| 08:05 | <zcorpan> | https://bugs.webkit.org/show_bug.cgi?id=76811 |
| 08:08 | <hsivonen> | https://twitter.com/#!/simukis/status/161097038719762432 |
| 08:09 | <zcorpan> | I accidentally h.264 |
| 08:14 | <MikeSmith> | "It just happens, that about every computer has H.264 libraries and they get linked accidentally" |
| 08:49 | <abarth> | Ms2ger: pretty harsh words for ap :( |
| 08:50 | <Ms2ger> | Deserved ones, IMO |
| 08:50 | <Ms2ger> | He keeps hurting the web for webkit's sake |
| 08:50 | <abarth> | what's the path forward here? we remove URL? |
| 08:50 | <Ms2ger> | Yeah |
| 08:50 | <abarth> | ok |
| 08:51 | <abarth> | seems reasonable, but I'm sorry I don't have the energy to argue with AP on this topic |
| 08:51 | <abarth> | too busy arguing with him about Content-Disposition |
| 08:52 | <Ms2ger> | Oh, I have lots of energy to argue ;) |
| 08:52 | <Ms2ger> | Now, to fight dom2html |
| 09:03 | <hsivonen> | speaking of WebKit and hurting the Web, did Apple fork WebKit for iBooks 2? |
| 09:03 | <hsivonen> | will all WebKit devs have to deal with -ibooks stuff living on WebKit trunk? |
| 09:04 | <hsivonen> | is there already a source drop of iBooks-modified WebKit? |
| 09:09 | <jgraham> | heycam: So I still don't understand how one is supposed to algorithmically determine what is a platform object and what is not, since the definition seems to be based on intent. |
| 09:10 | <jgraham> | Is it intended that platform objects have some hidden state that enables one to determine that they are platform objects? |
| 09:11 | <jgraham> | If so is it intended that e.g. a third party javascript library can never implement an IDL interface (since the browser wouldn't know how to check for the hidden flag) |
| 09:12 | <Ms2ger> | "An ECMAScript library that adds support for some Web API that is not natively supported by a browser would also be considered to expose platform objects." |
| 09:12 | <heycam> | jgraham, there's a restriction now that platform objects (ones that aren't for exceptions, or platform array objects) can only implement "normal" interfaces and user objects can only implement callback interfaces |
| 09:12 | <heycam> | jgraham, but yes because we aren't doing an explicit check on [[Class]] or whatever, there is some hidden state in there |
| 09:13 | <heycam> | jgraham, how to determine whether a given object is considered to implement a given interface isn't defined exactly |
| 09:13 | <jgraham> | heycam: Right I think that is my point. If a platform object is one that implements an interface and you can't tell what implements an interface you can't make platform objects |
| 09:13 | <heycam> | jgraham, but there is a requirement that say an object returned from `readonly attribute Node? parentNode;` is a platform object that implements Node |
| 09:14 | <heycam> | jgraham, just because platform objects are the only things that can implement (non-callback) interfaces |
| 09:15 | <heycam> | jgraham, tbh I was a bit lazy, and I think the concept of platform vs user objects could probably be eliminated, since they are always distinct anyway (user objects can't implement Node, platform objects can't implement EventListener) |
| 09:15 | <jgraham> | I am at least pretty sure than in the current state of the spec, the browser isPlatformObject check will basically amount to "is this implemented natively in the browser" |
| 09:16 | <Ms2ger> | Sounds like the right thing to do |
| 09:17 | <jgraham> | IIRC there is at least one situation where you need to take an arbitary object and determine if it is a platform object or not (overloading resolution with arrays, perhaps?) |
| 09:17 | <heycam> | "natively in the browser" is not exactly right; it's more like "has some hidden state variable set that only the given implementation would know" |
| 09:17 | <heycam> | jgraham, there you need to know whether it is a native Array object (easy to check), a platform array object, or a platform object that also supports indexed properties |
| 09:18 | <jgraham> | Either that is an equivalent statement or we will be in a bad state where some magic flags will have an effect on some browsers but not others |
| 09:18 | <heycam> | ok |
| 09:18 | <heycam> | there's probably a problem |
| 09:18 | <heycam> | let's say you have the native browser C++ implementation of some APIs |
| 09:18 | <heycam> | as well as a JS library that implements some other APIs |
| 09:18 | <heycam> | the C++ implementation has an operation that takes a sequence |
| 09:19 | <heycam> | the JS library vends an object that supports indexed properties |
| 09:19 | <heycam> | how is the C++ implementation to know that the JS library objects "implements" the IDL interface with indexed properties |
| 09:19 | <heycam> | or even what that other IDL interface is |
| 09:20 | <abarth> | hsivonen: i haven't seen any -ibook stuff in WebKit, but I don't venture into CSS land much |
| 09:20 | <heycam> | I don't think it's possible then for a native C++ implementation and a JS library implementation of APIs to coexist |
| 09:20 | <heycam> | at least if they don't share knowledge about these internal state bits |
| 09:22 | <heycam> | I should probably remove mention of JS libraries implementing APIs, and talk more generally about it not mattering what languages (JS or C++) implements given interfaces, just that they need to share this internal state information |
| 09:22 | <heycam> | jgraham, wdyt? |
| 09:22 | <abarth> | i don't see anything about ibooks on http://opensource.apple.com/ |
| 09:24 | <jgraham> | heycam: I am a bit skeptical about something that would allow random on-the-web javascript libraries to set some flags that have magic behaviour in some implementations. Of course if some browser wants to implement their DOM APIs in javascript or a javascript/C++ mix, that should be possible |
