 12:11 | <annevk> | matjas: preferably non-text in a way that works with ? or a circle |
| 12:12 | <annevk> | hsivonen: he's studying, only online during the evenings and weekends I think |
| 12:24 | <zcorpan> | hsivonen: bug in the test? |
| 12:27 | <zcorpan> | annevk: the JS logo seems a bit dull :-/ |
| 12:28 | <annevk> | don't let it keep you up |
| 12:34 | <zcorpan> | annevk: how about a mashup between the whatwg logo and the dutch ajax logo? |
| 12:35 | <annevk> | zcorpan: for XMLHttpRequest? |
| 12:35 | <zcorpan> | yeah |
| 12:35 | <annevk> | zcorpan: I think that could work |
| 12:35 | <annevk> | kinda obscure and fun |
| 12:35 | <annevk> | zcorpan: make it :) |
| 12:38 | <zcorpan> | copyright might be slightly problematic. not sure we want to rely on "fair use"s |
| 12:39 | <zcorpan> | s/s// |
| 12:39 | <zcorpan> | http://en.wikipedia.org/wiki/File:Ajax_Amsterdam.svg |
| 12:40 | <zcorpan> | LimeChat tries to show that url as an inline image and fails |
| 12:41 | <zcorpan> | i guess i can make a new logo resembling it |
| 12:58 | <Ms3ger> | hsivonen: you called, sir? |
| 12:59 | <annevk> | zcorpan: I suspect LimeChat of doing file extension sniffing |
| 12:59 | <zcorpan> | annevk: yeah |
| 13:00 | <hsivonen> | Ms3ger: it seems the test suite import script doesn’t generate the appropriate ^headers^ files |
| 13:01 | <hsivonen> | for *.bogus.css and *.windows1250.css |
| 13:02 | <hsivonen> | Ms3ger: should it, or should the HTTP server be changed to recognize .bogus. and .windows1250. ? |
| 13:02 | <Ms3ger> | Ah |
| 13:02 | <Ms3ger> | Mm, how do they work upsteam? htaccess? |
| 13:04 | <hsivonen> | Ms3ger: .htaccess, I believe |
| 13:05 | Ms3ger | thinks |
| 13:09 | <Ms3ger> | hsivonen: could you send me an email? I can probably try to fix it tonight |
| 13:10 | <hsivonen> | Ms3ger: ok |
| 13:12 | <Ms3ger> | Thanks |
| 13:21 | <zcorpan> | some images in https://github.com/joyent/node/pull/1580 are bitrotting :-( |
| 13:21 | <hsivonen> | Excercise: Apply Mark Pilgrim’s taxonomy of developers to this case: http://code.google.com/p/chromium/issues/detail?id=94369 |
| 13:39 | <zcorpan> | does the whatwg list prune attachments? |
| 13:41 | <annevk> | hsivonen: ah, the old argument of one URL per resource, or one for many |
| 13:41 | <annevk> | zcorpan: dunno, you could test |
| 13:41 | <annevk> | but better to link to resources |
| 13:42 | <zcorpan> | boris claimed he had attached a test case but i don't see it |
| 13:43 | <annevk> | hsivonen: reading a bit on it seems to be about content negotiation, the thing that trips up anyone everywhere |
| 14:35 | <GPHemsley> | annevk: I look at the spec locally when I'm editing it. |
| 14:36 | <jarek> | Hi |
| 14:36 | <jarek> | SVG 2 spec says that you can specify fill like this: |
| 14:36 | <jarek> | fill="#MyLightPurple" |
| 14:36 | <jarek> | but how does parser can know whether the fill value is IRI reference or hex color? |
| 14:37 | <jarek> | e.g. if fill="#ffffff" |
| 14:37 | <GPHemsley> | matjas, MikeSmith: Shouldn't bugs be filed in Bugzilla, not on GitHub? |
| 14:37 | <matjas> | GPHemsley: why? |
| 14:37 | <GPHemsley> | matjas: Because that's where the bugs for all the other specs are filed... |
| 14:37 | <matjas> | GPHemsley: but I’m much more comfortable with the GitHub Issues UI |
| 14:45 | <annevk5> | fwiw, GitHub issues is fine |
| 14:46 | <annevk5> | maybe we find out it works better and want to use it elsewhere |
