[11:31:57.0000] <annevk> Hixie: can't HTML just do that? [11:32:22.0000] <annevk> Hixie: or does the SVG WG still pretend it can have SVG without HTML? (because they cannot, they don't have an event loop and things) [11:32:23.0000] <Hixie> annevk: i could, but it seems weird for HTML to be defining some SVG-specific stuff, so i was thinking HTML and SVG would hook into something DOM specifies [11:32:49.0000] <Hixie> maybe i should just do it and not worry about it though [11:32:58.0000] <annevk> Hixie: I'm happy to blur the lines further, either way [11:33:06.0000] <SamB> /me has always found html <title> rather magical [11:34:57.0000] <SamB> (Well, ever since I discovered that you can't get the same results with custom XML and CSS) [11:37:52.0000] <Hixie> well all the useful elements in HTML are "magical" by that definition [11:39:22.0000] <SamB> in the sense that they have semantics as well as presentation, sure [11:40:25.0000] <Hixie> in the sense that the user agent is required to do stuff with them more than you can do with custom XML and CSS [11:40:43.0000] <Hixie> e.g. <form>, <a>, <input>, <video>, ... [11:41:01.0000] <SamB> you are clearly using a different sense of the term "useful" than I am [11:41:24.0000] <Hixie> why? [11:42:06.0000] <SamB> well, mine includes stuff like <h2> [11:43:35.0000] <Hixie> s/useful/interesting/? [11:48:04.0000] <SamB> anyway, it might have been 15 years ago that I got the idea "<title> is magical" into my head ... [11:49:28.0000] <Hixie> well, i'm not a fan of the word "magical", but <title> has non-mundane behaviour, certainly [11:49:45.0000] <Hixie> like <details>, <output>, <progres>, <canvas>, etc [12:05:41.0000] <manu-> TabAtkins: I totally love what you've done w/ the non-element selectors spec: http://dev.w3.org/csswg/selectors-nonelement/ [12:06:18.0000] <manu-> (W3C Stylesheets hacked) https://twitter.com/jirkakosek/status/451067752703197184/photo/1 [12:06:28.0000] <Ms2ger> manu-, *all dev.w3.org/csswg/ specs [12:07:23.0000] <manu-> It really is a step in the right direction... the dot-com bubble just messed everything up, glad to see a move back to fundamentals. [12:08:11.0000] <Ms2ger> It's kinda unimaginative... Same thing from last year [12:08:44.0000] <manu-> oooh, right... April fools - forgot about that... had turned off all of my news feeds waiting for it to blow over. [12:23:17.0000] <TabAtkins> Ms2ger: Yeah, I'd forgotten to prepare anything new. I'll actually do a new joke next year, I promise. ^_^ [12:51:17.0000] <annevk> TabAtkins: but will it resolve? [12:58:29.0000] <TabAtkins> annevk: It'll take a year, but yeah. [12:58:32.0000] <dglazkov> Hixie: with MessagePort, how does one side realize that the other side is no longer listening (due to untimely demise, for instance) [12:58:36.0000] <dglazkov> ? [12:58:59.0000] <annevk> dglazkov: you don't, you need to implement ping [12:59:15.0000] <annevk> dglazkov: reason is GC [12:59:32.0000] <dglazkov> okay [12:59:44.0000] <annevk> Which reminds me, at some point we need to figure out the SW story around garbage collection [13:00:24.0000] <SamB> SW? [13:00:26.0000] <dglazkov> why is GC being detectable a problem? [13:00:33.0000] <annevk> dglazkov: there have been some proposals on the position to not expose GC / make memory leaks likelier [13:00:56.0000] <SamB> /me does not like memory leaks [13:00:57.0000] <annevk> dglazkov: exposing the details of GC is bad as it might lead to having to define GC [13:01:06.0000] <annevk> dglazkov: ask TC39 [13:01:14.0000] <dglazkov> annevk: that doesn't seem bad, just hard [13:01:33.0000] <annevk> dglazkov: why not, we have changed GC strategies many many times over the years due to it not being exposed [13:01:40.0000] <annevk> dglazkov: if it's exposed we cannot change strategies [13:01:48.0000] <dglazkov> that's not necessarily true, right? [13:02:53.0000] <annevk> I'm not sure I can speak much to all of the constraints here. I do know it's been one of the things we have tried very hard to avoid ever doing. [13:02:56.0000] <SamB> hmm, that sounds kind of like a silly reason to have no way to detect a broken socket [13:03:29.0000] <SamB> just make sure the GC is already sufficiently hard to predict [13:03:54.0000] <annevk> "just" [13:04:04.0000] <SamB> what, you mean it isn't? [13:04:42.0000] <SamB> it's not like there's refcounting involved like in CPython ... [13:04:57.0000] <dglazkov> annevk: totally understand. It seems though the times are changing. With where asm.js is going, exposing GC in one way or another looks inevitable. [13:06:05.0000] <annevk> dglazkov: I don't know, would have to ask dherman, but exposing to asm.js does not necessarily mean we want to expose it higher-level too [13:06:33.0000] <dglazkov> yup [13:10:37.0000] <SamB> annevk: I guess you must be worried about exposing what happens when a whole compartment goes away? [13:11:07.0000] <annevk> SamB: exposing GC in general [13:12:38.0000] <SamB> what else is there that's presently deterministic enough that stuff might grow to depend on it? [13:13:31.0000] <annevk> Anything? Content has a tendency to rely on anything we expose, whether intentional or not [13:16:07.0000] <SamB> I mean, what other situations can cause objects to be collected more-or-less immediately, rather than whenever the GC finally kicks in? [13:18:02.0000] <annevk> Depends on your GC [13:24:08.0000] <smaug____> dglazkov: what has asm.js to do with gc? [13:25:25.0000] <smaug____> asm.js makes such js that gc doesn't have to run at all [13:28:43.0000] <smaug____> SamB: you might implement the whole system using a cycle collector and traditional refcounting [13:28:58.0000] <smaug____> SamB: then objects would usually die when the last ref goes away [13:29:18.0000] <smaug____> and in case of cycles, when the collector runs [13:29:32.0000] <SamB> smaug____: who would be stupid enough to actually do it that way [13:29:57.0000] <smaug____> what is wrong with that? [13:30:08.0000] <SamB> refcounting is actually pretty slow [13:30:25.0000] <smaug____> there are algorithms to optimize out most of it [13:30:37.0000] <smaug____> and GC is hard if you have lots of objects [13:30:52.0000] <smaug____> since GC deals with live objects and CC only possible garbage objects [13:31:09.0000] <SamB> hmm [13:32:03.0000] <jsbell> annevk FWIW: dbh = indexedDB.open('db', 1); indexedDB.open('db', 2).onsuccess = function(){alert('GC-or-equivalent must have happened');}; dbh = null; [13:32:24.0000] <smaug____> best performing systems are, at least per certain literature and research, combination of gc and cc [13:32:56.0000] <SamB> I guess refcounting *does* at least only do the extra work on objects that are probably of interest [13:32:57.0000] <annevk> jsbell: you don't define that in terms of event loops and lifetimes? [13:34:38.0000] <jsbell> annevk: The spec says "If the connection is GCed .... the connection is closed", which unblocks the second connection. [13:35:18.0000] <annevk> Oh my [13:35:34.0000] <smaug____> whaaat? [13:35:38.0000] <smaug____> /me files a spec bug [13:36:14.0000] <jsbell> Yeah. I was poking around for precedent for "what if a transaction could waitUntil() a Promise, but the Promise was never resolved" [14:01:06.0000] <annevk> https://www.youtube.com/watch?v=_eT7sl7bR1A oh man [14:01:21.0000] <annevk> I wonder if there's a 2048 homescreen [14:08:26.0000] <jgraham> annevk: Higher score == better apps you can launch? [14:08:45.0000] <jgraham> And 2048 gives you the browser [14:08:50.0000] <annevk> jgraham: yeah, maybe one screen has the game, and another has the apps available thus far [14:09:09.0000] <jgraham> No, the game tiles are the launchers [14:11:09.0000] <annevk> oh man [14:11:30.0000] <annevk> so at some point you can no longer access 1024 until you hit it again? [14:12:49.0000] <Ms2ger> Hmm? [14:13:08.0000] <annevk> the app corresponding to 1024 if that was unclear [14:13:23.0000] <annevk> I still haven't hit 8192 btw [14:13:30.0000] <annevk> got 4096 several times [14:13:40.0000] <annevk> I get sloppy after a while :/ [14:23:04.0000] <hober> threes ftw :) [14:27:52.0000] <zewt> refcounting is comparatively slow, but in most layers it doesn't matter today [14:28:11.0000] <zewt> in systems where you can get away with not breaking cycles (eg. not the web), it also has the huge advantage of being deterministic [14:28:46.0000] <Hixie> slow relative to what? managed memory and walking the entire heap every few seconds? [14:29:01.0000] <Hixie> or slow relative to a defined ownership model? [14:29:26.0000] <zewt> relative to mark and sweep, since you don't have to adjust refcounts all the time (though I guess that also depends on how smart you can be about optimizing those out) [14:29:47.0000] <Hixie> really? do you have data on that? [14:29:53.0000] <zewt> no, just my recollection [14:30:04.0000] <Hixie> refcounting is like 2 extra instructions per assignment [14:30:10.0000] <zewt> but if true, I still think it's irrelevant today, except maybe for really performance-sensitive stuff [14:30:11.0000] <Hixie> compared to walking the entire heap, that seems cheap. [14:30:15.0000] <jgraham> hober: There you are again promoting iOS rather than the open web ;) [14:30:50.0000] <SamB> Hixie: that doesn't happen every assignment though ... [14:30:58.0000] <Hixie> true, but still [14:31:08.0000] <zewt> it's an extra two writes, and in some (also non-web) cases an atomic write (don't know how expensive those are today) [14:31:20.0000] <Hixie> i have no data one way or the other, i just find it surprising [14:31:46.0000] <Hixie> i would have thought refcounting would win hands down in terms of raw cpu cycles spent on memory management over the lifetime of a task [14:31:49.0000] <zewt> i'm sure it depends massively on the access patterns and language, so I'm overgeneralizing for sure [14:32:22.0000] <Hixie> i mean, wouldn't a GC model require at least as many instructions per assignment as well? [14:32:34.0000] <zewt> not following [14:32:52.0000] <Hixie> i guess you could just know where the pointers are [14:33:01.0000] <Hixie> if it's a strongly typed system [14:33:27.0000] <Hixie> and walk it that way [14:33:51.0000] <zewt> if you have a memory model where you can walk around and look at pointers (like you need for that), "x = y" is simply what it says, set the storage for x to a pointer to y [14:34:07.0000] <Hixie> /me usually just uses a defined ownership model, which seems faster than either :-) [14:34:08.0000] <zewt> for refcounting, you have to say "if x was set to something, decrease its refcount" and then "increase y's refcount" after the assignment [14:34:18.0000] <zewt> which might also mean a branch [14:34:52.0000] <zewt> definitely faster, it just tends to lose on other fronts :P [14:35:34.0000] <Hixie> yeah, it's definitely not a perfect solution :-) [14:35:45.0000] <Hixie> i find it makes me think more about the design though [14:35:48.0000] <Hixie> which is a win [14:36:12.0000] <zewt> i grew up on C++, i guess I should be a little sad that I almost never use it these days [14:39:10.0000] <Hixie> i spent a decade or more writing lots of perl [14:40:00.0000] <zewt> i avoided that trap [14:40:09.0000] <Hixie> but recently i've come to realise that for anything serious, I really prefer a strongly typed non-GCed true OO natively compiled modern language [14:40:24.0000] <zewt> is there such a thing? [14:40:27.0000] <Hixie> http://ian.hixie.ch/programming/ [14:40:39.0000] <jgraham> (I think this is where someone says that there is no strongly typed OO language) [14:40:47.0000] <jgraham> (which indeed zewt did) [14:40:55.0000] <zewt> i've spent enough time at a high level to find things like closures hard to live without, and closures essentially require GC [14:41:07.0000] <zewt> jgraham: what I meant is "modern" [14:41:16.0000] <jgraham> Rust has closures [14:41:19.0000] <Hixie> C++ does closures without GC [14:41:33.0000] <Hixie> as do some variants of ObjectPascal (sadly not the variant i'm using, yet.) [14:41:46.0000] <zewt> i haven't touched any of the "C++digitdigit" things (since I don't think they'll go anywhere, not to the level of availability of real C++) [14:41:54.0000] <jgraham> (it doesn't do OO but unless you are implementing the DOM it doesn't seem to matter. And for implementing the DOM we will probably get single inheritance) [14:41:55.0000] <Ms2ger> Hixie, Rust might be interesting to you [14:42:20.0000] <gsnedders> Rust is an odd type of OO, though [14:42:49.0000] <Hixie> Ms2ger: i find the rust language to be a bit weird to read. [14:42:50.0000] <jgraham> Rust currently has typeclasses rather than inheritance [14:43:09.0000] <Hixie> Ms2ger: (too much like perl in its liberal use of unusual punctuation) [14:43:09.0000] <zewt> i have trouble visualizing closures without GC to handle the locals attached to the closure [14:43:16.0000] <gsnedders> zewt: I dunno, I thought this a few years ago, yet now almost all embedded platforms with a C++ compiler support C++11. [14:43:24.0000] <jgraham> Most of the punctuation is gone [14:43:28.0000] <Ms2ger> Hixie, that's been improved a little... @ is gone, at least [14:43:29.0000] <gsnedders> zewt: When a decade ago many didn't support stuff as basic as namespaces. [14:43:32.0000] <zewt> gsnedders: msvc? [14:43:33.0000] <jgraham> Well @ at least [14:43:44.0000] <gsnedders> jgraham: Is it really typecalsses? They didn't seem to quite be typeclasses last time I looked. [14:44:02.0000] <Hixie> Ms2ger: still has a long way to go. consider that my favourite language has so little punctuation it doesn't even use { } for blocks :-P [14:44:17.0000] <zewt> pascal? :P [14:44:23.0000] <zewt> (don't say yes) [14:44:26.0000] <gsnedders> zewt: They support most of it now [14:44:27.0000] <Ms2ger> Yes [14:44:27.0000] <Hixie> (old python was pretty good in this regard, though modern python has taken a bit of a turn for the punctuationy) [14:44:34.0000] <Hixie> zewt: modern variants thereof, but yeah [14:44:55.0000] <zewt> my only major, fundamental issue with python is the indentation block gimmick [14:45:12.0000] <jgraham> I have used perl (and lived to tell the tale) and Rust, and Rust doesn't use punctuation like Perl [14:45:16.0000] <Hixie> zewt: that and it's just not a fast language [14:45:27.0000] <Hixie> jgraham: yeah, it's not at the level of perl. [14:45:36.0000] <zewt> which means it structurally has no way to do inline functions, and it can't do the inverted-template thing of php and ruby [14:45:36.0000] <Ms2ger> Hixie, it does tend to follow C conventions for that, to avoid scaring the C programmers too much :) [14:45:49.0000] <gsnedders> jgraham: At least most things claim traits aren't quite type classes. [14:45:53.0000] <jgraham> Like the usage in Rust generally follows straightforward patterns [14:45:55.0000] <zewt> Hixie: it's extremely fast to develop in; fast to execute I don't know (except that it's never been a problem for me) [14:46:09.0000] <jgraham> gsnedders: Well I don't exactly know [14:46:23.0000] <Hixie> the problem with rust is that i look at some random rust and i can't tell what it's doing off the top of my head [14:46:27.0000] <Hixie> in that respect it's like perl [14:46:56.0000] <Hixie> e.g. code like: fn ascending<M: MutableMap<uint, uint>>(map: &mut M, n_keys: uint) { ... } [14:47:03.0000] <zewt> :| [14:47:08.0000] <Ms2ger> That's not the most readable of rust :) [14:47:09.0000] <gsnedders> /me finds that perfectly readable… [14:47:20.0000] <Hixie> the "map: &mut M" part in particular [14:47:29.0000] <Ms2ger> Argument map of type &mut M [14:47:36.0000] <gsnedders> (And I've not touched that much Rust) [14:47:50.0000] <Ms2ger> For M a type parameter that fulfills MutableMap<uint, uint> [14:47:50.0000] <Hixie> what is "&mut M" ? [14:47:57.0000] <Ms2ger> Mutable pointer to M [14:48:04.0000] <Ms2ger> (non-owning) [14:48:09.0000] <jgraham> A mutable borrowed pointer to M [14:48:10.0000] <Hixie> mutable pointer? [14:48:11.0000] <Hixie> as opposed to? [14:48:16.0000] <Ms2ger> Immutable [14:48:17.0000] <jgraham> An immutable pointer [14:48:27.0000] <gsnedders> Ms2ger: Oh, then I misunderstood what it meant, so maybe not so readable :) [14:48:31.0000] <Hixie> is an immutable pointer just a constant reference? [14:48:37.0000] <Ms2ger> Stronger [14:48:41.0000] <zewt> "immutable pointer" sounds like "reference" (c++) [14:48:48.0000] <jgraham> You can only have one mutable pointer to an object in scope at a time so that the compiler can reason about lifetimes [14:48:57.0000] <Ms2ger> It also guarantees that there are no mutable pointers to that value [14:49:03.0000] <jgraham> (that might be wrong) [14:49:20.0000] <Ms2ger> You can only have one mutable pointer to an object in scope at a time so that you don't have data races [14:49:36.0000] <jgraham> Yeah, that makes more sense [14:49:37.0000] <SimonSapin> Hixie: yes. Rust makes stuff constant by default, you need the 'mut' keyword to change that. (Opposite of C/C++ and the 'const' keyword.) [14:49:46.0000] <zewt> mutable pointer as in const foo *x or mutable pointer as in foo *const x? [14:50:13.0000] <Ms2ger> You can mutate what's behind it [14:50:13.0000] <jgraham> zewt: As in the thing it points to is mutable using that reference [14:50:18.0000] <Hixie> so "mut" is more like the opposite of C++ const than it is like pascal argument "var" or the opposite of pascal argument "const"? [14:50:44.0000] <jgraham> Rust doesn't have pointer arithmetic [14:50:45.0000] <Hixie> /me always thought the C++ "const" thing was a bit weird [14:50:49.0000] <Ms2ger> I haven't done pascal in 6 years or so, no idea :) [14:50:52.0000] <gsnedders> zewt: foo *const x if I remember which way around they are. :P [14:51:01.0000] <zewt> yeah I have to think about it too heh [14:51:02.0000] <gsnedders> zewt: i.e., &x is constant [14:51:21.0000] <Hixie> another line of random rust, no idea what this does at all: vec!