| 01:35 | <rniwa> | Domenic: it looks like we have another superfluous CEReactions on HTMLTextAreaElement's value |
| 01:35 | <Domenic> | rniwa: I think that can wipe out child nodes? |
| 01:35 | <rniwa> | Domenic: only defaultValue |
| 01:35 | <rniwa> | Domenic: value only touches the "raw value" or whatever |
| 01:35 | <Domenic> | Hmm right... hmm. I think I fixed this for input, maybe. |
| 01:36 | <rniwa> | Domenic: input doesn't really matter unless you're "customizing" with "is=" attribute but yeah |
| 01:36 | <rniwa> | Domenic: the same for input element |
| 01:36 | <Domenic> | Oh no input has a mode where it changes the attribute |
| 01:36 | <rniwa> | Domenic: .value doesn't mutate attribute |
| 01:36 | <Domenic> | It does for input sometimes |
| 01:36 | <rniwa> | Domenic: really? |
| 01:36 | <rniwa> | that's very strange |
| 01:37 | <rniwa> | oh well, it's input element... |
| 01:37 | <Domenic> | https://html.spec.whatwg.org/#dom-input-value-default |
| 01:37 | <Domenic> | Yeah textArea.value though seems non-CEReactions, I agree. |
| 01:37 | <KiChjang> | i think Domenic is trolling by linking people to full page specs |
| 01:38 | <KiChjang> | here's the multipage version https://html.spec.whatwg.org/multipage/#dom-input-value-default |
| 01:38 | <rniwa> | KiChjang: if you have Safari 10 with 15" MBP, the full spec opens like a charm ;) |
| 01:39 | <rniwa> | in fact, html5 spec page happens to be one of our perf tests. |
| 01:39 | <Domenic> | Yeah it's only Firefox that has a problem really |
| 01:39 | <KiChjang> | it doesn't render quick on chrome, but at least it doesn't freeze the entire browser |
| 01:40 | <rniwa> | Domenic: btw, https://github.com/w3c/webcomponents/issues/596 is a bigger issue |
| 01:40 | <rniwa> | Domenic: having CEReactions superfluously isn't as bad as missing one |
| 01:40 | <Domenic> | Right, for sure. |
| 01:41 | <Domenic> | Presumably that's why Chrome has a bug. |
| 03:19 | <rniwa> | annevk, Domenic: do you know what happened to CSSValue and CSSPrimitiveValue objects? |
| 03:20 | <rniwa> | they seem to be very widely supported (Gecko, Blink, & WebKit) |
| 03:20 | <rniwa> | but CSS OM spec seems to entirely ignore them |
| 03:22 | rniwa | oh, I guess blink dropped it a while ago: https://groups.google.com/a/chromium.org/forum/#!topic/blink-dev/3VmxWFzcyJc |
| 03:37 | <Domenic> | rniwa: we can add them to the list at the bottom of https://html.spec.whatwg.org/#cereactions |
| 03:37 | <rniwa> | Domenic: add webkitdropzone to that list... :( |
| 03:40 | <rniwa> | I love that document.body has setter but document.head is readonly |
| 03:40 | <rniwa> | what a great consistency! |
| 03:41 | <Domenic> | rniwa: please file an issue on anything we should add to that list; I gotta head to sleep but will fix all the CEReactions bugs tomorrow |
| 03:42 | <rniwa> | Domenic: okay |
| 03:42 | <rniwa> | Domenic: I'm going through my IDL files and figuring out things as we go |
| 03:42 | <rniwa> | Domenic: one weird thing is about |
| 03:42 | <rniwa> | Domenic: a bunch of attributes on Text |
| 03:42 | <rniwa> | Domenic: because WebKit still has child nodes on Attr we kind of have to have CEReactions on them |
| 03:42 | <rniwa> | not sure if you want to note that or not |
| 03:43 | <Domenic> | oh wow that's icky |
| 03:43 | <rniwa> | there is a lot of subtitles there. |
| 03:43 | <Domenic> | I guess worth noting in the bug and we can think about how to phrase it tomorrow |
| 03:43 | <rniwa> | sublities* |
| 03:43 | <rniwa> | a lot of subtitles LOL |
