| 00:00 | <sicking> | TabAtkins: http://nightly.mozilla.org/ |
| 00:00 | <TabAtkins> | That makes too much sense to be true.e |
| 00:00 | <sicking> | it'll even update you to the new nightly every day |
| 00:01 | <sicking> | it's the only thing i use |
| 00:01 | <sicking> | the only thing that is sad is that the release of new versions of firefox is much less exciting, since you've been using all the features for months already |
| 00:02 | <TabAtkins> | I think I'll survive. ^_^ I have nightlies here on my laptop, but I do most of my dev work on my work computer. |
| 00:05 | <karlcow> | sicking: you could download only the 6 months old nightly build as a rule ;) |
| 00:05 | <sicking> | heh :) |
| 00:06 | <sicking> | maybe we should set up a channel for that |
| 00:06 | <karlcow> | or a nightlyRoulette. Download a random nightly build |
| 00:37 | <sicking> | karlcow: hah, *that* would be awesome. And would likely generate a swarm of bugs from confused people |
| 01:49 | <GPHemsley> | How do I counter the argument that "IE doesn't support HTML5"? |
| 01:50 | <karlcow> | GPHemsley: which part of html5 |
| 01:50 | <GPHemsley> | there is no qualifier; it's just a blanket statement |
| 01:51 | <karlcow> | no browsers support the full stack of *proposed* features of html5 specifications and related |
| 01:52 | <GPHemsley> | right, but switching from HTML4 to HTML5 (primarily with the doctype, and then changing the code to validate) doesn't break anything, correct? |
| 01:55 | <miketaylr> | nope, unless you start putting in new elements, which won't be styled |
| 01:55 | <Dashiva> | Correct |
| 01:55 | <miketaylr> | but a little document.createElement('nav'), etc will fix that |
| 01:55 | <GPHemsley> | right |
| 01:55 | <Hixie> | GPHemsley: ask them if it supports HTML4... IE supports more parts of HTML5 than of HTML4. :-) |
| 01:56 | <GPHemsley> | Hixie: You're in the channel :P |
| 01:56 | <karlcow> | <!doctype html> should activate the strict mode in ie8, if I had my last browser class right. (if I'm wrong, it will be quickly rectified) |
| 01:56 | <Hixie> | GPHemsley: you didn't tell me what discussion it was :-P |
| 01:56 | Hixie | is in a ton of channels :-P |
| 01:56 | <GPHemsley> | :P |
| 01:58 | <karlcow> | then support also depends on what's in your pages. |
| 05:49 | <GPHemsley> | Is it possible for Bad Things™ to happen when simply switching an HTML4 doctype to HTML5? |
| 05:50 | <Hixie> | you can find people send you angry e-mails for no apparent reason |
| 05:50 | <Hixie> | does that count? |
| 05:50 | <wycats> | like what? |
| 05:51 | <wycats> | this was a fun commit: http://github.com/rails/rails/commit/01d92021e69f54def1ec8103b2b99f907dd88ec4 |
| 05:51 | <wycats> | look at the comments :) |
| 05:51 | <wycats> | "Holy crap!" |
| 05:52 | <wycats> | "mm, opinionated!" |
| 05:52 | <miketaylr> | omg _so_ opinionated |
| 05:53 | <wycats> | we should include a manifest by default ;) |
| 05:53 | <miketaylr> | :) |
| 06:08 | <GPHemsley> | Hixie: No, that doesn't count. :) |
| 06:09 | <GPHemsley> | wycats: I hope you later reinstated @lang? |
| 06:10 | <wycats> | GPHemsley: nope |
| 06:10 | <wycats> | should we have? |
| 06:11 | <GPHemsley> | Yeah, you should have a default language, especially if you already did before. ;) |
| 06:11 | <wycats> | this is master: http://github.com/rails/rails/blob/master/railties/lib/generators/rails/app/templates/public/404.html |
| 06:11 | <wycats> | patches accepted :P |
| 06:11 | <wycats> | I kid I kid |
| 06:12 | <GPHemsley> | I'd recommend it, but it doesn't matter to me... I don't use RoR ;) |
| 06:13 | <Hixie> | wycats: you can simplify the charset declaration to (using <meta charset="">) |
| 06:24 | <wycats> | Hixie: ya |
| 06:24 | <wycats> | that works on old browsers? |
| 06:24 | <Hixie> | yup |
| 06:24 | <Hixie> | that's why we allowed it |
| 06:24 | <wycats> | :-D |
| 06:24 | <wycats> | will fix |
| 06:25 | <wycats> | Hixie: what's your opinion about lang? |
| 06:25 | <Hixie> | it works because browsers had to support <meta name=http-equiv content=text/html; charset=...> where, if you look closely, you'll see that there are three tatributes |
| 06:25 | <Hixie> | and so browsers have to look for charset="" as an ottribute |
| 06:25 | <Hixie> | if you know the language, then set it |
| 06:25 | <wycats> | Hixie: lulz |
| 06:26 | <Hixie> | also apparently my typing sucks today |
| 06:26 | <Dashiva> | As opposed to when you try to type gauge? :P |
| 06:26 | <Hixie> | gauge |
| 06:26 | <Hixie> | woooo! |
| 06:26 | <wycats> | Hixie: that's hilarious |
| 06:26 | <Hixie> | i typed it right first time! |
| 06:27 | <wycats> | Hixie: I love these hax |
| 06:27 | <wycats> | like <!DOCTYPE html> turning on standards mode |
| 06:27 | <wycats> | in IE |
| 06:27 | <Hixie> | yeah |
| 06:27 | <wycats> | I don't know why, but it makes me warm and fuzzy |
| 06:29 | <Dashiva> | Hum |
| 06:30 | <Dashiva> | Wasn't there some discussion about early use of HTML5 doctype potentially leading to a "IE8 mode" lock in IE9? |
| 06:32 | <Hixie> | yeah well when google started using it on google.com I kinda gave up trying to stop it |
| 06:33 | <Dashiva> | Yeah, I used it myself too. Didn't seem like a winnable fight to begin with. |
| 06:37 | <Dashiva> | The @ping messages are informative, if only on unrelated things. I didn't know HTTP mandated UI. |
| 06:40 | <othermaciej> | good evening folks |
| 06:41 | <MikeSmith> | othermaciej: hey |
| 06:44 | <MikeSmith> | Hixie: I got a copy of this book last night - http://www.amazon.co.jp/HTML5-API入門-白石-俊平/dp/4822284220 |
| 06:44 | <MikeSmith> | I think it's the first printed book that's exclusively about HTML5 and related APIs |
| 06:46 | <MikeSmith> | the author, Shumpei Shiraishi, would like to mail you a copy of it |
| 06:46 | <Hixie> | he knows i can't read japanese, right? :-) |
| 06:47 | <MikeSmith> | yeah, sure |
| 06:47 | <Dashiva> | I wonder what revision of the spec it was made for |
| 06:47 | <MikeSmith> | just thought it would be nice for you to have |
| 06:47 | <Hixie> | MikeSmith: if he does want to mail me a copy, please thank him profusely on my behalf... he can mail it to any google office, c/o me, and it'll eventually get to me. |
| 06:48 | <Hixie> | http://www.google.com/corporate/address.html |
| 06:49 | <MikeSmith> | we were joking that he should just address it, "Hixie, Google, Mountain View, California" |
| 06:49 | <MikeSmith> | I'll pass on the address info to him |
| 06:49 | <Hixie> | "Hixie", i don't guarantee wull get to me :-) |
| 06:49 | <Hixie> | the rest would :-) |
| 06:49 | <MikeSmith> | hai |
| 06:49 | <Hixie> | but he's probably better off mailing it to tokyo or something |
| 06:50 | <Hixie> | since mailing books internationally can't be cheap |
| 06:50 | <Hixie> | especially books about html5, i mean, what is it, 20000 pages? :-) |
| 06:50 | <MikeSmith> | 325 pages |
| 06:52 | <MikeSmith> | anyway, I'll suggest to him that he get a copy to Takuya Oikawa and I guess either he could get it to you next time he's in Mountain View, or maybe pass it on to Ian Fette to bring back |
| 06:52 | <Hixie> | that works too |
| 06:52 | <Hixie> | though seriously |
| 06:52 | <Hixie> | we have inter-office mail |
| 06:53 | <Hixie> | :-P |
| 06:53 | <MikeSmith> | OK |
| 06:53 | <MikeSmith> | Dashiva: dunno what rev, but I know there are already a couple places where it's now out of sync with the spec |
| 06:53 | <MikeSmith> | e.g., it has a <details> example that uses <dt> and <dd> |
| 06:53 | <MikeSmith> | but anyway, he knew that going into it |
| 06:54 | <MikeSmith> | and the intro of the book of course warns that the spec is still under development and some things are still likely to change |
| 06:54 | <Dashiva> | I'm sure he'll be fine. He can release updated versions too. I'm thinking more about to what degree the books will stick around and confuse later. |
| 06:54 | <MikeSmith> | yeah, true |
| 06:55 | <Dashiva> | Maybe they should be Mission Impossible style self-destruct capability |
| 06:56 | <MikeSmith> | heh |
| 06:56 | <MikeSmith> | that's called "Kindle" |
| 06:56 | <MikeSmith> | that's called "Kindle" |
| 06:56 | <MikeSmith> | oops |
| 06:56 | <MikeSmith> | that's why they picked that name |
| 06:57 | MikeSmith | tries to remember which actual book it was that Amazon unilaterally self-destructed from users' kindles last year |
| 06:58 | <MikeSmith> | ah yeah |
| 06:58 | <MikeSmith> | Orwell's 1984 |
| 06:58 | <Hixie> | anyone have any idea what i meant in http://www.w3.org/Bugs/Public/show_bug.cgi?id=8999 ? |
| 06:59 | <Hixie> | oh maybe i meant that i should check if the origin is a tuple |
| 06:59 | <Hixie> | and if it's not, fail |
| 06:59 | <Hixie> | instead of failing if it's sandboxed |
| 06:59 | <Hixie> | that would make sense |
| 06:59 | <Hixie> | let's assume that's what i meant |
| 07:00 | Hixie | wonders how many bugs he's filed that later got marked NEEDSINFO before he made the script annotate who filed the bug |
| 07:02 | <Dashiva> | Looks like you wanted to split it up so sandboxing determines origin, and (only) origin determines .cookie |
| 07:15 | <Dashiva> | preload=autobuffer/playthrough doesn't help non-linear video (e.g. youtube choose-your-own adventure videos), or video where you expect the user to skip ahead for other reasons. |
| 07:19 | <annevk> | so XHR basically needs to return a special kind of Document |
| 07:19 | <annevk> | why was Document followed by object? seems redundant *shrug* |
| 07:24 | <wycats> | Hixie: http://www.whatwg.org/specs/web-apps/current-work/#resource-metadata-management |
| 07:24 | <wycats> | what do you mean by "miscellaneous"? |
| 07:25 | <Hixie> | "other" |
| 07:25 | <Hixie> | "things that didn't fit elsewhere" |
| 07:26 | <wycats> | don't they pretty much all reflect things about the HTTP request? |
| 07:26 | <wycats> | headers mostly? |
| 07:26 | <Hixie> | annevk: Document is an interface |
| 07:27 | <Hixie> | wycats: yeah, hence resource metadata management -- managing the metadata of resources |