(~"", ~"20") [14:51:29.0000] <zewt> looks like a smiley [14:51:40.0000] <jgraham> The ! means that vec is a macro [14:51:41.0000] <zewt> if a peculiar one [14:51:56.0000] <jgraham> ~"" is an owned pointer to the empty string [14:52:05.0000] <jgraham> No idea what the vec macro does [14:52:54.0000] <gsnedders> zewt: essentially all you have is foo * const and const foo * const in Rust [14:53:38.0000] <Ms2ger> vec!() is like [] [14:53:47.0000] <Hixie> anyway. i haven't studied rust much yet. i will eventually. i expect it'll be similar to Go, though, which is to say, a lot of interesting ideas, some really bad decisions, and odd syntax, the total of which isn't really interesting enough to me :-) [14:53:52.0000] <Ms2ger> It constructs a new Vec, which is a growable vector [14:54:03.0000] <Ms2ger> I hope we're getting better syntax for that at some point [14:54:16.0000] <zewt> how about [] [14:54:21.0000] <zewt> just saying :P [14:55:08.0000] <zewt> i'm skeptical of any language I can't read intuitively, without having to study first [14:55:22.0000] <Ms2ger> Yeah, that's used for a type that's currently being transitioned to being a non-growable vector [14:55:30.0000] <Ms2ger> It's confusing right now :) [14:56:06.0000] <zewt> (not to say read every nuance of, but to a reasonable level) [14:56:50.0000] <zewt> afk [15:22:14.0000] <SimonSapin> Hixie: we’d be interested in knowing what you consider bad decisions, with 1.0 coming soon [15:23:43.0000] <SamB> /me wonders: 1.0 of what? [15:24:00.0000] <SimonSapin> Rust [15:24:07.0000] <SimonSapin> http://www.rust-lang.org/ [15:24:31.0000] <SamB> /me does know roughly what rust is, even if he can't remember who is working on it [15:24:41.0000] <SamB> it's a better name than, uh, Go [15:25:33.0000] <SimonSapin> we try not to brand it as such, but it’s lead mostly by Mozilla Research [15:25:38.0000] <SamB> though admittedly still seems awfully common of a word to use for a software product's name [15:25:54.0000] <SimonSapin> worse than Perl, Java, Ruby, and Python? :) [15:26:02.0000] <SimonSapin> or C [15:26:14.0000] <SamB> I don't know why Perl is in that list [15:26:23.0000] <SamB> though you have a point [15:26:45.0000] <SimonSapin> oh, it’s spelled pearl in english, sorry [15:26:47.0000] <SamB> not that I've personally seen/heard Java used of the coffee all that often [15:27:36.0000] <SamB> /me is amused by VERA's first entry for perl [15:29:17.0000] <zewt> none as bad as calling a browser "chrome" [15:29:44.0000] <zewt> thus making it forever impossible to search for anything regarding firefox's chrome [15:34:25.0000] <zewt> just saw firefox's current fullscreen implementation for the first time, it's ... pretty terrible :| [15:34:46.0000] <zewt> every time i focus firefox on the other monitor a giant thing pops up FIREFOX IS NOW FULLSCREEN (which it already told me the first time) [16:34:02.0000] <Hixie> SimonSapin: using abbreviations like "fn" and "mod" instead of "function" and "module", using non-standard punctuation like || and !, having multiple string syntaxes, non-standard attribute syntax, that kind of thing [16:34:40.0000] <Hixie> SimonSapin: also i don't understand some decisions, which, unless they have good reasons, would also fall into that bucket, like needing to explicitly give the type name before a { } struct literal [16:41:25.0000] <Hixie> SimonSapin: (the non-standard punctuation thing extends to e.g. the way ~ is used. I'd personally think it'd be much more readable if it was keywords rather than punctuation.) [16:41:29.0000] <Hixie> hth [16:50:15.0000] <smaug____> zewt: file a bug [16:50:32.0000] <smaug____> zewt: not too many users have several monitors [16:50:45.0000] <smaug____> nor developers [16:52:55.0000] <smaug____> SimonSapin: btw looks like I agree with Hixie on bad design decisions of Rust. It is just too alien comparing to other languages, and mostly without any good reason. [16:53:47.0000] <Hixie> i should add that i have zero experience writing rust, so, you know, take what i said with a grain of salt ;-) [16:54:40.0000] <Hixie> but in general, where the language isn't being innovative, i would recommend sticking with what people know [16:54:49.0000] <Hixie> it'll make it easier for people to pick up the actually new stuff [16:55:45.0000] <Hixie> on a different note, i don't understand the difference between https://developer.mozilla.org/en-US/docs/NSS_Certificate_Download_Specification and https://wiki.mozilla.org/index.php?title=CA:Certificate_Download_Specification&action=history [16:55:55.0000] <Hixie> uh [16:56:08.0000] <Hixie> https://developer.mozilla.org/en-US/docs/NSS_Certificate_Download_Specification and https://wiki.mozilla.org/CA:Certificate_Download_Specification [16:57:08.0000] <annevk> Hixie: file a bug? [16:57:12.0000] <Hixie> on? [16:57:19.0000] <annevk> Hixie: MDN I suspect [16:57:58.0000] <Hixie> i don't really know what the bug is [16:58:55.0000] <annevk> Hixie: duplication of content [16:59:06.0000] <annevk> Hixie: without clarity on what is canonical [16:59:14.0000] <Hixie> it's not quite duplicate [16:59:15.0000] <Hixie> :-) [16:59:29.0000] <annevk> Not sure that makes it better [16:59:52.0000] <TabAtkins> Hixie: The "fn" abbreviation, at least, has a pretty good pedigree. Several languages use it. 2014-04-02 [17:00:03.0000] <Hixie> still unnecessary :-) [17:00:15.0000] <abarth> Hixie: can any browser actually compute the information required to "await a stable state" ? [17:00:16.0000] <Hixie> i'm sure you can find a precedent for pretty much anything [17:00:24.0000] <Hixie> abarth: how do you mean? [17:00:37.0000] <abarth> Hixie: we're trying to implement some things related to HTMLImageElement [17:00:48.0000] <Hixie> abarth: "stable state" is something you reach in the event loop. [17:00:50.0000] <abarth> Hixie: that the spec says to "await a stable state" but it seems very hard [17:00:53.0000] <Hixie> abarth: nothing to compute [17:01:04.0000] <abarth> what does "nothing to compute" mean? [17:01:13.0000] <abarth> there are always things scheduled in the event loop [17:01:18.0000] <Hixie> you don't have to compute anything to reach a stable state [17:01:32.0000] <Hixie> you just follow the loop processing model and it tells you when you're supposed to offer a stable state [17:01:44.0000] <Hixie> (which is basically any time you run microtasks) [17:01:55.0000] <Hixie> (but see the spec for specifics) [17:02:05.0000] <abarth> maybe I misunderstand the spec [17:02:16.0000] <abarth> but if there's an algorithm in the spec that spins the event loop [17:02:27.0000] <abarth> then isn't the stable state delayed until that algorithm completes? [17:03:17.0000] <Hixie> "provide a stable state" happens on step 8 of the event loop. [17:03:32.0000] <abarth> i see that [17:03:49.0000] <abarth> http://www.whatwg.org/specs/web-apps/current-work/#spin-the-event-loop [17:03:59.0000] <abarth> but it doesn't happen after step 6 of that algorithm [17:04:01.0000] <Hixie> "spin the event loop" returns to step 4, so the stable state will happen shortly after. [17:04:22.0000] <Hixie> that is, spin the event loop's step 7 returns the event loop to step 4 [17:04:35.0000] <Hixie> and from there it's quick work to reach 8. [17:04:52.0000] <Hixie> since 4, 5, and 6 are trivial and 7 is microtasks [17:05:04.0000] <abarth> i see [17:05:06.0000] <Hixie> (there's an open bug on making stable state into a microtask) [17:05:12.0000] <Hixie> (but that would make little difference) [17:05:29.0000] <abarth> so, when you spin the event loop, you actually return to the top-level event loop in the spec [17:05:52.0000] <abarth> so, why is there a microtask checkpoint at step 6 of spinning the event loop? [17:06:01.0000] <abarth> won't we get back to the main event loop and checkpoint there? [17:06:09.0000] <Hixie> yeah, there's an open bug on that too iirc [17:06:35.0000] <abarth> basically, the person submitting the CL is trying to use the microtask checkpoints to implement stable states [17:06:45.0000] <Hixie> (basically "spin" ends the current task, goes async (resuming the event loop), and then waits for the condition to be met and posts a continuation as a task. It never actually spins anything or resumes anything directly itself.) [17:06:47.0000] <abarth> and I'm trying to figure out if they're doing it right :) [17:06:51.0000] <Hixie> abarth: foolip? [17:07:09.0000] <Hixie> abarth: it's his bug asking for me to change the spec iirc [17:07:12.0000] <abarth> https://codereview.chromium.org/200923002/ [17:07:20.0000] <abarth> cbiesinger [17:07:33.0000] <abarth> apparently HTMLImageElement waits for a stable state before kicking off the load [17:07:33.0000] <Hixie> ugh, <picture> [17:07:39.0000] <Hixie> that really is gonna cause you all kinds of trouble [17:07:46.0000] <Hixie> <source> is a bad design [17:07:47.0000] <smaug____> Hixie: how could stable state be microtask? [17:07:54.0000] <Hixie> but anyway [17:07:55.0000] <smaug____> isn't stable state between tasks [17:08:04.0000] <Hixie> smaug____: microtasks are between tasks too [17:08:08.0000] <abarth> Hixie: oh, is it only in the <picture> spec that HTMLImageElement waits? [17:08:18.0000] <smaug____> there can be several microtasks in one task [17:08:53.0000] <abarth> smaug____: in the impl, we can tell the difference between the nested checkpoints and the top-level checkpoint [17:09:01.0000] <abarth> they're triggered by different systems [17:09:11.0000] <abarth> e.g., the scripting engine versus the event loop [17:09:16.0000] <Hixie> abarth: <img> as specced does a stable state too, <picture>'s problems are bigger than that. [17:09:38.0000] <Hixie> smaug____: yeah, there can be several synchronous sections (stable state consumers) too [17:09:44.0000] <Hixie> smaug____: it's all in the bugs [17:09:54.0000] <abarth> http://www.whatwg.org/specs/web-apps/current-work/#htmlimageelement step 8 says "await a stable state" [17:09:59.0000] <abarth> he's trying to implement that [17:10:06.0000] <smaug____> but how could stable state be microtask ? [17:10:24.0000] <Hixie> smaug____: i don't understand the question. why would it be hard? [17:10:26.0000] <smaug____> I don't want to start handling stable state stuff after each event listener call [17:10:31.0000] <Hixie> abarth: right [17:10:46.0000] <Hixie> abarth: you really want to coordinate with foolip (philipj). He's doing this stuff in blink. [17:10:57.0000] <Hixie> smaug____: comment on the relevant bug :-) [17:11:28.0000] <smaug____> which one is the relevant bug [17:12:10.0000] <abarth> Hixie: yes, he's CCed [17:12:43.0000] <abarth> Hixie: what is the underlying problem that waiting for a stable state here is solving? [17:12:48.0000] <smaug____> ah, https://www.w3.org/Bugs/Public/show_bug.cgi?id=24724 [17:13:10.0000] <Hixie> abarth: not wanting to do redundant work when a script modifies an attribute several times in a row [17:13:22.0000] <Hixie> smaug____: that's the one [17:13:24.0000] <abarth> Hixie: we sometimes load images synchronously out of the cache to avoid trashing layout [17:13:28.0000] <Hixie> smaug____: you found it faster than me :-) [17:13:31.0000] <abarth> because image width / height is an input to layout [17:13:37.0000] <Hixie> abarth: yeah, that's specced too iirc [17:13:50.0000] <Hixie> abarth: though the img loading model needs updating to handle the two-image thing, there's a bug on that too [17:14:24.0000] <abarth> I see the "list of available images" [17:14:31.0000] <Hixie> https://www.w3.org/Bugs/Public/show_bug.cgi?id=24958 is the bug iirc [17:16:04.0000] <abarth> Hixie: thanks for the info [17:17:31.0000] <Hixie> sorry i can't be more helpful, not sure what to tell you :-) [17:17:52.0000] <Hixie> i really would highly encourage you to push back on a multi-element image-loading feature [17:17:59.0000] <Hixie> it's a world of pain waiting for you [17:22:29.0000] <a-ja> <picture><img srcset> + negotiation <== sounds like fun [17:25:38.0000] <TabAtkins> Ah, I didn't realize I had been volunteered to give a lightning talk at Extensible Web Summit. [17:25:51.0000] <TabAtkins> That's fine, but I would have liked to be told by someone other than a Lanyrd update email. [17:26:09.0000] <Hixie> uh [17:26:13.0000] <Hixie> that's "nice" of them [17:36:32.0000] <TabAtkins> At least it's just a lightning talk, not a full-session talk. [17:36:43.0000] <TabAtkins> I can do a lightning talk with 5 minutes prep. [19:53:39.0000] <JosephSilber> Is there no way to prevent input elements from streching their parent td? [21:14:46.0000] <Rastus_Vernon> Aaaah, what is this? http://dev.w3.org/csswg/css-sizing/ [23:33:57.0000] <Ms2ger> I guess still half an hour left of April Fools [00:34:17.0000] <zcorpan> Hixie: it would be useful if you could be more specific about the problems with <picture>, maybe if you could read and review the current spec http://picture.responsiveimages.org/ (it's far from the same spec it was a year ago) [00:34:52.0000] <sangwhan> irccloud says i have a mention somewhere in the backlog but manual searching seems to disagree [00:35:15.0000] <zcorpan> Hixie: i think you're the only one at the moment pushing back [00:35:26.0000] <sangwhan> /me was hoping TabAtkins would have had a answer about prefixing [05:06:47.0000] <yoav> SimonSapin & TabAtkins: If you guys are around. I've got a question regarding http://dev.w3.org/csswg/css-syntax/#consume-a-string-token0 [05:08:37.0000] <yoav> If I understand correctly, under reverse solidus, the "if the stream starts with a valid escape" part is not needed [05:08:44.0000] <yoav> Is that correct? [05:22:12.0000] <zcorpan> can someone get rid of that stylesheet by now? [05:23:53.0000] <jgraham> zcorpan: The April Fools joke was actually that you thought it was only for a day [05:34:27.0000] <yoav> jgraham :D [05:35:51.0000] <zcorpan> instead it will get worse every day. in a few hours the text will start following the cursor [05:36:21.0000] <jgraham> Nah, Bert would never approve the use of javascript [05:37:50.0000] <zcorpan> then an image of Bert will also start following the cursor [05:38:08.0000] <jgraham> I think they'll go more subtle [05:38:34.0000] <jgraham> CSS Filters + Animations will be used to subtly blur the text [05:38:48.0000] <jgraham> So it will feel like it's going an and out of focus [05:39:18.0000] <jgraham> Because everyone knows that the CSS WG is actually run by a shady cabal of opticians [05:39:52.0000] <jgraham> Who else could these "Members that aren't interested in browser applications" be? [06:10:29.0000] <SimonSapin> zcorpan: done [06:10:54.0000] <zcorpan> thx [06:11:06.0000] <jgraham> You mean that Bert now follows the cursor? [06:11:51.0000] <SimonSapin> yoav: it’s needed in case you have '\' followed by a newline [07:04:51.0000] <IZh> Hi. :-) [07:05:48.0000] <IZh> How do you spell WHATWG? ;-) [07:06:51.0000] <IZh> W.H.A.T.W.G. or What W.G. or something else? [07:07:12.0000] <gsnedders> WHATWG. [07:07:23.0000] <SimonSapin> do you mean pronounce? [07:07:29.0000] <IZh> Yes. [07:08:27.0000] <IZh> Double U eight ey tee double U Gee? What double U Gee? [07:08:53.0000] <gsnedders> The latter most go for. I don't think there's any official rule, though. [07:11:29.0000] <zewt> double hat wig [07:11:57.0000] <IZh> Whatwig [07:16:43.0000] <yoav> SimonSapin: Isn't that case covered by the "Otherwise, if the next input code point is a newline, consume it." that comes right be before it? [07:17:46.0000] <SimonSapin> yoav: wait, is this for "consume a token", "consume an ident-like token", or "consume a string" ? [07:18:13.0000] <yoav> consume a string [07:18:18.0000] <SimonSapin> ah [07:18:22.0000] <SimonSapin> I was looking at the wrong place [07:19:18.0000] <SimonSapin> uhm, yes, it’s redundant now that we changed \<EOF> to not be an error [07:21:08.0000] <yoav> SimonSapin: OK, great! I'll delete that if without guilt, then :) [07:26:19.0000] <SimonSapin> yoav: Is this better? https://dvcs.w3.org/hg/csswg/raw-file/86568d6b9ce9/css-syntax/Overview.html#consume-a-string-token [07:26:31.0000] <SimonSapin> (It’ll take a few minutes to propagate to dev.w3.org.) [07:27:47.0000] <yoav> SimonSapin: That's great! [08:11:59.0000] <annevk> SimonSapin: what is evil utf-8? [08:12:22.0000] <SimonSapin> annevk: allows surrogates [08:12:35.0000] <annevk> SimonSapin: but is not CESU-8? [08:12:51.0000] <SimonSapin> it’s unclear whether CESU-8 allows lone surrogate [08:13:08.0000] <SimonSapin> I think it does not, as a consequence of UTF-16 not allowing them [08:13:10.0000] <annevk> SimonSapin: CESU-8 does not allow > U+FFFF [08:13:30.0000] <annevk> SimonSapin: that might be true in addition I suppose [08:13:42.0000] <SimonSapin> what CESU-8 does is encode non-BMP stuff as a surrogate pair, which doesn’t affect the value-space [08:14:36.0000] <SimonSapin> the point of "evil UTF-8" is to have the value space of JS strings [08:15:39.0000] <SimonSapin> I proposed it as half a joke, but interestingly roc seems to think it’s realistic https://groups.google.com/forum/#!msg/mozilla.dev.servo/1K2-Qy27e3A/Q9DWOO8Nt3EJ [08:16:43.0000] <annevk> Looks interesting [08:17:36.0000] <annevk> SimonSapin: a problem with evil utf-8 is that its value space is actually bigger [08:17:43.0000] <SimonSapin> how so? [08:17:58.0000] <annevk> SimonSapin: you can have paired surrogates as well as the actual code point [08:20:24.0000] <SimonSapin> good point [08:20:48.0000] <annevk> hsivonen: congratulations! [08:20:51.0000] <SimonSapin> so maybe it should be CESU-8 modified to allow lone surrogates [08:21:17.0000] <annevk> SimonSapin: the bad thing about that is that then you cannot just copy the input stream [08:21:40.0000] <annevk> SimonSapin: it's a tricky problem [08:21:42.0000] <SimonSapin> indeed [08:22:00.0000] <SimonSapin> would it make sense to only allow unpaired surrogates? [08:22:03.0000] <annevk> I guess you can accept that there will be two ways to represent certain code points [08:22:25.0000] <SimonSapin> that sounds bad [08:22:45.0000] <SimonSapin> makes == either wrong or hard to implement [08:22:49.0000] <annevk> It's a bit unclear to me how you can enforce that from JavaScript [08:23:07.0000] <annevk> If you can