should be renamed [15:31:03.0000] People misuse everything. [15:32:00.0000] There are lots of web pages on which it's semantically appropriate to use

for what you're calling the main content. There are many web sites where that's inappropriate from a semantic standpoint. It just depends on the page. [15:32:01.0000] and this is why the web is a toilet to dev for; lazy spec developers [15:32:02.0000] let me ask this: do you expect large corporations to wrap their page specific content in

? [15:33:00.0000] no. I expect them to wrap articles in

. [15:33:01.0000] if they would syndicate it as an independent item of content in a feed - then yes, I would expect them to wrap it in an

[15:33:02.0000] corporate web sites are not intending to distribute their page content sans context. you and i both know this. [15:33:03.0000] tantek: And that's equivalent to what hober said. ^_^ [15:33:04.0000] do you guys not see the gap i'm shining light on? [15:33:05.0000] key: if we rename

to , people will think it's for their main content, when it is in fact for syndication. [15:33:06.0000] key: we see that there's currently no eelemtn for "the main section of the page" [15:33:07.0000] the syndication attribute of article needs to be removed, leaving a generic [15:33:08.0000] TabAtkins - no hober said articles in

- that's a tautology :P [15:33:09.0000] but that has nothing to do with no, would be for any unit of content [15:34:01.0000] whole and complete [15:34:02.0000] key - that's what

is for [15:34:03.0000] we already have a generic "main content of page" element, it's [15:34:04.0000] heh [15:34:05.0000] Hixie: indeed, I misspoke [15:34:06.0000] yea, the spec sure is tight and clean. [15:34:07.0000] tantek: Only if you claim that the "article-ness" of a piece of content is defined solely through what it's wrapped in. [15:34:08.0000] and now you understand the difference between

and

[15:34:09.0000] hixie, there is a difference just as there is HEAD and HEADER [15:34:10.0000] tantek: If you assume that article-ness if an independent quality, then it makes sense. [15:35:00.0000] congratulations - quick, write it up in an FAQ :) [15:35:01.0000] 1 is for message, the other is semantic [15:35:02.0000] tantek: it's probably already in there [15:35:03.0000] is a historical artefact,

is to help authors not have to use

for a common class="header" case [15:35:04.0000] hober - if so, point to it :P [15:35:05.0000] /me is not particularly fond of either [15:36:00.0000] hixie, there is the structure of the message, then there is the semantic partitioning of the content (body) of the message [15:36:01.0000] uh [15:36:02.0000] you guys should /not/ be developing specs for the web [15:36:03.0000] you are welcome to do it in our stead :-) [15:36:04.0000] key - you're welcome to fork and do a better job :P [15:36:05.0000] lol [15:36:06.0000] i'm actually working on it [15:36:07.0000] the spec is under a license that allows forking for exactly this reason [15:37:00.0000] all you have to do is convince the browser vendors that you're doing a better job, and your spec will be the official one :-) [15:37:01.0000] or at least, as official as the whatwg's one today [15:37:02.0000] i'm sorry you guys want to squeeze everything web into the character of a blog. [15:37:03.0000] key - no need to strawman - https://secure.wikimedia.org/wikipedia/en/wiki/Straw_man [15:38:00.0000] i know what strawman means, and i did not ply that scheme [15:38:01.0000] tantek: looks like it's not in there. Hence my 'probably' :) [15:38:02.0000] having a tag so semantically rigid as article is in fact a poor choice [15:39:00.0000] Hey, it sure beats

[15:39:01.0000] This is why open-source licenses are neat. They give you a way to tell people to shut up while at the same time feeling smug and self-righteous about it. [15:39:02.0000] i like most of the idea behind it, but it shouldn't imply anything re syndication! [15:39:03.0000] HTML has lots of weird elements,

is on the less weird end of things [15:39:04.0000] so we lower our standards vs increasing them hober? [15:39:05.0000] i mean, wtf does that comment even mean? [15:39:06.0000] key: then just think of it being an independent section, and forget anyone said anything about syndication [15:39:07.0000] key: see the topic [15:39:08.0000] key, the fact is, you're not the one deciding what goes into the WHATWG's HTML spec. So you can either try to convince people, or agree to disagree. Just telling us we're wrong is not really very useful. [15:40:00.0000] key - your statement "you guys want to squeeze everything web into the character of a blog" is a misrepresentation of our positions [15:40:01.0000] hober, but you damn well know the 'syndication' reference will give cover to people who want to scrape and rape ppl's content [15:40:02.0000] key: the idea behind

is exclusively to allow syndication, so if you don't like that, you don't like the idea behind

. Which is fine, but you'll find it easier to convince us if you start from the standpoint of what you actually want than saying you agree to something you actually disagree with. [15:40:03.0000] wait, what? [15:40:04.0000] ergo strawman [15:41:00.0000] I'd call it more like a "flame". [15:41:01.0000] A straw man is really more when you *counter* an argument that your opponent doesn't hold. [15:41:02.0000] nah. moment, let me ponder [15:41:03.0000] key: Your entire conversation so far has been a massive pile of insult, distortion, and flame. This is not how you conduct conversations. [15:41:04.0000] Not when you merely accuse them of holding it. [15:41:05.0000] AryehGregor: yep you got it [15:41:06.0000] The whole point of the analogy is that straw men are easy to defeat, because they don't fight back. [15:41:07.0000] At least, not if you want to actually accomplish anything besides getting ignored. [15:42:00.0000] wait what? now we're condoning rape? [15:42:01.0000] If they aren't trying to counter the argument, it's not really appropriate to call it a straw man, IMO. [15:42:02.0000] . . . ? [15:42:03.0000] man, this conversation went sour fast [15:42:04.0000] relax [15:42:05.0000] I just happened to look here and we're talking about rape? [15:42:06.0000] there seems to be 2 sides to article [15:42:07.0000] what kind of logic is... oh [15:42:08.0000] 1. syndication being implied. 2. unit of content; whole and complete [15:42:09.0000] i like 2, i think 1 is less useful [15:43:00.0000] ok, so just don't worry about 1 then [15:43:01.0000] how does it affect you? [15:43:02.0000] would be semantically organizational, whereas

would be structurally organizational [15:43:03.0000] because we need a tag like it, semantically, but better [15:43:04.0000] hence [15:43:05.0000] the semantic of what you think of as is different from

, and this is orthogonal to the syndication thing [15:43:06.0000] if it's about syndication,

should inherit from , not

[15:44:00.0000] no, because it's sensible for a page to have many

s on it [15:44:01.0000] yes and? [15:44:02.0000] and, assuming I understand you when you talk about the nonexistent tag, it would only be once a page [15:44:03.0000] it's also sensible for an article to be spread over multiple pages [15:44:04.0000] not necessarily [15:44:05.0000] the same as article, just with no implication re syndication; which is just odd [15:45:00.0000] i mean fuck, why don't you guys just make a tag? would compliment well your blog tags [15:45:01.0000] I give up. [15:46:00.0000] key, are you trying to be constructive here? [15:46:01.0000] yes [15:46:02.0000] Then I regret to inform you that you're not very good at it. [15:47:00.0000] k [15:47:01.0000] takes ears to hear [15:47:02.0000] Uh, yeah. /ignore, dude. [15:47:03.0000] I'm pretty sure that whoever's listening, cursing at them and telling them they're incompetent is an ineffective rhetorical strategy. [15:47:04.0000] i tried reason, fell on deaf ears [15:48:00.0000] Well, that's why I asked you if you were trying to be constructive. [15:48:01.0000] The alternative would be that you had given up and were just venting. [15:48:02.0000] slightly [15:48:03.0000] it's frustrating to me. the web has so much promise that has yet to be realized [15:48:04.0000] we're decades into it now, it should be better than this [15:49:00.0000] the tag set should be complete and abstract, and not imply things irrelevant [15:49:01.0000] if you ask me, the semantic web is mostly going to be a failure [15:49:02.0000] future tense? [15:49:03.0000] jcranmer: No need to use the future tense. The "Semantic Web" was a failure. [15:49:04.0000] jcranmer: you're probably right. the temptation to try to create a tag for everything will overwhelm it [15:50:00.0000] should be a complete set of abstract tags, not tags for every kind of thing [15:50:01.0000] maybe the Semantic Web was something I didn't think it was [15:50:02.0000] It punted on all the hard issues like trust in favor of solving the more theoretically interesting ratholes. [15:50:03.0000] /me goes to check Wikipedia [15:50:04.0000] I don't think trust was the big issue. [15:50:05.0000] trust is a huge issue [15:50:06.0000] eg,