| 14:47 | <annevk5> | jarek: no need to wrap it in url()? |
| 14:48 | <jarek> | annevk5: SVG2 spec says "Paint servers are used by including an IRI reference in a ‘fill’ or ‘stroke’ property (i.e. fill="#MyLightPurple")." |
| 14:49 | <annevk5> | jarek: seems like a bug in the example |
| 14:50 | <jarek> | annevk5: also, on WebKit $element.style.fill always returns IRI references without url(), is this standard behavior? |
| 14:51 | <annevk5> | also seems bogus, but TabAtkins might know more |
| 14:52 | <annevk5> | connection is kinda flaky :/ |
| 14:54 | <GPHemsley> | annevk5, matjas: Well, one of the downsides is that we're all subscribed to everything by default. |
| 14:54 | <matjas> | GPHemsley: that’s easy to fix, no? |
| 14:55 | <annevk5> | sure, I unsubscribed |
| 14:55 | <GPHemsley> | annevk5, matjas: Bugzilla has much better user/e-mail management in general, IMO |
| 14:55 | <annevk5> | matjas is doing the work and wants to try something else |
| 14:56 | <matjas> | thanks for understanding, annevk5 :) |
| 14:56 | <annevk5> | he does the work and that is all there is to it imo |
| 14:56 | <matjas> | sorry for the inconvenience, GPHemsley |
| 14:56 | GPHemsley | shrugs |
| 14:58 | <GPHemsley> | meanwhile, the repo itself isn't being used? |
| 14:59 | <hsivonen> | GPHemsley: should I resend the feedback I wrote on bugzilla to the mailing list? |
| 14:59 | <GPHemsley> | hsivonen: Couldn't hurt |
| 14:59 | <hsivonen> | GPHemsley: ok |
| 15:00 | <hsivonen> | meanwhile: https://www.w3.org/Bugs/Public/show_bug.cgi?id=19925 |
| 15:01 | <GPHemsley> | No, Firefox; just because a site forces me to download a file does not mean I have "chosen to open" it. |
| 15:06 | <Smylers> | matjas: Small correction: in http://mathias.html5.org/specs/javascript/#annex-b the 3rd bullet point (escape) is already covered by the 2nd one (escape and unescape). |
| 15:07 | <matjas> | Smylers: whoops, fixed |
| 15:08 | <hsivonen> | Importing testharness.js tests is harder than I expected. |
| 15:09 | <hsivonen> | I’d probably have spent less time manually repurposing zcorpan’s tests as mochitests than I’ve now spent trying to grok the import process. :-( |
| 15:10 | <annevk> | GPHemsley: Look, we're not trying to reach uniformity in specification development. We are trying to document the web (and sometimes beyond) and people have different approaches to doing so. |
| 15:10 | <GPHemsley> | annevk: So... the non-use of the repo was on purpose? |
| 15:10 | <annevk> | GPHemsley: If we put restrictions on what they can do, they might not do it altogether or we end up where in 2012 we still publish in text/plain ASCII (as a matter of saying) |
| 15:11 | <annevk> | GPHemsley: We lack rules (other than being civil) on purpose. Lacking rules facilitates innovation and people having a chance to do what they came here to do. |
| 15:12 | <annevk> | GPHemsley: The non-use of the repo is temporary as I understand it. It will be used and http://javascript.spec.whatwg.org/ too within the next few days. |
| 15:12 | <GPHemsley> | OK, that's all I was asking. |
| 15:12 | hsivonen | wonders if MP3 is the new text/plain. if the IETF uses XMPP instead of IRC in order to eat their own dogfood, shouldn’t they use Opus instead of MP3 to eat their own dogfood, too. |
| 15:13 | <annevk> | GPHemsley: I thought I should elaborate on my earlier "let matjas do as he pleases", sorry if it went a bit too far :) |
