2014-05-01 [17:03:50.0000] death to target=_blank. [17:09:27.0000] zewt: that did always seem like a dumb idea [17:22:52.0000] MikeSmith: I submitted a couple of webkit patches for error handling: https://bugs.webkit.org/show_bug.cgi?id=132407 and https://bugs.webkit.org/show_bug.cgi?id=132412 [17:23:51.0000] cabanier: cool [17:24:04.0000] /me takes a look [17:25:58.0000] MikeSmith: that should make WK turn green for 4 more tests [17:28:00.0000] cabanier: excellent [17:29:33.0000] cabanier: thanks for taking time on the testing stuff, and raising the bugs, and the patches [18:14:54.0000] hober: i'm not sure what that is saying :-| [18:23:54.0000] cabanier: trybots are indicating that test is still failing, right? [18:28:07.0000] Hixie: https://www.w3.org/Bugs/Public/show_bug.cgi?id=25521 is a bit unsettling [18:28:21.0000] not that I disagree with the rationale [18:29:47.0000] it's just, if we end up changing the platform do the point where authors can't get any useful linting/static-checking feedback done on their markup sources, that would suck [18:31:25.0000] leading eventually to total societal breakdown [18:31:33.0000] "In accordance with prophecy" [18:33:44.0000] MikeSmith: yes, I missed some tests [18:33:57.0000] cabanier: ah ok [18:36:14.0000] MikeSmith: i have no idea what that bug is saying [18:37:24.0000] MikeSmith: web components aren't valid in html currently. i presume that the web components people have some sort of solution for that, e.g. via the inheritance thing () [18:37:28.0000] Hixie: "For these reasons, all of these content model rules should probably be discarded entirely" it says [18:37:30.0000] what's the wiki syntax for external links? [18:37:46.0000] i need to put a period after a link and i don't want it to be included in the link [18:37:54.0000] Hixie: I think you're being generous in that assumption ;-) [18:38:02.0000] MikeSmith: that seems like a non-starter. I mean, we can't make

