2014-03-01 [16:00:27.0000] oh god it has its own clippy [16:00:37.0000] ? [16:00:41.0000] "pegman" heh [16:00:50.0000] old maps had pegman too... how is it like clippy? [16:01:19.0000] corniness? heh [16:02:40.0000] ? [16:02:48.0000] it's just a ui widget to drop you into street view [16:02:51.0000] how is it corny? [16:02:56.0000] /me wonders why browsers don't have a "find an anchor to link to" feature ... [16:03:14.0000] zewt: (i'm not arguing it isn't, i just don't understand what you mean) [16:03:59.0000] a goofy character suddenly introducing itself seems corny to me, heh [16:04:01.0000] anyway, um, http://www.whatwg.org/specs/web-apps/current-work/multipage/obsolete.html#non-conforming-features provides some VEEERY terse recommendations [16:04:26.0000] zewt: he introduced himself? interesting, i haven't seen that [16:04:35.0000] i wonder if they added the introduction after i'd been using it already [16:04:36.0000] "hey, I'm pegman" or something like that (dismissed it already) [16:04:53.0000] agreed that that's corny. or cute, you know, whatever floats your boat :-P [16:05:10.0000] SamB: happy to add more... file bugs with suggested text [16:05:16.0000] at least it didn't slowly animate on screen, and then when told to go away, slowly animate off [16:06:48.0000] i think the thing that made clippy "clippyesque" isn't so much the animation, or the cuteness, or even the slowness of the interaction, it's that it just wasn't that useful, so it would interrupt you in ways that were just annoying. [16:08:47.0000] I have no idea whether the new pegman intro is useful, since I reflexively dismissed it as soon as it appeared [16:09:54.0000] in general, I think one should avoid modal things that pop up to show you how to use new studd [16:09:59.0000] s/studd/stuff/ [16:10:05.0000] Hixie: well, I suspect that more detailed recommendations would be more approriate in some of the linked sections; probably this one should just explain how to tell which link(s) will provide such details [16:10:50.0000] SamB: that works too. basically whatever it is you want, file a bug saying what it is :-) [16:11:00.0000] astearns__: yeah, the "learn new maps!!!" dialog that popped up at first was bad like that [16:11:34.0000] since when I first turn it on I just want to poke around first (so I dismiss it), and of course now I don't know how to get it back if I wanted to look at it now (not that I really want to) [16:13:38.0000] zewt: That *precise* feedback was just echoed by some in internal Plus today. [16:13:54.0000] "Warm welcome" tutorials are often worthless for that reason. [16:14:09.0000] /me goes to look up the checkout directions ... [16:14:48.0000] itym "get out of the way, damn it" tutorials [16:15:19.0000] at least maps's did (some are very insistant) [16:17:42.0000] so basically, such tutorials should start like this: 1. To re-open this tutorial, [...] 2. To close this tutorial, [...] [16:23:03.0000] /me wonders where he can find an tag in use ... [17:13:01.0000] /me is kind of freaked out to see that Mozilla still loads Java DLLs in-process ... though thinking about it for a bit, he does remember hearing about Mozilla having some very special APIs basically just for the Java "plugin" ... [17:15:49.0000] /me tries not to speak in the third person [17:16:36.0000] /me finds that ACTION tends to look funny otherwise [17:17:45.0000] /me prefers PRIVMSG [17:20:31.0000] it *is* a PRIVMSG; it just has some funky stuff in it that ... [17:34:41.0000] /me sees that zewt is familiar with the concept :-) [21:43:26.0000] zewt: Why the fuck are you sending a bunch of CTCP shit? [21:43:42.0000] to personally irritate you [21:43:56.0000] mission: success [21:46:59.0000] zewt!