| 05:22 | <jamesr> | anyone around? looking at a canvas patch for webkit and it looks like firefox does the wrong thing for (stroke|fill)Text when maxWidth is specified as 0 |
| 05:22 | <jamesr> | it does the draw as if the value was not specified at all |
| 05:24 | <jamesr> | wondering if it's just an oversight or if it's difficult to distinguish between not-specified numbers and '0' in that idl implementation |
| 05:24 | <roc> | I think it's our bug |
| 05:24 | <roc> | if (aMaxWidth > 0 && totalWidth > aMaxWidth) { |
| 05:26 | <jamesr> | opera gets 0 right but not negative numbers |
| 05:26 | <jamesr> | roc: guessing aMaxWidth is already coerced to a numerical type at that point? |
| 05:31 | <roc> | yeah, but we can get around it |
| 05:31 | <roc> | you have to be able to, for drawImage anyway |
| 05:32 | <jamesr> | aight |
| 05:33 | <jamesr> | negative values throw a weird exception |
| 05:41 | <jamesr> | Philip`: yt? |
| 05:43 | <jamesr> | Philip`: if you get a chance to see this i think you should add some canvas tests for negative maxWidth, NaN maxWidth, and for globalCompositeOperation + text drawing. opera fails on the first, the second seems unspecified (although i might have missed a step in the conversion routine that fixes it) and the last fails pretty hardcore in webkit |
| 05:43 | <jamesr> | or if i can figure out the harness maybe i could write some tests |
| 05:46 | <jamesr> | roc: or do you happen to know how a specified value of NaN is supposed to behave when passed in to a function specified as taking a double in IDL? |
| 05:47 | <roc> | That depends on the function |
| 05:47 | <jcranmer> | you throw an exception saying "double check your damn math" |
| 05:47 | <roc> | canvas generally just ignores the call |
| 05:47 | <roc> | we've had some debates about this inside and outside Mozilla |
| 05:47 | <roc> | I tend to favour ignoring the call, for a couple of reasons |
| 05:48 | <roc> | 1) it's the bottom of the "this site 'works' in browser X, but not Firefox, you suck" slippery slope |
| 05:48 | <jamesr> | iirc olliej prefers that as well, his argument being that it's easy to get a NaN on an intermediate step of some complex calculation and it's harsh to throw an exception as that'll typically kill all draw calls that would happen afterwards |
| 05:48 | <roc> | that was my point #2 :-) |
| 05:49 | <roc> | that is especially true for coordinate and matrix parameters |
| 05:49 | <jamesr> | when you 'ignore the call' do you completely ignore it or still apply the globalCompositeOperation (i.e. do you clear out everything if the gCO was "copy")? |
| 05:49 | <roc> | we ignore it completely |
| 05:49 | <roc> | I could go either way on that |
| 08:26 | <annevk-cloud> | dglazkov, what is the difference between a DOM element and DOM content? |
| 08:26 | <annevk-cloud> | dglazkov, http://wiki.whatwg.org/wiki/Component_Model_Behavior_Attachment could use some examples :) |
| 09:27 | <boblet> | crazy q: is it possible to disable Moz’s HTML5 parser via JS for a site? |
| 09:29 | <annevk> | no |
| 09:30 | <boblet> | cool |
| 09:40 | <annevk> | Ms2ger, http://lists.w3.org/Archives/Public/www-archive/2011Jul/0003.html |
| 09:40 | <annevk> | ... and http://lists.w3.org/Archives/Public/www-archive/2011Jul/0004.html |
| 09:50 | <boblet> | annevk: the questions we get at H5D, itellya ;) |
| 09:52 | <annevk> | I read a sitepoint discussion this morning where zcorpan was in discussion with a bunch of people believing HTML5 will fail soon |
| 09:56 | <Onderhond> | What was their alternative? |
| 10:01 | <annevk> | Something would come along... |
| 10:02 | <annevk> | The mutation events thread be long, geez |
| 10:38 | <annevk> | http://tools.ietf.org/html/draft-weber-iri-guidelines looks somewhat promising |
| 10:38 | <annevk> | though replacing the whole set of RFCs with a standard about URLs still makes more sense to me |
| 10:50 | <MikeSmith> | hmm, yeah, that doc looks good |
| 10:50 | MikeSmith | is behind on reading the public-iri list |
| 10:56 | <MikeSmith> | annevk: I added Cross-Origin Resource Embedding Exclusion to platform.html5.org |
| 10:57 | <MikeSmith> | but I think that spec should just be renamed "The From-Origin header" |
| 10:57 | <MikeSmith> | hmm, I guess I should also add CSP |
| 11:01 | <annevk> | yeah I gues |
| 11:02 | <annevk> | s |
| 11:02 | <annevk> | The long term plan is to take "fetch" from HTML, CORS, and COREE; and merge them all into a single document about fetching resources |
| 11:03 | <MikeSmith> | yeah |
| 11:03 | <MikeSmith> | that would be good |
| 11:50 | <annevk> | not sure if sylvia is here, but <? does not work in HTML |
| 11:51 | <Ms2ger> | How unfortunate. |
| 13:06 | <Ms2ger> | "4. Pre-processing Arbitrary Unicode Strings" |
| 13:06 | Ms2ger | wonders if that spec will ever manage to become a RFC |
| 13:07 | <annevk> | written by the chair of the group if I'm not mistaken |
| 13:07 | <annevk> | should have a chance |
| 13:39 | <annevk> | Anyone used http://prizes.org/ for getting people to write specifications? |