| 07:27 | <annevk> | yay, feedback from dbaron |
| 07:28 | <annevk> | Hixie, I don't see how that matters here, but ok |
| 07:29 | <wycats> | Hixie: I guess it's fine :) |
| 07:42 | <Hixie> | if whoever filed http://www.w3.org/Bugs/Public/show_bug.cgi?id=9000 is here, can you tell me what "and so on" refers to? |
| 07:45 | <Lachy> | Hixie, I guess it's referring to the attributes that were listed in the HTML 4.01 DTD as being reserved for future use, but which are still undefined and non-conforming |
| 07:46 | <Lachy> | and whoever filed the bug thinks they should be listed in that section |
| 07:47 | <Hixie> | ...HTML4 reserves an attribute "dataformatas" [sic] |
| 07:47 | <Hixie> | that's awesome |
| 07:47 | <annevk> | might have been some IE stuff |
| 07:48 | <Hixie> | it's the typo i was referring to |
| 07:48 | <Hixie> | <table datapagesize> is reserved in HTML4, too |
| 07:49 | <Lachy> | maybe it wasn't a typo. Maybe it really is meant to be DataFormatAs which makes some sense given the allowed values are "plaintext" and "html" |
| 07:49 | <Hixie> | oohhhh |
| 07:49 | <Hixie> | valid |
| 07:50 | <annevk> | agreed that DataFormatas is cooler |
| 07:51 | <Lachy> | I just wonder what possible use it was meant for. It's supposedly reserved on a whole bunch of seemingly unrelated elements |
| 07:52 | <Hixie> | actually they're not reserved if you examine the DTD closely |
| 07:52 | <Hixie> | but IE did implement datasrc and datafld, iirc |
| 07:52 | <hsivonen> | Hixie: regarding the SVG <font> but: are you also looking at the SVG in table in SVG case heycam found? |
| 07:53 | <Hixie> | wasn't that fixed already? |
| 07:53 | <hsivonen> | maybe |
| 07:53 | <Lachy> | looks like http://www.html5reference.com/ has turned into a potentially useful site now. Last time I looked at it, it looked more like a spam site. |
| 07:53 | <hsivonen> | my mental snapshot of bug state could be stale |
| 07:53 | <Hixie> | hsivonen: i seem to recall i fixed the "reset the insertion mode" algorithm a while back |
| 07:53 | <Hixie> | to handle that case |
| 07:53 | <hsivonen> | Hixie: ok |
| 07:54 | <Hixie> | if i'm right, and didn't screw it up, it ended up being way simpler than i expected |
| 07:54 | <Hixie> | iirc there was even a comment in the source or something saying "oh yeah i should fix this" |
| 07:54 | <Hixie> | or maybe it was in some of my notes |
| 07:54 | <Hixie> | anyway i think it's fixed |
| 07:55 | <Hixie> | for the </font> and </a> thing unless you object i think i'll do it the way i described in the bug |
| 07:56 | <hsivonen> | hmm. I need to think a bit |
| 07:56 | <Hixie> | k |
| 07:56 | <Hixie> | you have until i get back to that bug :-) |
| 07:56 | <Hixie> | probably a few days at least |
| 07:56 | <Hixie> | maybe a month or so |
| 07:56 | <hsivonen> | so far, the end tag handling hasn't had to dispatch on foreigness |
| 07:57 | <Hixie> | on foreigness of the current node you mean? |
| 07:57 | <Hixie> | i don't see a way around that particular problem |
| 07:57 | <hsivonen> | I meant the mode |
| 07:57 | <hsivonen> | currently, the end tag just looks at the secondary mode |
| 07:58 | <Hixie> | ah |
| 07:58 | <Hixie> | well |
| 07:58 | <Hixie> | that's the bug :-) |
| 07:58 | <hsivonen> | so... |
| 07:59 | <hsivonen> | the solution is a bit annoying, because it goes back and forth between being 'in foreign' -specific at step 2 |
| 07:59 | <hsivonen> | and using the secondary mode in step 4 |
| 07:59 | <Hixie> | yeah i'm not hugely fond of the solution |
| 08:00 | <Hixie> | i'm very open to a solution that just fails for foreign end tags that don't match the current node, btw |
| 08:00 | <Lachy> | wtf? "The xml declaration and indicates that the content is a HTML document. The name parameter of the xml declaration must be html, matching the name of the top element an HTML document." -- http://www.html5reference.com/tags/xml-declaration/ |
| 08:00 | <Hixie> | Lachy: dude now my head hurts |
| 08:01 | <Lachy> | oh, I think what's happened is the author has copied and pasted from the DOCTYPE page to make it, and just done a search and replace. it says basically the same thing here http://www.html5reference.com/tags/doctype-declaration/ |
| 08:02 | hsivonen | tries to think of cases where step 4 has to walk a lot of stack nodes |
| 08:05 | <hsivonen> | Hixie: what's the purpose of "but there is no element in scope that has a namespace other than the |
| 08:05 | <hsivonen> | HTML namespace" |
| 08:06 | <hsivonen> | ? |
| 08:06 | <MikeSmith> | hsivonen: I applied Carey Evan's build.py patch and just now got it working.. I now have a working v.nu instance running on my windows VM and everything seems to be working as expected |
| 08:06 | <hsivonen> | MikeSmith: awesome. |
| 08:07 | <Hixie> | hsivonen: it's a poor-man's "reset the insertion mode" |
| 08:07 | <MikeSmith> | hsivonen: OK if I go ahead and commit his patch? |
| 08:07 | <hsivonen> | MikeSmith: ok |
| 08:07 | <MikeSmith> | cool, thanks |
| 08:07 | <Hixie> | hsivonen: it's to make <table><svg></table> drop out of foreign content, while leaving <svg><table><svg></table> in foreign content. |
| 08:07 | <Hixie> | (that same text is already in the spec) |
| 08:08 | <hsivonen> | oh |
| 08:08 | hsivonen | wonders if there are some bad time-complexity properties here with repeated stack walking with deep stacks |
| 08:09 | <othermaciej> | Hixie: it really does appear to be undefined what URL decomposition attributes return on <a href="http://a[b].example.org/">;, looking at HTML5 plus its normative references |
| 08:10 | <othermaciej> | the WEBADDRESSES parsing algorithm doesn't always return values for the components, but HTML5 assumes it does |
| 08:10 | <othermaciej> | http://www.w3.org/html/wg/href/draft#parsing-urls vs http://dev.w3.org/html5/spec/Overview.html#terminology-0 |
| 08:10 | <annevk> | yeah |
| 08:10 | <hsivonen> | Hixie: I *think* having the loop start in 'in foreign', go to secondary and loop back is going to suck |
| 08:11 | <annevk> | I was just about to reopen that bug and clarify, but something came up inbetween |
| 08:11 | <hsivonen> | Hixie: I don't have any better ideas to suggest though |
| 08:11 | <othermaciej> | annevk: I just did reopen the bug but feel free to add more info |
| 08:12 | <hsivonen> | Hixie: because I have the modes as switch-cases and not as function pointers |
| 08:12 | <Hixie> | othermaciej: no, the url decomposition attributes in that case all return the empty string, it's well-defined. |
| 08:12 | <Hixie> | othermaciej: specifically, this is because of the "Otherwise, the attribute must return the empty string" in the "On getting" paragraph. |
| 08:13 | <othermaciej> | Hixie: I see |
| 08:13 | <Hixie> | othermaciej: (because if it doesn't parse, it can't be an absolute URL, by definition) |
| 08:13 | <othermaciej> | Hixie: I missed that since it's separate from the definitions |
| 08:13 | <annevk> | is that what UAs do? |
| 08:13 | <annevk> | guess that works |
| 08:13 | <othermaciej> | annevk: it's what Safari and Chrome do for the example URL that Julian gave |
| 08:13 | <annevk> | wfm then |
| 08:13 | <othermaciej> | annevk: but not Firefox or Opera, I think because they actually do successfully parse that URL |
| 08:14 | <Hixie> | othermaciej: the table is meaningless without the paragraph from which i just quoted :-) |
| 08:14 | <Hixie> | hsivonen: yeah |
| 08:14 | <Hixie> | hsivonen: definitely open to better ideas |
| 08:14 | <Hixie> | hsivonen: not sure what to do though |
| 08:14 | <Hixie> | hsivonen: how do you jump to the secondary mode for end tags today? |
| 08:15 | <hsivonen> | Hixie: the mode is always the secondary mode and 'in foreign' is a separate flag |
| 08:15 | <Hixie> | ah |
| 08:15 | <hsivonen> | Hixie: so the end tag handler switches directly on mode |
| 08:15 | <hsivonen> | and at the end checks if it needs to flip 'in foreign' off |
| 08:17 | <hsivonen> | It's interesting how https://bugzilla.mozilla.org/show_bug.cgi?id=214476 gets a lot of dupes |
| 08:18 | <annevk> | othermaciej, ok, guess it depends on how we define parsing then |
| 08:18 | <annevk> | (and when that gets implemented... I'm guessing that'll take a while) |
| 08:19 | <hsivonen> | how actively is AIR tracking WebKit trunk? |
| 08:20 | <fantasai> | annevk: What are the rules and conventions for writing / submitting patches to html5lib? |
| 08:20 | <othermaciej> | no idea |
| 08:20 | <othermaciej> | AIR has not fed the vast majority of their changes upstream |
| 08:20 | <othermaciej> | they have their own perforce repostiory somewhere |
| 08:20 | fantasai | couldn't find anything about How to submit a patch or What a patch must contain before it's accepted |
| 08:20 | <annevk> | fantasai, prolly filing a bug |
| 08:20 | <fantasai> | ok, done |
| 08:20 | <annevk> | I haven't been involved for a while |
| 08:21 | <annevk> | jgraham can prolly help you out more, or maybe Philip` |
| 08:21 | <fantasai> | k |
| 08:21 | <annevk> | btw google/on2: http://on2.com/index.php?id=472&news_id=698 |
| 08:21 | <annevk> | now all we need is a public plan from google... |
| 08:22 | <MikeSmith> | hsivonen: ↑ seems they finally got around to not adjourning the shareholders meeting early.. |
| 08:22 | <hsivonen> | MikeSmith: yeah. nice. |
| 08:23 | hsivonen | is also interested in Google stating what they are up to now |
| 08:25 | <fantasai> | jgraham: Can you comment on html5lib bugs 47, 52, and 135 wrt what remains to be done to finish off their patches? |
| 08:25 | <Hixie> | anyone have any suggestions on how to fix http://www.w3.org/Bugs/Public/show_bug.cgi?id=9014 ? |
| 08:26 | <Hixie> | we can't hook it into the parser, since the actual end tag might be missing |
| 08:27 | <Hixie> | not to mention what happens if the element is moved around while still on the stack |
| 08:27 | <annevk> | if the end tag is missing, should it be instantiated at all? |
| 08:28 | <annevk> | isn't that much like <script> without end tag? |
| 08:28 | Hixie | tests |
