| 01:44 | <TabAtkins> | WHY IS IT SO DAMNED ADDICTING TO DRIVE THIS PROFILE DOWN??? |
| 01:45 | <TabAtkins> | Just spent 10 minutes shaving off .15%, because I knew it was there and it was annoying me. |
| 01:50 | <tantek> | sounds like well designed gamification |
| 01:56 | <TabAtkins> | It is. Making numbers smaller is so much fun! |
| 01:56 | <TabAtkins> | Especially when, after a change, the number is so small the box disappears from the profile completely. |
| 06:57 | <annevk> | mounir: screen.orientation seems undefined |
| 07:08 | <annevk> | Anyone here run Chrome stable? Would still like to know what http://software.hixie.ch/utilities/js/live-dom-viewer/?saved=3191 does |
| 07:33 | <zcorpan> | SimonSapin: also test quirks mode for the font size thing |
| 07:35 | <zcorpan> | annevk: i have opera next 24 which is like chrome 37, which says Uncaught ReferenceError: TextDecoder is not defined on line 1 |
| 07:36 | <annevk> | zcorpan: oh, that's even worse |
| 07:37 | <zcorpan> | SimonSapin: and monospace |
| 07:38 | <zcorpan> | SimonSapin: and non-16px default font size |
| 07:39 | <zcorpan> | i recall there being something funny in that area when writing the quirks spec (although i didn't specify the quirk) |
| 08:14 | <annevk> | Domenic: JakeA: is creating a tee synchronously acceptable? |
| 08:14 | <annevk> | Domenic: JakeA: that's what we require with clone()... |
| 08:16 | <JakeA> | Good question |
| 08:17 | <JakeA> | I guess it depends on what's synchronously available on the stream. It isn't a problem with the Request/Response API as it currently stands |
| 08:18 | <annevk> | Yeah, I remember Domenic saying that reading from a stream could be synchronous... |
| 08:20 | <annevk> | hsivonen: you must love that bug |
| 08:20 | <annevk> | hsivonen: the response to utf-8 advocacy end of 2003 is fun |
| 08:21 | <annevk> | "lol That would be a good solution :) I personnaly use the iso-8859-15 charset which is a slight improvement of iso-8859-1 ..." |
| 08:29 | <hsivonen> | annevk: whoa. I didn't even notice that comment. |
| 08:49 | <SimonSapin> | TabAtkins: what profiler do you use? |
| 08:56 | <annevk> | Ms2ger: I get "[Errno 13] Permission denied: '/tmp/spec'" these days for our Makefile; Apple apparently requires root for that now |
| 08:56 | <annevk> | Ms2ger: can we make that work in some other way? |
| 09:04 | <Ms2ger> | Will look later |
| 09:53 | <annevk> | jgraham: lazy Python help request; is there a better way to do this: |
| 09:53 | <annevk> | id = re.match(r"^/tools/test.py\?from=\d+&to=(\d+)$", os.environ["REQUEST_URI"]) |
| 09:53 | <annevk> | print id.group(1) |
| 10:15 | <annevk> | well it works, that's enough for me |
| 10:40 | <hsivonen> | annevk: too much success: http://mxr.mozilla.org/mozilla-central/source/browser/components/loop/content/shared/libs/sdk.js#9210 |
| 10:41 | <hsivonen> | I guess I should file a bug not to ship that stuff with Loop and to use built-in stuff |
| 10:42 | <annevk> | hsivonen: copy from mxr.mozilla.org/mozilla-central/source/browser/components/loop/content/shared/libs/sdk.js#9210 |
| 10:43 | <annevk> | hsivonen: sorry, copy from jsbell's work |
| 10:43 | <annevk> | hsivonen: so they could just use TextDecoder / TextEncoder |
| 10:44 | <annevk> | hsivonen: I think Firefox OS had a similar problem |
| 10:45 | <annevk> | hsivonen: but that doesn't appear on mxr |
| 10:46 | <annevk> | hsivonen: is there a bug on mxr not using TLS? |
| 10:46 | <annevk> | hsivonen: hmm, make that HSTS |