make it so that you only allow them unpaired, that would be interesting [08:23:28.0000] <SimonSapin> when decoding a JS string, turn paired surrogates to 4-bytes sequences [08:24:41.0000] <annevk> There might be some tricky edge cases, where we store strings as a list and you use += with surrogates [08:25:29.0000] <annevk> But yeah, that seems sensible, especially since it should be the edge case, over time people will use the new Unicode escapes or utf-8 directly [08:27:43.0000] <SimonSapin> I don’t understand the list thing [08:28:19.0000] <SimonSapin> but I think roc’s idea it to make this invisible from script, .charAt() would still give you UCS-2 units [08:31:04.0000] <jgraham> SimonSapin: If you have a String stored as a fragment + pointer to next fragment [08:31:24.0000] <jgraham> Which I think spidermonkey sometimes does to make append fast [08:32:11.0000] <SimonSapin> jgraham: ok, what about it? [08:32:33.0000] <jgraham> So if you had var a = low_surrogate + high_surrogate [08:33:15.0000] <jgraham> You would have a string represented as (low_surrogate, pointer) (high_surrogate, null) [08:33:24.0000] <jgraham> But semantically it would be one character [08:34:22.0000] <jgraham> (I'm not sure this would be a big problem in practice) [08:37:15.0000] <annevk> Presumably we would not use evil-utf-8 for those situations [08:50:32.0000] <dglazkov> good morning, Whatwg! [09:32:00.0000] <SimonSapin> annevk: roc’s idea seems to be to use it for everything [09:32:57.0000] <annevk> SimonSapin: you mean he'd change all string types in SpiderMonkey? I somewhat doubt that [09:34:24.0000] <arunranga> annevk, hai! I think I'd like to explore your idea of an "object list" associated with Blobs a bit more. [09:34:32.0000] <SimonSapin> https://groups.google.com/forum/#!msg/mozilla.dev.servo/1K2-Qy27e3A/Q9DWOO8Nt3EJ [09:34:57.0000] <arunranga> annevk, doesn't this mean that user agents have to analyze code beforehand? [09:35:26.0000] <Hixie> zcorpan: i went through it when srcset was first being discussed years ago. [10:10:56.0000] <annevk> arunranga: why? [10:11:23.0000] <arunranga> annevk, how is the object list constructed exactly? [10:12:34.0000] <annevk> arunranga: it's just something a blob has [10:12:47.0000] <annevk> arunranga: and other objects can use to make sure they can read the blob [10:13:35.0000] <arunranga> annevk, ok, but how do other objects get inserted into the list? How will a Blob deterministically add objects to it? [10:14:26.0000] <annevk> arunranga: the other objects would add themselves to it as part of their algorithms [10:14:39.0000] <annevk> arunranga: e.g. FormData would do it as part of serializing itself [10:15:36.0000] <arunranga> annevk, I see. So, if the Blob is CLOSED, an object can NOT add itself to the Blob's list. But pre-existing objects can still get serviced. [10:16:04.0000] <arunranga> annevk, and CLOSED blobs might be GC'd when their object lists are empty. [10:16:11.0000] <annevk> yes [10:17:06.0000] <arunranga> annevk, I think this makes sense. [10:17:24.0000] <arunranga> /me walks off to spec it [10:42:47.0000] <zcorpan> is there no way to serialize an "xml" document as html? [10:43:01.0000] <zcorpan> i thought we'd extended XMLSerializer to support that? [10:45:20.0000] <gsnedders> Create an HTML document, adopt node, then use that? [11:00:57.0000] <zcorpan> yeah, but that seems like a workaround [11:01:06.0000] <zcorpan> should be able to serialize it directly [11:27:33.0000] <Hixie> <Hixie> zcorpan: i went through it when srcset was first being discussed years ago. [11:29:39.0000] <odinho> That was different. [11:30:12.0000] <SamB> what's an "xml" document, and how does it differ from an XML document [11:31:12.0000] <odinho> Proposal was very different back then. -- The way <picture> is being done now is vastly simpler. And the most interesting/needed parts of <picture> does not use the tag at all. The 90% use case is <img> with "sizes" and "srcset" (albeit a different srcset than yours, the "w" modifier tells how wide the image is) [11:31:56.0000] <zcorpan> SamB: it doesn't differ. but it doesn't need to have come into being from parsing XML source. e.g. `new Document()` creates an XML document [11:32:17.0000] <zcorpan> SamB: whereas document.implementation.createHTMLDocument('foo') create an HTML document [11:32:23.0000] <odinho> <picture> is only metadata to an <img> that does the actual work. [11:39:33.0000] <zcorpan> Hixie: in http://lists.w3.org/Archives/Public/public-whatwg-archive/2012May/0247.html i see concern over verbosity, which i guess is still valid, but it turned out people didn't like the more terse solution with attributes on <img> [11:39:59.0000] <zcorpan> Hixie: is it just verbosity or something else? [11:41:19.0000] <Hixie> search for "multiple-element" in http://lists.w3.org/Archives/Public/public-whatwg-archive/2012Aug/0070.html [11:42:16.0000] <Hixie> multiple-element structures are just a nightmare to manage [11:42:36.0000] <Hixie> <video><source> and <select><option> are great examples [11:42:45.0000] <Hixie> even <object> with its impact on descendants [11:42:50.0000] <SamB> zcorpan: is that the one with the crazy attribute naming? [11:42:52.0000] <Hixie> they all suck so much [11:44:18.0000] <SamB> Hixie: nightmare from which side(s)? [11:44:27.0000] <Hixie> eveyone [11:44:34.0000] <Hixie> authors, implementors, specs [11:44:43.0000] <SamB> so, um, what's better? [11:44:47.0000] <zcorpan> Hixie: i think <picture> sucks much less than <video> [11:44:59.0000] <Hixie> SamB: anything that only involves one element, ideally one attribute. [11:45:15.0000] <Hixie> zcorpan: i'd hope we'd have a higher bar than that... [11:45:17.0000] <SamB> Hixie: that can perhaps get a bit TOO terse [11:45:57.0000] <SamB> though at least it doesn't have the crazy attribute names [11:46:20.0000] <Hixie> SamB: an attribute just takes a string, there's nothing a priori terse or verbose about it [11:46:37.0000] <Hixie> you could put an entire C++ program in there... [11:46:37.0000] <SamB> hmm [11:46:49.0000] <Hixie> or the machine code equivalent [11:46:57.0000] <Hixie> or the java equivalent [11:46:58.0000] <SamB> maybe I'm just used to not being allowed newlines between " and " [11:47:11.0000] <Hixie> just to give three examples along the line of "terse-verbose" [11:47:20.0000] <zcorpan> Hixie: sure. we tried different things and so far it seems this sucks the least, and it seems implementors are on board now [11:47:26.0000] <zcorpan> Hixie: anyway, looking at the email... [11:47:28.0000] <SamB> Hixie: which one was supposed to be the terse one [11:47:46.0000] <Hixie> zcorpan: yeah, well, i just hope someone else has to clean up the mess, cos i don't want to, that's for sure [11:47:57.0000] <Ms2ger> I'm afraid it's your job [11:47:57.0000] <Hixie> zcorpan: took me long enough to clean up <select>'s mess [11:48:13.0000] <zcorpan> Hixie: "they introduce the need for much more elaborate error handling," is handled fine in the spec, it's not really much more elaborate [11:48:13.0000] <Ms2ger> And maybe better to clean it up while it can still be changes [11:48:18.0000] <Ms2ger> changed* [11:48:42.0000] <zcorpan> Hixie: i'm pre-emptively working on having it clean :-) [11:48:57.0000] <zcorpan> Hixie: but i have to monkey-patch html which sucks [11:50:07.0000] <zcorpan> Hixie: "the processing model has to deal with changes more complicated than just "change" (what if an element is added or removed, or moved?)" - true, but it's covered. it's "just" a list of mutations that trigger an algorithm [11:50:38.0000] <Hixie> zcorpan: would you like to just own the entire <img> section? I can just have it automatically import from some doc you maintain [11:50:43.0000] <hober> zcorpan: when you say implementors are on board now, who are you referring to? [11:53:15.0000] <Hixie> zcorpan: i'm sure it's covered, but how many bugs does it have? how many edge cases have you forgotten? how many race conditions will we run into? [11:53:24.0000] <Hixie> zcorpan: we are _still_ fixing <source>, years later [11:53:28.0000] <zcorpan> hober: mozilla and blink at least are working on implementing now. not sure about apple and microsoft but i don't think there have been negative signals [11:53:53.0000] <Hixie> zcorpan: and the bugs we're finding aren't any simpler than they were at first, in fact i'd say they're much more insidious and hard to fix now than they were when we started [11:54:01.0000] <Hixie> zcorpan: if i could go back, i'd never introduce <source> [11:54:56.0000] <zcorpan> Hixie: yeah i agree about <video>'s <source>, but <picture>'s <source> isn't the same. i'm not saying there won't be bugs or that it will be easy this time, but i think i've learned the lesson about <video><source> :-) [11:55:12.0000] <Hixie> what i'm saying is that the lesson i learnt is "never again" [11:55:24.0000] <Hixie> i'm serious about the offer to just give you ownership of <img> [11:56:25.0000] <Hixie> (also, i assume there's no actual <source> in this <picture> thing? you're using <img>, right? not <source>?) [11:56:27.0000] <zcorpan> that's ok with me. let's wait a bit with that though to see how things shake out [11:56:37.0000] <zcorpan> it's <source> [11:56:53.0000] <Hixie> please don't overload <source> [11:57:20.0000] <zcorpan> what's the problem with that? <source> in itself doesn't do anything. also the attributes are different [11:57:56.0000] <Hixie> i don't want to have to deal with a bug where <source> is in a shadow dom and then grafted into a <video> while the <picture> processing model is running and then on the next microtask it's moved to a different shadow dom with an <audio> [11:58:24.0000] <zcorpan> we could call it something else i guess, but <img> would be bad for back-compat [11:58:30.0000] <Hixie> (also, you say <picture> is simpler, but you have _three_ elements?) [11:58:52.0000] <zcorpan> yeah [11:58:57.0000] <Hixie> that's not simpler. [11:59:05.0000] <Hixie> you're gonna have so much pain. [11:59:30.0000] <Hixie> anyway. please avoid interacting with other elements so we never have to worry about the case above. [11:59:41.0000] <Hixie> e.g. overloading existing elements [11:59:57.0000] <Hixie> let's keep this very self-contained [12:00:07.0000] <zcorpan> i'll file a bug about the <source> tag name, thanks [12:09:29.0000] <zcorpan> https://github.com/ResponsiveImagesCG/picture-element/issues/133 [12:14:01.0000] <zcorpan> Hixie: let's go on with the old email, in case there's something more that isn't addressed already [12:15:07.0000] <zcorpan> Hixie: "it introduces complexities in the algorithms to deal with unexpected text nodes, comment nodes, PIs, etc." - that's handled, it just skips over anything that isn't a <source> or the <img> element itself [12:16:50.0000] <Hixie> by "it's handled" in all these cases you presumably mean "i think i have handled that because it does in fact need to be handled with this design, though it wouldn't have to be if we avoided a multiple element design, and i've no idea if there won't be subtle bugs found with it later" [12:16:56.0000] <zcorpan> Hixie: "it introduces some complexity in the parser, ..." - that's handled, the algorithm starts when <img> is seen (it's <img> that does the "work"), and at that point all information is already there. anything after it is ignored [12:17:18.0000] <Hixie> what if the img is in a different document than the <picture>? [12:17:27.0000] <Hixie> does it still handle it then? [12:18:05.0000] <zcorpan> yeah then the <picture> wouldn't be a parent of the <img> so it would act like it was just the <img> and no <picture> [12:19:09.0000] <Hixie> my point is that these are questions that just don't come up if you don't do a multiple-element design [12:21:22.0000] <zcorpan> ok. yep that's true [12:23:38.0000] <zcorpan> Hixie: thanks for the feedback [12:24:01.0000] <Hixie> zcorpan: i've put markers in the spec for the part that i'm ready to hand off to you. let me know when you're ready to take it, and i'll replace that segment with a #include that merges your source in. [12:24:24.0000] <Hixie> that way you shouldn't have to worry about hooks. [12:24:28.0000] <Hixie> or monkeypatching. [12:24:47.0000] <zcorpan> ok cool [12:25:15.0000] <zcorpan> i'm ready when you've fixed my bugs :-D [12:27:05.0000] <Hixie> zcorpan: if you're going to be fiddling with that stuff anyway, might be best for you to just fix them as part of rejigging that part of the spec :-) [12:27:38.0000] <zcorpan> Hixie: sorry, not ready yet! :-P [12:27:52.0000] <Hixie> :-) [12:31:11.0000] <zewt> multi-element images is bizarre and out of place because of the two xml-like axes for data (sub-elements and attributes), html has always chosen attributes [12:32:34.0000] <Hixie> not always, but every time it's chosen elements it has been a mistake, imho. [12:59:20.0000] <SamB> /me wonders about title of this section: http://www.whatwg.org/html#the-document-object [13:07:40.0000] <SamB> I mean, why's that called Document and not HTMLDocument [13:10:31.0000] <SamB> /me somehow doubts the people responsible for the package in question are at fault WRT the non-free junk [13:10:55.0000] <SamB> er. [13:10:57.0000] <SamB> ECHAN [13:11:02.0000] <SamB> /me hates it when he does that [13:27:34.0000] <smaug____> SamB: some people think there should be only Document, not Document and HTMLDocument [13:27:52.0000] <smaug____> that doesn't match the reality too well atm [13:37:32.0000] <SamB> it would perhaps be easier to work towards that by starting with the way things are now? [13:37:52.0000] <Hixie> smaug____: it's not so much that we think there should only be one, it's that having two makes no sense in a multi-namespace world [13:38:29.0000] <Hixie> SamB: specs point to where we should be, not where we are (or were) [13:41:06.0000] <smaug____> well, it is not clear what kinds of documents XHR should return for example [13:41:30.0000] <Hixie> if there's only one kind of document, it seems pretty clear :-) [13:41:32.0000] <smaug____> adding all the legacy html stuff there doesn't make much sense [13:41:45.0000] <smaug____> since lots of those APIs would be just broken [13:42:04.0000] <annevk> So yes, somebody needs to go through the various APIs returning documents and make suggestions [13:42:40.0000] <annevk> Though I'm not super convinced that not exposing certain members here and there will actually be a large benefit [13:43:09.0000] <Hixie> i don't understand [13:43:14.0000] <Hixie> why would we want to hide certain APIs? [13:43:18.0000] <Hixie> just have them return null, etc [13:43:27.0000] <Hixie> pretty sure we already do that anyway [13:43:48.0000] <Hixie> e.g. if you hold onto a Document that's been ungrafted from browsing contexts [13:44:06.0000] <annevk> We do, we also discussed this on the WHATWG list [13:44:26.0000] <Hixie> yeah [13:44:32.0000] <annevk> I'm not sure what camp I'm in, I'd like this to move forward though [13:44:33.0000] <Hixie> i don't really understand what's controversial at this point [13:44:36.0000] <Hixie> i thought this was all long resolved [13:45:00.0000] <smaug____> really? I thought this was never resolved [13:45:23.0000] <smaug____> but was just a thing which was spec'ed to be different what implementations do [13:45:35.0000] <smaug____> but I don't have a strong opinion on this one [13:48:19.0000] <SamB> claiming that we already have everything on Document doesn't seem to have made it happen, somehow ... [13:48:32.0000] <Hixie> nobody is claiming that [14:05:57.0000] <SamB> /me wonders why nobody minds SVG being multi-element [14:09:15.0000] <Hixie> more than half the e-mails i have pending in my "canvas" folder are from adobe [14:09:40.0000] <Hixie> and the majority of the rest are from junov... [14:10:05.0000] <astearns> /me perhaps we should hire junov... [14:23:40.0000] <SamB> /me wonders how they add stuff to Java's DOM APIs ... [14:37:44.0000] <SamB> hmm, I don't suppose anyone would much care if *these* actually stayed on HTMLDocument until the heat death of the universe: http://www.whatwg.org/specs/web-apps/current-work/multipage/obsolete.html#Document-partial [14:37:59.0000] <Hixie> HTMLDocument doesn't exist anymore according to the HTML Spec. [14:38:35.0000] <Hixie> per spec, document instanceof HTMLDocument returns true only because Document === HTMLDocument [14:38:52.0000] <Hixie> ("For historical reasons, Window objects must also have a writable, configurable, non-enumerable property named HTMLDocument whose value is the Document interface object.") [14:39:55.0000] <SamB> hmm, that wasn't in the index [14:41:10.0000] <Hixie> which index would it be in? it's not an interface, event, element, attribute, or content model [14:42:40.0000] <SamB> it *used* to be an interface though [14:43:30.0000] <Hixie> yes 2014-04-03 [18:24:53.0000] <benschwarz> hey @ Hixie - someone just hit me up on IRC saying "The file http://www.whatwg.org/specs/web-apps/current-work/images/robots.jpeg has wrong MIME-type text/html. The web-server config have to be corrected." [18:26:40.0000] <TabAtkins> Chrome's network tab says it's image/jpeg. [18:30:47.0000] <benschwarz> TabAtkins: heh. I didn't check [18:31:00.0000] <benschwarz> so does curl [18:31:06.0000] <benschwarz> it also says Content-Language: en-GB-x-Hixie [18:36:49.0000] <benschwarz> TabAtkins: this time next week we get to kick back in sunny Melbourne [19:09:16.0000] <TabAtkins> benschwarz: Yessssss [19:09:40.0000] <TabAtkins> benschwarz: And I found nail polish that appears to be the exact shade of the cssconf.au page background. ^_^ [19:09:59.0000] <benschwarz> we saw ! [19:10:09.0000] <benschwarz> we posted it in our hipchat [19:10:18.0000] <benschwarz> we're thinking a booth setup [19:10:25.0000] <benschwarz> where we get our nails did' [19:10:33.0000] <TabAtkins> +1 [19:22:18.0000] <zewt> hate@hipchat [20:44:04.0000] <Hixie> benschwarz: weird, dunno why they're asking you. anyway, i fixed those earlier today when someone was asking about it. [21:42:30.0000] <MikeSmith> Hixie: thanks for the rel="shortcut icon" change. I'll implement it in the validator this week [21:42:42.0000] <Hixie> np [21:42:59.0000] <Hixie> sorry i've been slow recently. i should be back on more reliably next week. [21:43:07.0000] <Hixie> (then i disappear again for a week, iirc) [22:56:54.0000] <zcorpan> hober: http://status.modern.ie/#/ says picture is under consideration. but it's not clear to me if that means anything. looks like they just copy stuff from chromestatus [22:57:56.0000] <zcorpan> hober: also http://lists.w3.org/Archives/Public/public-whatwg-archive/2014Jan/0037.html [01:21:58.0000] <ondras> can someone please point me to the spec where :host pseudo-class is defined? [01:30:59.0000] <MikeSmith> ondras: TabAtkins [01:31:24.0000] <MikeSmith> scoping spec [01:35:12.0000] <ondras> MikeSmith: okay thanks [01:35:17.0000] <ondras> TabAtkins: ? [01:35:47.0000] <MikeSmith> his spec [01:37:34.0000] <ondras> MikeSmith: well where can I find it then? [01:38:32.0000] <Ms2ger> http://drafts.csswg.org/css-scoping/ [01:39:25.0000] <Ms2ger> zcorpan, re "The harness should do as little as possible."