| 13:41 | <Onderhond> | Long shot (and off-topic). |
| 13:41 | <Onderhond> | But I remember a news site that used the right site of the browser to give an indication of content below the fold. |
| 13:41 | <Onderhond> | * right side |
| 13:41 | <Onderhond> | Any idea if this is a common pattern, what it's generally called and where to find it? :) |
| 13:42 | <annevk> | NY Times has/had such an overlay on the bottom right, not sure what it was for anymore |
| 13:45 | <Onderhond> | Shame they change their site a lot, I'll check archive.org :) |
| 15:07 | <Onderhond> | Is it feasible to use input[type=range] for a fixed set of values? (radius around a point of a map: 1km, 5km, 15km, 25km, 50km, 75km) |
| 15:08 | <annevk> | input type=radio and select are for that |
| 15:09 | <Onderhond> | Currently there's no way to define a fixed set on values. |
| 15:09 | <Onderhond> | Myeah, figured so. |
| 15:10 | <Onderhond> | Sadly "those in power" prefer a fancier (and screen estate-reducing) input type. |
| 15:10 | <annevk> | need shadow DOM |
| 15:10 | <annevk> | or XBL |
| 15:11 | <annevk> | or whatever will solve that |
| 15:11 | <bga_> | but http://www.whatwg.org/specs/web-forms/current-work/#the-datalist |
| 15:12 | <wilhelm> | Your use case makes sense, but you'll need a bit of additional logic. <input type='range' min='0' max='5' step='1'> plus a script that converts indexed values of 0-5 to 1-75km. |
| 15:14 | <Onderhond> | Hmmm, sounds reasonable too. |
| 15:17 | <annevk> | bga_, that's suggestions, you cannot limit the control to those values |
| 15:18 | <bga_> | ah |
| 15:18 | <Onderhond> | data-fixedVals="1,5,15,25,50,75" and use js to snap to those values ... I guess that would be an option ? |
| 15:19 | <annevk> | data-fixed-vals |
| 15:19 | <annevk> | becomes dataset.fixedVals |
| 15:19 | <annevk> | and that could work, bit hacky |
| 15:20 | <Onderhond> | kay, thanks for the input :) |
| 15:20 | <Onderhond> | (hah pun!) |
| 18:11 | <karlcow> | http://www.innoq.com/blog/st/2010/03/uddi_rip.html |
| 18:25 | <TabAtkins> | I think I just need to write off everyone with "Leif" in their name as bad. |
| 18:27 | <othermaciej> | you probably shouldn't judge a person as "bad" based on their participation in a standards group in any case... |
| 18:27 | <Ms2ger> | I've only needed to do that with one |
| 18:27 | <Ms2ger> | Or I didn't realize there were different people |
| 18:28 | <TabAtkins> | othermaciej: I'm not. I'm judging them by their name. |
| 18:28 | <TabAtkins> | And I'm doing it to *all* people with "Leif" in their name. |
| 18:28 | <TabAtkins> | 'cause I'm racist. |
| 18:28 | <TabAtkins> | namist. |
| 18:29 | Ms2ger | fires TabAtkins |
| 18:29 | <TabAtkins> | YOU CAN'T FIRE ME, I QUIT MOZILLA |
| 18:30 | TabAtkins | thinks he remembers that Ms2ger is Moz. |
| 18:30 | <annevk> | Is that actually just for W3C purposes or are you meaningfully employed? |
| 18:30 | <Ms2ger> | Just W3C red tape |
| 18:36 | <miketaylr> | aww TabAtkins not all Leif's are bad |
| 18:37 | <miketaylr> | s/Leif's/Leifs/ |
| 18:37 | <Ms2ger> | leiftaylr? |
| 18:37 | <TabAtkins> | HE'S A CRYPTO-LEIF |
| 18:37 | <TabAtkins> | GET HIM |
| 18:38 | <miketaylr> | indeed leif taylor http://www.flickr.com/photos/11691204@N00/5701898006/in/photostream |
| 18:38 | Ms2ger | was never really fond of the name "Tab" either |
| 18:38 | miketaylr | hides his babby from tab |
| 18:38 | <TabAtkins> | I challenge you to find Tabs that are worth vilifying. |
| 18:38 | <TabAtkins> | Besides my dad. |
| 18:42 | <Hixie> | right, what's next |
| 18:43 | <Hixie> | microdata, the command api thing, rtc, and dialogs are on my list |
| 18:43 | <Hixie> | am i missing anything? |
| 18:43 | <Ms2ger> | Meh, dialogs can wait |
| 18:44 | <Hixie> | let's start with microdata |
| 18:56 | <Hixie> | you have to love the way the w3c on the one hand cries about how forks are bad and that's why we shouldn't use a free license, and then on the flip side they seem to have no problem making forks. |
| 18:57 | <AryehGregor> | I thought we discussed. Forks of W3C specs are what's bad, regardless of whether the W3C spec happened to come first or second. |
| 18:57 | <AryehGregor> | s/discussed/discussed this/ |
| 18:57 | <Hixie> | clearly |
| 18:58 | <Hixie> | and if all the forks occur within w3.org, it's fine too |
| 18:58 | <AryehGregor> | Yes, because then they (the good guys) control all of them, so it's okay. |
| 18:58 | <Hixie> | so tired of dealing with people who put process ahead of the web |
| 18:58 | <AryehGregor> | The only problem is when untrusted evil people make forks of W3C specs. |
| 18:58 | <Ms2ger> | Also, generalizations |
| 19:02 | <Hixie> | actually i should rephrase that. it's not so much that people put process ahead of the web. It's that they think the process helps the web, despite evidence to the contrary. |
| 19:03 | <AryehGregor> | I was going to say something like that, but I've argued with you enough already about the motives of people who disagree with us. |
