| 00:26 | <volkmar> | Hixie: about the input type change algorithm (for value management) |
| 00:27 | <volkmar> | it looks like we loose the value with type='file' |
| 00:27 | <volkmar> | if input.type='text', .value='foo', .type='file', .value would be "" |
| 00:27 | <volkmar> | because we don't save the value to the content attribute when switching to a file type |
| 00:28 | <volkmar> | why not doing that? |
| 00:29 | <tabatkins> | volkmar: Because you could just script-fill the value of the type=text with a filepath, then change it to type=file. If the input is visually hidden, it can be smuggled inside another form, so the user ends up accidentally submitting files from their hard drive they didn't know about. |
| 00:29 | <Hixie> | the <input type=file> always ignored the /value/ |
| 00:30 | <volkmar> | TabAtkins: hmmm, not what i meant |
| 00:30 | <Hixie> | volkmar: .value would still be "" even if we saved the /value/ to the value="" content attribute |
| 00:30 | <Hixie> | volkmar: because .value for type=file has nothing to do with either the /value/ concept or the value="" attribute |
| 00:30 | <volkmar> | Hixie: yes, the problem is if i go back to <input type='text'> i loose the previous value |
| 00:30 | <Hixie> | so? |
| 00:31 | <volkmar> | why not saving the value to the content attribute when we change teh type to 'file' considering 'file' shouldn't use it but it will be re-used if the type is changed to something else |
| 00:31 | <Hixie> | why bother? |
| 00:32 | <Hixie> | what's the use case? |
| 00:32 | <Hixie> | why does it matter? |
| 00:32 | <volkmar> | Hixie: it doesn't really matter, just sound weird than it's required for default and default_on and not filename |
| 00:33 | Hixie | points to the topic |
| 00:33 | <Hixie> | if that's the only thing that sounds weird about the forms model in HTML, you haven't looked closely :-P |
| 00:33 | <Hixie> | personally I think what's weird is that we have to change things at all |
| 00:33 | <Hixie> | not that it gets reset to "" |
| 00:34 | <Hixie> | unfortunately if we don't reset things, it breaks backcompat |
| 00:34 | <volkmar> | yeah... |
| 00:34 | <Hixie> | anyway if you think it should change, file a bug explaining why |
| 00:34 | <Hixie> | i don't make changes based on irc discussions :-) |
| 00:34 | <Hixie> | at least not while i'm making other changes :-) |
| 00:35 | <volkmar> | Hixie: i don't think it _should_ be changed, just wondering if there were a reason |
| 00:35 | <volkmar> | i don't really care actually |
| 00:35 | <Hixie> | no reason particularly |
| 00:35 | <Hixie> | it just didn't seem to matter for compat |
| 00:35 | <Hixie> | bz and i looked at what seemed to matter and just changed that |
| 00:35 | <volkmar> | fine then |
| 09:15 | <annevk> | really: http://html5.org/tools/web-apps-tracker?from=5413&to=5414 :/ |
| 09:15 | <annevk> | charset overrides is one of the worst ideas anyone ever came up with |
| 09:15 | <annevk> | and has given us security issues (and is still giving them to anyone with UTF-7 support) |
| 10:17 | <Philip`> | If someone wants to argue about completeness of rel registries, maybe they should look at which of the common values in http://philip.html5.org/data/rel-rev-200904.txt are in each registry |
| 10:18 | <Philip`> | I expect someone who's using the registry to either understand a random web document or avoid collisions when choosing their own names will care much more about rel="edituri" than about rel="child" |
| 10:21 | <espadrine> | What are those french 'rel's at the end of the page? |
| 10:27 | <Philip`> | espadrine: People mixing up rel with src, I expect |
| 10:31 | <Philip`> | espadrine: http://www.tourdumonde.over-blog.com/photo-257570-saopaulo-placadese-cathedrale_jpg.html |