| 08:29 | <Hixie> | data:text/html,<!DOCTYPE html><object type="application/x-shockwave-flash"><param name=movie value="http://www.macromedia.com/shockwave/download/triggerpages_mmcom/flash.swf">; |
| 08:29 | <Hixie> | chrome plays it |
| 08:29 | <Hixie> | firefox doesn't even with an end tag... |
| 08:30 | <Hixie> | opera plays it |
| 08:30 | <annevk> | in theory it could be dangerous or something, no? |
| 08:30 | <Hixie> | IE plays it |
| 08:30 | <annevk> | with the ddos thing and certain parameters not making it |
| 08:31 | <Hixie> | i suppose in theory, maybe, but i can't see how in practice |
| 08:32 | <annevk> | could do end tag or EOF |
| 08:32 | <annevk> | oh well |
| 08:36 | <Hixie> | firefox plays it as well (at least, it plays my debug plugin, i couldn't get it to work with flash) |
| 08:40 | <Hixie> | i have no idea how to test this |
| 08:41 | <Hixie> | hsivonen: by inspection, can you figure out when <object>s are instantiated in gecko? (i assume there's some parser magic involved) |
| 08:41 | <hsivonen> | Hixie: do you mean with the old parser? |
| 08:41 | <Hixie> | either probably |
| 08:41 | <Hixie> | unless you broke plugins |
| 08:41 | <hsivonen> | Hixie: I've done that but I thought I'd fixed it :-) |
| 08:43 | <hsivonen> | Hixie: there are two code paths, it seems |
| 08:43 | <hsivonen> | Hixie: one for parser-inserted case and one for the rest |
| 08:44 | <hsivonen> | so in the parser-inserted case, this code runs http://mxr.mozilla.org/mozilla-central/source/content/html/content/src/nsHTMLSharedObjectElement.cpp#196 |
| 08:44 | <hsivonen> | and per the comment, BindToTree takes care of the other case |
| 08:45 | <Hixie> | and this "done adding children" thing fires when? |
| 08:45 | <Hixie> | any time you pop it off the stack? |
| 08:45 | <hsivonen> | Hixie: yeah |
| 08:46 | <Hixie> | thanks |
| 08:46 | <jgraham> | fantasai: 47 and 52 require me (or someone else) to look at the sanitizer code again and verify that the patch is correct. 135 doesn't seem to have a patch |
| 08:46 | <hsivonen> | the "have notified" argument depends on whether the parser has returned control out of the parser between the start tag and the end tag |
| 08:47 | hsivonen | goes back to studying http://mxr-test.konigsberg.mozilla.org/mozilla-central/source/intl/chardet/src/nsMetaCharsetObserver.cpp |
| 08:47 | <jgraham> | fantasai: Sadly html5lib and the sanitizer in particular have no been top priority recently |
| 08:47 | <jgraham> | for me at least |
| 08:48 | <Hixie> | hsivonen: that's basically an implementation detail, right? |
| 08:48 | <Hixie> | something to do with creating frames or something? |
| 08:48 | <hsivonen> | Hixie: yeah, but I figured I'd mention it just in case you walk through the code |
| 08:48 | <Hixie> | thanks |
| 08:54 | <Hixie> | zcorpan, you make my life hard |
| 08:54 | <Hixie> | (specifically in this instance because of http://www.w3.org/Bugs/Public/show_bug.cgi?id=9014 ) |
| 08:55 | <othermaciej> | WebKit makes an effort to do something along those lines |
| 08:55 | <Hixie> | everyone seems to |
| 08:55 | <othermaciej> | we have a finishParsingChildren() method |
| 08:56 | <othermaciej> | though now I wonder what happens w/ an <object> element created purely through dom calls |
| 08:59 | <Hixie> | what stops you from instantiating a plugin when you set an orphan <div>'s innerHTML to something that includes an <object>? |
| 08:59 | <Hixie> | i guess the code checks that it's in a document |
| 08:59 | <Hixie> | hmm... |
| 08:59 | <hsivonen> | Hixie: it does |
| 08:59 | <Hixie> | innerHTML in the spec uses a document |
| 08:59 | <Hixie> | i guess i need to check for a document with a browsing context |
| 09:00 | <zcorpan> | i wonder if plugins are instantiated sync while scripts are running or not |
| 09:00 | <Hixie> | yeah that should work |
| 09:01 | <Hixie> | given how much time i'm spending trying to make flash work in the spec, you'd think adobe would be more supportive |
| 09:01 | <othermaciej> | hsivonen: btw it would be really useful if you could send any review comments you have about the impact on table parsing of that Change Proposal I linked in the next 7 hours or so, so the A11Y TF has them by their telecon time (since they will be discussing it), but if you don't have time today that's fine |
| 09:01 | <othermaciej> | Hixie: as much time as Apple has spent trying to make Flash run really well in Safari (and other Mac browsers)? |
| 09:02 | <hsivonen> | othermaciej: oh. ok. I didn't realize it was so time-sensitive |
| 09:02 | <Hixie> | othermaciej: probably less time than that, but, indeed |
| 09:03 | <hsivonen> | othermaciej: which list should I address my comments to? |
| 09:04 | <othermaciej> | hsivonen: <public-html-a11y⊙wo> |
| 09:04 | <hsivonen> | othermaciej: ok |
| 09:07 | <zcorpan> | http://www.w3.org/mid/4B7C91AC.4020301⊙ac - i wonder if it's because no-one cares about said xhtml spec |
| 09:08 | <Hixie> | so um |
| 09:08 | <Hixie> | hsivonen, othermaciej: about this <object> thing |
| 09:09 | <Hixie> | hsivonen, othermaciej: do browsers also have this magic for XML parsers? |
| 09:09 | <othermaciej> | Hixie: the only magic we have at the parser level afaik is notifying an element when all its children have been parsed, and I believe we do that in both HTML and XML parsing |
| 09:09 | <Hixie> | k |
| 09:10 | <Hixie> | zcorpan: given how often people complain to me about colour schemes i pick, i'm sure that's it :-) |
| 09:10 | <othermaciej> | Hixie: well, the other bit of magic being that the element knows if it was created by the parser |
| 09:10 | <othermaciej> | Hixie: this is actually the exact same two pieces of magic we use for <script> elements |
| 09:10 | <Hixie> | k |
| 09:13 | <hsivonen> | interesting. in Opera, unknown elements don't uppercase their tagName |
| 09:13 | <zcorpan> | hsivonen: that has been fixed |
| 09:13 | <zcorpan> | though we don't have localName and namespaceURI right yet |
| 09:14 | <hsivonen> | zcorpan: ok. I have an old build |
| 09:14 | <hsivonen> | zcorpan: and it keeps telling me that an update was downloaded but not applied |
| 09:17 | <annevk> | What is Larry talking about? He revealed his own private emails before: http://masinter.blogspot.com/2010/02/html5-and-w3c-priorities.html |
| 09:17 | <hsivonen> | whoa. IE8 does that too for elements enabled with document.createElement |
| 09:18 | <hsivonen> | annevk: what's Larry talking about where? |
| 09:18 | <zcorpan> | hsivonen: ie also supports /> for such elements |
| 09:19 | <annevk> | hsivonen, see http://lists.w3.org/Archives/Public/public-html/2010Feb/0622.html |
| 09:20 | hsivonen | is amused how @othermaciej misspelled @johnallsopp and now various people are tweeting to the nonexistent "johallsop" twitter account |
| 09:21 | <othermaciej> | I think he was might have been implying anyone citing the URLs of archive messages or implying anything about their relevance to the matter at hand (e.g. me) had failed to meet the W3C Member confidentiality policy |
| 09:21 | <othermaciej> | hsivonen: oops |
| 09:22 | <Hixie> | othermaciej: a propos of nothing, notice how on my blog post i carefully didn't imply anything about the three e-mails i cited. |
| 09:23 | <annevk> | othermaciej, ah I see, but by posting that he violated the rules too o_O |
| 09:23 | <hsivonen> | https://twitter.com/mattmay/status/9143644762 |
| 09:24 | <Hixie> | the whole secret list thing is such a joke |
| 09:24 | <annevk> | othermaciej, well, "too", not sure who else did |
| 09:24 | <Hixie> | w3c should just be public |
| 09:24 | <annevk> | hsivonen, that's bs |
| 09:24 | <annevk> | hsivonen, but I can't say why :/ |
| 09:25 | <Hixie> | i can understand having a secret list for minor administrative affairs, but for anything to do with objections, charters, etc, it's just so last-decade |
| 09:25 | <Hixie> | more than last-decade |
| 09:25 | <Hixie> | it's last-century |
| 09:25 | <Hixie> | since we're now in a new decade! |
| 09:25 | <Dashiva> | Well, if it's last century it's also last millennium! |
| 09:25 | <othermaciej> | annevk: doesn't the email by Larry that you cited appear to refute the twitter that hsivonen just linked? |
| 09:25 | <Hixie> | othermaciej: that reminds me. Are there any objections outstanding at this point, now that adobe has retracted theirs? |
| 09:26 | <Hixie> | Dashiva: indeed! |
| 09:26 | <othermaciej> | Hixie: yes |
| 09:26 | <othermaciej> | Hixie: well, let me rephrase |
| 09:26 | <othermaciej> | there is at least one claimed Formal Objection outstanding |
| 09:26 | <Hixie> | i saw one that sam essentially said was going to be ignored |
| 09:26 | <othermaciej> | there are a number of objections outstanding (including Adobe's) which were not raised as Formal Objections |
| 09:26 | <Dashiva> | Hixie: I think the intention is that the list _is_ for administrative affairs. However, the list existing at all allows for other uses, and nobody on the outside will be able to tell. |
| 09:26 | <othermaciej> | I just made a list of all objections raised and whether they had been addressed to the satisfaction of the objector |
| 09:26 | <Dashiva> | (With regard to html-wg practices) |
| 09:27 | <Hixie> | Dashiva: there are hundreds of private lists |
| 09:27 | <othermaciej> | the chairs will go over it tomorrow |
| 09:27 | <annevk> | othermaciej, true |
| 09:27 | <Hixie> | Dashiva: at w3c |
| 09:27 | <annevk> | othermaciej, though it doesn't explain the reason |
| 09:27 | <Hixie> | othermaciej: so we're missing the thursday publication window? |
| 09:27 | <Dashiva> | Hixie: Added a qualifier about html-wg |
| 09:27 | <Hixie> | othermaciej: i can redate them for tuesday if so |
| 09:28 | <othermaciej> | Hixie: we're going to go over them at 10 AM tomorrow, right after the telecon, can't say definitively whether we are missing the window until we've done that |
| 09:28 | <Hixie> | ah ok |
| 09:28 | <Dashiva> | The main argument for private lists seems to be that the people paying for them want them private, which is a real-world concern as real as they come |
| 09:28 | <othermaciej> | we will post our conclusions publicly |
