| 00:29 | <Hixie> | hahahahha |
| 00:30 | <Hixie> | wakaba_ calls the U01234 notation that hsivonen and I came up with the "Uhhhhh" notation |
| 00:30 | <Hixie> | I love it |
| 00:32 | <gsnedders> | Uh? |
| 00:32 | <gsnedders> | :P |
| 00:38 | <annevk> | Hixie, last checkin uses a double dot to end a sentence |
| 01:58 | <Hixie> | annevk: which sentence? |
| 01:59 | <Hixie> | found it |
| 02:15 | <Hixie> | wow are you seeing this ietf-http-wg thread? |
| 02:16 | <gsnedders> | Hixie: CSRF? |
| 02:16 | <Hixie> | yeah |
| 02:16 | <Hixie> | i think roy just said "i am always right" |
| 02:16 | <Hixie> | "i need no evidence" |
| 02:17 | <gsnedders> | Yeah, I think so too. |
| 02:20 | <Philip`> | Hooray, now all my fonts work with no reported conversion errors and no reported parsing errors from Pango and no visual difference from the original fonts in any browser as far as I can see |
| 02:20 | <Hixie> | maybe i should ask vint cerf to weigh in on the discussion and say "well with my 30 years of experience with the Internet, I outrank you" |
| 02:20 | <Philip`> | with all the kerning and ligatures and positioning-acute-accents-over-'i'-so-it-lines-up-with-the-dot and stacking half a dozen diacritics on top of each other and doing Arabic and all that stuff |
| 02:21 | <Hixie> | Philip`: nice |
| 02:21 | <Dashiva> | Hixie: Disregarding how productive (or not) it would be, it would be awesome |
| 02:21 | <Philip`> | and I've only found four bugs in Font::TTF, which is still pretty good |
| 02:22 | <Hixie> | Dashiva: if only i wasn't concerned about wasting his time :-) |
| 02:22 | <Philip`> | But this is much more painful than I initially expected it to be :-( |
| 02:22 | <Dashiva> | Hixie: He's spent 16 years on this, I'm sure a few minutes can sneak past :P |
| 02:23 | <Hixie> | Dashiva: i meant vint's time |
| 02:24 | <Dashiva> | But yeah, probably no point in getting into an argument over who has the most weight on an irrelevant scale |
| 02:24 | <annevk> | His argument about changing Referer is somewhat appealing, though I suppose getting those proxies updated will be hard. |
| 02:25 | <Dashiva> | He hasn't addressed path information in referer that I can see |
| 02:25 | <Hixie> | his whole argument is spurious. authors can't rely on referer or origin, but when they see origin, they can use it as a hard-and-fast rule. with referer, they can't ever do that. |
| 02:25 | <Hixie> | the whole point isn't to fix every browser or every server, it's to make it possible for server-browser combinations to be safer |
| 02:27 | <jruderman> | Philip`: i'm not a fan of text engines letting you stack a dozen diacritics on top of each other. it's kind of a security hole. |
| 02:27 | <Philip`> | jruderman: What makes it a security hole? |
| 02:27 | <jruderman> | Philip`: https://bugzilla.mozilla.org/attachment.cgi?id=288292 and https://bugzilla.mozilla.org/show_bug.cgi?id=supercombiner |
| 02:27 | <annevk> | I'm not sure I like the way how Adam did the draft. I would have preferred that when or when not Origin is included is stated somewhere else (e.g. in CORS). Currently it seems to introduce a change as with his draft even same origin requests would contain the Origin header, though maybe that is a good thing... |
| 02:28 | <Hixie> | annevk: tell him |
| 02:28 | <jruderman> | Philip`: lets you make stuff unreadable when you're only supposed to be able to stick in a line of text |
| 02:28 | <Hixie> | annevk: cc me or the whatwg list |
| 02:28 | <Philip`> | jruderman: Ah, right |