| 10:48 | <hsivonen> | annevk: dunno. There are random hostnames under mozilla.org that break with https, so at least includeSubdomains on mozilla.org is infeasible for the time being |
| 10:49 | <annevk> | I guess I'll just assume someone else is trying to make it work |
| 10:49 | <annevk> | hsivonen: if you want I can file a bug on Loop, no big deal |
| 10:50 | <hsivonen> | annevk: I filed one already |
| 10:50 | <hsivonen> | annevk: I CCed you |
| 10:50 | <annevk> | cool |
| 10:50 | <MikeSmith> | hsivonen: annevk fwiw now, today I will set up a redirect of that dvcs.w3.org url to the current whatwg spec (and same for any other stale dvcs.w3.org URLs). but can't til I'm back home |
| 10:51 | <hsivonen> | MikeSmith: interesting turn of events. Thanks. |
| 10:52 | <annevk> | What URLs are we talking about? I must have missed something |
| 10:53 | <MikeSmith> | annevk: the gecko code comment in hsivonen's mxr link |
| 10:54 | <annevk> | MikeSmith: ah okay, it seems to redirect a bit already |
| 10:55 | <MikeSmith> | Ah ok |
| 11:09 | <MikeSmith> | btw a week or so back, I already set up such a redirect for the d |
| 11:09 | <GPHemsley> | it's more than "a bit"... but it's an HTML redirect |
| 11:09 | <MikeSmith> | ...old dvcs.w3.org draft of the URL spec |
| 11:12 | <MikeSmith> | annevk: also btw as far as I know the Director has no blanket objection to normative references to whatwg drafts |
| 11:13 | <annevk> | MikeSmith: so what is the problem then? |
| 11:16 | <MikeSmith> | annevk: well sorry to be coy but I just think it's just worth knowing that the problem could be something other than that |
| 11:19 | <MikeSmith> | if somebody were to ask the Director directly maybe that could help to clear things up some |
| 11:19 | <MikeSmith> | maybe Domenic could ask for some clarification about it at the TAG f2f |
| 11:55 | <annevk> | hsivonen: your mxr find gave me an idea |
| 11:55 | <annevk> | https://github.com/search?q=encoding.spec.whatwg.org |
| 11:55 | <annevk> | there's a lot of hits |
| 12:59 | <darobin> | does the SVG on https://whatwg.org/ *have* to be embedded with <object>? |
| 12:59 | <darobin> | I mean seriously, who does that |
| 13:00 | <darobin> | poor little harmless SVG |
| 13:03 | <gsnedders> | darobin: Hixie_ |
| 13:04 | <darobin> | gsnedders: I'm shocked |
| 13:04 | <darobin> | it's almost like torturing kittens or something |
| 13:05 | <mathiasbynens> | annevk: use quotes https://github.com/search?q="encoding.spec.whatwg.org"&type=Code |
| 13:05 | <zcorpan> | darobin: what's wrong with <object>? |
| 13:06 | <darobin> | zcorpan: it's eeeeeeevil |
| 13:06 | <darobin> | it bullied me when I was little |
| 13:06 | <zcorpan> | poor darobin |
| 13:07 | darobin | sniffles, nods |
| 13:08 | <zcorpan> | i guess the <img> is also downloaded though |
| 13:19 | <annevk> | if you embed SVG through <object>, I'd guess the problem is that you create a browsing context rather than an image |
| 13:19 | <annevk> | i.e. bloat |
| 13:24 | <zcorpan> | time to switch to the <img src=png srcset=svg> hack |
| 13:26 | <Domenic> | annevk: JakeA yes sync tee is no problem |
| 13:32 | <zcorpan> | hmm, making https://images.whatwg.org/abstract.png in svg means having a jpg in a data: URL, right? at least if it should work in <img> |
| 13:33 | <annevk> | hsivonen: how can Mozilla work without original iso-8859-1? HTTP requires it... |
| 13:33 | <annevk> | hsivonen: I guess we simply do byte manipulation there? |
| 13:34 | <annevk> | zcorpan: yes |
| 13:35 | <zcorpan> | annevk: thx |
| 13:44 | <JakeA> | Domenic: \o/ |
| 13:47 | <annevk> | JakeA: did you see my cache mode update to Fetch? |