| 09:25 | <heycam> | jgraham, right so I think that's all I want to enable |
| 09:25 | <heycam> | jgraham, i.e. avoid "host object" vs "native object" terminology |
| 09:25 | <jgraham> | OK, so as long as it is clearly scoped to that, IO think it is fine |
| 09:25 | <heycam> | jgraham, so I should forget about random JS libraries |
| 09:25 | <jgraham> | s/IO/I/ |
| 09:25 | heycam | will have a go at writing that tomorrow |
| 09:26 | <jgraham> | Great, thanks |
| 09:32 | <hsivonen> | abarth: the iBooks stuff is based on WebKit, though, right? It would be really weird for Apple to implement the common stuff a second time. |
| 09:32 | <abarth> | dunno |
| 09:32 | <abarth> | i bet they have an internal fork |
| 09:32 | <jgraham> | Pretty sure that at least some older Apple ePub reader was basically WebKit |
| 09:32 | <abarth> | it would be tough to land that proprietary stuff in WebKit |
| 09:33 | <heycam> | Ms2ger, "I propose that we add a pointer to the contemporary specification to the following specifications: […] DOM 2 Events (D3E) " |
| 09:33 | <heycam> | Ms2ger, slip of the tongue? :) |
| 09:34 | <heycam> | Ms2ger, or is the bracketed spec what the old spec should be pointing to (not the shortname of the obsolete spec) |
| 09:34 | <heycam> | Ms2ger, (if so, nm me!) |
| 09:34 | <hsivonen> | abarth: it would be cool to see the Apple WebKit team oppose to landing that stuff as vigorously as they opposed to landing the Dart stuff. :-) |
| 09:35 | <abarth> | there's certainly some tension in the webkit community around these sorts of issues |
| 09:36 | <abarth> | the project is wrestling with how to be innovative without going in every direction at once |
| 09:38 | <abarth> | (i should also say that I don't know much specifically about ibooks except what i read in articles like http://www.zdnet.com/blog/bott/how-apple-is-sabotaging-an-open-standard-for-digital-books/4378 ) |
| 09:38 | <heycam> | tbh I don't really see a problem with iBooks doing this, if it doesn't impact WebKit more generally -- if Apple wishes to maintain an internal fork for their experimental syntax, that's up to them |
| 09:38 | <heycam> | it might even provide good feedback to the CSS WG for the standards based features |
| 09:39 | <heycam> | (this is wrt to web features btw; I have no idea whether their -ibook-* features are detrimental to EPUB standardisation efforts) |
| 09:40 | <abarth> | heycam: sure, but presumably the LGPL would compel them to release the source |
| 09:41 | <heycam> | abarth, sure, but I wouldn't expect that source to be used as a web browser engine for other products |
| 09:48 | <asmodai> | Must say I like Opera Mobile the most on my Android devices. |
| 09:48 | <asmodai> | Especially great how it reopens all my tabs after it got pushed out of active memory. Most other browsers give you a clean slate when you start the browser again. |
| 09:48 | <hsivonen> | heycam: this sort of thing could be detrimental to writing Gecko, Presto or Trident-based ebook readers. |
| 09:49 | <hsivonen> | heycam: and if one believes that everything is moving to the Web, ebook readers will be Web apps, too |
| 09:49 | <heycam> | hsivonen, that is true, and I guess I was just thinking about this from a web browser engine PoV rather than eBook |
| 09:50 | <hsivonen> | asmodai: Fennec Native reopens the tabs when reopening after being killed by the OS. |
| 09:50 | <hsivonen> | asmodai: (though there's a bug that makes the frontmost tab reopen twice.) |
| 09:50 | <asmodai> | hsivonen: I found Fennec to be not so good :( |
| 09:50 | <heycam> | hsivonen, who is to say Apple (or other WebKit contributors) are not willing to implement eventually unprefixed, standards-cooked versions of features they implemented with -ibook-* prefixes |
| 09:50 | <asmodai> | I tried normal, beta, aurora, and nightly. |
| 09:50 | <hsivonen> | asmodai: XUL or Native? |
| 09:50 | <asmodai> | Mmm, good question |
| 09:50 | <hsivonen> | asmodai: when did you try Nightly? |
| 09:51 | <asmodai> | about 4-5 days ago or so |
| 09:51 | <hsivonen> | heycam: to a large extent, the position of MS Office is based on support for .doc, not on support for ISO OOXML |
| 09:51 | heycam | sucks because he complains (mostly in his head, sometimes in public forums) about Firefox Native vs Firefox XUL |
| 09:51 | <asmodai> | hsivonen: downloaded the apk from a mozilla page |
| 09:51 | heycam | should help improve Firefox Native rather than complain, of course |
| 09:52 | <hsivonen> | asmodai: recently? |
| 09:52 | <hsivonen> | oh. 4-5 day |
| 09:52 | <asmodai> | hsivonen: 4-5 days ago |
| 09:52 | <hsivonen> | asmodai: that's Fennec Native |
| 09:52 | <hsivonen> | asmodai: please file bugs about stuff that sucks |
| 09:52 | <asmodai> | heycam: I gave up on that, I got way too many projects. |
| 09:52 | <asmodai> | hsivonen: Will do |
| 09:52 | <asmodai> | hsivonen: It definitely performs faster though |
| 09:52 | <asmodai> | hsivonen: Than previous versions |
| 09:53 | <heycam> | I with my Galaxy S2 didn't find a huge performance difference |
| 09:53 | <asmodai> | heycam: on my Galaxy Tab 10.1 it was at least |
| 09:53 | <asmodai> | Didn't test on my S2 |
| 09:53 | <heycam> | and there definitely are many more broken things for me in Native UI vs the old XUL UI |
| 09:53 | <heycam> | but! I have started to file or CC myself to bugs, so that's better than nothing :) |
| 09:54 | heycam | only updated last week once the intiial Sync implementation had landed |
| 09:54 | <hsivonen> | I still use Fennec XUL for sites that require login |
| 09:54 | <heycam> | hsivonen, forms don't work for you? |
| 09:54 | <heycam> | hsivonen, or is this HTTP auth? |