| 10:31 | <Philip`> | <img src="http://img.over-blog.com/630x470-000000/0/44/26/71//09-bresil/saopaulo-placadese-cathedrale.jpg"; alt="saopaulo-placadese-cathedrale.jpg" class="played" rel="0/44/26/71//09-bresil/saopaulo-placadese-cathedrale.jpg" height="470" width="630" /> |
| 11:00 | <Hixie> | annevk: i'm open to better ideas |
| 11:01 | <annevk> | allow the HTTP charset parameter? |
| 11:01 | <annevk> | I guess that is not very author friendly |
| 11:02 | <annevk> | though how about marking this as a v2 feature? |
| 11:02 | <annevk> | we'll deploy <track>, see how it works out in the first year and if enough problems occur we go for it |
| 11:03 | <annevk> | cause people will likely need to move SRT files around anyway and maybe making the adjustments is easy enough for them |
| 11:03 | <Hixie> | if the idea is to be compatible with legacy SRT, MIME types and UTF-8 only are both failures |
| 11:03 | <Hixie> | if it's not, we don't need the override, indeed |
| 11:04 | <Hixie> | people seemed pretty much in favour of the former though (haven't yet sent my reply to that thread, working on it still) |
| 11:05 | <annevk> | well yeah, it's tricky -- i want to be compatible too but I also do not want to introduce more features such as charset |
| 11:06 | <Hixie> | if you can think of an attack vector, mail the list and i'll remove it |
| 11:06 | <Hixie> | security trumps compatibility |
| 11:06 | <Hixie> | bed time now |
| 11:06 | <Hixie> | nn |
| 12:58 | <annevk> | Ms2ger, yo, where is PARSE_ERR from? |
| 12:58 | <annevk> | is that DOM L&S? if so, isn't that obsolete? |
| 12:59 | <Ms2ger> | Yes/Dunno |
| 12:59 | <annevk> | k, I think it should be |
| 12:59 | <annevk> | but maybe something else uses PARSE_ERR too... hmm |
| 13:01 | <Dashiva> | My google only finds L&S, and the reference in HTML5 |
| 13:01 | <annevk> | HTML5 mentions them? |
| 13:02 | <annevk> | figures |
| 13:02 | <annevk> | I believe the WHATWG wiki does too |
| 13:02 | <annevk> | guess we just tried to be complete |
| 13:02 | <Ms2ger> | I'm blaming you, annevk |
| 13:02 | <annevk> | Ms2ger, what is this "tempTests" in DOMImpl...createDocument..exceptions? rewriting that test now |
| 13:02 | <Ms2ger> | http://wiki.whatwg.org/index.php?title=Exception_Codes&diff=next&oldid=2625 |
| 13:02 | <annevk> | Ms2ger, win |
| 13:03 | <annevk> | i'll annotate the wiki with "should be obsolete" |
| 13:04 | <Ms2ger> | Does any ua have them? |
| 13:04 | <annevk> | opera might |
| 13:04 | <annevk> | but i'm campaigning for nuking our (i believe somewhat broken) dom ls support |
| 13:04 | <annevk> | ok, breakfast time o_O |
| 13:04 | <annevk> | back later |
| 13:05 | <Ms2ger> | annevk, ask gsnedders :) |
| 13:32 | <annevk> | Ms2ger, guess i should use hg log more often |
| 13:32 | <annevk> | :) |
| 13:34 | <Ms2ger> | annevk, and all my tests have link rel=author :) |
| 13:37 | <annevk> | aah so i'm adding the wrong credits here |
| 13:39 | <Ms2ger> | Though I think I added a few cases recently |
| 14:06 | <llrcombs> | if a page has a .manifest file, shouldn't the UA check for a new version of each content file? |
| 14:16 | <llrcombs> | (just see if it's changed, not reload the whole file unless it hasn't changed) |
| 14:16 | <llrcombs> | amirite? |
| 16:42 | <llrcombs> | hello? |
| 17:04 | <Dashiva> | llrcombs: Better to just say or ask what you came for right away. People will get back to you :) |
| 17:05 | <Philip`> | It looks like he did ask, 3 hours ago, and nobody got back yet :-p |
| 17:07 | <Dashiva> | I just did |
| 17:13 | <llrcombs> | if a page has a .manifest file, should the UA check for a new version of each content file on load? |
| 17:24 | <llrcombs> | Dashiva; getting back? |
| 17:27 | <Dashiva> | I'm not well versed in offline, but my understanding is that content files are only checked when the manifest changes (and not necessarily then either) |