| 09:28 | <Hixie> | i'll be offline til 1pm, i think, but i'll make sure to get online promptly after my meeting then so i can do any tweaks that pubteam need if we do publish tomorrow |
| 09:29 | <Hixie> | Dashiva: yeah the paying thing is another problem... |
| 09:29 | <Dashiva> | But even getting them to admit that is a step forward, IMO |
| 09:31 | <othermaciej> | our conclusions may include such things as requests to make minor changes to the document, explanations of why we would proceed over objections, perhaps suggested publication dates to put on the drafts if that ends up != Feb 18, and any other relevant information |
| 09:32 | <annevk> | i will likely not be around to make any edits |
| 09:32 | <annevk> | to html5-diff |
| 09:32 | <annevk> | i can also not attend the telcon |
| 09:33 | <Hixie> | opera is delaying html5!!!!!111oneone |
| 09:33 | <annevk> | well I could, but I want to go skiing on my new skis before I leave Norway |
| 09:33 | <annevk> | would make a great headline |
| 09:33 | <zcorpan> | s/delaying/blocking/ |
| 09:34 | <Hixie> | oh right, sorry |
| 09:34 | <Dashiva> | And in a private, secret forum such as IRC |
| 09:35 | <zcorpan> | opera prefers skiing over publishing html5 |
| 09:35 | <hsivonen> | othermaciej: I sent email to the accessibility TF |
| 09:35 | <annevk> | i updated XHR to prevent documents from giving or eating cookies |
| 09:35 | <othermaciej> | thank you sir |
| 09:36 | zcorpan | now wants to eat a cookie |
| 09:36 | <hsivonen> | othermaciej: is the list of objections at a URL? |
| 09:38 | <Hixie> | annevk: what does http://www.w3.org/Bugs/Public/show_bug.cgi?id=9017 mean? |
| 09:38 | <othermaciej> | hsivonen: all the ones I noted are in this month's archive of public-html, I strongly suspect the chairs will post a list of all of them tomorrow |
| 09:38 | <zcorpan> | mmm, good thing my fiancee is a confectioner |
| 09:39 | zcorpan | always has home baked cookies available |
| 09:39 | <othermaciej> | hsivonen: I'm not 100% sure yet if my list is complete and correct |
| 09:39 | <hsivonen> | othermaciej: ok |
| 09:40 | <annevk> | Hixie, your doc says "If the origin of the appropriate Document is not a scheme/host/port tuple" |
| 09:40 | <othermaciej> | basically the ones I spotted are various comments about status sections, Larry's scope objections, Martin's objections to canvas based on accessibility, and Krzysztof's duplication objections to Microdata |
| 09:40 | <annevk> | Hixie, well a) there might not be a document (see bug on workers) and b) the doc origin and xhr origin can differ I believe |
| 09:41 | <othermaciej> | (plus a few that I think are already resolved) |
| 09:41 | <annevk> | Hixie, so using the origin parameter of the fetch algorithm would be nicer |
| 09:41 | <Hixie> | oh i see |
| 09:41 | <Hixie> | ok |
| 09:41 | <Hixie> | thanks |
| 09:42 | <Hixie> | that means going through and actually setting origin everywhere |
| 09:42 | <Hixie> | that would break the Origin header stuff |
| 09:42 | <Hixie> | hmm |
| 09:43 | <Hixie> | ok well i'll deal with it later |
| 09:43 | <Hixie> | nn |
| 09:43 | <annevk> | you could let it default to the Doc origin if it was not given |
| 09:43 | <annevk> | kk |
| 09:43 | <annevk> | for CORS I use the origin arg too btw, same concerns |
| 09:43 | <annevk> | (CORS is using fetch) |
| 09:46 | <zcorpan> | the user agent must act as if the element was |
| 09:46 | <zcorpan> | + a <a href=#stack-of-open-elements>stack of open elements</a>.</p> |
| 09:46 | <zcorpan> | Hixie: is the word "in" missing there? |
| 09:47 | <othermaciej> | MikeSmith: are you about? |
| 09:47 | <MikeSmith> | yup |
| 09:47 | <othermaciej> | MikeSmith: I suspect Henri's email to public-html-a11y may be stuck in moderation |
| 09:47 | MikeSmith | checks |
| 09:47 | <othermaciej> | do you have the power to unstick, or do you know someone who can? |
| 09:48 | <MikeSmith> | If it turns out it needs moderation, I can |
| 09:50 | <hsivonen> | MikeSmith: it is stuck |
| 09:51 | <hsivonen> | (I got a note from a bot) |
| 09:53 | <MikeSmith> | OK, it seems I can't actually unstick it. |
| 09:54 | <MikeSmith> | hsivonen: the subscriber list is auto-generated from the W3C group DB |
| 09:54 | <MikeSmith> | and you are not in the DB for the group |
| 09:54 | <othermaciej> | there's no way to authorize a post by a non-subscriber? |
| 09:54 | <MikeSmith> | apparently not |
| 09:54 | <MikeSmith> | I had some discussion with systems team when this happened before |
| 09:55 | <hsivonen> | Should I just resend to public-html? |
| 09:55 | <MikeSmith> | I can't remember the reason why it doesn't show up in the moderation interface |
| 09:55 | <MikeSmith> | hsivonen: yeah, either that or I can add you to the group |
| 09:55 | <othermaciej> | hsivonen: sure, or you can send it to me and I can forward |
| 09:55 | <othermaciej> | or what MikeSmith said |
| 09:55 | <hsivonen> | It seems simplest if I just resend to public-html |
| 09:56 | <MikeSmith> | ok |
| 09:56 | <annevk> | annoying set up for a list... |
| 09:57 | <hsivonen> | sent |
| 09:58 | <othermaciej> | I'll just send an FYI link to the TF list so they don't miss it |
| 10:23 | <MikeSmith> | Hixie: thanks for the r4783 change (about value of empty attributes implicitly being the empty string) |
| 10:58 | <hsivonen> | meta charset is weirder than I had thought |
| 11:11 | <gsnedders> | Meh. My stomach aches as if of nervousness, yet I think it's just air in it making it sore. Peh. |
| 11:11 | <gsnedders> | And this isn't the right window. |
| 11:14 | <zcorpan> | hsivonen: what's with meta charset? |
| 11:15 | zcorpan | wonders why the sublime player doesn't work in opera |
| 11:25 | zcorpan | files https://bugzilla.mozilla.org/show_bug.cgi?id=546905 |
| 11:29 | <hsivonen> | zcorpan: the interaction with document's script-examinable charset, document.write and location.reload() |
| 11:30 | <hsivonen> | and scripts blocking the parsing of document.written content |
| 11:30 | <hsivonen> | in Gecko (old parser), meta charset in synchronously-parsed document.write doesn't cause renavigation |
| 11:31 | <hsivonen> | but if an external script has blocked parsing and the document.written meta charset is parsed as an event loop task later, it can cause a charset renavigation |
| 11:31 | <zcorpan> | renavigation really sucks |
| 11:31 | <hsivonen> | to a stream that contains the concatenation of the document.write arguments |
| 11:31 | <hsivonen> | it sure does |
| 11:32 | <hsivonen> | I was told Gecko ignored meta charset in document.write, so I implemented that |
| 11:32 | <hsivonen> | but things are a bit more complicated |
| 11:32 | <zcorpan> | do you log in the error console when you do renavigation? |
| 11:32 | <hsivonen> | nope |
| 11:33 | <hsivonen> | renavigating a document.open()ed stream implicitly scares me |
| 11:33 | <hsivonen> | but changing things scares me too |
| 11:42 | zcorpan | also files http://code.google.com/p/chromium/issues/detail?id=36120 and https://bugs.webkit.org/show_bug.cgi?id=35088 |
| 13:30 | <hsivonen> | http://www.w3.org/QA/2010/02/working_group_publication_requ.html#c186978 |
| 13:30 | <hsivonen> | "In the present situation and unlike what I said during the HTML teleconference, we will not hold the publication of the documents if the procedural concerns are not resolved." |
| 13:36 | <annevk> | ah, but my skiing plans will throw it all of the rails |
| 13:42 | <hsivonen> | annevk: what kind of disruptive skiing plans do you have? |
| 13:45 | <annevk> | apparently the plan is to publish later today while I'm trying to improve my downhill skiing skills |
| 13:46 | <annevk> | maybe I'll drop by the office after skiing for a few minutes to see if anything needs to be done |
| 13:47 | zcorpan | looks at http://tools.ietf.org/html/rfc3803 and https://trac.annodex.net/wiki/HttpHeaders |
| 13:47 | <zcorpan> | i wonder if mozilla plan to support those sans x- |
| 13:48 | <Dashiva> | How can the time value be exact if it only supports integer seconds? |
| 13:48 | <zcorpan> | it can't |
| 13:49 | <zcorpan> | https://developer.mozilla.org/en/Configuring_servers_for_Ogg_media indicates mozilla supports float values |
| 13:49 | <gsnedders> | Dashiva: How can a time value be exact if it is restricted to any non-fractional set? |
| 13:51 | <Dashiva> | "The time value should be exact" seems like a useless requirement, considering |
| 13:52 | <Dashiva> | gsnedders: If the frame time divides into a power of ten, it will work at some point of increased resolution :) |
| 13:53 | <Philip`> | Is there no way to distinguish unknown duration from infinite duration? |
| 13:53 | <gsnedders> | Dashiva: Indeed, but that isn't exact for something 1/3s long |
| 13:54 | <zcorpan> | Philip`: in the dom api, NaN means unknown and Infinite means infinite. but a browser can't tell the difference between a streaming resource and a normal resource when the server doesn't support byte range requests, i think |
| 13:55 | <Philip`> | Seems like a Content-Duration header ought to handle that case |
| 13:55 | <Philip`> | so that UAs do have some way to know |
| 13:56 | <Dashiva> | 1*DIGIT | 'Infinity' | 'NaN' |
| 13:56 | <zcorpan> | yeah |
| 13:56 | <zcorpan> | 1*DIGIT ["." 1*DIGIT]? | ... |
| 13:57 | <Philip`> | You don't need anything explicit for unknown, you can just not send the header |
| 13:58 | <Dashiva> | What if the web server is misconfigured to send a content-duration always, and you need to override it? |
| 13:58 | <Philip`> | You get a better web server |
| 13:58 | <Dashiva> | Opera had a bug like that |
| 13:59 | <Dashiva> | text-overflow only supported ellipsis, not clip |
| 13:59 | <Dashiva> | So once it was set, it was impossible to override it to return to the default |
| 14:23 | <Dashiva> | Going by the arguments that using @ping on links is bad because a) link implies safe and b) link tracking is non-safe, doesn't that mean anyone doing link tracking is already violating it? |
| 14:26 | <Philip`> | People already doing link tracking are using GET, so it must be safe |