| 19:03 | <AryehGregor> | Lots of people think procedures are good and anarchy is bad, in web standards as in many other things. |
| 19:04 | <Ms2ger> | Given that people do care about the process, I'm not sure saying "whatever, publish it as a rec, svg got away with it" is helpful |
| 19:05 | <AryehGregor> | It comes up a lot in politics too. And the administration of large organizations. |
| 19:06 | <Hixie> | Ms2ger: the process as a whole isn't being helpful |
| 19:07 | <Hixie> | Ms2ger: the right answer is to change it or drop it, but nobody at w3c that i've spoken to seems interested in even discussing considering it |
| 19:10 | <AryehGregor> | Because to change the process so massively would be incredibly hard, because of the process. |
| 19:10 | <AryehGregor> | Since you'd have to go through the process to change the process. |
| 19:10 | <AryehGregor> | So they only tolerate superficial patches that don't really solve the problems. |
| 19:10 | <Hixie> | not doing something because it would be hard is possibly the _worst_ reason not to do it |
| 19:10 | <AryehGregor> | Hey, you're preaching to the choir here. |
| 19:11 | <Hixie> | sorry, just frustrated :-) |
| 19:11 | <AryehGregor> | But realistically, the AC is reactionary and will never vote to upturn all the things we need to upturn unless they have a gun held to their head. |
| 19:11 | <Hixie> | ok, i'm done reading my e-mail, frustration will now dissipate |
| 19:11 | <Hixie> | foolip: you around? |
| 19:11 | <Ms2ger> | (which I wouldn't recommend) |
| 19:12 | <AryehGregor> | Which wouldn't you recommend, holding a gun to their head? |
| 19:12 | <Ms2ger> | Yeah |
| 19:12 | <AryehGregor> | Even then they might not be willing to accept all the changes we need, because many of them run directly against their interests as organizations. E.g., letting implementers run everything. |
| 19:12 | <Hixie> | what would you recommend? (personally i've stopped caring enough to even consider finding something to convince them to change, so there's no risk of my doing what AryehGregor suggests) |
| 19:13 | <AryehGregor> | As I've said, I don't think they'll ever change, I think the only way out is to abandon ship. Strikingly similar to TBL's account of how he split from the IETF. |
| 19:13 | <AryehGregor> | That was in large part because they spent ages arguing about stupid procedural issues. |
| 19:13 | <Ms2ger> | You can't publish specs from a prison, so I'd rather you didn't end up there |
| 19:14 | <Hixie> | i meant what would Ms2ger recommend :-) |
| 19:14 | <Ms2ger> | I'm fine with convincing MS to join the cabal myself |
| 19:14 | <Philip`> | I thought people often wrote books in prison, so why not specs? |
| 19:15 | <Hixie> | Philip`: no internet access |
| 19:15 | <annevk> | Wait, WebApps is going to patch Ian's drafts and move them ahead independently of the WHATWG drafts? |
| 19:15 | <Ms2ger> | Philip`, publish, not write |
| 19:15 | <annevk> | That is going to be interesting |
| 19:15 | <AryehGregor> | Hixie, who says? I bet prisoners in low-security prisons get Internet access for good behavior. |
| 19:15 | <Hixie> | annevk: "interesting" isn't the word i would pick :-P |
| 19:15 | <annevk> | Been on vacation btw? |
| 19:15 | <AryehGregor> | Actually, I bet they'd love it if you could work productively from prison and make a decent salary. You'd think so, right? Must be better than stamping out license plates or whatever. |
| 19:15 | <Philip`> | http://www.ehow.com/how_6629174_write-book-prison-published.html has a handy step-by-step guide |
| 19:16 | <Hixie> | annevk: four day weekend in the US |
| 19:16 | <Ms2ger> | AryehGregor, sorry, that wouldn't get through the prison's process |
| 19:17 | <AryehGregor> | Okay, so are implementers going to come out and say they'll ignore the W3C specs if they patch them independently of Hixie? That's really what needs to happen to force the W3C to act sane, implementers need to take a stand. |
| 19:17 | <annevk> | Hixie, aah, nice |
| 19:17 | <annevk> | I'm gonna have one of those starting somewhere tomorrow, in Berlin :) |
| 19:17 | <Ms2ger> | AryehGregor, browsers will match the W3C spec more closely than the WHATWG one, no? |
| 19:17 | <Hixie> | AryehGregor: in this particular case, the implementations are already doing what the patch is intended to do |
| 19:17 | <Hixie> | AryehGregor: i just haven't gotten around to implementing that fix because it's of trivial importance |
| 19:17 | <AryehGregor> | Ah, I see. |
| 19:18 | <AryehGregor> | Well, I wouldn't say it's trivial, the spec doesn't match reality. It's just not blocking anything important. |
| 19:18 | <Hixie> | the whatwg spec will eventually follow suit; in the meantime, god knows what'll happen if i fix actually important bugs |
| 19:18 | <Hixie> | AryehGregor: right |
| 19:18 | <AryehGregor> | Well, whatever. |
| 19:19 | <Ms2ger> | So, how much time would have been saved if you just did that a couple of weeks ago instead of complaining about the W3C? |