| 02:28 | <Hixie> | annevk: i can back you up if he disagrees :-) |
| 02:28 | <annevk> | Hixie, yeah, I will do it when I'm more awake :) |
| 02:30 | Philip` | realises that despite all the fancy typographic effects working correctly, he's still got a font which renders ")" upside-down :-( |
| 02:31 | <jruderman> | how do you know it's upside-down? isn't it symmetric wrt vertical flips? |
| 02:32 | <Dashiva> | Maybe the vertical positioning changes when it's upside down |
| 02:32 | <Philip`> | I mean it's rotated 180 degrees |
| 02:33 | <Philip`> | I think it's actually the '(' glyph that's meant to be rotated by 180 degrees but isn't rotated at all |
| 02:37 | <Dashiva> | There's no actual ) glyph, just a copy instruction? |
| 02:38 | <Philip`> | The ')' glyph is a composite glyph, whose single component is the '(' glyph, and supposedly with some rotational transform that seems to be disappearing in my conversion process |
| 02:40 | <Philip`> | I wish these file formats hadn't been designed for 16-bit Windows where you had approximately zero RAM |
| 03:05 | <annevk> | and Adam pwns Roy again |
| 03:05 | <annevk> | now it's really bedtime :) |
| 03:05 | <annevk> | nn |
| 03:08 | <Hixie> | adam is shockingly good at this |
| 03:09 | <Hixie> | Philip`: what characteristic makes them suitable for zero ram? |
| 03:18 | <Philip`> | Hixie: There's things like a ARG_1_AND_2_ARE_WORDS flag which means the next two values in the structure are stored as 2-byte values rather than 1-byte values, and scale transforms can be stored as 1 or 2 or 4 values depending on some other flags |
| 03:18 | <Philip`> | presumably all to save a tiny bit of space |
| 03:19 | Philip` | fixes two more bugs in Font::TTF, and now his bracket works correctly |
| 03:25 | <Hixie> | aah |
| 03:30 | <Philip`> | http://philip.html5.org/demos/font/tests.html |
| 03:31 | <Philip`> | (Firefox definitely wins on quality, as far as I can tell) |
| 03:33 | <Philip`> | (But annoyingly the Essays1743 font has now stopped working on Windows, after I fixed it to preserve the -1 scale on ')') |
| 03:34 | <Philip`> | (Oh, actually, that's not true, I think it's never worked) |
| 03:34 | <wakaba_> | hi Hixie |
| 03:35 | <wakaba_> | about my mail (s/FDDF/FDEF/g (noncharacter code points).) |
| 03:35 | <wakaba_> | In http://www.whatwg.org/specs/web-apps/current-work/multipage/parsing.html#preprocessing-the-input-stream |
| 03:35 | <wakaba_> | U+FDD0 to U+FDDF are parse errors, but U+FDE0 to U+FDEF are also noncharacter |
| 03:36 | <wakaba_> | see http://www.unicode.org/charts/PDF/UFB50.pdf |
| 04:57 | <Hixie> | wow what kind of crazy character is U+FDFD |
| 04:58 | <heycam> | why "ARABIC LIGATURE BISMILLAH AR-RAHMAN AR-RAHEEM" of course :) |
| 04:58 | <Hixie> | have you seen the canonical rendering?! |
| 04:58 | <Hixie> | jesus |
| 04:58 | <Hixie> | wakaba_: thanks |
| 04:59 | heycam | looks at the glyph in the above pdf |
| 04:59 | <heycam> | it must be a phrase or something? |
| 05:00 | <heycam> | seems no crazier than the Zapfino glyph :) |
| 05:00 | <Hixie> | Zapfino isn't one Unicode character! |
| 05:02 | <gavin> | http://en.wikipedia.org/wiki/Basmala |
| 05:02 | <Hixie> | Philip`: crashed webkit :-) |
| 05:03 | <heycam> | ah, interesting |
| 08:59 | <roc> | Philip`: ouch, you crashed ATSUI! |
| 08:59 | <roc> | perhaps the same crash that Hixie saw with Webkit, if he's on Mac |
| 09:42 | <Philip`> | Hixie: roc: That doesn't sound good |
| 09:42 | <Philip`> | I'll blame Apple :-) |
| 09:42 | <roc> | yeah |