| 14:46 | <danbri> | whatwg wiki seems super-slow today, some problem with it? |
| 14:46 | <danbri> | (who hosts it btw?) |
| 14:46 | <gsnedders> | danbri: hixie |
| 14:47 | <gsnedders> | danbri: It's a Dreamhost dedicated server, AFAIK |
| 14:47 | <danbri> | oh, dreamhost ... my commiserations |
| 14:47 | <gsnedders> | The dedicated servers are fine, from what I've been told |
| 14:47 | <gsnedders> | It's only if your on the shared servers that Dreamhost is diabolical |
| 14:47 | <Philip`> | Apparently not, given the performance of the site :-p |
| 14:47 | <gsnedders> | (in general; there seem to be a few that are fine) |
| 14:48 | <paul_irish> | gsnedders: i have the inverse view, actually. hate the VPS's and love the Shared, which they try so very hard to move you off of |
| 14:48 | <Philip`> | although probably any dedicated server would have the same problems |
| 14:48 | gsnedders | has a VPS with screen, irssi, and Apache serving thereshouldbenored.com on it |
| 14:49 | <gsnedders> | I need to move gsnedders.com on to it, but I haven't got round to that |
| 14:49 | <gsnedders> | (but not on Dreamhost, either) |
| 14:49 | <danbri> | i moved most of my stuff onto ec2 but didn't finish completely the dreamhost exodus yet :( |
| 14:50 | <danbri> | ec2 is expensive but likeable |
| 14:50 | <Philip`> | EC2 doesn't seem like good value if you want a constant amount of server power |
| 14:51 | <danbri> | i'd like to record rel="meta" in the whatwg wiki, since it has been used in html to link to foaf files for years ... would that be ok? |
| 14:51 | <Philip`> | (though it's good for flexibility) |
| 14:51 | <danbri> | the smallprint in wiki says either go thru Microformats process or be in a w3c cr or REC, ... which it is via http://www.w3.org/TR/2008/REC-rdfa-syntax-20081014/#relValues |
| 14:51 | Philip` | has an EC2 server that he runs for approximately one hour per week |
| 14:51 | <danbri> | but rather than pull in all those values it's really just meta i care about, to document existing practice |
| 14:51 | <hsivonen> | gandi.net provides Xen VMs at a competitive price |
| 14:52 | <hsivonen> | they are subject to French law though |
| 14:52 | Philip` | has been happy with gandi.net for irssi and web hosting |
| 14:52 | <hsivonen> | dunno how big a problem it is in practice if you host a public wiki in France and someone posts Nazi stuff on the wiki |
| 14:56 | <Philip`> | http://www.gandibar.net/post/2010/01/05/An-active-new-year - "We are surely going to open more data centers, notably in the United States, which is the world's largest market for this technology (cloud infrastructure), and the UK." |
| 14:57 | <Philip`> | danbri: You can always put rel=meta there as "Proposal", regardless of whether there's a spec |
| 14:57 | <Philip`> | which is probably a good first step |
| 14:57 | <danbri> | i'll do that |
| 14:58 | <danbri> | i don't have the energy for one of those giant blahblah rdfa blahblah microdata threads that arise, so i'm a bit wary of putting a foot in the crossover zone |
| 14:58 | <Philip`> | I wouldn't personally consider that #relValues thing to be a definition of the feature |
| 14:58 | <danbri> | but i think it needs documenting and has been used in good faith over the years |
| 14:58 | <mhausenblas> | ahm, what is the status of http://wiki.whatwg.org/wiki/RelExtensions anyway - will this be taken up or is this for documentation only? |
| 14:58 | <Philip`> | since "Refers to a resource that provides metadata, for instance in RDF." is really quite vague |
| 14:59 | <danbri> | the concept of metadata is inherently vague |
| 14:59 | <danbri> | but people still go to conferences about it ;) |
| 15:00 | <Philip`> | It could be less vague by saying what kind of data format consumers are likely to expect, and how it's meant to be related to the linking page, or something, I guess |
| 15:01 | <danbri> | ah i remember now, http://wiki.foaf-project.org/w/Autodiscovery ... tantek dug out some old history |
| 15:02 | <danbri> | i won't add it now, but will make a tidier doc to link to |
| 15:02 | danbri | makes notes |
| 15:02 | <Philip`> | If I look for <link rel=media type=?> then I see |
| 15:02 | <Philip`> | 3306 application/rdf+xml |
| 15:02 | <Philip`> | 99 application/rdf xml |
| 15:02 | <Philip`> | 18 application/xbel+xml |
| 15:02 | <Philip`> | 13 application/rss+xml |
| 15:02 | <Philip`> | 2 application/apml+xml |
| 15:02 | <Philip`> | 1 application/xml |
| 15:02 | <Philip`> | So it looks quite like it's really RDF |
| 15:03 | <danbri> | yes |
| 15:03 | mhausenblas | adds a pointer from http://wiki.foaf-project.org/w/Autodiscovery to http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData/AutoDiscovery |
| 15:03 | <Philip`> | except for a small number of exceptions and larger number of typos |
| 15:03 | <danbri> | from that foaf wiki page, "It makes a (previously unarticulated) assumption: information about the author of a document can also be considered metadata about the document itself. So long as the RDF/XML describes some relationship between Person and Document, it should be true that everything we learn about the Person also adds to our indirect knowledge of the document." |
| 15:04 | <danbri> | in http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/#transport "The recommended relation type for this purpose is REL="meta"" |
| 15:05 | <Philip`> | mhausenblas: http://www.w3.org/html/wg/tracker/issues/27 |
| 15:05 | <mhausenblas> | thanks Philip` |
| 15:06 | <Philip`> | mhausenblas: I think the idea in the related discussions was to switch to http://tools.ietf.org/html/draft-nottingham-http-link-header-07#section-6.2 perhaps |
| 15:07 | <mhausenblas> | +1 |
| 15:07 | <Philip`> | but I don't remember if anything was concluded yet |
| 15:07 | <mhausenblas> | can you check back with someone in charge, Philip`, please? |
| 15:07 | mhausenblas | would be very interested |
| 15:08 | <Philip`> | I don't think anybody is in charge here :-) |
| 15:09 | <mhausenblas> | hehe, Philip`, very funny :P |
| 15:09 | <gsnedders> | Besides, nobody knows the trouble I've seen. |
| 15:09 | <gsnedders> | So being in charge wouldn't help |
| 15:09 | gsnedders | typed that, then forgot how he was going to make it a relevant and witty comment |
| 15:09 | <Philip`> | mhausenblas: http://lists.w3.org/Archives/Public/public-html/2010Jan/1006.html |
| 15:09 | gsnedders | fails at humour |
| 15:10 | <Philip`> | mhausenblas: http://lists.w3.org/Archives/Public/public-html/2010Feb/0225.html |
| 15:10 | <Philip`> | I think those are the most relevant posts |
| 15:10 | <mhausenblas> | thanks |
| 15:12 | <Philip`> | danbri: I guess adding things to the RelExtensions page isn't incredibly useful now, since it seems it'll be obsoleted |
| 15:14 | <danbri> | it will? |
| 15:14 | <Philip`> | Yes |
| 15:14 | <danbri> | ah, the rollercoaster ride that is standards |
| 15:14 | <Philip`> | See previous dozen lines :-) |
| 15:14 | danbri | clickin' |
| 15:14 | <Philip`> | about switching to a non-HTML-specific rel registry |
| 15:16 | <danbri> | seems a good idea to me |
| 15:35 | <MikeSmith> | .me wonders if anybody here done much with Hadoop |
| 15:35 | <MikeSmith> | it looks pretty interesting |
| 15:37 | <Dashiva> | If you have a few thousand nodes to work with :) |
| 15:37 | <MikeSmith> | well, I got my spam botnet |
| 15:38 | <MikeSmith> | I could always repurpose that for other things |
| 16:06 | <jgraham> | So if you have <img src=""> its complete attribute should be false per spec, right? |
| 16:07 | <jgraham> | What about <img> (no src attribute) |
| 16:18 | <Dashiva> | MikeSmith: I use Opera dragonfly, I suppose other browsers' inspection tools are similar |
| 16:20 | <MikeSmith> | Dashiva: |
| 16:20 | <MikeSmith> | yeah |
| 16:20 | <MikeSmith> | though I think the UI in Web Inspector is maybe better than most |
| 16:21 | <MikeSmith> | or maybe it's just because I've just become most used to it |
| 16:23 | <MikeSmith> | Dashiva: I'm not sure what Norm means by "complex CSS interactions" |
| 16:23 | <Dashiva> | I assume which styles override which, and which apply at all |
| 16:24 | <MikeSmith> | OK |
| 16:27 | <MikeSmith> | Dashiva: so I would think that anything that showed the computed style and also indicated where the styles were built from (external stylesheets, embedded ones, inline styles, UA stylesheet) would make that doable |
| 16:28 | <Dashiva> | And also useful to show which styles might have applied, but were overridden by other styles |
| 16:37 | <MikeSmith> | Dashiva: yeah |
| 16:43 | <gsnedders> | Hixie: Your JS is broken :( |
| 16:43 | <gsnedders> | Hixie: "Running a script" isn't clickable. Nothing happens. No list of instances. |
| 16:43 | <Philip`> | Are there any instances? |
| 16:44 | <Philip`> | Clicking on http://www.whatwg.org/specs/web-apps/current-work/multipage/scripting-1.html#running-a-script works for me |
| 16:44 | <Philip`> | and shows one reference |
| 16:44 | <Philip`> | in Opera 10.10 |
| 16:44 | Philip` | wonders if the next versions will be 10.11 and 10.100 |
| 16:45 | gsnedders | wasn't using the multipage version |
| 16:46 | <gsnedders> | WFM in multipage |
| 16:47 | <Philip`> | Maybe your browser hasn't finished loading the single-page one yet |
| 16:47 | <gsnedders> | It has |
| 16:48 | <gsnedders> | Reproduces in 10.10, Chromium god-knows-what-version-they're-on-now-on |
| 16:57 | <drclue> | With the On2 thing done , any word on the plan for VP8 and such? |
| 16:58 | <Philip`> | None yet |
| 16:58 | <Philip`> | (The merger isn't quite done yet, it was just approved) |
| 16:58 | <drclue> | I'll be happy to see this thing done , but patient enough to see it done well |
| 17:00 | <Philip`> | Has Google ever explicitly hinted at what they want to do with it? |
| 17:01 | <drclue> | AFAIK Google has been pretty mum , but I like anyone else can conjecture that in order to be of value , they are going to have to release it somehow |
| 17:02 | <drclue> | I know he Mozilla folks are waiting for the cake to go ding too |
