| 00:07 | <Hixie> | benschwarz: wfm |
| 00:09 | <Hixie> | hm, turns out we have a help.whatwg.org already, looks like it's run by MikeSmith |
| 00:11 | MikeSmith | wakes up |
| 00:11 | <Hixie> | and a syntax.whatwg.org, i wonder what that was for |
| 00:11 | <MikeSmith> | I'm just to the first one to have put anything up on help.whatwg.org I guess |
| 00:12 | <MikeSmith> | I'm happy if anybody wants to put whatever else they'd like there |
| 00:12 | <Philip`> | http://forums.whatwg.org/viewtopic.php?t=11 |
| 00:13 | <MikeSmith> | Hixie: I think syntax.whatwg.org might have been intended for hosting that relaxng and datatype library for the validator |
| 00:13 | <Philip`> | "I found the DTD file for the HTML 5.0 Draft. <http://syntax.whatwg.org/sgml/html5core+wf2/pre1/dtd>"; |
| 00:13 | <MikeSmith> | *the relaxng schemas |
| 00:13 | <Philip`> | MikeSmith: That was http://syntax.whattf.org/, I thought |
| 00:13 | <MikeSmith> | Philip`: yeah |
| 00:13 | <Hixie> | syntax.whatwg.org is empty as far as i can tell |
| 00:13 | <Hixie> | i guess i'll delete it for now |
| 00:13 | <Hixie> | can always add it back if it's actually useful |
| 00:56 | <Hixie> | http://www.w3.org/mid/AANLkTikqcJmCoHfRo1JwzW+s_jp5PsmaZ3mHnEEwq_Cn⊙mgc pretty much summarises the problem with making decisions based on consensus |
| 00:57 | <jamesr_> | Hixie: are you not in good standing with the AX TF? |
| 00:58 | <Hixie> | dunno, what is "good standing" for the tf? |
| 00:58 | <Hixie> | that wasn't what i was referring to, though |
| 00:58 | <Hixie> | i just meant, the idea of deciding whether a group agrees to something or not being resolved by defining the group as those who agree |
| 00:59 | <Hixie> | either through gerrymandering or attrition |
| 01:01 | <jamesr_> | speaking of, do you read the public-canvas-api list or know what they are up to? |
| 01:16 | <bga_> | js asi is evil |
| 01:18 | <MikeSmith> | hober: congrats (saw your tweet) |
| 01:21 | <bga_> | brendaneich console.log\n(1, eval)(); |
| 01:22 | <jaket> | great word, gerrymandering |
| 01:23 | <Hixie> | jamesr_: i read it, wasn't aware they were up to anything |
| 01:23 | <Hixie> | jamesr_: anything in particular? |
| 01:24 | <jamesr_> | Hixie: they keep talking about a shadow DOM in canvas and i'm not sure what they mean if it isn't fallback content |
| 01:26 | <Hixie> | they seem to be discussing what the spec says without realising the spec has said it for several months now |
| 01:26 | <Hixie> | the idea is that the fallback content of a <canvas> is supposed to be focusable even when the <canvas> isn't falling back |
| 01:26 | <Hixie> | so that you can make the fallback reflect the image |
| 01:27 | <Hixie> | this lets you get (for free) things like tabbing around controls |
| 01:27 | <Hixie> | works in conjunction with the drawFocusRing() method to automatically render focus rings on controls on the canvas |
| 01:32 | <brendaneich> | bga_: that is not JS ASI |
| 01:32 | <brendaneich> | bga_: the first rule of ASI-club is there's no insertion if there is no error |
| 01:34 | <bga_> | a || (a = b)\n(b && c) || d |
| 01:35 | <brendaneich> | bga_: where is the error? |
| 01:35 | <bga_> | try in web inspector |
| 01:36 | <Hixie> | hober: will you be working on things like http://www.w3.org/Bugs/Public/show_bug.cgi?id=10694 ? |
| 01:36 | <brendaneich> | bga_: what happens? i'm not talking about web inspector, i'm talking about JS |
| 01:37 | <bga_> | in js file too |
| 01:37 | <bga_> | v8 concat two lines |
| 01:38 | <bga_> | and tries to call function |
| 01:38 | <bga_> | because sees call operator |
| 01:41 | <bga_> | in ff4 too |
| 01:41 | <brendaneich> | bga_: why do you think there's an error requiring semicolon insertion? |
| 01:41 | <brendaneich> | bga_: remember the first rule of ASI fight-club |
| 01:41 | <brendaneich> | if there's no error, there is no insertion |
| 01:41 | <brendaneich> | bga_: you're thinking the \n should cause a ; to be inserted? |
| 01:41 | <brendaneich> | v8 and ff4 are correct |
| 01:43 | <bga_> | yes but many ppl do not insert ; in js code |
| 01:43 | <bga_> | its trap :) |
| 01:43 | <aho> | it's fine to use menu for menus, isn't it? |
| 01:43 | <bga_> | like return\n{} |
| 01:46 | <bga_> | js should has minimal count of traps for newbies imho |
| 01:46 | <aho> | i mean, it says those "actions" can be links... and going to a different page is an immediate action, isn't it? |
| 01:47 | <aho> | bga_, jslint helps |
| 01:48 | <bga_> | aho its spike-nail |
| 01:48 | <aho> | what's that supposed to mean? |
| 01:50 | <bga_> | english dictonary fail. sorry |
| 01:51 | <bga_> | i mean crutch |
| 01:51 | <aho> | yes, it is |
| 01:52 | <aho> | but if it's integrated into your text editor or IDE, it's not much of an issue |
| 01:54 | <bga_> | may be is better to fix lang issue instead keep "backward compatibility"? |
| 01:54 | <aho> | you can't really fix js |
| 01:55 | <aho> | it's out there... gotta deal with it |