... Didn't you suggest adding the div id=log? :) [01:39:36.0000] <Ms2ger> (I know, not nearly as intrusive as a script) [01:40:22.0000] <zcorpan> Ms2ger: yeah. i realize that adding a div can confuse things, but it's easy to opt-out [01:40:39.0000] <Ms2ger> And we already add the stylesheet, I think [01:40:47.0000] <zcorpan> also it only does that after the tests have finished [01:41:03.0000] <Ms2ger> Anyway, just throwing that out there :) [01:41:04.0000] <zcorpan> at which point it can do whatever, it won't affect the tests' result [01:41:43.0000] <zcorpan> maybe it could inject *report at that point also [01:42:24.0000] <Ms2ger> I think report wants to be running earlier [01:42:31.0000] <zcorpan> yeah [01:42:56.0000] <Ms2ger> /me goes back to work [04:45:08.0000] <manus> Hello! I have a couple quick questions around tabindex and various notions of focus, specifically regarding behaviors/rules for overflowed elements.. would here be an appropriate place to ask them? [04:45:36.0000] <manus> I've already done a bit of googling and light reading and testing and am still slightly confused about something. [04:46:33.0000] <MikeSmith> manus: fine to just go ahead and ask here [04:48:19.0000] <manus> Cool, thanks MikeSmith. Is there any defined/documented behavior for keyboard focus (as opposed to "normal" focus) for elements such as scrollable divs? [04:48:59.0000] <manus> I made a jsfiddle here: http://jsbin.com/sizofuse/7/edit?html,output [04:49:28.0000] <manus> and results are surprising, especially when tabindex is involved, and when nesting scrollable divs [04:50:14.0000] <manus> I've noticed browsers implementing different behaviors, and I just wanted to know if there's a part of this that hsa been, or will ever be, standardized [04:50:36.0000] <MikeSmith> SteveF: ↑☃ [04:51:33.0000] <MikeSmith> manus: no clue personally but if you hang out here for a bit somebody might be able to give you an informed answer [04:52:12.0000] <manus> Thanks, MikeSmith! I'll lurk here for a while, though may be away to lunch at some point [04:54:13.0000] <jgraham> So I think that generally the order in which various things get focus is supposed to be standardised, and the focus model itself is getting standardised, although I don't exactly know what would make scrollables different [04:56:15.0000] <MikeSmith> zcorpan: fyi I landed support to make rel="shortcut icon" checking match the current spec [04:56:23.0000] <MikeSmith> https://github.com/validator/syntax/commit/6befdd67ecf87754fc592cc0ac526d5c9c3c3345#diff-2 [04:56:34.0000] <MikeSmith> and pushed to http://validator.w3.org/nu/ [04:57:48.0000] <manus> jgraham: thanks for a response. I think scrollables have a different notion of focus, in terms of, for example, keyboard scrolling. if you click a div that has scrollable content, does it have "focus" from a standards perspective? are there different kinds of focus in play here? [04:58:05.0000] <jgraham> manus: I don't know too much about this :) [04:58:40.0000] <jgraham> Hixie was working on it, but you will have to wait another 4-5 hours for him to be around, I guess [04:58:48.0000] <manus> jgraham: ok, thanks anyway. :] It's just that current browser implementations don't agree, and I want to file a bug somewhere, but I don't know if this behavior is even specified [04:59:06.0000] <jgraham> manus++ for wanting to file a bug ;) [05:03:19.0000] <zcorpan> MikeSmith: "Error: Bad value shortcut icon for attribute rel on element link: If the shortcut keyword is present, the rel attribute's entire value must be shortcut icon. The string shortcut is not an absolute URL. " [05:03:32.0000] <MikeSmith> yeah [05:03:40.0000] <zcorpan> MikeSmith: i follow the message up to the last sentence [05:03:41.0000] <MikeSmith> that's because RDFa I think [05:03:51.0000] <zcorpan> @_@ [05:04:15.0000] <MikeSmith> yeah RDFa tokens in rel to be absolute URLs [05:04:46.0000] <MikeSmith> the checkers emits two separate error messages [05:05:18.0000] <MikeSmith> from two different datatype-checking classes that are unaware of each other [05:05:53.0000] <zcorpan> MikeSmith: but why doesn't it pretend that "shortcut" is registered keyword? [05:06:18.0000] <zcorpan> it doesn't complain that "icon" isn't a URL [05:07:20.0000] <MikeSmith> right, because it only proceeds with the absolute-URL check if the keyword check fails [05:07:49.0000] <MikeSmith> and if the keyword check fails the code in that class throws [05:08:51.0000] <MikeSmith> I don't have any means after it throws to proceed as if "shortcut" is registered [05:08:56.0000] <zcorpan> ok.. "Bad value icon shortcut for attribute rel on element link: If the shortcut keyword is present, the rel attribute's entire value must be shortcut icon. The string icon is not an absolute URL." [05:09:43.0000] <MikeSmith> hmm that's not good [05:10:48.0000] <zcorpan> ":" seems to pass the URL check [05:11:09.0000] <MikeSmith> sadly, RDFa [05:11:43.0000] <MikeSmith> that passes because it's a valid CURIE [05:12:09.0000] <MikeSmith> per the gigasmic mess that the RDFa specs are [05:13:21.0000] <zcorpan> do you support CURIEs? [05:13:58.0000] <MikeSmith> the W3C service does [05:14:54.0000] <MikeSmith> for whatever attributes where the RDFa says they're allowed [05:15:02.0000] <zcorpan> ;_; [05:15:08.0000] <MikeSmith> yup [05:15:54.0000] <zcorpan> anyway, it'd be nice if the url/curie checker somehow skipped the tokens that are registered keywords [05:16:16.0000] <MikeSmith> yeah I will see if I can make it do that somehow [05:16:30.0000] <zcorpan> maybe ax them out of the list before passing on to that checker or something [05:18:41.0000] <zcorpan> MikeSmith: rel=":}]%" also validates so it doesn't seem like it checks the url part much [05:21:59.0000] <zcorpan> MikeSmith: is safe_curie supported? where is that expected instead of curie? [06:34:29.0000] <SteveF> zcorpan: windows Progressive Disclosure Controls http://msdn.microsoft.com/en-us/library/windows/desktop/aa511487.aspx [06:40:35.0000] <MikeSmith> zcorpan: sorry lost my connection for a while [06:41:19.0000] <MikeSmith> zcorpan: safe_curie is supported yeah [06:42:28.0000] <MikeSmith> and I guess the reason rel=":}]%" validates is because ":}]%" is valid curie per RDFa [06:43:07.0000] <MikeSmith> prefixless curie I guess [06:43:54.0000] <MikeSmith> I vaguely recall something about prefix-lacking curies being allowed in RDFa [06:45:23.0000] <MikeSmith> so anyway I think the validator is not checking ":}]%" as a URL in that case but instead as a curie [06:45:56.0000] <MikeSmith> and the validator considers it a valid curie [06:46:16.0000] <MikeSmith> rightly or wrongly [06:46:46.0000] <MikeSmith> but I think rightly, since I'm just using whatever production for curie is given in the current RDFa spec [09:17:14.0000] <zcorpan> MikeSmith: the part after the colon is "url" stuff so i figured % was not allowed [11:41:45.0000] <annevk> Realms are actually interesting I just realized. They're a way to get a global without a browsing context... I wonder if that's going to lead to a bunch of issues [11:42:13.0000] <annevk> I think we carefully check for "associated browsing context" mostly. But if we have checks for "associated global" instead that might be trouble [14:32:50.0000] <cbiesinger> Hixie: hey [14:33:05.0000] <cbiesinger> Hixie: I was wondering if there's any progress on https://www.w3.org/Bugs/Public/show_bug.cgi?id=24724 ? [16:05:20.0000] <annevk> cbiesinger: it came up here the other day, doubt it [16:05:32.0000] <annevk> cbiesinger: Hixie would be available more next week btw [16:06:26.0000] <smaug____> cbiesinger: why would you want to merge microtask and stable state [16:06:33.0000] <smaug____> they are conceptually very different [16:07:00.0000] <annevk> smaug____: to have less concepts [16:07:01.0000] <smaug____> though, perhaps something which is using stable state could use microtask stuff [16:08:07.0000] <smaug____> I can see use cases for running something at the end of the current task [16:08:23.0000] <smaug____> for perf reasons for example 2014-04-04 [23:35:53.0000] <zcorpan> "Polyglot markup begins a comment with either "<!" or " Request.context [04:07:06.0000] <annevk> jungkees: I'd like to define those objects in Fetch in due course [04:07:23.0000] <annevk> jungkees: since Fetch also defines the underlying concepts it makes sense to couple them I think [04:07:39.0000] <jungkees> In due course, I agree [04:08:09.0000] <annevk> jungkees: as for the APIs defined now, we should not expose synchronous [04:08:17.0000] <jungkees> We're just defining it in SW as for now and may hook it into the part of the fetch later on I guess [04:08:19.0000] <annevk> jungkees: I'm not sure if timeout should be on Request [04:08:41.0000] <annevk> jungkees: yeah, we can make the switch when I get around to it I suppose [04:09:10.0000] <annevk> jungkees: also bookkeeping details such as redirectCount should be not exposed [04:09:35.0000] <annevk> jungkees: forceSameOrigin seems redundant with Mode [04:09:47.0000] <jungkees> Right. I think we better make a ticket in SW issue list for that so that people are aware of what you think of as a direction.. and your plan for it [04:10:10.0000] <annevk> jungkees: there's a general ticket on Fetch integration [04:10:30.0000] <jungkees> alright. then let's keep updating it upon our discussion [04:11:17.0000] <jungkees> annevk: and.. do we need to add "CORS-with-forced-preflight" in the Mode enum? [04:11:41.0000] <annevk> jungkees: I guess [04:11:55.0000] <jungkees> alright [04:12:27.0000] <annevk> jungkees: it's weird if you're the one making the request, but if the request came from XHR, it makes sense to expose that [04:16:45.0000] <jungkees> annevk: Alright I'll add it. And found https://github.com/slightlyoff/ServiceWorker/issues/242; think it is the right place to discuss the plan from this point on [04:17:44.0000] <annevk> jungkees: yeah sure, but lets use IRC when questions come up, much quicker [04:18:04.0000] <jungkees> annevk: Ok [06:01:14.0000] <annevk> jgraham: hey, can you walk me through a pull request merge again? [06:01:40.0000] <Ms2ger> "Click the button"? [06:01:58.0000] <annevk> Ms2ger: I'd prefer one commit in the history [06:02:08.0000] <jgraham> annevk: What are you trying to merge? [06:02:23.0000] <annevk> jgraham: https://github.com/whatwg/dom/pull/12 [06:02:36.0000] <annevk> Ms2ger: otherwise we get two tweets [06:03:36.0000] <jgraham> annevk: OK, github makes this a bit difficult because there are multiple repos involved [06:04:11.0000] <jgraham> So, locally, you can either add Aryeh's remote, or use one of the scripts on the internet that undoes some GH magic [06:04:18.0000] <jgraham> The first option is easier [06:04:50.0000] <jgraham> git remote add aryeh git⊙gc:ayg/dom.git [06:05:00.0000] <jgraham> git fetch origin [06:05:05.0000] <jgraham> git fetch aryeh [06:05:41.0000] <jgraham> git checkout aryeh/allow-filter-recursion [06:05:50.0000] <jgraham> git rebase origin/master [06:05:57.0000] <jgraham> git push origin HEAD:master [06:06:12.0000] <jgraham> git checkout master [06:06:21.0000] <jgraham> git merge --ff-only [06:06:38.0000] <jgraham> (the last two steps aren't really necesssary) [06:09:01.0000] <Domenic_> Let me find my bash script, it is nice [06:09:09.0000] <Domenic_> I type "pr 42" in the current repo and it figures it out [06:10:04.0000] <jgraham> Yeah, like I say magic scripts exist. Does it also allow you to land PRs? [06:10:33.0000] <annevk> seems almost easier to just apply https://github.com/ayg/dom/commit/61544aa440d0ef6cff10776a952a9562aa67e295.patch [06:11:07.0000] <jgraham> Well not really [06:11:45.0000] <jgraham> I mean apart from adding the remote there's only one step that would really replace [06:11:53.0000] <jgraham> Umm, that didn't make sense [06:12:05.0000] <jgraham> Anyway the point is that it wouldn't be much easier [06:12:33.0000] <Domenic_> https://gist.github.com/domenic/11371574 [06:12:45.0000] <jgraham> (in this case you could even optimise by checking out master; updating it to origin/master, cherry-picking aryeh's commit and pushing) [06:13:15.0000] <jgraham> Domenic_: that doesn't work [06:13:21.0000] <Domenic_> jgraham: I use it every day. [06:13:26.0000] <jgraham> annevk specifically wanted to not merge [06:13:33.0000] <Domenic_> jgraham: it rebases first [06:13:36.0000] <Domenic_> fast-forward merge [06:14:09.0000] <jgraham> Oh, OK. Would be way clearer with --ff-only there [06:14:18.0000] <Domenic_> sure, sounds good. [06:16:02.0000] <annevk> "pr: cannot open 12, No such file or directory" [06:17:27.0000] <Domenic_> Hmm. source ~/.bashrc ? [06:17:52.0000] <Domenic_> I guess I haven't tested this on non-Windows systems... [06:21:42.0000] <Domenic_> oh there's already a /usr/bin/pr on my Mac [06:21:59.0000] <Domenic_> so a different name i guess [06:23:26.0000] <annevk> I feel like such a noob. I changed the name, but it still does not work [06:23:56.0000] <darobin> annevk: did you remember to source your bashrc in that console? [06:24:29.0000] <annevk> Oh, Mac uses .profile? [06:24:33.0000] <annevk> http://apple.stackexchange.com/questions/12993/why-doesnt-bashrc-run-automatically [06:25:01.0000] <darobin> yes [06:25:13.0000] <annevk> that worked [06:25:20.0000] <darobin> you have to adapt "bashrc" to your platform :) [06:26:43.0000] <Domenic_> my platform is the highly-popular Git Bash on Windows, so I can see why this might have been confusing -_- [06:27:10.0000] <darobin> lol [06:31:07.0000] <annevk> Domenic_: ta [06:44:31.0000] <jgraham> Domenic_: I guess you are the wrong person to ask how you get bash and python working in the same shell in windows? I was looking at getting the web-platform-tests running but something required both git and python, which didn't seem to want to work together in the msys.git shell I had [06:45:11.0000] <Domenic_> jgraham: I think Git Bash + installing Python using the installer has worked fine for me in the past. [06:46:06.0000] <jgraham> Hmm, OK [06:46:12.0000] <jgraham> maybe I just had path problems then [06:46:29.0000] <jgraham> I think this was the point at which the VM locked my computer [06:46:41.0000] <jgraham> and I have been to scared to try since [06:47:45.0000] <annevk> jgraham: https://www.google.com/search?q=virtuelenv+windows ? [06:47:57.0000] <annevk> jgraham: even corrects typos [06:48:15.0000] <arv> annevk: dom.spec.whatwg.org is serving an empty document at the moment [06:48:55.0000] <annevk> hmm, the script works, I wonder if GH changed something [06:49:13.0000] <annevk> seems like GH fucked up my syncing [06:49:19.0000] <annevk> I just turned fetch blank [06:49:31.0000] <annevk> thanks arv, I'll investigate [06:51:56.0000] <annevk> well fuck, if I fetch https://raw.githubusercontent.com/whatwg/fetch/master/Overview.html it works, if I fetch it from the server, it doesn't [06:53:42.0000] <Domenic_> what does "fetch it from the server" mean? [06:53:45.0000] <Domenic_> curl? [06:54:27.0000] <annevk> curl -k [06:55:00.0000] <darobin> annevk: any chance you're using https://github.com/whatwg/fetch/raw/master/Overview.html instead of the link you posted above? [06:55:19.0000] <darobin> (GitHub recently changed that and is redirecting) [06:55:28.0000] <annevk> nope [06:55:58.0000] <annevk> but actually curl directly works, it's just curl -k x > y that fails [06:56:06.0000] <annevk> within a shell script [06:56:41.0000] <darobin> does the script fail when invoked directly, or only under cron? [06:57:20.0000] <darobin> annevk: can you try using -o output-file.html instead of > ? [06:57:49.0000] <annevk> darobin: it doesn't necessarily fail, it just downloads no data [06:58:05.0000] <darobin> annevk: by fail I meant not do what you want [06:59:10.0000] <annevk> -o doesn't make a difference [06:59:14.0000] <annevk> nothing in the error log [06:59:17.0000] <darobin> mmmmm [06:59:37.0000] <darobin> does that curl work with anything else? [06:59:53.0000] <zewt> even if you isolate it? not like there's any difference to curl... [07:00:04.0000] <darobin> can you maybe paste what you're using? [07:00:22.0000] <annevk> #!