*@* added to ignore list. [21:47:28.0000] That kind of fucktard reply would have resulted in a permanent ban and k-line if sane people were running this shit. [21:47:59.0000] good to know teenagers still inhabit IRC. [22:07:28.0000] /me thinks the w3c validator should perhaps use validator.nu for more stuff ... [22:24:47.0000] AlexBones: hmm, your sarcasm-o-meter seems busted [22:42:21.0000] Just tired of all the obnoxious people. [22:47:19.0000] SamB: what more should the w3c validator be using validator.nu for? [22:49:04.0000] /me thinks its HTML 4 schemas are a bit more precise than the DTDs that onsgmls uses ... [22:50:02.0000] ah [22:50:39.0000] I was exposing HTML4 checking as an option at http://validator.w3.org/nu/ but I've since remove it [22:50:48.0000] *removed [22:52:02.0000] I removed the whole Presets menu actually [22:52:49.0000] so it just chooses a schema based on the MIME type of the document and doesn't provide an option in the UI to override that [22:54:10.0000] so by default it now does validation against the latest HTML spec (aka HTML5), even for documents that have a HTML4 doctype [22:56:05.0000] validator.nu also always does that by default now too, even for documents with HTML4 doctypes. It just still also provides the Presets option to allow users to override that behavior if they want [22:58:51.0000] anyway I was maintaining the validator.nu HTML4 schema for a while and responding to bug reports and requests about it but I'm no longer interested in putting any time into it [23:00:33.0000] I'm not sure why anybody should want to do HTML4 validation other than just out of curiosity -- to see, e.g., what was an error in HTML4 but not in HTML now, or whatever [00:17:25.0000] hmm, what license are the WHATWG logos under? [07:48:59.0000] (yeah, nothing obnoxious at all about calling someone a "fucktard") [11:58:45.0000] Thoughts? A this-is-responsive style page about a11y http://jonathantneal.github.io/this-is-a11y/ [13:08:42.0000] hmm, I think http://resources.whatwg.org/ needs a COPYING file. Also, the FAQ doesn't seem to contain anything about licensing ... [15:33:07.0000] SGML/XML used by linguists is uniformally horrifying. It makes me want to cry all the time. :( 2014-03-02 [16:23:17.0000] gsnedders: I find all sgml documents to be rather frustrating. [16:24:22.0000] I would dig a future of native slim-lang parsing http://slim-lang.com/ [18:55:05.0000] /me finds the typical DSSSL stylesheet to be refreshingly straightforward [18:55:35.0000] (note: they usually don't have any actual tags in them) [08:46:05.0000] I’m writing a short article to describe the differences between the Content-Language header and the lang attribute. Is it accurate so far? https://gist.github.com/jonathantneal/9309367 [10:42:23.0000] JonathanNeal: http://www.whatwg.org/specs/web-apps/current-work/multipage/elements.html#language is based on lang attributes, but when that’s missing it uses Content-Language as the final fallback [10:53:26.0000] I don’t know if the difference means anything in practice [10:53:51.0000] such as, any tool interpreting them differently [10:57:29.0000] :lang? [11:06:23.0000] :lang() treats them the same, other than priority [11:09:56.0000] SimonSapin, do you know if the lang attribute supports multiple languages? [11:10:07.0000] It doesn’t [11:10:34.0000] but when falling back to Content-Language, "the language of an element" doesn’t either [11:11:02.0000] see link above [11:12:27.0000] I believe Content-Language supports multiple attributes http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.12 [11:12:37.0000] yes [11:13:09.0000] but not when it’s used as the fallback for the language of an element, per the HTML spec [11:13:29.0000] maybe search engines use Content-Language [11:15:52.0000] my point is that "Don’t do this because it’s Wrong" isn’t very helpful if it doesn’t make any difference in practice. [11:17:25.0000] Oh, do you think I suggesting something like that in my post? [11:18:45.0000] not really [11:19:46.0000] but "This also means a document intending to teach French to English readers would use Content-Language: en", what happens if I use Content-Language: fr in this case? [11:20:20.0000] More angels would dance on the head of a pin [11:22:51.0000] SimonSapin, in that case, a translation service might recommend translating the french portions into English, despite the intent to keep them? [11:23:23.0000] JonathanNeal: than that’s what the article should say [11:24:35.0000] all the better if you have example