| 01:55 | <bga_> | i mean new es6 |
| 01:55 | <aho> | js is the least of my problems though |
| 01:56 | <bga_> | js has alot of traps, bugs and fails imho |
| 01:56 | <aho> | it's very straightforward imo |
| 01:57 | <bga_> | or i too critical |
| 01:57 | <Hixie> | man there's all kinds of things that aren't specced right for type=email multiple |
| 01:57 | <aho> | i write lots of js and it's generally completely pain-free |
| 01:58 | <aho> | oh it's the Hixie :) is it fine to use menu for menus? :f |
| 01:58 | <Hixie> | hm? |
| 01:58 | <Hixie> | what do you mean by "menus"? |
| 01:58 | <aho> | top/side navigation bar thingies |
| 01:58 | <aho> | those things which are typically done with ul elements |
| 01:59 | <Hixie> | "The menu element represents a list of commands." |
| 01:59 | <Hixie> | if your "navigation bar" is a list of commands, then sure |
| 01:59 | <aho> | and a command can be a link... and going to some site is some immediate action... |
| 02:00 | <aho> | so... uhm... it's fine, isn't it? |
| 02:00 | <Hixie> | <a href=""></a> is a command, per the html spec, yes |
| 02:02 | <aho> | thing is, i kinda prefer using menu instead of ul for those things, since i wont have to re-reset them and because i think menu makes a little bit more sense than ul |
| 02:02 | <Hixie> | from the html spec's point of view, either is defensible |
| 02:02 | <aho> | thank you :x |
| 02:03 | <Hixie> | <menu> really is intended for drop-down menus (type=toolbar) and context menus (type=context), the list state (no type attribute) is basically only allowed for historical reasons |
| 02:03 | <Hixie> | <nav> is the important thing to have in your DOM if you have a navigation list |
| 02:03 | <aho> | ye, nav>menu>li>a ish |
| 02:04 | <Hixie> | whether you do <nav><p><a>...</a><p><a>...</a></nav>, <nav><ul><li><a>...</a><li><a>...</a></ul></nav>, <nav><menu><p><a>...</a><p><a>...</a></menu></nav>, or <nav><menu><li><a>...</a><li><a>....</a></menu></nav> is really up to you |
| 02:05 | <aho> | kay :) |
| 02:07 | <aho> | bga_, the closure compiler also helps you to write somewhat more robust js. there is also the option to skip js entirely (e.g. by using GTW, pyjamas, and things like that) |
| 02:08 | <bga_> | i know |
| 02:09 | <bga_> | es4 was very good |
| 02:34 | <brendaneich> | bga_: opinions vary on that |
| 02:35 | <brendaneich> | bga_: js has only a few such "I expected ASI here but i forgot the ASI-club rules" cases -- return is another, yet python (and ruby i think) have that one too |
| 02:35 | <brendaneich> | bga_: being only somewhat line oriented is a hard spot, i agree -- significant newlines all over, or nowhere, seem better |
| 02:35 | <brendaneich> | bga_: CoffeeScript goes hard the significant-newlines way |
| 02:36 | <brendaneich> | harder than python even -- no colon to separate condition from consequent, e.g. (if foo:\n bar() in python is if foo\n bar in coffee) |
| 04:20 | <hober> | Hixie: maybe! I'll let you know after I start. :) |
| 04:51 | <erlehmann> | Hixie, what does “our TF ducks in line” mean in that mail? |
| 05:11 | <Hixie> | erlehmann: beats me |
| 06:00 | <karlcow> | sweet Error500 http://tools.ietf.org/html/draft-ietf-httpstate-cookie |
| 09:19 | <hsivonen> | Hixie: do you happen to recall how many pages you analyzed when you concluded that the parser should break out of foreign lands when seeing a start tag for an old HTML element? |
| 10:29 | <annevk> | http://blog.mozilla.com/rob-sayre/2011/01/04/skating-to-where-the-puck-is-going-to-be/ -- more reasons for having competing browsers |
| 10:29 | <annevk> | at some point I should collect them |
| 10:30 | <annevk> | (though these are somewhat more generic, not really tied to having different rendering engines) |
| 11:17 | <annevk> | hober, btw, anything in particular you'll be working on? |
| 11:29 | <jgraham> | annevk: This is Apple. If he tells you he will have to kill you. And everyone you have ever spoken to. Including himself. Not really worth it. |
| 11:45 | <annevk> | jgraham, so the interwebs? |
| 11:46 | <annevk> | I see their evil plan now :) |
| 11:47 | <annevk> | But I know what plenty of people at Apple are doing... Maybe I should do something about security... |
| 12:11 | <hsivonen> | still hoping that one day Ubuntu gets their QA act together... |
| 12:20 | <MikeSmith> | hsivonen: I think they could if they really wanted to |
| 12:21 | <MikeSmith> | it's just that they don't really want to |
| 12:21 | <MikeSmith> | calculated tradeoff at your expense |
| 12:21 | <jgraham> | hsivonen: Ubuntu has QA? |
| 12:21 | <jgraham> | But, seriously, I imagine doing QA for Ubuntu would be a nightmare |
| 12:22 | <MikeSmith> | heh |
| 12:24 | <jgraham> | I mean there is a roughly infinite amount of hardware to test, you don't control much of the code you are shipping, and it's not really clear how you automate all the things you might want to test |