| 19:19 | <AryehGregor> | I'd like to see them try that in the HTMLWG. |
| 19:19 | AryehGregor | concurs with Ms2ger that it would be simpler if you had just prioritized this stuff from the beginning to shut people up, since it's a legitimate request |
| 19:19 | <AryehGregor> | You prioritize chairs' requests in the HTMLWG, why not WebApps too? |
| 19:19 | <Hixie> | i'm not going to start prioritising things in the order that people complain about them |
| 19:19 | <Hixie> | that way lies madness |
| 19:19 | <Ms2ger> | This way, too |
| 19:20 | <AryehGregor> | Well, it might be madness, but it would save people a lot of trouble. |
| 19:21 | <Hixie> | what would save people trouble is just not worrying about the process |
| 19:21 | <AryehGregor> | You only have to prioritize complaints by chairs and W3C administrative people, and only if they sound threatening. |
| 19:21 | <Hixie> | this entire issue only matters because Art has suddenly decided it should be in LC |
| 19:21 | <AryehGregor> | Well, yes, but given that we aren't going to get that, no point in worrying about it. |
| 19:41 | <annevk> | smaug____, you could allow it only to be called from a top level browsing context |
| 19:42 | <smaug____> | annevk: rph ? |
| 19:42 | <smaug____> | yeah, that could be one, ugly, solution |
| 19:45 | <TabAtkins> | If someone posts a bug asserting something is missing, and it's not missing, is that WORKSFORME or INVALID? |
| 19:46 | <AryehGregor> | TabAtkins, http://dev.w3.org/html5/decision-policy/decision-policy-v2.html#bug-resolutions |
| 19:46 | <AryehGregor> | WORKSFORME in the HTMLWG. |
| 19:46 | <AryehGregor> | INVALID is only for total garbage, or if the filer withdraws it. |
| 19:46 | <TabAtkins> | Done, thanks. |
| 19:49 | <annevk> | smaug____, doesn't seem so ugly to me |
| 19:50 | <smaug____> | annevk: at least it is not backwards compatible, if currently there isn't such limitation |
| 19:52 | <annevk> | I doubt you'd even want iframe content to be allowed to ask for such permissions |
| 19:52 | <annevk> | E.g. when that happens for geolocation you want to present special UI as well |
| 20:07 | <dbaron> | Hixie, fyi, it's a three day weekend in the US, and a 4 day weekend in Google-land |
| 21:05 | <AryehGregor> | Hixie, should I resolve WF3 stuff LATER, move it to HTML.next, what? |
| 21:05 | <AryehGregor> | Stuff in the "we want this, but not until there's better WF2 support" category. |
| 21:06 | Ms2ger | thinks that's what target milestones should be for |
| 21:08 | <AryehGregor> | The latest target milestone is "Rec". Also, I don't know who's supposed to set those. |
| 21:08 | <Ms2ger> | Yeah, they're broken, IMO |
| 21:09 | <timeless> | AryehGregor: generally i find RESOLVED LATER = disaster |
| 21:09 | <timeless> | it's better to just use a TM to get things off your plate |
| 21:09 | <AryehGregor> | TM? |
| 21:10 | <Ms2ger> | Target milestone |
| 21:10 | <AryehGregor> | Oh. Well, it all depends on how Hixie's workflow operates here. |
| 21:10 | <AryehGregor> | But anyway, we don't have a suitable milestone. |
| 21:10 | <AryehGregor> | We'd have to move it to HTML.next. |
| 21:11 | <Ms2ger> | That's an even bigger disaster, but oh well :) |
| 21:12 | <jamesr> | AryehGregor: can you just leave the bugs open? |
| 21:12 | <AryehGregor> | jamesr, that's what I'm asking Hixie. I'll do whatever he wants. |
| 21:12 | <Ms2ger> | That is what you're being paid for after all, no? :) |
| 21:12 | <AryehGregor> | I'm pretty sure that HTMLWG procedure requires they be resolved at some point, though, or the component is changed. |
| 21:12 | <AryehGregor> | No, I'm not being paid for this, alas. |
| 21:13 | <jamesr> | AryehGregor: leaving it open for now != leaving it open forever |
| 21:14 | <AryehGregor> | The thing here is that we don't know when exactly we'll want to do it. It depends when browsers have good WF2 support. |
| 21:14 | <Ms2ger> | Never |
| 21:15 | <Ms2ger> | :) |
| 21:25 | <Hixie> | AryehGregor: either mark them WF3 but otherwise leave them, or mark them WF3 and move them to LATER if you think there's no chance browser vendors would want to work on them soon |
| 21:25 | <AryehGregor> | Hixie, okay. |
| 21:25 | <Hixie> | if they're _really_ pie in the sky and things that we shouldn't even consider soon, I guess move them to HTML.next |
| 21:26 | <Hixie> | but let's not move things into that bucket if there's a chance we might have to solve them sooner rather than later |
| 21:32 | <AryehGregor> | Okay. |
| 21:32 | <AryehGregor> | I'll leave it open. |
| 21:32 | <zewt> | "If the applications you are completing require electronic signature you may be asked to download or update Java version 1.4.2 or higher after you sign in." fffff |
| 21:34 | <moo-_-> | "may be" |
| 21:34 | <moo-_-> | zewt: you still have your chances! |
| 21:34 | <moo-_-> | zewt: run for it |
| 21:34 | <zewt> | there will be no survivors |
| 21:35 | <moo-_-> | zewt: I so love Crockford's Javascript presentation where he dissed Java |