| 09:54 | <hsivonen> | heycam: credential sync and/or autofill don't work for me |
| 09:55 | <heycam> | hsivonen, oh. autofill works for me on the ~1 site that I view on the phone that needs a login. |
| 09:55 | <hsivonen> | I'm not planning on re-entering the credentials I've entered on desktop |
| 09:55 | <asmodai> | One thing that's ticking me off with some recent browsers is when I copy URLs it's not giving me the scheme |
| 09:56 | <hsivonen> | asmodai: yeah, that's annoying with Opera Mobile |
| 09:56 | <hsivonen> | in Fennec Native, I can't copy the URL at all, though :-) |
| 09:56 | <heycam> | asmodai, occasionally that happens with me in Firefox, but most of the time it does the right thing |
| 09:56 | <hsivonen> | asmodai: desktop Firefox has more and more advanced heuristics for this |
| 09:56 | <heycam> | hsivonen, currently nightlies having broken long tap makes them somewhat unusable for me :) |
| 09:57 | <hsivonen> | asmodai: but with desktop Firefox, you can avoid the problem by disabling the scheme hiding |
| 09:57 | <asmodai> | Sucks when I am making notes in Evernote and need to add complete URLs |
| 09:57 | <hsivonen> | I disabled the scheme hiding in Firefox on desktop |
| 09:57 | <asmodai> | Will check that. It's a bit annoying when you need to provide them to applications that do some sort of regexp matching on URLs and thus fail :) |
| 09:58 | heycam | goes to cook dinner and watch Hewitt v Djokovic on the telly |
| 09:58 | <hsivonen> | that Opera Mobile prefills "www." is also terribly annoying when trying *paste into* Opera Mobile |
| 09:59 | <asmodai> | especially since www. is hardly common anymore |
| 10:00 | <asmodai> | so many sites droped it or use some other subdomain/host |
| 10:03 | <hsivonen> | Hixie: why aren't CR and CRLF treated as line breaks in title tooltips? |
| 10:08 | <hsivonen> | huh. how can the spec be wrong about CR and CRLF in tooltips and I don't find a bug report filed by me about it? |
| 10:08 | <hsivonen> | http://hsivonen.iki.fi/test/moz/title-attribute.html I have an existing test case |
| 10:09 | <zcorpan> | hsivonen: maybe you traveled back in time. in the future, you will file a bug about it |
| 10:10 | <jgraham> | I remember there was already some discussion about this |
| 10:11 | <jgraham> | tbh, I thought it was hsivonen who requested the current behviour |
| 10:11 | <hsivonen> | really? |
| 10:11 | <hsivonen> | seems unlikely since I have a pending Firefox patch that expects CR and CRLF to be breaks |
| 10:13 | <jgraham> | Well I'm not saying I am right |
| 10:14 | <jgraham> | That's just what I thought |
| 11:52 | <oal> | Does hsla colors support half percent values, like 20.5%, or only whole numbers? |
| 11:58 | <Philip`> | oal: http://www.w3.org/TR/css3-color/#hsla-color says some components are "a percentage" which presumably refers to http://www.w3.org/TR/CSS2/syndata.html#percentage-units which refers to http://www.w3.org/TR/CSS2/syndata.html#value-def-number which allows fractional values |
| 11:58 | <Ms2ger> | As usual, don't expect too much of CSS specs |
| 11:59 | <zcorpan> | short answer is "Yes" |
| 11:59 | <jgraham> | You are in a mazy of twisty specs, all alike |
| 12:00 | <Ms2ger> | You are likely to be eaten by glazou |
| 12:01 | <zcorpan> | glazou is the W3C grue? |
| 12:01 | <Ms2ger> | Just The WG's |
| 12:01 | <zcorpan> | ah |
| 12:11 | <oal> | Hmm, thanks, Philip` |
| 12:12 | <oal> | Just noticed google's closure library's regular expressions only match whole numbers |
| 12:13 | <Philip`> | Sounds like a bug |
| 12:16 | Philip` | presumes it should be something kind of like hsla\(\s*[-+]?([0-9]*\.)?[0-9]+\s*,\s*[-+]?([0-9]*\.)?[0-9]+%\s*,\s*[-+]?([0-9]*\.)?[0-9]+%\s*,\s*[-+]?([0-9]*\.)?[0-9]+\s*\) |
| 12:17 | <oal> | http://code.google.com/p/closure-library/source/browse/trunk/closure/goog/color/alpha.js#367 |
| 12:18 | <Philip`> | That's totally bogus since e.g. "0000" is a synactically valid number |
| 12:18 | <bga> | Philip` \d ! |
| 12:18 | <Philip`> | plus they don't allow the right whitespace |
| 12:18 | <bga> | its shorter |
| 12:18 | <Philip`> | and they do allow "HSLA(...)" which I think they shouldn't |
| 12:19 | <Philip`> | Oh, actually I suppose it looks like they should |
| 12:46 | <hsivonen> | awesome. Gecko's HTML to plain text conversion treats <multicol> as a block break |
| 12:50 | <Ms2ger> | hsivonen, ever noticed http://mxr.mozilla.org/mozilla-central/source/parser/htmlparser/src/nsHTMLTags.cpp#206 ? :) |
| 12:51 | <hsivonen> | Ms2ger: no. I'm looking at nsElementTable.cpp |
| 12:54 | <Ms2ger> | Heh, I never realized we had two nsHTMLElement classes |
| 12:55 | <hsivonen> | <listing>, OTOH, is something the spec knows about |
| 12:57 | <hsivonen> | curiously, <dd> and <dt> aren't blocks as far as old Gecko is concerned |
| 16:35 | <SamDutton> | Just noticed minor typo in http://www.whatwg.org/specs/web-apps/current-work/multipage/the-video-element.html#text-track-api |
| 16:36 | <SamDutton> | var sfx = new Audio('sfx.wav'); |
| 16:36 | <SamDutton> | var sounds = a.addTextTrack('metadata'); |
| 16:36 | <Ms2ger> | SamDutton, file a bug |
| 16:37 | <SamDutton> | thanks — excuse my ignorance, but where is the right place to file a bug for typos in the spec? |
| 16:37 | <jgraham> | Little grey box at the botoom right of the spec |
| 16:37 | <jgraham> | *bottom |
| 16:38 | <SamDutton> | ah — got it — thanks |
| 16:51 | Ms2ger | wonders why someone would abbreviate "deterministic" as "determinist." |
| 16:53 | <Philip`> | Maybe "d11c" sounds a bit too rude |
| 17:25 | <AryehGregor> | This seems to render very incorrectly in WebKit relative to Gecko: data:text/html,<!doctype html><div style="width:200px;height:200px;background:gray;-webkit-transform:perspective(200px) rotateX(45deg);-moz-transform:perspective(200px) rotateX(45deg)"> |