of existing translation services that do behave that way [11:26:15.0000] Great. [11:29:13.0000] if there are no web-compat reasons to do otherwise (which there may be), content-language and should be the same thing [11:29:52.0000] it's hard enough to get people to tag languages in the first place, without having multiple different ways with different nuances, and some in HTML headers and others inline [11:41:10.0000] +1 [11:42:45.0000] though for HTML’s purposes they’re already the same, and CSS only uses them through HTML’s definition [11:52:05.0000] zewt, SimonSapin, thanks for the input. I’ve made some changed based on our conversation. Do you think this reads more accurately? https://gist.github.com/jonathantneal/9309367 [11:53:38.0000] JonathanNeal: don’t link to HTML 4 [11:57:02.0000] JonathanNeal: "Google determines ...", is that Search or Translate? The blog post you link to seems to be about Search [12:06:43.0000] SimonSapin: Thanks, I will do some research and clarify that. [13:01:20.0000] JonathanNeal: SimonSapin is right that you shouldn't reference HTML4 [13:02:04.0000] that spec should not be trusted as anything reflecting reality [13:02:25.0000] I guess you should just use http://tools.ietf.org/html/draft-ietf-httpbis-p2-semantics-26#section-3.1.3.2 instead [13:03:00.0000] there's supposed to be an updated RFC of the HTTP spec published soon [13:03:21.0000] I think -26 is the final editor's draft [13:04:18.0000] JonathanNeal: or instead maybe best to just use https://www.w3.org/International/questions/qa-html-language-declarations [13:04:49.0000] MikeSmith: Thanks. I'm using that document currently to reference more of what I'm writing. [13:05:17.0000] https://www.w3.org/International/questions/qa-html-language-declarations#metadata [13:05:31.0000] JonathanNeal: ok [13:06:31.0000] btw the stuff under https://www.w3.org/International/ is generally trustworthy [13:07:19.0000] since Richard Ishida knows whereof he speaks, and he keeps that content up to date [13:09:19.0000] Basically - English is the default language of the web. The lang attribute describes the language of actual content. It falls back on content-header. Content-header describes the language of the intended audience. It can describe multiple languages, but in practice, keep pages monolingual. [13:09:37.0000] https://www.w3.org/International/questions/qa-http-and-lang seems good too [13:09:40.0000] content-language header [13:25:24.0000] /me wonders again what license applies to the stuff on http://resources.whatwg.org/ ... [13:28:39.0000] That's a good question [13:30:13.0000] I don't think anyone took the time to slap a license on it [13:31:42.0000] Probably look up who contributed to https://github.com/whatwg/resources.whatwg.org and send them email cc www-archive to see if they agree to relicensing [13:36:16.0000] /me was just going to report a bug against HTML even though it's not really HTML-related ... but I guess that would also work [13:38:25.0000] my first inclination was to file in issue against the github repository, but it has issues disabled :-( [13:39:15.0000] /me wonders if these logos are trademarks [13:40:16.0000] The XHR one may be somewhat problematic [13:45:43.0000] Hey. [13:48:16.0000] Is there a WebMUX implementation anywhere? [14:06:06.0000] never heard of it [14:07:18.0000] Last draft seems to be from 1999. [14:07:27.0000] Also called w3mux sometimes, or SMUX originally. [14:14:09.0000] is that related to SPDY at all? [14:15:10.0000] Some multiplexing ideas might be similar, but I think that's about it. [14:16:12.0000] I think the implementation is part of the ILU. Also from 99. [14:20:23.0000] Hmm, that doesn't look like a generally usable implementation. Tied to the ILU, whatever that does. [14:24:26.0000] Is anyone aware of any simple, implemented, multiplexing protocols? [14:30:29.0000] are any of you opinionated between public domain and MIT? [14:34:43.0000] Jonathan: Copyright notice shown? [14:36:31.0000] In general, in picking a license. I have committed projects to the public domain, and I haven’t had any problems. On the other hand, I’ve