| 21:37 | <zewt> | at least if it required Flash i already have that installed ... but almost nobody uses Java on webpages anymore so I don't |
| 21:38 | <Hixie> | hsivonen: yt? |
| 21:44 | <TabAtkins> | zewt: It's a very bad idea to turn on Java in your browser anyway, due to the attack vectors it opens. |
| 21:44 | <TabAtkins> | (I only have Java on my comp at all for Minecraft.) |
| 21:44 | <zewt> | that's what VMware is for |
| 21:44 | <moo-_-> | TabAtkins: I really don't care about attack vectors as Linux user |
| 21:44 | <moo-_-> | TabAtkins: but I care when my browser freezes for 60 seconds |
| 21:51 | <Hixie> | moo-_-: you don't care if your machine gets owned? :-) |
| 21:52 | <TabAtkins> | Java attack vectors are not windows-specific. What do you once you exploit the vector is platform-specific, but the actual holes opened up by Java are often cross-platform. |
| 21:52 | <TabAtkins> | s/are not/are often not/ |
| 21:52 | <jamesr> | write once, owned everywhere |
| 21:52 | <Hixie> | write once, debug everywhere |
| 21:53 | <jamesr> | actually a number of java sploits are windows-specific, but that's a factor of where the targets are, not where the exploits are |
| 21:59 | <Hixie> | anyone know if schema.org implementations support itemref=""? |
| 22:01 | <annevk> | AryehGregor, when mutations are sorted out that'll be sorted out |
| 22:02 | <Hixie> | looks like google's richsnippet testing tool does, at least |
| 22:02 | <AryehGregor> | annevk, which part? |
| 22:02 | <Hixie> | though it exposes it in the data model for some reason |
| 22:02 | <annevk> | AryehGregor, but basically we don't have to define the methods in terms of each other to add the mutation stuff |
| 22:02 | <AryehGregor> | No, but it makes things simpler. |
| 22:02 | <annevk> | Hixie, from what I read on schema.org they have proprietary extensions to e.g. the <time> element |
| 22:02 | <AryehGregor> | That way there's only one point where we need to define these things. |
| 22:03 | <annevk> | AryehGregor, if they all use a single callback you have that too |
| 22:03 | <Hixie> | annevk: i think those were mistakes, dunno if they fixed them or not |
| 22:03 | <annevk> | Hixie, and they use some kind of extension model that is incompatible with the API |
| 22:03 | <AryehGregor> | annevk, that'd be good enough for me. |
| 22:03 | <Hixie> | annevk: that, there was discussion of on whatwg |
| 22:03 | <annevk> | yeah |
| 22:03 | Hixie | is still thinking of whether to just replace <time> with something else for "computer representation" |
| 22:03 | <AryehGregor> | But it would still be more complicated AFAICT. |
| 22:03 | <Hixie> | btw speaking of the mutation stuff, what's the status on that? are we close to a solution everyone likes? |
| 22:04 | <annevk> | Hixie, no consensus whatsoever it seems |
| 22:04 | <Hixie> | k |
| 22:04 | <TabAtkins> | Hixie: I end up wanting to use <time> with a unix timestamp anyway, because it's easier to work with that than an rfc timestamp. |
| 22:04 | <annevk> | Hixie, there's like very simple mutation listeners versus complex ones that keep some kind of diff-state around |
| 22:05 | <Hixie> | TabAtkins: yeah |
| 22:05 | <Hixie> | annevk: i've no opinion on the matter, i'm just waiting for someone to make a decision :-) |
| 22:05 | <annevk> | AryehGregor, I don't really like defining one method in terms of another |
| 22:05 | <annevk> | Hixie, me too |
| 22:05 | <annevk> | Hixie, my plan is to put them in DOM Core, whatever "them" is |
| 22:05 | <Hixie> | cool |
| 22:06 | <AryehGregor> | annevk, then define some abstract steps like "alter the data" that accept the same arguments as replaceData(). |
| 22:06 | <annevk> | AryehGregor, yeah, could do that I suppose |
| 22:06 | <TabAtkins> | Hixie: I do suspect it's useful to have a generic "machine representation". I'd be afraid of it being misused to embed arbitrary data a la RDFa's @content attribute, but shrug. |
| 22:07 | <annevk> | AryehGregor, I haven't looked at that section in a long time, other than minor cleanup |
| 22:07 | <TabAtkins> | That's probably not a bad enough fear to really care. |
| 22:07 | <Hixie> | TabAtkins: well we already have <meta> for that |
| 22:07 | <TabAtkins> | Hixie: For the former or the latter? |
| 22:07 | <Hixie> | TabAtkins: but people want to do <span itemprop=a content=b>x</span> instead of <meta itemprop=a content=b>c for some reason |
| 22:08 | <Hixie> | (wish i understood why) |
| 22:08 | <TabAtkins> | Yeah, I dunno. |
| 22:08 | <Hixie> | (i meant for the latter) |
| 22:09 | <TabAtkins> | Hixie: However, I still kinda like <time> for its application to the atom extraction algo. ^_^ |
| 22:11 | <TabAtkins> | Hixie: Also, would a generic machine-representation element have some way of indicating the datatype? <time> was useful in that it indicated the data was definitely a datetime, not an arbitrary string. |
| 22:11 | <Hixie> | dunno, depends on what hte use cases are |
| 22:12 | <Hixie> | for things like microformats or microdata, you don't need the type, since the prop name gives you the type |