valid, it would never do anything understandable. [18:38:09.0000] eligrey: it's just metawiki syntax, I think [18:38:31.0000] eligrey: [#FooBar] or such [18:38:44.0000] same for extenal links? [18:38:49.0000] i don't edit wikipedia, sorry :p [18:38:57.0000] eligrey: oh external [18:39:18.0000] eligrey: [http://foo link text here] I think [18:39:23.0000] thanks [18:40:08.0000] the wiki markup thing gives me a list of all the different types of markup [18:40:21.0000] you would think hovering over each item would give me a description, but no [18:40:25.0000] they all say "Click on the character or tag to insert it into the edit window" [18:40:34.0000] so helpful :) [18:40:43.0000] MikeSmith: anyways thanks for that [18:40:46.0000] wikis all suck [18:40:51.0000] eligrey: np [18:43:36.0000] Hixie: yeah I guess in the proposed regime there'd still need to be a few prohibitions on markup that just would never make sense [18:44:12.0000] Hixie: but the bug seems pretty clear to meーshe's saying that the "optionally a caption element, followed by zero or more colgroup elements, followed optionally by a thead element, followed optionally by a tfoot element, followed by either zero or more tbody elements or one or more tr elements, followed optionally by a tfoot element (but there can only be one tfoot element child in total), optionally intermixed with one or more script-supporting elem [18:44:18.0000] oops [18:44:25.0000] well anyway, that part [18:44:39.0000] your comment cut off in the middle of the spec quote [18:44:45.0000] she's saying that's too constrained in the face of web components [18:45:15.0000] Hixie: was just quoting the spec, nothing more [18:45:25.0000] hadn't meant to quote the whole thing [18:45:29.0000] ah k [18:45:54.0000] i don't think that's right [18:45:59.0000] was going to put some ellipsis in there, since I'm pretty sure you're familiar somewhat with the part that follows [18:46:05.0000] i mean, sure, you'll want to allow elements to override , etc [18:46:10.0000] yeah [18:46:18.0000] but those elements still need to be real s at some level [18:46:20.0000] as in is="" [18:46:24.0000] or some similar solution [18:48:17.0000] yeah I guess it would help there if she could give an actually markup example [18:48:22.0000] I'll post a comment [18:48:35.0000] oh wait she did [18:48:43.0000] https://github.com/angular/angular.js/issues/7295 [18:48:47.0000] /me reads [18:49:09.0000] oh geez man [18:49:09.0000] that's about the parser [18:49:13.0000] not the content models [18:49:13.0000] yeah [18:49:17.0000] bingo [18:49:21.0000] yup [18:51:17.0000] it's amazing how many people confuse those two things [18:51:30.0000] i mean, that's not a criticism or anything [18:51:39.0000] i'm honestly just amazed at how confusing this apparently is [18:54:03.0000] so do you think there's an opportunity to rephrase and/or add additional examples to the spec [18:54:54.0000] there's always the opportunity to rephrase and add examples [18:55:06.0000] post suggestions to http://whatwg.org/newbug :-) [18:55:55.0000] correct, but i would say that if "this" is an issue which has confused "many people", that might bump this opportunity higher on the priority list. [18:57:00.0000] oh, that particular issue [18:57:13.0000] there's already entire sections that try to explain it [18:57:21.0000] i think teh confusion is mostly amongst people who haven't read the spec [19:00:59.0000] the parsing behavior is just inherently confusing [19:01:09.0000] yeah that too [19:01:28.0000] and even when you read the spec, you don't want to believe that's how things actually work [19:01:36.0000] but it's the way people assume the content models have anything to do with that which is what i'm mostly talking about [19:01:51.0000] because, how dumb would that be, for people to create something that works that way [19:01:58.0000] Hixie: yeah [19:02:17.0000] Hixie: maybe we could put some icon there [19:02:24.0000] like the fingerprinting icon [19:02:33.0000] to say, this is not a UA requirements [19:03:30.0000] I suggest this icon: https://avatars2.githubusercontent.com/u/568252?s=140 [19:04:10.0000] argh! stupid firefox :( crashes on the single page spec page. haha MikeSmith [19:04:36.0000] MikeSmith: well, that's what developers.whatwg.org is supposed to be, really [19:04:39.0000] my firefox doesn't crash on it, just takes a long time to load [19:04:47.0000] Hixie: true [19:04:54.0000] MikeSmith: but the problem is people assume that the authoring requirements _are_ UA requirements [19:05:00.0000] right [19:05:04.0000] i mean, it's UA requirements this contributor is looking for [19:05:17.0000] so saying "don't look here, this is for UAs" might even be what is making them look at content models [19:05:30.0000] maybe we need a version with all the authoring requirements suppressed, and just have the UA requirements [19:05:33.0000] seriously [19:06:07.0000] that would be... interesting [19:06:09.0000] which I realize might require marking stuff with class=author [19:06:10.0000] a lot of work to do though [19:06:14.0000] yeah [19:06:26.0000] I remember when you did the class=impl change [19:12:55.0000] there's a pretty deep difference between implementation requirements and conformance criteria, the normative language that basically says "you must do this (but if you don't, everything will still work in a precisely defined way)" has always felt like a bad use of "must" to me [19:13:30.0000] don't really know how it could be fixed... [19:16:02.0000] zewt: well the "Content model" sections don't contain any musts anyway [19:16:16.0000] or anything else normative-ish looking [19:16:32.0000] yeah, i just have one "must" in the definition of "content model" [19:16:45.0000] oh? [19:16:49.0000] /me wonders which [19:17:17.0000] "content model. A normative description of what content must be included as children and descendants of the element." [19:17:31.0000] well, replying to authoring requirements vs. UA requirements [19:17:53.0000] it's actually redundant with "An HTML element must have contents that match the requirements described in the element's content model." a few paragraphs later [19:27:29.0000] Hixie: as another data point about the confusion: https://www.w3.org/Bugs/Public/show_bug.cgi?id=25501 [19:28:58.0000] confusion about "content models" vs types of content (flow, phrasing, etc.) [19:29:49.0000] and asking that "empty" be defined as a content type [19:30:34.0000] so yeah it is confusing [19:31:49.0000] that's a different problem. That's the problem of people who forget the english language while reading the spec. :-P [19:48:16.0000] Hixie: not so fair, man :-) [19:48:39.0000] come on, that bug is basically just the guy saying he doesn't know what "empty" means [19:48:47.0000] it's kind of terms-of-art usage of English [19:49:00.0000] Hixie: yeah true in that case [19:49:13.0000] it does come down to that [19:49:25.0000] i guess the real problem is assuming that a word is a term of art when it's just english [19:49:37.0000] hmm yeah [19:49:45.0000] but i don't know what else the term would mean [19:49:46.0000] sometimes empty just means empty [19:49:52.0000] yup [19:50:09.0000] there is the void vs empty thing though [19:50:30.0000] well "void" doesn't mean anything clear, so assuming it's a term of art seems reasonable :-) [19:50:31.0000] which is good, and an improvement over what we had before [19:50:37.0000] sure [21:09:38.0000] sweet lord there's a lot of script data states [21:31:55.0000] +1 for a for-UAs version of the spec, containing the actually-normative stuff. [02:30:40.0000] Hixie: aaah, we already have listener observation of sorts with http://xhr.spec.whatwg.org/#upload-events-flag [02:30:47.0000] /me feels bad [06:52:56.0000] annevk: i wonder if it would be web-compat to change that to "set the upload events flag if the xhr.upload property has been accessed" [06:53:48.0000] zewt: possible [06:54:08.0000] There's also http://xhr.spec.whatwg.org/#garbage-collection [06:54:10.0000] a different sort of lameness, but maybe a lesser one [06:54:56.0000] that seems okay, i think... [06:55:31.0000] (also a little weird that accessing .upload would mean requests complete even if you throw away the object when they wouldn't otherwise, but not catastrophic) [06:55:49.0000] oh, you mean another event listener check. hmm [06:56:32.0000] personally I intuitively thought that XHR would always complete the request if I let go of it, even if I wasn't listening for anything on it [06:56:46.0000] is there a reason to not just never GC the object while the request is in the air? [06:58:33.0000] could see interop issues there, eg. function send_keepalive() { var xhr = new XHR(); xhr.open("http://api.server.com/ping"); xhr.send(); } may or may not actually send the ping, depending on GC (if I understand correctly) [06:58:33.0000] EventSource and WebSocket do the same; they might need it more, granted [06:59:09.0000] makes more sense for streaming things that would never close on their own [06:59:14.0000] Heh, interesting case [07:00:34.0000] zewt: actually in that case it would deliver the ping [07:00:48.0000] zewt: at HEADERS_RECEIVED the ping is already at the other side [07:01:35.0000] zewt: which does argue that the upload related check is wrong... [07:05:57.0000] annevk: can you clarify the and/or gruoping in that section [07:06:27.0000] An ... (state is opened and send() flag is set), (state is RECEIVED), or (state is LOADING and one of the following is true)? [07:06:35.0000] also grouping [07:07:10.0000] (that's what it seems like based on what it's trying to do, just a bit unobvious from a naive reading) [07:07:13.0000] Hehe, source has "Based on EventSource and WebSocket. Not sure what I am doing." [07:08:28.0000] might argue that this is okay from an event API standpoint, because the difference is unobservable to script (it just means "if there are listeners to hear it, do keep going"), but ... [07:09:29.0000] so in this case i guess what it's really doing is saying "if nobody is listening, you don't actually have to read the whole POST body" [07:10:02.0000] In a tree falls in the forest... [07:10:04.0000] maybe that's arguably an implementation detail anyway [07:10:09.0000] *If, dammit [07:10:43.0000] forest.dispatchEvent(new Event("TreeFell")); [07:12:46.0000] surprised that websocket is spying on event listeners, i thought that was only done with event handlers [07:12:51.0000] zewt: added a commit [07:13:37.0000] zewt: it's only about the response body [07:14:10.0000] zewt: might be observable from the server if the UA actively kills the connection [07:14:43.0000] right, but there are plenty of things we seem okay with being observable from the server that we're not in script... [07:14:59.0000] (resource caching behavior, etc) [07:16:29.0000] the upload events flag is the bigger one, though [07:19:00.0000] afk, heading to work [07:23:06.0000] Yeah, not sure how to fix that or if [07:23:49.0000] "Invariants" seem to be screwed over left and right [08:12:08.0000] annevk: you forgot to add `s around your HTML tags so your issue makes no sense :P [08:13:50.0000] Domenic_: fixored [08:14:01.0000] Domenic_: seems weird GH removes stuff it doesn't do anything with [08:16:24.0000] Domenic_: what behavior-only objects are you talking about? [08:16:28.0000] /me hasn't seen many [08:17:24.0000] annevk: I'm trying to get the spec editor to tell me if there's hidden state, but https://dvcs.w3.org/hg/webcrypto-api/raw-file/tip/spec/Overview.html#dfn-SubtleCrypto seems at first glance to be behavior-only. [08:19:17.0000] Domenic_: with functions that access privileged APIs elsewhere somehow? [08:19:29.0000] annevk: I don't understand the question? [08:20:02.0000] Domenic_: https://dvcs.w3.org/hg/webcrypto-api/raw-file/tip/spec/Overview.html#dfn-SubtleCrypto-method-encrypt can't be implemented without "magic" [08:20:30.0000] annevk: why do you say that? It just runs some well-specified algorithms, which are accessible to any Turing machine. [08:22:06.0000] Domenic_: as in, all those functions want to share some algorithms, such as "normalize" [08:22:20.0000] (which appears to be a broken link...) [08:22:21.0000] Sure, functions can call other functions ... [08:24:40.0000] Domenic_: anyway, any other examples? [08:26:53.0000] annevk: https://dvcs.w3.org/hg/quota/raw-file/tip/Overview.html#idl-def-StorageQuota, given that supportedTypes is a constant (not mutable state) [08:28:07.0000] That has NoInterfaceObject, so can you distinguish it from an ordinary object? [08:30:24.0000] annevk: I believe Object.getPrototypeOf(navigator.storageQuota) would give back a StorageQuota, not Object. [08:30:34.0000] even if window.StorageQuota doesn't exist. [08:30:58.0000] Ah, I suspect that's probably true [08:31:33.0000] Yeah, first sentence of http://heycam.github.io/webidl/#interface-prototype-object [08:38:08.0000] Modules and IDL support for them is what we need for a lot of this stuff [08:38:17.0000] indeeeed [08:38:26.0000] probably worth waiting on implementations for that though. [08:38:31.0000] Do modules have a concept of being scoped to particular types of realms? [08:38:43.0000] E.g. we can't have DOM in workers... [08:39:08.0000] I imagine different realms would have different built-in module loaders, and thus different built-in modules. [09:21:40.0000] good morning, Whatwg! [09:37:12.0000] cwilso, I got the eternal question from our implementor... Why XML and Http in XHR? [09:37:20.0000] Length? [09:40:12.0000] Conventions [09:40:33.0000] It was HttpRequest shipped as part of the XML library [09:41:02.0000] see /topic ? [09:41:45.0000] :) [09:42:11.0000] though I guess actually people are actually after the story [09:42:15.0000] +sometimes [09:42:22.0000] -actually [09:42:41.0000] I'm certainly not going to suggest implementing XmlHttpRequest in Servo [09:42:55.0000] Well no [09:43:12.0000] If you were picking the name you would probably choose Request [09:43:32.0000] And not implement this API at all [09:43:36.0000] You'd pick fetch and you'd pass it a Request [09:43:40.0000] oh wait [09:54:55.0000] the inspecting of event listeners is to make GC not observable [09:57:36.0000] grr is there nothing less crappy than gettext for python localization [10:00:09.0000] is there ever anything less crappy than gettext? [10:00:25.0000] I mean, for the same purpose [10:00:54.0000] /me prepares to take notes [10:01:17.0000] dunno, i just want something not crappy [10:01:53.0000] well, what about it is your problem? [10:04:36.0000] at the moment it happens to be fighting with pygettext, which apparently has no support for extracting comments for strings [10:06:21.0000] that does look bad [10:09:49.0000] zewt: have you tried xgettext? [10:13:47.0000] zewt: it appears to be *intended* to work with Python, and it of course supports such things [10:43:50.0000] Ms2ger: why do we have readyState in XMLHttpRequest as a `short`? [10:44:29.0000] Why not? [10:45:10.0000] Ms2ger: it runs from 0-4. And I don't think that's going to increase much -- at least not above 8 [13:19:51.0000] annevk: GH doesn't "remove stuff it doesn't do anything with". If you don't wrap HTML elements in `s, *they're HTML elements*, and so you don't see them in the visible text obviously. [13:40:09.0000] i don't know any good reason to specify ints as "shorts" in most cases, they're just integers [13:40:39.0000] SamB: works a bit better, but seems to suck at recursively handling file trees... [13:40:50.0000] TabAtkins: it probably sanitizes them though. [13:41:00.0000] zewt: isn't that what find is for [13:41:06.0000] Domenic_: Possibly, yeah. [13:41:09.0000] sure, but i'm having to jump a lot of hoops [13:41:12.0000] I know it allows and such through. [13:41:43.0000] i need to say xgettext -j to make it combine with a previous execution (in case xargs splits it into multiple invocations)... but that means I need to delete the file before running it, so it doesn't join with a previous execution... but if I say -j when the file *doesn't* already exist, it throws an error [13:41:43.0000] yeah, what the heck is short [13:41:57.0000] (are there any specs that already use short for something?) [13:42:14.0000] zewt: hmm. [13:42:17.0000] so i have to delete the file, then only add -j on the second and further invocations, which i don't know how to do with xargs [13:42:38.0000] there's a -D "add DIRETORY to list for input files search" which sounds like what I want, but it doesn't seem to actually work [13:42:45.0000] zewt: maybe --files-from= helps? [13:43:14.0000] it also says "If input file is -, standard input is read", which should let me avoid xargs entirely, but that doesn't seem to work either [13:43:40.0000] I think -D is for if you don't know where the files actually are [13:44:00.0000] possibly to make things "easier" for out-of-tree builds [13:44:11.0000] the word "footgun" comes to mind [13:44:19.0000] hence the scare quotes [13:45:17.0000] I don't know, maybe it's actually possible to use it in a useful way [13:45:21.0000] wonder if the python gettext module has code to compile .po's so i can do it at runtime, precompiling to .mo is pointless for me (it's a server) [13:46:14.0000] http://docs.python.org/library/gettext should have the answer [13:46:24.0000] oh well, guess i'll write a script to invoke xgettext, it'll take less time [13:46:58.0000] zewt: so did you try --files-from=<(find ...) [13:47:22.0000] wouldn't i have to output the file list to a temp file (maybe --files-from=- works) [13:47:53.0000] oh hey that does seem to work [13:47:57.0000] yes it does [13:48:11.0000] it doesn't know what the heck to do with CWEB's .w files, but boy does it find them [13:48:41.0000] i think my stuff is encapsulated enough that it won't run into a bunch of stuff to confuse it [13:48:56.0000] I was just trying a random example that'd work in the tree I was in [13:49:11.0000] like when I was trying to package a web view's javascript files inside a Unity package, which Unity promptly interpreted as JS and tried to run as Unity code [13:49:15.0000] thanks, stop that [13:49:22.0000] mostly in case I was not remembering my shell syntax correctly [13:49:48.0000] gettext is still very... 90s? [13:50:12.0000] woah, copyright starts in '95 ? [13:50:27.0000] I was expecting '8x [13:54:04.0000] hmm, changelog for GCC starts in '91 with "Freshly created ChangeLog."; no idea when gcc was created [13:54:52.0000] "GCC 1.0 was released in 1987" [13:55:16.0000] /me goes to do "git log Makefile.in" in the binutils-gdb repo, which is subsetted from the old src/ repo ... [13:56:06.0000] /me tries again "git log -- Makefile" ... [13:57:11.0000] nothing [14:48:31.0000] "1.0 released in 1987" suggests they started on it circa 1972 [14:58:51.0000] hmm [14:59:04.0000] it's apparently been a while since i read my bugmail [14:59:10.0000] 4,767 e-mails... [14:59:14.0000] (and that's after filtering) [15:25:00.0000] TabAtkins: can you use "internal slots" and [[x]] in http://dev.w3.org/csswg/css-font-loading/, instead of "internal attributes" and [x] ? [15:25:18.0000] TabAtkins: we want to encourage more people to do that kind of thing, but it's harder to argue for when everyone is doing something different. [15:29:48.0000] Domenic_: Absolutely. [15:30:02.0000] Though using [[]] is kinda annoying in Bikeshed. Hmm. [15:30:14.0000] I think I can write [[foo]]. [15:30:43.0000] TabAtkins: awesome, thanks! And, uh, wow that sounds hard. [15:32:34.0000] It's just that Bikeshed treats [[foo]] as a biblio ref. [15:33:00.0000] ah tricksy [15:33:00.0000] (A behavior inherited from the old CSSWG preprocessor, and shared by Anolis, I think.) [15:33:21.0000] ah i see, so a behavior that would move peoples' cheese if changed [15:34:17.0000] The only thing I can think of is inventing something arcane that gets translated to [[x]], e.g. \\x// [15:34:34.0000] Yeah, it would break virtually every Bikeshedded spec if changed. [15:34:48.0000] What kind of cheese is this, and — assuming it is nice cheese — could they move it this way? [15:34:57.0000] What I might do instead is add a metadata field to let you specify things that aren't biblio refs. [15:35:06.0000] For now, though, I'm just adding a comment to break things up. [15:35:30.0000] We do want to add internal slot declarations to WebIDL of some sort [15:35:37.0000] Cool. [15:36:30.0000] Just checked - ReSpec uses that syntax too. [15:36:39.0000] waaah waaaaah [15:36:55.0000] Any chance y'all could come up with a syntax that doesn't directly impinge on every extant spec preprocessor? [15:37:05.0000] i mean, we could ask ES to change, but seems unlikely... [15:37:11.0000] or we could just be inconsistent with ES... [15:37:17.0000] but they're supposed to be the same concept [15:37:17.0000] I'm fine with the latter. [15:37:26.0000] don't suppose you could ask them not to process the IDL for bibliographic references [15:37:48.0000] SamB: It won't just appear in IDL. [15:37:54.0000] i'm a bit surprised that it doesn't leave [[x]]s alone if there's no x in the references section. [15:37:57.0000] It's scattered throughout the spec, every time you reference the slot. [15:38:12.0000] Domenic_: A good preprocessor *tells* you you've made a typo when it can't find "x" in the refs database. [15:38:26.0000] yeah, that makes sense. [15:38:27.0000] And they *generate* the references sections (that's why preprocessor exist) [15:38:43.0000] oh right i forgot about that feature. [15:38:51.0000] using the database of well-known references [15:38:54.0000] Yup. [15:39:09.0000] You people writing specs in GHMarkdown are missing out on a lot. ^_^ [15:39:16.0000] I guess I'd say use {{x}} in source and have preprocessor convert to [[x]] [15:39:17.0000] heh [15:39:20.0000] /me is getting closer to having most of Markdown implemented in Bikeshed. [15:39:29.0000] I never quite understood how that part was supposed to work with each thing in its own repository, but then I'm thinking of bibTeX ... [15:39:42.0000] SamB: What part? [15:39:58.0000] the part where you have a big database [15:40:13.0000] You keep a master database of refs that everyone updates. [15:40:28.0000] easier with spec tools [15:40:43.0000] The biblio dbs are maintained automatically. Bikeshed's linking database is done automatically, though. [15:40:53.0000] Once it knows a spec's location, it stays up-to-date. [15:41:07.0000] SamB: I don't understand what you mean. [15:41:21.0000] so what you're saying is, there's only one place we have to change [[HTML]]'s URL in, and then everything will be better... [15:41:26.0000] Domenic_: Yes. [15:41:42.0000] ([[HTML]] already points to the proper spec in Bikeshed's DB.) [15:41:50.0000] nice [15:42:21.0000] (Though [[HTML5]] points to the W3C spec.) [15:43:07.0000] Domenic_: I can always give you a Bikeshed crash-course if you want. It works fine with the GH workflow - 's spec is Bikeshedded, and uses a gh-pages as master for displaying. [15:45:57.0000] TabAtkins: probably a good idea, when it's time for me to get serious about prettifying streams. I'll let you know ^_^ [15:46:22.0000] kk, but Bikeshed is helpful during initial writing too, as it makes sure you don't typo links and such. [15:46:29.0000] Let's you draw railroad diagrams. [15:46:32.0000] Other cool things. _^ [15:58:05.0000] [[HTML5]] should be a fatal error :-P [16:53:45.0000] Domenic_: Why are [[Foo]] things called private slots rather than private attributes? [16:54:50.0000] TabAtkins: attributes is a WebIDL-ism. [16:55:03.0000] They used to be called internal data properties in ES [16:55:20.0000] then we thought that was confusing since you can't e.g. getOwnPropertyDescriptor them [16:55:24.0000] also it was long [16:55:28.0000] so they became internal slots [16:55:31.0000] Okay. [16:55:56.0000] Anyway, just pushed the change to Font Loading. [16:56:10.0000] May take a few minutes to show up. [16:57:44.0000] yaaaay :) thanks man 2014-05-02 [17:05:17.0000] Happy to help get terminology confluence. [18:10:30.0000] TabAtkins: does HTML5 emit the appropriate deprecation warnings? [18:10:51.0000] or, yeah, fatal error haha [18:11:42.0000] No, I don't do anything fancy with biblio refs. [18:11:47.0000] Though I probably should. [22:22:54.0000] cabanier: fyi see my review comments on that addHitRegion test https://critic.hoppipolla.co.uk/showcomments?review=1239&filter=all [22:23:06.0000] MikeSmith: saw it [22:23:19.0000] yeah I wish I hadn't merged that [22:23:32.0000] well, I guess it's not a bad test [22:23:56.0000] anyway, if he's not keen on updating it, I'll do it myself [22:24:35.0000] but regardless the source for it needs to be in the yaml file with the sources for the other tests [22:24:54.0000] cabanier: btw thanks for your other canvas tests and merges [22:25:38.0000] MikeSmith: np. The WebKit people want me to fix their IDL compiler instead of just patching the code so it passes the tests [22:25:52.0000] oh geez [22:26:00.0000] MikeSmith: I hope that won't suck up too much time... [22:26:01.0000] cabanier: that's a lot of work man [22:26:08.0000] well I think it will [22:26:17.0000] but I hope it won't :-) [22:26:42.0000] MikeSmith: it's actually not so bad. I already fixed the compiler but I worry that I'm now changing the behavior of dozens of interfaces [22:26:49.0000] oh [22:27:02.0000] well it would be nice to have it fixed [22:27:13.0000] MikeSmith: yes [22:27:42.0000] if the current behavior of those interfaces is not conformannt, better to break it as early as possible rather than later [22:27:50.0000] MikeSmith: yeah [22:27:51.0000] I would think [22:28:12.0000] MikeSmith: I'm finding a lot of issues where the wrong exception is thrown, or not thrown at all [22:30:06.0000] cabanier: good that you're finding those, but sad that nobody else has taken time to do it yet so you now get stuck with it [22:31:09.0000] yeah [22:31:32.0000] MikeSmith: I hope to get it in before Safari forks [22:31:45.0000] MikeSmith: otherwise it's an another year before it would update [22:44:30.0000] cabanier: by "before Safari forks" you mean before the branch for the next release? [22:44:45.0000] MikeSmith: yes [22:44:49.0000] ok [23:19:20.0000] TabAtkins: fwiw i think anolis doesn't do [[foo]] except if you give a flag like --enable-w3c-crazy-substitutions or some such [23:34:54.0000] speaking of empty content model, doesn't the spec in fact allow whitespace and comments (and PIs) for "empty"? [23:36:17.0000] "Inter-element whitespace, comment nodes, and processing instruction nodes must be ignored when establishing whether an element's contents match the element's content model or not" [23:36:44.0000] zcorpan: if it's possible to parse them that way, sure ... [23:37:01.0000] so not exactly for in HTML syntax [23:37:01.0000] SamB: what do you mean? [23:37:19.0000] oh ok [23:37:43.0000] well you could create the DOM with script and forget about the syntax [23:37:52.0000] true [23:39:05.0000] so a definition of "empty" that reminds the above exception might actually be good to have [23:39:13.0000] Hixie: ^ [00:21:42.0000] who was it that was reminding us other day that it's a bad idea to create new specs that load the precious first bytes of our documents down with yet more stuff? https://docs.google.com/document/d/17jg1RRL3RI969cLwbKBIcoGDsPwqaEdBxafGNYGwiY4/edit [00:23:48.0000] do I read correctly that requires each document instance to have [00:24:14.0000] ... both a new meta@name element and a new link@rel element [00:26:34.0000] with a new content attribute whose value the UA has parse and that's an arbitrarily long list of arbitrarily long selector expressions [02:37:00.0000] anyone know what happened to hspace/vspace on table in gecko? http://software.hixie.ch/utilities/js/live-dom-viewer/?saved=2984 looks like it's not supported anymore but https://bugzilla.mozilla.org/show_bug.cgi?id=725646 is not FIXED yet [02:43:59.0000] http://www.browserstack.com/screenshots/2b6be040b79a037b7d05c52f555295e5828762cf ... [02:44:29.0000] oh only quirks mode [02:51:26.0000] I like the aggressive "Standardization plan" in https://docs.google.com/document/d/17jg1RRL3RI969cLwbKBIcoGDsPwqaEdBxafGNYGwiY4/edit#heading=h.tndapzbk8g3p [06:30:37.0000] MikeSmith: it was slightlyoff [06:31:11.0000] https://twitter.com/PointedEars2/status/462218792576499714 o_O [06:32:03.0000] annevk: never argue with trolls/idiots/etc — people might not know the difference [06:34:06.0000] They'll pull you down to their level and beat you on experience? [06:34:44.0000] damn right [06:34:48.0000] heh [06:35:09.0000] Domenic_: yeah I know, it was more of rhetorical question :-) [06:35:41.0000] Haha OK, just woke up, rhetorical devices are too subtle for me right now... [06:37:19.0000] heh [06:38:07.0000] Interesting. Do they really think the only part of this that needs standardization is the rel types? [06:42:07.0000] Domenic_: I reckon they know better. I think they just need to add more detail to that section yet [06:42:20.0000] or I hope that's the case at least [06:42:44.0000] Ah, MikeSmith, the eternal optimist [06:43:19.0000] Ms2ger: that's me to a T [06:43:41.0000] photo of me being optimistic: http://goo.gl/5uUnK [06:50:06.0000] darobin: these look like issues with the WebIDL parser rather than syntax issues, am I right? [06:50:08.0000] https://github.com/slightlyoff/ServiceWorker/issues/259 [06:53:04.0000] ^ pointedears reminds of comp.lang.javascript. He was already trolling back then. [06:56:43.0000] tobie__: On an entirely different topic, I think to get the documentation for testharness and so on onto testthewebforward.org we will need to move away from using GitHub-Pages directly (but can still use Jekyll). This is because GHP doesn't support any kind of useful include mechanism. What we want is basically http://octopress.org/docs/plugins/render-partial/ [06:57:24.0000] So we could do something like {% render_partial ../_resources/testharness.js/docs/api.md %} [06:57:54.0000] oh... [06:58:07.0000] includes don't let you do that/ [06:58:10.0000] ? [06:58:38.0000] (I guess they're scoped to the include folder) [06:58:49.0000] Yeah, they seem to be scoped [06:59:00.0000] (sounds like a wise thing to do) [06:59:14.0000] yeah works for me. [06:59:22.0000] Can we still get auto-deploy? [07:00:51.0000] I'm not sure. I think you will have to build locally, push, and it will deploy [07:00:59.0000] aiui Domenic_ uses this setup [07:01:28.0000] mmm. [07:01:52.0000] that sucks. [07:01:56.0000] tobie__: yes, and these are all easy fixes too [07:01:59.0000] feel free to file [07:02:06.0000] darobin: ok [07:02:21.0000] happy to look over it if you have pointers. [07:02:29.0000] it's a blocker for me atm [07:04:11.0000] darobin: filed https://github.com/darobin/webidl2.js/issues/5 [07:04:48.0000] jgraham: concerned about the lack of auto-deploy feature. [07:05:00.0000] jgraham: could we have a script or something to help? [07:05:25.0000] if not, please check with rhauck before you go ahead. [07:05:51.0000] tobie__: So I don't know exactly what the situation is, I haven't used either GitHub-Pages or Octopress before [07:06:08.0000] k [07:06:23.0000] But of course I don't plan to change anything without agreement that it's an improvement [07:06:24.0000] you can't have a special build of jekyll on gh-pages [07:07:13.0000] so if you want to add stuff, you;re effectively loosing the auto-deploy feature (you have to build locally and push the static files only) [07:07:41.0000] http://octopress.org/docs/deploying/github/ [07:07:50.0000] that said, you can build a script that does it for you (listens to changes on master, builds, and pushes to gh-pages) [07:08:20.0000] jgraham: right [07:08:36.0000] /me running out. back later. [07:09:00.0000] Yeah, I think it's worth fixing the deploy issue to get the docs in sync with upstream [07:23:42.0000] darobin: "Wants to keep the web in the Dark Ages" makes for a cool Twitter bio [07:24:00.0000] annevk: heck yay! I'd totally go for it man [07:24:05.0000] I'm almost jealous there [07:25:06.0000] Let's just drop that "almost" [07:26:46.0000] hahaha [07:40:26.0000] "Wants to bomb the Web back into the Stone Age" would be even better [07:47:06.0000] He's not American [07:56:17.0000] jgraham: tobie__: you could also use travis CI to run the deploy [07:56:48.0000] use it as a post-commit hook to run the appropriate script [07:57:02.0000] use the encrypted credentials feature to include necessary keys for pushing to gh-pages [11:34:17.0000] foolip: i'm surprised that browsers append attributes to elements as they're parsing the attributes, since that means that they are likely creating redundant elements. [11:34:17.0000] foolip: for example, if you're parsing , then you need to pass a token with {start tag, name:isindex, attirubtes: foo=bar} to the tree construction, which then doesn't create an "isindex" element [12:53:21.0000] does canPlayType() return a: state, kind, mode, reason, rule, or type [12:53:51.0000] maybe it's a new thing, a result [13:21:03.0000] can people actually use canPlayType()? Doesn't it return things like "maybe"? [13:23:04.0000] vine.co uses mediaelement.js which uses it like so: https://github.com/johndyer/mediaelement/blob/master/src/js/me-featuredetection.js#L37-L44 [13:23:09.0000] an api that returns "maybe" and "probably" would have to be in the running for least usable api ever [13:24:02.0000] Well it's honest [13:24:29.0000] (doesn't look like a very realistic api input, either--video files often have alternative data formats, so a list of codecs used by the file doesn't tell you much) [13:24:45.0000] (no doubt that's a partial cause of the silly results...) [13:25:03.0000] The point is that the browser generally doesn't know if the media framework will actually be able to play the video without trying to play it [13:25:22.0000] (i assume the main cause is that you also need to know things like the encoding profile, not just the codec) [13:25:50.0000] my parenthetical quota has been exceeded [13:25:57.0000] mediaelement's optional "source chooser" plugin uses it like so: https://github.com/johndyer/mediaelement/blob/666b6adaea247a5f7dd1b788190a100daa4f26a3/src/js/mep-feature-sourcechooser.js#L53-L55 [13:26:02.0000] vine.co doesn't use that [13:26:56.0000] yeah the only meaningful result is the "definitely can't" one i guess [13:27:11.0000] /me marks zewt down as "not a lisp programmer" [13:27:31.0000] there are far more useless APIs in the web platform than canPlayType [13:27:56.0000] it's a competitive field [13:28:16.0000] in other news, i just typed "" when trying to write "" [13:28:19.0000] blimey [13:28:28.0000] I remember a long time ago reading an interview with Hixie where he was talking about all the vestigial things. I think the useless pushState argument was near the top of his list? Anyway it was refreshing honesty for someone new to all this. [13:28:49.0000] ("long time" ~ "a year"?) [13:29:12.0000] the title argument? [13:29:24.0000] (if I want to set the title too, I'll ... just set the title) [13:30:14.0000] "... which leads to APIs that make no sense, like pushState(), which has a required argument that is ignored." [13:30:18.0000] http://html5doctor.com/interview-with-ian-hickson-html-editor/ [13:31:17.0000] none of the arguments are ignored, right? [13:32:55.0000] Firefox ignores title [13:33:44.0000] Spec says: "Note: The title is purely advisory. User agents might use the title in the user interface." [13:33:48.0000] the title argument is ignored by everyone [13:33:53.0000] it's the saddest thing [13:34:31.0000] Do you think the spec should be more explicit about that? [13:34:44.0000] I guess the argument is they could put it in some session history UI [13:34:52.0000] But they definitely shouldn't change document.title [13:38:07.0000] they should put it in some session history UI [13:38:26.0000] if it was guaranteed to be ignored, i could say pushState(null, null, url) instead of pushState(null, document.title, url) ... 95% of the time the URL is all I want to change anyway [13:39:52.0000] that said, i'm sure i've used the title argument expecting it to change the title at some point, and if that works in some browsers and is ignored in others, that's obviously bad [13:41:39.0000] i think it should be a normative requirement that it doesn't change the same things document.title changes [13:41:45.0000] to avoid your overcaution [13:42:21.0000] sorry, my what? :) [13:42:51.0000] Your caution [14:17:40.0000] Domenic_: smart suggestion re relying on Travis CI [14:34:51.0000] hey [15:09:59.0000] gsnedders: fyi, the tokeniser tests don't check any of the script data states, or the cadatasection state [15:37:09.0000] dear firefox stop letting tabs steal focus with alert() [15:37:51.0000] granted chrome is now even worse than firefox with modals, back in the dark ages [15:43:45.0000] wow, chrome alerts are even more broken than I knew [15:44:44.0000] if one *window* has an alert open, focusing another browser window (not tab) lets me access that tab (if it's a different origin), but browser chrome is wedged (can't change tabs or access the address bar), and if I switch to another window that shares a context with the one with an alert open, the other browser window is completely frozen 2014-05-03 [18:05:35.0000] Hixie: yeah, I know individually the coverage is not great [18:05:46.0000] Hixie: the sum of both the tokenizer and tree construction is good [18:07:55.0000] (this is one of the things I want to fix this summer) [18:45:17.0000] gsnedders: k [18:46:28.0000] gsnedders: i was just amused because i've been slowly writing code as each test fails, and then suddenly i was passing all the tokeniser tests, and i had like 15 "XXX" bits in my tokeniser still... [20:33:15.0000] woah [20:33:26.0000] ... [20:33:32.0000] i bet old UAs didn't parse that like we do now [20:34:27.0000] i wonder how i can test that [20:34:32.0000] i haven't seen an old browser in years [20:42:00.0000] Hixie: I seem to remember thet [20:42:30.0000] *there being some thing that let you test in Netscape 4 [20:42:37.0000] /me looks [20:43:21.0000] dunno how you'd even test that in n4 [20:43:51.0000] http://virtuallyfun.superglobalmegacorp.com/?p=3866 [20:45:00.0000] server-side thing, I guessーit just generates PNGs you can view [20:45:02.0000] anyway [20:47:00.0000] Hixie: I don't understand what's so exceptional about how gets parsed [20:47:25.0000] looking at the DOM for it I see pretty much what I'd expect [20:49:59.0000] MikeSmith: it's invalid xml, for example. [20:50:11.0000] MikeSmith: no charrefs before the root element [20:50:18.0000] ok [20:50:24.0000] ah yeah [20:51:43.0000] in other news, I'm reading the statement "it is extremely rare in JavaScript to rely on checking a specific exception type" and wondering why that's true [20:51:58.0000] I mean I assume that's just stating common practice [20:52:43.0000] that is, that web devs don't typically check for specific exceptions, something either just throws or it doesn't [20:52:47.0000] I guess [20:52:58.0000] well, because javascript as a language sucks at it, presumably [20:53:07.0000] doesn't really suck at that [20:53:21.0000] compared to other languages [20:53:22.0000] does it? [20:53:38.0000] i mean i guess some have special syntax for different handlers for different classes of exceptions [20:53:40.0000] https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/try...catch didn't even know about the "var if condition" syntax, is that widely supported? [20:53:52.0000] never seen it in code [20:54:16.0000] makes me wonder why we go to the trouble of defining specific exceptions if in practice nobody cares, or no code changes any behavior based on what the exception is [20:54:24.0000] filtering exception handlers by class is exception handling 101, i've always found it bizarre that javascript has no syntax for it (but maybe the "if" syntax is meant for that) [20:54:27.0000] oh wow [20:54:31.0000] i didn't know about that either [20:54:41.0000] trying it [20:55:02.0000] Note: This functionality is not part of the ECMAScript specification. [20:55:18.0000] all that matters is whether it's implemented, heh [20:55:34.0000] so maybe people don't try to do it because they already know it doesn't work reliably? or the language doesn't provide what's needed for proper exception handling? [20:55:40.0000] parse error in firefox [20:55:47.0000] s/language/platform [20:56:15.0000] maybe because most day-to-day errors come via error events, rather than exceptions? [20:56:24.0000] (since most code is still async in the UI thread, not worker stuff) [20:56:56.0000] most exception handlers i can think of writing are just "catch all errors to log them to