| 13:47 | <annevk> | JakeA: Domenic: if you could review Fetch's cache mode update that'd be great, if it's fine I'll add the corresponding API |
| 13:54 | <JakeA> | annevk: looking now |
| 14:07 | <Domenic> | Will look at |
| 14:19 | <zcorpan> | ok so what needs updating with the kitchensink? xbl -> web components, dom core/dom events -> dom, uri/iri -> url |
| 14:19 | <zcorpan> | (is web components above html?) |
| 14:24 | <SimonSapin> | zcorpan: how much should CSSOM serialization preserve the origin form of things? E.g. keywords instead of the equivalent numerical value (when there is one) |
| 14:25 | hemanth | did a talk on generators at JSChannel :) http://www.youtube.com/watch?v=ZrgEZykBHVo |
| 14:27 | <gsnedders> | zcorpan: Components is surely alongside it? |
| 14:31 | <Ms2ger> | Components should be wrapping all around it, clearly |
| 14:31 | <Ms2ger> | And superimposing itself on everything else |
| 14:31 | <annevk> | zcorpan: part of, eventually |
| 14:31 | <annevk> | zcorpan: ECMAScript -> JavaScript |
| 14:31 | <jgraham> | Something, something, webcomponents describing itself, something, inception, something |
| 14:32 | <jgraham> | JavaScript™ |
| 14:32 | <gsnedders> | ®, please! |
| 14:35 | <zcorpan> | SimonSapin: does http://dev.w3.org/csswg/cssom/#serialize-a-css-component-value answer your question? |
| 14:35 | <annevk> | zcorpan: Notifications / Fullscreen / Sensor APIs? |
| 14:35 | <annevk> | zcorpan: GIF / PNG |
| 14:36 | <zcorpan> | annevk: can you give me ascii art for what it should say where? :-) |
| 14:36 | <SimonSapin> | zcorpan: not really. It’s rather, should "parse a declaration" preserve the original component values, or some approximation? |
| 14:37 | <JakeA> | annevk: Btw, developers I've asked would expect fetch('/foo') to include cookies |
| 14:37 | <SimonSapin> | parse+serialize |
| 14:37 | <JakeA> | annevk: and still pull faces after I explain why it doesn't |
| 14:37 | <SimonSapin> | or can it do anything, as long as it preserves property-specific semantics? |
| 14:39 | <zcorpan> | SimonSapin: i haven't studied serialization in detail. i know the spec has bugs. so i'm not sure. my guess is that you don't need to preserve the original component values |
| 14:39 | <zcorpan> | SimonSapin: e.g. for <color> you don't roundtrip keywords |
| 14:40 | <gsnedders> | my memory of this stuff is web compat constraints are behind UA sniffing |
| 14:40 | <SimonSapin> | gsnedders: what does that mean? |
| 14:41 | <jgraham> | SimonSapin: "servo is screwed" |
| 14:41 | <zcorpan> | SimonSapin: it can't do anything that preserves semantics, you have to follow the serialization rules (insofar they aren't wrong) |
| 14:41 | <SimonSapin> | thanks jgraham for your usual optimism :) |
| 14:41 | <gsnedders> | SimonSapin: if your UA string claims you're IE, the expected answer is "rgb(255, 255, 255)"; if you're Firefox, the expected answer is "black"; if you're Chrome, the expected answer is "rgba(255,255,255,1)" |
| 14:42 | <gsnedders> | SimonSapin: if you deviate from this, you break the web |
| 14:42 | <gsnedders> | SimonSapin: if you're Servo, good-fucking-luck |
| 14:42 | <SimonSapin> | wut |
| 14:42 | <jgraham> | But of course UA string isn't the only way of doing UA sniffing |
| 14:42 | <gsnedders> | SimonSapin: what's so wut? did you forget to leave your sense of logic at the door? |
| 14:43 | <SimonSapin> | my mistake :) |
| 14:43 | <jgraham> | It's a pity that people didn't parse the results and convert to a canonical form rather than UA sniff :( |