| 09:42 | <roc> | we have discovered other crashing bugs in ATSUI |
| 09:42 | <roc> | Apple's response was "use the Core Text APIs instead" |
| 09:42 | <roc> | which aren't supported in 10.4, but oh well |
| 09:43 | <Philip`> | I think these aren't even invalid font files now (except maybe a tiny litle bit) |
| 09:43 | <Philip`> | *little |
| 09:43 | <roc> | it doesn't matter, the bad guys can send invalid font files |
| 09:44 | <Philip`> | It seems to work alright in Safari on Windows, though the combining diacritics completely mess up the actual layout and make everything overlap |
| 09:49 | Philip` | wonders if the bug should be reported or something |
| 09:58 | <roc> | which bug? |
| 09:58 | <roc> | the last ATSUI crasher bug we reported, via our super-duper paid support channel, was rejected with "Use Core Text intead" |
| 10:00 | <jruderman> | roc: so we can't even *pay* apple to fix its security bugs? |
| 10:00 | <roc> | I don't think we presented it as a security bug |
| 10:00 | <jruderman> | oh |
| 10:00 | <jruderman> | well, was it a security bug? |
| 10:00 | <roc> | hard to tell |
| 10:00 | <roc> | no source remember |
| 10:01 | <Philip`> | If it crashes WebKit too, I guess that's something Apple has to fix regardless of what APIs they use internally |
| 10:01 | <roc> | I hope that if we had an exploit, we could force them to fix it |
| 10:01 | <roc> | Philip`: they could switch Webkit to use Core Text |
| 10:02 | <roc> | we have a patch for that, it was just a couple of weeks of work (for someone very experienced and smart, anyway) |
| 10:02 | <roc> | We'll probably have that in FF3.2 for Leopard users |
| 10:03 | <roc> | no doubt Core Text has its own bugs, but hopefully they'll fix those |
| 10:03 | <Philip`> | Or they'll tell you to use the new incompatible text API in 10.6 |
| 10:04 | <jruderman> | Mac OS X 10.4 (Tiger) itself will probably be out of security support by the time we ship firefox 3.2 |
| 10:04 | <jruderman> | Philip`: hehe |
| 10:29 | <Philip`> | Hmm, it's a kind of random crash, which makes it hard to work out what characters are causing it :-( |
| 10:29 | <Philip`> | (but it seems to be the Doulos SIL font) |
| 10:49 | <Philip`> | Ah, looks like it's breaking on the input i\x{0331} |
| 11:11 | <Philip`> | http://philip.html5.org/tests/font/atsui-kern-crash.html |
| 11:12 | <Philip`> | That seems to be about as reliable as I can make it |
| 11:12 | <Philip`> | and it crashes WebKit and Firefox |
| 11:12 | <Philip`> | (in the latest nightlies) |
| 11:15 | <Philip`> | and it crashes Opera too |
| 11:15 | <Philip`> | (all only on OS X) |
| 11:16 | <Philip`> | and I can't see anything I'm doing wrong, so I can't think of any workaround I could implement :-( |
| 12:21 | <Philip`> | (Hmph, OS X doesn't like my GPOS tables) |
| 12:21 | Philip` | wonders when the pain will end |
| 13:32 | <annevk> | I was wondering, for menus, would it be fine to just have <nav> <a ...>...</a> <a ...>...</a> ... </nav> ? |
| 13:32 | <annevk> | would remove a lot of cruft |
| 13:33 | <annevk> | it doesn't really state that lists are supposed to be used in HTML5, though there is an example to that effect and lists have been used traditionally |
| 13:49 | <annevk> | hey cool, the other Ian from Google I met a few times joined the WG |
| 13:53 | <annevk> | http://lists.w3.org/Archives/Public/public-xhtml2/2009Jan/0065.html |
| 13:53 | <annevk> | http://lists.w3.org/Archives/Public/public-xhtml2/2009Jan/0064.html |
| 13:57 | <takkaria> | wow, a new xhtml2 draft too |