the server, because we don't really except to get any exceptions at all" [20:57:04.0000] which is definitely unlike most other languages i work with [20:57:32.0000] zewt: I see, ok [20:57:35.0000] that makes sense [20:57:40.0000] (platforms, rather) [20:57:55.0000] /me wonders what Domenic_ would say [20:58:12.0000] I'm not an idiomatic JavaScript programmer [20:58:28.0000] i know i write way more exception handlers in python than JS, in any case [20:58:51.0000] and i'd go nuts if python didn't have class filtering [20:59:19.0000] well, for sure I write a ton in Java [20:59:28.0000] not that I'm proud of it [20:59:36.0000] that's just what Java forces on me [20:59:42.0000] java shoves them down your throat, heh [20:59:53.0000] i tend to make all of my exceptions unchecked in java to avoid that nonsense [21:00:23.0000] well for me it works well within the Java universe [21:01:07.0000] e.g., the validator code is exception handling stacked on exception handling all over the place [21:01:32.0000] but I guess that's the nature of that particular app [21:01:56.0000] inherent in it, in that its entire purpose is to report errors of varying kinds [21:02:09.0000] i find it's one of the many things that makes java a nightmarish, developer-hostile environment seemingly designed by academics with zero practical experience [21:02:20.0000] (re: java is probably my single most hated language) [21:02:38.0000] (combination of really terrible language + forced to use it for Android) [21:02:47.0000] to me it's more a fact of life I guess [21:03:02.0000] I suppose I should question it terribleness more [21:03:26.0000] but it's like trying to fix the government or something [21:03:30.0000] i picture some college kid just being told by his CS201 prof "all errors should be declared explicitly and you should handle every error deliberately" going on to bake that nonsense into a langauge to force it on everyone else [21:04:02.0000] also language [21:04:03.0000] oh well yeah [21:04:22.0000] it shouldn't set a precedent for anything else, that's for sure [21:04:33.0000] though I guess it unfortunately has [21:05:32.0000] java should serve as a cautionary tale to other languages, and nothing more [21:05:33.0000] heh [21:06:52.0000] but getting back to the platform, the vast majority of the failures in the DOM test suite apparently come down to implementations not throwing the exceptions that the spec says they should throw [21:07:31.0000] and I can say that holds true for some other parts of the web-platform-tests suite [21:08:13.0000] and those wrong-exception-type bugs tend to stick around for a long time without anybody caring enough to fix them [21:08:53.0000] I assume that's in part because webdevs don't complain about those bugs and demand they be fixed [21:09:05.0000] they're not pain points, in practice [21:09:18.0000] i think separate exception types for "programming error"-class exceptions are probably less useful in general [21:09:27.0000] aok [21:09:35.0000] *ok [21:09:35.0000] you don't usually try to catch different classes of programming errors, you just log them and fix the bug... [21:09:44.0000] I see [21:10:06.0000] and i'm guessing, but DOM exceptions like "you tried to stick a node where it can't go" are probably programming errors? [21:10:52.0000] s, s/you/i/ i guess heh [21:11:34.0000] I guess they are, yeah [21:12:06.0000] but honestly I guess I don't think about that much when I'm reviewing tests or writing them while reading the spec [21:12:09.0000] I guess I should [21:12:40.0000] having them under a base class like python's RuntimeError might be useful (to distinguish them as programming errors), but finer than that I'm not sure [21:12:55.0000] yeah, that [21:13:23.0000] of course Java does things that way too, to a degree [21:13:41.0000] well, Python's RuntimeError and Java's RuntimeError (RuntimeException? been a while, I forget) are different beasts [21:13:48.0000] ah OK [21:14:06.0000] python's means (to my interpretation) "programming error", java's means "unchecked error that you don't have to declare in the exception specifier", right? [21:14:21.0000] yeah pretty much, as far as Java [21:15:40.0000] netsplittin' like it's 1996 [21:16:14.0000] oh, irccloud dying [21:16:32.0000] amusing that you can tell who uses irccloud even with masked hostnames [21:16:37.0000] * timeless (sid4015@firefox/developer/timeless) Quit (Ping timeout: 240 seconds) [03:44:08.0000] Hixie: Specifically it's because those tokenizer states postdate Philip` generating a huge number of tokenizer tests [11:05:51.0000] the `if` syntax in `catch` is a Mozilla-specific extension. MDN is not very clear on these sorts of things. [11:06:21.0000] JS developers do rarely filter exception types, but IMO this is because most exceptions they deal with that could be recovered from result from async ops. (E.g., network failures, permission errors, ...) [11:07:21.0000] I am hopeful that ES will grow better exception-filtering syntax; in the meantime, `promise.catch(e => { if (e.name !== "NetworkError") { throw e; } /* recover from e */ })` will work OK [11:08:03.0000] or `try { result = await promise; } catch (e) { if (e.name !== "NetworkError") { throw e; } /* recover from e */ }` with await syntax [11:08:20.0000] Related es-discuss thread http://esdiscuss.org/topic/try-catch-conditional-exceptions-in-light-of-generators [11:08:29.0000] that syntax needs to be shot into the sun [11:09:09.0000] also http://esdiscuss.org/topic/error-objects-in-w3c-apis [11:09:16.0000] zewt: which syntax? [11:09:21.0000] "e => {" [11:10:53.0000] I sure am glad we don't shoot everything zewt dislikes into the sun [11:10:59.0000] lol yes, arrow functions are <3 [11:11:15.0000] completely unobvious syntax [12:17:08.0000] Throwing the wrong kind of exception is less worse than throwing no exception at all. [12:22:54.0000] Yeah, but it's rather hard to codify the badness of failures in tests. And it's only considered acceptable because people have learnt not to depend on it [12:23:14.0000] If javascript wasn't a bit rubbish, it would be quite important [12:24:01.0000] not sure about that; the APIs tend to depend on people catching exceptions much less, too (at least the async versions) [12:24:15.0000] (less than APIs in most other platforms) [13:40:59.0000] Hi. Why no UNIX timestamp support in datetime fields? It is easier to parse. [14:54:01.0000] Hixie: I was mistaken, the parser doesn't set the attributes one by one, it has them all ready and sets them after creating the element. It was after that point I tried adding a hook that showed up in performance tests. [14:56:43.0000] maybe there is some trick I could use to make Blink match the spec, but it would be a special-case, since almost everywhere parser-set and script-set attributes are handled in the same code path, which is nice. [14:58:38.0000] how hard I try to find that trick depends on how much you hate any of the alternatives that don't have a special case 2014-05-04 [20:29:47.0000] Ooooh! I just realized the W3C cheatsheet supports CSS at-rules. [20:30:01.0000] http://www.w3.org/2009/cheatsheet/#search,@media [20:31:10.0000] Hum, but it doesn’t link to the specification… and my primary use for that cheatsheet is getting quickly to the specification of any HTML element, CSS property, HTML attribute, etc. [22:19:11.0000] if i sent a message to the mailing list with an unregistered email and then register after the fact, must i re-send the message? [22:19:34.0000] or will it go properly through after registration? [22:19:52.0000] eligrey: did you get a bounce yet? [22:20:13.0000] yeah, just now [22:20:31.0000] those tend to contain some kind of instructions [22:22:23.0000] i don't want to spam everyone [22:22:46.0000] can i just send to whatwg⊙lwo when i resend? [22:23:54.0000] maybe you should give special delivery instructions to your mailer daemon ... [22:23:54.0000] as long as the subject is the same it'll keep its context right? [22:24:07.0000] er, message ID is important [22:24:13.0000] in the archives [22:24:48.0000] ideally resend with the same message ID [22:25:47.0000] SamB: i don't think i can configure that header in gmail [22:26:23.0000] It has SMTP access; you could do magic that way. Or you could repost with an apology. [15:43:25.0000] did someone really just say "we should do something we thought was a bad idea because the web has more competition now" 2014-05-05 [23:40:38.0000] TabAtkins: ping (re: colors 4 nit) [23:46:52.0000] TabAtkins: s/constrast/contrast/ in section 10.5 header (and in toc) [02:28:03.0000] "Started work on the next version of http://HTML5test.com . First addition is [bits not in the HTML spec]" [03:36:48.0000] Ms2ger: you're making the common beginner mistake of misinterpreting that stylized ampersand character as a five [03:58:47.0000] is there somewhere a list of css box types, or how to call them. inline, block, replaced inline etc [04:05:08.0000] or is the whole concept of replaced still not properly defined [04:59:57.0000] It seems that on web-developer's version of the document visited links lasts only minutes on my Adnroid device. It's strange... [05:05:05.0000] jgraham: critic didn't like https://github.com/w3c/web-platform-tests/pull/959 [05:30:50.0000] would it make sense to move `atob`/`btoa` to ECMAScript? http://esdiscuss.org/topic/native-base64-utility-methods → would be good if some WHATWG folks chimed in [05:52:24.0000] mathiasbynens: Claude Pache's attitude makes me sad and want to not read through the whole thing :-( [05:53:20.0000] read anyway [05:54:24.0000] zcorpan: tbh i don’t get his point at all. why can “raw binary strings” not be fed to atob? octets only go up to 0xFF [05:54:45.0000] which is exatly what `atob` supports [05:55:19.0000] whatever happened to base64 string <-> TypedArray ? i thought someone proposed extending atob/btoa to do that [05:56:05.0000] was that mixed in with the encoding api stuff? since you also want streaming for base64 [05:56:35.0000] i forget where that left off (it's comparable to string encodings, but not quite a direct fit, iirc) [05:57:32.0000] http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2013-August/040364.html [06:00:54.0000] away [06:01:02.0000] yeah, that morphed into "make it part of TextEncoder/TextDecoder" later in the thread, with the oddity that array->base64 is a "decoder" instead of an "encoder" [06:07:43.0000] seem to recall some discussion since then but can't think of where [07:20:29.0000] Hixie: can you have a look at https://www.w3.org/Bugs/Public/show_bug.cgi?id=25542 ? [08:30:58.0000] High rest time isn't exposed in web workers!? [08:31:11.0000] Where do I file that bug? (Who manages that spec?) [08:32:26.0000] Hixie [08:32:32.0000] there's high rest time? i want in! [08:32:41.0000] could use some rest [08:32:46.0000] Oh, I was going to send zcorpan chocolate [08:33:45.0000] oh, nice [08:34:18.0000] /me should find his way to a post office [08:35:21.0000] Domenic_: you probably want to email http://lists.w3.org/Archives/Public/public-web-perf/ [08:36:36.0000] darobin: sounds good. Given that the spec is already a Rec, will this cause large amounts of work to happen? ("Performance timing v2"?) [08:37:26.0000] Domenic_: not necessarily; the last time someone found a bug with a WebPerf Rec they spinned a new Rec out in 6 days [08:38:06.0000] Woah, didn't know that was possible. [08:38:10.0000] Cool, thanks. [08:38:41.0000] Domenic_: that's why I often say that the Process isn't the problem, you can do a hell of a lot with it; the problem is mostly with the culture [08:40:17.0000] WebPerf++ for creating a new Rec in 6 days [08:41:23.0000] We should publish HTML in WebPerf [08:43:14.0000] :D [08:43:14.0000] Ms2ger: I think that could join the suggestion I made that the HTML WG would use the WHATWG mailing list for technical discussion at that big party with interesting designer drugs :) [08:43:32.0000] Is the syntax [Exposed=Window,Worker]? [08:43:35.0000] #lazyirc [08:44:21.0000] darobin, by the time the HTMLWG notices and starts flam^Wdiscussing, the Rec will have been shipped [08:44:36.0000] Ms2ger: my thoughts exactly *MUAHAHAHAHA* [08:45:26.0000] Domenic_: yeah, but there's an open spec bug about the comma [08:45:56.0000] [Exposed=Window`Worker] [08:47:22.0000] [Exposed=Window┏(°.°)┛Worker] [08:52:08.0000] zcorpan++ [09:11:31.0000] Hixie: intentional that is not allowed in the content model? [09:15:49.0000] good morning, Whatwg! [10:25:46.0000] zcorpan: There were also requests for string->TypedArray (assuming the default 16-point code units of JS strings). Strings are the most compact way to ship binary data inline in scripts. [10:26:45.0000] a-ja: Fixed typo, thanks. [10:49:52.0000] zcorpan: dunno. yet more reasons i hate multi-element designs. [10:50:41.0000] Hixie: i filed a bunch of bugs about script-supporting elements and content models [10:50:52.0000] great [10:50:53.0000] :-P [10:51:05.0000] you're welcome :-) [10:51:21.0000] maybe you should have made template="" an attribute [10:51:27.0000] and script="" [10:55:33.0000] hello, i'm sitting around at work twiddling my thumbs hitting f5 once ina while as google docs 502s for me [10:55:38.0000] welcome to cloud [10:57:09.0000] TabAtkins: i'm not sure exactly what you're describing, but it sounds horrible [10:57:55.0000] incidentally, i think base64 deflates down to basically its original size, which makes it a silly way to send nontrivial amounts of binary data, but not an uncompact one [10:58:34.0000] zewt: Interesting. It makes sense that it should compress to roughly its original size, I guess - you're only using 6 bits of each byte, after all. [11:00:03.0000] i don't recall how deflate's algorithm works, but if it's bitwise rather than bytewise, it'd probably pick up the encoding pretty precisely, too [11:01:20.0000] Pretty sure it's bitwise. [11:01:24.0000] But I could be wrong. [11:01:35.0000] 1048576 bytes of random data becomes 1076467 bytes, so about 3% overhead [11:03:27.0000] doesn't google images do that? [11:03:32.0000] for instant cats [11:07:50.0000] Yeah, I think so. [11:12:13.0000] zcorpan: what about https://www.w3.org/Bugs/Public/show_bug.cgi?id=25542 ? [11:13:03.0000] Hixie: if you remember how the spec comment 4 talks about came into being (if it was there when you edited the spec) [11:13:57.0000] i'd have to look at the blame [11:14:02.0000] i don't recall off the top of my head [11:14:13.0000] ok [11:30:20.0000] i guess we shouldn't tell @PointedEars2 about the quirks spec [11:32:23.0000] Might go all pointed ears about it [11:50:29.0000] zcorpan: Uh, yeah seems to be broken [11:50:43.0000] I have never understood that particular error :( [11:51:06.0000] jgraham: context? [11:51:20.0000] oh the PR [11:51:24.0000] 12:05 < zcorpan> jgraham: critic didn't like https://github.com/w3c/web-platform-tests/pull/959 [11:52:01.0000] Invalid history rewrite: No commit on the rebased branch references [11:52:02.0000] remote: the same tree as the old head of the branch. [11:53:53.0000] Which I guess suggests a move-type rebase is being misinterpreted as a in-place rebase [11:54:02.0000] Which does suggest a way to fix it… [11:56:17.0000] does "move type" mean you put the results in a different ref or something? [11:57:19.0000] SamB: By "Move type" I mean you go from A-B-C-B1-B2 -> A-B-C-D-E-B1'-B2' [11:57:50.0000] jgraham: oh [11:58:18.0000] why does it even try to notice the latter? [11:58:43.0000] The latter? [11:58:47.0000] in-place [11:59:05.0000] which I assume means more like -> A-B-C-B2'-B1' ? [11:59:37.0000] In-place is typically like A-B-C-B1-B2 -> A-B-C-B12 [11:59:46.0000] whatever [11:59:59.0000] It needs to know that the branch history doesn't match what's in git [12:00:50.0000] ah [12:01:25.0000] jgraham: btw is there some unstable test you want me to look at? [12:01:25.0000] Critic has a truly immutable history of the branch because all the previous commits have extra data attached to them like comments [12:02:12.0000] so I guess it'd want to notice if you reorder commits too, so that it can reattach the comments? [12:02:54.0000] zcorpan: I had to disable the tests in http://w3c-test.org/workers/semantics/structured-clone/ because they were playing merry hell with the next test [12:03:19.0000] what, you can't get a fresh instance of the thing-under-test? [12:03:23.0000] Also, they behave differently on my local computer compared to infrastructure [12:03:36.0000] eeinteresting [12:03:39.0000] jgraham: intredasting [12:03:44.0000] SamB: I could, but there is a tradeoff between isolation and performance [12:03:51.0000] jgraham: true [12:04:29.0000] (I could probably add a restart-after feature to the test harness so that tests known to behave badly could still be run, but I haven't, yet) [12:05:31.0000] jgraham: do you think it would help if it was split up to lots of separate files? [12:07:05.0000] zcorpan: Well then at least I could just disable the subset that actually cause problems [12:07:43.0000] yep. might also be easier to figure out what the problem actually is [12:08:09.0000] I *suspect* there is a gecko bug here, but I am *so* close to getting a complete set of green testruns that I haven't wanted to investigate everything [12:08:10.0000] or maybe the problem goes away altogether when splitting, which is both good and bad [12:08:21.0000] Yeah [12:08:52.0000] Well I guess we still have the old version if it does, although people will be less motivated to care [12:09:27.0000] yeah [12:09:47.0000] maybe i should dress it up as an acid test instead :-P [12:40:30.0000] my kittens it's depressing seeing the number of changes the htmlwg make to the spec that are just bogus [12:40:41.0000] dare I ask? [12:40:45.0000] not just thing i disagree with, i mean, things where the change is not even what the htmlwg intends [12:40:56.0000] jtcranmer: i'm going through bug mail looking at old checkins [12:41:28.0000] any choice examples? [12:41:45.0000] https://github.com/w3c/html/commit/d601f6af9914aa0dadd3277c8771ed46995f61de is my favourite so far [12:42:07.0000] replaces "must" with "need", so it's no longer normative [12:42:36.0000] if read literally, it actually gives the wrong effect (e.g. if the contents are "\n\n\n", it suggests that the markup should be "\n\n", whereas it should be "\n\n\n\n") [12:43:00.0000] plus it talks about intent rather than describing the mapping normatively (referring to the actual contents) [12:43:03.0000] etc [12:43:05.0000] it's just a microcosm of error [12:45:37.0000] so... htmlwg is run by a bunch of phenomenal idiots [12:45:55.0000] good to know that I don't have to worry about it [12:46:00.0000] I’m not clear on how the previous MUST is a reasonable author conformance requirement [12:46:07.0000] unless someone tries to convince me that I need to write an HTML parser to process email [12:46:12.0000] it tells you conditionally what to do if you want a certain effect [12:46:49.0000] /me stares at the change [12:46:54.0000] that is a description of implementation behavior, not of author requirements [12:46:56.0000] othermaciej: that section is essentially telling you how to serialise a DOM [12:48:00.0000] Oh, I can’t tell what section - I assumed the “by the author” meant its an authoring requirements section [12:48:28.0000] it's the syntax section, describing how you serialise a DOM [12:48:37.0000] as opposed to the parsing section [12:48:57.0000] so it's for "authors" as opposed to "UAs" but it's still normative :-) [12:49:12.0000] the original statement was mildly problematic [12:50:08.0000] mmm, nope, its in 12.1 Writing HTML documents, nothing about serializing a DOM there [12:50:33.0000] the new statement is slightly unclear as well [12:50:33.0000] afaict this is the section conformance checkers should use to check conformance of any document [12:50:40.0000] Section 12.3 is Serializing HTML Fragments [12:50:41.0000] /me sighs [12:50:57.0000] othermaciej: 12.1 is a description of how you serialise a dom [12:51:02.0000] this is why I like C++'s method of explictly including [Note: ] fragments [12:51:14.0000] othermaciej: "conformance checkers must use the requirements given in the next section ("parsing HTML documents")." [12:51:35.0000] jtcranmer: we have "note" fragments too. in green, even. [12:52:05.0000] so you could say [Note: to make a 