| 12:25 | <MikeSmith> | add to that the fact that you have lots of users demanding that you get "stable" updates out to them as often as possible |
| 12:26 | <hsivonen> | MikeSmith: I'm slowly digging myself out from under my todo list of unpushed patches. I just pushed a couple of yours to the htmlparser repo. |
| 12:26 | <hsivonen> | (It looks like approvals on Gecko 2.0 patches are slow, so I went ahead and allowed the htmlparser repo get ahead of m-c) |
| 12:26 | <MikeSmith> | ah cool |
| 12:26 | <MikeSmith> | thanks |
| 12:27 | <MikeSmith> | hsivonen: I've goth a couple other patches queued up for your review, but don't want to swamp you |
| 12:28 | <jgraham> | (Microsoft have the advantage of roughly infinite amounts of money to buy hardware, control over most of the relevant code, and multi-year release cycles) |
| 12:29 | <jgraham> | (Apple have the advantage of control over the hardware and software, control over the code, and multi-year release cycles) |
| 12:29 | <jgraham> | (as well as roughly infinite money) |
| 12:29 | <hsivonen> | jgraham: hardware support regressions aren't the only thing I'm unhappy about |
| 12:29 | <hsivonen> | jgraham: they also shipped with broken gconf and leaky nautilus and leaky vinagre |
| 12:30 | <hsivonen> | about on every third logout, my ~/.gconf/ gets corrupted and I have to ssh into the box from my Mac to restore .gconf from a backup |
| 12:31 | <MikeSmith> | wow |
| 12:31 | <hsivonen> | also, it's a good idea to kill vinagre and nautilus at least daily |
| 12:31 | <MikeSmith> | that's pretty bad |
| 12:31 | <jgraham> | hsivonen: Right, that's bad. Although leakiness is one of the harder things to pick up. But my point was that the difficulty level of shipping non-buggy Ubuntu is high |
| 12:31 | <jgraham> | Especially compared to OSX |
| 12:32 | <jgraham> | This is a problem, not an excuse |
| 12:32 | <gsnedders> | Leaky nautilus? I hadn't noticed. |
| 12:32 | <hsivonen> | jgraham: yes and yes |
| 12:32 | <gsnedders> | Admittedly, I rarely use Nautilus. |
| 12:32 | <jgraham> | gsnedders: Nautilus provides the desktop by default |
| 12:32 | <jgraham> | Even if you don't use the file manager |
| 12:33 | <jgraham> | (which I don't) |
| 12:33 | <jgraham> | (my nautilus seems to be well behaved though. Mind you I am on 10.04) |
| 12:33 | <jgraham> | (dunno what vinagre is) |
| 12:34 | <hsivonen> | jgraham: vinagre is the bundled VNC client |
| 12:36 | <annevk> | https://bitbucket.org/ms2ger/ -- why does Web DOM Core have this ?! sign behind it? |
| 12:37 | <MikeSmith> | annevk: that's a logo that was added manually |
| 12:37 | <Philip`> | https://bitbucket-assetroot.s3.amazonaws.com:443/c/photos/2010/Sep/04/logo_avatar.png |
| 12:37 | <Philip`> | Sounds like a logo or avatar |
| 12:37 | <MikeSmith> | yeah, you can change it in the admin settings for the repo |
| 12:38 | <MikeSmith> | hey jgraham btw, over the break I made a bunch of updates to the HTML ES5 spec |
| 12:39 | <MikeSmith> | http://sideshowbarker.github.com/es5-spec/ |
| 12:39 | <MikeSmith> | for one thing, all of the errata are not incorporated into the main body of the document |
| 12:39 | <MikeSmith> | though they remain available as annotations also |
| 12:40 | <MikeSmith> | for anybody who wants to know what changes since edition 5 was first published |
| 12:40 | <MikeSmith> | I also incorporated all the additional changes that were made for edition 5.1 |
| 12:40 | <MikeSmith> | and made those available as annos as well |
| 12:41 | <annevk> | MikeSmith, oh ok |
| 12:41 | <MikeSmith> | jgraham: and I added annos for all the stuff from http://wiki.whatwg.org/wiki/Web_ECMAScript |
| 12:42 | <jgraham> | MikeSmith: Awesome |
| 12:42 | <MikeSmith> | plus I added a whole bunch of internal hyperlinks |
| 12:42 | <hsivonen> | aargh. why is bugzilla.validator.nu so slow? Is someone attacking it again? |
| 12:43 | <MikeSmith> | jgraham: for the hyperlinks, I basically just added hyperlinks for all the stuff that Jason Orendorff has hyperlinked in his http://people.mozilla.org/~jorendorff/es5.html doc |
| 12:43 | <MikeSmith> | (which seems to be 404 now) |
| 12:44 | <hsivonen> | looks like the validator process is eating up CPU... |
| 12:45 | <hsivonen> | no runaway memory behavior though, so I guess I'll let whoever is validating stuff do their thing |
| 12:47 | <jgraham> | MikeSmith: Links seem to be sadly hard to distinguish visually |
| 12:47 | <jgraham> | In Opera at least |
| 12:47 | <MikeSmith> | yeah, I need to restyle them |
| 12:47 | <MikeSmith> | suggestions welcome |
| 12:48 | <MikeSmith> | the problem is that in some algorithms, there are lots of links |
| 12:48 | <MikeSmith> | along with boldfaced stuff |
| 12:48 | <jgraham> | Tyhat's true in HTML5. Just making them blue or whatever (no underline) would work |
| 12:48 | <MikeSmith> | OK |
| 12:48 | <MikeSmith> | hang on, I'll fix it now |
| 12:54 | <MikeSmith> | jgraham: please reload and lemme know if it's an improvement or not |