| 17:26 | AryehGregor | isn't sure why |
| 17:28 | AryehGregor | is quite sure Gecko is right here, not WebKit |
| 17:28 | <Ms2ger> | Gecko is always right ;) |
| 17:28 | <AryehGregor> | I filed like half a dozen bugs against Gecko in its transforms alone in the last few weeks, so I think that's not quite the case. :) |
| 17:50 | <dglazkov> | good morning, Whatwg! |
| 18:41 | <AryehGregor> | The lack of rebase in hg means I have to make sure to pull before committing to avoid pointless merge commits. |
| 18:41 | <AryehGregor> | Whereas in git, I can commit, then if there were other commits, rebase after pulling. |
| 18:41 | <AryehGregor> | Tiresome. |
| 18:42 | <AryehGregor> | hg wants you to not commit. |
| 18:42 | <AryehGregor> | It wants you to use a whole separate patch-management system instead of just having them be the same. |
| 18:42 | AryehGregor | grumbles |
| 18:42 | <AryehGregor> | (of course, I can always use hg rebase, I guess, but it's eaten my data before, so I'd prefer not) |
| 18:43 | <AryehGregor> | (rebase in git is safe: you can always easily recover any commits that get overwritten) |
| 18:43 | <AryehGregor> | (in fact, pretty much everything in git is safe except git gc, barring bugs) |
| 18:43 | <AryehGregor> | (which I've admittedly run into, but at least it's not by design, and upgrading fixed it) |
| 18:43 | <AryehGregor> | (. . . fortunately, without much data loss, since I had clones) |
| 18:51 | <bga> | https://dvcs.w3.org/hg/audio/raw-file/tip/streams/StreamProcessing.html |
| 18:51 | <MikeSmith> | AryehGregor: there is a rebase extension for hg |
| 18:51 | <AryehGregor> | MikeSmith, IIRC, it ate my data when I rebased and changed my mind. |
| 18:51 | <MikeSmith> | I have used it and it seems to pretty well work |
| 18:52 | <MikeSmith> | oh shit |
| 18:52 | <MikeSmith> | damn |
| 18:52 | <AryehGregor> | I might be misremembering. |
| 18:52 | <MikeSmith> | hmm |
| 18:52 | <AryehGregor> | In git, if you rebase, the old commit still sticks around for a while. |
| 18:52 | <MikeSmith> | I see |
| 18:52 | <AryehGregor> | As long as you can find the commit id, you're fine. |
| 18:52 | <AryehGregor> | Which you usually can. |
| 18:52 | <AryehGregor> | IIRC, the hg rebase extension actually destroys the old commit. |
| 18:52 | AryehGregor | double-checks |
| 18:52 | <MikeSmith> | I think it might yeah |
| 18:53 | <MikeSmith> | but the times I have tried it worked the way I wanted it to |
| 18:53 | <MikeSmith> | in that, I could rebase after committing to set things straight |
| 18:53 | <MikeSmith> | a |
| 18:53 | <AryehGregor> | When I used it, it sent me to vim to merge something or other, and I had no idea what it wanted me to do, so I did ":q!". |
| 18:53 | <MikeSmith> | and I could then push without multiple-head problem |
| 18:53 | <MikeSmith> | ah |
| 18:53 | <AryehGregor> | Apparently, it interpreted this as "resolve the conflict by discarding all changes" or something. |
| 18:53 | <AryehGregor> | And my work was permanently lost. |
| 18:54 | <MikeSmith> | oh geez |
| 18:54 | <MikeSmith> | yeah, that ain't good |
| 18:54 | AryehGregor | might be misremembering, but thinks it was something like that |
| 18:55 | <MikeSmith> | it sounds not totally unlikely |
| 18:55 | <MikeSmith> | I have never had it prompt me to do a merge on a rebase but that could just be because I've never had any conflicts to deal with |
| 18:55 | <AryehGregor> | git treats commits as objects that are kept lying around and eventually garbage-collected if there are no references to them. |
| 18:55 | <MikeSmith> | I see |
| 18:55 | <AryehGregor> | hg treats them as permanent records of some kind, kept in an append-only database, or something like that. I think. |
| 18:56 | <AryehGregor> | So I guess it's not natural for hg to keep around no-longer-used commits. |
| 18:56 | <AryehGregor> | In git it's very natural; everything is an object identified by a hex id, and objects are never destroyed except by garbage collection, as a rule. |
| 18:56 | <AryehGregor> | Anyway, back to work. |
| 18:58 | <dglazkov> | I like useless hg commits. They're great reminders that we are humans and thus have imperfections. |
| 18:58 | <dglazkov> | http://dvcs.w3.org/hg/webcomponents/rev/9c8f2d89a85e |
| 18:59 | <MikeSmith> | I just like for the commit history to remain purty |
| 19:00 | <MikeSmith> | without all kinds of unnecessary branches and merges |
| 19:00 | <dglazkov> | :) |
| 19:04 | AryehGregor | also misses git add -i |
| 19:12 | <MikeSmith> | what does add -i do? |
| 19:13 | <MikeSmith> | btw, as much as you dudes hate XSLT, you probably don't want to know that all the git help pages are generated using some DocBook-to-groff XSLT code that I wrote |
| 19:14 | <Ms2ger> | MikeSmith, hah, seriously? |
| 19:14 | <MikeSmith> | yup |
| 19:14 | <MikeSmith> | try a "vim /opt/local/share/man/man1/git-add.1.gz" or equivalent some time |
| 19:15 | <MikeSmith> | the hg man pages use the same code |
| 19:15 | <MikeSmith> | as do the man pages for many Debian utils |
| 19:15 | <zewt> | i vaguely remember using xslt for some college mess a decade ago |
| 19:15 | <MikeSmith> | e.g., the ones apt-get and friends |
| 19:15 | <MikeSmith> | xslt is a horror |
| 19:16 | <MikeSmith> | groff is a horror too |
| 19:16 | <Philip`> | Sounds like they're a good fit for Git |
| 19:16 | <MikeSmith> | touche |
| 19:18 | <zewt> | that's why i like svn--it's actually not a horror |
| 19:19 | <zewt> | that's pretty much what it takes to create something which isn't a horror: deploy a horror for a decade or so (cvs), then rewrite it from scratch with what you learned |
| 19:20 | <MikeSmith> | svn is worse than cvs |
| 19:20 | <zewt> | uh |
| 19:20 | <zewt> | no. heh |