/bin/sh [07:00:22.0000] <annevk> echo "Content-type: text/plain" [07:00:23.0000] <annevk> echo "" [07:00:24.0000] <annevk> curl -k https://raw.github.com/whatwg/fetch/master/Overview.html > index.html [07:00:26.0000] <annevk> echo "teehee" [07:00:57.0000] <Domenic_> i think you need -l to follow redirects [07:01:01.0000] <Domenic_> raw.github.com redirects these days [07:01:12.0000] <annevk> running curl outside the shell script works fine [07:01:13.0000] <Domenic_> sorry -L [07:01:46.0000] <Domenic_> i am surprised that it works outside. but unsurprised that it fails inside. [07:02:33.0000] <zewt> that curl (without -k) works fine for me [07:03:02.0000] <zewt> there's no redirect [07:03:14.0000] <annevk> adding -L helps [07:03:21.0000] <zewt> (with the githubusercontent URL earlier) [07:03:42.0000] <zewt> err, confused--the url you pasted in that script != the well fuck URL [07:03:54.0000] <annevk> yup [07:04:09.0000] <annevk> I wonder what is stable these days [07:04:32.0000] <Domenic_> it's frustrating they didn't have a switchover period [07:04:57.0000] <zewt> seems like a bug that curl doesn't follow redirects by default [07:09:30.0000] <annevk> arv: thanks fixed [07:09:38.0000] <annevk> Domenic_: looks like they changed URLs twice already [07:09:44.0000] <arv> Thanks annevk [07:09:58.0000] <annevk> Domenic_: some of my scripts had -L and are fetching from github.com and not raw.github.com [07:10:12.0000] <Domenic_> ah yeah back in the day [07:10:31.0000] <annevk> I added -L, removed -k as that seems bad, and have not updated URLs; I should probably write a more generic script to handle all this at some point [07:10:54.0000] <zewt> were you expecting a stable web api D: [07:12:03.0000] <annevk> My experience writing web services is generally bad [07:12:09.0000] <annevk> And I'm doing the most trivial of things [07:23:15.0000] <annevk> MikeSmith: autocomplete for cc list is awesome, thanks for adding [07:27:08.0000] <annevk> foolip: hey, yt? [07:27:18.0000] <annevk> foolip: are you still maintaining the splitter script? [07:29:10.0000] <annevk> Okay, so I think the fix is to replace .//*[@class="head"] with something that does subscring matching [07:29:18.0000] <annevk> XPath has no support for classes so... [07:30:27.0000] <annevk> Hmm [tokenize(@class,'\s+')='atag'] [07:30:31.0000] <annevk> But that's XPath 2.0 [07:31:00.0000] <annevk> Otherwise contains(@class,'atag') which has obvious flaws but might be good enough [07:33:14.0000] <annevk> contains() is not even supported here [07:33:19.0000] <annevk> LOL standards [07:34:42.0000] <jgraham> Which XPath lib are you using? [07:35:35.0000] <annevk> I have no idea [07:35:53.0000] <annevk> I'm trying to debug this over ssh in nano :/ [07:36:42.0000] <annevk> jgraham: etree? [07:38:55.0000] <annevk> jgraham: https://docs.python.org/2/library/xml.etree.elementtree.html#xml.etree.ElementTree.Element.find afaict [07:42:03.0000] <jgraham> annevk: Element.find != xpath [07:42:16.0000] <jgraham> It's like a very crappy sub-xpath thing [07:42:31.0000] <annevk> jgraham: https://code.google.com/p/html5/source/browse/trunk/spec-splitter/spec-splitter.py#94 I see [07:42:45.0000] <annevk> I'll repeat: LOL standards [07:43:30.0000] <jgraham> annevk: You should be able to use .xpath I think [07:44:03.0000] <jgraham> http://lxml.de/xpathxslt.html#xpath [07:45:04.0000] <foolip> annevk: oh hi [07:45:18.0000] <jgraham> http://xmlsoft.org/html/libxml-xpathInternals.html should tell you which functions are implemented [07:45:25.0000] <foolip> annevk: I guess I don't, I haven't had cause to fiddle with it [07:46:10.0000] <annevk> jgraham: I replaced it with //header [07:46:30.0000] <annevk> jgraham: seems to work [07:46:38.0000] <jgraham> OK [07:48:04.0000] <annevk> MikeSmith: do you have interest in moving spec-splitter to github.com? [07:48:12.0000] <annevk> Hixie: are you planning on rewriting the spec-splitter? [07:55:53.0000] <annevk> foolip: ok, thanks [08:23:06.0000] <annevk> Maybe we should call it "unicodelatin1" [08:23:32.0000] <annevk> Unicode actually calls that block Latin-1 Supplement, so not too far off [08:29:36.0000] <SimonSapin> annevk: fwiw, cssselect implements class selectors as [@class and contains(concat(' ', normalize-space(@class), ' '), 'head')] [08:30:36.0000] <annevk> SimonSapin: you don't have XPath 2.0? [08:30:44.0000] <SimonSapin> nope [08:30:47.0000] <SimonSapin> 1.0 [08:31:05.0000] <SimonSapin> that’s what libxml2 (used by lxml) implements [08:31:17.0000] <annevk> you should do ' head ', no? [08:31:40.0000] <SimonSapin> ah, yes [08:31:53.0000] <annevk> XPath is weird [08:32:00.0000] <SimonSapin> https://github.com/SimonSapin/cssselect/blob/master/cssselect/xpath.py#L532-L534 [08:32:33.0000] <SimonSapin> translating Selectors to XPath is an horrible, horrible idea [08:33:25.0000] <annevk> I think I can agree to that without having attempted it [08:33:27.0000] <SimonSapin> and does not actually work (some Selectors are impossible to express in XPath AFAICT.) [08:34:10.0000] <annevk> Well everything that requires observing state would require some newer form of XPath that allows custom functions [08:34:30.0000] <SimonSapin> observing state? [08:34:41.0000] <annevk> :hover [08:35:00.0000] <jgraham> XPath does allow custom functions, no? [08:35:08.0000] <jgraham> Or is that only XPath 2? [08:35:15.0000] <SimonSapin> Yeah, no. Don’t even go down that road. [08:35:40.0000] <jgraham> Which road? The custom functions road? [08:35:42.0000] <SimonSapin> :hover just never matches in cssselect [08:35:45.0000] <jgraham> Or :hover? [08:35:50.0000] <MikeSmith> annevk: yeah we should move the splitter source to github [08:35:51.0000] <SimonSapin> jgraham: implementing :hover in XPath [08:36:02.0000] <jgraham> Right, that would be weird [08:36:23.0000] <jgraham> It depends on a non-obsevable bit that can't be set in most implementations [08:36:25.0000] <annevk> MikeSmith: if you still have that Google repo locally, might be worth deleting the stuff we already moved, to make it more clear it's no longer being maintained there [08:36:55.0000] <MikeSmith> annevk: ok [08:39:18.0000] <dglazkov> good morning, Whatwg! [08:49:04.0000] <JonathanNeal> Hixie: are alts not being respected in document outline for things like <h1><img alt="Title" src="path/to/image.jpg"></h1> ? [08:50:02.0000] <Domenic_> Oooh good question [08:55:43.0000] <jgraham> JonathanNeal: The spec doesn't say anything about that [08:56:06.0000] <jgraham> the spec says that the <h1> element is the heading element for that section [08:56:19.0000] <jgraham> It doesn't say how to extract the text content of the heading, afaict [08:59:10.0000] <JonathanNeal> Okay, was told there might be an issue, which is why I asked. So, when a heading contains <img alt="Title">, what do browsers do with the <img> presently? What (if any) document tells them to use the alt attribute? [09:00:42.0000] <SteveF> JonathonNeal: browsers don't do anything, javascript implementations and http://validator.w3.org/nu/ don't pick up the alt text [09:00:45.0000] <jgraham> As far as I know there isn't any document that tells them to use alt. Or to use the .textContent of the heading [09:01:17.0000] <jgraham> SteveF: Seems like a good time to submit some PRs [09:01:24.0000] <annevk> JonathanNeal: you should probably file a bug, I think there's an XXX marker in the specification close to the definition "text content" but I might misremember [09:01:36.0000] <SteveF> example http://validator.w3.org/nu/?showoutline=yes&doc=http%3A%2F%2Fwww.apple.com%2Fiphone%2Faccessories%2F [09:02:14.0000] <SteveF> jgraham: raise a bug more likely [09:04:48.0000] <SteveF> jgraham: "As far as I know there isn't any document that tells them to use alt. Or to use the .textContent of the heading" right, which is suboptimal [09:05:09.0000] <JonathanNeal> annevk: thanks. This is probably my first time filing a bug, so I apologize for its messiness https://www.w3.org/Bugs/Public/show_bug.cgi?id=25493 [09:11:27.0000] <annevk> Domenic_: if JavaScript had immutable arrays, people would use them [09:11:39.0000] <Domenic_> annevk: maybe. maybe not. [09:11:47.0000] <Domenic_> the language is *so* mutable [09:12:35.0000] <Domenic_> tbh I think frozen arrays are cool but I am mostly echoing wycats___ and dherman. [09:12:38.0000] <jgraham> FWIW it looks like the right patch in validator.nu is to add a switch in https://github.com/validator/validator/blob/master/src/nu/validator/servlet/OutlineBuildingXMLReaderWrapper.java#L427 that checks inElement and in case HEADING_CONTENT_ELEMENT checks for a HTML img element with a alt attribute and, if one exists, adds its characters to currentSection.getHeadingTextBuilder() [09:12:59.0000] <Domenic_> and i see their point about frozen sucking in general and so the fact that it works by accident on arrays not meaning we should use it. [09:13:26.0000] <annevk> Well, I agree that frozen is weird [09:14:07.0000] <annevk> I wonder if Event.path is always a set [09:14:28.0000] <annevk> Or if shadow DOM makes it a graph, I guess it's still a set [09:15:25.0000] <zewt> python has immutable arrays (tuples), but the only reason I really use them is convention [09:15:45.0000] <zewt> (i think they're faster to create too, but i rarely write python for performance) [09:20:39.0000] <MikeSmith> jgraham: yeah that sounds right [09:37:08.0000] <annevk> So the main problem with the current request concept setup is that there's a bunch of it describing a request, and there's a bunch of it that is processing input to build up more of the request [09:38:02.0000] <annevk> E.g. "author headers" is input for modifying the headers of a request per a given set of rules [09:38:24.0000] <Domenic_> TabAtkins: haha wow that old grid spec exceeded my expectations when I read "barely translated C++ code" [09:39:08.0000] <TabAtkins> Now imagine fantasai and I poring over it for two hours, scribbling notes on a whiteboard next to us, then each of going to separate desks to translate it into our own words, so that later we can compare algos and make sure we got things right. [09:43:03.0000] <TabAtkins> Domenic_: You can still see the intermediate state of the more complicated part of the algo at http://dev.w3.org/csswg/css-grid/#translated-algo , where I tried a more direct translation into English while fantasai tried a more idiomatic translation (which turned out to be wrong in several details, which is why we did separate passes). [10:42:58.0000] <annevk> wycats___: feel like we've gone full circle, nobody is surprised I think [10:43:04.0000] <annevk> wycats___: just annoyed [11:20:54.0000] <annevk> TabAtkins: you're hilarious [11:21:01.0000] <annevk> https://twitter.com/tabatkins/status/460840981185581056 [11:21:19.0000] <TabAtkins> ^_^ [11:22:25.0000] <Domenic_> I really liked https://twitter.com/tabatkins/status/460839870995263488 [11:23:25.0000] <annevk> heh [11:25:14.0000] <Domenic_> Do CGs get the same patent commitments as e.g. webapps? Seems doubtful. [11:26:01.0000] <TabAtkins> When you publish a FSD, you get sign-offs from everyone in the CG, and that's equivalent in power to the sign-offs we get by default when publishing through normal W3C process. [11:26:31.0000] <Domenic_> so if the CG membership was as large as webapps then it would be equivalent. [11:26:55.0000] <annevk> The main problem is that CG is opt-in and WGs are opt-out [11:26:57.0000] <TabAtkins> Yeah. [11:27:38.0000] <annevk> Note though that none of the plagiarized work made it to REC and so none of it is protected [11:28:26.0000] <Domenic_> Yeah. [11:28:29.0000] <Domenic_> That's the worst part. [11:32:08.0000] <annevk> I think I got more Twitter notifications in the past hour than in the rest of the year [12:59:44.0000] <estellevw> question on http://www.whatwg.org/specs/web-apps/current-work/multipage/states-of-the-type-attribute.html#range-state-(type=range) - multiple attribute. [12:59:59.0000] <estellevw> example code [13:00:00.0000] <estellevw> <input type=range min=0 max=24 value=0,24 step=1.0 ... [13:00:01.0000] <estellevw> oninput="o1.value = valueLow + ':00'; o2.value = valueHigh + ':00'"> [13:00:06.0000] <estellevw> doesn't have the multiple attribute set [13:00:31.0000] <estellevw> is it necessary to include multiple, or does the comma in the value suffice [13:00:31.0000] <estellevw> ? [13:00:59.0000] <Domenic_> good question... [13:01:04.0000] <estellevw> also, any implementers? [13:01:05.0000] <Domenic_> looks like a spec bug on first glance [13:01:27.0000] <annevk> estellevw: spec bug [13:01:42.0000] <annevk> estellevw: although it does say ... so you could assume multiple to be specified there... [13:04:10.0000] <estellevw> I did assume it was a spec example bug, but wanted to be sure. [13:25:55.0000] <Ms2ger> "We will be having our scheduled DOM L3 Events call tomorrow (October 1st)." [13:58:09.0000] <Ms2ger> "I was going to make an XSLT markup joke, but then I remembered that Twitter only allows 140 characters." [13:58:14.0000] <Ms2ger> Eric Meyer++ [14:04:17.0000] <IZh> http://html5.validator.nu/?doc=http%3A%2F%2F192.168.0.1%2F [14:04:54.0000] <IZh> It seems that validator doesn't check for private networks. [14:05:27.0000] <IZh> It could be used to guess web-servers on nearby local PCs. [14:14:12.0000] <IZh> http://html5.validator.nu/?doc=http%3A%2F%2Flocalhost%3A65536 [14:14:20.0000] <IZh> Intetnal error. [14:37:40.0000] <Hixie> annevk: eventually, but not any time soon [14:41:21.0000] <Hixie> zcorpan: top line goes into summary, second and subsequent lines into the comment, iirc [14:57:14.0000] <SamB_> er, as I understand it you need to leave a blank line after the summary [15:04:40.0000] <Hixie> on what? [15:05:21.0000] <SamB> Hixie: or some things think the next line is also summary [15:05:32.0000] <Hixie> i'm talking about the bug form [15:05:37.0000] <SamB> oh! [15:05:58.0000] <SamB> /me was thinking git ... [15:06:32.0000] <SamB> what bug form is that? I thought bugzilla had separate fields for each of those [15:10:33.0000] <Hixie> the one on the spec [15:10:37.0000] <Hixie> html spec [15:11:45.0000] <SamB> ah [15:12:47.0000] <SamB> /me hasn't looked at that for a few days; he's been using an even older computer than usual because the fan on his usual one had gotten WAAAY too noisy and he was waiting for a replacement in the mail ... [15:12:59.0000] <Hixie> heh [15:21:21.0000] <Domenic_> 500 on http://html5.org/tools/web-apps-tracker?from=8582&to=8583 [15:35:10.0000] <Domenic_> (fixed itself) [15:46:33.0000] <Hixie> umm.... [15:46:34.0000] <Hixie> hey anne [15:46:41.0000] <Hixie> the tracker is missing some entries :-) [15:47:18.0000] <Hixie> e.g. r8580 2014-04-29 [18:21:08.0000] <zewt> guh @ urllib.urlencode vomitting when passed unicode [19:08:43.0000] <MikeSmith> cabanier: even with the canvas.focusring.enabled pref set in FF, we still have tests that fail http://w3c.github.io/test-results/2dcontext/complete-fails.html [19:09:23.0000] <MikeSmith> oh hmm [19:09:33.0000] <MikeSmith> maybe not http://www.w3c-test.org/2dcontext/drawing-paths-to-the-canvas/drawFocusIfNeeded_001.html [19:46:08.0000] <MikeSmith> cabanier: regenerated http://w3c.github.io/test-results/2dcontext/less-than-2.html [19:46:20.0000] <MikeSmith> and http://w3c.github.io/test-results/2dcontext/complete-fails.html now shows zero [19:58:45.0000] <cabanier> MikeSmith: thanks! I'll see if I can fix some of the FF issues [22:04:43.0000] <MikeSmith> gsnedders: that "All the Dashes in Web Typography" thing is missing ー [22:05:54.0000] <MikeSmith> U+30FC [03:00:09.0000] <annevk> Hixie: http://html5.org/r/8580 seems to be there? [03:01:27.0000] <annevk> Hixie: note that I only track changes to /webapps/source [04:00:24.0000] <annevk> JakeA / jungkees: do we have diagrams somewhere of the various fetch flows? [04:01:06.0000] <annevk> JakeA / jungkees: in particular I wonder what the best way is to define the event.default() semantic [04:02:45.0000] <annevk> <img> -> invokes fetch -> invokes SW -> invokes default -> invokes fetch recursively but with "use SW flag" unset or some such? [04:03:50.0000] <jgraham> /me gets annevk a copy of dot [04:10:20.0000] <JakeA> annevk: Do you specifically want diagrams or would rough algorithms fo? [04:10:22.0000] <JakeA> do* [04:16:08.0000] <annevk> JakeA: rough works I suppose [04:16:52.0000] <annevk> JakeA: I had this idea of making the request concept much simpler, but then I realized that is probably not possible because we have this recursion thing going on, so at least some of the data needs to stay preserved [04:22:47.0000] <hsivonen> annevk: do we have any single-byte encoding (other than x-user-defined) that defines mappings for the entire upper half? [04:23:20.0000] <annevk> hsivonen: we need "unicodelatin1" or some such that will have that [04:23:28.0000] <hsivonen> non-control mappings that is [04:23:33.0000] <annevk> hsivonen: for networking [04:23:35.0000] <hsivonen> annevk: that won't help me now [04:23:47.0000] <hsivonen> annevk: I'm thinking of an encoding I could use with gedit today [04:24:06.0000] <hsivonen> the best I have so far is windows-1251, which has one undefined code point [04:24:27.0000] <hsivonen> so I have to be really unlucky not to be able to roundtrip stuff with that one [04:24:41.0000] <JakeA> annevk: I'll write an algo for event.default() later this week (I'm booked up today/tomorrow) [04:24:48.0000] <hsivonen> use case: editing patches that as a whole are both invalid UTF-8 and invalid windows-1252 [04:25:12.0000] <annevk> JakeA: I'll do something as well [04:25:53.0000] <annevk> hsivonen: http://encoding.spec.whatwg.org/index-x-mac-cyrillic.txt looks complete [04:26:06.0000] <hsivonen> annevk: cool thanks [04:26:42.0000] <hsivonen> MacRoman, too, actually [04:26:43.0000] <hsivonen> wow [04:27:11.0000] <hsivonen> it's been so long since I've used MacRoman for something useful that I've forgotten that Mac encodings use all the available space [04:27:53.0000] <annevk> macintosh does have a PUA mapping [04:27:59.0000] <hsivonen> aargh. gedit doesn't support Mac encodings [04:28:18.0000] <hsivonen> it has stuff like UTF-32 and TCVN, but no Mac encodings at all [04:29:31.0000] <hsivonen> maybe I should use VISCII for this use case [04:33:38.0000] <SimonSapin> hsivonen: what’s the use case? [04:34:39.0000] <zcorpan> SimonSapin: http://krijnhoetmer.nl/irc-logs/whatwg/20140429#l-299 [04:55:59.0000] <zcorpan> jgraham: how do i take the tests here https://dvcs.w3.org/hg/quirks-mode/ and move them to web-platform-tests retaining history? [04:57:55.0000] <zcorpan> it seems the w3c haven't forked the quirks mode spec yet [04:58:13.0000] <MikeSmith> /me adds that to his TODO list [04:59:08.0000] <jgraham> zcorpan: Probably git-hg clone to create a git repo and then git subtree add to add it to wpt [04:59:32.0000] <jgraham> Going to experiment