| 12:57 | <MikeSmith> | hsivonen: I don't remember if I pinged you set about this or not, but the one other parser patch I have is for implementing zcorpan 's request for reporting the names and locations of unclosed elements |
| 12:57 | <jgraham> | MikeSmith: Much better |
| 12:57 | <MikeSmith> | cool |
| 12:58 | <MikeSmith> | jgraham: if you have suggestions for other things / classes of things that should be hyperlinked, lemme know |
| 12:58 | <MikeSmith> | hsivonen: https://bitbucket.org/validator/htmlparser-patches/qseries?apply=t&qs_apply=unclosed-elements |
| 12:58 | <jgraham> | MikeSmith: Random sampling suggests you don't link all instances of e.g. ToInt32 |
| 12:58 | <jgraham> | MikeSmith: http://sideshowbarker.github.com/es5-spec/#x9.4 |
| 12:58 | jgraham | is away now |
| 13:00 | MikeSmith | peruses ToInt32 instances |
| 13:00 | <MikeSmith> | hsivonen: the raw patch is at https://bitbucket.org/validator/htmlparser-patches/raw/8bb1e0fbd733/unclosed-elements |
| 13:01 | <MikeSmith> | (other URL is for the line-numbered/ syntax-highlighted pretty-printed version) |
| 13:01 | <MikeSmith> | hsivonen: and deployed at http://www.w3.org/html/check for testing |
| 13:02 | <hsivonen> | MikeSmith: thanks. taking a look now... |
| 13:12 | <hsivonen> | hmm. I think I'll write my review comments as a patch that applies on top of yours |
| 13:12 | <hsivonen> | there are some tweaks I'm inclined to make |
| 13:13 | <hsivonen> | eventually, it might make sense to try to expose the implied tags to the application through some long-term-supported API |
| 13:14 | <hsivonen> | MikeSmith: NetBeans patches the parser in order to maintain a data structure similar to the one you are adding |
| 13:14 | <MikeSmith> | OK |
| 13:15 | <MikeSmith> | I hope that patch is a decent basis for whatever else you have in mind to add |
| 13:15 | <MikeSmith> | I really wasn't sure if it was the right approach or not |
| 13:16 | <hsivonen> | MikeSmith: my review comments are trivialities like the exact placement of the // [NOCPP[ stuff |
| 13:16 | <MikeSmith> | ok |
| 13:16 | <MikeSmith> | cool |
| 13:22 | <hsivonen> | MikeSmith: do I understand correctly that there isn't a bug number to go with this patch? |
| 13:22 | <hsivonen> | (not to imply that there should be; just checking) |
| 13:23 | <MikeSmith> | hsivonen: there is a bug |
| 13:23 | MikeSmith | looks for the number |
| 13:24 | <karlcow> | hmmm not sure it matters but… opinions? |
| 13:24 | <karlcow> | in adam spec http://tools.ietf.org/html/draft-ietf-httpstate-cookie-20, there are algorithm for |
| 13:24 | <karlcow> | * creating cookies by the server |
| 13:24 | <karlcow> | * storing cookies by the user agent |
| 13:24 | <karlcow> | * sending cookies to the server |
| 13:24 | <karlcow> | * parsing the cooking sent by the server |
| 13:24 | <MikeSmith> | hsivonen: http://bugzilla.validator.nu/show_bug.cgi?id=689 |
| 13:24 | <karlcow> | but no algorithm for what servers should do when parsing the value sent by the client aka what should they do with "Cookie: …" |
| 13:24 | <karlcow> | s/cooking/cookies/ |
| 13:25 | <hsivonen> | MikeSmith: thanks |
| 13:27 | <annevk> | karlcow, servers can do whatever they want |
| 13:27 | <MikeSmith> | karlcow: |
| 13:27 | <MikeSmith> | " The semantics of individual cookies in the Cookie header are not |
| 13:27 | <MikeSmith> | defined by this document. Servers are expected to imbue these |
| 13:27 | <MikeSmith> | cookies with application-specific semantics." |
| 13:27 | <MikeSmith> | http://tools.ietf.org/html/draft-ietf-httpstate-cookie-20#section-4.2.2 |
| 13:28 | <karlcow> | there is a discussion about Cookie.py module and what they should do for parsing it, what is legal or not, etc. |
| 13:29 | <karlcow> | and the different languages libraries are looking at each other |
| 13:29 | <karlcow> | Java will do one thing, ruby another one and python another one. |
| 13:33 | <MikeSmith> | karlcow: what are any of them doing at the library level with the header contents? |
| 13:33 | <MikeSmith> | seems like they should just be treating it as an opaque string |
| 13:33 | <MikeSmith> | and just passing it on as is to whatever app is using the library code |
| 13:34 | <MikeSmith> | abarth deserves some extra credit for getting the word "imbue" into a spec |
| 13:37 | <karlcow> | python's Cookie.py rejects colon, Java servlet allows it, perl allows it, I do not about ruby |
| 13:38 | <annevk> | hsivonen, s/2010/2011/ ? |
| 13:38 | <MikeSmith> | seems like Cookie.py is wrong then |
| 13:38 | <annevk> | hsivonen, in the license file |
| 13:38 | <karlcow> | http://cpansearch.perl.org/src/GAAS/libwww-perl-5.835/lib/HTTP/Cookies.pm |
| 13:38 | <annevk> | why do you need interop between programming libraries? |
| 13:39 | <karlcow> | MikeSmith: it's why I was wondering if it should be documented for people developing libraries |