| 17:28 | <Dashiva> | You shouldn't take my word for it, though |
| 18:06 | <llrcombs> | hmmm |
| 18:07 | <llrcombs> | not much provision for when there's a new version of the file then, eh? |
| 18:38 | <annevk> | hmm, I screwed up all the help links in tests |
| 18:38 | <annevk> | but I don't really believe in those anyway |
| 18:39 | <annevk> | or in author for that matter... |
| 18:39 | <annevk> | Ms2ger, any particular reason we are using those fields? |
| 18:40 | <Ms2ger> | Blame the CSSWG |
| 18:40 | <Ms2ger> | You can get rid of them if you want |
| 18:41 | <annevk> | cool |
| 18:43 | <annevk> | not quite sure if rewriting the tests is the most efficient approach rather than just starting over but I guess overall it's not too much work |
| 18:43 | <annevk> | and it's almost done |
| 18:47 | <annevk> | jgraham, you around? how do you test for ECMAScript errors using your framework? |
| 18:47 | <annevk> | e.g. error instanceof TypeError |
| 18:48 | <annevk> | i.e. how to rewrite this code to use assert_throws: |
| 18:48 | <annevk> | try { |
| 18:48 | <annevk> | document.body.removeChild({'a':'b'}) |
| 18:48 | <annevk> | assert_unreached() |
| 18:48 | <annevk> | } catch (e) { |
| 18:49 | <Ms2ger> | assert_true(e instancof TypeError) |
| 18:49 | <Ms2ger> | } |
| 18:52 | <annevk> | ? |
| 18:52 | <annevk> | it's instanceof afaict |
| 18:53 | <Ms2ger> | It is |
| 19:02 | <annevk> | hmm |
| 19:02 | <annevk> | document.body.removeChild(null) makes the framework collapse? |
| 19:02 | <annevk> | i wonder if i do something wrong |
| 19:03 | <annevk> | ooh, I guess it's a bug in the framework |
| 19:03 | <annevk> | assert_throws is continuously buggy |
| 20:00 | <annevk> | yay |
| 20:00 | <annevk> | all existing stuff is fixored |
| 20:00 | <annevk> | now all the new stuff needs doing |
| 20:02 | <llrcombs> | if a page has a .manifest file, should the UA check for a new version of each content file on load? If not, when should it? |
| 20:03 | <annevk> | only when the manifest changes |
| 20:04 | <llrcombs> | so how would I designate that a file had been updated? |
| 20:04 | <llrcombs> | (i.e. a .js file with a new version) |
| 20:04 | <annevk> | you change the manifest |
| 20:04 | <annevk> | e.g. change a comment or do something like foo.js?v=2 |
| 20:04 | <llrcombs> | or add whitespace? |
| 20:05 | <annevk> | could work |
| 20:05 | <llrcombs> | seems flawed |
| 20:05 | <annevk> | we can make it more complicated later |
| 20:05 | <llrcombs> | suggestion: allow a .manifest to include last-modified datetimes for each file |
| 20:05 | <annevk> | that would be complicated |
| 20:05 | <llrcombs> | so the UA can figure out if it should grab a new rev |
| 20:06 | <llrcombs> | until then, I can't use manifests for my webapps/widgets/etc |
| 20:06 | <annevk> | why? |
| 20:06 | <annevk> | you can just print the timestamp as comment, no? |
| 20:07 | <llrcombs> | yeah, but then it'd grab a new version of every file for every load |
| 20:07 | <annevk> | the server can return 304 like it can normally |
| 20:07 | <llrcombs> | defeats the purpose |
| 20:07 | <annevk> | if your file changes all the time you should prolly not put it in the manifest |
| 20:08 | <llrcombs> | no, I'm debugging |
| 20:08 | <llrcombs> | and I'll end up updating it to make new versions |
| 20:11 | <llrcombs> | you know, mantinence, new features... |
| 20:11 | <llrcombs> | bugfixes |
| 20:12 | <annevk> | sure |
| 20:12 | <annevk> | so add manifest when most of the other stuff is done |
| 20:12 | <annevk> | I don't really see the problem here |
| 20:13 | <llrcombs> | here's my idea: |
| 20:14 | <llrcombs> | the manifest file could be generated by a script, or just manually edited, so it include last-modified dates for each file in it |