| 15:13 | <hsivonen> | at least MP3 does not sound as bad as text/plain looks |
| 15:14 | <GPHemsley> | hsivonen: Though I did have trouble attempting to listen to it in-browser; haven't investigated why that was, though. |
| 15:14 | <GPHemsley> | annevk: S'ok |
| 15:17 | <zcorpan_> | hsivonen: :-( why is it hard? |
| 15:18 | jgraham | hopes this is "because I haven't done it before" |
| 15:19 | <jgraham> | Maybe it could be a web service of some sort? |
| 15:22 | <hsivonen> | jgraham: because I can’t import failures from the try run |
| 15:23 | <hsivonen> | jgraham: because the failure setup script doesn’t behave like the README seems to say |
| 15:23 | <hsivonen> | jgraham: or because our build system changed and the log format is now different |
| 15:23 | <hsivonen> | jgraham: because .bogus. is not automagic |
| 15:24 | <hsivonen> | jgraham: because Ms2ger forgot one Makefile line in his patch that I expected to have that sort of thing covered |
| 15:25 | <hsivonen> | jgraham: because syncing the tree leads to a "REBUILD ALL THE THINGS" sort of phenomenon |
| 15:30 | <jgraham> | hsivonen: I am not entirely sure, but it sounds like bugs rather than an intrinsic problem? |
| 15:30 | <zcorpan_> | .bogus. not being automagic, i.e. .htaccess not working for you, seems like something we should resolve in some way. not sure how. |
| 15:31 | <zcorpan_> | the other issues seem less actionable from my and jgraham's perspective |
| 15:32 | <zcorpan_> | .htaccess works for opera, microsoft, w3c. but mozilla and webkit don't use an apache server for testing currently |
| 15:42 | <hsivonen> | zcorpan_: I didn’t mean to suggest these problems were actionable by you or jgraham |
| 15:42 | <zcorpan_> | hsivonen: yeah i understand that |
| 15:43 | <zcorpan_> | hsivonen: we're interested in improving the situation with tests, so it's good to know what the issues are |
| 16:52 | <GPHemsley> | Hmm... either Gecko doesn't allow SVG as favicons, or I broke something. |
| 16:58 | <GPHemsley> | hsivonen: I responded. |
| 17:33 | GPHemsley | wonders why html5lib rearranges the attributes on <link> |
| 17:34 | <SimonSapin> | GPHemsley: are the attributes in a dictionary? Python dicts are unordered |
| 17:34 | <GPHemsley> | wait... html5lib alphabetizes the attributes? o_0 |
| 17:35 | <GPHemsley> | SimonSapin: Nope, they are definitely alphabetized. |
| 17:35 | <GPHemsley> | alt height src width |
| 17:35 | <GPHemsley> | href rel type |
| 17:41 | <gsnedders> | GPHemsley: html5lib does not guarantee attribute order |
| 17:41 | <gsnedders> | GPHemsley: Any order is purely coincidental. |
| 17:41 | <GPHemsley> | riiiight |
| 17:41 | <gsnedders> | (Different tree-builders give different orders) |
| 17:41 | <GPHemsley> | ah |
| 17:42 | <gsnedders> | (Fundementally we cannot because none of the third-party tree-builders we rely on don't!) |
| 17:42 | <gsnedders> | s/don't/do/ |
| 17:42 | <GPHemsley> | I see |
| 17:54 | <jsbell> | Anyone know why ArrayBuffers are not one of the types enumerated in the structured clone algorithm? http://www.w3.org/TR/html5/common-dom-interfaces.html#safe-passing-of-structured-data |
| 17:54 | <jsbell> | i.e. was there discussion which led to them being excluded, or has the algorithm just not been updated since WebGL landed? |