| 13:39 | <hsivonen> | annevk: oh. right. the patch was made in 2010. Thanks. |
| 13:39 | <karlcow> | these libraries being used for clients and servers |
| 13:41 | <annevk> | happened to see the change as it came by for "WHATWG" on twitter |
| 13:41 | <karlcow> | annevk: because there are used in the design of clients and servers? |
| 13:41 | <karlcow> | (just asking and trying to see if miss something in the reasonning) |
| 13:42 | <annevk> | I think what you miss is when you port an application from one language to another cookie semantics are the least of your troubles |
| 13:43 | <annevk> | if the libraries are used for clients as well obviously that part needs to be changed... |
| 13:43 | <annevk> | if they want to work with other servers anyway |
| 13:43 | <karlcow> | there are always used in both ways. |
| 13:44 | <karlcow> | sometimes by the same application such as a proxy |
| 13:44 | <MikeSmith> | hsivonen: I have http://twitter.com/vnu_changes set up to tweet commit notifications from all repos under https://bitbucket.org/validator |
| 13:44 | <karlcow> | hmmm difficult |
| 13:45 | <MikeSmith> | hsivonen: (which includes a mirror of the htmlparser repo) |
| 13:51 | MikeSmith | steps away for a bit |
| 13:59 | <annevk> | omg |
| 13:59 | <annevk> | hybi :/ |
| 13:59 | <annevk> | I should just unsubscribe |
| 14:00 | <annevk> | "A bit of pragmatism" -- lets ignore your security concerns that are irrelevant in my opinion and move on |
| 14:00 | <annevk> | I have Arrested Development to entertain me in such ways, no need for real live to be like it |
| 14:17 | <hsivonen> | MikeSmith: hmm. wouldn't it make sense to put the locators on the stack nodes instead of introducing these two new lists? |
| 14:17 | <MikeSmith> | yeah |
| 14:18 | <hsivonen> | MikeSmith: OK. I'll write a patch for that on top of your patch |
| 14:18 | <MikeSmith> | thanks |
| 14:18 | <MikeSmith> | btw, I made a minor update to the patch |
| 14:19 | <MikeSmith> | to remove the System.err debugging cruft |
| 14:20 | <hsivonen> | MikeSmith: I had already removed those lines in my follow-up patch ;-) |
| 14:20 | <MikeSmith> | yeah, I figured you probably had |
| 14:21 | <MikeSmith> | I didn't notice them til I re-read through the patch after I pinged you |
| 14:36 | <annevk> | hsivonen, fwiw, just replied to sam ruby |
| 14:53 | <hsivonen> | annevk: do you mean on public-html-xml? |
| 14:56 | <annevk> | yes |
| 14:56 | <annevk> | just in case you were replying |
| 14:56 | <annevk> | but if you have something else to add... |
| 14:58 | <hsivonen> | I'll add one observation |
| 14:59 | <hsivonen> | actually, maybe it's better that I don't |
| 15:08 | <hsivonen> | ah. good old java.lang.ClassCastException: org.xml.sax.InputSource cannot be cast to nu.validator.xml.TypedInputSource |
| 15:13 | <hsivonen> | the tweaked implementation correctly flags the start tag of a formatting element if a clone of that formatting element was unclosed |
| 15:13 | <hsivonen> | I hope users appreaciate that instead of getting confused |
| 15:41 | <annevk> | http://adaptivethemes.com/are-blocks-section-or-div-an-html5-conumdrum -- optional headers? |
| 15:41 | <annevk> | I wonder how that works |
| 15:43 | <Philip`> | "Can you cite where it states that sections must have a heading?" "lets just say that’s my position on HTML5 usage - if your <section> does not have a heading then its highly improbable that its a section" |
| 15:43 | <Philip`> | Sounds like he's just making problems for himself |
| 15:46 | <annevk> | oh right, we changed that |
| 15:48 | <annevk> | jgraham, should we file an issue on HTML5 or WebKit for that markup serialization issue? |
| 15:48 | <annevk> | having thought about it some more I can't really see why any such script would not just use a.href rather than a.innerHTML |
| 15:59 | annevk | wonders if joining #html5 is worth it |
| 16:03 | Workshiva | wonders if that should be ##html5 |
| 16:07 | <aho> | brendaneich, saw your proxies talk the other day. read the slides first, didn't really get it. it made a lot of sense when i finally saw the talk though. proxies are awesome indeed. :) |
| 16:13 | <annevk> | Workshiva, can channels start with a #? |
| 16:16 | <aho> | #javascript redirects to ##javascript for example |
| 16:16 | <annevk> | learned something today |
| 16:17 | <annevk> | I wish I had lots of time to read the IRC archives |
| 16:17 | <annevk> | would make a compilation of quotes from 2010 or so |
| 16:18 | <Philip`> | Ask Mr Last Week to do it, he seems to have plenty of spare time to read the logs |
| 16:19 | <Philip`> | Hmm, now his blog says "Some readers of this blog have contacted Google because they believe this blog's content is objectionable. In general, Google does not review nor do we endorse the content of this or any blog." |
| 16:22 | <annevk> | well, last I checked the stuff he highlights is only borderline relevant to what I would like to extract |