had some issues with MIT licensed stuff because of opinions on how attribution should be made. [14:37:53.0000] I can't say I've dwelt much on licenses. [14:38:24.0000] How's attribution different to any other license with a similar clause? [14:56:54.0000] JonathanNeal: not personally; some people say they prefer MIT for the warranty disclaimer, but I don't know how that's better than just putting a warranty disclaimer in by itself [14:58:02.0000] What warranty is there to public domain? [14:59:07.0000] Public domain is my way of not dealing with any of that, and ensuring that other’s don’t complicate it either. [14:59:26.0000] ZiNC: Depends on what country you're in [15:00:05.0000] Source or target? :) Just seems odd there'd be any warranty to random code you release for free. [15:00:59.0000] ZiNC: Most of the implied warranties in most juristictions apply only to sales. [15:01:32.0000] That's what I'd expect. [15:01:50.0000] I had an issue with MIT attribution. Someone worried folks might be technically breaking the license every time they forked the project. And then there’s another infamous story I’d rather not get into where people were acting rather threatening toward me when I changed the language of a forked project. [15:02:23.0000] JonathanNeal: I'll also point out wrt public domain that not all juristictions allow one to put something into the public domain. [15:03:11.0000] gsnedders: I did not know that! Anything in particular I should know about that? I apply a lot of JS to public domain. [15:03:14.0000] How can that be? [15:04:46.0000] In quite a few juristictions "public domain" only exists as a state that works pass into after author's life + 70 (or more) years. [15:05:32.0000] And there's no concept of disowning copyright of a work [15:05:53.0000] You can always release anonymously. :) [15:06:37.0000] But you still own copyright. [15:07:16.0000] Ability to release anonymously or pseudononymously is one of the irrevoccable moral rights granting to the author by the Berne Convention. [15:07:27.0000] JonathanNeal: If you're concerned about it, see CC0. [15:08:44.0000] (IANAL, etc.) [15:12:53.0000] SimonSapin, MikeSmith, thanks again. I’ve updated https://gist.github.com/jonathantneal/9309367 to reflect some of the information and links you gave me. [15:13:16.0000] yeah, CC0 is a public domain dedication with failover to a simple permissive license, which is about the best that can be done [15:14:24.0000] modulo the details of constructing the license so that it will be considered legal everywhere but still provide as much freedom as possible [15:25:52.0000] anyway, I just reported ... 2014-03-03 [19:27:42.0000] SamB: what do you want to do with the resources? [19:31:06.0000] Well, logo.svg would look nice on the WHATWG article, obviously, though I guess it could be done as "fair use" on en.wp ... [19:36:44.0000] (note that commons doesn't seem to require any trademark-related rituals) [19:41:05.0000] of all the pointless ... [19:41:09.0000] [19:44:26.0000] pretty sure that would just be fair use regardless of copyright [19:48:03.0000] but that's only actually allowed on the English wikipedia [20:20:48.0000] hmm, you guys haven't been keeping the PNGs up to date ... [20:56:55.0000] well, "fair use" isn't something that exists everywhere [20:57:02.0000] as I (incompletely) understand things [21:02:10.0000] zewt: yes, that would be why it's not allowed on the other wikipedias. even though they're all hosted in the US. [21:03:13.0000] shouldn't be allowed on any of them [21:05:53.0000] well, it's not FUN on any of them [22:30:25.0000] its's too bad PNG only supports up to 256-entry palettes ... [23:41:55.0000] hmm, rsvg does not grok logo-xhr.png ... [02:11:58.0000] foolip: could you review a MediaSource test? [02:57:36.0000] MikeSmith: I could try [02:58:38.0000] foolip: cool, gimme a minute I'll get you the critic URL [02:59:13.0000] foolip: https://critic.hoppipolla.co.uk/r/911 [02:59:29.0000] actual PR is https://github.com/w3c/web-platform-tests/pull/708 [03:01:46.0000] MikeSmith: oh look, you're the author [03:07:35.0000] heh [03:07:50.0000] foolip: yeah should I have disclosed that first? [03:21:54.0000] MikeSmith: not really, I just assumed things :) [03:22:25.0000] :-) [03:28:42.0000] MikeSmith: you should ask Ms2ger for review as well [03:29:47.0000] No you shouldn't ;) [03:29:52.0000] /me lunches [03:33:21.0000] MikeSmith: you have some issues to look at now anyway :) [03:34:58.0000] foolip: cool -- thanks much [04:00:15.0000] MikeSmith, see html/dom/interfaces.html, fwiw [04:01:54.0000] Ms2ger: yeah I've looked at it before [04:02:04.0000] I will look at it more [04:53:25.0000] foolip: Thanks -- from your review comments I realize I can simplify the test quite a bit (since after all it really only needs to test the IDL support). So I'll make the change you suggested about putting all of the setup in an async test and running the IDL tests after that's done. (And push a change later for re-review but I think that next review won't cost much time.) [08:28:30.0000] gah, I meant to say rsvg doesn't like logo-xhr.svg obviously [09:12:06.0000] good morning, Whatwg! [12:40:31.0000] [13:04:31.0000] where do I report a bug in ... [13:05:01.0000] SamB, please report bugs to www-svg⊙wo but note that nobody really looks at /TR/SVGMobile2 -- check http://svgwg.org/svg2-draft/paths.html instead [13:05:13.0000] heycam: ah [13:12:06.0000] anyway, it seems to refer to both "instructions" and "commands" without making it clear what, if anything, is the difference between them [13:15:46.0000] Hi Ms2ger, are you around? [13:16:04.0000] Briefly [13:16:13.0000] SamB, (without looking) no difference [13:16:38.0000] I wonder: are there any specs that define !important UA style sheet rules? [13:16:50.0000] You recently closed https://bugzilla.mozilla.org/show_bug.cgi?id=978924 but even your test case shows inconsistencies in event firing across browsers. I've changed the spec., but I think the bug still stands. [13:17:15.0000] I think your test case should be added to it. [13:19:23.0000] arunranga, yeah, Chrome is wrong, it dispatches async [13:21:19.0000] /me hmmm… you're right that abort() is synchronous… [13:57:28.0000] Hixie: http://esdiscuss.org/topic/es6-tasks-and-taskqueues#content-3 [13:58:08.0000] microtasks only have one queue per event loop, btw [13:59:02.0000] as far as HTML goes, i have no desire to pretend it's distinct from ES... my goal was to hook in the ES spec into the HTML spec [13:59:13.0000] by making use of hooks provided by the ES spec [14:00:25.0000] oh i thought there were several microtasks queues last i looked ... custom elements, then table sorting, then mutation observers, etc. [14:00:51.0000] they're all getting collapsed into one [14:00:52.0000] i feel like that order is a bit strange given how underspecified ES is [14:00:56.0000] ah cool [14:01:11.0000] i'm waiting for mutation observers to jump onte the train before doing table sorting [14:01:41.0000] has anyone even announced intent to implement table sorting? :( :( :( [14:02:08.0000] no :-( [14:02:13.0000] tons of author demand, though [14:04:42.0000] siiigh [14:48:56.0000] Domenic_: ping. [14:49:05.0000] (about es6 task vs microtask issue) [14:58:08.0000] oh, I figured out why logo-xhr was so troublesome ;-) [15:15:24.0000] rafaelw: pong [15:19:37.0000] do you understand how modules work? [15:19:56.0000] I just skimmed the spec, but I'm still pretty fuzzy on the actual mechanics. [15:20:16.0000] I might have naively expected modules to need to enqueue Task work (as opposed to Microtask work) [15:20:21.0000] but maybe they don't. [15:20:23.0000] any idea? [15:29:33.0000] rafaelw: from my attempt to understand/model module loading and skimming the spec and pawing through https://github.com/jorendorff/js-loaders the behavior is described in terms of Promises, so it should only be microtasks, but I'm equally curious in an authoritative answer. [15:29:54.0000] rafaelw: my understanding is they delegate everything to promises, yeah. [15:30:23.0000] jorendorff was quite pleased when he realized he wouldn't have to do any of the event loop stuff himself :P [15:30:34.0000] but presumably, Task work is taking place, right? IOW, when