| 22:12 | <annevk> | only if you know the propname, no? |
| 22:12 | <TabAtkins> | I was thinking probably the same thing. |
| 22:12 | <TabAtkins> | annevk: If you don't know the propname, what are you doing with the data? |
| 22:13 | <TabAtkins> | (Besides just archiving it, where you don't need the datatype.) |
| 22:13 | <annevk> | hmm |
| 22:13 | <annevk> | not having generic processing supposedly was a problem with microformats |
| 22:13 | <TabAtkins> | That's generic *parsing*. |
| 22:14 | <Hixie> | what would you do with the type? |
| 22:14 | <TabAtkins> | Which was a problem. |
| 22:14 | <annevk> | Hixie, store things in different ways |
| 22:14 | <TabAtkins> | (In that you couldn't even *extract* microformats without knowing the format.) |
| 22:14 | <Hixie> | annevk: well so long as there's a 1:1 mapping from the value to the storage mechanism, you can still do that |
| 22:14 | <TabAtkins> | annevk: Possibly good point. datetimes can be stored much more efficiently than a string timestamp. |
| 22:14 | <annevk> | you could do some queries solely based on type, not sure how much meaning you can extract from it |
| 22:15 | <Hixie> | annevk: anyway, if there's a concrete use case, i'm open to it |
| 22:15 | <Hixie> | i don't know of one |
| 22:15 | <annevk> | export to Atom is/was one |
| 22:16 | <Hixie> | yeah, pubdate="" would be sacrificed if we did this |
| 22:16 | <Hixie> | not sure that's a huge problem |
| 22:16 | <Hixie> | we can always redefine the mapping to atom using microdata again |
| 22:16 | <Hixie> | btw, at the risk of errecting a bike shed and stepping in a can of paint, the main reason i haven't done this change is that i've no idea what the replacement element would be called |
| 22:16 | <Hixie> | <data>? |
| 22:16 | <TabAtkins> | Given that I'm already annotating my blogposts with microformats... |
| 22:16 | <TabAtkins> | <machine> |
| 22:16 | <TabAtkins> | <itemdata> |
| 22:17 | <TabAtkins> | <meta> |
| 22:17 | <Hixie> | it's not microdata specific, so i'd rather avoid <item*> |
| 22:17 | <TabAtkins> | It's not? |
| 22:17 | <Hixie> | <meta> is void |
| 22:17 | <Hixie> | nah, it could at least be used with microformats as well |
| 22:17 | <TabAtkins> | Ah, right. |
| 22:17 | <TabAtkins> | Microformats is stuck with html4 syntax anyway. |
| 22:17 | <Hixie> | they don't have to be |
| 22:17 | <annevk> | I'm not sure what the replacement element would look like |
| 22:18 | <TabAtkins> | If they break away, might as well just use microdata syntax entirely. |
| 22:18 | <annevk> | If it is solely defined by its attributes maybe just use <span> and <div> |
| 22:19 | <annevk> | gives you a reason to like them again ;) |
| 22:19 | <TabAtkins> | Having an inline-default and block-default version would be nice. |
| 22:19 | <Hixie> | i'd rather not overload them with semantic meaning, it's messy |
| 22:20 | <Hixie> | also it complicates the microdata algorithm if now <span> has three ways of having a value |
| 22:20 | <Hixie> | bad enough having two |
| 22:20 | <Hixie> | (itemscope and textContents) |
| 22:21 | <annevk> | Just an idea. Again, I've no idea what this thing would look like |
| 22:23 | <Hixie> | it would replace <p><time itemprop=start datetime="2005-01-05">Yesterday</time> it was <meta itemprop="weight" value="5">three weight-units</p> with: |
| 22:24 | <Hixie> | <p><data itemprop=start value="2005-01-05">Yesterday</data> it was <data itemprop="weight" value="5">three weight-units</data></p> |
| 22:24 | <Hixie> | (and would drop pubdate="") |
| 22:25 | <TabAtkins> | I like the consistency somewhat. |
| 22:26 | <TabAtkins> | But still think it's basically just for Microdata, and so should have "item" in its name. |
| 22:27 | <Philip`> | <annotate itemprop=start ...> |
| 22:29 | <Hixie> | TabAtkins: it doesn't have to just be for microdata. e.g. someone could use it to include machine-readable data in the page for their own purposes, much like data-* |
| 22:29 | <Hixie> | TabAtkins: in fact this is very similar to data-* |
| 22:29 | <Hixie> | *ponders* |
| 22:30 | <Hixie> | i guess people could just use data-* if they want their own thing |
| 22:30 | <TabAtkins> | Hixie: Indeed. |
| 22:30 | <Hixie> | maybe microdata is the only compelling use case for this |
| 22:30 | <Hixie> | hmm |
| 22:31 | <TabAtkins> | I guess <itemdata> (or whatever) shouldn't be able to accept @itemscope, though, right? |
| 22:31 | <TabAtkins> | Since it, by definition, can't represent further microdata items? |
| 22:32 | <Hixie> | interesting point |
| 22:32 | <Hixie> | this leads me to the thing i wanted to speak to hsivonen about, which came up in the discussion of the command api |
| 22:32 | <Hixie> | namely, the global attributes: should we make them less global? |
| 22:32 | <Hixie> | it doesn't make much sense to allow tabindex="" or accesskey="" or contenteditable="" on, say, <track> |
| 22:33 | <TabAtkins> | Yeah, true. |