| 20:14 | <llrcombs> | so when offline, the UA uses its cached versions of everything |
| 20:14 | <llrcombs> | and when online, it just checks the manifest against its stored manifest |
| 20:14 | <llrcombs> | to see if there's a new version of anything |
| 20:14 | <annevk> | yeah, I just told you you can do that |
| 20:14 | <llrcombs> | if there is, it just downloads THAT file |
| 20:15 | <llrcombs> | otherwise, it keeps using its cached version |
| 20:15 | <annevk> | it doesn't download the whole file anyway |
| 20:15 | <llrcombs> | the whole cached entity, or the whole .manifest? |
| 20:15 | <annevk> | the former |
| 20:16 | <annevk> | it just does a request and the server can say 304 |
| 20:16 | <gsnedders> | For all those people nagging me about Anolis while I've been away, the following statement is true: 10:40 #whatwg: < jgraham> annevk: gsnedders has explicitly lost interest in anolis |
| 20:16 | <annevk> | SPDY will prolly optimize away the remaining bits |
| 20:16 | <llrcombs> | :D @ SPDY |
| 20:16 | <llrcombs> | but no one is implementing... |
| 20:17 | <llrcombs> | so it makes a request for every file, even those in the cache manifest? |
| 20:17 | <annevk> | yawn |
| 20:17 | <Ms2ger> | gsnedders, are you still going to put anolis2 somewhere? |
| 20:17 | <llrcombs> | sorry |
| 20:17 | <llrcombs> | I guess I'm not a caching expert |
| 20:17 | <gsnedders> | Ms2ger: Let me do that riht now |
| 20:20 | <gsnedders> | Ms2ger: http://hg.gsnedders.com/anolis2/ |
| 20:20 | gsnedders | has no idea whether it even works at the moment |
| 20:22 | Ms2ger | wonders what prompted "I <3 namespaces." |
| 20:22 | <gsnedders> | Ms2ger: See the diff |
| 20:55 | <jgraham> | Oh anne left |
| 20:55 | <jgraham> | dammit |
| 20:56 | <jgraham> | for the record assert_throws(new TypeError(), document.body.removeChild({'a':'b'})) |
| 20:56 | <jgraham> | should work |
| 20:58 | jgraham | wonders what the bug is |
| 20:59 | <jgraham> | I guess I could investigate |
| 20:59 | <jgraham> | But I should probably do the thing that Sam wanted first |
| 21:00 | <Philip`> | Surely that won't work since it'll call removeChild before calling assert_throws? |
| 21:01 | Ms2ger | assumes function() { document.body.removeChild({'a':'b'}) } |
| 21:02 | <Philip`> | Ah, that would seem more sensible and less like magic |
| 21:02 | <jgraham> | Er, yeah right |
| 21:02 | <jgraham> | Lots of function() {} wrappers needed everywhere |
| 21:03 | jgraham | wishes javascript had a Haskell-like syntax for anonymous functions |
| 21:04 | <jgraham> | It would make this kind of thing much less visually noisy |
| 21:09 | <gsnedders> | jgraham: Are you starting to admit Haskell is a good language? :P |
| 21:28 | <jgraham> | gsnedders: WTF? |
| 21:28 | <jgraham> | I have *nver* said Haskell was a bad language |
| 21:28 | <jgraham> | **never* |
| 21:28 | <jgraham> | I sai it has a steep learning curve |
| 21:28 | <jgraham> | +d |
| 21:35 | <gsnedders> | You've said it in such a way as to imply it is bad as a result |
| 21:48 | <jgraham> | Only if I was being severely misunderstood |
| 21:48 | <jgraham> | Which is possible if I wasn't being clear |
| 21:48 | <jgraham> | (being hard to learn is a bad property. It is not necessarily enough to create a bad whole) |
| 21:52 | <gsnedders> | jgraham: what happened about the t-shirt, btw? |
| 21:56 | <jgraham> | gsnedders: Not sure |
| 21:57 | <gsnedders> | jgraham: Feel free to just keep it till New Year, fwiw |
| 21:57 | <jgraham> | gsnedders: It wouldn't fit |
| 21:57 | <jgraham> | :) |
| 21:58 | <gsnedders> | ;P |
| 23:51 | <annevk> | jgraham, thanks, will try |
![http://img.over-blog.com/630x470-000000/0/44/26/71//09-bresil/saopaulo-placadese-cathedrale.jpg"](http://img.over-blog.com/630x470-000000/0/44/26/71//09-bresil/saopaulo-placadese-cathedrale.jpg"){kind=link}