| 08:34 | <annevk> | heycam|away: isn't EventListener the wrong name as that clashes with DOM where it means something else? |
| 08:35 | <annevk> | heycam|away: other than that I guess it's okay |
| 12:39 | <asmodai> | Using canvas to remake old Amiga demos: http://www.abyss-online.de/cruisin/ |
| 12:41 | <bga_> | hehe const |
| 13:21 | <annevk> | http://lists.w3.org/Archives/Public/www-style/2011Dec/0482.html Plan B :) |
| 13:31 | <Ms2ger> | Isn't it like Plan M by now? |
| 13:34 | <annevk> | yeah maybe |
| 14:00 | <bga_> | usecase of FileSystem and native ui |
| 14:00 | <bga_> | http://i5.pixs.ru/storage/5/8/9/Snimok2011_1083103_3655589.png |
| 14:00 | <bga_> | :( |
| 14:01 | <bga_> | pseudo antivirus |
| 14:24 | <zcorpan> | annevk: one site is not enough data to make a decision on BOM, i think |
| 14:26 | <annevk> | zcorpan: WebKit/Trident is I think |
| 14:29 | zcorpan | is still not convinced |
| 14:30 | gsnedders | isn't convinced either |
| 14:31 | <gsnedders> | Avoiding layering violations is preferable. If Gecko/Presto can get away without violating it, all the better. |
| 14:31 | <gsnedders> | s/all the better/that implies it is feasible/ |
| 14:31 | <zcorpan> | hmm, BOM wins over content-type for <script src> |
| 16:21 | <AryehGregor> | annevk, I don't see how to define the constants more similarly to compareDocumentPosition. compareDocumentPosition uses the constants for its return value, compareBoundaryPoints uses them for a parameter. |
| 16:23 | <AryehGregor> | Also, can you or someone else tell me if IE9 fails any tests here? Gecko and WebKit both pass everything (yay). http://aryeh.name/tmp/webapps/DOMCore/tests/submissions/AryehGregor/Range-compareBoundaryPoints.html |
| 16:24 | <Philip`> | AryehGregor: My IE9 gets 11994 pass, 2646 fail |
| 16:25 | <AryehGregor> | Hmm. |
| 16:25 | <Philip`> | (Lots of "assert_not_equals: Creating argument range threw an exception got disallowed value null" and some others) |
| 16:25 | <AryehGregor> | Ah, okay. |
| 16:26 | <AryehGregor> | So it's failing on the range creation, not the compareDocumentPosition results, probably. |
| 16:26 | <AryehGregor> | I'll look later if I remember. |
| 16:26 | <AryehGregor> | But I'm probably not going to go with IE against Gecko+WebKit anyway, so it's unlikely I actually need to look at the results. |
| 16:26 | <Philip`> | There's some others like "assert_equals: Wrong return value expected 1 but got -1" |
| 16:27 | <annevk> | AryehGregor: that somehow the <dfn> end up in prose and not the IDL |
| 16:27 | <annevk> | AryehGregor: but I don't really know how that should be done either |
| 16:27 | <annevk> | so it's fine to leave it until we think of something |
| 16:27 | <AryehGregor> | Oh. |
| 16:27 | AryehGregor | looks |
| 16:27 | <AryehGregor> | Philip`, oh, that sounds more interesting. Could you give me an example of the test name for one of those? |
| 16:27 | AryehGregor | should get his IE9 working again sometime today, if all goes well . . . |
| 16:28 | <AryehGregor> | Maybe IE9 matches the current spec instead of Gecko/WebKit? . . . |
| 16:28 | <annevk> | I wonder if Jacob is responsible |
| 16:29 | <annevk> | can probably ask him |
| 16:29 | <annevk> | dunno if he works today though |
| 16:29 | <Philip`> | AryehGregor: "1,39,2" (is that a sufficiently unique part of the name?) |
| 16:29 | <AryehGregor> | Yes. |
| 16:29 | <AryehGregor> | That's why I put the identifier there. :) |
| 16:29 | <Philip`> | Ah, good, so I don't have to type it all out onto this computer :-) |
| 16:31 | <AryehGregor> | That seems weird. Might be a test bug. |
| 16:31 | <AryehGregor> | I'll try to remember to look into it. |
| 16:34 | <annevk> | I wish it was not the end of December |
| 16:35 | <annevk> | it's making it hard to make progress on work |
| 16:36 | <AryehGregor> | If I just made a commit in hg and want to tweak the commit message, is there any easy way to do that, like git commit --amend? Or do I have to hg rollback and retype the whole message? |
| 16:37 | <annevk> | unless you have extensions, I think the latter |
| 16:38 | <annevk> | but I don't really know a lot about Mercurial |