| 19:20 | <zewt> | svn is better than cvs in every way except one (purging old data) |
| 19:20 | <zewt> | (which seems to be a result of them going "nobody could ever possibly want to do that!", which is nonsense, and designing a system which is incapable of it) |
| 19:21 | <MikeSmith> | if I could drop an atomic nuclear H-bomb on svn and completely obliterate it from the face of the earth forever, I would stop whatever I'm doing right now and do that instead |
| 19:22 | <MikeSmith> | but that's just me |
| 19:22 | <zewt> | i'd be happy to do that, to cvs :) |
| 19:25 | <Ms2ger> | I'd do that to svn, cvs, git and bzr |
| 19:28 | <MikeSmith> | I think President Obomba or somebody already did a predator-drone surgical strike on bzr |
| 19:28 | <MikeSmith> | you mention of bzr is the first I remember hearing of it for a long time |
| 19:28 | <Philip`> | zewt: What's wrong with serialising your entire repository and passing it through svndumpfilter and making everybody throw out their working copies and check it out again? |
| 19:29 | <Ms2ger> | MikeSmith, bugzilla uses it :/ |
| 19:31 | <MikeSmith> | Ms2ger: thanks for one more data point about what makes bugzilla a suboptimal bug-tracking system! |
| 19:31 | <Ms2ger> | Hah |
| 19:31 | <MikeSmith> | and I thought my expectations could be lowered no further |
| 19:31 | <Ms2ger> | Well, do you have better alternatives? |
| 19:36 | <MikeSmith> | no |
| 19:36 | <MikeSmith> | all open-source bug-tracking systems suck hard pretty much equally |
| 19:38 | <MikeSmith> | when I worked at Openwave formerly phone.com formerly software.com we there had a bug-tracking system that some engineers had developed in-house and that was absolutely brilliant |
| 19:39 | <MikeSmith> | I tried to convince some product-management weenies to make it into an actual product that the company could sell |
| 19:39 | <MikeSmith> | but they were not interested |
| 19:39 | <MikeSmith> | because they only cared about stuff that had potential to bring in gazillions of dollars |
| 19:40 | <Ms2ger> | ... and they hired you? ;) |
| 19:41 | <MikeSmith> | I was working on an e-mail server product than and the CEO then actually said stuff like "e-mail is a zero-billion dollar market" as a justification for not investing in that product |
| 19:42 | <MikeSmith> | Ms2ger: they hired my originally just as a technical writer to work on docs |
| 19:42 | <MikeSmith> | so I was pretty much like, say, a lamp |
| 19:42 | <MikeSmith> | or a lampshade |
| 19:42 | <Ms2ger> | Linux, Apa... Doesn't make sense, does it? |
| 19:43 | <MikeSmith> | whatever random piece of furniture |
| 19:43 | <Ms2ger> | Oh |
| 19:43 | AryehGregor | cannot get -webkit-perspective to work for some reason, grr. |
| 19:43 | <AryehGregor> | Indeed, this demo doesn't work for me in Chrome 17 dev: http://www.webkit.org/blog-files/3d-transforms/perspective-by-example.html |
| 19:43 | <AryehGregor> | Can anyone reproduce? |
| 19:45 | <AryehGregor> | Hmm: http://code.google.com/p/chromium/issues/detail?id=54702 |
| 19:46 | <AryehGregor> | It seems like Safari and Chrome treat transforms differently. |
| 19:46 | <AryehGregor> | Well, I can run Safari in my Windows VM just like IE, I guess . . . |
| 19:47 | <MikeSmith> | Ms2ger: jackass CEO would visit our dev office sometimes and walk into the break room and randomly quiz us with questions like, What are the top three strategic priorities of our company? (Which I responded to by saying, "Make money?" and he said "NO!" and I said "OK, reduce costs?" and again he said NO and then told us what the priorities really were and none of us had any clue what he was describing despite hearing sounds coming from h |
| 19:47 | <MikeSmith> | mouth.) |
| 19:47 | <MikeSmith> | AryehGregor: that demo works for me on OSX |
| 19:47 | <AryehGregor> | MikeSmith, in Chrome or Safari? |
| 19:47 | <MikeSmith> | Chrome |
| 19:48 | <MikeSmith> | 17.0.963.12 dev |
| 19:48 | <MikeSmith> | which is a little out of date |
| 19:48 | <AryehGregor> | Huh. |
| 19:48 | AryehGregor | tests in Chrome on Windows |
| 19:48 | <MikeSmith> | lemme try latest Canary |
| 19:48 | <MikeSmith> | could be some regression |
| 19:49 | <AryehGregor> | Doesn't work for me in Chrome 18.something on Windows . . . maybe it's dependent on GPU support or something crazy like that? |
| 19:50 | <MikeSmith> | Nothing that Chrome does is crazy |
| 19:50 | <MikeSmith> | by definition |
| 19:50 | <MikeSmith> | it can only do good |
| 19:51 | <MikeSmith> | anyway, seems to work fine for me also in current Canary |
| 19:51 | <MikeSmith> | 18 whatever |
| 19:52 | <MikeSmith> | as long as I understand what the demo is supposed to do |
| 19:52 | <MikeSmith> | which understanding is limited to, I move my mouse over the purty things and they move |
| 19:52 | <AryehGregor> | Sliding the slider should do something. |
| 19:52 | <AryehGregor> | Doesn't do anything for me in my Windows VM either. |
| 19:52 | <AryehGregor> | Hmmmm. |
| 19:52 | <AryehGregor> | (in Safari 5 or Chrome dev) |
| 19:53 | <AryehGregor> | I suspect this has to do with GPU support. |
| 19:53 | <AryehGregor> | Since my VM has no 3D acceleration, and my Linux driver may or may not be good enough. |
| 19:53 | AryehGregor | tries on his laptop, which has Linux with a different video card/driver |
| 19:54 | <AryehGregor> | Doesn't work there either. |
| 19:54 | <AryehGregor> | Kind of hard for me to test the features if they don't work. |
| 19:54 | <MikeSmith> | fwiw, sliding the slider does do something. But The behavior above 1000 on that slider seems pretty pointless since it looks all pretty much the same to me |
| 19:54 | <AryehGregor> | Most of the 3D transforms themselves seem to work, though . . . |
| 19:54 | <AryehGregor> | Yes, that's expected. |