. i mean wtf, took me 3 minutes of thinking to come up with being superior [15:50:07.0000] I don't think it was the fatal one, though. [15:50:08.0000] it's also known as "spam" [15:51:00.0000] You *must* solve trust. If you don't, the entire thing falls down immediately from the weight of spam. [15:51:01.0000] I think the big issue was almost no one actually produced useful machine-readable data repositories. [15:51:02.0000] the other huge issue is that it's nigh on impossible to get authors to write good machine-readable content [15:51:03.0000] the basic problem is that once you define content as being unstructured [15:51:04.0000] the SW has punted on both of these issues in avour of syntax bikesheds. [15:51:05.0000] you can never go back and decide that it can be structable [15:51:06.0000] er, structurable [15:51:07.0000] i maintain a hope that NLP will save the day [15:52:00.0000] cos god knows we're out of other options :-P [15:52:01.0000] NLP? [15:52:02.0000] I don't think you need to do full NLP [15:52:03.0000] /me has to force himself to not read that as "neurolinguistic programming". [15:52:04.0000] key: Natural Langage Processing [15:52:05.0000] oh [15:52:06.0000] heh, what do you think orwell's newspeak is for? [15:52:07.0000] reduce the speech-base, you make it easier to implement NLP [15:53:00.0000] although my attempt to extract HTML text by how the HTML was structured did turn out to be mildly failtastic [15:53:01.0000] now, granted, having about a weekend to implement it didn't help things [15:55:00.0000] jcranmer: oh i'm not saying you need full AI. With each step along the NLP path, you get results. The better the NLP the more impressive the results can be. [15:56:00.0000] NLP is more or less AI-complete, though. [15:56:01.0000] an example of incomplete NLP: translate.google.com [15:56:02.0000] "The article element represents an independent section of a document, page or site. It is suitable for content like news or blog articles, forum posts or individual comments." from http://www.alistapart.com/articles/previewofhtml5 [15:56:03.0000] Full NLP, sure. Google gets by pretty decently without AI, though. [15:56:04.0000] see how people are interpreting this tag guys? [15:56:05.0000] key: correctly? [15:57:00.0000] they aren't discussing the issue of implied syndication [15:57:01.0000] that's fine [15:57:02.0000] implied syndication is a perversion of this semantic need [15:58:00.0000] we need a way to denote a part of a web page as the main content area of the page, sans menu, extras, etc [15:58:01.0000] what we don't need, is to have implied syndication along for the ride [15:58:02.0000] i don't even mind the name article, rather the spec definition including said implied syndication [15:59:00.0000] we have rss if people want to syndicate their content, why wedge that kind of stuff into html? [15:59:01.0000] you keep saying "a way to denote a part of a web page as the main content area of the page" 2011-02-08 [16:00:00.0000] as if that has anything to do with

[16:00:01.0000] the main content area of the page, sans menu, extras, etc, is with

and

removed. [16:00:02.0000] key, it kind of seems like you are coming in here with an attitude problem [16:00:03.0000] key: is that because you are frustrated and not sure how to express your point, or just because you're the kind of guy to come into a channel and throw off attitude? [16:00:04.0000] key: if it's the former, I can give you advice [16:01:00.0000] shall i answer? or would you rather just keep spewing from the mouth [16:01:01.0000] the best way to effect change in the HTML spec is to write up use cases [16:01:02.0000] I'm guessing it's #2, in which case you should leave [16:01:03.0000] obviously you're the kind of guy who just asks rhetorical questions [16:01:04.0000] more of your assumption [16:01:05.0000] just because it's an open standards process and an open channel, doesn't entitle you to come in here and be a jerk [16:01:06.0000] straw man [16:02:00.0000] anyway, if you actually want to create change and not just vent, write up your use cases [16:02:01.0000] send to the mailing list [16:02:02.0000] start a discussion [16:02:03.0000] show data from real sites [16:02:04.0000] that is what works [16:03:00.0000] hober, the reason i say that is that 'content' is one of the biggest outline parts that exist [16:03:01.0000] along moreso than header/footer, even [16:03:02.0000] without content, you wouldn't have nav, or any of the rest [16:04:00.0000] hixie, are you saying that's implied by ? [16:04:01.0000] but that's an entirely different issue than any problem you may have with

[16:04:02.0000] my point is that we wouldn't need to create a new tag to fill this need, we could just clean up the definition of

and most likely change its name [16:05:00.0000] we already have a tag for this [16:05:01.0000] it's [16:06:00.0000] so the main content area shouldn't show up in the outlining algorithm? [16:06:01.0000] or when you draw out a web interface outline, you don't have a special area for "content" ? [16:06:02.0000] http://www.brucelawson.co.uk/2010/html5-articles-and-sections-whats-the-difference/#comment-667884 [16:06:03.0000] he raises good points i'd say. opinions? [16:08:00.0000] also, http://www.brucelawson.co.uk/2010/html5-articles-and-sections-whats-the-difference/#comment-680580 [16:08:01.0000] the outlining algorithm already handles all this [16:08:02.0000] see the spec [16:08:03.0000] (re spanning of logical content units across multiple pages) [16:09:00.0000] also, http://www.brucelawson.co.uk/2010/html5-articles-and-sections-whats-the-difference/#comment-688922 [16:10:00.0000] what i find is, when there is some issue that many are confused of, it isn't necessarily anything more than the topic being poorly constructed [16:10:01.0000] like the 7 series 'all nav in 1 button' bmw issue a few years back [16:11:00.0000] have you guys taken a step back and considered if perhaps article is poorly defined and constructed as a concept? [16:16:00.0000] No. We have irreversibly sunk into a bottomless sea of self-righteous delusion, never to emerge until we have destroyed the entire web platform through our willful ignorance. [16:16:01.0000] Sorry. [16:16:02.0000] can you please be serious? [16:17:00.0000] ok, who had AryehGregor down in the pool for "who will go sarcastic first"? [16:17:01.0000] No again. Sorry. [16:17:02.0000] hand in your bets people [16:17:03.0000] Hixie, I'm pretty sure I was sarcastic at least once before that. [16:17:04.0000] ah, must have missed it :-) [16:17:05.0000] yea, and i'm the jerk [16:17:06.0000] key, seriously, this is not a big deal either way. Everyone's going to use the element however they feel like, we can't stop it. [16:17:07.0000] i try to have a serious discussion and get the peanut gallery taking a shit [16:17:08.0000] If it's not named so wisely, well, it's hardly the first in that regard. [16:18:00.0000] key: your tone has not been one that one would describe as a serious discussion, as many people have already pointed out to you [16:18:01.0000] so why don't we just use tables for layout? if proper tag use doesn't matter [16:18:02.0000] key: and people have now tuned you out because of your tone [16:18:03.0000] oh my tone? does it hurt your spineless jellyfish bodies? [16:18:04.0000] That has quantifiable accessibility impact. [16:18:05.0000] fucking children [16:18:06.0000] case in point [16:18:07.0000] tough to reason with a channel full of queens [16:19:00.0000] key, your tone is inefficient for human communications. you are hereby /ignored. [16:19:01.0000] tantek, your pseudo-intellect bores me [16:19:02.0000] The annoying thing about /ignore is you still have to read the other people talking to him. [16:19:03.0000] Then you get curious about what he said. [16:19:04.0000] i'll just leave [16:19:05.0000] Which defeats the point. [16:19:06.0000] At least for me. [16:19:07.0000] irssi lets you /ignore the responses too [16:19:08.0000] which is pretty awesome [16:19:09.0000] Only if they contain his name, I assume? [16:20:00.0000] yeah [16:20:01.0000] AryehGregor - btw - granted - you're correct, that was more of a flame than a strawman. [16:20:02.0000] it doesn't have NLP yet :-) [16:20:03.0000] /me apologizes for key, i'm the one that mentioned this room earlier over in #html5 [16:20:04.0000] tw2113: no worries [16:21:00.0000] i was at work so didn't have ability to give full attention to any friendly arguements he had [16:21:01.0000] and don't know quite enough to give authoritative voice :D [16:22:00.0000] Actually, should be fairly easy for NLP to detect both such trivial name-calling and above-average use of expletives to auto-categorize a participant as potentially unproductive. [16:22:01.0000] above-average *frequency* that is, rather than above-average sophistication [16:22:02.0000] could be a useful IRCbot [16:22:03.0000] send those messages to /dev/null? [16:23:00.0000] as a first implementation. second, connect an Eliza-style backend to attempt to educate the unproductive participant into being more productive [16:23:01.0000] saving humans the time of doing so [16:24:00.0000] i don't know how much formal introductions are encouraged, but i'm michael [16:24:01.0000] btw, where do you guys stand on capitalized tag names? [16:24:02.0000] rather, up-cased [16:24:03.0000] or ? [16:25:00.0000] personally, i do [16:25:01.0000] The spec is intentionally indifferent [16:25:02.0000] i think letter case has been made irrelevant [16:25:03.0000] anyone can correct me if wrong though [16:25:04.0000] wow, key wasn't gone long [16:26:00.0000] spec says you can any case [16:26:01.0000] had a new thought [16:26:02.0000] yea im just asking personal prefs [16:26:03.0000] i know what the spec says, heh [16:26:04.0000]