| 22:42 | <AryehGregor> | On the other hand, <style contenteditable>head, style { display: block }</style> is awesome. |
| 22:43 | <jamesr> | heycam: (or anyone else) what's the difference in w3 issue tracker land between RAISED and OPEN? |
| 22:43 | <heycam> | jamesr, it's whatever you want it to be :) |
| 22:43 | <jamesr> | is OPEN sort of like an accepted state (we think we need to do something but haven't yet)? |
| 22:43 | <heycam> | that's how I tend to use it, yeah |
| 22:44 | <Hixie> | AryehGregor: yeah, there are some things for which it makes sense even though it's dubious. |
| 22:44 | <heycam> | it's probably not that important though; the other specs in the WG haven't been using tracker for their issues |
| 22:44 | <AryehGregor> | contenteditable probably should be allowed on any non-void element. |
| 22:45 | <AryehGregor> | The way I'm defining it, editing hosts themselves aren't modifiable, only their descendants are, so on a void element it makes no sense. |
| 22:45 | <TabAtkins> | AryehGregor: <video contenteditable>? |
| 22:45 | <AryehGregor> | Well, that would be a totally different feature. |
| 22:45 | <AryehGregor> | Oh, I see. |
| 22:45 | <AryehGregor> | You mean that should be invalid too. |
| 22:45 | <TabAtkins> | Possibly, yeah. |
| 22:46 | <AryehGregor> | Yeah, things whose children have no visible effect should also logically not be allowed to be contenteditable. |
| 22:47 | <nlogax> | when using drag and drop, how do you prevent a drop at drop time? if you decide that was not a good place (e.g when sorting stuff by dragging, and you didn't drag it far enough) |
| 22:47 | <jamesr> | heycam: aight, i won't worry about it then |
| 22:47 | <nlogax> | so that it glides back to whence it came, like when you drop it somewhere non-droppable |
| 22:47 | <jamesr> | it seems if i close an issue via the web ui it doesn't generate an email |
| 22:47 | <TabAtkins> | nlogax: On dragenter, don't preventDefault? |
| 22:47 | <jamesr> | hmm. i suppose that's ok |
| 22:47 | <nlogax> | TabAtkins: hmm. but i don't know where it will be dropped when that fires |
| 22:49 | <jamesr> | heycam: i also think that email about psychology experiments + knowing the vblank time is important |
| 22:49 | <jamesr> | heycam: but before addressing that i really want to propose a time interval API for everything that doesn't suck |
| 22:49 | <heycam> | jamesr, I haven't read that mail yet |
| 22:50 | <jamesr> | it's good stuff |
| 22:51 | <jamesr> | but i want to talk about time intervals on the whatwg list |
| 22:51 | <jamesr> | to make sure it gets a broad audience since it's important for the whole platform |
| 22:51 | <heycam> | what do you mean by time intervals here? |
| 22:51 | <heycam> | jamesr, I saw this recently, also related: http://blog.fishsoup.net/2011/06/30/frame-timing-the-simple-way/ |
| 22:51 | <jamesr> | you can't tell how much time has elapsed between two callbacks using, f'instance, Date.now() |
| 22:51 | <zewt> | timedelta |
| 22:52 | <heycam> | jamesr, ah right, yeah we should get on that |
| 22:52 | <zewt> | there was some discussion about that recently--don't remember if it was only in here or on the list too |
| 22:53 | <jamesr> | one of the timing APIs specs a better clock internally, but doesn't expose it. webaudio has a better clock too, but it's specific to webaudio |
| 22:53 | <jamesr> | mozilla had (has?) the animationStartTime global, but i'm not sure how that responds to system clock changes |
| 22:53 | <heycam> | jamesr, I am not sure either |
| 22:53 | <zewt> | audio can sometimes want to use a clock owned by the audio hardware or drivers, so having that separate might make sense anyway |
| 22:54 | <zewt> | afk |
| 22:54 | <nlogax> | whatever happens when pressing escape is what i would like to do, but with JS :) |
| 22:54 | <nlogax> | but can't find it |
| 23:15 | <Hixie> | foolip: http://www.whatwg.org/specs/web-apps/current-work/complete.html#associating-names-with-items look good? |
| 23:15 | <Hixie> | foolip: (i am about to start updating the other algorithms to break loops, not done that yet) |
| 23:23 | <Hixie> | what's the term meaning the set that contains all items included in the graph if you start at one node and just walk the graph |
| 23:23 | <Hixie> | the something something |
| 23:24 | <AryehGregor> | The largest connected component containing that node? |
| 23:24 | <Hixie> | there's a shorter term for it |
| 23:24 | AryehGregor | doesn't know any graph theory, can't help more than that |
| 23:25 | <Hixie> | man i am completely blanking here |
| 23:25 | <heycam> | are you thinking of the term "transitive closure"? |
| 23:25 | <Hixie> | yes! |
| 23:25 | <Hixie> | thank you. |
| 23:26 | <heycam> | I think "transitive closure" by itself doesn't necessarily mean just walk the graph; I think it only means something if you give it some relation |
| 23:26 | <heycam> | so I'm not exactly sure how to use it correctly |
| 23:26 | <Hixie> | yeah my definition wasn't great |
| 23:26 | <Hixie> | but it got me the term i was looking for, so good enough! |