some bytes actually load off the network, something gets queue to handle the bytes being loaded and continue the loading pipeline? [15:30:48.0000] gets queued [15:31:00.0000] Domenic_: it was a miracle [15:31:06.0000] Domenic_: and not because i'm lazy although i am [15:31:30.0000] Domenic_: but because instead of specifying details of behavior, i could lean on the conventions baked into the Promise design [15:38:09.0000] rafaelw: yes, if I am understanding correctly. When the bytes arrive you call EnqueueTask("PromiseTasks", ...stuff that causes the bytes to be used...) [15:39:09.0000] Hmm. ok, so maybe the es6 can leave the outer event loop as an implicit concept. [15:40:27.0000] And what is the "Script Evaluation Tasks" task queue used for? [15:41:07.0000] I haven't seen that one... 2014-03-04 [16:52:08.0000] Hixie: what can I do to convince you to add an element.createOutline() to the spec [16:52:35.0000] for generating just a static Outline object [16:52:55.0000] (I mean with not live nodes) [16:55:08.0000] what is it for? [17:26:41.0000] hey othermaciej [17:26:46.0000] hi MikeSmith [17:27:37.0000] element.createOutline would be for using the outline algorithm to generate and Outline that can could then be used to, say, add a TOC to a document [17:28:32.0000] the actuall adding-the-TOC part would be application-specific to the code of whatever Web application [17:28:53.0000] that is, Web authors would have to roll their own for that [17:29:20.0000] but at least they'd have a built-in native way to get the Outline to work with [17:29:52.0000] instead of N different Web authors needing to write their own implementations of the outline algorithm in their own JS [18:01:30.0000] MikeSmith: convince me a couple of UAs would implement it :-) [18:02:44.0000] othermaciej: please tell me and Hixie you would implement createOutline [18:03:17.0000] why is dynamically creating a TOC at runtime useful? [18:04:24.0000] I would tend to think doing a pass over the whole document when the page is loaded is not a great thing to do, perf-wise [18:04:49.0000] better to have your document already include a TOC [18:05:04.0000] sure [18:05:22.0000] but I think the same could be said for a lot of things you can do on the client side [18:05:27.0000] if you don't care for that, and somehow really need to dynamically create a TOC (maybe you are a word processing app), you can implement the Outline algorithm or similar algorithm of choice in JS [18:05:33.0000] othermaciej: well, the method could do it on demand [18:05:39.0000] right [18:05:59.0000] perf is not a good argument to make it built in, because it's bad for perf anyway (compared to precomputing your TOC) [18:06:15.0000] convenience might be an argument if there was huge demand for this specific functionality [18:06:42.0000] there's nothing corresponding in popular native APIs so a priori I would assume it's not a necessary part of a good platform [18:06:53.0000] but to be fair, the Web has more things that blur the lines between app and document [18:07:25.0000] the web is the only scriptable platform that has scripting in any serious sense, isn't it? [18:07:31.0000] s/scriptable/document/, sorry [18:07:51.0000] I have not previously heard of demand for this particular functionality though so I can't say it would be a high priority [18:09:47.0000] yeah I guess I should try to find some JS libraries that are providing something similar already [18:10:32.0000] "TOC is a jQuery plugin that will automatically generate a table of contents for your page" [18:10:36.0000] http://projects.jga.me/toc/ [18:10:56.0000] https://github.com/jgallen23/toc [18:11:22.0000] actually it seems like there are at least half a dozen jQuery TOC plugins [18:12:11.0000] so at least for anybody actually using those already, the perf issue's already there [18:13:00.0000] /me wonders why entirely non-normative sections get the implementation status in those boxes, instead of just bug boxes ... [18:13:03.0000] wow even a Drupal jQuery TOC plugin [18:13:50.0000] SamB: because the annotation mechanism doesn't distinguish between normative