 that starts with an empty line, two linebreaks would be inserted, as the first one is semantically invisible.]

[12:52:16.0000] 
othermaciej: e.g. "The next few characters of a start tag must be the element's tag name"

[12:52:24.0000] 
othermaciej: an element is something from a DOM

[12:52:40.0000] 
Hixie: come on, there’s enough genuine errors that you don’t have to make bad faith arguments

[12:52:48.0000] 
?

[12:52:53.0000] 
this isn't a bad faith argument

[12:53:12.0000] 
just happened to be my favourite of the run i had looked at

[12:54:03.0000] 
12.1 is not about serializing a DOM, its authoring conformance requirements for correct syntax; there is literally no mention of serialization

[12:54:25.0000] 
it is true that a serializer would also be required to output correct syntax

[12:54:32.0000] 
but there’s no reference to a source DOM anywhere in there

[12:54:33.0000] 
the word "serialialisation" isn't used, sure. but that's what it's describing nonetheless.

[12:55:14.0000] 
the whole section is phrased in terms of how you describe a tree of elements

[12:55:22.0000] 
elements only exist in DOMs

[12:56:18.0000] 
if you're writing markup from scratch, do you first imagine the DOM and then serialize that? :-)

[12:56:39.0000] 
its restrictions on valid syntax, not specifically instructions for serializing

[12:56:59.0000] 
annevk: are you aware of any test cases for the Encoding Standard? specifically for the legacy encodings

[12:57:20.0000] 
mathiasbynens: http://dump.testsuite.org/encoding/ has tests

[12:57:41.0000] 
mathiasbynens: needs a lot more though, darobin might have written some more maybe?

[12:58:08.0000] 
(those tests need to be checked for accuracy by the way)

[12:59:03.0000] 
and port to web-platform-tests?

[12:59:06.0000] 
you could argue that any time you make a document from scratch you are implicitly serializing an imaginary DOM, but that would not be the way most folks think about it

[12:59:26.0000] 
speaking of which, I am having trouble figuring out what this means: “A table element must not contain tr elements, even though these elements are technically allowed inside table elements according to the content models described in this specification. (If a tr element is put inside a table in the markup, it will in fact imply a tbody start tag before it.)”

[12:59:33.0000] 
annevk: ta

[12:59:36.0000] 
on whom is that must requirement?

[12:59:43.0000] 
Authors

[12:59:54.0000] 
but authors are allowed to write 

[13:00:01.0000] 
Yeah

[13:00:13.0000] 
But that doesn't lead to a table containing a tr

[13:00:22.0000] 
At least not if "contain" means "child"

[13:00:53.0000] 
is it an obscure way to say your document is invalid if you use DOM methods to insert a  as a direct child of a 
?

[13:01:10.0000] 
othermaciej: you can also use XML

[13:01:17.0000] 
no, that section doesn't apply to DOM methods

[13:01:22.0000] 
that section specifically does not apply to XML

[13:01:34.0000] 
Then what is it talking about?

[13:01:45.0000] 
othermaciej: it's saying that a DOM that is serialised according to that section cannot have a  in a 


[13:02:00.0000] 
othermaciej: it's an additional restriction on the content model

[13:02:08.0000] 
othermaciej: specifically for DOMs that are serialised per this section

[13:02:12.0000] 
you're not allowed to imagine  as a child of 
 when writing your text/html

[13:02:33.0000] 
yeah, the DOM you're serialising is usually just an imagined one, that's a good way to view this

[13:02:35.0000] 
you have to imagine  as child of  as child of 
, and then it's ok to write it as 
 :-)

[13:02:38.0000] 
So any serializer is required to output 
 instead of Mtable>?

[13:03:07.0000] 
no, it means if you pass a DOM to this section, it cannot have a TR as a child of a TABLE

[13:03:17.0000] 
what zcorpan said

[13:03:32.0000] 
you have to imagine  as child of  as child of 
, and then it's ok to write it as 


[13:03:53.0000] 
having a conformance requirement on the input to an algorithm makes no sense (assuming arguendo that this even describes an algorithm)

[13:04:13.0000] 
how would you even check if someone’s imaginary DOM is valid?

[13:04:18.0000] 
content models are nothing but conformance requirements on the inputs to algorithms

[13:04:34.0000] 
well it might not be imaginary

[13:04:53.0000] 
content models are requirements for correct syntax of a textual representation of HTML

[13:05:14.0000] 
othermaciej: it's planned for the next iteration of Google Glasses

[13:05:18.0000] 
a content model requirement that can’t be checked in the serialized output has no effect

[13:07:26.0000] 
so basically, you run the parser FIRST then worry about the content model ...

[13:07:58.0000] 
othermaciej: you run their imaginary DOM through a validator, obviously

[13:09:01.0000] 
if it fails your imagination validator, are you allowed to correct the imaginary error? (and then write exactly what you would have if you hadn’t fixeed the error?)