| 13:57 | <annevk> | http://www.w3.org/2009/01/21-xhtml-minutes.html is also interesting |
| 13:57 | <annevk> | takkaria, yeah, there are some new features in it since last TR draft it seems, though it's also quite broken and incomplete |
| 13:58 | <annevk> | apparently they replace <script charset> with <script encoding> which takes a comma-separated list so that it's more useful than HTML4?! |
| 13:58 | <annevk> | wtf, the only reason charset is there is to override the encoding of the script (if that's not defined); it's a hack |
| 14:00 | <annevk> | and with all their focus on being backwards compatible they still have things such as "... from a processing point of virew, img and span are identical" |
| 14:00 | <annevk> | and then still deliver it as text/html, how anyone can even consider this plan remotely sane is beyond me |
| 14:01 | <takkaria> | they're saying you can serve as text/html? |
| 14:01 | <hsivonen> | "Steven: If we say that @lang has no meaning, it wouldn't actually break anything" |
| 14:02 | <hsivonen> | I think that illustrates priorities rather aptly |
| 14:02 | <annevk> | yeah, it's hilarious |
| 14:04 | <annevk> | http://lists.w3.org/Archives/Public/public-xhtml2/2009Jan/0073.html |
| 14:05 | <annevk> | http://lists.w3.org/Archives/Public/public-xhtml2/2009Jan/0074.html (about lang and xml:lang) |
| 14:06 | <hsivonen> | if lang is the one that works, getting rid of xml:lang seems to be the sensible thing to do |
| 14:07 | <takkaria> | their priority of constituencies rule is different |
| 14:07 | <takkaria> | consider theoretical purity over specifiers over implementers |
| 14:09 | <annevk> | seems to me they're just buiding a toy language |
| 14:11 | <hsivonen> | looks like they are using http://www.w3.org/1999/xhtml as the namespace for XHTML2 |
| 14:13 | annevk | is reminded of http://annevankesteren.nl/2003/07/xhtml-20-spec-summary :) |
| 14:13 | <annevk> | "but XHTML2.0 has http://www.w3.org/2002/06/xhtml2 and that's very unlikely to change" |
| 14:14 | <annevk> | also http://www.goer.org/Journal/2003/07/index.html#04 "XHTML2 is a big specification (430 KB and counting)" |
| 14:14 | <karlcow> | annevk: http://www.w3.org/MarkUp/tracker/actions/44 |
| 14:14 | <karlcow> | :) |
| 14:15 | karlcow | wonders what would be the content of this |
| 14:17 | <takkaria> | do the XHTML2 people intend to get implementors to implement the spec at some point? |
| 14:18 | <takkaria> | I'm not quite sure what their strategy is there |
| 14:18 | <annevk> | they implement things in three schema languages |
| 14:19 | <annevk> | karlcow, me too... we'll see I suppose; it seems they mostly discuss things in telcons so I suppose it might take a while before they get there |
| 14:21 | <annevk> | takkaria, there are some plug-ins too, I think, e.g. sidewinder or whatever the name was |
| 14:21 | <karlcow> | takkaria: "implementors" is a very wide definition. Yes I'm pretty sure they already have implementors working on it, but maybe they are not the implementers which matter to you, aka different communities. |
| 14:21 | <annevk> | aka not relevant on the Web :p |
| 14:22 | <karlcow> | I didn't say that ;) and do not agree with the hsivonen definition of the Web ;) |
| 14:23 | <annevk> | I think he was misunderstood |
| 14:23 | <karlcow> | takkaria: a sure fact that is XHTML 2 is not implemented in Safari, Opera, Firefox and IE, as far as I know. |
| 14:25 | <karlcow> | ouch bad english as usual. |
| 14:25 | karlcow | reboots his brain serializer |