| 17:03 | <drclue> | Rupert Murdoch and M$ can't be too happy |
| 17:04 | <Philip`> | Hmm, http://www.on2.com/docs/on2-google-merger-faq.pdf says interesting things |
| 17:04 | <Philip`> | "based on prior discussions, we assume that if the merger is not consummated Google may undertake to build or acquire its own codec that would compete with On2 rather than pursue a significant license or other material relationship with On2." |
| 17:04 | <drclue> | Of course Adobe is probably getting the welcome wagon from beta-max |
| 17:05 | <Philip`> | "Google has informed us, and has asked that we tell our customers and partners, that it will continue to make VP8 available after the merger is consummated." |
| 17:05 | <Philip`> | Also "As previously disclosed, On2 has experienced, and expects to continue to experience, difficulty in hiring and retaining highly skilled employees with appropriate qualifications." so it sounds like On2 is valuable for its IP, not for its employees |
| 17:05 | <drclue> | That's a non-announcement announcement |
| 17:06 | <drclue> | The VP8 thing I mean |
| 17:07 | <drclue> | Google has long ago said they have a home for the engineers |
| 17:10 | <drclue> | Google will need to satisfy the content vendors with a DRM wrapper, and probably need to figure a way to keep the OGG hood ornament or otherwise get the recognition going. I'm sure they have been working on all this since last year. I'm still like a kid waiting for the opening of christmas gifts |
| 17:11 | <Philip`> | Why DRM? Lots of people seem happy with H.264 which doesn't have any |
| 17:12 | <drclue> | Well, I think there is a larger picture to CODEC life and while I personally dislike DRM , displacing H.264 will be easier with a DRM option that I hope nobody uses |
| 17:14 | <Sidnicious> | Like EPUB, which Apple's now using for its bookstore with DRM. |
| 17:15 | <Philip`> | Would being better quality than H.264, and free from patent licensing requirements, and supported by the top video web site and one of the top few browsers, not be sufficient to displace it in this context? |
| 17:16 | <gsnedders> | If Google say no-DRM or it's not on YouTube, that may have an effect. |
| 17:16 | <MikeSmith> | anybody know if there's any existing hardware support for VP8 (or VP6 or 7) on mobile-device chipsets? |
| 17:17 | <drclue> | Being 40% better than h.264 and un-encumbered are all good things, but I'm sorta curious how this will play out when Rupert has to take of the latex of the MPEG-LA patent pool members |
| 17:17 | <gsnedders> | Rupert? |
| 17:17 | <drclue> | Yes Rupert is always hiding in the corners |
| 17:19 | <drclue> | [MikeSmith] I noticed on On2's site that they do have interactions in the mobile market , but actually having the CODEC embedded widely is something that having Google's stamp on it will make occur |
| 17:19 | <Philip`> | MikeSmith: http://www.on2.com/index.php?id=439&news_id=641 - "On2 has worked closely with ARM to ensure suitability for porting to current and future ARM single and multi-core processors in mobile devices." |
| 17:19 | <Philip`> | makes it sound like they're focused on improving software support, rather than expecting hardware support |
| 17:20 | <MikeSmith> | hmm |
| 17:20 | <Philip`> | gsnedders: http://www.mckinnonsc.vic.edu.au/la/it/rupert/rupert.gif |
| 17:20 | <drclue> | Well, On2 had no market clout like h.264 , but Google changes that |
| 17:20 | <MikeSmith> | the "It's so good it doesn't need hardware support" pitch |
| 17:21 | <gsnedders> | Philip`: I was presuming that Rupert wasn't the intended one |
| 17:21 | <MikeSmith> | which I think falls into the "Let's believe that when we see it" category |
| 17:21 | <drclue> | [Phillip] That Ruppert image needs some fangs and some simpsons burns features |
| 17:21 | <Philip`> | MikeSmith: Do they make that pitch explicitly? |
| 17:22 | <Philip`> | I interpreted it as meaning "We're not going to mention that we're not going to have widespread hardware support, because that's obvious and not something to dwell on, so we're making the best of the situation by optimising for mobile CPUs" |
| 17:23 | <Philip`> | stated in a way that focuses on the positive parts but without lying about it |
| 17:24 | <drclue> | I don't think they have made the we so good we don't need hardware pitch , but I think that it will happen as a natural part of evolution |
| 17:28 | <MikeSmith> | I meant the "It's so good it doesn't need hardware support" thing is just a general response that's made any time the question about hardware support is asked about something that doesn't have hardware support |
| 17:29 | <MikeSmith> | I much hope that it is usable on average mobile devices without hardware support |
| 17:29 | <MikeSmith> | it will certainly make things easier |
| 17:30 | Philip` | wonders how much specialised hardware support actually helps H.264, compared to general purpose CPUs with decent SIMD instruction sets |
| 17:30 | <drclue> | Well , considering that OGG is basically VP3 and Flash (.flv) is basically VP6 , and that H.264 is weaker than VP8 , chances are pretty good that it would work fine in mobiles |
| 17:32 | <drclue> | Anyways , I gotta jet folks , got to get out of my bus and hook up the garbage trailer for a dump run. I think I have room for Adobe in there still :) |
| 17:35 | <AryehGregor> | Philip`, my impression is that the specialized hardware is cheaper and/or uses less power than a good enough CPU. |
| 17:36 | <Philip`> | If you already have a good enough CPU, how significant is the difference in power? |
| 17:38 | Philip` | assumes people will have plenty of other reasons for getting CPUs that are fast enough for software video decoding |
| 17:45 | <Sidnicious> | Is the behavior of the enter key in form inputs standardized? I'm not finding it. |
| 17:46 | <drclue> | A few more moments I have while I wait for the truck to get here. Anytime one has to emulate a procedure that could have been implemented in hardware , there is going to be some overhead, but a lot depends on the usage pattern. We are quickly heading towards new usage patterns that will be making more use of video but at present the battery life trade off is minimal |
| 17:47 | <gsnedders> | Sidnicious: No |
| 17:47 | <gsnedders> | Sidnicious: That's UA behaviour, nothing that should be spec'd, as it has no effect on interoperability |
| 17:48 | <AryehGregor> | gsnedders, sure it does. Authors want to know what will happen when users hit enter. |
| 17:48 | <AryehGregor> | I've seen crazy stuff like display:none submit buttons to try to get browsers to submit forms when the user hits enter. |
| 17:49 | gsnedders | wonders what happens on devices where you have no enter button |
| 17:49 | <wycats> | Hixie: aboot/ |
| 17:49 | <wycats> | ? |
| 17:49 | <wycats> | AryehGregor: sweet |
| 17:49 | <AryehGregor> | What happens to onclick events if you have no mouse? :) |
| 17:49 | gsnedders | guesses Hixie is still asleep |
| 17:50 | <Sidnicious> | gsnedders: Hurm. I'm not sure that I totally agree. Right now, testing in WebKit at least, pressing enter when an input is focused submits the form by activating the next input type submit in the document, including its value. button elements with type submit are skipped. |
| 17:51 | <AryehGregor> | I don't see why this shouldn't be specced, personally. |
| 17:51 | <AryehGregor> | This is clearly user-/author-visible behavior that affects functionality. |
| 17:51 | <wycats> | I think we should explicitly mention at data-* that it's supported by legacy browsers |
| 17:52 | <drclue> | I have always wished that there was an attribute on forms for specifying field advance keys |
| 17:52 | <Philip`> | AryehGregor: onclick isn't for clicks, it's for activation |
| 17:52 | <Philip`> | as far as I'm aware |
| 17:52 | <AryehGregor> | That's my point. |
| 17:52 | <Philip`> | Oh |
| 17:52 | <Philip`> | I missed gsnedders question |
| 17:53 | <Philip`> | so I didn't realise yours was rhetorical |
| 17:53 | <Philip`> | s//'/ |
| 17:53 | <Philip`> | wycats: Almost all of data-* isn't supported by legacy browsers |
| 17:54 | <Philip`> | (i.e. the element.dataset API) |
| 17:54 | <Philip`> | so it doesn't seem useful to promote its use now |
| 17:54 | <wycats> | wrong ;) |
| 17:54 | <wycats> | Rails added support for UJS |
| 17:54 | <wycats> | and we use data-* |
| 17:54 | <wycats> | and people constantly ask "will it work in IE6" |
| 17:54 | <Philip`> | If people are going to use custom attributes anyway, they'll know that custom attributes work, so they'll know that data-* works in exactly the same way |
| 17:55 | <Philip`> | What do they mean by "work"? |
| 17:55 | <wycats> | Philip`: exactly |
| 17:55 | <drclue> | Shovels dirt on IE6's grave and hucks a loogy |
| 17:55 | <wycats> | people are just burned by IE6 |
| 17:55 | <wycats> | and they're like "omg HTML5... that won't work for a million years" |
| 17:55 | <wycats> | they're scared we're doing something bad |
| 17:55 | <wycats> | :P |
| 17:56 | <drclue> | The HTML5 won't work for a million years thing of late seems to come mostly from the Adobe propaganda machine. |
| 17:56 | <wycats> | drclue: nah |
| 17:56 | <wycats> | people are just burned |
| 17:57 | <drclue> | I for one will cry not , when I kick IE6 off my Ubuntu desktop. Always a lot of extra work to keep it in the audiance |
| 17:57 | <Philip`> | wycats: Seems like it'd be easy for you to say that the extra markup won't affect anything at all, and scripts that want to use it will work fine as long as they use getAttribute instead of dataset |
| 17:58 | <Philip`> | and that's more likely to answer their question than a statement in the HTML spec which nobody reads |
| 17:59 | <wycats> | Philip`: I say it all the time |
| 17:59 | <wycats> | just saying it's a common question |
| 17:59 | <Philip`> | Do those people say they've read the HTML5 spec? |