,
all equivalent [16:26:05.0000] I do everything lowercase. [16:26:06.0000] Because it's easier to read. [16:26:07.0000] And also it compresses better, if you want to be obsessive. [16:26:08.0000] ya? interesting. i find my style more readable [16:26:09.0000] personal preference is personal preference [16:26:10.0000] i capitalise the first letter, because my finger can't get off the shift key fast enough after typing the < [16:26:11.0000] And also because uppercase tag names put me in the mind of 1990s HTML, before XML was introduced and all the trendy people started using lowercase. [16:27:00.0000] :O which one holds more data weight? capital or lowercase? [16:27:01.0000] ew i hate capitalized style the most :P [16:27:02.0000] I even use instead of . [16:27:03.0000] NO YOU'RE BOTH WRONG, YOU NEED TO USE RANDOM CAPITALIZATIONS [16:27:04.0000] at least be consistent! [16:27:05.0000] e.g., sCRipT [16:27:06.0000] kiddy [16:27:07.0000] I do that in tests. [16:27:08.0000] i made my first cat shirt today [16:27:09.0000] There are some examples in the spec that do it too, I think, just to make the point. [16:27:10.0000] save file size! use the smallest characters possible [16:27:11.0000] elements with odd numbers of attributes should be uppercase, and even numbers of attribute should be lowercase [16:27:12.0000] so you can tell at a glance whether the number of attributes is odd or even [16:27:13.0000] haha [16:28:00.0000] oh come on, you can throw more data into that tag name [16:28:01.0000] AryehGregor: none of the examples in the spec go quite as far as random case in the middle of tags :-P [16:28:02.0000] i'm not quite THAT crazy [16:28:03.0000] that's silly. everyone knows UPPERCASE TAGS ARE FOR SHOUTING. [16:28:04.0000] tantek: so but ? [16:28:05.0000] Hixie, they should! [16:29:00.0000] Hixie, bingo [16:29:01.0000] Hixie: throw one in when you write an example on April 1 [16:29:02.0000] tantek, isn't intended for person to person communication, rather machine to machine, so the upcase convention of SHOUTING doesn't apply [16:29:03.0000] it's syntactic, not semantic [16:29:04.0000] /me ducks [16:29:05.0000] it's quixotic [16:29:06.0000] haha [16:30:00.0000] now you're just being silly [16:30:01.0000] /me points to the topic [16:30:02.0000] Well, if your HTML isn't meant to be read by humans, then why do you care about case? [16:30:03.0000] Figure out what's fastest to parse or compresses best or something. [16:30:04.0000] (AFAIK, XHTML went with lowercase because it compresses better in typical documents.) [16:30:05.0000] (Slightly.) [16:30:06.0000] i think you guys would benefit from having my mental stimulation regularly; i think i'll hang out here [16:31:00.0000] /me sits down, pulls up a recliner [16:31:01.0000] Just try not to violently insult us all more than, say, once a week. [16:31:02.0000] ok, deal [16:31:03.0000] and preferably not on days that I have Abstract Algebra [16:31:04.0000] my head hurts enough after that class [16:31:05.0000] avoid days that end in "ay" [16:31:06.0000] jcranmer, oh, you're taking abstract algebra? As part of a computer science major or such? [16:31:07.0000] mm, get your ambiant calculus on [16:32:00.0000] AryehGregor: yes [16:32:01.0000] I wound up semi-inadvertently specializing in abstract algebra for my (math) bachelor's and master's degrees. [16:33:00.0000] Every ideal is contained within a maximal ideal!!! [16:33:01.0000] if you want a detailed explanation on why all normal subgroups are kernels of some surjective homomorphism [16:33:02.0000] (proof: Zorn's lemma) [16:33:03.0000] I am not your person [16:33:04.0000] btw, if i have a corporate web site (doesn't publish/syndicate content), would it be best to wrap the content area with
then have nested sections within it? or do the
with nested
jig? [16:33:05.0000] That's trivial. [16:33:06.0000] Just take the natural map from G to G/N. [16:33:07.0000] for what it's worth, i think the internet has syndication coming out its ass, so article is actually quite the best term for it [16:33:08.0000] Then that's a surjective homomorphism, and N is the kernel. [16:34:00.0000] If you have any questions, ask in #math. They like people who are advanced enough to actually be asking stuff about group theory. [16:34:01.0000] tw2113: i'm inclined to agree, but what about the issue of a single logical article spanning multiple pages, and hence multiple
instances [16:34:02.0000] (as opposed to the flood of calc and pre-calc and numerical computing and programming) [16:34:03.0000] isn't algebra pretty abstract in its normal state? what is _abstract_ algebra? [16:34:04.0000] what's wrong with that? [16:34:05.0000] It's algebra where you prove stuff about abstract *types* of structures. [16:35:00.0000] like lambda calculus? [16:35:01.0000] Dunno any lambda calculus. [16:35:02.0000] and even type taken abstract, ie abstract character [16:35:03.0000] Hixie: it [16:35:04.0000] But regular algebra is where you make abstract statements about, e.g., the real numbers. [16:35:05.0000] or dimensional analysis? [16:35:06.0000] 's where you decide that abstract vector spaces are too concrete [16:35:07.0000] nod, what AryehGregor said [16:35:08.0000] In abstract algebra, you make statements about more abstract structures, like groups. [16:35:09.0000] so you decide to prove things using as few requirements as possible [16:36:00.0000] The real numbers under addition are a group, so are the nonzero real numbers under multiplication, so are the invertible matrices under multiplication, so are zillions of things. [16:36:01.0000] tw2113: i've decided to use article with nested sections [16:36:02.0000] I would have mentioned the hard-to-understand theorem we had in class today [16:36:03.0000] but I forgot what it actually stated [16:36:04.0000] have at it [16:37:00.0000] i liked the link you sent me that defined them as: article; whole unit of related content. section; related contents. div; null semantic [16:37:01.0000] something about using homomorphisms to inductively prove that abelian groups have an element of order divisible by p, where p is some prime or something [16:37:02.0000] it's the content semantic breakdown i'm looking for [16:37:03.0000] one of those from html5doctor? [16:37:04.0000] yea [16:37:05.0000] jcranmer: oh i understand why your head hurts [16:37:06.0000] or at least, one i found via it [16:38:00.0000] good resource there [16:38:01.0000] word [16:38:02.0000] A group is any set G together with a binary operation * on the set such that 1) for any x, y, z, (x * y) * z = x * (y * z); 2) there is some e so that for all x, x*e = e*x = x; 3) for any x, there is a y so that x*y = y*x = e. [16:38:03.0000] Then you prove stuff about them. [16:38:04.0000] jcranmer, the order of any nontrivial group element is divisible by p for some prime p. [16:38:05.0000] ah, apparently this is Cauchy's Theorem [16:38:06.0000] Oh, Cauchy's theorem. [16:38:07.0000] That's not just for abelian groups, though. [16:39:00.0000] well, we only proved it for finite abelian groups [16:39:01.0000] It's really a weak corollary of Sylow's theorems. I can't remember how to prove it offhand. [16:39:02.0000] It's true for all finite groups. [16:39:03.0000] write something for 1/10 of the page in your notebook, that's how you prove it [16:39:04.0000] That's only a few lines. [16:39:05.0000] tw2113: let's say i have an article with 5 pages. i wrap the first page of article in the template's
. now let's say i want to provide a link to the 4 other pages. would i put a
within the