| 14:44 | <gsnedders> | speaking of which, has anyone done anything about Error.stack and UA-dependent expectations there? |
| 14:46 | <SimonSapin> | gsnedders: was that for computed or specified values? |
| 14:48 | <gsnedders> | SimonSapin: computed, I think |
| 14:54 | <JakeA> | annevk: ok, https://twitter.com/jaffathecake/status/514425450194411520 suggests developers don't really know what to expect re cookies |
| 14:56 | <Ms2ger> | That they're delicious delicacies! |
| 15:00 | <annevk> | JakeA: they should eat them, clearly |
| 15:01 | <JakeA> | nom nom nom |
| 15:13 | <TabAtkins> | annevk: Chrome stable for 3191 is Uncaught TypeError: Failed to construct 'TextDecoder': The encoding label provided ('hz-gb-2312') is invalid. |
| 15:13 | <TabAtkins> | SimonSapin: Just cProfile, piped through gprof2dot. `bikeshed profile` generates a profile for you now, if you've got the right stuff installed (documented in its help info). |
| 15:14 | <SimonSapin> | cool |
| 15:14 | <SimonSapin> | I also like xdot, for browsing gprof2dot output |
| 15:15 | <annevk> | TabAtkins: thanks, I guess I'd need to write a standalone test for hz-gb-2312 if I really wanted to know whether it was gone |
| 15:16 | <TabAtkins> | SimonSapin: Yeah, thought about doing that, but with --root and --leaf I have enough control so far that it's not overwhelming. |
| 15:16 | <SimonSapin> | TabAtkins: what are these? |
| 15:16 | <TabAtkins> | Kinda wanna PR gprof2dot to allow you to exclude subtrees, so I can drop everything below the html5lib entrypoint. That's about half the nodes on the graph. |
| 15:17 | <TabAtkins> | --root specifies which node you want to display as root; trims everything without it as ancestor, and everything above it. --leaf specifies that you only want to see call chains that bottom out in the specified node. |
| 17:17 | <annevk> | https://wiki.whatwg.org/wiki/TLS |
| 17:17 | <annevk> | contributions welcome |
| 17:23 | <willchan> | annevk: i think package analysis is a typo. perhaps you meant packet analysis. or traffic analysis. |
| 17:23 | <willchan> | and it might be more precise to say increased confidentiality |
| 17:26 | <wanderview> | JakeA: ping |
| 17:27 | <JakeA> | wanderview: hey (on a train so dodgy connection) |
| 17:27 | <wanderview> | JakeA: np, just curious how the cache spec updates coming? :-) |
| 18:25 | <annevk> | willchan: thanks fixed |
| 18:54 | <zcorpan> | what's the state of the art accessibility best practice for svg (for https://images.whatwg.org/abstract.png ) ? |
| 19:06 | <annevk> | zcorpan: seems better to provide it in alt=? |
| 19:09 | <zcorpan> | annevk: yes, maybe so |
| 19:19 | <zcorpan> | annevk: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3200 |
| 19:35 | <zcorpan> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3201 |
| 19:36 | <Hixie_> | SVG and MathML at this point should be in the top box |
| 19:37 | <Hixie_> | also you should add some more things |
| 19:37 | <Hixie_> | WebIDL |
| 19:37 | <Hixie_> | JPEG (on the side) |
| 19:37 | <Hixie_> | GIF, PNG |
| 19:37 | <Hixie_> | Encodings (below) |
| 19:37 | <Hixie_> | CSP (side) |
| 19:38 | <Hixie_> | TLS (below) |
| 19:42 | <zcorpan> | Hixie_: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3202 |
| 19:43 | <Hixie_> | nice |
| 19:43 | <zcorpan> | oops forgot webidl |
| 19:43 | <caitp> | kitchen sink is getting pretty full there |
| 19:43 | <Hixie_> | caitp: i think you may have misread the diagram |
| 19:44 | <Hixie_> | the sink hasn't changed here |
| 19:44 | <caitp> | it must be rendering very wrong here on netsurf |