| 14:27 | <annevk> | karlcow, though now they base it on the XHTML namespace it suddenly "works" |
| 14:28 | <karlcow> | :D |
| 14:29 | <Philip`> | That's the magical power of namespaces! |
| 14:30 | <myakura> | though some incompatible changes has been made throught the spec. like <img> can has now content! |
| 14:30 | <karlcow> | I was not for this change of namespace. It is unfortunate because I think he does more harm to namespaces by diluting their meaning |
| 14:30 | <karlcow> | unfortunately |
| 14:30 | myakura | remembers he's the member of the wg. cough. |
| 14:31 | <annevk> | :) |
| 14:31 | <karlcow> | s/he/it/ |
| 14:32 | karlcow | needs food. It is obvious. |
| 14:32 | <annevk> | I have the feeling that due to the committee process it now tries to be both bold and new and backwards compatible leading to a very poor specification because it accomplishes neither |
| 14:33 | <Philip`> | If there's a problem with two different groups assigning different meanings to the namespace http://www.w3.org/1999/xhtml, clearly we need to add a layer of indirection so you can bind the short name 'http://www.w3.org/1999/xhtml'; to one of several alternative expanded references that identify exactly which version of the namespace you want |
| 14:34 | <takkaria> | karlcow: well, I wonder where it will be implemented if not in web browsers that people use |
| 14:35 | <takkaria> | it amuses be a little bit that in practice XHTML2 will be dependent on HTML5 for text/html parsing |
| 14:35 | <karlcow> | takkaria: there are plenty of spaces for *documents* on the (all types) Web |
| 14:36 | <Philip`> | Perhaps we could define the value of xmlns to be a CURIE, and then use xmlns:http="..." to redefine the meaning of xmlns="http://www.w3.org/1999/xhtml"; |
| 14:38 | <karlcow> | It might be possible that it would not create issues in the end if the communities of usage are clearly separate (wishful thinking) exactly like vocabularies in different communities. cow in argentina and cow in india don't have the same meaning ;) |
| 14:38 | <karlcow> | which leads us to my initial comment: I need food for my soul :p now. |
| 14:38 | <karlcow> | bbl |
| 14:41 | <takkaria> | well, documents using xhtml2 are all very well and good but if they only rely on features that are in previous versions of xhtml then I guess I don't know what the point is |
| 15:33 | <annevk> | heh |
| 15:33 | <annevk> | mark birbeck removed some tweets |
| 15:33 | <annevk> | compare http://search.twitter.com/search?q=html5 and http://twitter.com/markbirbeck |
| 15:34 | <annevk> | http://friendfeed.com/e/d969a84b-621d-6eda-e009-ffd2d34c8c87/Yet-another-interesting-blog-post-comments-about/ |
| 15:35 | <annevk> | unfortunately it's not interesting at all |
| 15:35 | <annevk> | so I guess that's why he removed it |
| 15:41 | <gsnedders> | http://search.twitter.com/search?q=html5+OR+%22html+5%22 is a more useful search |
| 15:42 | <annevk> | i use separate tabs for them |
| 15:48 | gsnedders | wishes you could follow searches on Twitter :P |
| 16:27 | <karlcow> | gsnedders: you can follow searches. There is an rss feed for them or maybe I have not understood the question |
| 16:27 | <gsnedders> | karlcow: I mean as if they were a user on twitter |
| 16:36 | <annevk> | http://www.webscienceman.com/2009/01/24/html-xhtml-html5-future-html/ |
| 16:40 | <hsivonen> | annevk: seems a bit exaggerated to compare HTML5 to the Massacre of the Innocents |
| 16:43 | <annevk> | yeah... |
| 16:43 | <hsivonen> | is there any test case data on how local(...) matches on font names in different environments? |