| 18:01 | <drclue> | I think the really nice thing about HTML5 is that what it brings to the table will put web experiences so far ahead of where they are that IE will either have to fish or cut bait becoming to web browsing what their paint program is to desktop publishing |
| 18:02 | <wycats> | "does anybody know who was responsible for putting rails3 on HTML5? When was the decision made?" |
| 18:02 | <wycats> | we also switched to the HTML5 doctype |
| 18:02 | <wycats> | lulz |
| 18:02 | <paul_irish> | hahah |
| 18:02 | <wycats> | Philip`: I would be able to link to it |
| 18:02 | <hsivonen> | Hixie: sorry. I missed your question about object in XML. yes, it uses the same doneaddingchildren call. |
| 18:03 | <Philip`> | wycats: Seems like it'd be better to link to some kind of wiki page that can describe the current implementation status |
| 18:03 | <Philip`> | since anything the spec said would be quickly obsolete |
| 18:04 | <paul_irish> | hsivonen: you really need to subset the fonts you're using with @font-face |
| 18:04 | <Philip`> | (which I expect is why it currently says approximately nothing at all about browser support for any feature) |
| 18:04 | <paul_irish> | like really bad. :) |
| 18:04 | <wycats> | Philip`: in this case, the choice was made in order to have support with older browsers |
| 18:04 | <wycats> | but sure |
| 18:04 | <wycats> | that would be fine |
| 18:04 | <Philip`> | wycats: Well, the implementation status in what are currently considered older browsers :-) |
| 18:04 | <hsivonen> | paul_irish: work over the atlsntic and back in firefox. webkit needs a fix |
| 18:05 | <hsivonen> | works, even |
| 18:05 | <wycats> | Philip`: sure :) |
| 18:05 | <hsivonen> | Atlantic |
| 18:05 | <paul_irish> | hsivonen: you think webkit shouldnt do the invis-text? most designers implementing @font-face tend to disagree, fwiw. |
| 18:06 | <hsivonen> | paul_irish: I think the webkit behavior is bad |
| 18:06 | <paul_irish> | i think the 'right' solution is bandwidth dependent. |
| 18:06 | <Philip`> | I think it should default to a standard font while it's downloading, and then automatically morph each glyph into its new shape over a second or two once it's downloaded |
| 18:06 | <paul_irish> | but regardless you shouldnt have 2MB of fonts required to download every time i want to view your 40k page :) |
| 18:07 | <paul_irish> | Philip`: yah!! transition: glyph-morph 4s ease-in-bounce; |
| 18:07 | <wycats> | paul_irish: it's great for mobile! |
| 18:07 | <gsnedders> | Use transitions for whole reflows! What could be better! |
| 18:07 | <wycats> | paul_irish: just use the cache manifest :P |
| 18:08 | <wycats> | gsnedders: AMAZING |
| 18:23 | <drclue> | While I have some misgivings about some of the HTML5 spec (did not make a list) , overall I like it, and look forward to the hours I'll spend with it |
| 18:24 | <Philip`> | Feedback on misgivings may be appreciated :-) |
| 18:27 | <wycats> | drclue: what specifically? |
| 18:29 | <drclue> | One of these moments I'll pull the spec back in front of me and put the lead in my pencil. It just seemed that some of the spec goes off in an odd direction. I would rather see the spec go forward as it is than to keep it bottled up. My opinions on the spec were formed during an evenings read and I really did not internalize it. It seemed to try and please so many people that it sorta went over the edges of what I figure the boundries of |
| 18:29 | <Dashiva> | Oh, irony |
| 18:29 | <Dashiva> | >it sorta went over the edges of what I figure the boundries of [cut off] |
| 18:30 | <wycats> | luuuuulz |
| 18:30 | <Philip`> | Pencil? |
| 18:30 | <Philip`> | I hope you didn't print out the spec :-p |
| 18:31 | <drclue> | There are separation points between the various standards that give each it's realm and certain separations of function , data and presentation that form the minor misgivings , but like I said as a whole I like it and think it should go forward |
| 18:32 | <drclue> | Anyways , the ruck is here the trailer hooked up so I gotta jet |
| 18:41 | <AryehGregor> | That sounds like a mixed metaphor. Or, perhaps, a mixed half-metaphor. |
| 18:43 | <Dashiva> | Maybe the truck has a jet pack |
| 18:48 | gsnedders | needs to stop trying to understand Dijkstra's algorithm and go home |
| 18:48 | <Philip`> | Dijkstra's algorithm is fun |
| 18:49 | <Philip`> | You just add into your set whatever vertex is closest to your current set, and then repeat until finished - it's not that complicated really :-) |
| 18:51 | <gsnedders> | So if I know the distance of every node from the initial, and I ahve a specific destination, how do I get the shortest path from it? |
| 18:52 | <gsnedders> | I guess I just follow the path of the shortest distance |
| 18:53 | <gsnedders> | You need to know which node from each node gives the shortest distance, though |
| 18:53 | <gsnedders> | Right? |
| 18:54 | gsnedders | should just steal someone at work tomorrow with a CS degree who thinks they understand it? |
| 18:54 | <Philip`> | Every time you add a new node into the set, or reduce its cost, you store a link to its predecessor node |
| 18:54 | <gsnedders> | Yeah |
| 18:54 | <Philip`> | and when you reach the destination you trace backwards over those links |
| 18:54 | <gsnedders> | That's what I mean |
| 18:55 | <gsnedders> | At the first time you reach the destination? |
| 19:01 | <gsnedders> | Yes, I think |
| 19:01 | <gsnedders> | Philip`: Is that right? |
| 19:01 | <Philip`> | You only ever 'reach' a node when it's the closest unvisited node to your current visited nodes |
| 19:02 | <Philip`> | so when you first reach the destination, all other unvisited nodes must be further away |
| 19:02 | <Philip`> | so a path through them cannot be better |
| 19:02 | <gsnedders> | Indeed |
| 19:02 | <gsnedders> | That's what I thought |
| 19:02 | <gsnedders> | Yay |
| 19:32 | <franksalim> | who knows the latest about ByteArray in ECMAScript? |
| 19:47 | <TabAtkins> | gsnedders: Do you understand the A* algorithm? |
| 19:48 | <TabAtkins> | Dijkstra's is just A* with the heuristic always returning 0. |
| 19:48 | <TabAtkins> | (Which, since 0 will always underestimate the actual distance, is guaranteed to find a shortest path, per normal A* rules.) |
| 19:51 | <AryehGregor> | Some Wikipedia article like "Software bloat" used to have a screenshot of Microsoft Word with all menus enabled, so there was like twenty pixels to actually write in. |
| 19:51 | <AryehGregor> | I can't find it now. I want that picture. :( |
| 19:55 | <miketaylr> | AryehGregor: somewhat related, http://blogs.law.emory.edu/elsit/files/2009/02/iemess211.jpg |
| 19:58 | <MikeSmith> | franksalim: http://wiki.commonjs.org/wiki/Binary |
| 19:58 | <MikeSmith> | (if you've not already seen it) |
| 20:18 | <Philip`> | TabAtkins: I think trying to understand algorithm X by first understanding algorithm Y which is more complex than X and has X as a special case, does not seem like the easiest approach :-) |
| 20:21 | <TabAtkins> | Philip`: For some reason I found A* to be really simple and intuitive, so it was actually easier for me to understand than Dijkstra's. |
| 20:21 | <TabAtkins> | Plus, it parallel's the joke: |
| 20:23 | <TabAtkins> | "A physicist and a mathematician were at a physics conference. The physicist turned to the mathematician and said 'Man, I just can't follow all this talk about 4-dimensional stuff.'. The mathematician replied 'Ah, it's quite easy! First imagine an n-dimensional space, then just set n to 4!'" |
| 20:24 | <TabAtkins> | Though, dammit, extraneous apostrophe in the line before. |
| 20:24 | <TabAtkins> | >_< |
| 20:36 | Philip` | usually encounters A* in the context of 2D tile grids, and Dijkstra in the context of matrices over abstract algebras, so they feel like very different things even though they're not |
| 21:08 | <Sidnicious> | I was just most of the way through writing to the list when I realized that that form submission issue I was asking about is already in the spec <http://www.whatwg.org/specs/web-apps/current-work/multipage/association-of-controls-and-forms.html#implicit-submission>; :) |
| 21:10 | <Sidnicious> | I'm a bit miffed that there's no way to specify a default button, though. |
| 21:12 | <franksalim> | MikeSmithX, thanks. I had seen that page. I didn't realize that was now the most active discussion on the topic |
| 21:18 | <Sidnicious> | Actually, other than backwards compat, is there a reason there should be a default button at all? |
| 21:27 | <TabAtkins> | Sidnicious: Is there a reason there *shouldn't* be a default button? I use the fact that forms will submit with their default button all the time when I'm cruising the web. |
| 21:29 | <Sidnicious> | TabAtkins: Submit, yes, use a default button, no. Especially if default means first-in-tree-order, not the-one-that-the-developer-decides-is-default. |
| 21:29 | <TabAtkins> | And if the developer doesn't know how to set a default/doesn't know you can/doesn't care/doesn't exist anymore? |
| 21:30 | <TabAtkins> | I see the appeal of being *able* to specify a default button, but you shouldn't *rely* on such a specification when we already have a perfectly serviceable default behavior. |
| 21:30 | <Philip`> | Authors will be less surprised if everyone submits the form via the button - it stops them forgetting about the case where the user hits enter instead of clicking |
| 21:31 | <Philip`> | (and so e.g. it works as expected if they have <input type=submit disable> in the form) |
| 21:31 | <Sidnicious> | I like the behavior of forms without submit buttons: just submit the darn thing without pretending that a button was clicked. |
| 21:32 | <Sidnicious> | The current behavior just takes away information, makes it impossible to tell the difference without nasty JavaScript hacks. |
| 21:35 | <TabAtkins> | What difference? Between the user hitting Enter, and them clicking the first-submit-input-in-tree-order? |
| 21:35 | <Sidnicious> | Correct. |
| 21:36 | <TabAtkins> | Why do you care about that difference? (I understand caring about *which* submit button was pressed in javascript, which you've brought up before.) |