| 19:46 | <zcorpan> | Hixie_: ok http://software.hixie.ch/utilities/js/live-dom-viewer/saved/3203 |
| 19:50 | <zcorpan> | caitp: i tried installing netsurf and opened the link. it didn't run the script in live-dom-viewer it seems. then i tried pasting the svg into the textarea, and netsurf crashed |
| 19:50 | <caitp> | i would be very surprised if netsurf ran it |
| 19:50 | <caitp> | misread jokes compounded by other misread jokes |
| 19:50 | <caitp> | :p |
| 19:51 | <caitp> | humour on IRC is hard because you really can't hear the tone of voice |
| 19:51 | <caitp> | it takes dedication and practice to figure out when someone is being genuine in their remarks |
| 19:53 | <zcorpan> | well, it was fun trying a new browser that seems to be about as useful as netscape 4 |
| 19:54 | <zcorpan> | maybe next i should try servo |
| 19:59 | <zcorpan> | Hixie_: should i file a bug about the image? |
| 20:09 | <Ms2ger> | zcorpan, do try servo, and expect cssom tests from jdm :) |
| 20:09 | <Domenic> | Can we moderate L2L |
| 20:11 | <Hixie_> | zcorpan: sure |
| 20:12 | <Hixie_> | Domenic: on it |
| 20:14 | <caitp> | what has he been doing? i've only seen him on es-discuss |
| 20:14 | <Hixie_> | he posted to the whatwg list and his posts are incoherent |
| 20:15 | <Domenic> | he's on all the lists -_- |
| 20:15 | <caitp> | well look at that, he cares about the state of the web and such |
| 20:15 | <caitp> | enough to participate in all the lists |
| 20:16 | <Hixie_> | so....... how do i clear the STS headers of a web site in safari or firefox? |
| 20:16 | <zcorpan> | Hixie_: the new image is 2x for the bitmap part and the bitmap is base64-encoded, yet the image is almost half the size of the old png version :-) |
| 20:16 | <Hixie_> | zcorpan: wow |
| 20:18 | <annevk> | Hixie_: https://domparsing.spec.whatwg.org/ can redirect to https://dvcs.w3.org/hg/innerhtml/raw-file/tip/index.html |
| 20:18 | <annevk> | Hixie_: seems like we have nobody that is interested in maintaining it and Travis is actually doing an okay job |
| 20:18 | <annevk> | Hixie_: bit unfortunate about the loss of CC0, but so be it I guess |
| 20:18 | <Ms2ger> | I'll update the git repo with something tomorrow |
| 20:19 | <Hixie_> | cool, ok |
| 20:19 | <annevk> | ta |
| 20:19 | <tantek> | annevk: wait what? did we lose a whatwg editor resource so the work needs to be picked up by W3C? |
| 20:19 | <Ms2ger> | tantek, more like the whatwg editor admitted to himself he wasn't going to have time for it |
| 20:20 | <tantek> | Ms2ger - perfectly reasonable. |
| 20:20 | <zcorpan> | shouldn't we fork travis' spec? |
| 20:20 | <Hixie_> | zcorpan: ... |
| 20:20 | <Hixie_> | zcorpan: forking bad |
| 20:20 | Ms2ger | eyes zcorpan |
| 20:20 | <Ms2ger> | I have ban power here, don't I? |
| 20:20 | <zcorpan> | :-) |
| 20:20 | <Hixie_> | banning bad too! |
| 20:20 | <Hixie_> | damnit, zcorpan tricked me |
| 20:20 | <Ms2ger> | Pah |
| 20:20 | Ms2ger | withholds chocolate |
| 20:21 | <zcorpan> | Ms2ger: hey! |
| 20:21 | <Hixie_> | ffs, dreamhost |
| 20:21 | <Ms2ger> | Hixie_, see, that works better :) |
| 20:21 | <Hixie_> | you can set an http:// url to redirect, but setting an https:// redirect requires that you actually do full hosting then do the .htaccess thing |
| 20:22 | <Hixie_> | annevk: can you make domparsing redirect on your end |
| 20:22 | <zcorpan> | withholding chocolate bad |
| 20:22 | <Hixie_> | annevk: i've done it for http:// |
| 20:22 | <Hixie_> | and am updating specs. |
| 20:23 | <tantek> | isn't there a meta tag for that? ;) |