| 23:26 | <jamesr> | do you want the connected component containing the node? |
| 23:27 | <jamesr> | assuming this is an undirected graph |
| 23:27 | <Hixie> | the term i was looking for is transitive closure, i was just having a mind blank |
| 23:30 | <TabAtkins> | Hixie: "transitive closure" doesn't make sense by itself, for the reason heycam gave - it needs a relation. |
| 23:31 | <Hixie> | yes, i am aware :-) |
| 23:31 | <TabAtkins> | If your relation is "neighbor of", then taking the transitive closure gives you the largest connected component including your starting node, which is what you want. |
| 23:33 | <TabAtkins> | (Via the fact that, after taking the transitive closure, all reachable nodes now return true for the new relation you defined.) |
| 23:36 | <jamesr> | i think "largest" is redundant, connected component implies it contains all connected nodes |
| 23:36 | <TabAtkins> | No it doesn't. |
| 23:37 | <TabAtkins> | The largest connected component probably includes a lot of smaller connected components. |
| 23:37 | <Hixie> | tab speaks the truth |
| 23:37 | <TabAtkins> | (You're allowed to ignore neighbors if you want, after all.) |
| 23:37 | Hixie | defines a conformance criteria in terms of a constraint rather than steps, for once |
| 23:37 | <TabAtkins> | FINALLY |
| 23:37 | <TabAtkins> | ^_^ |
| 23:37 | <Hixie> | :-D |
| 23:38 | TabAtkins | keeps defining conformance criteria in terms of "use a cycle detector". |
| 23:38 | <Hixie> | heh |
| 23:39 | <Hixie> | specifically: |
| 23:39 | <Hixie> | "All itemref attributes in a Document must be such that the graph formed from representing each item in the Document as a node in the graph and each property of an item whose value is another item as an edge in the graph connecting those two items does not contain any cycles." |
| 23:39 | <Hixie> | that's a pretty horrible sentence, mind you |
| 23:39 | <Hixie> | let me rephrase it into english |
| 23:39 | <jamesr> | wut |
| 23:39 | <TabAtkins> | Yay, I was (barely) able to parse it using my native english facilities! |
| 23:40 | <TabAtkins> | (Rather than the deductive variant of english I consciously learned later.) |
| 23:41 | <Hixie> | well that's a new error from anolis |
| 23:42 | <Hixie> | (some sort of httplib error) |
| 23:52 | <Hixie> | if a vcard A has as agent a vcard B and vice versa |
| 23:52 | <Hixie> | what should B's AGENT line say? |
| 23:53 | <Hixie> | (A's AGENT line is B's entire vCard) |
| 23:54 | <Hixie> | AGENT;VALUE=TEXT:ERROR ? |
| 23:55 | <annevk> | HTML should start using Ms2ger's Anolis |
| 23:55 | <Hixie> | i use whatever jgraham uses |
| 23:55 | <Hixie> | pimpmyspec baby |
| 23:55 | <Hixie> | what does ms2ger's do that's different? |
| 23:55 | <annevk> | xspec xref |
| 23:55 | <Hixie> | how? |
| 23:56 | <annevk> | using https://bitbucket.org/ms2ger/specification-data |
| 23:56 | <annevk> | some kind of repository with references and linkable terms in other specs |
| 23:57 | <Hixie> | please tell me that's automatically generated somehow |
| 23:57 | <annevk> | and then you use <span data-anolis-ref>HTML</span> for a reference |
| 23:57 | <AryehGregor> | It is for all specs except HTML, I think. |
| 23:57 | <Hixie> | personally i'd much rather cross-spec xrefs be opt-in rather than automatic... what happens if we have a conflict? or don't we? |
| 23:57 | <annevk> | and <code data-anolis-spec=domcore>Node</code> |
| 23:57 | <Hixie> | oh |
| 23:57 | <AryehGregor> | You have to say which spec you're referring to. |
| 23:57 | <annevk> | for terms |
| 23:58 | <annevk> | the dictionary maps terms to fragment identifiers |
| 23:58 | <annevk> | it's auto-generated most of the time, but for IETF specs that doesn't really work |
| 23:58 | <Hixie> | i don't mind listing the terms once, but i'm definitely not going to keep track of what spec things are where all the time |
| 23:58 | <AryehGregor> | The makefiles for DOM Core and DOM Range generate the data files for them automatically. |
| 23:58 | <Hixie> | can't we make this all automatic? |
| 23:58 | <AryehGregor> | I suggested we shouldn't need data-anolis-spec. |
| 23:58 | <AryehGregor> | Conflicts should be handled statically somehow when parsing the data files. |
| 23:59 | <AryehGregor> | Or just be an error on use, or whatever. |
| 23:59 | <annevk> | Yeah, maybe we can get rid of data-anolis-spec |
| 23:59 | <annevk> | by having specs use less conflicting terms |
| 23:59 | <AryehGregor> | But personally I was too lazy to fix it, so I gave up on typing it out manually and wrote a preprocessor. |
| 23:59 | <Hixie> | right now what i do is have a section of the spec that basically "imports" the terms from the other spec |
| 23:59 | <Hixie> | this makes it work for print media too |
| 23:59 | <AryehGregor> | Even without data-anolis-spec, typing stuff like title=dom-Document-createElement all the time is murderous. |
| 23:59 | <AryehGregor> | This works for print media as well. |
| 23:59 | <AryehGregor> | It uses generated content. |