| 16:38 | <annevk> | or git |
| 16:38 | <annevk> | or svn or cvs |
| 16:38 | <annevk> | :) |
| 16:39 | <AryehGregor> | :( |
| 16:39 | Philip` | doesn't remember hearing of anything better than "hg qimport -r tip -n somename; hg qref -e; hg qfinish" |
| 16:39 | <AryehGregor> | Figures. |
| 16:39 | <Philip`> | (which should allow incremental editing of the current message) |
| 16:40 | <Philip`> | (but is a weird way to do it) |
| 16:41 | <Philip`> | (Hg seems to generally assume that commits imply some kind of commitment to the change, whereas git seems to avoid any commitment until you push) |
| 16:41 | <AryehGregor> | The latter is kind of logical, given that it's normally private until you push . . . |
| 16:41 | <Philip`> | The former is kind of logical given that it's called "commit", though |
| 16:56 | <zewt> | i'm not touching hg or "bzr" or any of the other zillion versioning systems; there's been an unreasonable proliferation of them in the last few years and everyone seems to just expect people to learn all of them |
| 16:56 | <zewt> | i just stick to svn and git |
| 16:57 | <AryehGregor> | I am starting to get severely annoyed by compiz messing up regularly on my laptop. |
| 16:57 | <AryehGregor> | Sigh. |
| 16:57 | <AryehGregor> | zewt, I stick to git when I can, but the W3C insists on hg . . . |
| 16:57 | <michel_v> | zewt: it's annoying when you want to check a project's code and you have to install yet another VC system... |
| 16:58 | <zewt> | someone actually got offended when i told him that sorry, no, i'm not learning to use bzr to test a patch for his obscure project, sorry |
| 17:02 | <AryehGregor> | wtf, this laptop is unusable. Argh. |
| 17:02 | AryehGregor | stabs Linux |
| 17:02 | <zewt> | linux on a laptop? heh |
| 17:02 | <AryehGregor> | Oh, now Alt-Tab started working again. |
| 17:02 | <AryehGregor> | Sometimes it stops, and Chrome doesn't respond to clicks . . . grr. |
| 17:03 | <AryehGregor> | To be fair, I should stab Ubuntu. I'm sure this is a Unity bug. |
| 17:03 | <zewt> | i gave up on ubuntu, but i only used it for servers |
| 17:05 | <AryehGregor> | I know. |
| 17:11 | <annevk> | the short term for the thing I mentioned yesterday is apparently "Ruby's postulate" |
| 17:11 | <annevk> | per http://www.comp.leeds.ac.uk/roger/Famous/eponymy_content.html |
| 17:29 | <AryehGregor> | Can someone with IE9 access tell me whether it passes the tests 0,0,{4,5,6,7,9,10,11,16,etc.} here? Or does it give a whole bunch of "IndexSizeError required" errors? http://aryeh.name/tmp/webapps/DOMCore/tests/submissions/AryehGregor/Range-compareBoundaryPoints.html |
| 17:30 | <AryehGregor> | It seems like Gecko, WebKit, and Opera all throw InvalidStateError rather than IndexSizeError if both are applicable. |
| 17:30 | <AryehGregor> | I want to check if IE does the same thing. |
| 17:33 | <AryehGregor> | I just changed the test, so my previous question is invalid now. |
| 17:47 | <AryehGregor> | gwicke, didn't know you hung out here. Are you active in MediaWiki development these days? |
| 17:48 | <gwicke> | AryehGregor: yes, I am working on a parser for the visual editor effort |
| 17:48 | <AryehGregor> | Nice. |
| 17:48 | <gwicke> | https://www.mediawiki.org/wiki/Future/Parser_development |
| 17:49 | <AryehGregor> | I was an active MW developer from like 2007 to 2010, so I guess we didn't overlap, but obviously I saw your name all over the skin code. |
| 17:49 | <AryehGregor> | Now I work on specs. |
| 17:50 | <gwicke> | not a bad thing to work on |
| 17:50 | <gwicke> | especially as things move towards html again |
| 17:53 | <gwicke> | I really hope that contenteditable and dom transactions become usable some day |
| 17:59 | <AryehGregor> | gwicke, contenteditable is the primary thing I'm paid to work on, as it happens. I wrote a mostly-complete spec, but now it's up to implementers to implement it. |
| 17:59 | <AryehGregor> | (I also follow DOM transactions, but rniwa does the actual spec work there) |
| 17:59 | <gwicke> | heh- nice! |
| 18:00 | <gwicke> | is there some interest in implementing it? |