| 16:44 | <hsivonen> | should I say local("Linux Libertine Italic") or also local("LinLibertineI")? |
| 16:47 | <annevk> | jdaggett would know |
| 16:49 | <Philip`> | I wish I had some way to debug why my GPOS table seems to be entirely ignored on OS X |
| 16:51 | <hsivonen> | It seems there are three plausible ways to identify a font: |
| 16:51 | <hsivonen> | 1) Family, space, Style |
| 16:51 | <hsivonen> | 2) Full name |
| 16:51 | <hsivonen> | 3) PostScript name |
| 16:52 | <hsivonen> | ah, it's in the spec although I didn't see it when I was scanning it the first time |
| 16:52 | <hsivonen> | "For TrueType and OpenType fonts, the full font name as defined in the font name table is used to reference a given face. Additionally, for TrueType and OpenType fonts user agents may optionally support Postscript name lookup on platforms where that is appropriate." |
| 16:53 | <hsivonen> | although it seems that all of Firefox, Safari and Opera on Mac are sensitive to the Family name field when it comes to traditional non-@font-face use... |
| 17:01 | <hsivonen> | It seems I can't do Web authoring without getting myself into a situation that calls for test cases :-/ |
| 17:07 | <hsivonen> | http://hsivonen.iki.fi/test/font-local.html |
| 17:07 | <hsivonen> | wow. totally surprising results |
| 17:10 | <hsivonen> | so far, it seems that using the postscript name works best and using the full name is harmful in Opera |
| 17:11 | <hsivonen> | this is on Mac |
| 17:24 | <Philip`> | Hmm, maybe OS X is automatically translating "i{combining grave accent}" into "{i with grave accent}" without even looking at the font's substitution tables |
| 17:24 | <Philip`> | Is there some kind of Unicode normalisation that does that? |
| 17:25 | <gsnedders> | NFD |
| 17:25 | <gsnedders> | Oh, wait |
| 17:25 | <gsnedders> | NFC |
| 17:25 | <gsnedders> | NFD does the opposite |
| 17:26 | <gsnedders> | s/opposite/inverse/ |
| 17:26 | <hsivonen> | eww. MgOpen Cosmetica looks awful on Windows. Hinting gone badly wrong. |
| 17:28 | <Philip`> | So... I have a list of characters (by which I mean codepoints) that the user says they want. I need to work out which glyphs are required for those characters. If OS X is seemingly NFCing the input text and therefore the characters it tries to render are not the characters the user asked for, then that's a bit annoying :-( |
| 17:29 | <Philip`> | I guess I need a copy of the normalisation tables, and if some string "xy" normalises to "z" and the user asked for "x" and "y" then I'll have to include "z" |
| 17:29 | <hsivonen> | if I have multiple styles of a @font-face font within a family, Opera 10 uses the last @font-face per family for all styles |
| 17:29 | <hsivonen> | which sucks pretty badly |
| 17:29 | Philip` | wonders if that is a sane thing to do |
| 17:46 | <gsnedders> | Wow, the latest mr last week post is really not funny |
| 17:47 | <Philip`> | s/the latest (.*) post/$1/ ? |
| 17:47 | <gsnedders> | :) |
| 17:48 | <Dashiva> | Maybe he's trying to make a statement related to "I liked X better when it was underground." |
| 18:28 | <Philip`> | Hooray, now i-grave works on OS X too |
| 18:29 | <Philip`> | The only thing that doesn't quite work is Arabic in DejaVu Sans on OS X, and I don't care enough to fix it now |
| 18:35 | <gsnedders> | Is there any reason why <ul><ins><li></ins></ul> isn't conforming? |
| 20:31 | <Philip`> | IE appears to have a limit of 32 embedded fonts per page |