| 17:55 | <gsnedders> | "If input is an object that another specification defines how to clone" |
| 17:55 | <gsnedders> | The Typed Array specification defines it. |
| 17:56 | <GPHemsley> | I have to say... SVG syntax is extremely opaque |
| 17:56 | <jsbell> | Ah, yes. |
| 17:56 | <gsnedders> | (It was previously in HTML, I believe) |
| 17:56 | <jsbell> | thx |
| 18:00 | <gsnedders> | GPHemsley: fantasai was wanted a guaranteed-order for attributes, FWIW. I think insertion order. |
| 18:01 | <GPHemsley> | gsnedders: if "insertion order" means "the way I typed it", then I concur |
| 18:01 | <dglazkov> | good morning, Whatwg! |
| 18:02 | <gsnedders> | GPHemsley: document order |
| 21:24 | <SimonSapin> | lang="en-US-x-Hixie" humm… |
| 21:28 | <FireFly> | SimonSapin: see http://ian.hixie.ch/bible/english :p |
| 21:44 | <bholley> | abarth: yt? |
| 21:45 | <annevk> | GPHemsley: I recommend browsers not supporting SVG as favicon as a bug in that browser and not working around it, but up to you |
| 21:46 | <abarth> | bholley: hi |
| 21:47 | <bholley> | abarth: hey, so I've been talking with bz, mrbkap, and jst about the Location stuff |
| 21:47 | <abarth> | ok |
| 21:49 | <bholley> | abarth: the general consensus is that people want to land it, given that we have the patches ready, and given that there's currently no accord among different UAs and the spec |
| 21:49 | <bholley> | abarth: and try to get in touch with MS in parallel to see what their position is |
| 21:50 | <abarth> | bholley: that sounds like a reasonable choice from your perspective |
| 21:51 | <abarth> | I doubt that anyone from MS would object to your making Gecko more like Trident |
| 21:51 | <bholley> | abarth: I don't want to be a jerk about it though :-( |
| 21:51 | <bholley> | abarth: it's just not clear to me how to resolve the current impasse here |
| 21:52 | <abarth> | bholley: the impasse is only that the different engines are constructed differently, so the engineering tradeoffs are different |
| 21:53 | <abarth> | its good that we have diversity in implementations, but it does lead to situations like this where what's of benefit to one engine is a cost to another |
| 21:53 | <abarth> | I don't think you should feel bad about aligning gecko's behavior with IE and opera |
| 21:54 | <bholley> | abarth: right. And Gecko's in a position to tilt the ship one way or another by throwing its weight a given direction, whereas WebKit isn't |
| 21:54 | <abarth> | my guess is that we won't change WebKit to match here unless there is a compatibility reason to do so |
| 21:54 | <abarth> | the engineering costs are pretty high |
| 21:55 | <abarth> | its possible that as ECMAScript evolves, the costs will get lower |
| 21:55 | <bholley> | abarth: yeah. It doesn't seem like a very high-priority issue |
| 21:55 | <abarth> | e.g., with harmony proxies and such |
| 21:55 | <bholley> | abarth: similar to aligning on document.domain behavior |
| 21:55 | <bholley> | abarth: (per the spec issue I raised a few months ago - similar roots in our different implementations) |
| 21:57 | <abarth> | document.domain I feel more passionate about :) |
| 21:58 | <abarth> | the consequences of changing its behavior are more likely to cause compatibility problems |
| 21:58 | <abarth> | (and I want the API to die in a fire) :) |
| 21:58 | <bholley> | abarth: heh, yeah |
| 21:58 | <bholley> | abarth: While we're talking, do you have any thoughts on the revocation issue? |