| 19:55 | <AryehGregor> | I think. |
| 19:56 | <AryehGregor> | Right. 1000px is quite far away already, and getting further away will make no big difference. The distortion is related to one over the distance, with a perspective of infinity being a normal-looking webpage. |
| 19:56 | <AryehGregor> | Anyway, I'm annoyed. |
| 19:56 | AryehGregor | tries on a non-virtual Windows machine |
| 19:56 | <AryehGregor> | MikeSmith, thanks for helping out! |
| 19:56 | <jamesr_> | AryehGregor, on chrome, if GPU stuff doesn't seem right look at about:gpu |
| 19:57 | <jamesr_> | it'll tell you which features can/cannot be enabled |
| 19:57 | <AryehGregor> | jamesr_, thanks! |
| 19:57 | <AryehGregor> | 3D CSS: Unavailable. Hardware acceleration unavailable |
| 19:58 | <AryehGregor> | NVIDIA cards with nouveau drivers in Linux are crash-prone.: 94103 |
| 19:58 | <jamesr_> | yup |
| 19:58 | <jamesr_> | welcome to linux |
| 19:58 | <AryehGregor> | jamesr_, is there no software rendering fallback for 3D CSS? |
| 19:58 | <AryehGregor> | It seems to work in Firefox. |
| 19:58 | <jamesr_> | where the drivers are crashy and the user doesn't matter |
| 19:58 | <AryehGregor> | (presumably in software) |
| 19:58 | <jamesr_> | i'd actually think firefox is just using the driver |
| 19:58 | <jamesr_> | we don't have s/w fallback rendering for 3D css on linux |
| 19:59 | Philip` | didn't think Nouveau worked well enough for any kind of 3D at all |
| 19:59 | <AryehGregor> | Philip`, I enabled the scary experimental Gallium3D stuff, IIRC. |
| 19:59 | <AryehGregor> | Also, seems like mesa for Intel is *also* crash-prone, yay! |
| 20:00 | <jamesr_> | the only drivers that seem reasonable on linux are the very recent binary blob ones, based on the reports we've received |
| 20:00 | <AryehGregor> | Matches my experience gaming. |
| 20:00 | <AryehGregor> | NVIDIA binary works okay, everything else is garbage. |
| 20:01 | <AryehGregor> | From about:support in Firefox: "GPU Accelerated Windows 0" |
| 20:01 | <AryehGregor> | But it still supports 3D CSS. |
| 20:01 | <jamesr_> | what does 0 mean? |
| 20:02 | <Ms2ger> | None |
| 20:02 | <Philip`> | It's the typical representation of the number zero |
| 20:02 | <Ms2ger> | It's a count |
| 20:03 | <AryehGregor> | jamesr_, according to http://blog.mozilla.com/joe/2010/11/10/how-to-tell-if-youre-using-hardware-acceleration/, it seems it means "no GPU acceleration". |
| 20:05 | <Hixie> | hsivonen: i thought the issue of newlines in title="" was already handled |
| 20:06 | Philip` | hasn't experienced many serious compatibility problems with 3D game development on Linux, except for non-Gallium R300 drivers breaking when using shaders, and one version of binary NVIDIA drivers crashing any multithreaded application |
| 20:06 | <Philip`> | and a few behavioural bugs that got fixed after reporting them |
| 20:07 | <Philip`> | but I'd not be surprised if doing anything slightly more complex (like using GLSL) hit loads more bugs |
| 20:10 | <Hixie> | anyone know what's up with irc.w3.org? |
| 20:11 | <MikeSmith> | Hixie: working for me |
| 20:11 | <Hixie> | i get "Unable to connect server irc.w3.org port 6665 [Connection timed out]" from norway |
| 20:12 | <MikeSmith> | but I have today seen reports from people saying they are not able to get to W3C servers |
| 20:12 | <MikeSmith> | hmm |
| 20:12 | <Hixie> | k, i'll just leave it for now then |
| 20:12 | <MikeSmith> | yeah |
| 20:12 | <MikeSmith> | some routing problems for sure |
| 20:21 | <AryehGregor> | So of course my wife's Windows laptop has outdated graphics drivers, so browsers don't want to use GPU acceleration on it either. |
| 20:21 | <AryehGregor> | And I don't have an admin password. Sigh. |
| 20:40 | <AryehGregor> | Argh, tons of the 3D transform spec isn't well-defined. |
| 20:48 | AryehGregor | wonders if he can take over and rewrite the transform specs |
| 20:50 | <TabAtkins> | jacobolu_: Argh, you're not only reading the /TR draft, you're reading an *old* /TR draft! There have been four revisions since then. >_< |
| 20:51 | <TabAtkins> | AryehGregor: DO IT |
| 20:51 | <TabAtkins> | DO IT NOW |
| 20:51 | <AryehGregor> | How does that work? Do we have to have some WG discussion or something? |
| 20:51 | <AryehGregor> | I mean, it has a bunch of editors already. |
| 20:52 | <TabAtkins> | AryehGregor: The chairs decide officially, but in practice it's an "anyone object?" vote. |
| 20:52 | <jamesr_> | AryehGregor, who? |
| 20:52 | <jamesr_> | are editors, that is? |
| 20:52 | <TabAtkins> | Submit the request now, so we can vote on it on wednesday. |
| 20:52 | <AryehGregor> | jamesr_, a bunch of people from Apple. |
| 20:52 | <AryehGregor> | TabAtkins, I'll have to run it past my Mozilla overlords first. |
| 20:52 | <jamesr_> | names? |
| 20:53 | <TabAtkins> | jamesr_: You can just look, you know. |
| 20:53 | jamesr_ | hunts |
| 20:53 | <AryehGregor> | jamesr_, http://dev.w3.org/csswg/css3-transforms/#transform-origin-property |
| 20:53 | <TabAtkins> | jamesr_: http://dev.w3.org/csswg/ |
| 20:53 | <AryehGregor> | TabAtkins, is there some way to get the version history? |
| 20:53 | <jamesr_> | sloooooooooooooow |
| 20:53 | <TabAtkins> | AryehGregor: Yeah, but it's painful. |
| 20:54 | <AryehGregor> | Let me guess, it requires a CVS checkout? |
| 20:54 | <TabAtkins> | http://dev.w3.org/cvsweb/csswg/ |
| 20:54 | <TabAtkins> | Or a CVS checkout, yes. |
| 20:56 | <jamesr_> | AryehGregor, i don't think any of those people are actively working on this |
| 20:56 | <jamesr_> | the spec, anyway |
| 20:56 | <AryehGregor> | It doesn't look like it. |
| 20:56 | <jamesr_> | except for possible chris marrin, but in the last discussion of plane intersection he said that he doesn't think css should define the actual rendering |