, and expect user agents to understand that as an article menu? [16:39:06.0000] actually, it's a lot [16:40:00.0000] I use graph paper for notes [16:40:01.0000] One of Sylow's theorems states that if G has order n, and p^m divides n, and p^(m + 1) does not divide n, then G has a subgroup of order p^m. [16:40:02.0000] and I can squeeze three lines of text per square [16:40:03.0000] Then all elements of that subgroup have orders that are powers of p, by Lagrange's theorem. [16:40:04.0000] So just raise one to a suitable power and you get an element of order p. [16:40:05.0000] hey, Lagrange's theorem! [16:40:06.0000] There's surely a much more elementary proof of Cauchy's theorem, though. [16:40:07.0000] a name I recognize [16:41:00.0000] Yeah, that's one of the most basic theorems. [16:41:01.0000] Where by "basic" I mean "important" or "fundamental". [16:41:02.0000] the size of a subgroup is a divisor of the size of a group [16:41:03.0000] Yep. [16:41:04.0000] Thus also the order of an element is a divisor of the size of the group, since any element generates a cyclic subgroup of its own order. [16:41:05.0000] someone correct me if i'm wrong, but i think if the nav is within the parent article tag, then ideally, the browser would associate the nav with that article [16:42:00.0000] You can stack up lots of theorems that let you restrict what finite groups can look like. Once you get to Sylow's theorems, you can do lots of precise classification. [16:42:01.0000] Like "there's exactly one nonabelian group of order 21". [16:42:02.0000] (finite abelian groups are classified much more easily, using the fundamental theorem of finite abelian groups; they're all direct products of cyclic groups) [16:42:03.0000] Anyway, have fun. [16:43:00.0000] tw2113: Yeah, you can associate it with the nearest ancestor article. [16:43:01.0000] I once tried to make a math joke in a programming channel and no one got it. :( [16:43:02.0000] /me is always somewhat ashamed to admit that he's bad at higher math. [16:43:03.0000] /me recalls that he was going to look up what the topic of Wednesday's class would be to prevent a brain fart at the beginning of class [16:43:04.0000] there ya go key [16:43:05.0000] Though to be fair, the math I do know still puts me in like 95th percentile at least. [16:43:06.0000] ok nice [16:44:00.0000] (It was a pun on someone misspelling "discreet" as "discrete". I contrasted it to the trivial topology, but apparently no one knew basic topology.) [16:44:01.0000] (For future reference, the discrete topology is the finest possible topology, while the indiscrete or trivial topology is the coarsest.) [16:44:02.0000] tw2113, TabAtkins, so this is how nested nav's are rational?; that they associate to nearest sectioning ancestor in their tree? [16:44:03.0000] you see [16:44:04.0000] (Although I don't really know much topology, sadly.) [16:44:05.0000] from what i gather, it's like that for most of them [16:44:06.0000] topology is considered magic to the rest of us [16:45:00.0000] Topology is neat. [16:45:01.0000] Neat but magic. [16:45:02.0000] At least at a basic level, it seems much slicker to me than group theory. [16:45:03.0000] I barely held my way through differential equations [16:45:04.0000] Algebra generally involves more computation than topology, and less visual thinking. [16:45:05.0000] I can get a hold of abstract algebra if I actually do some stuff with it [16:45:06.0000] Differential equations is pretty much just memorizing types of equations and practicing so you can solve them fast, isn't it? [16:46:00.0000] I took a differential equations course for the heck of it. Pretty boring. [16:46:01.0000] I don't know how I got an A in that class [16:46:02.0000] You can't really understand any math unless you do stuff with it. [16:46:03.0000] I somehow managed to be able to do it without understanding what I'm doing it [16:46:04.0000] tw2113: as witnessed by section being nestable? [16:46:05.0000] i mean, in regards to your mention of recursiveness [16:46:06.0000] I can spout out the names of stuff, and I have a vague idea of how it works [16:46:07.0000] jcranmer, you realize that all basic undergrad math courses teach you how to do stuff without understanding it? [16:47:00.0000] /me doesn't remember how to solve anything but linear ordinary differential equations with constant coefficients [16:47:01.0000] no, no, calculus you understand [16:47:02.0000] AryehGregor: yea isn't that a pity? [16:47:03.0000] makes a bunch of calculators, not thinkers [16:47:04.0000] /me could maybe do polynomial coefficients -- that's Taylor series, right? [16:47:05.0000] well, I was taught calculus at a more theoretical level [16:47:06.0000] You can sort of understand calculus, but not really, unless you do it all with proofs. [16:47:07.0000] we did proofs [16:47:08.0000] yes key [16:47:09.0000] And actually are required to know all the proofs. [16:47:10.0000] oh god, we did proofs [16:47:11.0000] we stopped short of doing real analysis [16:48:00.0000] but I do recall proving taylor's theorem [16:48:01.0000] The key question is: were you required to prove things that weren't discussed in class? I.e., did you have exercises of the form "prove this thing you've never seen before"? [16:48:02.0000] yep [16:48:03.0000] Okay, then it sounds like a serious course. [16:49:00.0000] that was also very prevalent in abstract vector spaces [16:49:01.0000] In my college, only math majors took courses like that, as far as I could tell. [16:49:02.0000] Mostly only pure math majors. [16:49:03.0000] tw2113: ok so let's say you have your nav bar in your header. so you put
in
. does this associate the menu to the header vs the whole document/site? [16:49:04.0000] I even tripped up in one theorem because I assumed only finite-dimensional vector spaces [16:49:05.0000] Infinite-dimensional vector spaces aren't nearly as well-behaved. [16:49:06.0000] well, it could be countably infinite-dimensional and still work [16:50:00.0000] Hah. [16:50:01.0000] the proof broke down for uncountably infinite [16:50:02.0000] What result were you using that held for only countable dimensions? [16:50:03.0000] I forget [16:50:04.0000] depends, which header are we talking about? [16:50:05.0000] let me check my noteboke [16:50:06.0000] tw2113:
tag [16:50:07.0000] for the site? for the article? [16:50:08.0000] site [16:50:09.0000] /me can't think of any relevant ones offhand, that couldn't be patched up with some axiom of choice [16:51:00.0000] Maybe you used induction or recursion? [16:51:01.0000] (all i do on my site header is show the current date+time) [16:51:02.0000] then the nav within that one, would be associated with the header, which i assume would be assocaited with the site [16:51:03.0000] That's my bet. [16:51:04.0000] being the top level header [16:51:05.0000] Like Graham-Schmidt. [16:51:06.0000] Only works in countable dimensions. [16:51:07.0000] ok so you don't think a nav in a root level header would be 'buried' from proper prominence ? [16:51:08.0000] considering screen readers, etc [16:52:00.0000] and if you use WAI-ARIA stuff, you can give that nav a role and it'd be known as the primary nav to assisted browsing [16:52:01.0000] If you try to extend it, you have to subtract off infinitely many pieces from a vector, which is impossible. [16:52:02.0000] ah, I did an explicit enumeration of the elements in the basis [16:52:03.0000] tw2113: sure, but i'm starting with logical organization first [16:52:04.0000] And then did what with it? [16:52:05.0000] You can get away with that stuff sometimes in uncountable dimensions, as long as you're using the axiom of choice. [16:52:06.0000] no i don't think it'd end up buried [16:52:07.0000] (If you aren't, then you can't assume a basis to begin with in uncountable dimensions, so no point in going there.) [16:53:00.0000] thanks tw2113 [16:53:01.0000] (Since the axiom of choice is equivalent to "All vector spaces have a basis.") [16:53:02.0000] The question was to prove that [16:53:03.0000] To prove what? That the axiom of choice is equivalent to all vector spaces have a basis? [16:53:04.0000] If S is a subset of a vector space V, and T is a subspace of V, then prove that L(S) is a subset of T. [16:54:00.0000] Oh, okay. [16:54:01.0000] L being what? [16:54:02.0000] the set of linear combinations of S [16:54:03.0000] well, linear combinations of elements of S [16:55:00.0000] Then it's clearly wrong. Let S be V, for instance, and T be a proper subspace. [16:55:01.0000] Maybe you meant S is a subset of T, and prove that L(S) is a subspace of T? [16:55:02.0000] sorry, S is a subset of T [16:55:03.0000] I was trying to find out today if we there was a discussion about a "type" attribute for "input" element which would pop a map to select a location and sends back latitude/longitude [16:56:00.0000] Anyway, it follows directly from the definition of a subspace, which is basically "contains zero and is closed under linear combinations". Unless you define subspaces differently. [16:56:01.0000] Definitely not a basis-dependent question, not a good idea to bring those into it. [16:57:00.0000] karlcow, it's been discussed, but the consensus so far has been that it's best not to add new features like that barring strong demand, until we have good implementations of the existing ones. [16:57:01.0000] I had assumed that S was finite [16:57:02.0000] (plural) [16:57:03.0000] well, it probably would have worked if S was countably infinite [16:57:04.0000] karlcow: In other words, let the implementations catch up first. [16:57:05.0000] ok [16:58:00.0000] karlcow: That said, I know we have some internal interest in doing type=location. [16:58:01.0000] So when we do add it, Webkit should be happy about it. [16:58:02.0000] TabAtkins: do you remember links do discussions? [16:58:03.0000] /me has to step out suddenly [16:58:04.0000] karlcow: No, I just know water-cooler talk. [16:58:05.0000] I also have to leave now, so I can give the car to my wife for the night. [16:59:00.0000] And I'm off the computer for the night. [16:59:01.0000] jcranmer, have fun with your abstract algebra! [17:01:00.0000] will do [17:16:00.0000] anyone got IE9 handy? [17:17:00.0000] yup [17:17:01.0000] i don't even have windows handy :D [17:19:00.0000] http://gsnedders.html5.org/outliner/process.py is failing, does anyone know of another outliner? [17:22:00.0000] boogyman: can you install the ahem font (http://www.w3.org/Style/CSS/Test/Fonts/Ahem/) and then go to http://software.hixie.ch/utilities/js/canvas/ and click "download" and tell me what the picture looks like? [17:22:01.0000] boogyman: assuming it even works in IE9 [17:23:00.0000] on it like white on rice [17:24:00.0000] did it download a script that has a try { } block? [17:24:01.0000] and lots of fillRects? [17:24:02.0000] and a font line that uses Ahem? [17:25:00.0000] oh i misread what you said [17:25:01.0000] i thought you said "it looks like white on rice" :-) [17:25:02.0000] which i assumed meant it showed nothing :-) [17:29:00.0000] i don't think it's working properly [17:30:00.0000] that was my worry [17:30:01.0000] oh well, thanks anyway [17:31:00.0000] p://i52.tinypic.com/e6vzpf.jpg [17:32:00.0000] http://i52.tinypic.com/e6vzpf.jpg even [17:32:01.0000] that's after clicking "download"? [17:32:02.0000] yes [17:32:03.0000] hm, hold on one sec [17:33:00.0000] paste the text at http://junkyard.damowmow.com/448 into the textarea instead [17:33:01.0000] use http://junkyard.damowmow.com/449 [17:33:02.0000] it has the right encoding set [17:34:00.0000] now i see xep with some coloured lines [17:35:00.0000] this is after installed the Ahem font? [17:36:00.0000] yes [17:36:01.0000] weird [17:36:02.0000] wonder what's going on [17:37:00.0000] boogyman: what does http://junkyard.damowmow.com/450 look like in IE9? [17:37:01.0000] give me a second, i'll install it directly in my root Fonts folder, I had a symlink to another location [17:37:02.0000] ah ok [17:39:00.0000] 450 has an X then a black square [17:39:01.0000] ok excellent [17:40:00.0000] did the canvas thing change? [17:40:01.0000] http://i56.tinypic.com/2d7vl02.jpg [17:40:02.0000] fantastic, thanks [17:40:03.0000] stupid symlink [17:41:00.0000] so the one time i was hoping microsoft might have not followed the spec, they did follow it :-) [17:41:01.0000] ah well, can't complain i guess [17:41:02.0000] haha [17:42:00.0000] should i use within
? [17:43:00.0000] you can [17:43:01.0000] if is what you actually need [17:43:02.0000] see the spec [17:43:03.0000] boogyman: what is that web interface? [17:44:00.0000] I'm running the crap that is M$ Vista [17:45:00.0000] no i mean, looks like a function declaration in a web page, with a text area for the code of it [17:45:01.0000] like you can code a function through a web interface or osmething [17:45:02.0000] ah that, Ian would be able to give you that answer [17:45:03.0000] key: it's http://software.hixie.ch/utilities/js/canvas/ [17:46:00.0000] oh cool man, nice tool [17:46:01.0000] btw, i love switzerland; especially lausanne [17:48:00.0000] i love switzerland too, so pretty [17:48:01.0000] but i prefer california, as a place to live :-) [17:49:00.0000] which part? [17:50:00.0000] bay area [17:52:00.0000] ah, yea. i lived in palo alto in the .com days, had a blast [17:52:01.0000] are you back in .ch? [17:53:00.0000] nope, bay area [17:53:01.0000] ah [17:54:00.0000] Ian, what do you mean lived .com :) it's just the beginning of the .com days [17:55:00.0000] that was mr key :-) [18:05:00.0000] /me bows [18:07:00.0000] http://honolulu.craigslist.org/mau/cto/2198949378.html [18:07:01.0000] might snag that [18:07:02.0000] anyone else into vintage surf rides? [18:17:00.0000] http://www.pastie.org/private/vjpeztiw7fkzjzsfoxpua <- any critiques of my outline? [00:57:00.0000] Constructor(HTMLFormElement form) [00:57:01.0000] is the proper way to write it down right? [01:00:00.0000] /me wonders how many of the fields in the IANA registration are long obsolete [01:02:00.0000] what's the preferred way to make your content be constrained to 960 px wide and centered in ? [01:02:01.0000] body { max-width:960px; margin:0 auto } ? [01:02:02.0000] i was thinking something like
and putting the styling on article [01:02:03.0000] whatever suits you then [01:03:00.0000] well i like your idea [01:03:01.0000] i just didn't know we could do that [01:06:00.0000] aah damnit [01:06:01.0000] there's no hook in HTML5 for "constructing the form data set" [01:11:00.0000] http://www.w3.org/Bugs/Public/show_bug.cgi?id=12005 [01:16:00.0000] would you guys use
to wrap your web page? (let's say if you wanted to fix its width and center it) [01:16:01.0000] I think div could be better for wrapper? [01:17:00.0000] That would be a div, indeed [01:17:01.0000] oh, hm [01:18:00.0000] nice. something like