[13:09:56.0000] 
of course

[13:10:18.0000] 
I guess it’s saying that you are allowed to write 
, but you MUST imagine there is a  there

[13:11:42.0000] 
othermaciej:  and are similar actually

[13:12:19.0000] 
but they happen to have the same content model in xhtml and text/html

[13:12:59.0000] 
so in XHTML, you are allowed to write 
 
 without imagining the ?

[13:13:04.0000] 
yes

[13:13:05.0000] 
wait you mean text/html has its own content models, not just quirky parsing rules?

[13:13:37.0000] 
SamB: yeah there are a few differences between text/html and xhtml content models

[13:13:48.0000] 

 is one

[13:13:54.0000] 
And noscript

[13:13:55.0000] 


[11:17:55.0000] 
Hixie: and the previous one.

[11:18:30.0000] 
Hixie: I mean, the shorter example do not adding any new information to the longer one.

[11:18:43.0000] 
yeah, for the dev spec one you'll need ben. Maybe try asking him on twitter? He hangs out there more than here IIRC.

[11:18:48.0000] 
looking...

[11:19:47.0000] 
hm yeah those do seem a bit redundant

[11:21:00.0000] 
Yes.

[11:26:33.0000] 
jgraham: ta

[11:26:54.0000] 
IZh: Ben said you should email him

[11:27:33.0000] 
By the way, I have been for years on IRC channels where everyone said Hi/Bye, and that was ok. It was impolite if you are not saying high when you entered the room. ;-) That was in 1996-2000.

[11:27:52.0000] 
annevk: Thank you.

[11:27:54.0000] 
IZh: customs change ;-)

[11:27:56.0000] 
IZh: don't worry about the cranky people here :-P

[11:28:17.0000] 
haha

[11:30:41.0000] 
Hixie++

[11:33:03.0000] 
IZh: fixed the examples for you

[11:36:32.0000] 
Hixie: Thanks.

[11:37:40.0000] 
Hixie, http://www.whatwg.org/specs/web-apps/current-work/multipage/browsers.html#dom-document-0 confuses me, want to elaborate? :)

[11:37:59.0000] 
which part, the "newest" part?

[11:38:26.0000] 
Yeah

[11:38:43.0000] 
http://www.whatwg.org/specs/web-apps/current-work/multipage/history.html#initialise-the-document-object

[11:38:46.0000] 
first step

[11:39:44.0000] 
hm, seems i broke my "affected topics" detector

[11:39:49.0000] 
will have to look into that after lunch

[11:40:19.0000] 
Hixie: The paragraph " When used in HTML documents, the allowed content model" near the end of iframe's section seems to me not human readable. :-) I read it 5 times before understand.

[11:40:43.0000] 
please file a bug - i gotta go to a meeting

[11:40:48.0000] 
(you can use the widget in the bottom left to file a bug)

[11:41:02.0000] 
Hixie, "and change the document attribute of the Window object..." sounds like the COMEFROM style bz always complains about :)

[11:41:34.0000] 
you can file a bug too :-P

[11:41:37.0000] 
and now i'm gone :-P

[11:41:38.0000] 
Hixie: it's not a bug. It's just a very long sentence.

[11:44:30.0000] 
Hixie, oh, the "open bug" link is nice

[11:49:46.0000] 
IZh: http://nohello.com/

[11:51:41.0000] 
/me is planning something in Bikeshed to help reduce COMEFROM effects.

[11:52:41.0000] 
I have never heard any complains from people I said Hello to...

[11:52:49.0000] 
IZh: Except me.

[11:53:20.0000] 
Read that link. This is an async medium. It's really very annoying if I'm in a room and have to deal with a ping and then wait for someone to actually ask their question.

[11:53:31.0000] 
IZh I think most people are able to context switch and can take Hello as just a greeting, but maybe not everyone

[11:53:34.0000] 
And it's even worse when I'm not in the room because of differing timezones, and have to deal with multiple content-free pings.

[11:55:02.0000] 
TabAtkins: My greetings are typically personalized. I don't expect answers from someone else when I talk personally.

[11:55:27.0000] 
This doesn't seem to have been true for your pings of Hixie and others in this room over the past weeks.

[11:55:50.0000] 
And it's also frustrating when you repeatedly look for one person to ask a question to, when you have no idea if someone else can answer your question as well or better.

[11:56:03.0000] 
Just email people if you really, *really* want to talk to a single person.

[11:56:56.0000] 
When I pinged Hixie, I know for sure that only he can answer. The same story with Ben.

[11:57:49.0000] 
Except that you're wrong, because you just asked Hixie, and he said he didn't know the answer.

[11:58:05.0000] 
Sorry for annoying.

[11:58:18.0000] 
It would seem to be an extraordinarily rare circumstance for someone to know they have no idea what the answer to a question is, but be certain of exactly who can answer it.

[11:58:49.0000] 
I'm just going to /ignore you if you keep it up, no skin off my back.

[12:04:25.0000] 
TabAtkins: I'll try to be less annoying. But I want to note, that I don't need any help. I want to help and to contribute. So I will loose nothing in any case.

[12:05:03.0000] 
Feel free to help and contribute, that's great.  Just do it by pinging people *with a message*, so you can get your answer in one day rather than multiple attempts over a week or longer. ^_^

[12:05:33.0000] 
Domenic - wow someone made a one page website of that. Ok, I've added it as sub-variant of "Don't presence query". tantek.com/w/CommunicationProtocols#Dontpresencequery

[12:24:16.0000] 
First rule of the WHATWG club: Never say "Hi" to TabAtkins. ;-)

[12:31:55.0000] 
you've just pinged him again, dangnabbit

[12:32:02.0000] 
or her, idk

[12:34:06.0000] 
Tab is a he :)

[12:41:12.0000] 
it's not good to make assumptions, always cover your bases with or clauses

[12:47:35.0000] 
i assumed annevk was a guy (though I did check before committing to a pronoun)

[12:47:36.0000] 
True, you never know what people are on the internet.  Like Ms2ger. :]  But I've met Tab.

[12:47:52.0000] 
(call it statistics)

[12:48:00.0000] 
zewt: I thought he was a girl before he came into my office and was a guy. :P

[12:48:24.0000] 
I either saw a picture or the pronouns on his blog

[12:48:31.0000] 
or possibly on IRC

[12:49:12.0000] 
er, wait, probably not actual pronouns

[12:49:15.0000] 
Not that it changed anything. Gender is a bit unimportant, but we (I) seem to devote brain cells to it anyway. :P

[12:50:32.0000] 
it's unimportant until you hurt peoples feelings and then you feel terrible =(

[12:51:50.0000] 
it matters as far as our common language being one dependent on gender-specific pronouns :P

[12:51:56.0000] 
Yeah.

[12:55:50.0000] 
suppose I should really say sex-specific pronouns, since gender means something different when talking about language

[12:57:39.0000] 
Can somebody summarize the current state of non-Flash-based audio support in modern browsers? As well as the actual standard?

[12:58:04.0000] 
Do we yet have multi-voice (basically unlimited samples played at once) and panning?

[12:58:07.0000] 
And volume.

[12:58:19.0000] 
Those three features are the absolute minimum IMO.

[12:58:51.0000] 
I'd love being able to modify samples as they play, for example to make it feel like you are underwater.

[12:58:52.0000] 
Or in a church.

[12:58:53.0000] 
i don't think you ever get unlimited high quality mix channels

[12:59:03.0000] 
Unlimited or "high enough".

[12:59:19.0000] 
Last time I checked, it was only 1 voice, which is retarded at a 1940s level.

[12:59:41.0000] 
Flash can apparently play "basically unlimited" samples.

[12:59:46.0000] 
have you seen http://hya.io/#/ /

[12:59:54.0000] 
been a while since I looked at it, but I recall it worked pretty well

[13:00:16.0000] 
Huh?

[13:00:19.0000] 
Did you even read what I asked?

[13:00:22.0000] 
I did

[13:00:32.0000] 
I'm not looking for advice on a library. I'm wondering about the state of the standard and the implementations.

[13:00:41.0000] 
I already use a different library, which uses a Flash hack.

[13:00:41.0000] 
er, it's not advice or a library

[13:00:48.0000] 
it's an example application which mixes multiple audio channels

[13:00:53.0000] 
and it sounds pretty good

[13:01:18.0000] 
Hmm...

[13:01:28.0000] 
Well, it's no good if it runs for me but not others.

[13:01:34.0000] 
IE users must have it too.

[13:01:53.0000] 
coolbot95: No, that's not how standards work.

[13:01:59.0000] 
i hear internet explorer is real popular in the musician and sound design community

[13:02:31.0000] 
caitp: I'm not sure if that's supposed to be a joke.

[13:02:37.0000] 
it is

[13:02:38.0000] 
coolbot95: That's how web dev works. You think you can't shut out IE. But that's how it is with standards, they can't be implemented right after they're made.

[13:02:56.0000] 
caitp: What makes you think game players are "musicians" or "sound designers"?

[13:03:18.0000] 
well i'm sure there's some overlap on the venn diagram

[13:04:16.0000] 
anyways the point is that web audio has been used and is being used, and it's not perfect, but it sounds ok

[13:04:30.0000] 
you can do cool stuff with it

[13:08:52.0000] 
If it doesn't have panning, volume and multi-voice, it's useless for games.

[13:10:26.0000] 
with web audio you're basically setting up a mixer. the mixer isn't without limits, but you can do a lot with it

[13:10:49.0000] 
although I think there are some bugs in blink's implementation which could cause problems for long-lived applications right now

[13:21:44.0000] 
I have no idea what you mean.

[13:21:47.0000] 
I just know what I've seen.

[13:21:51.0000] 
Or heard.

[13:22:08.0000] 
okay, lets make another example

[13:22:34.0000] 
so a long time ago, a few companies used to make video game consoles

[13:23:05.0000] 
one of them was Sega, and in the late 80s, they produced one which controlled sound using an 8bit microcontroller hooked up to a very cheap sound chip

[13:23:32.0000] 
this imposed some restrictions on what sort of sounds could be made, it was essentially a cheap DX7

[13:24:31.0000] 
with 6 FM channels, you'd have either one carrier and one modulator, or one carrier and 2 modulators, and you could have 2 or 3 of those combinations pretty much

[13:24:55.0000] 
this imposed a limit on the number of voices you could make, and a limit on the timbre of those voices. and yet, people still did awesome stuff with them

[13:25:18.0000] 
even with a very limited mixer, you can still make marketable, quality stuff

[13:49:10.0000] 
zewt: Nope, gender-specific is the right word for it.  "sex" and "gender" are distinct concepts.

[13:59:53.0000] 
saying "gender" next to "pronoun" implies the linguistic meaning of the word "gender", which it's not

[14:00:03.0000] 
heh, unstable sort detected in chrome

[14:00:39.0000] 
i have the network inspector open, and a few requests are bouncing around randomly, seems like the request timestamp is the same and they're not sorting for display in a consistent way

[14:01:54.0000] 
... isn't it worrisome that they have the same timestamp?

[14:02:08.0000] 
not sure, probably depends on the resolution

[14:02:32.0000] 
it could easily get two URLs to fetch in the same millisecond, if they come in the same chunk of parsing the HTML

[14:02:52.0000] 
oh sure, but why would it have such a low resolution?

[14:03:14.0000] 
no idea how it works, but that's what you'd get if you were using Date

[14:03:22.0000] 
seriously?

[14:03:45.0000] 
Date.now() gives integer milliseconds

[14:03:52.0000] 
Date does only down to a whopping millisecond?

[14:06:53.0000] 
i'd be more surprised at using an unstable sort; every sort I've seen in years has been stable, you'd have to go hunting for one that isn't (or roll it yourself, but who does that)

[14:08:01.0000] 
Quick?

[14:09:12.0000] 
zewt: hmm, sort(1) has a flag for wanting stability

[14:09:24.0000] 
implying that the default might be unstable

[14:10:18.0000] 
python, javascript, c++ sorts are all stable, i don't think i've ever come across a case where having a stable sort isn't better than not

[14:10:51.0000] 
huh, looks like GNU sort is intentionally antistable

[14:10:57.0000] 
when given the choice between "predictable behavior" and "unpredictable behavior"...

[14:10:59.0000] 
unless you ask for it to be stable

[14:11:13.0000] 
or, wait

[14:11:28.0000] 
never really noticed with /usr/bin/sort, since i rarely use it for multi-level sorts

[14:11:35.0000] 
"no ordering options other than --reverse" probably just means "ignore everything but reverse", doesn't it

[14:11:45.0000] 
s/ reverse/ --reverse/

[14:12:52.0000] 
blimey

[14:13:04.0000] 
crikey

[14:13:43.0000] 
/me waves

[14:13:55.0000] 
hi

[14:14:12.0000] 
/me waves bye!

[14:30:16.0000] 
This chair rattles...

[14:30:21.0000] 
I hate rattly shitty chairs.

[14:31:37.0000] 
zewt: Regardless, that's just a matter of a correct term being ambiguous in a particular situation.  "Sex-specific pronouns" is still just incorrect.

[14:32:07.0000] 
Also, I've never seen a gendered pronoun, in the sense of a gendered noun.

[14:32:29.0000] 
"gendered pronoun" always means "referring to a specific gender of person", afaict.