| 18:01 | <AryehGregor> | Ehsan Akhgari of Mozilla and Ryosuke Niwa of Google (for WebKit) are very interested, but it will require major rewrites with significant compat risk, so they're proceeding very cautiously in actually doing it. |
| 18:01 | <AryehGregor> | Only Gecko and WebKit have anyone working on their editing code at all, AFAICT. |
| 18:01 | <AryehGregor> | Although if they start converging on the spec, I expect MS to follow suit. |
| 18:02 | <gwicke> | that is good to hear, and I really hope that they push it through |
| 18:02 | <AryehGregor> | Anyway, my editing spec work is basically blocked waiting for implementer feedback, so I spend most of my time on other things. (Like right now, making the definition of Range.compareBoundaryPoints more precise and writing lots of tests for it.) |
| 18:03 | <gwicke> | the visual editor we are working on tries to do without contenteditable, but of course that is a lot of work |
| 18:04 | <AryehGregor> | What do you use instead? Canvas? A regular DOM where you try to intercept all keystrokes? |
| 18:04 | <gwicke> | regular DOM |
| 18:05 | <gwicke> | and a hidden textarea behind it for pasting |
| 18:05 | <zewt> | every manual gui editor ever implemented in javascript sucks in some nontrivial way :| |
| 18:05 | <AryehGregor> | The problem with that is that you then lose all the platform features users expect, like platform-specific (perhaps customized!) text navigation, and IMEs, and things like that. |
| 18:05 | <gwicke> | but I am mostly working on the parser, so I only know some stuff at the periphery |
| 18:05 | <AryehGregor> | Google Docs mostly pulls it off, but only because they have Google's resources, and they still have lots of problems. |
| 18:05 | <zewt> | even gmail's editor has its serious annoyances |
| 18:06 | <zewt> | probably the highest-profile editor of its type in existance |
| 18:06 | <AryehGregor> | Most of the successful WYSIWYG editors (TinyMCE, CKEditor, etc.) use contenteditable and then hack around browser incompatibilities, AFAIK. |
| 18:07 | <AryehGregor> | zewt, Gmail's editor uses contenteditable. |
| 18:07 | <gwicke> | I try to stay compatible with contenteditable, as I am not that convinced of the manual DOM approach either |
| 18:07 | <gwicke> | the parser produces an HTML DOM first, and only converts that to an editor-specific WikiDom format at the end before passing that to the editor |
| 18:07 | <zewt> | it's pretty bad about undo and cursor positions, which is a day-to-day irritant |
| 18:08 | <zewt> | (the cursor position after undo seems to be selected via a cryptographically-secure prng) |
| 18:08 | <AryehGregor> | Manual DOM will seem to work okay in development, because it will work with all the keyboard shortcuts and things that the developers are used to. Then you expose it to your users and find that it's completely unusable for CJK, doesn't respect OS preferences you never heard of, undo breaks in a million cases you didn't think of, . . . |
| 18:09 | <AryehGregor> | Can someone tell me how IE9 does on the tests here starting with "3,3,"? http://aryeh.name/tmp/webapps/DOMCore/tests/submissions/AryehGregor/Range-compareBoundaryPoints.html |
| 18:09 | <AryehGregor> | I really need to get IE9 working again. Blech. |
| 18:09 | <zewt> | you really need to be able to run IE. heh |
| 18:09 | <zewt> | ^ |
| 18:10 | AryehGregor | is considering whether to change IndexSizeError to NotSupportedError, since Opera throws that, Gecko throws some non-DOMException, and WebKit doesn't throw |
| 18:10 | <AryehGregor> | But if IE throws IndexSizeError, of course, I should keep it. |
| 18:12 | AryehGregor | realizes his wife's Vista laptop is sitting five feet behind him |
| 18:13 | <AryehGregor> | But a) I don't have a login and b) I think she prefers IE8. Blech. |
| 18:14 | <AryehGregor> | annevk, Philip`, can you tell me how IE9 does on the tests here starting with "3,3,"? http://aryeh.name/tmp/webapps/DOMCore/tests/submissions/AryehGregor/Range-compareBoundaryPoints.html |
| 18:20 | <Philip`> | AryehGregor: 3,3,(4|5|6) says it threw object "Error: Invalid argument." |
| 18:21 | <Philip`> | Also 3,3,(11|16|17|22|23|28|34|35) and like half the rest but I got bored of counting |