| 20:32 | <takkaria> | oh noes! |
| 20:32 | <Philip`> | It's not very good when you're trying to make a page that demonstrates lots of different fonts :-p |
| 20:39 | <jruderman> | i wonder how they chose they limit |
| 20:40 | <Philip`> | I guess it might actually be a font limit, not an embedded font limit |
| 20:41 | <Philip`> | and it might not be precisely 32, but it looked close for me |
| 20:41 | <Philip`> | s/close/close enough/ |
| 20:42 | <Philip`> | Fortunately I can't find 32 free (modifiable, distributable) fonts that have TrueType outlines and work cross-platform and don't like incredibly rubbish, so the limit isn't much of a problem |
| 21:26 | <Hixie> | so can anyone think of a DOM API other than XHR, the Location object, and various APIs that map to element attributes (like HTMLAnchorElement.href) that allows the author to give a relative URL that then needs to be resolved? |
| 21:27 | <Hixie> | addEventSource() |
| 21:28 | <Hixie> | the data grid getRowImage() api |
| 21:28 | <Hixie> | window.open() |
| 21:29 | <Hixie> | registerProtocolHandler and registerContentHandler |
| 21:31 | <Hixie> | pushState() |
| 21:31 | <Hixie> | resolveURL() |
| 21:32 | <Hixie> | removeEventSource(), duh |
| 21:33 | <Hixie> | postMessage() |
| 21:33 | <Hixie> | wow, surprisingly many of these are new |
| 21:34 | <jruderman> | setting style properties, if that counts |
| 21:35 | <Hixie> | i'm really looking for things that depend on the script to get the base uri, and i think in that case the style sheet provides the base uri |
| 21:38 | <jruderman> | document.load |
| 23:18 | <Hixie> | jruderman: what's document.load()? |
| 23:20 | <jruderman> | https://developer.mozilla.org/en/DOM/document.load i guess |
| 23:27 | <Hixie> | oh, dom3 l&s should be ignored :-) |
| 23:33 | <hendry> | http://static.guim.co.uk/sys-images/Guardian/Pix/pixies/2008/04/28/crops_question_pixie.jpg |
| 23:44 | <jruderman> | Hixie: i noticed that html5 defines a few quirks mode behaviors. are you going to try to define most/all of them? |
| 23:44 | <jruderman> | https://developer.mozilla.org/en/Mozilla_Quirks_Mode_Behavior |
| 23:44 | <jruderman> | https://bugzilla.mozilla.org/show_bug.cgi?id=475191 made me wonder |
| 23:47 | <jruderman> | oh "Some obsolete legacy attributes parse colors in a more complicated manner." |
| 23:48 | <Hixie> | the idea is for the spec to be a complete description of how to implement a user agent (web browser, command line tool, search engine, whatever) in a manner that is compatible with existing content |
| 23:48 | <karlcow> | jruderman: there was partial implementation of DOM 3 Load and Save in Firefox, opera betas and Konqueror in… 2005 at least if I remember correctly |
| 23:48 | <Hixie> | so that no implementor ever feels the need to do anything not in the spec for compatibility |
| 23:49 | <karlcow> | the products implementing it were Xerces, X-Hive, pxdom |
| 23:50 | <jruderman> | Hixie: so a quirk shared by all modern browsers is likely to get included |
| 23:50 | <karlcow> | http://www.w3.org/2003/10/DOM-Level-3-LS-implementations.html |
| 23:52 | <Philip`> | jruderman: I believe zcorpan looked at quirks-mode colour parsing a while ago |
| 23:53 | <Philip`> | jruderman: http://simon.html5.org/test/html/rendering/color-attributes/ and http://simon.html5.org/specs/html-color-attributes might be relevant |
| 23:55 | <Hixie> | jruderman: yeah. and hopefully the browsers will all converge on one set of (standardised) quirks |
| 23:58 | <jruderman> | Philip`: nice, thanks |