| 20:56 | <TabAtkins> | No, it's another one of the specs that Apple dropped on the ground after they got it far enough in the WG that they can claim they're implementing a standard. |
| 20:57 | <dglazkov> | ouch. |
| 20:57 | <TabAtkins> | Same thing happened with Transitions and Animations, which is why dbaron and sylvain are editors now. |
| 21:03 | <dbaron> | AryehGregor, you may want to see the thread following http://lists.w3.org/Archives/Public/public-fx/2011OctDec/0195.html for the different transforms specs |
| 21:04 | <AryehGregor> | TabAtkins, is there anything special I have to know if I'm a CSS spec editor? |
| 21:04 | <AryehGregor> | dbaron, hmm, thanks. |
| 21:04 | AryehGregor | looks |
| 21:04 | <jamesr_> | AryehGregor, well there's the nozzle |
| 21:04 | <jamesr_> | but otherwise it's like anything else |
| 21:05 | <TabAtkins> | AryehGregor: Read some of the things at the bottom of http://wiki.csswg.org/spec |
| 21:05 | <TabAtkins> | But otherwise, we can fill you in as necessary. |
| 21:05 | <TabAtkins> | And then turn that into documentation for the future. |
| 21:06 | <TabAtkins> | On that note, I should write a quick page about calc() and percentages. |
| 21:06 | <AryehGregor> | dbaron, FWIW, based on what I've seen, it makes sense to me to advance 2D as it stands and then make a new unified spec with 3D that will progress more slowly. |
| 21:06 | <TabAtkins> | +1 |
| 21:06 | <AryehGregor> | 2D Transforms can almost go to PR as-is: it has two interoperable implementations of virtually everything in my testing. |
| 21:06 | <AryehGregor> | Except CSSMatrix. |
| 21:06 | <jamesr_> | TabAtkins, also FYI the w3c does not publish standards |
| 21:06 | <AryehGregor> | And there are a couple of things I didn't test. |
| 21:07 | <jamesr_> | it produces recommendations. orgs like ECMA and ISO produce standards |
| 21:07 | <jamesr_> | (in the context of that obsolescence notice thread with the grazy glenn) |
| 21:07 | <jgraham> | man I thought I missed a perfectly good VCS flamewar and then you went and mentioned specs still in CVS, so I still get to have a good laugh at others' misery. Thanks! |
| 21:07 | <TabAtkins> | You're welcome. |
| 21:07 | <jamesr_> | i don't plan to jump into that mess but The More You Know |
| 21:08 | <AryehGregor> | TabAtkins, also, if I took over editing, do you think they'd let me use hg? |
| 21:08 | <TabAtkins> | jamesr_: I honestly don't give a fuck. The correct answer is "we care about the web, not whatever industry you're referring to". |
| 21:08 | <jgraham> | I think W3C now claims that it does make standards |
| 21:08 | <jgraham> | AryehGregor: Unles the CSS WG is more fucked than I realised, the rules are "(s)he who does the work makes the rules" |
| 21:08 | <TabAtkins> | AryehGregor: We... prefer to keep things together. That said, last week we talked about switching to Hg. I think dbaron wrote a guide for it, so hell, try it out. |
| 21:09 | <jgraham> | TabAtkins: I have heard several times that you are switching to hg |
| 21:09 | <jgraham> | From pliniss at least |
| 21:09 | <TabAtkins> | jgraham: Except last week it was more than one person saying so. |
| 21:10 | <jgraham> | But seriously if you are expecting people to use CVS to edit your specs... well I think it wouldn't be a positive thing |
| 21:10 | <jgraham> | We generally want *more* ediors |
| 21:10 | <jamesr_> | requestanimationframe uses hg. it's fine |
| 21:10 | <jgraham> | jamesr_: Not CSSWG |
| 21:10 | <TabAtkins> | Yeah, so I think Aryeh should go ahead and use it. |
| 21:11 | <jamesr_> | jgraham, i know, but the w3 setup is fairly sane |
| 21:11 | <jgraham> | Well sort of |
| 21:11 | <jgraham> | I am sort of considering suggesting that we move testsuite development to github so that we could have a semi-functional review system. But I fear I would be tilting at windmills |
| 21:12 | <TabAtkins> | No, do so, please. |
| 21:12 | <jgraham> | The problem with having a totally kickass internal code review setup is that everything else looks lame |
| 21:12 | <jgraham> | And teh fact that we have essentially nothing for testsuites is super lame |
| 21:12 | <dglazkov> | jgraham: why not adopt WebKit's review system? It's open source. |
| 21:13 | dglazkov | doesn't in fact volunteer :P |
| 21:13 | <jgraham> | dglazkov: I guess without knowing that it is tied to svn and doesn't meet my requeirements for a usable system |
| 21:14 | <jgraham> | The first is a showstopper, the second just means that I would whine a lot |
| 21:14 | <dglazkov> | jgraham: it works both with git and svn, and the SCM parts are abstracted out, so it should be easy to write hg support. |
| 21:14 | <dglazkov> | and webkit-patch is freakishly nice. |
| 21:14 | <jgraham> | dglazkov: Pointer? |
| 21:15 | <dbaron> | TabAtkins, http://wiki.csswg.org/spec/hg isn't at all finished |
| 21:15 | <jgraham> | Anyway moving to github would be like 0 effort if people would agree to it |
| 21:15 | <TabAtkins> | dbaron: Ah, there it is. Could you link that from /spec? |
| 21:17 | <dglazkov> | jgraham: here's an example: https://bugs.webkit.org/attachment.cgi?id=123585&action=review |
| 21:17 | <jgraham> | And the code review system there isn't perfect but it has basically the right idea (a code review is a series of commits that you would like to apply. You kep adding commits until you have something that is r+ Then you merge your commits into the main codebase. Github uses multiple repositories, you can also use multiple branches) |
| 21:17 | <jgraham> | (there === github) |
| 21:19 | <jgraham> | (I think gerrit also has this model but... java) |
| 22:26 | <jacobolu_> | TabAtkins: oh. that's where someone linked me, sorry. is there a better link for crossfade? |