| 18:21 | <Philip`> | The others pass |
| 18:22 | <kennyluck> | Does anyone have a list of pages with <base> elements in real use? |
| 18:24 | <AryehGregor> | Philip`, great, thanks. |
| 18:25 | <AryehGregor> | Philip`, it says something like 'threw object "Error: Invalid argument." that is not a DOMException INDEX_SIZE_ERR: property "code" is equal to undefined, expected 1' or similar? |
| 18:26 | <Philip`> | AryehGregor: Exactly that |
| 18:26 | <AryehGregor> | Great, thanks. |
| 18:26 | <Philip`> | (...prefixed by the "assert_throws: blah blah function "..."") |
| 18:28 | <AryehGregor> | Right. |
| 18:28 | <AryehGregor> | So I'm actually making the spec such that Opera is the only one that passes certain tests. Yay Opera. |
| 18:28 | <AryehGregor> | (that only happens when all other browsers behave in some way I can't standardize, like throwing nonstandard exception types) |
| 18:29 | <Philip`> | (You could always standardise nonstandard exception types) |
| 18:30 | <AryehGregor> | Of course, Opera fails 11,012 tests total, as opposed to 544 for Gecko and 616 for WebKit. But at least it passes 3,3,4! |
| 18:30 | <AryehGregor> | Philip`, I'm not likely to standardize an exception type called NS_ERROR_ILLEGAL_VALUE. |
| 18:30 | <AryehGregor> | The NS_ is a bit of a giveaway. |
| 18:35 | <Philip`> | HTML5 already standardises the handling of strings like "-//Netscape Comm. Corp.//DTD HTML//", so why not? |
| 18:38 | <zewt> | if lots of people inadvertently depend on one and not the other, then there's a pretty big difference |
| 18:43 | <AryehGregor> | Philip`, if the web actually depended on it, okay . . . |
| 18:44 | <AryehGregor> | Okay, now what should I work on? |
| 18:44 | AryehGregor | scrutinizes Range methods |
| 18:45 | <AryehGregor> | commonAncestorContainer! |
| 18:45 | <annevk> | browsers are going to change their exception types anyway |
| 18:46 | <annevk> | at least that's been the assumption thus far |
| 18:50 | <Philip`> | Does anyone ever depend on exception types at all? |
| 18:50 | <AryehGregor> | Well, my tests do. |
| 18:50 | <AryehGregor> | So if browsers don't interoperate, they'll fail my tests. |
| 18:50 | <AryehGregor> | :) |
| 18:50 | <Philip`> | Anyone other than you :-) |
| 18:51 | <Ms2ger> | My tests :) |
| 18:55 | <AryehGregor> | Okay, commonAncestorContainer tests were pretty quick to write. |
| 19:06 | <Ms2ger> | I think I had a couple already, though probably not up to your standards |
| 19:07 | <AryehGregor> | Oh, right, I remember that. |
| 19:07 | AryehGregor | looks |
| 19:08 | <AryehGregor> | I don't think I test exactly the same cases as you, because my document fragments are all empty. |
| 19:09 | <AryehGregor> | I only actually test 60 different ranges. |
| 19:09 | <AryehGregor> | I need to write a testRangesExtended with a couple thousand ranges for things like this where we're only running one test per range, probably. |
| 19:10 | <Ms2ger> | Uh-oh :) |
| 19:10 | <AryehGregor> | Presumably I'd generate it programmatically from testNodes. |
| 19:10 | <AryehGregor> | Just take the Cartesian product with itself and then knock out all the ones that aren't valid Ranges. |
| 19:11 | <AryehGregor> | Not entirely trivial to do, so I'll put it off for later. |
| 19:14 | <gsnedders> | AryehGregor: You mean our handling of Ranges has a few more bugs than other browsers? Colour me surprised! |
| 19:14 | <AryehGregor> | gsnedders, I know, shocking. |
| 19:15 | <AryehGregor> | I think some of them are actually DOM bugs that make the tests give incorrect expected values. |
| 19:16 | <gsnedders> | Eh, that would surprise me a bit more. |
| 19:22 | <AryehGregor> | It's mostly not that, though. |
| 19:22 | <AryehGregor> | I think there are only a handful of those. |
| 23:26 | <divya> | dglazkov: (reading the spec again) http://dvcs.w3.org/hg/webcomponents/raw-file/tip/spec/shadow/index.html "A valid selector fragment fragment may contain" what is the criteria for it? |
| 23:27 | <divya> | dglazkov: is it just not having pseudo-elements and :hovers? basically any selectable element? |