| 17:05 | <Rik`> | http://html5.com weirdly redirects to http://www.apple.com/html5/ |
| 17:05 | <Rik`> | I can't remember if it always did that |
| 17:08 | <annevk> | no, I guess Apple bought it |
| 17:08 | <annevk> | maybe I should have taken it when it was still free |
| 17:09 | <Rik`> | domains owned by apple (i've checked apple.com and itunes.com) seems to have more info in whois data |
| 17:10 | <annevk> | (It was free, then taken by some other party that did not do anything. They offered it for some amount and I was not interested. I guess then this happened.) |
| 17:33 | <Philip`> | Too late to get html6.com already |
| 17:34 | <Philip`> | Looks like html11.com is free, if you want to plan that far ahead |
| 17:38 | <annevk> | I wonder if Microsoft owns htmlnext |
| 17:38 | <annevk> | oh free http://whois.domaintools.com/htmlnext.com |
| 17:48 | <bga_> | Philip` http://html6.by.ru/ :) |
| 17:50 | <annevk> | I would mostly like html.org |
| 17:50 | <annevk> | everything with 5 in the name is just temporary |
| 17:51 | <annevk> | (naming html5lib html5lib was also not very forwardlooking) |
| 17:54 | <Philip`> | html5lib is a misleading name because it doesn't implement support for geolocation or CSS3 at all |
| 18:25 | <hober> | annevk: here's the job description: http://jobs.apple.com/index.ajs?BID=1&method=mExternal.showJob&RID=50510&CurrentPage=1 |
| 18:25 | <hober> | annevk: now you have as much of an idea about what I'll be working on as I do :) |
| 18:33 | <othermaciej> | Safari team still has openings, if anyone wants more where that came from |
| 18:57 | <dglazkov> | .. and so does Chrome team :P |
| 19:14 | dbaron | wishes he didn't miss interesting threads due to their having the subject line "Re: [whatwg] whatwg Digest, Vol 82, Issue 10" |
| 19:42 | <Hixie> | hsivonen: O-of-billions, but it was a long time ago now |
| 19:42 | <Hixie> | hsivonen: though I still have a framed printout of one of the pages I found stuck to my fridge |
| 19:43 | <Hixie> | what's more interesting than the number of pages scanned is the fraction of pages that had issues |
| 19:43 | <Hixie> | iirc the number was small, but non-zero |
| 19:44 | <Hixie> | the list of elements in the spec has html comments in the source listing which elements were found to be problematic and which were added just for completeness and to make writing parsers easier |
| 19:44 | <Hixie> | (e.g. iirc h1 was a problem but h6 probably wasn't, but someone asked that they be treated the same to make parsers simpler) |
| 19:55 | <AryehGregor> | How can I throw a DOMException from JavaScript code? |
| 19:56 | <Hixie> | i don't think you car |
| 19:56 | <Hixie> | can |
| 19:56 | <AryehGregor> | Hmm, okay. |
| 19:56 | <Hixie> | ask ms2ger to give you a constructor on DOMException :-) |
| 20:10 | <AryehGregor> | Yay, atob() has totally different error handling in different browsers. |
| 20:19 | <Hixie> | figures |
| 20:20 | <Hixie> | as a general rule, I tend to prefer majority implementations, biased slightly towards market share, and biased slightly against sillyness |
| 20:20 | <Hixie> | if that helps |
| 20:23 | <karlcow> | not sure to understand http://www.w3.org/html/wg/tracker/issues/149 for the examples. Is the change proposal to write all the examples, or to say that the examples have to be changed. |
| 20:42 | <Hixie> | benschwarz: btw, while you're doing the developers.whatwg.org stuff, if you come across anything in the spec that is mis-labeled (class=impl when it shouldn't be, or vice versa, or something that just plain is more confusing in the author case for some reason), then don't hesitate to let me know so i can fix it |
| 20:47 | <AryehGregor> | So, the breakdown is as follows: 1) Opera is super-lenient, and never throws. 2) Firefox is very strict, and atob() throws for most if not all things that btoa() can't produce (e.g., wrong number of trailing =). 3) WebKit is moderately strict, and throws only if the atob() input contains an invalid character or has an = except in a run at the end or has length 1 mod 4 after stripping trailing =. 4) IE9 is completely insane and I have no id |
| 20:47 | <AryehGregor> | ea what it's doing, my entire test suite now fails (mostly with the error "'atob'/'btoa' is undefined"). |
| 20:47 | <AryehGregor> | Opinions? |
| 20:48 | <AryehGregor> | WebKit is the least intuitive, since it tolerates some errors but not others. I'm not sure whether strict errors are good or bad here, so I guess if it's a toss-up between Firefox and Opera, Firefox wins by market share. |
| 20:48 | <AryehGregor> | But I'm generally inclined to think that it's nicer to be tolerant. |
| 20:49 | <AryehGregor> | On a side note, IE9 thinks btoa("") is invalid, and it's not totally clear that's wrong from the RFC (since that's vague), so I'm leaning toward writing my own btoa() spec text. |
| 20:53 | <AryehGregor> | WebKit also doesn't like strings that consist only of ='s. |