and non-normative sections [18:15:24.0000] I suppose that is the most likely explanation, yes [18:16:28.0000] but that leaves me wondering how some sections have opted out, which leaves me looking for my login info ... [18:17:03.0000] opted out? no sections have annotations by default [18:17:31.0000] the only way for an annotation gets added it by somebody taking time to manually add one [18:20:17.0000] How come it's 2014 and we still can't use proper HTTP verbs in HTML? [18:22:06.0000] JosephSilber: why should HTML be using HTTP verbs in the first place? [18:22:37.0000] SamB: Don't you want to use a form to update a record? [18:23:02.0000] do you mean verbs like BREW or somesuch? [18:23:16.0000] PUT/PATCH/DELETE [18:24:33.0000] those are rather exotic verbs, why would you expect javascript to be trusted with them? [18:24:56.0000] SamB: JavaScript already is. Was talking about HTML. [18:31:28.0000] HTTP verbs are so last century [18:31:43.0000] Can't tell whether you're serious. [18:32:11.0000] Hixie: sarcasm doen't come across well in writing. [18:33:10.0000] more seriously, HTTP should just have two verbs, GET and POST. Endpoints should be scripts. URIs for APIs, not resources. [18:33:41.0000] resources are too abstract to be given URLs in a practical world. [18:34:49.0000] Hixie: So I take it you aren't a fan of the RESTful mindset? [18:35:40.0000] you could say that. [18:35:58.0000] REST has caused thousands of web authors to waste time on a dead-end idea. [18:36:26.0000] Then what's the difference between GET and POST. [18:36:30.0000] caching? [18:36:43.0000] bigger payload? [18:36:54.0000] Hixie: I kind of think HEAD is useful too [18:38:03.0000] the APIs i design tend to be a single URL (let "URI" die already, everything's a URL), posted JSON, with the whole body of the command in the JSON [18:38:12.0000] JosephSilber: GET is idempotent and thus (more) cachable. [18:38:20.0000] zewt: there ARE a few URNs in use [18:38:41.0000] SamB: HEAD, in practice, too often returns bogus data to be valid. [18:38:42.0000] they work pretty well for namespace URIs [18:38:44.0000] Can we at least get POST on anchor tags? [18:38:45.0000] puts the whole body of the request in a single vocabulary (a JSON dictionary), which is so much simpler than some of it being in a request body, some of it in the URL, some of it in the HTTP request type... [18:39:22.0000] Hixie: i've had webkit in iOS cache POST :( [18:39:29.0000] "surprise!" [18:39:35.0000] /me would love an ISBN URL scheme [18:39:54.0000] JosephSilber: POST is an action, links are supposed to be safe to click. [18:40:04.0000] JosephSilber: another way of describing GET and POST is GET is a link, POST is a button. [18:40:24.0000] Hixie: theoretically, but in practice you can style it however you like. [18:40:38.0000] e.g. a delete button shouldn't need a hidden form behind it. [18:40:47.0000] well sure, you can also make a link look like a text field... [18:40:56.0000] buttons don't need
s anymore [18:41:17.0000] How would you delete a record without a form (and no js) [18:41:18.0000] ? [18:41:22.0000] Hixie: well, ever since it's been possible to attach scripts to clicking things that look like links, that's sort of gone out the window [18:42:07.0000] (not that "looks like a link" itself means much these days; google search is apparently removing the link styling from result headers, which drives me nuts) [18:43:21.0000] google has a lot of nerve making bad changes as well as good changes ;-P [18:43:29.0000] JosephSilber: [18:53:48.0000] oops, missed an attribute [18:53:58.0000]
[18:54:18.0000] Hixie: http://jsfiddle.net/k3nvR/ [18:54:40.0000] JosephSilber: that has type=submit [18:54:52.0000] JosephSilber: you want it to be type=button to not submit [18:55:15.0000] Oh. So it defaults to submit. [18:55:20.0000] (see http://whatwg.org/html#attr-button-type ) [18:55:50.0000] that settles it. [18:56:33.0000] Why do we need an empty form around the button? [18:56:59.0000] doesn't have to be around, you can point to it using form="". but there has to be one, because form submission is actually done by the
, not the