with the London transport network now, back online in a bit [04:59:48.0000] <zcorpan> jgraham: ok thx [05:00:52.0000] <zcorpan> MikeSmith: incentive pays for the number of forked specs eh? [05:01:06.0000] <zcorpan> that's why there were 27 canvas specs [05:01:15.0000] <zcorpan> /me figured it out [05:01:21.0000] <MikeSmith> perceptive [05:01:51.0000] <MikeSmith> how to migrate a mercurial repo to subdirectory of an existing git repo and retain the history) [05:02:06.0000] <MikeSmith> oofs [05:02:42.0000] <MikeSmith> zcorpan: I meant to say, what jgraham said [05:02:47.0000] <MikeSmith> about migrating that repo [05:03:14.0000] <MikeSmith> that's what I did to add the conformance-checkers/ stuff [05:05:15.0000] <zcorpan> MikeSmith: ok great [05:08:40.0000] <zcorpan> pointer to git-hg ? [05:16:43.0000] <MikeSmith> http://hg-git.github.io/ [05:17:09.0000] <annevk> Oh my, I was just wondering whether my seatbelt was fastened while sitting in the office [05:19:20.0000] <Ms2ger> Prepare for liftoff... [05:26:22.0000] <zcorpan> MikeSmith: thanks. i got an error when trying to install it :-( [05:26:54.0000] <zcorpan> Running dulwich-0.9.6/setup.py -q bdist_egg --dist-dir /tmp/easy_install-BsoHD7/dulwich-0.9.6/egg-dist-tmp-wHtFoJ [05:26:54.0000] <zcorpan> clang: error: unknown argument: '-mno-fused-madd' [-Wunused-command-line-argument-hard-error-in-future] [05:26:54.0000] <zcorpan> clang: note: this will be a hard error (cannot be downgraded to a warning) in the future [05:26:54.0000] <zcorpan> error: Setup script exited with error: command 'cc' failed with exit status 1 [05:27:07.0000] <zcorpan> /me gotta go [05:35:52.0000] <annevk> http://www.w3.org/TR/2014/WD-tracking-dnt-20140424/#exceptions-javascript-api wtf [05:41:10.0000] <Ms2ger> "Push Server protocol: Mozilla and Google to kick-off a new draft at the IETF to standardize it." [05:41:20.0000] <Ms2ger> Because that went so well the last time? [05:42:25.0000] <annevk> Ms2ger: I don't understand that either [05:43:33.0000] <annevk> Ms2ger: both companies hired too much IETF apologists I suspect [05:43:52.0000] <Ms2ger> That may be the case [05:47:33.0000] <Ms2ger> Heh [05:48:14.0000] <Ms2ger> Apparently w3.org thinks I'm in a "Test Framework Developers group" with URL http://w3c-test.org/framework/ [05:48:44.0000] <Ms2ger> MikeSmith, you can throw me out of that :) [05:49:59.0000] <MikeSmith> Ms2ger: oh man I should just remove that whole group [05:50:30.0000] <Ms2ger> Also, Testsuite Maintainers group [05:51:20.0000] <annevk> DNT :-( [05:52:05.0000] <annevk> Why the fuck does the W3C still allow groups to just go and invent some weird APIs without oversight? http://www.w3.org/2011/tracking-protection/drafts/tracking-dnt.html [05:52:19.0000] <annevk> Isn't this what "The Team" is supposed to be good at. Helping out? [05:53:01.0000] <annevk> Sorry MikeSmith, bit frustrated [05:53:43.0000] <MikeSmith> preaching to the choir [05:54:03.0000] <MikeSmith> well I guess I shouldn't say that [06:06:03.0000] <MikeSmith> and I didn't even know the DNT spec introduced any APIs. but I can't blame anybody except myself for that, since all it requires is me to actually read the spec [06:06:22.0000] <beverloo> annevk, tricking me into giving feedback for DNT...?! ;-) [06:06:52.0000] <MikeSmith> good god almighty StoreSiteSpecificExceptionPropertyBag [06:08:06.0000] <MikeSmith> annevk: fwiw I also never got any heads up about that stuff from the team contacts for the group [06:08:10.0000] <MikeSmith> not that I remember [06:08:18.0000] <annevk> beverloo: it could use it :-) [06:09:03.0000] <annevk> There ought to be a meme for "Don't know what the fuck I'm doing. Let's ask for feedback during Last Call" [06:09:20.0000] <Ms2ger> Does the CSSWG have a logo we could use? [06:09:28.0000] <MikeSmith> anyway I'm pretty sure I would remember something mentionining "hey btw we're adding StoreSiteSpecificExceptionPropertyBagW [06:09:49.0000] <MikeSmith> *somebody mentioning to m [06:09:52.0000] <MikeSmith> *me [06:10:19.0000] <Ms2ger> /me wonders if MikeSmith got distracted from those silly groups [06:11:54.0000] <MikeSmith> my single interest in the tracking group is pretty much just reading Roy's messages to the WG list, and the responses to him, for the entertainment value [06:12:15.0000] <MikeSmith> anyway I blame the TAG [06:12:53.0000] <Ms2ger> Ooooh, more Hixie on www-archive [06:12:55.0000] <annevk> Ms2ger: http://www.w3.org/Style/Woolly/woolly-icon.svg [06:13:19.0000] <Ms2ger> Good old Woolly [06:57:49.0000] <SimonSapin> Ms2ger: shepazu talked about making a new one [06:59:22.0000] <annevk> SimonSapin: you suggested inlining "redirect count" and such at one point, right? [06:59:34.0000] <SimonSapin> annevk: context? [06:59:37.0000] <annevk> SimonSapin: I just looked into doing that, and it seems to require quite a bit more text than the current solution [06:59:45.0000] <annevk> SimonSapin: http://fetch.spec.whatwg.org/ [06:59:58.0000] <SimonSapin> annevk: that was probably someone else [07:02:18.0000] <gsnedders> Someone should write a wiki page or something with the varying problems of various standards groups [07:05:04.0000] <annevk> There's http://wiki.whatwg.org/wiki/IETF [07:05:16.0000] <MikeSmith> Diagnostic and Statistical Manual of Mental^W Standard-Group Disorders [07:12:24.0000] <Domenic_> Hixie: what's an FSA draft? [07:12:53.0000] <annevk> Domenic_: http://www.w3.org/community/about/agreements/final/ [07:13:22.0000] <Domenic_> Ah, so Hixie is saying that WHATCG snapshots should be published more often? [07:13:53.0000] <annevk> We haven't really done it yet [07:14:38.0000] <annevk> See http://lists.w3.org/Archives/Public/www-archive/2014Apr/0039.html for reasons [07:14:43.0000] <Domenic_> That seems like a weak position to argue from :-/ [07:15:44.0000] <annevk> W3C has no REC of plagiarized material, we have no FSA of original work [07:16:27.0000] <annevk> Agreed that it's not ideal, I'm mostly trying to find a way to simplify the fetch algorithm [07:16:58.0000] <Domenic_> haha [08:02:09.0000] <Domenic_> Hmm is Firefox actually shipping any promise-based APIs yet? Or did they just ship Promises to be awesome? [08:26:59.0000] <Hixie> annevk: weird, it works now [08:27:16.0000] <annevk> Hixie: maybe DreamHost tried turning it off and on again [08:28:50.0000] <Hixie> heh [08:29:00.0000] <Hixie> i was getting the same as now just with some missing lines on the tracker index [08:29:15.0000] <annevk> oh [08:30:45.0000] <SamB> so is the "leave your sense of logic at the door" referring to the expectation that things will be logical? (Otherwise it seems like a really lousy suggestion.) [08:31:13.0000] <Hixie> basically [08:31:14.0000] <jgraham> SamB: Yes [08:31:29.0000] <Domenic_> Why do Asian websites use <marquee> so much? I've heard this fact but never understood why. [08:31:50.0000] <jgraham> People often come here and say things like "the platform would make more sense if X" where X is some thing that we can't do for compat reasons [08:32:19.0000] <jgraham> And when I say "people" I mean "everyone up to and including Hixie" [08:32:37.0000] <jgraham> So it is helpful to be able to say "see /topic" [08:32:50.0000] <jgraham> Particularly when it's Hixie or someone else that should know better ;) [08:33:03.0000] <Hixie> jgraham++ :-) [08:35:13.0000] <dglazkov> good morning, Whatwg! [08:36:31.0000] <annevk> Domenic_: I only have anecdotal data, but sites in Asia tend to be a lot more busy [08:37:24.0000] <annevk> Domenic_: people in that part of the world might enjoy that more, I've no idea [08:37:48.0000] <SamB> jgraham: yeah, that's what I figured [08:38:12.0000] <SamB> that it referred to the crazy stuff that must be done for compat [08:38:54.0000] <SamB> ... so what did Asia have to say about the death of <blink>? [09:07:01.0000] <cwilso> Domenic_: <marquee> was a way to get vertical text layout in IE, well before vertical text layout was supported. [09:07:23.0000] <Domenic_> cwilso: ah nice, that makes a lot of sense! [09:08:20.0000] <cwilso> A very twisted hacky kind of sense, but yes. :) [09:09:18.0000] <cwilso> (redacted braindump of "@fonts", vertical-text fonts that had each glyph rotated and the baseline on the side, because that's how vertical text used to be done.) [09:10:25.0000] <annevk> I'm somewhat surprised we still don't have vertical text if it's that important [09:11:24.0000] <SimonSapin> annevk: we’re working on it, but It’s Hard. [09:11:45.0000] <SimonSapin> feel free to send feedback on http://dev.w3.org/csswg/css-writing-modes/ ;) [09:13:38.0000] <Ms2ger> Or implement it in Servo :) [09:27:13.0000] <SimonSapin> Ms2ger: I hear that part is up to me [09:54:10.0000] <Hixie> TabAtkins: is there a spec for 'display-box' yet that has browser adoption of any kind? [09:57:10.0000] <annevk> Hixie: http://dev.w3.org/csswg/css-display/#the-display-box [09:57:20.0000] <annevk> Hixie: haven't seen intent to implement for it anywhere [10:03:03.0000] <cwilso> annevk: yeah, vertical text the right way - or at least, a different-and-much-less-wrong way - has been cycling around for quite a while. IE did vertical text ca IE5 or so, if memory serves, using CSS writing-mode (I think, it's been a loooong time.) [10:03:32.0000] <cwilso> not necessarily any relation to current work SimonSapin mentioned. :) [10:04:46.0000] <SimonSapin> We’re working on display-box: https://bugzilla.mozilla.org/show_bug.cgi?id=907396 [10:25:49.0000] <TabAtkins> Hixie: Spec is what Anne pointed out. No implementations yet, though Blink is leaning toward one. [10:26:01.0000] <Hixie> k thanks [10:26:42.0000] <MikeSmith> Domenic_: a lot of Japanese mobile sites used marquee effect back in the WAP days. not using <marquee> but instead by some WAPish thing which name I forget. I wonder if if maybe they liked the effect so much that it ended up getting back-migrated to desktop sites. [10:39:45.0000] <Hixie> bholley: any update on https://www.w3.org/Bugs/Public/show_bug.cgi?id=20701 ? [10:40:32.0000] <bholley> Hixie: new Window bindings got delayed slightly - they might even be done this week. I was on jury duty yesterday, so I missed the status meeting [10:40:39.0000] <Hixie> k [10:40:51.0000] <bholley> Hixie: but yeah, just give it another couple of weeks [10:40:57.0000] <Hixie> awesome [10:41:00.0000] <bholley> Hixie: as soon as that stuff lands I'll land the new XO behavior [10:47:23.0000] <Ms2ger> bholley, I don't think there was one [10:47:38.0000] <bholley> Ms2ger: do you know the status of the Window bindings? [10:47:53.0000] <Ms2ger> Nope [10:48:04.0000] <Ms2ger> I asked jst to relay it to me if he figured out [10:52:45.0000] <Ms2ger> Oh, the buttons in the HTML spec don't look so awful now [10:54:14.0000] <Hixie> really? i think they're still pretty terrible [10:54:22.0000] <Hixie> i just don't know what to do that'd be better [10:56:04.0000] <Ms2ger> They looked more awful before :) [10:56:07.0000] <Hixie> looks particularly bad at the top of the PDF version for some reason [11:19:46.0000] <cabanier> MikeSmith: ping [11:35:18.0000] <cabanier> MikeSmith: http://www.w3c-test.org/2dcontext/fill-and-stroke-styles/2d.pattern.image.broken.html is supposed to throw an exception but the test is currently looking for a null return value [11:36:19.0000] <cabanier> MikeSmith: chrome throws an exception in my local testing but not for the test in your depot [11:36:36.0000] <IZh> IZh: Hi. What you think about swapping page title and page navigation of developer's version pages? [11:39:47.0000] <IZh> It is strange to me to scroll to the top of the page and do not see its title. [11:40:11.0000] <IZh> Currently the title is below the navigation. [11:40:27.0000] <SamB> that sounds like a good idea to me [11:40:30.0000] <SamB> Hixie: ^ [11:40:42.0000] <IZh> http://developers.whatwg.org/text-level-semantics.html [11:40:53.0000] <Hixie> benschwarz: ^ [11:41:11.0000] <SamB> oh, that version [11:41:12.0000] <SamB> sorry [11:41:17.0000] <SamB> /me forgot what "developers" meant [11:42:02.0000] <SamB> (confused it with "implementors" or something) [11:43:29.0000] <IZh> Also there are no links to Back and Top. Only Next. [11:43:55.0000] <IZh> I mean Back and Up. [12:02:48.0000] <Hixie> hsivonen: any news on the encoding stuff? [12:06:12.0000] <SamB> /me still thinks those link rels should be in the spec, even if they aren't currently used by much except old-school HTML generation tools -- they are GOOD rels which are probably never going to go away [12:07:21.0000] <Hixie> "never going to go away" implies that they're not already gone [12:07:38.0000] <Hixie> in fact it implies that at some point they arrived [12:07:45.0000] <Hixie> and it's not clear to me that they ever did in any meaningful sense [12:08:00.0000] <SamB> well, okay, so yeah, there is that [12:10:10.0000] <SamB> /me supposes if he really wants to help he should implement a toolbar that actually *works* that uses them ... [12:11:42.0000] <SamB> ... hmm, how come PDFs get all the fancy outline sidebars anyway ... [12:28:09.0000] <jcgregorio> https://github.com/google/canvas-5-polyfill [12:28:36.0000] <jcgregorio> I just launched that, feedback welcome :-) [13:02:17.0000] <cabanier> MikeSmith: http://www.w3c-test.org/2dcontext/fill-and-stroke-styles/2d.pattern.repeat.null.html is also testing an incorrect assumption. passing null to createPattern should not throw [13:03:33.0000] <cabanier> MikeSmith: the drawFocus tests will work on chrome canary with the pref turned on [13:03:53.0000] <Ms2ger> Oh really? [13:05:28.0000] <Ms2ger> Interesting [13:05:41.0000] <Ms2ger> PR? [13:44:37.0000] <cabanier> Ms2ger: were you talking to me? [13:45:17.0000] <Ms2ger> Yes [13:45:49.0000] <cabanier> :-) [13:46:27.0000] <cabanier> Ms2ger: are you saying createPattern should throw? [13:46:39.0000] <Ms2ger> Not anymore, no [13:46:46.0000] <cabanier> good [13:46:47.0000] <Ms2ger> So, want to fix the test? :) [13:47:01.0000] <Ms2ger> And add one for "null" as opposed to null; that should still throw [13:47:04.0000] <cabanier> yes. How do I do that? [13:47:44.0000] <Ms2ger> web-platform-tests in general or canvas tests in particular? [13:47:57.0000] <cabanier> the canvas tests [13:47:58.0000] <cabanier> brb [13:48:58.0000] <Ms2ger> Edit 2dcontext/tools/tests2d.yaml [13:49:06.0000] <Ms2ger> The format is pretty straightforward [13:49:20.0000] <Ms2ger> And then run 2dcontext/tools/gentest.py [14:37:13.0000] <IZh> It seems it is impossible to specify dates B.C. because of year must be positive. [14:38:20.0000] <IZh> I don't need it, but in some historic documents it is possible. [14:48:04.0000] <gsnedders> IZh: IIRC there was no use-case for being able to explicitly mark them up, given we rarely know dates [14:56:11.0000] <Hixie> yeah, if you go back more than a few 100 years, it gets realllllly complicated [14:56:18.0000] <Hixie> because e.g. our calendar didn't exist back then [14:56:56.0000] <gsnedders> Calendar is the least of the worries, really. [14:58:05.0000] <TabAtkins> Yeah, just be glad we ended up allowing Proleptic Gregorian. [15:03:42.0000] <IZh> Okay. :-) [15:07:05.0000] <IZh> What about colors? There are more and more monitors capable to display 30-bit colors. [15:07:20.0000] <Hixie> yeah [15:07:22.0000] <Domenic_> Wasn't there some smooth animating verison of scrollIntoView() being specced somewhere? [15:07:28.0000] <Hixie> color is hard [15:07:35.0000] <Hixie> too many places in the platform assume 32 bit colour [15:07:41.0000] <Hixie> (including alpha) [15:07:58.0000] <TabAtkins> IZh: Colors can be whatever depth you want. You're not limited to the hex notation. [15:08:09.0000] <IZh> There is often lack of support by modern OSes, but I believe it will be more wide supported soon. [15:08:09.0000] <TabAtkins> Whether the browser actually uses anything bigger than 32-bit is up to it. [15:08:13.0000] <Hixie> you are in a lot of places [15:08:33.0000] <Hixie> e.g. rgb(1,1,1) and #XXX are both limited to 32 bit [15:08:34.0000] <TabAtkins> Domenic_: It's in CSSOM-View, I believe. [15:08:46.0000] <TabAtkins> Hixie: *Some* syntaxes are. [15:08:52.0000] <TabAtkins> Other syntaxes aren't. [15:08:53.0000] <Hixie> and it underlies a lot of the assumptions [15:09:36.0000] <TabAtkins> ...like? [15:09:42.0000] <IZh> What syntax is suitable for 30-bit colors (10 bit per channel)? [15:09:44.0000] <Domenic_> TabAtkins: that was it, thanks. Nice spec. I assume no implementers? :P [15:09:56.0000] <Hixie> like, <input type=color> is a 32 bit color [15:10:01.0000] <Hixie> (well, 24 bit) [15:10:04.0000] <TabAtkins> IZh: Use rgb() with percentages, or use hsl(). [15:10:23.0000] <TabAtkins> Hixie: Yeah, it exposes the color as a hex string, so 8bits per. [15:11:02.0000] <Hixie> and canvas exposes colours using rgba(n,n,n,m) 0<=n<=255 [15:11:34.0000] <Hixie> i forget what cssom does [15:11:40.0000] <TabAtkins> So HTML makes a bunch of assumptions and only exposes colors as 8bit per. CSS doesn't. [15:12:51.0000] <Hixie> oh nobody mentioned CSS. CSS might allow it, sure. [15:13:34.0000] <Hixie> CSSOM serialises to 32bit also, it looks like [15:13:38.0000] <zewt> doubt we'll see greater than 8bpc color in (consumer) displays any time soon [15:14:03.0000] <Hixie> so same problem there [15:14:20.0000] <zewt> could be sort of interesting to do temporal dithering on a 24bpp display to get higher resolution color, heh [15:14:32.0000] <IZh> I think the best way is a percentages with enough precision to select particular value up to 16-bit per channel -- the highest depth possible in some image formats [15:15:13.0000] <TabAtkins> Hixie: Where? [15:15:14.0000] <IZh> zewt: I have 30-bit capable display. ;-) [15:15:20.0000] <zewt> (which is what my plasma TV seems to do) [15:15:39.0000] <Hixie> TabAtkins: http://dev.w3.org/csswg/cssom/#serializing-css-values [15:15:55.0000] <zewt> "consumer", they're beyond rare heh [15:16:11.0000] <TabAtkins> rgb() no longer requires only integers in the N,N,N format. [15:16:49.0000] <IZh> zewt: not so rare as you think. :-) You can search for 30-bit colors [15:17:01.0000] <TabAtkins> (Got that fixed in Colors 4. Recent, so impls probably haven't caught up yet.) [15:17:08.0000] <zewt> they're precisely as rare as i think :) [15:17:35.0000] <IZh> zewt: the problem now is that only few apps like Photoshop can use it :-) [15:17:55.0000] <Hixie> TabAtkins: ah. i hope that doesn't break any scripts relying on the serialisation to be rounded off to integers. [15:19:01.0000] <TabAtkins> Me too! [15:19:11.0000] <zewt> even photoshop is bad at it and turns lots of stuff off [15:19:13.0000] <TabAtkins> Because I'm mainly trying to fix *my* scripts that do naive math and forgte to floor things to integers. [15:19:57.0000] <TabAtkins> Canvas is extra-bad, since accidentally setting thing to "rgb(127.5, 127.5, 127.5)" gives you pure black. [15:20:03.0000] <TabAtkins> Very annoying to debug unless you learn to recognize it. [15:20:31.0000] <IZh> zewt: what you mean? I read that all is ok. But I have no time yet to play with it by myself. [15:22:19.0000] <zewt> i wonder if it would be useful to allow serializing colors to [0,255] floats to truncate the decimal place early, so (1/65535) can be output as something like 0.004 instead of 0.003921568627 [15:22:55.0000] <zewt> allow truncating to any value that gives the same result when converted to a 16-bit int, or something like that [15:23:20.0000] <TabAtkins> It's annoying that the divisor isn't 256, as that makes for nice pretty decimals. [15:23:31.0000] <TabAtkins> 10-bit color just has .25, .5, .75 in that case. [15:24:10.0000] <zewt> 1/256 is 0.00390625, which is beyond my ability to recognize as 1/power of two, heh [15:24:28.0000] <TabAtkins> The values are already scaled to 1/256, so that's just "1". [15:25:09.0000] <TabAtkins> The issue is that the divisor right now isn't 1/256, but 1/255, which has worse decimal serialization. [15:25:22.0000] <zewt> no, 1/256 (0.00390625) is the value you'd get for 1/65536 after converting it to [1,256] [15:25:35.0000] <TabAtkins> Oh, that's 16-bit channels. [15:25:41.0000] <TabAtkins> Sure. [15:25:49.0000] <TabAtkins> ([0,256]) [15:26:05.0000] <zewt> i'll pass on a 257-color mode :P [15:26:23.0000] <TabAtkins> Aw, come on. [15:26:27.0000] <zewt> IZh: many filters only work in 8-bit, and tons don't work in 32-bit [15:26:37.0000] <zewt> (they're probably gradually fixing them) [15:26:59.0000] <zewt> but the main point of 16-bit editing in photoshop is editing, not display [15:27:36.0000] <zewt> (which I've used many times; helps prevent artifacts cropping in when you're doing a lot of filtering) [15:28:12.0000] <IZh> zewt: 1. You can see raw image from your camera with more wider dynamic range. [15:28:46.0000] <zewt> 16-bit color doesn't give you more dynamic range, it gives you higher resolution in the same range [15:29:16.0000] <IZh> Agree. [15:29:21.0000] <zewt> do i have to [15:29:31.0000] <IZh> Finer gradients [15:29:51.0000] <zewt> one of photoshop's dumb things is that it doesn't (as far as I know) support dithering gradients [15:30:39.0000] <IZh> With 30-bit mode you don't need it. ;-) [15:30:50.0000] <zewt> not sure what you're trying to argue :P [15:31:09.0000] <IZh> Although you have a 4k display :-) [15:31:23.0000] <zewt> (no I don't) [15:31:36.0000] <IZh> Me too. Alas. [15:32:07.0000] <zewt> afk [15:33:40.0000] <astearns> zewt: the gradient tool options has a dither checkbox (never used it, though) [15:40:51.0000] <IZh> What about hex noration, where bits are "left-aligned"? [15:41:50.0000] <IZh> I mean that ff, ff4, ff40 will give the same value ff when converted to 8-bit. [15:42:22.0000] <IZh> I mean, most significant bit first [15:42:57.0000] <TabAtkins> IZh: What are you talking about? [15:43:04.0000] <IZh> Colors [15:43:24.0000] <TabAtkins> Yes. What are you talking about, with this discussion of left-aligned bits? [15:43:50.0000] <IZh> I propose new encoding :-) [15:43:55.0000] <IZh> Of colors [15:44:55.0000] <TabAtkins> Why? [15:45:41.0000] <IZh> To support different depth without very long floating point numbers. [15:46:21.0000] <cabanier> zewt: sure it does [15:46:29.0000] <cabanier> support dithering gradients [15:47:02.0000] <IZh> TabAtkins: Just an idea. [15:48:07.0000] <cabanier> IZh: is this just to reduce banding? nothing's stopping a browser from making that optimization [15:48:54.0000] <cabanier> IZh: colors don't have to be in high precision for that. Adobe apps have been doing that for 18 years [15:50:03.0000] <IZh> cabanier: I believe that sooner 30-bit displays will conquer the market [15:50:27.0000] <cabanier> ok [15:50:38.0000] <IZh> cabanier: So you will need to specify particular color. [15:50:46.0000] <cabanier> IZh: why? [15:52:00.0000] <IZh> cabanier: For image manipulation, for example [15:53:27.0000] <cabanier> IZh: with color managments, you will never get the actual color value on the display [15:55:08.0000] <IZh> cabanier: I think, this is orthogonal things [15:55:48.0000] <cabanier> IZh: 30 bit gives you more consistency over the color range so you don't lose precision [15:57:15.0000] <IZh> cabanier: It gives the ability to set, for example, rgb(1023, 1022, 1000) color [15:57:59.0000] <zewt> oh yeah, it's gradient layer styles that don't dither [15:59:53.0000] <cabanier> zewt: there's a dither button for gradient layer styles [16:00:36.0000] <zewt> ah, that's new [16:01:03.0000] <zewt> (i see it in CS6, i do a lot in CS4) [16:01:05.0000] <cabanier> Photoshop CS6 [16:01:07.0000] <cabanier> ah [16:08:25.0000] <cabanier> IZh: css color allows decimal points for color values. That should work for you, right? [16:12:34.0000] <IZh> cabanier: yes. The only thing is long syntax. 7/1024 = 0.0068359375 vs 013 [16:17:41.0000] <IZh> I mean 01C :-) [16:21:57.0000] <zewt> have browsers implemented high-color support in other systems (canvas, CSS colors, gradients and shadows, image rendering, ) ... well okay [16:26:02.0000] <cabanier> zewt: no :-) not even decent color support except maybe Safari [16:26:50.0000] <zewt> i presumed that, it was a leading question to "then lobby for that before asking for new features" :) 2014-04-30 [17:25:38.0000] <MikeSmith> cabanier: thanksーturned on the Chrome "Enable experimental canvas features" flag and re-generated the test report http://w3c.github.io/test-results/2dcontext/less-than-2.html [17:27:04.0000] <MikeSmith> cabanier: about http://www.w3c-test.org/2dcontext/fill-and-stroke-styles/2d.pattern.repeat.null.html & http://www.w3c-test.org/2dcontext/fill-and-stroke-styles/2d.pattern.image.broken.html ーthanks also, I'll fix those and open a PR [17:29:46.0000] <MikeSmith> cabanier: ah I see you opened a PR already [17:37:26.0000] <MikeSmith> cabanier: hmm that new http://web-platform.test:8000/2dcontext/fill-and-stroke-styles/2d.pattern.image.broken.html test fails too [17:37:48.0000] <MikeSmith> "that is not a DOMException INVALID_STATE_ERR: property "code" is equal to undefined, expected 11" [17:44:40.0000] <MikeSmith> cabanier: yeah seems that still nobody actually conforms to the spec on this yet, right? [17:44:44.0000] <MikeSmith> nobody throws [17:50:12.0000] <MikeSmith> cabanier: wait in the case where there is not image at a given URL, the image is not going to be in the "broken" state is it? it just remains "unavailable" [17:55:56.0000] <MikeSmith> an nah I see it does end up as "broken" [18:03:37.0000] <MikeSmith> cabanier: fyi https://bugzilla.mozilla.org/show_bug.cgi?id=405761 "drawImage should throw INVALID_STATE_ERR on invalid image, not NS_ERROR_NOT_AVAIL" [18:03:57.0000] <MikeSmith> which is not createPattern but I guess it's the same issue, as far as Firefox goes [19:28:45.0000] <cabanier> MikeSmith: It's weird. when I try the createPattern case with my own test file in chrome canary, it throws the right exception, but not in the test [19:29:08.0000] <cabanier> MikeSmith: I opened a firefox bug on it: https://bugzilla.mozilla.org/show_bug.cgi?id=1003380 [19:29:33.0000] <cabanier> MikeSmith: I got an r+ for the patch but the test is failing. Likely because the image is not in the broken state when the test runs [19:45:58.0000] <MikeSmith> /me looks at cabanier bug [19:49:29.0000] <MikeSmith> cabanier: sounds like you've got another Firefox bug to fix :-) by the time you're done you'll have re-written all the image-processing logic in Firefox. And by the time you're done with that, Hixie will have landed the image-processing changes in the spec and your changes will be out of date already :-) [19:49:47.0000] <cabanier> MikeSmith: I marked 405761 as a duplicate [19:49:53.0000] <MikeSmith> ok [19:49:57.0000] <cabanier> :-) [19:50:24.0000] <MikeSmith> so about the createPattern test, maybe there's a bug in the test [19:50:44.0000] <MikeSmith> but I dunno what it'd be [19:50:55.0000] <cabanier> MikeSmith: I don't know what's going on. I can't tell from stepping through. [19:51:06.0000] <cabanier> MikeSmith: it seems the image state is fragile [19:51:36.0000] <MikeSmith> sounds like [21:05:42.0000] <Hixie> MikeSmith: aw man, img loading. i don't suppose you want to spec that do you. [21:05:49.0000] <Hixie> /me is so lot looking forward to speccing that [21:05:54.0000] <Hixie> uh, not, not lot [21:19:07.0000] <MikeSmith> Hixie: I don't want to spec it :) [21:19:21.0000] <Hixie> bummer [21:19:47.0000] <cabanier> Hixie: didn't you already spec it? [21:20:07.0000] <cabanier> or is it not detailed enough? [21:20:13.0000] <Hixie> see recent mails to the list, i linked to the relevant bugs [21:22:43.0000] <MikeSmith> Hixie: I guess you probably don't necessarily _want_ to spec it either. But as usual if you don't, nobody else going to [21:22:56.0000] <Hixie> :-) [22:13:45.0000] <Hixie> woot, my parser is up to passing 5000 tests. [22:13:50.0000] <Hixie> no idea how many tests there are... [22:14:01.0000] <Hixie> and about 4000 of those seem to be entity tests... [22:14:05.0000] <Hixie> but anyway... [22:28:53.0000] <zcorpan> Hixie: html parser? [23:25:04.0000] <zcorpan> Hixie: now the spec looks nice! [00:19:18.0000] <hsivonen> Hixie: no complaints about the TLD-based guessing yet [00:19:50.0000] <hsivonen> Hixie: meanwhile, Firefox 29 shipped, so I should nugde the telemetry folks about running the analysis for 27 and 28 now [00:20:47.0000] <hsivonen> SimonSapin: the use case for VISCII is editing an mq patch as text in gedit when the patch patches both UTF-8 files and windows-1252 files, so the patch as a whole won't decode cleanly as either [00:21:38.0000] <hsivonen> SimonSapin: VISCII a) is supported by gedit, b) assigns all the bytes and c) doesn't do bidi or other munging [00:23:08.0000] <Ms2ger> Speaking of obscure use cases :) [00:32:20.0000] <zcorpan> https://xkcd.com/1172/ [04:08:38.0000] <jgraham> zcorpan: You should get TabAtkins to sign up as a reviewer for the quirks mode spec [04:08:54.0000] <zcorpan> TabAtkins: ^ [04:09:21.0000] <jgraham> +tests [04:18:59.0000] <zcorpan> mathiasbynens++ for record turnaround time re CSS.escape spec changes [04:41:29.0000] <mathiasbynens> zcorpan: Mozilla just landed an initial implementation based on the old spec, bad timing :/ [04:42:31.0000] <zcorpan> that's ok, it'll just escape some things that are not necessary anymore (but were necessary per css2.1) [04:42:45.0000] <zcorpan> (before errata) [04:43:08.0000] <zcorpan> i don't like errata [04:45:54.0000] <MikeSmith> "In accordance with prophecy" should be the new meme [04:47:01.0000] <MikeSmith> zcorpan: errata are the new substantive changes [04:49:04.0000] <MikeSmith> we should just give up on trying to create an XML-ER spec and just have the XML Core group publish it all in successive pieces as "errata" for the XML 5th edition spec [04:54:37.0000] <zcorpan> what if you need to errata the errata? can it go deeper? [04:55:45.0000] <zcorpan> errata #1: fix the link in the errata to this list of errata [04:56:45.0000] <zcorpan> (obviously each level should be harder to find) [04:58:30.0000] <zcorpan> ((the next level is uncommitted)) [05:00:21.0000] <zcorpan> (((limbo is not written down anywhere))) [05:22:05.0000] <MikeSmith> hah [05:33:23.0000] <annevk> http://what-if.xkcd.com/94/ <3 [06:05:57.0000] <annevk> mathiasbynens: so what if I pass in a surrogate to CSS.escape()... I get it back? Does not seem entirely correct [06:18:00.0000] <mathiasbynens> annevk: do you mean that whenever CSSOM uses the term “character”, it means scalar value/symbol? [06:18:20.0000] <mathiasbynens> annevk: or should CSSOM be changed so that it is [06:25:08.0000] <annevk> mathiasbynens: I wonder if the behavior for CSS.escape("\uD800") is what we want [06:25:48.0000] <annevk> mathiasbynens: it not defining what it means by character is an issue I suppose [06:26:02.0000] <mathiasbynens> agreed [07:25:37.0000] <zewt> wow, really? gmail's editor now appears to be swallowing control-pgup/pgdn, preventing me from changing tabs with the keyboard while editing an email [07:28:53.0000] <SamB> zewt: time to switch to basic html? [07:30:19.0000] <SamB> (it's not as bad as it sounds -- there's still some JavaScript for autocompletion, it's just still possible to use the page if it doesn't work) [07:44:38.0000] <mathiasbynens> /me still wishes Gmail would make the annoying hard-wrapping optional for plaintext emails [08:07:16.0000] <annevk> I'd rather they wrap the whole thing [08:07:23.0000] <annevk> http://lists.w3.org/Archives/Public/public-pointer-events/2014AprJun/0020.html is ugly [08:09:10.0000] <annevk> JakeA: seems like the SW spec is broken for interfaces? [08:11:05.0000] <JakeA> annevk: ugh, yes [08:11:16.0000] <JakeA> will see which commit busted that [08:17:36.0000] <JakeA> annevk: I've rolled it back to the last good commit while I figure out what's happening [08:29:37.0000] <JakeA> annevk: New version of polymer broke it. Asked Alex to look into it. gh-pages version is working now though [08:31:32.0000] <annevk> JakeA: ta [08:31:52.0000] <annevk> Polymer seems like such overkill for a spec [08:37:47.0000] <Domenic_> I like the idea of killing preprocessors via custom elements. [08:42:35.0000] <mathiasbynens> annevk: that’s hypermail’s fault, though [08:46:39.0000] <Hixie> gsnedders: yt? [08:46:53.0000] <Hixie> gsnedders: are you doing test coverage analysis of the parser by any chance? [08:54:34.0000] <Ms2ger> https://twitter.com/mattur/status/461240899503407105 [08:55:30.0000] <annevk> Pretty sure Philip` would have spotted that before they even started to attempt writing Polyglot [08:56:02.0000] <Hixie> anyone want to take over maintaining the blog? it's the main user of CPU on our machine these days. [08:59:39.0000] <jgraham> Hixie: 4 posts in 2 years? Seems like you could just shoot it in the head [08:59:49.0000] <annevk> Hixie: maintaining how? [09:00:13.0000] <jgraham> Maybe try to dump the content somehow first [09:00:18.0000] <Hixie> maintaining like maybe transitioning us to something that doesn't take huge amounts of CPU just to report content that, as jgraham points out, has only changed 4 times in 2 years [09:00:35.0000] <annevk> /me attempts to reach blog.whatwg.org [09:01:11.0000] <Hixie> the lowest hanging fruit would be making /feed static somehow [09:01:47.0000] <Domenic_> Octopress has worked very well for me. (Static blog generator.) [09:01:51.0000] <annevk> I can't even get to the server at this point [09:01:59.0000] <Hixie> load average: 35.36, 20.42, 11.41 [09:02:05.0000] <Domenic_> My preference for Node toolchains over Ruby ones would make me want to try Ghost, but Octopress is known-working. [09:02:14.0000] <Hixie> about 19 of the top 20 processes on the machine are: [09:02:24.0000] <Hixie> 4897 17693 lhunt 0:00.96 0.3 262m 34m 1.8 ? php54.cgi [09:02:29.0000] <Hixie> where "lhunt" is the user that runs the blog [09:02:31.0000] <annevk> Pretty sure WP has plug-ins to make most stuff stable [09:02:43.0000] <annevk> s/stable/static/ [09:02:53.0000] <annevk> Just need to get in... [09:02:58.0000] <Hixie> yeah i'd be shocked if WP didn't have something like that [09:03:25.0000] <Hixie> making the RSS static was one of the first optimisations i made to my own blog software decades ago :-) [09:03:43.0000] <annevk> /me updates blog [09:04:54.0000] <jgraham> Oh, look http://octopress.org/docs/plugins/render-partial/ sounds like what I want for the testthewebforward website [09:05:39.0000] <jgraham> Domenic_: Any idea if that works directly with github static pages or whatever they're called [09:05:55.0000] <jgraham> testthewebforward is using raw Jekyll [09:06:01.0000] <Domenic_> jgraham: yup, any static file server. I host http://domenic.me/ with GitHub pages + a CNAME file. [09:06:21.0000] <Domenic_> AFAICT Octopress is just Jekyll + some blog-specific helpers [09:06:22.0000] <jgraham> Domenic_: Right, but I think GitHub have some auto-build thing [09:06:32.0000] <Domenic_> jgraham: yeah, but I don't use that, I just push the files. [09:06:40.0000] <Domenic_> after building locally [09:06:51.0000] <jgraham> Domenic_: Right, but testthewebforward does [09:07:00.0000] <jgraham> Which is what I want to change [09:07:01.0000] <Domenic_> jgraham: oh i see what you're saying. Cool! [09:08:04.0000] <jgraham> Looks like the answer is no :( [09:08:14.0000] <cabanier> Ms2ger: can you do merge your own pull request? [09:08:19.0000] <cabanier> s/do// [09:08:36.0000] <jgraham> cabanier: Which repo? [09:08:50.0000] <cabanier> jgraham: w3c/web-platform-tests [09:09:16.0000] <jgraham> cabanier: If it is reviewed on github or critic then feel free to merge [09:09:57.0000] <cabanier> jgraham: how do I ask for review there? Do I assign it to someone? [09:10:00.0000] <annevk> Hixie: blog.whatwg.org is super slow [09:10:12.0000] <Hixie> yes [09:10:23.0000] <jgraham> cabanier: critic automagically asks for review [09:10:25.0000] <Hixie> that's why i mentioned it eaclier :-) [09:10:28.0000] <Hixie> earlier even [09:10:34.0000] <jgraham> cabanier: https://critic.hoppipolla.co.uk/r/1416 [09:10:56.0000] <gsnedders> Hixie: no, but it's trivial to do from html5lib [09:11:25.0000] <gsnedders> Hixie: except the fact that that doesn't necessarily