| 20:24 | <TabAtkins> | Hixie_: Mind banning L2L from the whatwg lists? He's a persistent and crazy troll in all the other standards lists. |
| 20:25 | <Hixie_> | his mod bit has already been set and i've already contacted him privately |
| 20:25 | <TabAtkins> | kk, cool |
| 21:13 | <annevk> | Hixie_: in my experience if you make the http one a redirect, the https one redirects too |
| 21:14 | <annevk> | Hixie_: not sure how that works |
| 21:17 | <annevk> | Will make it work tomorrow either way |
| 22:02 | <caitp> | "Specifying a character set early for your HTML documents allows the browser to begin executing scripts immediately." is this actually true? surely the browser doesn't wait for the entire document to arrive before trying to guess the encoding |
| 22:03 | <TabAtkins> | caitp: No, it wait for 1024 bytes to arrive. |
| 22:03 | <caitp> | that seems more likely |
| 22:04 | <caitp> | but the first 1024 bytes of the packet are probably going to arrive in the very first packet anyways, so it seems like it wouldn't make much difference latency-wise |
| 22:04 | <willchan> | caitp: https://code.google.com/p/page-speed/wiki/SpecifyCharsetEarly |
| 22:05 | <caitp> | cool thanks |
| 22:14 | <TabAtkins> | Domenic: Some more improvements to Bikeshed's handling of IDL stuff and particularly those [[HiddenField]] things. |
| 22:14 | <TabAtkins> | Domenic: I already told you that Bikeshed lets you escape [[foo]], Markdown style, as \[[foo]]. |
| 22:15 | <TabAtkins> | Domenic: Now, if you're using the IDL shortcut link syntax, you don't even need to do that: {{[[foo]]}} looks terrible, but it's a straightforward application of the syntax to the term you want to link. ^_^ |
| 22:46 | <TabAtkins> | Hixie: Is there a term for "the interval between two rAF checkpoints"? |
| 22:51 | <zewt> | cool. restarted chrome and it decided to only restore tabs from one window out of three |
| 22:51 | <zewt> | history diggin' time |
| 22:58 | <Hixie_> | TabAtkins: "page refresh period"? dunno, but one day there probably will be. it's variable, of course. |
| 22:58 | <Hixie_> | annevk: the http ones _all_ redirect for http, and none of the https ones do... |
| 22:59 | <TabAtkins> | Hixie_: Yeah, that's fine. We just want to be able to only fire a single animation iteration event per "frame", but "frame" isn't a meaningful term. |
| 22:59 | <Hixie_> | TabAtkins: oh i think the solution for that is probably to be called as part of the event loop page rendering thing we'll be adding |
| 23:00 | <Hixie_> | TabAtkins: i can give you a flag if the page will be getting rAF or whatever |
| 23:00 | <TabAtkins> | Hixie_: No need to hook into the guts of an algorithm, just want some simple way to say "UAs must not fire more than one iteration event per element per animation per "frame". " |
| 23:01 | <Hixie_> | TabAtkins: that's fire for a "must not" but presumably you also want a "must" that actually fires it |
| 23:01 | <Hixie_> | TabAtkins: and that must needs to say when it's fired, etc |
| 23:01 | <Hixie_> | especially relative to rAF, tasks, microtasks, etc |
| 23:01 | <TabAtkins> | Yeah, the animations spec is already attempting to hook into rAF for *when* to fire things. |
| 23:01 | <TabAtkins> | WebAnim, that is. |
| 23:02 | <Hixie_> | well if it hooks into when, you don't need a when not |
| 23:03 | <birtles> | (that part of the WebAnim spec is pretty awkward/hand-wavy at the moment) |
| 23:04 | <Hixie_> | hopefully i'll have the hooks for that done in a few weeks |
| 23:05 | <birtles> | that'd be great |