[14:42:31.0000] 
i think it depends who you ask, since people come up with all kinds of different ways to refer to tom neuwirth and his stage personality! but it's getting a bit tumblr in here, time to move onto other endeavors

[14:45:50.0000] 
so... how do i find out if mozilla has implementor interest in a feature?

[14:46:03.0000] 
tantek, Ms2ger, sicking, annevk: ping ^

[14:46:12.0000] 
for example https://www.w3.org/Bugs/Public/show_bug.cgi?id=25663

[14:46:54.0000] 
Ask

[14:47:16.0000] 
Since mounir is gone...

[14:47:42.0000] 
ask whom?

[14:48:14.0000] 
#content

[14:48:24.0000] 
i have like a hundred of these...

[14:48:38.0000] 
Oh dear

[14:48:54.0000] 
https://www.w3.org/Bugs/Public/buglist.cgi?list_id=36950&query_format=advanced&resolution=---&target_milestone=Needs%20Impl%20Interest

[14:49:09.0000] 
one of the other vendors could send an intent to ship on a mailing list or bug?

[14:49:16.0000] 
Ugh

[14:49:19.0000] 
bz :)

[14:49:56.0000] 
i try cc'ing the people that https://wiki.mozilla.org/Modules/Core says i should ping, but i don't often get a response that way

[14:50:33.0000] 
Yeah, people are busy :)

[14:53:03.0000] 
https://www.w3.org/Bugs/Public/show_bug.cgi?id=25699 (appcache related) would be particularly useful to get feedback on asap

[14:53:22.0000] 
I don't know who does appcache

[14:53:25.0000] 
Maybe smaug____ does

[14:54:48.0000] 
Hixie: sadly i don't know that we have anyone really owning form controls any more

[14:55:10.0000] 
Hixie, I'd ping sicking re: anything appcache / offline / storage related (they're quite interconnected) - and there he is

[14:55:18.0000] 
Hixie: jwatt and bz might be good people to ask

[14:55:57.0000] 
Hixie: regarding anything appcache, my answer is basically "we're unlikely to invest in appcache at all at this point"

[14:56:15.0000] 
Hixie, your general meta approach of querying a handful of mozilla folks about who to ping is a reasonably useful late-binding approach to figuring out who to ping about what in particular

[14:56:29.0000] 
we're hoping to deprecate it fairly agressively once SW ships. I don't have an ETA on that though

[14:56:51.0000] 
yeah. this particular bug is about mitigating a potential security issue.

[14:56:57.0000] 
tantek: k

[14:57:34.0000] 
Hixie: yeah. That might be the only exception. I skimmed the bug but it wasn't clear to me that it was serious enough. My gut instinct is that it isn't

[14:58:35.0000] 
Hixie: I could check with out security team. That's probably the right thing to do

[15:04:25.0000] 
Ms2ger: appcache? not me

[15:04:55.0000] 
smaug____, yeah, we found sicking :)

[15:21:31.0000] 
that frustration of finding a tool that seems to work well, until you use it for a couple days to find lots of rough edges and people complaining about them in trackers


2014-05-16
[17:37:43.0000] 
attempts at isolating firefox 29 chewing cpu constantly: fruitless

[17:38:03.0000] 
seems like it's just a small, constant amount of cpu per tab (and I have lots of tabs), since no matter how I close blocks of tabs, cpu usage drops linearly

[17:42:18.0000] 
zewt: does it not have a profiler?

[17:44:57.0000] 
i don't know, i can't find any equivalent of chrome's task manager (which is a lot easier to implement in a process-model browser, of course)

[17:45:39.0000] 
but stress level tangibly increased by having my cpu fan spinning nonstop

[17:49:58.0000] 
zewt: huh, I thought it had one of those now

[17:50:31.0000] 
if it does I haven't tripped over it yet

[17:50:40.0000] 
there's a profiler, but it looks like a regular per-page profiler

[18:00:39.0000] 
it has a profiler

[18:00:53.0000] 
https://developer.mozilla.org/en-US/docs/Performance/Profiling_with_the_Built-in_Profiler

[18:02:15.0000] 
zewt: yeah, you'll want to profile the chrome

[18:02:33.0000] 
you might need to do that special out-of-process thing

[18:03:06.0000] 
"sorry mario, your devtools are in another process!"

[18:03:55.0000] 
boogyman: your cloak isn't too effective if you join channels before it kicks in ...

[18:05:09.0000] 
SamB: the cloak is "there" because i donated.

[18:05:18.0000] 
a lot of networks just mask everyone's hostname

[18:09:19.0000] 
(you have to donate for privacy?)

[18:10:46.0000] 
well, as soon as i turned the profiler on, the problem went away

[18:10:51.0000] 
some things never change

[18:11:24.0000] 
maybe they shouldn't have made it possible to turn it off ;-P

[04:18:59.0000] 
JakeA: did you follow the discussion I had with Domenic about ResponsePromise -> Promise?

[04:19:33.0000] 
JakeA: jungkees: I think he made a reasonable point that we should just have fetch() return a Promise for a Response, no subclassing warranted

[04:20:44.0000] 
/me reads

[04:23:50.0000] 
annevk: can you share some pointer to the discussion?

[04:26:05.0000] 
jungkees: http://krijnhoetmer.nl/irc-logs/whatwg/20140515#l-420

[04:28:09.0000] 
annevk: ResponsePromise isn't doing anything special right now. It has toBlob on it, which is handy but not so handy that I want to get into the promise subclassing mess :D

[04:28:37.0000] 
annevk: If we use a Promise initially, how much breakage would happen if we switched to a subclass later?

[04:29:14.0000] 
annevk: Not just thinking of ResponsePromise, but also the cache methods that return a promise for an array - the possibility of making them something both promise-like and async iterable later

[04:37:31.0000] 
"[charter] Joint work with TAG on their Packaging on the Web spec?; deadline May 21" worst thread in a while

[04:38:16.0000] 
JakeA: returning a subclass later is thought of as being reasonably safe

[04:38:17.0000] 
You sound like comic store guy's more reasonable cousin

[04:38:36.0000] 
JakeA: we're considering it for query/queryAll for instance (return Array for now, Elements later)

[04:39:57.0000] 
/me shakes hands with annevk & Domenic in a weird 3-way handshake

[04:40:12.0000] 
annevk: consider ResponsePromise dropped

[04:41:00.0000] 
cool

[04:43:22.0000] 
annevk: We renamed purpose to context right? Are you still interested in having that on Request objects?

[04:43:30.0000] 
(rather than event objects where it is now)

[04:44:49.0000] 
JakeA: I think grouping it all on Request is fine

[04:45:24.0000] 
JakeA: We could either have a more pure "HTTP-like Request" class and a bunch of parameters, or we could just group them all together as we've done so far

[04:46:20.0000] 
JakeA: the pure style is somewhat attractive I have to say, but basically more involved

[04:46:36.0000] 
annevk: It's called "context" now right?

[04:46:41.0000] 
JakeA: yes

[04:46:54.0000] 
http://fetch.spec.whatwg.org/#concept-request-context

[04:47:43.0000] 
JakeA: we could also call it "API" I suppose

[04:47:57.0000] 
JakeA: but that's not a great property name

[04:54:14.0000] 
annevk: https://github.com/slightlyoff/ServiceWorker/issues/279

[05:01:14.0000] 
hello!

[05:01:43.0000] 
I’m looking for a suitable meta tag name for monitoring purpose but can’t anything, but the pingdom

[05:02:05.0000] 
is there any hash, uid, monitoring meta name I’ve missed?

[05:18:59.0000] 
how is inline SVG (in HTML code) supposed to react when using innerHTML?

[05:19:16.0000] 
Firefox parses it using the HTML parser, Chrome using the SVG parser…

[05:19:24.0000] 
espadrine, are you sure?

[05:19:30.0000] 
pdr: https://thefiletree.com/espadrine/bugs/svg-innerHTML/index.html?plug=none

[05:19:59.0000] 
it breaks because the HTML pass doesn't recognize it as valid

[05:20:05.0000] 
at least, I think

[05:20:33.0000] 
also, things like inserting  yields 

[05:21:41.0000] 
according to the dev tools

[05:21:42.0000] 
espadrine, chrome is parsing that with the html parser in the same way as the inline svg case. I'm not sure what's up with firefox though.

[05:23:04.0000] 
pdr: so, chrome does things right, and firefox is the one with the bug?

[05:23:08.0000] 
I'll file it.

[05:23:34.0000] 
espadrine, try to dig into what firefox is doing first. But I do think chrome is correct in this case

[05:28:50.0000] 
espadrine: firefox has a known bug; puts elements in the wrong namespace, IIRC.

[05:29:50.0000] 
https://bugzilla.mozilla.org/show_bug.cgi?id=886390

[05:29:52.0000] 
davve: good to know; I'm asking on #developers at mozilla

[05:45:00.0000] 
espadrine: Talk to hsivonen

[05:50:37.0000] 
espadrine: i guess what firefox does matches an older revision of the spec

[05:52:57.0000] 
zcorpan: yep. The updated spec is more intuitive.

[06:03:32.0000] 
ok then, anyone here who can help me getting a wiki account?

[06:05:44.0000] 
nedt: I don't own any wikis, but you probably need to state which wiki you want an account on & what permissions you're wanting

[06:05:52.0000] 
(& why)

[06:05:57.0000] 
http://wiki.whatwg.org/

[06:06:07.0000] 
because of the meta tag I asked about an hour ago

[06:06:34.0000] 
nedt: I think annevk can help

[06:06:44.0000] 
Certainly Hixie can later

[06:08:24.0000] 
I would be even more happy if there is an existing meta extension that gives a simple hash for monitoring etc. without being specific to a service. But couldn’t find one

[06:12:28.0000] 
What's the hash used for?

[06:13:57.0000] 
mostly for monitoring. i.e. verifing a site is up and everything needed to output the page (cms, db, …) is working.

[06:14:34.0000] 
there is the pingdom metaextension of the google-site-verification, but using them for an other service feels like misusing

[06:14:35.0000] 
nedt: pm with a username/password

[06:14:48.0000] 
nedt: not password, email address

[06:15:08.0000] 
/me always forgets

[06:20:32.0000] 
jgraham: I think you should be able to create accounts too

[06:21:09.0000] 
Oh really?

[06:31:43.0000] 
jgraham: http://wiki.whatwg.org/index.php?title=Special:UserLogin&type=signup

[06:32:53.0000] 
annevk: Interesting

[06:33:24.0000] 
most longtime users of the wiki can do it

[07:24:13.0000] 
TabAtkins: SimonSapin: querySelector/querySelectorAll can pass unpaired surrogates into the "CSS" parser; problem?

[07:25:41.0000] 
Nope, not a problem. As part of parsing, the input stream gets cleaned up, which'll eliminate unpaired surrogates.

[07:26:12.0000] 
annevk: CSSOM (both per spec and in impls) too. Currently they go through unchanged. I don’t like it, but I don’t know if we can change it

[07:26:50.0000] 
TabAtkins: I don,t belive we hav ethat particular cleanup

[07:27:25.0000] 
(sorry for the mangling, SSH is not very responsive over mobile in a moving train)

[07:27:42.0000] 
Oh, you're right, we only kill nulls.

[07:27:49.0000] 
Hm, I bet we could fix that.

[07:28:04.0000] 
Keep things consistent between input and escapes.

[07:28:59.0000] 
SimonSapin: Why would you think we couldn't change it?  Where do you think unpaired surrogates might be used?  Only possible useful spot would be in Selectors, and I doubt there are many pages with purposefully unpaired surrogates in class/etc.

[07:30:57.0000] 
hmm, come to think of it, if you change that, run it by bz

[07:31:13.0000] 
I could see people assigning random strings to id/class and expect that to work

[07:34:11.0000] 
TabAtkins: not that it’s useful, but who know what crazy things people are doing in JS

[07:34:22.0000] 
it’s just selectors, it’s all of CSSOM

[07:34:32.0000] 
it’s not just selectors, it’s all of CSSOM

[07:35:14.0000] 
annevk: If we change it at the preprocessing level, it'll still work (you'll be setting a different string, but also querying the same, different, string).

[07:35:25.0000] 
but maybe we can get away with it. I’d be happy to

[07:35:32.0000] 
SimonSapin: None of CSSOM is happy with unpaired surrogates except selectors and a few custom idents.

[07:36:14.0000] 
Which means that either you're explicitly putting unpaired surrogates in your markup (crazy?!?) or you're setting them in JS (and they'll get cleaned up in the same way, so they continue to match).

[07:36:39.0000] 
TabAtkins: could you bring it up at the f2f? I’ll try to join by phone in your afternoons

[07:36:41.0000] 
annevk: Oh wait, nm, they wouldn't get cleaned up if you set class/id.

[07:37:00.0000] 
SimonSapin: Add it to the wiki?

[07:37:37.0000] 
TabAtkins: you can't do it through markup, has to be JS

[07:37:55.0000] 
TabAtkins: no text encoding allows for unpaired surrogates

[07:37:57.0000] 
annevk: Ah, good.

[07:38:26.0000] 
yeah, anything from Encoding is clean

[07:38:27.0000] 
Then maybe we can make a coordinated change between DOM and CSS, get unpaired surrogates cleaned up everywhere at once?