| 21:38 | <Sidnicious> | Google has the "Google Search" button on the left, and the "I'm Feeling Lucky" button on the right of their search page, but what if the designer had wanted them the other way around? |
| 21:39 | <Sidnicious> | The Mac OS traditionally has the default button in a form on the right, with the alternate buttons filling in to the left. |
| 21:39 | <Sidnicious> | Linux, too. |
| 21:40 | <Hixie> | default button will likely be in a future version |
| 21:40 | <Hixie> | baby steps! |
| 21:42 | <Sidnicious> | Heh, all right. As long as you consider not assuming a default button when it isn't specified. |
| 21:45 | <beowulf> | evening |
| 21:49 | <TabAtkins> | Sidnicious: I don't understand your objections. Again, why shouldn't you assume a default button when it's not specified explicitly? |
| 21:50 | <Philip`> | Sidnicious: Seems unlikely the current behaviour could change, because it'd break compatibility |
| 21:51 | <Sidnicious> | I know. |
| 21:56 | <Sidnicious> | TabAtkins: I just don't think it's author-friendly behavior. Being handed the value of a button that wasn't clicked is not intuitive. |
| 21:57 | <Sidnicious> | I can't find any benefit to that behavior. |
| 21:57 | <TabAtkins> | Really? I rely on submit buttons to signal to scripts that they're receiving the result of a form submission, not a normal page navigation. |
| 21:58 | <TabAtkins> | At the moment, if I have multiple submits I just make sure that the best default one comes first, though I would appreciate being able to specify the default explicitly. |
| 21:59 | <annevk> | Hixie, oh please do write HTTP5 |
| 21:59 | <TabAtkins> | Aren't we very nearly just writing Internet5 then? |
| 21:59 | <othermaciej> | TCP5! |
| 21:59 | <Sidnicious> | TabAtkins: Really? IE 6 and 7 don't submit a default button. |
| 22:00 | <Sidnicious> | So if you're relying on that... |
| 22:00 | <TabAtkins> | !_! Really? In that case, I've got some potentially broken forms. |
| 22:00 | <Sidnicious> | http://dev.deeptechinc.com/sidney/share/submit_button_roulette.html |
| 22:00 | <Sidnicious> | (test page I've been using to test different browsers' current behavior) |
| 22:01 | <Hixie> | othermaciej: i was reading about TCP recently... apparently the spec for TCP is pretty far from actual implementations too, much like HTTP and HTML4 are |
| 22:01 | <TabAtkins> | >_< |
| 22:01 | <Hixie> | othermaciej: so maybe we do need a TCP5! |
| 22:01 | <Hixie> | not my expertise though |
| 22:01 | <Hixie> | so i ain't doing it |
| 22:01 | <Sidnicious> | TabAtkins: Actually, wait, that was a different behavior. |
| 22:02 | <Sidnicious> | No versions of IE submit a default button :) |
| 22:02 | <othermaciej> | Hixie: or we could just move to SCTP |
| 22:02 | <othermaciej> | (though I don't actually know if SCTP reflects reality) |
| 22:02 | <TabAtkins> | If no versions of IE do that, then why is it specced, precisely? |
| 22:02 | <Hixie> | othermaciej: SCTP isn't backwards compatible, is it? |
| 22:03 | <othermaciej> | Hixie: fair enough, it would be an XHTML2 type solution |
| 22:03 | <TabAtkins> | And... yup, no version of IE sends a submit button if you hit Enter. |
| 22:03 | <TabAtkins> | So, um, Hixie, how was that the behavior that made it into html5? |
| 22:04 | <Necrathex> | what happens with just <button>s and no <input type="submit"> ? |
| 22:04 | <Hixie> | TabAtkins: hm? |
| 22:04 | <TabAtkins> | IE doesn't submit a default submit button if you submit a form with Enter. The spec says to submit the first submit button you find in tree order. |
| 22:05 | <Hixie> | yes? |
| 22:05 | <TabAtkins> | So, I'm wondering how that behavior got specced? Did everyone else submit the default submit input or something? |
| 22:06 | <Sidnicious> | Necrathex: Same thing in IE8. 6 and 7 send the name and content (not value) of <button>s and <button type=button>s all the time, treat <button type=submit>s like <input type=submit> as far as I can tell. |
| 22:06 | <Hixie> | TabAtkins: i disagree with the first statement |
| 22:06 | <Hixie> | TabAtkins: <!DOCTYPE html>...<form><input><input type=submit></form> |
| 22:06 | <Hixie> | focus the input |
| 22:06 | <Hixie> | hit enter |
| 22:06 | <Hixie> | the form submits |
| 22:07 | <TabAtkins> | Yes it does. But if you put a name on the submit, it doesn't show up in the submitted data. |
| 22:07 | <Hixie> | oh, i see |
| 22:08 | <Hixie> | that was so that disabling the first button stopped submission |
| 22:09 | <TabAtkins> | Was there any compat testing to make sure that pages don't depend on nothing being submitted there? |
| 22:09 | <TabAtkins> | (I happen to have many scripts which *do* depend on the default input getting included in the submitted data, so this is good for me, but still...) |
| 22:09 | <Hixie> | opera and firefox do what the spec says, so i assume so |
| 22:12 | <Necrathex> | Sidnicious: i don't see the name of any button or submit when i press enter in IE with your test page |
| 22:12 | <Necrathex> | only appears when i click it |
| 22:16 | <TabAtkins> | Necrathex: That's the expected behavior, yes. |
| 22:17 | <annevk> | TabAtkins, whether or not the trickery is justified, the proposal is for a pseudo-element, not pseudo-class |
| 22:18 | <TabAtkins> | annevk: Yes? That's what's appropriate here, after all. |
| 22:19 | <annevk> | I think :placeholder::value or some such would make more sense |
| 22:19 | <annevk> | it's the same box |
| 22:19 | <annevk> | after all |
| 22:21 | <TabAtkins> | annevk: I suppose if you think of the input itself being in a placeholder state, that makes sense. Most of us are thinking of the placeholder itself being an element *within* the input. |
| 22:24 | <annevk> | "most of us", really? |
| 22:24 | <TabAtkins> | Apparently. |
| 22:24 | <annevk> | adele from webkit disliked changing it to a pseudo-element |
| 22:24 | <TabAtkins> | (Going by the talk on the list.) |
| 22:24 | <annevk> | most people on the web when Apple first announced this feature didn't say it should've been a pseudo-element... |
| 22:25 | <TabAtkins> | I think most people on the web don't understand the difference between pseudoclasses and pseudoelements. ^_^ |
| 22:25 | <TabAtkins> | On the other hand... the ::value pseudoelem is from the UI Module, right? |
| 22:26 | <TabAtkins> | It might indeed be nice to just lean on that, and allow full styling of the input based on whether or not it's in the placeholder state. |
| 22:27 | <Sidnicious> | So, silly question to Hixie et. al... I like the versionless doctype, but what do you plan to do when, some day, you want to make a backwards-incompatible change to the spec? |
| 22:27 | <TabAtkins> | Sidnicious: Either *don't* make a backwards-incompatible change, or fully break compat and make a new language. |
| 22:27 | <Sidnicious> | Don't make a backwards-incompatible change *ever*? |
| 22:28 | <TabAtkins> | Ideally, yes. In practice, don't make backwards-incompatible changes that would affect enough pages to be a problem for implementors. |
| 22:28 | <annevk> | in practice we can make certain changes |
| 22:28 | <annevk> | HTML is pretty flexible |
| 22:29 | <Hixie> | Sidnicious: we can't make changes that are backwards-incompatible with deployed content anyway |
| 22:30 | <Hixie> | Sidnicious: (and we don't need to worry about backwards-incompatible changes that don't break any deployed content, so we just make them) |
| 22:34 | <annevk> | time to gome; nn |
| 22:54 | <Hixie> | oops, i just repeated exactly what sicking said, but in less eloquent terms |
| 22:54 | <Traveler1> | Hello. |
| 22:54 | <Traveler1> | Do you guys recommend to add a header tag around the hgroup? |
| 22:55 | <sicking> | Hixie: no worries, it happens all the time ;) |
| 22:55 | <sicking> | just kidding |
| 22:55 | <Traveler1> | <hgroup> <h1>title</h1> <h2>subtitle</h2> </hgroup> |
| 22:55 | <sicking> | Hixie: where at? |
| 22:55 | <Traveler1> | Should I add an extra header tag? |
| 22:55 | <AryehGregor> | If you don't need it, don't add it. |
| 22:56 | <AryehGregor> | Most <hgroup>s aren't part of a <header> |
| 22:56 | <AryehGregor> | . |
| 22:56 | <Traveler1> | Hm |
| 22:56 | <TabAtkins> | Yeah, <header> isn't needed most of the time. It's just for when you are grouping a bunch of stuff into a "header". |
| 22:57 | <AryehGregor> | Generally speaking, don't use extra tags unless you have some particular reason. It's largely pointless. |
| 22:57 | <Traveler1> | Right. |
| 22:57 | <TabAtkins> | On the other hand, if you're going to end up putting in a <div id=header> or similar, might as well use <header> there. |
| 22:57 | <AryehGregor> | Exactly. |
| 22:57 | <Hixie> | sicking: the websocket thing |
| 22:57 | <Traveler1> | No way, as AryehGregor said, no point. |
| 22:58 | <TabAtkins> | I'm not saying definitely do it. Just if you are planning on grouping things like that, then use <header> rather than a <div>. <header> was introduced *precisely* to replace that sort of <div> usage. |
| 22:59 | <AryehGregor> | So for instance, I use a <header> on aryeh.name because I want to have possibly a few headings at the top of the page, with a particular background and such, so I need an extra element. So I made it <header> instead of <div id=header> or something. |
| 22:59 | <TabAtkins> | Hmm, looks like I can swap from using this single display table (to get a good two-column layout) to using display:table pretty painlessly. It handles itself pretty well in IE7 when I patch it with float. |
| 23:00 | <Traveler1> | What about if I have two links on the header? Should I use a nav tag? Btw the links are just for viewing the page in english or another lang |
| 23:01 | <Traveler1> | nvm |
| 23:01 | <TabAtkins> | If they are navigation for the site, use <nav>. If not, don't. ^_^ |
| 23:59 | <TabAtkins> | Man, why didn't I do this earlier? The layout table is now officially banished from my company's site. I just need to push a bit further and do the new sectioning elements rather than divs, and I'll be happy leaving this company. |