| 22:26 | <TabAtkins> | jacobolus: First, always check http://dev.w3.org/csswg/ for CSS specs. In particular, http://dev.w3.org/csswg/css3-images |
| 22:27 | <TabAtkins> | cross-fade() has been punted to level 4 in recent drafts, but it was present in an improved form in the draft following the one you were looking at. |
| 22:30 | <jacobolus> | gotcha |
| 22:30 | <jacobolus> | I was going to say, I don't see it there at all :) |
| 22:30 | <TabAtkins> | But I'm still pretty sure your objection earlier was wrong. |
| 22:30 | <jacobolus> | it's mainly a question of wording/definitions |
| 22:30 | <TabAtkins> | Right now I use plus(dissolve(A), dissolve(B)) |
| 22:31 | <jacobolus> | what's dissolve? |
| 22:31 | <TabAtkins> | defined in p-d |
| 22:32 | <jacobolus> | oh, I see it |
| 22:32 | <jacobolus> | TabAtkins: yes, that's a correct formula then |
| 22:33 | <TabAtkins> | It should amount to the "linearly interpolate each pixel's color components". |
| 22:33 | <jacobolus> | the language in the spec before said something like over(dissolve(A, α), dissolve(B, 1-α)) |
| 22:33 | <TabAtkins> | Yes. |
| 22:33 | <jacobolus> | but over is not the same as plus |
| 22:33 | <TabAtkins> | Indeed. |
| 22:33 | <TabAtkins> | But over(dissolve(A,alpha), B) isn't right either. |
| 22:34 | <jacobolus> | it is if the two images are opaque |
| 22:34 | <TabAtkins> | Yes, but I can't rely on that assumption. |
| 22:34 | <jacobolus> | it isn't if they're semitransparent :) |
| 22:34 | <jacobolus> | that's true |
| 22:34 | <jacobolus> | as I said, that's the only real use case I see for the spec |
| 22:35 | <jacobolus> | cross-fading between semitransparent images |
| 22:35 | <TabAtkins> | What specifically do you mean by "the spec" here? |
| 22:35 | <jacobolus> | I mean, the `cross-fade` css property |
| 22:36 | <TabAtkins> | The major use of cross-fade() is to allow 'transition' to work on background-image. |
| 22:36 | <jacobolus> | that is, the only real use case that isn't already covered by just having variable opacity |
| 22:36 | <jacobolus> | gotcha |
| 22:37 | <TabAtkins> | Rather, that's the reason I made it. It has other minor uses, of course. |
| 22:37 | <jacobolus> | just layering two boxes works in most instances |
| 22:39 | <TabAtkins> | We try to avoid making people add multiple boxes to achieve common effects. ^_^ |
| 22:40 | <jacobolus> | that's fair :) |
| 23:19 | <dglazkov> | in HTML spec, is there a notion of a document viewport, it's existence and difference in behavior of elements in that document due to viewport exists or not? |
| 23:20 | <dglazkov> | like, if you create var d = document.implementation.createHTMLDocument(); does the spec say how <img> or <iframe> must behave if they are in tree d? |
| 23:21 | <Hixie> | i'm not sure what you are asking |
| 23:21 | <Hixie> | viewport is a CSS concept relating to visual rendering |
| 23:21 | <Hixie> | you might be looking for browsing contexts? |
| 23:22 | <dglazkov> | hmmm. possibly. |
| 23:22 | <Hixie> | to answer your second question, yes. |
| 23:22 | <dglazkov> | found it. Awesome. |
| 23:22 | <dglazkov> | thanks! |
| 23:40 | <dglazkov> | Hixie: ok, I can't find how image element should work in absence of browsing context: http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-1.html#embedded-content-1 |
| 23:42 | <dglazkov> | Gecko, WebKit, and IE all have slightly different interpretations of what should happen with attributes: http://jsfiddle.net/zL9P2/2/ |
| 23:42 | <Hixie> | can you be more specific regarding what you mean by "work"? |
| 23:43 | <dglazkov> | Hixie: ^^^ |
| 23:43 | <Hixie> | i don't see anything in that example that should be affected by whether or not there's a browsing context |
| 23:44 | <Hixie> | the only interesting thing going on in that example is url resolution and reflection in an idl attribute |
| 23:44 | <dglazkov> | right -- and whether an image is loaded or not |
| 23:45 | <dglazkov> | IE resolves URLs right away, whether in browser contextless doc or not |
| 23:45 | <dglazkov> | IE9 that is |
| 23:46 | <Hixie> | nothing in that tests checks the network side of things as far as i can tell |
| 23:46 | <Hixie> | nor does anything check when the url gets resolved |
| 23:46 | <Hixie> | as far as i can tell |
| 23:46 | <dglazkov> | well, no, but you can look at the console. |
| 23:47 | <Hixie> | well, browsers can do whatever they want in the console :-P |
| 23:48 | <Hixie> | anyway to answer what i think you mean to be your question, the spec says "A user agent that obtains images immediately must synchronously update the image data of an img element whenever that element is either created with a src attribute [...]" |
| 23:48 | <Hixie> | which means that even a document fetched via XMLHttpRequest would result in network traffic if it contained an <img> |
| 23:48 | <Hixie> | this is probably not what the spec should require |
| 23:49 | <Hixie> | ("update the image data" doesn't check if there's a browsing context) |
| 23:56 | <dglazkov> | also interesting stuff on url resolution. Gecko does the seemingly sanest thing and reports back both attribute and property value unresolved when the browsing context doesn't exist |
| 23:56 | <dglazkov> | WebKit doesn't report a property value, which is probably wrong. |
| 23:57 | <dglazkov> | IE9 happily resolves against document URL regardless of the state. |
| 23:57 | <dglazkov> | hate results, but love the attitude. |
| 23:58 | <Hixie> | i don't see why browsing context existence would have any effect on url resolution in idl attribute values |
| 23:58 | <Hixie> | (i do see why it should be a precondition for network traffic) |
| 23:59 | <Hixie> | (file a bug if you want that fixed) |
| 23:59 | <dglazkov> | k. I am just trying to find the most analogous thing to inert chunks of DOM for HTML templates. |