[07:38:32.0000] 
it’s only JS strings

[07:41:02.0000] 
anything you can write/append to in chuncks, there is a risk that each half of a surrogate pair be in a different chunck, but is still ends up paired correctly

[07:41:24.0000] 
TabAtkins: for DOM in general that's a no :(

[07:42:14.0000] 
annevk: What do you think about just cleaning up id/class and maybe attrs?

[07:42:48.0000] 
TabAtkins: I wonder if anyone is willing to take a perf hit on that and whether it's actually worth it

[07:43:17.0000] 
TabAtkins: kinda depends too on what you see as the bottom layer, if it's JS, getting rid of the unpaired surrogates is kinda pointless

[09:39:24.0000] 
annevk (or anyone): what decides whether an API wants ScalarValueString? It seems to be something to do with URLs?

[09:42:29.0000] 
never heard of it

[09:43:36.0000] 
Sounds like something annevk made up

[09:43:56.0000] 
Since he's the only person that uses "Scalar Value" to mean something related to unicode

[09:44:55.0000] 
i use "scalar value" in the parts of the spec that interface with Unicode...

[09:49:15.0000] 
Hixie: No, you use "Unicode scalar value" and define a synonym

[09:49:23.0000] 
that's what i mean

[09:49:42.0000] 
Your only use of "scalar value" is in the normal meaning

[09:49:42.0000] 
i basically do "import scalar value as..." :-)

[09:49:53.0000] 
oh anne uses it for a non-unicode meaning?

[09:50:20.0000] 
No, annevk uses "scalar value" for "unicode scalar value"

[09:51:12.0000] 
so his use is the normal meaning too

[09:51:18.0000] 
by "normal meaning" I mean "unrelated to unicode"

[09:51:55.0000] 
you just quoted the part of the spec where i use it in the unicode sense :-)

[09:51:57.0000] 
The only use of "scalar value" in HTML is "The meter element also does not represent a scalar value of arbitrary range"

[09:51:57.0000] 
i'm confused

[09:52:12.0000] 
plus the two uses i just mentioned, where i'm interfacing with the unicode spec

[09:52:23.0000] 
No, that's *unicode* scalar value

[09:52:57.0000] 
"unicode scalar value" contains "scalar value"...

[09:53:04.0000] 
this conversation is making my head hurt

[09:55:13.0000] 
I don't see why it's confusing

[09:55:29.0000] 
well part of it is i've not yet woken up :-)

[09:55:39.0000] 
There are two distinct terms; "scalar value" and "unicode scalar value"

[09:56:07.0000] 
to me, "unicode scalar value" contains the term "scalar value", and both contain the term "scalar".

[10:06:34.0000] 
good morning, Whatwg!

[10:17:59.0000] 
jgraham: those might seem distinct at the markup level, but in English it is just as Hixie says

[10:19:47.0000] 
It really isn't

[10:19:56.0000] 
though reading the scrollback it does seem like if you mean a special unicode concept, you should either use the word Unicode or state up front that you're going to be omitting it from the term for brevity

[10:20:07.0000] 
They are different compound nouns

[10:39:06.0000] 
ScalarValueString is the new name for [EnsureUTF16] DOMString apparently

[10:56:47.0000] 
Hixie: Hi. There is a bug in acknowledgements section.

[10:57:18.0000] 
... Alexey Feldgendler, Алексей Проскуряков (Alexey Proskuryakov), Alexis Deveria,...

[10:58:22.0000] 
IZh: ?

[10:59:07.0000] 
ap: Why there are entities instead of characters?

[11:00:05.0000] 
IZh: oh. that I don't know or care about - it looks correct in a browser :)

[11:00:34.0000] 
http://www.whatwg.org/specs/web-apps/current-work/multipage/acknowledgments.html#acknowledgments

[11:01:25.0000] 
ap: Hmm... Perhaps mobile browser is buggy... Wait.

[11:02:02.0000] 
No. The same bug.

[11:03:23.0000] 
The problem is bacause the ampersand was escaped as &. Double escaping problem.

[11:04:36.0000] 
IZh: I see now. It's correct on http://www.whatwg.org/specs/web-apps/current-work/, but yes, double escaping in acknowledgments.html

[11:05:02.0000] 
Yes.

[11:05:10.0000] 
much appreciated (hopefully Hixie will see and fix this) :)

[11:07:01.0000] 
I have filed a bug.

[12:43:20.0000] 
IZh, ap: yeah, known bug with the multipage splitter. I'm slowly working on revamping my pipeline which will fix this.

[12:48:46.0000] 
 Hixie: the page splitter double escapes?

[12:49:07.0000] 
the page splitter has some parsing/serialising bugs

[12:49:38.0000] 
right now the pipeline isn't using a modern HTML parser throughout, which has all kinds of weird bugs

[13:03:23.0000] 
if i have a suggestion for a test for the html test suite, where do i put it?

[13:03:25.0000] 
http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3026

[13:03:43.0000] 
Hixie: A suggestion for a test rather than a test?

[13:03:53.0000] 
well i haven't made it use the complicated harness stuff

[13:04:01.0000] 
Hixie: https://github.com/w3c/web-platform-tests/issues

[13:04:07.0000] 
thanks

[13:04:19.0000] 
Hixie: feel free to assign to me

[13:05:06.0000] 
Well the way it's written it would have to be a reftest, but I guess you could fix that

[13:05:20.0000] 
no idea how to assign the bug

[13:05:22.0000] 
https://github.com/w3c/web-platform-tests/issues/993

[13:06:10.0000] 
in other news, has there been any movement on providing a "directory" API to go with the File API?

[13:06:13.0000] 
You might not be allowed to assign the bug because github is strange

[13:06:19.0000] 
Anyway, I did it

[13:06:52.0000] 
github--

[13:29:33.0000] 
Hixie: the "new" filesystem API seems to be directory based

[13:30:15.0000] 
I am not sure how I feel about that; most native FS APIs I have used are just plain filename based, with e.g. a directory listing API that accepts a path, but no first-class realization of directories

[13:35:46.0000] 
Domenic: how much buy-in does it have?

[13:50:42.0000] 
Hixie: hard to say, but from what I've seen it has tentative buy-in from Mozilla and Google? Not quite finished yet though so I imagine they're taking wait-and-see positions...

[13:52:12.0000] 
Domenic: wait who's speccing it if not a browser vendor?

[13:52:59.0000] 
Hixie: looks like Mozilla. I probably shouldn't be trying to summarize something I've only half paid attention to...

[13:53:11.0000] 
heh

[14:24:14.0000] 
so someone once complained of the spec, "can't find html5 counterpart of many html 4.01 specifications"

[14:24:17.0000] 
i wonder what they meant

[14:26:48.0000] 
it may be *harder* to find

[14:27:33.0000] 
pretty meaningless without specifying "many"

[14:33:02.0000] 
some other feedback: "information is there but in planty so you end up reading all night when you need just few lines"

[14:33:06.0000] 
sounds about right

[14:36:28.0000] 
also i love the people who ask for a way to jump back to the top dearly

[14:36:30.0000] 
but darlings

[14:36:33.0000] 
learn about the "home" key

[14:36:34.0000] 
please

[14:37:41.0000] 
MikeSmith: you any idea how we could adjust the spec to address things like "should be easier to check questions such as can a p tag contain a div" ?

[14:41:33.0000] 
Interesting, yeah, I've found myself asking that question a lot

[14:41:44.0000] 
in developers edition you can follow links pretty easily

[14:41:48.0000] 
for content models

[14:42:00.0000] 
nearly all the restrictions about what tags can contain what other tags or not were some of the dumbest things to put into HTML

[14:42:06.0000] 
did nothing but confuse people

[14:42:28.0000] 
so the more we can get rid of those, the better

[14:42:43.0000] 
p is a great example of wrongness

[14:44:33.0000] 
i dunno about that

[14:44:49.0000] 
the restrictions aren't really restrictions, they're just cases that make no sense and are likely an indication of an error

[14:45:22.0000] 
e.g. putting an ordered list in an 
 is very unlikely to be what the author meant, and so it's helpful if we can flag that

[14:45:45.0000] 
Domenic: is it not easy to follow the links on the big version of the spec?

[14:58:11.0000] 
Domenic: where it needs to be put into something that does not expect unpaired surrogates

[14:58:25.0000] 
Domenic: URL and Encoding are such layers

[15:19:47.0000] 
/me pokes at the style sheet some more

[15:19:55.0000] 
tried to add more White Space.

[15:34:30.0000] 
Is there any standard regarding getting the amount of memory JS consumes in a web page? I'm using memory.performance in Chrome and I was wondering if there is anything I should look at in terms of standards

[15:51:30.0000] 
benjamingr: what kind of standards?

[15:51:38.0000] 
benjamingr: (the answer is probably "no")

[15:51:55.0000] 
benjamingr: like, requirements on how much a browser should use for a given page?

[15:57:40.0000] 
Hixie:  as in, be able to get the amount of available memory for the page in the browser, or a vague indication of it

[15:58:01.0000] 
even the browser probably doesn't know that

[15:58:03.0000] 
oh, i see

[15:58:30.0000] 
maybe in the web perf wg's set of specs?

[15:58:35.0000] 
In chrome, I can get it via the performance.memory object

[15:58:36.0000] 
i don't know of anything offhand

[16:12:48.0000] 
whatwg spec editors: if your spec is too big for browsers to do the transitions i just added without making scrolling painful, add class=big to your  element.


2014-05-17
[19:50:26.0000] 
Hixie: about "should be easier to check questions such as can a p tag contain a div" no ideas jump to mind

[19:51:50.0000] 
but, well, the content model is a hyperlink that says "phrasing content" and if you click on that link you get http://www.whatwg.org/specs/web-apps/current-work/multipage/elements.html#phrasing-content

[19:52:25.0000] 
which is a complete list of elements that are allowed as phrasing content

[19:52:42.0000] 
so not sure how much easier you can make it than you already have

[19:53:52.0000] 
yeah i dunno either

[19:54:36.0000] 
Hixie: I guess instead of that hyperlink, the complete list of those elements could be repeated at each place where you have "content model: hyperlink" now. But I don't think most people would think that'd be in improvement.

[19:56:22.0000] 
or maybe the problem is that whoever reported that instead wants to have the list of elements that are _not_ allowed. Maybe they don't understand that by not being listed, that means they're not allowed.

[19:56:38.0000] 
But if so I don't think most people have that confusion problem.

[19:57:37.0000] 
so anyway "p can't contain div" seems pretty clear from what you have now. About as clear as you can practically make it.

[19:57:51.0000] 
yeah maybe

[19:58:09.0000] 
one thing i'm thinking of doing is having popups appear when you hover over a link, essentially telling you what the link points to

[19:58:14.0000] 
no idea how i'm gonna do that exactly

[19:58:17.0000] 
but that could solve this

[19:58:46.0000] 
ah yeah, that would be helpful

[19:59:51.0000] 
for the content model links, popups with the list of elements for "phrasing content" and "flow content" would be good

[20:00:42.0000] 
and also for whatever the other classes of content are that are referenced in the content model sections

[20:33:36.0000] 
/me wonders what transitions Hixie is talking about

[20:43:27.0000] 
load the multipage version of the spec

[20:43:30.0000] 
or the dom spec

[20:48:33.0000] 
MikeSmith: see them?

[21:48:08.0000] 
Hixie: stepped out for a bit, looking now

[21:49:33.0000] 
Hixie: page transitions?

[21:49:43.0000] 
gradients at the end?

[21:56:34.0000] 
Anyone in here have experience with createShadowRoot?

[23:16:33.0000] 
MikeSmith: move your cursor off the page for a bit

[23:17:30.0000] 
Hixie: ok, trying now

[23:18:20.0000] 
Hixie: ah the link thing?

[23:18:32.0000] 
not just links, but yeah

[23:18:32.0000] 
I thought that was a browser bug :)

[23:18:37.0000] 
yeah

[23:18:46.0000] 
but now that I know what it is I like it

[23:18:51.0000] 
some people seem to want less colour

[23:18:52.0000] 
so...

[23:19:03.0000] 
ah yeah it's very slow transition

[23:19:07.0000] 
personally i use the colour to figure out what's going on

[23:19:12.0000] 
Hixie: bingo

[23:19:41.0000] 
These are the 4 “element query” techniques I’ve heard of. Am I missing others? https://gist.github.com/jonathantneal/d4e27ac9a65a426e6870

[23:21:08.0000] 
Hixie: yeah the color is there for genuine functional purposes. I think the people who don't like are people who don't actually use the spec, or don't need to. And for them it just hurts their purist aesthetic sensibilities about what a good document should look like in general, I guess

[23:21:19.0000] 
yeah

[23:21:28.0000] 
i also added a lot of whitespace, which i think works ok

[23:22:16.0000] 
Hixie: yeah I think we can all agree that more whitespace is better

[23:22:23.0000] 
within reason

[23:22:55.0000] 
sure yeah

[23:23:05.0000] 
it did add about 50 pages to the pdf version...

[23:25:56.0000] 
Hixie: well anybody who's nuts enough to try to print out the whole PDF doesn't have anybody to blame but themselves

[23:26:05.0000] 
hah

[23:32:55.0000] 
hmm wchen adding Gecko support for