| 20:53 | <AryehGregor> | I think I'll go with Firefox for now. |
| 20:53 | <gsnedders> | AryehGregor: Never throwing causes site-compat issues, FYI |
| 20:53 | <AryehGregor> | Ah, okay. |
| 20:53 | <AryehGregor> | That settles it. |
| 20:57 | <AryehGregor> | Actually, Firefox doesn't throw errors for everything you might expect. |
| 20:57 | <AryehGregor> | So I think I'll go with WebKit, since it throws in fewer cases and therefore is simpler to spec and implement. |
| 20:57 | gsnedders | looks in a, "do I want to know?" sort of way |
| 20:59 | <jgraham> | gsnedders: I thought you knew everything |
| 21:00 | <AryehGregor> | But maybe WebKit isn't web-compatible either . . . hmm. |
| 21:00 | AryehGregor | looks in their Bugzilla |
| 21:02 | <AryehGregor> | Hixie, any suggestions? I think WebKit's behavior is simpler to spec, but perhaps it's less site-compatible. |
| 21:03 | <Hixie> | AryehGregor: hard to say |
| 21:03 | <Hixie> | AryehGregor: any idea what IE is doing? |
| 21:03 | <Hixie> | (i feel your pain, when it comes to debugging IE) |
| 21:03 | <AryehGregor> | Hixie, I was trying to test IE9 before, but now something I did in my test suite causes it to fail all tests with cryptic error messages. |
| 21:03 | <AryehGregor> | I'll try doing more limited debugging and see if I can work out what it does, though. |
| 21:04 | <Hixie> | weird |
| 21:04 | <Hixie> | also try IE8, maybe it's just an IE9 beta bug |
| 21:04 | <AryehGregor> | The test framework doesn't even work in IE8. |
| 21:04 | <Hixie> | that's problematic :-) |
| 21:04 | <AryehGregor> | It's possible that the actual IE bug is in ES5 features or something, since I don't really know what parts of JavaScript are new or whatever. |
| 21:05 | <Hixie> | from your descriptions, it sounds like webkit is the sanest, but it also sounds like you might have some room to try to come up with something more logical than anyone currently implements |
| 21:05 | <AryehGregor> | I don't think we need to care so much about IE, to be honest . . . it's usually in different code paths and it's already committed to breaking compat on every release with compatibility mode. |
| 21:05 | <Hixie> | you'd be surprised how many pages just assume IE and don't have other code paths |
| 21:06 | <Hixie> | but that's more about IE6 or 7 than IE9 |
| 21:06 | <Hixie> | remember that the goal is for the spec to be compatible with legacy content, not legacy UAs |
| 21:10 | <AryehGregor> | Okay, IE9 is giving "'btoa' is undefined" for no discernible reason. |
| 21:10 | <aho> | lameness |
| 21:11 | <aho> | that's the reason :v |
| 21:11 | <Hixie> | gotta go for lunch, but i can try to help out when i get back |
| 21:14 | <AryehGregor> | In compat mode I get "Object expected" instead. |
| 21:14 | <AryehGregor> | I know btoa() exists in IE, I was using it before . . . |
| 21:14 | AryehGregor | will return to this in an hour or two |
| 21:16 | <AryehGregor> | Does Microsoft have any official documentation for their JavaScript function implementation . . . ? |
| 21:17 | <AryehGregor> | This is just slightly incomplete: http://msdn.microsoft.com/en-us/library/6fw3zxcx(v=VS.94).aspx |
| 21:20 | <webr3> | is the paste event standardized/ing ? |
| 21:48 | <Hixie> | webr3: not yet, but we really should |
| 21:51 | <webr3> | hixie, best place to nudge re paste event (and aligning w/ drag and drop, and aligning TextInput event which seems to differ, and correlations to input event) |
| 21:51 | <Hixie> | nowhere useful to nudge, we don't have an editor for it |
| 21:51 | <webr3> | for events 3? |
| 21:51 | <Hixie> | for copy/paste |
| 21:52 | <Hixie> | imho dom events should just define the events model, not the actual events |
| 21:52 | <Hixie> | almost every event it tries to define it gets wrong somehow |
| 21:52 | <Hixie> | or duplicates requirements elsewhere, or doesn't specify in enough detail for interop, etc |
| 21:53 | <webr3> | have to agree there, either the place for all events, or just the model, but if all it'd have to be more of a registry.. would be god to have a central place to look them all up and ensure they were all aligned when being defined though |
| 21:54 | <Hixie> | events can't sanely be specced in a registry |
| 21:54 | <Hixie> | events are like exceptions |
| 21:54 | <Hixie> | how they fire is deeply integrated with the things they are talking about |
| 21:54 | <Hixie> | the place to spec them is in the processing model for the relevant feature |
| 21:54 | <Hixie> | doesn't make sense to spec them separately |
| 21:54 | <webr3> | at some level surely they can, for instance ensuring certain properties existed on all events of type input etc |
| 21:55 | <webr3> | it'll be even more of a pita if every single event has a different interface for the properties |
| 21:55 | <Hixie> | the interfaces can be specced wherever, those are almost trivial |
| 21:55 | <Hixie> | i'm talking about the rules for when the events fire |