or? [01:19:00.0000] key: yeah, same way you use article [01:20:00.0000] ok awesome [01:20:01.0000] thanks [01:20:02.0000] does div mean division? [01:21:00.0000] The div element has no special meaning at all. It represents its children. It can be used with the class, lang, and title attributes to mark up semantics common to a group of consecutive elements. [01:22:00.0000] i thought it meant division, and was the most generic container tag there is [01:22:01.0000] the above is the spec's definition :) [01:23:00.0000] i like divisino [01:23:01.0000] divisino [01:23:02.0000] division [01:23:03.0000] would it be better to use class="document" or id="document" ? [01:23:04.0000] s/better/precise [01:23:05.0000] It'd be a container, not a document [01:23:06.0000] why do you need it at all? [01:23:07.0000] to fix the width of my web page, and center it [01:24:00.0000] you already have two wrapper elements (html and body) [01:24:01.0000] is it supported to fix the width of body? isn't that something controlled by browser window width? [01:24:02.0000] it's supported since ie6 [01:25:00.0000] if you care about ie5, you need a div [01:25:01.0000] is it possible to fix the width of body, yet still have the body bgcolor affect the entirety of the browser window? [01:25:02.0000] yes [01:25:03.0000] ok nice. so what would the css look like to do something like this? [01:51:00.0000] /me wonders if the scrollback is worth reading [01:51:01.0000] /me wasn't convinced from skimming it [01:52:00.0000] no [01:55:00.0000] annevk: I think it is quite OK to have those tests in seperate files fwiw [01:56:00.0000] From my sample of one, I expect that they are not very well constructed tests [01:56:01.0000] , however [01:57:00.0000] Microsoft dumps all their poor quality tests again [01:58:00.0000] at some point I should write about that, so there's at least some perspective about them on the web [02:10:00.0000] /me wonders what normatively specifies the toString of a DOM object [02:10:01.0000] Does WebIDL cover that [02:10:02.0000] ? [02:10:03.0000] think so [02:16:00.0000] AryehGregor, why are you reminding me of Sylow? :( [02:18:00.0000] /me agrees with hsivonen about script src/readyState [02:21:00.0000] zcorpan: maybe worth pointing out on the list if you dress it up more than "+1" [02:29:00.0000] done [02:43:00.0000] ok so if i wrap my web page document in a
, i'll then modify #document to fix the width of the web page content and center it in body. sound proper? [02:43:01.0000] fwiw, Michael Kay commented on the HTML is the new HTML5 post [02:44:00.0000] ? [02:48:00.0000] im having some trouble figuring out how to structure my web page [02:49:00.0000] the html [02:51:00.0000] key, #html5 is probably a better place [02:52:00.0000] it's more conceptual than those guys can probably handle [02:52:01.0000] i need the spec authors' ear [02:52:02.0000] no you don't [02:55:00.0000] i have a main menu at the top of my page, then a section which has a section menu + the content. so would i do nav id=main-menu, then section id="section", then nav, then article? [03:17:00.0000] events... [03:18:00.0000] registration / dispatch / event flow for nodes / canceling... [03:21:00.0000] http://www.pastie.org/private/z4e91xoa8dx8twu4etl2ua [03:21:01.0000] how would you guys html that up? [04:50:00.0000] trying to define events anew is tough [04:50:01.0000] meh [05:01:00.0000] I forgot how much a pain synthetic events were [05:01:01.0000] create / initialize / dispatch [05:01:02.0000] teehee [05:02:00.0000] Schadenfreude? [05:02:01.0000] Not the best quality in a spec editor [05:03:00.0000] Although rather a common one if existing specs are any guide [05:05:00.0000] you missed somewhere [05:05:01.0000] sarcasm even [05:07:00.0000] I want to add the event model to Web DOM Core [05:07:01.0000] I guess I should just start writing and start over a couple of times [05:07:02.0000] just like with XMLHttpRequest [05:08:00.0000] I have the basic model sorted out, but there's a lot of variables [05:08:01.0000] (and then hope my description is better than DOM Events, otherwise there's no gain here) [05:12:00.0000] [05:22:00.0000] if you can't write it, draw it [05:23:00.0000] Or fingerpaint. [05:23:01.0000] That should be fun [05:24:00.0000] Get all specs people together with those big squeezy tubes of poster paint [05:24:01.0000] And give everyone a red plastic apron [05:26:00.0000] And paint the bikeshed [05:26:01.0000] events are so complex... [05:27:00.0000] once you reach dispatch [05:27:01.0000] there's a whole path [05:27:02.0000] and for each object in the path [05:27:03.0000] there's listeners [05:27:04.0000] which you need to filter and keep ordened [05:27:05.0000] and then execute synchronously [05:27:06.0000] and depending on what that execution does [05:27:07.0000] you may need to stop [05:27:08.0000] or continue just for this object [05:28:00.0000] or continue in general but not follow the link [05:28:01.0000] Hmm, does onfoo come before or after addEventListener? [05:28:02.0000] it depends on when you set it [05:28:03.0000] Does the order events are added have meaning normally? [05:29:00.0000] Yes [05:29:01.0000] Depending what you mean "normally" [05:29:02.0000] listeners for an event have an order [05:29:03.0000] and it's important [05:29:04.0000] otherwise stopImmediatePropogation() would not make sense [05:30:00.0000] not that I ever used that and not that this is not highly over engineered, but you know, it is important [05:45:00.0000] what happens when events are dispatched in subtrees? [05:45:01.0000] hey MikeSmith [05:45:02.0000] hola [05:49:00.0000] e.g. if you have a DocumentFragment with an element in it and you register a capture listener on DocumentFragment for 'click' and then invoke a.click() [05:50:00.0000] I suppose that might potentially do weird things anyway given that the is not in a document [05:50:01.0000] :/ [05:50:02.0000] And I guess people rely on whatever happens now [05:52:00.0000] interesting [05:52:01.0000] in Opera that works [05:53:00.0000] in Firefox/Chrome has no method click()?! [05:53:01.0000] Wut [05:53:02.0000] http://software.hixie.ch/utilities/js/live-dom-viewer/saved/830 [05:55:00.0000] http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Cscript%3E%0A%20w%28%22click%22%20in%20document.createElement%28%22a%22%29%29%0A%20%3C%2Fscript%3E [05:56:00.0000] no idea what is going on here [05:58:00.0000] if I change it to a synthetic events it seems events work in any kind of tree though [05:59:00.0000] s/synthetic events/synthetic event [06:00:00.0000] That's only on button and input [06:02:00.0000] click()? [06:02:01.0000] MikeSmith: do you happen to have demo cases that trigger the code path that makes Jing report what would have been allowed based on the schema? [06:02:02.0000] oh, I wonder why [06:02:03.0000] it works in Opera [06:03:00.0000] hsivonen: some in http://dev.w3.org/html5/tests/validation/full/invalid/missing-attributes/ [06:03:01.0000] if that's what you mean [06:04:00.0000] Ms2ger, you sound unhappy to be reminded of Sylow. That's sad. :( [06:04:01.0000] MikeSmith: that's a spec extract, though [06:05:00.0000] MikeSmith: what happened to adopting George Bina's Jing patch for generating messages from the schema? [06:05:01.0000] /me has terrible memory [06:06:00.0000] AryehGregor, the theorems are fine, the proof isn't ;) [06:07:00.0000] MikeSmith: ooh. now I see it. doh. [06:07:01.0000] hsivonen: I reduced George's patch to only reporting the element and attribute names [06:07:02.0000] Heck if I remember the proof. [06:07:03.0000] hsivonen: George's original dumped a string representation of the content model [06:07:04.0000] I've seen it two or three times, but I don't remember it at all. [06:07:05.0000] annevk, https://bugzilla.mozilla.org/show_bug.cgi?id=583514 [06:08:00.0000] IIRC it's pretty tedious, the sort of proof you want to skim for form's sake instead of really understanding and remembering. [06:08:01.0000] My prof disagreed [06:08:02.0000] Heh. [06:08:03.0000] Worst proof I remember seeing in a textbook was the Lindemann-Weierstrass Theorem. [06:09:00.0000] It