| 21:59 | <bholley> | abarth: per http://lists.w3.org/Archives/Public/public-script-coord/2012AprJun/0120.html |
| 22:03 | <abarth> | there's very little chance that we'd implement revocation in WebKit |
| 22:03 | <abarth> | its not actually needed for security |
| 22:03 | <abarth> | and we don't have any mechanism with which to implement it |
| 22:04 | <bholley> | abarth: do you think it's problematic to be unaligned here? |
| 22:04 | <abarth> | yes, i'd much prefer user agents to interoperate, especially on security checks |
| 22:06 | <bholley> | abarth: so the issue is that WebKit has no way to keep track of cross-scope references? |
| 22:07 | <abarth> | WebKit has no wrappers |
| 22:07 | <abarth> | especially not CrossOriginWrappers |
| 22:07 | <bholley> | abarth: they're just raw JS refs? |
| 22:18 | <jgraham> | Lack of interop. on th ocation object would be sadness |
| 22:18 | <jgraham> | *the location |
| 22:45 | <annevk> | jgraham: Opera caused lack of interop on IDNA |
| 22:45 | <annevk> | jgraham: not sure which is worse |
| 22:48 | <annevk> | for those interested in IDNA, http://www.alvestrand.no/pipermail/idna-update/2012-November/date.html has the "good" stuff |
| 22:49 | <annevk> | it's kinda sad IDNA is not a solved problem by now, but it's interesting enough |
| 23:02 | <zcorpan> | annevk: WIP http://software.hixie.ch/utilities/js/live-dom-viewer/saved/1911 |
| 23:03 | <annevk> | genius |
| 23:04 | <zcorpan> | with this in the textarea: |
| 23:04 | <zcorpan> | M17,66C21,63,43,76,37,68 |
| 23:04 | <zcorpan> | M33,75a8,8,0,0,16,8 |
| 23:04 | <zcorpan> | M50,84C68,49,57,90,63,77, |
| 23:04 | <zcorpan> | 66,58,75,87,71,69, |
| 23:04 | <zcorpan> | 68,62,74,65,78,66 |
| 23:04 | <zcorpan> | M65,62C57,64,64,71,67,64 |
| 23:04 | <zcorpan> | M47,60L43,69,44,69,51,66 |
| 23:04 | <zcorpan> | M50,52C47,60,53,54,52,60, |
| 23:04 | <zcorpan> | 50,68,62,60,57,62, |
| 23:04 | <zcorpan> | 58,54,65,60,59,54, |
| 23:04 | <zcorpan> | 48,42,20,25,29,37 |
| 23:06 | <annevk> | neat |
| 23:07 | <annevk> | you're better than I am at this |
| 23:07 | <annevk> | my skills fall apart when things go round :-) |
| 23:07 | <annevk> | "skills" |
| 23:11 | <zcorpan> | if you look at where the points go in a bezier curve, you kinda know where to click, and then i just tweak the values a bit |
| 23:12 | <annevk> | make a short documentary, "Creation of the XMLHttpRequest logo" |
| 23:12 | annevk | would watch it |
| 23:15 | <zcorpan> | "instead of using a proper authoring tool i hacked together my own, inefficient, and annoying-to-use "tool" in live dom viewer" |
| 23:16 | <zcorpan> | maybe i should work on an svg authoring tool that sucks less |
| 23:17 | <annevk> | "use Live DOM Viewer for all the things" |
| 23:17 | <annevk> | I do that too |
| 23:18 | <gsnedders> | x-x-big5 not in HTML in any more? |
| 23:18 | <annevk> | sometimes I need to write some HTML, so I open Live DOM Viewer |
| 23:19 | <annevk> | gsnedders: HTML is going to use http://encoding.spec.whatwg.org/ |
| 23:21 | <gsnedders> | Sorry, 386 occuring. |
| 23:21 | <Hixie> | the quirk and encoding logos are still the best ones |
| 23:39 | <zcorpan> | annevk: quirks.spec doesn't seem to update |
| 23:58 | <GPHemsley> | annevk: It was my own personal workaround, especially since the PNG was already there. I can certainly file a browser bug on the issue, though. |