| 21:57 | <webr3> | gotcha, i was talking more about the methods and properties on certain classes of event and ensuring they were aligned, .data vs .getData vs .clipboardData.getData(), with types without and so forth |
| 21:58 | <Hixie> | getting them all aligned is mostly a lost cause, but yeah, for the simpler interfaces i'm fine with dom events speccing them -- there's not much to spec, the attributes are readonly and just return what the initialisers were given and there are no methods |
| 21:58 | <Hixie> | other than the initialisers |
| 22:00 | <AryehGregor> | Can anyone please explain to me why http://aryeh.name/tmp/test.html in IE9 doesn't work? |
| 22:01 | <Hixie> | put a letter before the <script> to make sure it ends up in the body |
| 22:01 | <AryehGregor> | It won't work in the head? |
| 22:01 | <Hixie> | dunno, but try |
| 22:01 | <AryehGregor> | No difference. |
| 22:01 | <AryehGregor> | For those who don't want to view source, it was: <!doctype html><script>alert(btoa("test"));</script> |
| 22:01 | <AryehGregor> | Console gives error "'btoa' is undefined". |
| 22:02 | <Hixie> | what does alert(window.btoa) give ? |
| 22:02 | <AryehGregor> | undefined . . . |
| 22:02 | <AryehGregor> | Surely IE can't just not implement these? Some of the tests passed before. |
| 22:02 | <Hixie> | try removing the doctype? |
| 22:02 | <AryehGregor> | Still undefined. |
| 22:03 | <webr3> | i didn't even know btoa and atob were standardized :| |
| 22:03 | <AryehGregor> | They aren't, that's what I'm doing. |
| 22:03 | <webr3> | lol, that explains it |
| 22:04 | <AryehGregor> | Okay, it really is behaving as though btoa() and atob() aren't defined. |
| 22:04 | <AryehGregor> | But I could swear it wasn't before . . . |
| 22:04 | <Hixie> | the web suggests it's not supported in IE |
| 22:07 | <AryehGregor> | Well, that explains why I was so confused. |
| 22:08 | <AryehGregor> | So I guess I'll spec what Firefox does. |
| 22:13 | <Hixie> | seems reasonable |
| 22:17 | <annevk> | I was gonna sleep, but then I read the logs... |
| 22:17 | <annevk> | hallvors is speccing copy/paste |
| 22:17 | <Hixie> | oh, cool |
| 22:17 | <annevk> | I think he already posted on public-webapps, will soonish be an official draft |
| 22:18 | <Hixie> | let's add it to http://wiki.whatwg.org/wiki/Specifications_that_apply_to_Web_browsers |
| 22:20 | <webr3> | annevk, cool ty for answer |
| 22:20 | <annevk> | http://thedailywtf.com/Articles/XMLd-XML.aspx o_O |
| 22:21 | <Hixie> | annevk: is there a link to hallvors' draft spec? |
| 22:21 | <annevk> | Hixie, did you see http://wiki.whatwg.org/wiki/Specs ? |
| 22:21 | <annevk> | Hixie, it's not checked in yet I think |
| 22:21 | <Hixie> | annevk: yeah, all that information is on the other page now. Saw Specs too late, sorry. |
| 22:21 | <annevk> | http://lists.w3.org/Archives/Public/public-webapps/2010OctDec/att-1067/cbapispec.html is the public one |
| 22:21 | <annevk> | I mean latest published in public |
| 22:21 | <Hixie> | why doesn't he edit it live |
| 22:21 | <annevk> | he probably does |
| 22:22 | <Hixie> | stop working on secret specs people, edit it all liiiiveee :-P |
| 22:22 | <annevk> | but I don't think he's a W3C account and all yet |
| 22:22 | <annevk> | or a personal website |
| 22:22 | <annevk> | but I don't know for sure |
| 22:22 | <Hixie> | i guess i should make /Specs redirect to the other page now |
| 22:22 | <annevk> | wfm |
| 22:23 | <annevk> | though yours has a long name... |
| 22:23 | <Hixie> | well if it's a redirect the other one will work fine too :-) |
| 22:23 | <Hixie> | there |
| 22:23 | <Hixie> | http://wiki.whatwg.org/wiki/Specs |
| 22:24 | <annevk> | I like how it has links to discussion / bugs / etc. |
| 22:24 | <Hixie> | feedback and bugs are the two links i actually wanted that page for |
| 22:24 | <Hixie> | feedback, mainly |
| 22:25 | <annevk> | I created Specs mostly because I lost track of where the spec was hosted |
| 22:25 | <annevk> | but the formatting was way complex |
| 22:25 | <annevk> | well, wiki editing is way complex |
| 22:26 | <Hixie> | yeah, wiki tables are a pain |
| 22:28 | <Hixie> | we should clean up http://wiki.whatwg.org/wiki/Companion_specifications to make sure the stuff with editors is on the /Specs page now and remove it from the Companion page |
| 23:20 | <_bga> | http://visual6502.org/JSSim/index.html |
| 23:26 | <MikeSmith> | hsivonen: I remember now that the reason I implemented the unclosed-element reporting using additional lists instead of just putting the locators on the stack nodes |
| 23:27 | <MikeSmith> | the reason was, to have a way to essentially keep state about which unclosed elements had already been reported |
| 23:28 | <MikeSmith> | because without that, many or most unclosed element instances cause duplicate error messages to be emitted |
| 23:39 | <benschwarz> | Hixie: Roger. |
| 23:52 | <Evet> | is it possible to build RIAs for ie 5.5+? |