was several pages of completely ad hoc arguments involving a mix of Galois theory and complex analysis. [06:09:01.0000] At the time I didn't know any complex analysis, so I didn't even try. [06:09:02.0000] The result is very neat, though. [06:30:00.0000] what if you remove an element from the DOM that still has event listeners attached to it while you just dispatched an event and that element was in its path? [06:32:00.0000] I guess I need to test that script readyState stuff before commenting on it further [06:33:00.0000] I remember there being some compat issues btw with IE script loading but it's been too long :/ [06:35:00.0000] current DOM event spec seems to suggest stuff is still dispatched [06:36:00.0000] sounds like making garbage collection a pain... [06:38:00.0000] the current DOM Events draft from shepazu also does not seem to take into account registering and unregistering of event handlers while other event handlers on the same object are being executed [06:38:01.0000] I am pretty sure that ought to "just work" [06:39:00.0000] which means the "event node path" is some kind of static list whereas everything else is pretty much live [06:53:00.0000] so basically [06:53:01.0000] when an event is dispatched on a node [06:53:02.0000] you calculate its "event node path" and store it [06:54:00.0000] then you start walking that, set currentTarget and target appropriately, and process the event listeners for each object in the "event node path" sequentially, taking modifications into account as they happen [06:55:00.0000] all the while looking at the event object each step of the way to see if propagation has stopped, propagation has stopped immediately, or the event has been canceled [06:56:00.0000] hmm, still not past this? http://html5doctor.com/html5-living-standard/ [07:03:00.0000] a lot of the complexity is basically dispatching on nodes [07:05:00.0000] does a realistic spec for progress events exist? [07:05:01.0000] Anne has one [07:06:00.0000] http://dev.w3.org/2006/webapi/progress/ [07:06:01.0000] Ms2ger: thanks [07:07:00.0000] mind you, Gecko has extensions and nobody has documented them so far or told me whether they need to be [07:08:00.0000] :-( [07:16:00.0000] They're in the DOM LS spec [07:18:00.0000] DOM LS is obsolete [07:21:00.0000] Ms2ger, I might just check in some notes on Events as a comment [07:21:01.0000] Sure [07:24:00.0000] if listeners are live you can create infinite loops [07:24:01.0000] fun [07:25:00.0000] Infinite loops aren't hard, you can do them with loops [07:25:01.0000] sigh. the progress events for scripts in IE are useless: http://hsivonen.iki.fi/test/moz/script-readystate.html [07:26:00.0000] there's no event when readyState transitions to "loaded" [07:26:01.0000] there's only an event for "complete" after the script has run [07:29:00.0000] I do think this rewrite will make events easier to understand [07:29:01.0000] teehee [07:32:00.0000] Hmm. How does one select text nodes in XPath? e.g. if I want the text node siblings of a given node? [07:32:01.0000] (XPath 1.0) [07:34:00.0000] does postmessage allow you to talk directly from one iframe to another x-domain without the parent window listening? [07:34:01.0000] ::preceding-sibling()::text() or some such? [07:34:02.0000] /me forgot the syntax [07:35:00.0000] preceding-sibling::text() it seems [07:38:00.0000] Ah, my mistake was assuming that * was short for node() [07:39:00.0000] should just rtfs :p [07:39:01.0000] Well I was :p [07:40:00.0000] But I hadn't internalised the concept of principle node type [07:40:01.0000] lies [07:40:02.0000] *principal [07:41:00.0000] me wonders how often security principles and security principals get confused [07:42:00.0000] ouch. IE seems to execute external scripts synchronously if the script has already finished fetching [07:55:00.0000] hmm [07:56:00.0000] nothing in both HTML and DOM Events suggests that "activation behavior" cancels the event once the first element has been activated as far as I can tell... [07:56:01.0000] I guess I can look at that later [08:00:00.0000] smaug____, the default action of an event, that happens when the event is through its entire propagation path? [08:00:01.0000] /me is in a meeting [08:06:00.0000] this is pretty much though what sucks about the current spec [08:07:00.0000] it's very hard to tell how things happen [08:07:01.0000] http://www.mobl-lang.org/ [08:24:00.0000] hmm [08:24:01.0000] I guess default action is about object / event type / event trusted combination [08:24:02.0000] not event target or such things [08:25:00.0000] though default actions are probably only triggered in the target/bubble phase [08:41:00.0000] so per [08:41:01.0000] http://software.hixie.ch/utilities/js/live-dom-viewer/?%3Cbody%20onchange%3D%22event.preventDefault()%3Bw(event)%22%3E%3Cinput%20type%3Dradio%3E [08:41:02.0000] you cannot prevent the default action in the bubbling phase [08:42:00.0000] actually, when I move that onchange to the it does not work either [08:43:00.0000] ah, change is not cancelable [08:43:01.0000] click is [08:43:02.0000] until the very last moment [08:45:00.0000] okay, so while the default action can happen earlier it can be undone until the very end [08:57:00.0000] Ms2ger, https://bitbucket.org/ms2ger/web-dom-core/changeset/6c7a0c465d83 [08:57:01.0000] Nice [08:58:00.0000] I think I tackled the difficult part [08:59:00.0000] I guess tomorrow I should draft all the concept-blahdieblahs [09:05:00.0000] feel free to hack on it btw [09:05:01.0000] might be online later tonight if I get tired [09:33:00.0000] jgraham: Some Chrome engineers want to submit bidi tests to the html5 test suite. What's the best way to go about this? [09:34:00.0000] TabAtkins, push to Google's submission folder in hg and send email to the list [09:34:01.0000] (You need to be a WG member) [09:37:00.0000] What Ms2ger said [09:37:01.0000] Also, make markp fix his tests [09:37:02.0000] Okay. The hg repo is just part of the W3C repo, right? [09:37:03.0000] dvcs.w3.org [09:38:00.0000] /html [09:39:00.0000] Is there a format guide? [09:40:00.0000] For what, the tests? [09:40:01.0000] Maybe on the wiki somewhere [09:41:00.0000] Basically the theory is either javascript tests using testharness.js (controversial), self-describing reftests (controversial) or manual tests (unwelcome) [09:41:01.0000] In tests/resources/testharness.js you will find documentation for that harness [09:42:00.0000] and apisample.html should give some idea of how to use it [09:42:01.0000] Yay. [09:42:02.0000] s/controversial/IE team doesn't like/, s/unwelcome/IE team likes/ [09:43:00.0000] Well the first controversial is more like "Mozilla team doesn't like" [09:44:00.0000] Well, they are rather verbose [09:55:00.0000] TabAtkins: http://www.w3.org/html/wg/wiki/Testing is the wiki page btw [09:57:00.0000] Ah, thank you. [09:57:01.0000] It doesn't actually tell you anything useful about how to write a test [10:01:00.0000] Or rather about the types of test that are allowed [10:02:00.0000] I just told them to either use the harness (and pointed them to the links you gave me) or make reftests (and pointed them to CSS's guides on how to make those). [10:03:00.0000] TabAtkins: Sounds perfect [10:04:00.0000] (CSS requires self-describing reftests, right?) [10:04:01.0000] It does now. [11:16:00.0000] Are there implementors here that write prototypes for firefox or chrome? [11:17:00.0000] i'm not sure what you're asking, but i'm fairly sure the short answer is yes. [11:17:01.0000] http://www.w3.org/Bugs/Public/show_bug.cgi?id=9557#c4 [11:17:02.0000] Maybe someone here that does that could prototype