perfectly match the spec, as it is slightly out of date [09:11:32.0000] <cabanier> jgraham: got it! [09:11:36.0000] <TabAtkins> annevk: The behavior for CSS.escape("\uD800") is that you get a string containing U+FFFD, right? [09:11:51.0000] <gsnedders> Hixie: and what are you looking for? line coverage? branch coverage? path coverage? [09:12:07.0000] <Hixie> gsnedders: mostly coverage of each of the switch statements' branches, i think [09:12:17.0000] <Hixie> gsnedders: just to make sure we've covered all of them at least once [09:12:27.0000] <annevk> TabAtkins: no, you get U+D800 back with the current API, though it depends on what character means whether or not it crashes [09:13:02.0000] <gsnedders> Hixie: ping me in half an hour and I'll check what html5lib currently gives [09:13:07.0000] <Hixie> gsnedders: i can probably instrument my own code to check later if you haven't, i was just curious if you were looking into it. could be a useful thing to look at. [09:13:15.0000] <TabAtkins> Hmm. I'd expect either U+FFFD or "\d800", which escapes to a U+FFFD character. [09:13:21.0000] <gsnedders> Hixie: it's trivial to do with html5lib, and I've done it before [09:13:26.0000] <Hixie> gsnedders: ah ok, cool [09:13:32.0000] <TabAtkins> Depending on how escape() is defined. [09:13:37.0000] <gsnedders> Hixie: like nosetests --coverage or something, I forget the argument needed [09:13:44.0000] <Hixie> cool [09:13:52.0000] <annevk> Hixie: cache plugin is on and updated now... [09:13:54.0000] <Hixie> /me wishes he could say the same of FPC [09:13:54.0000] <gsnedders> but yeah, me -> uni [09:13:58.0000] <annevk> Hixie: hopefully it works better? [09:14:17.0000] <Hixie> (i think you have to use like valgrind or something to do coverage testing for FPC programs) [09:14:20.0000] <Hixie> annevk: cool [09:14:36.0000] <Hixie> annevk: does that make /feed basically static? [09:15:09.0000] <Hixie> looks like yes [09:15:17.0000] <Hixie> there's a huge comment at the end of the feed now talking about the cache status [09:15:35.0000] <Hixie> i'm guessing that's not necessary :-) [09:16:27.0000] <annevk> hmm [09:16:50.0000] <Hixie> thanks for doing this [09:16:52.0000] <Hixie> it should be a huge help [09:21:08.0000] <jgraham> cabanier: You want to add your email to https://critic.hoppipolla.co.uk/home so you get notifications [09:24:19.0000] <annevk> Hixie: I'll look into disabling debug [09:24:27.0000] <cabanier> jgraham: done. let me know if I'm not using the toold correctly [09:25:57.0000] <Hixie> annevk: cool [09:26:05.0000] <annevk> Hixie: the admin panel is sooooo slow [09:28:18.0000] <Hixie> the load average is dropping [09:28:28.0000] <Hixie> but it's still 4.64, 7.54, 8.89 [09:29:30.0000] <jgraham> cabanier: Looks like you are doing just fine at the moment. I suspect Ms2ger is particularly interested in "null" rather than a randomly chosen invalid value though (I didn't check the other test, I'm just assuming that it doesn't use "null") [09:31:10.0000] <cabanier> jgraham: yes, it uses a bogus value [09:31:26.0000] <cabanier> jgraham: maybe he's worried that someone tests for the "null" string? [09:31:39.0000] <annevk> Hixie: should be gone now; if you're logged in you might still see it or maybe it's another kind of artifact [09:31:59.0000] <Hixie> annevk: k [09:32:02.0000] <Hixie> annevk: thanks! [09:32:35.0000] <annevk> Domenic_: tools used to fetch DTDs, brought netscape.com nearly to its knees back in the RSS days [09:32:54.0000] <Domenic_> wowwww [09:33:23.0000] <Hixie> made w3c suffer too [09:33:46.0000] <annevk> Yeah, W3C might still suffer I guess [09:33:57.0000] <cabanier> jgraham: I'll add that test. It's easy [09:38:52.0000] <beverloo_> MikeSmith, thanks for the push repo! [09:42:45.0000] <cabanier> jgraham: I think I messed something up [09:45:08.0000] <zcorpan> TabAtkins: r? https://critic.hoppipolla.co.uk/r/1424 [09:45:27.0000] <zcorpan> TabAtkins: also would you be OK with setting up a filter for quirks-mode/ ? [09:47:25.0000] <TabAtkins> Dunno how to do so, but I'm fine with such. [09:49:54.0000] <TabAtkins> annevk: Yeah, W3C still has caching infrastructure in place to handle DTD queries super-cheaply. [09:50:39.0000] <jgraham> cabanier: Seems like you created a new review rather than pushing a commit to your previous branch? [09:51:55.0000] <jgraham> cabanier: Right, the normal mode of operation is to address all the comments in that review and not merge until it says "Accepted" in big letters [09:52:11.0000] <cabanier> jgraham: I directly committed something in the master github. I have no idea how that happened. [09:54:25.0000] <cabanier> jgraham: ok. I should not have merge that pullrequest yet either [09:54:35.0000] <zewt> can just reset it if it's not pushed [09:55:07.0000] <jgraham> cabanier: Yeah, I'm not sure what you did here, but basically no harm done I thnk [09:55:15.0000] <cabanier> weird [09:56:06.0000] <gsnedders> Hixie: http://stuff.gsnedders.com/html5lib-coverage-20140430/ is what html5lib-python currently has. html5lib.html5parser and html5lib.tokenizer are probably the only really interesting cases [09:57:31.0000] <zcorpan> TabAtkins: https://critic.hoppipolla.co.uk/home click "add filter", type quirks-mode/ in path, make sure it's the right repo and type is "reviewer", click save [09:57:59.0000] <zcorpan> TabAtkins: https://critic.hoppipolla.co.uk/tutorial?item=filters if you feel like reading about it [09:58:07.0000] <zcorpan> TabAtkins: and thanks :-) [09:58:11.0000] <TabAtkins> kk [09:58:38.0000] <gsnedders> Hixie: so we're missing tests for <![FOO[ where <![CDATA[ is parsed, some case in the AAA, and a few cases of branches always being taken [09:59:33.0000] <zcorpan> jgraham: wtf happened here? https://critic.hoppipolla.co.uk/r/1423 https://critic.hoppipolla.co.uk/r/1424 https://critic.hoppipolla.co.uk/r/1425 https://critic.hoppipolla.co.uk/r/1426 [09:59:34.0000] <Hixie> gsnedders: neat [10:01:01.0000] <gsnedders> Hixie: a few of those cases I'm not sure it's even possible to not take the branch quickly looking [10:01:09.0000] <jgraham> zcorpan: My guess is that MikeSmith was trying to help me set up wptrunner in critic, running into problems and pressing the "test hook" button in w-p-t which re-sends the last request [10:01:19.0000] <gsnedders> Hixie: (proof is left as an exercise to the reader) :) [10:01:22.0000] <Hixie> gsnedders: dead code in html5lib? [10:01:25.0000] <zcorpan> jgraham: oh [10:01:39.0000] <zcorpan> ok [10:02:03.0000] <gsnedders> Hixie: the branch is always taken, so could maybe hoist the body of the if statement and get rid of the if statement [10:02:24.0000] <zcorpan> MikeSmith: wanna drop the excess reviews? :-) [10:03:20.0000] <gsnedders> Hixie: anyhow, time for me to go back to^W^W^W start revising for exam tomorrow :) [10:03:52.0000] <Hixie> gsnedders: enjoy! [10:05:13.0000] <jgraham> MikeSmith: I dropped them, no need for you to do anything [10:19:05.0000] <MikeSmith> jgraham: OK [10:19:25.0000] <MikeSmith> jgraham: yeah your guess was correct [10:19:34.0000] <MikeSmith> monkey vs typewriter [10:31:40.0000] <TabAtkins> What's the correct repo in critic for quirks-mode tests? [10:32:49.0000] <jgraham> TabAtkins: web-platform-tests [10:32:57.0000] <Ms2ger> Since... yesterday? [10:33:22.0000] <jgraham> Today [10:33:41.0000] <jgraham> Possibly depending on timezone or sleeping pattersm [10:33:42.0000] <MikeSmith> beverloo_: glad to help [10:33:45.0000] <jgraham> *patterns [10:33:47.0000] <TabAtkins> Cool, thanks. [10:34:10.0000] <MikeSmith> Hixie: let's move the blog over to hosting at w3.org! [10:34:22.0000] <MikeSmith> that'll probably work out well [10:34:26.0000] <jgraham> MikeSmith: Are we still stuck with the wptrunner webhook thing? [10:34:40.0000] <MikeSmith> jgraham: yeah I'm stumped at least [10:34:54.0000] <Ms2ger> MikeSmith, ....interesting idea [10:35:32.0000] <MikeSmith> jgraham: if Robin can get some time to look at it he might be able to figure it out. I have no clue myself at this point [10:35:36.0000] <jgraham> MikeSmith: Did you try choosing a different content type and then switching back, like darobin suggested? [10:35:39.0000] <MikeSmith> jgraham: I can look more in the morning my time [10:35:46.0000] <MikeSmith> jgraham: yeah did that [10:35:48.0000] <jgraham> MikeSmith: OK [10:35:55.0000] <jgraham> MikeSmith: Thanks for the help [10:36:22.0000] <jgraham> Probably we should just ask github at this point [10:36:30.0000] <MikeSmith> yeah [10:37:17.0000] <MikeSmith> jgraham: that config UI makes me want to kick somebody in the face [10:37:53.0000] <Ms2ger> Hey [10:37:59.0000] <Ms2ger> The previous UI was "use the API" [10:38:15.0000] <MikeSmith> I like that UI better [10:38:30.0000] <Hixie> MikeSmith: heh [10:38:30.0000] <MikeSmith> at least somebody didn't get paid money to spend time developing that UI [10:39:12.0000] <MikeSmith> Hixie: speaking of blogs btw the blog thing on w3.org also sucks up most of the CPU. except in that case it's multiplied by 150 times. Or however many CGs there are. It's nuts [10:39:18.0000] <MikeSmith> so, yay for blogs [10:40:27.0000] <gsnedders> Hixie: you just running Apache? [10:40:50.0000] <Hixie> MikeSmith: nice [10:40:52.0000] <Hixie> gsnedders: yeah [10:41:43.0000] <gsnedders> Hixie: should be able to throw mod_cache in front of teh blog, I think [10:42:03.0000] <MikeSmith> cabanier: if/when you have time, please review https://critic.hoppipolla.co.uk/r/1239 [10:42:13.0000] <MikeSmith> cabanier: which is for https://github.com/w3c/web-platform-tests/pull/851 [10:42:16.0000] <Hixie> gsnedders: configuring apache on dreamhost is a pain [10:42:23.0000] <gsnedders> Hixie: gl;hf ;P [10:42:29.0000] <Hixie> :-) [10:52:20.0000] <MikeSmith> jgraham: please try again on that hook right now if you have time [10:52:50.0000] <MikeSmith> when I went to check it now it was set back to no-JSON and with SSL turned back on [10:53:01.0000] <MikeSmith> which I re-switched it back again [10:56:23.0000] <jgraham> MikeSmith: Checking [10:56:41.0000] <jgraham> MikeSmith: Worked :) [10:56:53.0000] <MikeSmith> sweet [10:57:17.0000] <MikeSmith> so I can now go to sleep feeling like I actually accomplished something useful today [10:57:26.0000] <jgraham> Heh [10:57:35.0000] <jgraham> Sleep well :) [12:03:14.0000] <cabanier> MikeSmith: that test looks good to me [14:47:24.0000] <IZh> Hixie: Hi. [14:48:35.0000] <IZh> Hixie: My script reported that your URL returned HTTP 504 status for last document. [14:51:24.0000] <Hixie> 504! [14:51:30.0000] <Hixie> /me tries to remember what 504 is :-) [14:51:38.0000] <Hixie> oh gateway timeout [14:51:40.0000] <Hixie> yeah [14:51:44.0000] <Hixie> server was having issues this morning [14:53:47.0000] <IZh> Should my script handle this by repeating sending? [14:56:35.0000] <Hixie> nah [14:56:43.0000] <Hixie> i would just ignore any errors you get back [14:57:01.0000] <Hixie> i'll update the spec again soon enough and then it'll regenerate [15:01:26.0000] <IZh> Ok :-) [15:11:48.0000] <eligrey> i'd like to request an account for the wiki (desired username: eligrey) [15:12:17.0000] <eligrey> the first edit i'd like to make is writing up a navigator.cores proposal [15:12:42.0000] <eligrey> (which exposes the amount of total hardware threads available to the system) [15:13:47.0000] <eligrey> the wiki says to talk to "these permanent autoconfirmed members" (redlink) :/ [15:19:46.0000] <Hixie> eligrey: e-mail? [15:19:55.0000] <eligrey> me⊙ec [15:20:19.0000] <Hixie> done [15:24:24.0000] <eligrey> ok thanks [16:06:29.0000] <Hixie> anyone got a recent IE? what does http://software.hixie.ch/utilities/js/live-dom-viewer/?saved=2883 look like? [16:07:25.0000] <boogyman> by recent you mean nightly build or 11? [16:08:37.0000] <boogyman> i have 10.0.9 [16:10:29.0000] <boogyman> http://minus.com/i/bpp3b2KgaK6Vy @ Hixie for IE 10.0.9 [16:10:39.0000] <Hixie> thanks [16:10:48.0000] <Hixie> interesting [16:10:57.0000] <Hixie> so safari and chrome do cross-origin font loads, but IE and mozilla do not [16:35:30.0000] <hober> Hixie: http://www.w3.org/2011/10/31-webapps-minutes.html#item02

2014-04-01 [17:00:39.0000] for the window.onerror example, are we talking about parsing an error log later on, or programmatically doing something with the magic string? [17:01:52.0000] TabAtkins: why 60 million, btw? [17:01:59.0000] TabAtkins: what's your definition of "expect"? [17:02:04.0000] 36^5, the sqrt of 36^10. [17:02:07.0000] 50% expectation [17:02:28.0000] oh, i see [17:02:35.0000] the alphabet only actually has 16 characters [17:02:40.0000] so 16^5 [17:02:51.0000] Oh, you're only genning hex? [17:02:55.0000] I didnt' notice that [17:03:09.0000] That's only about 1M then. [17:03:40.0000] er wait [17:03:47.0000] not 16... [17:04:00.0000] oops [17:04:02.0000] it is 16 [17:04:08.0000] but it's supposed to be more [17:04:10.0000] 26 or so [17:04:13.0000] 2346789abcdefghjkmnpqrtwxy [17:04:50.0000] Ah, the easy-to-distinguish alphanums. [17:05:01.0000] yeah, that's the idea [17:05:19.0000] hm [17:05:25.0000] i need this to be a power of 2 for the logic to work better [17:05:37.0000] /me wonders whether to add or remove [17:06:35.0000] What's your logic? [17:07:34.0000] geenerate hmax sha 512 of increasing number, use the first n bits to generate a 10 character string [17:07:53.0000] Interesting. [17:08:04.0000] i can add z5suv back in but i'm still short a character... [17:08:12.0000] just reading /dev/urandom will probably give you equally non-repeating results, without depending on a last-used-number state [17:08:12.0000] Why not just generate a num between 0 and 26^10, then encode it in base 26? [17:08:18.0000] really don't want to add il1o0 [17:08:51.0000] And yes, the lowercase z/s aren't really confusable for 2/5. [17:09:02.0000] (since /dev/urandom these days is basically just a crypto hash on top of entropy) [17:09:37.0000] i like to avoid actual randomness in my code [17:09:58.0000] You're effectively grabbing randomness by taking an arbitrary chunk of a hash. [17:10:37.0000] right [17:10:45.0000] reproducible "randomness" [17:10:49.0000] i prefer to avoid state that has to be persisted, myself [17:12:57.0000] also a bunch of new complexity if two clients might race and generate from the same index [17:13:29.0000] Yup. [17:14:32.0000] (which itself can be really hard to write tests for; i hate writing tests for things like locking and atomicity) [17:14:42.0000] locking files is a solved problem :-) [17:15:21.0000] with a solution that's much more complex than open('/dev/urandom').read(16) :----------) [17:15:45.0000] i guess it comes down to what your environment provides, and how much you trust it [17:16:03.0000] (i don't include "make sure /dev/urandom is really random" in my tests, after all) [17:17:00.0000] ("really random" as in "not /dev/zero") [17:17:09.0000] (in case someone was going to nit about PRNGs) [17:17:30.0000] Any system that doesn't implement a decent /dev/urandom would be *super* insecure anyway. [17:18:19.0000] i tend to suspect security inexperience in people who still use /dev/random because it's "more secure", heh [17:18:59.0000] locking files is like lines of code including importing the relevant constants and unlocking... [17:19:05.0000] like three lines, even [17:19:36.0000] ok, now using a 32 character alphabet. [17:20:07.0000] using the first 5 bits of each of the first ten bytes of the hmac sha 512 of numbers in sequential order. [17:20:20.0000] bbiab [17:21:55.0000] maybe your environment has additional helpers for the busywork (blocking if the lockfile already exists, cleaning up if the pid contained in the lockfile doesn't exist, etc) [17:22:18.0000] It's perl, so that's plausible. [17:22:33.0000] also the need to atomically write the state file containing the last-used index (write new file, sync, rename), which is something all languages really should provide, but few do [17:24:06.0000] While you're locked, that works anyway, right? [17:27:07.0000] if the machine crashes, you don't want to reboot and come back up with a zero-byte file [17:27:24.0000] Ah, right. [17:27:32.0000] (restore from backup and you lose that day's index delta) [17:28:10.0000] ''.join('23456789abcdefghijkmnpqrstuvwxyz'[ord(x) >> 3] for x in hashlib.sha512(open('/dev/urandom').read(32)).digest()[0:10]) # one-liners are evil, even in python [20:59:02.0000] i ended up testing 30,058,310 codes with the new algorithm, no dupes. [21:00:29.0000] (by tab's maths, there would be a 50% chance of a dupe at about 33 million, if i understand it right) [22:25:18.0000] man, the tokeniser sure does have a lot of states [23:53:09.0000] TabAtkins: is unprefixing transitions *and* transforms both covered under the umbrella of this bug? https://code.google.com/p/chromium/issues/detail?id=154772 [00:12:57.0000] Hixie: sounds like a plan, will devote some time to it eventually. I need to do a write up with use cases, possibilities (e.g. generated vs. human readable), impact on existing/future specs, etc. [03:00:14.0000] TabAtkins: ping [03:07:12.0000] TabAtkins: unping....will catch you later [03:07:55.0000] The new css spec style sheet is so funny... [03:22:25.0000] looks like the support status markers on CSS specs haven't been updated to reflect the demise of Presto and Blink forking off of WebKit [03:41:39.0000] MikeSmith: I don't know what to do about bugzilla.validator.nu cookies. The urlbase and cookiepath settings look right to me. [03:52:22.0000] oh, hsivonen is alive [03:53:57.0000] is this where I'm supposed to say that rumors of my demise have been exaggerated? [03:58:29.0000] There were rumours? [04:32:33.0000] hsivonen, oh, better news than your death, congratulations :) [04:32:47.0000] Ms2ger: thanks [04:34:23.0000] what's the news? [04:35:01.0000] zcorpan: a baby (that is, I'm now back from paternity leave) [04:35:30.0000] hsivonen: ok, congrats! [04:35:36.0000] zcorpan: thanks [04:37:47.0000] congratulations [04:38:06.0000] smaug____: thanks [06:57:29.0000] wow. there's now a community group for MPEG-2 in-band captioning in