that in one of the browsers. Would be nice. [11:21:00.0000] I made a post to the public-webapps mailing list also since Doug said it would gain more interest that way. Curious if more people are interested nowadays. Seemed like only one or two people cared a year ago for mouse capture. [11:21:01.0000] Well, mouse capture is obviously a *requirement* for serious gaming on the web. You simple can't do mouselook without it. [11:25:00.0000] one day, i will make a shock site using all the new APIs :3 [11:40:00.0000] TabAtkins, tell me about it. I made that bug report a year ago when I was working on WebGL tests thinking it would magically get added. :P [12:36:00.0000] hsivonen: yt? [12:50:00.0000] hsivonen, just remembered, good mail re rdfa in html being subject to second guessing under html wg, fair and valid point - will be interested to see follow up responses (speaking for myself of course) [13:09:00.0000] is XSLTProcessor defined anywhere? [13:09:01.0000] Doubt it [13:11:00.0000] Hixie, possibly in some form in http://www.w3.org/TR/xslt20/ (I'd guess) [13:11:01.0000] i don't see any dom stuff at all in there [13:12:00.0000] Mozilla has docs [13:12:01.0000] from the lack of citation of a standard, I'd assume there is none [13:13:00.0000] likewise, would be surprised if there was dom stuff tbh [13:14:00.0000] are theer any specs or work done related to different interface devices, like barcode scanners etc? [13:14:01.0000] [13:14:02.0000] Barcode scanners are just keyboard devices. [13:15:00.0000] In general, though, yeah, . I've been poking around, looking for someone to help spec using with HID devices. [13:15:01.0000] So that we can, for example, get access to game controllers. [13:15:02.0000] lol I hadn't even noticed the element - better read up [13:16:00.0000] Don't read too much into what the spec says. [13:16:01.0000] game controllers should just be event handlers [13:16:02.0000] the implied permissions model of doesn't make sense for game controllers [13:17:00.0000] game controllers seem more like the keyboard, mouse, accelerometer or compass [13:17:01.0000] i think i'm convinced by the arguments by othermaciej, roc, and jamesr_, that we should just do things like ff4's geolocation rather than have an element for selecting a device [13:17:02.0000] though clickjacking has to be resolved first [13:17:03.0000] What do you mean, like just exposing an API for it directly, rather than through an element? [13:17:04.0000] yeah [13:18:00.0000] I agree with that. [13:18:01.0000] i think we should start with some actual concrete use cases [13:18:02.0000] the right API for game controllers is probably not the same as the API for a webcam [13:18:03.0000] concrete cases are easy [13:18:04.0000] (for webcam) [13:18:05.0000] othermaciej: So, I'm sympathetic to your position. Any ideas on how you would listen for arbitrary controller buttons? [13:18:06.0000] gmail's video chat [13:18:07.0000] no idea [13:19:00.0000] TabAtkins: this is actually already specced, it's a combination of the mousewheel and keyboard event interfaces imho [13:19:01.0000] what i think would be even more useful is exposing the mac remote [13:19:02.0000] to web pages [13:19:03.0000] jamesr_: The current API is just a way of asking the user for a hardware stream. The individual stream objects can have specialized interfaces. [13:19:04.0000] Hixie: Where is it specced? [13:19:05.0000] i'm tired of having to use remote desktop to pause hulu instead of just being able to use the remote [13:19:06.0000] TabAtkins: yeah, that's kind of not the point. the API you want and the permissions around it depend on _what_ is in the stream [13:20:00.0000] jamesr_: A valid point. [13:20:01.0000] TabAtkins: the old dom3 note had a simple way of exposing any key-like input [13:20:02.0000] i think treating everything as "a stream of data from the hardware" is throwing away lots of useful information [13:20:03.0000] I don't know anything about what data is typically exposed by a game controller [13:20:04.0000] TabAtkins: not sure where the mousewheel stuff is [13:20:05.0000] when the sorts of data are really different [13:20:06.0000] and i don't know of anyone that really wants to solve this for every type of hardware in the world. people care about particular use cases (webcam, microphone, controller) that will probably need different APIs [13:20:07.0000] of course if you're talking about modern controllers with their accellerometers and so forth, that's a different ballgame [13:21:00.0000] old school controllers have buttons/triggers and the analog stick [13:21:01.0000] jamesr_, the idea is that you get an object that implements a specialized API [13:21:02.0000] but in some cases buttons are themselves pressure-sensitive/analog, not binary on/off [13:21:03.0000] Ms2ger: then what's the value of at all? [13:21:04.0000] Getting that object [13:21:05.0000] would it even be possible to define an api that could be used by all devices of one category? or would it have to be more of an on demand loadDeviceAPI(foo) type thing [13:21:06.0000] (like crypto in firefox) [13:22:00.0000] jamesr_: the value was the permissions model. but i don't think anyone is advocating for anymore. [13:22:01.0000] some kinds of devices have a meaningful common interface [13:23:00.0000] it's a wonder there isn't already a game device api somewhere from game manufacturors that could be ported and used (game devices just being an example) [13:23:01.0000] Apple remote would be easy to expose as just key presses [13:23:02.0000] I think it's "USB" [13:23:03.0000] othermaciej: yeah [13:23:04.0000] othermaciej: we should do that [13:23:05.0000] Well, "HID". [13:23:06.0000] othermaciej: it would be awesome [13:24:00.0000] HID is a simplification of generic USB, and you can simplify it further. [13:24:01.0000] USB is not generally a useful application-level interface for many kinds of USB devices [13:24:02.0000] remote api for html? that'd be sweet, especially w/ tv and web wgs coming up [13:25:00.0000] TabAtkins: Barcode scanning input needs to be differentiated from keyboard input to avoid need for control focus. [13:26:00.0000] darrel: I wouldn't know. My old company's product just installed a keylogger and listened for Alt+F12 to indicate that it should steal focus. [13:26:01.0000] is there a set of requirements for what barcode scanners would need? if something was down I doublt it'd be the msot difficult thing to cater for, and probably worth it given all the html based pos terminals [13:27:00.0000] TabAtkins: Alt+F12 was encoded into the barcode? [13:27:01.0000] darrel: Yes. [13:28:00.0000] TabAtkins: Our enterprise app reads direct from the serial port and treats it as distinct input and we don't mess with the user's focus. [13:28:01.0000] Oh, your scanner plugs into the serial? Ours was, like I said, just a USB device that masqueraded as a keyboard. [13:28:02.0000] TabAtkins: Interesting workaround though. [13:29:00.0000] darrel, almost sounds like you're professionals [13:29:01.0000] You could open up Notepad and scan barcodes in. [13:29:02.0000] Ms2ger: hehe [13:29:03.0000] othermaciej: i was thinking we should map the remote keys to up, down, left, right, and space (i'm guessing apple won't want to expose the menu key since it goes into front row), with the events having .location set to DOM_KEY_LOCATION_REMOTE (0x06, a new value) [13:30:00.0000] Hixie: Would they be events just fired at window or something? [13:30:01.0000] just regular key events, fired like normal keyboard events at whatever has focus [13:30:02.0000] Hixie: when/if we do this, we'd probably want to give