| 03:43 | <Domenic> | you should just remove child nodes of Attr :) |
| 03:44 | <Domenic> | shave all the yaks |
| 03:44 | <rniwa> | Domenic: maybe one day |
| 03:45 | <rniwa> | Domenic: e.g. I just found some apple products rely on getPropertyCSSValue to exist |
| 03:45 | <rniwa> | Domenic: so we can't get rid of that method. |
| 07:35 | <nox> | Does that mean that in WebKit, attributes can have children? |
| 07:37 | <Ms2ger> | The one Text node, probably |
| 07:38 | <nox> | Ms2ger: What Text node? oO |
| 07:38 | <Ms2ger> | The spec called for the Attr to have children representing the value of the attribute |
| 07:39 | <Ms2ger> | Like if you had foo="a b", that would be [Text(a), EntityReference(nbsp), Text(b)] |
| 07:40 | <Ms2ger> | But browsers never implemented EntityReference, so we ended up with Attr always having a single Text child |
| 07:43 | <nox> | Ms2ger: I'm disgusted. |
| 08:02 | <jgraham> | nox: Well done, you passed the DOM Developer test |
| 08:03 | <nox> | jgraham: Ah ah. |
| 08:08 | <MikeSmith> | haha |
| 08:15 | <nox> | I mean, I was already disgusted by Attr: Node, |
| 08:15 | <nox> | but I didn't know it actually has a child. |
| 08:16 | <annevk> | mathiasbynens: aaah, while walking this morning I had exactly this as title "Premature Polyfills" for a theoretical blogpost |
| 08:16 | <Ms2ger> | nox, I ripped that out from Gecko at some point |
| 08:16 | <annevk> | mathiasbynens: guess you'll have to write one now |
| 08:16 | <nox> | Ms2ger: Interesting. |
| 08:17 | <nox> | jgraham: Don't quote me on this, |
| 08:17 | <nox> | jgraham: but document.write should be a matter of weeks at most in Servo. :) |
| 08:24 | <annevk> | JakeA: https://twitter.com/jaffathecake/status/791192371954343940 is pretty much Twitter |
| 08:25 | <JakeA> | hah, I suppose it is |
| 08:28 | <MikeSmith> | annevk: I am curious to hear more about the "Premature Polyfills" thing |
| 08:28 | <MikeSmith> | mathiasbynens brought up something earlier |
| 08:28 | <MikeSmith> | ? |
| 08:28 | Ms2ger | quotes nox |
| 08:29 | <JakeA> | MikeSmith: I was about to ask Domenic to clarify some stuff around that, but just catching up with the thread |
| 08:29 | <nox> | Ms2ger: What? |
| 08:29 | <nox> | Oh. :) |
| 08:29 | <mathiasbynens> | MikeSmith: https://github.com/jonathantneal/document-promises/issues/4 |
| 08:29 | <MikeSmith> | annevk: I vaguely recall we talked face to face some about problems with polyfills but cannot remember the details.. |
| 08:29 | <MikeSmith> | mathiasbynens: thanks, looking now |
| 08:31 | <mathiasbynens> | MikeSmith: some terminology discussion in replies here https://twitter.com/mathias/status/791003447856394241 TL;DR everyone has their own definition of shim |
| 08:31 | MikeSmith | looks there too |
| 08:33 | <MikeSmith> | ah, @WebReflection .. |
| 08:40 | <nox> | Holy hell that thread. |
| 08:42 | <MikeSmith> | haha |
| 08:42 | <nox> | What the hell is a stage-0 transpiler? |
| 08:42 | <MikeSmith> | yeah I am still rading through it |
| 08:43 | <annevk> | I suspect stages refer to TC39's process |
| 08:44 | <MikeSmith> | I have trouble catching a few of @WebReflection’s references |
| 08:44 | <MikeSmith> | but I guess I’m not going to let it worry me |
| 08:45 | <MikeSmith> | anyway, glad to see Andrew Betts weighing in there |
| 08:55 | <JakeA> | nox: eg https://goo.gl/Is18qq |
| 09:11 | <jgraham> | nox: I am totally quoting you on that :) |
| 09:12 | annevk | sets calendar reminder |
| 09:20 | <zcorpan> | well that's... also a way to parse HTML... https://www.browserstack.com/screenshots/9b8c537e3590fbe42688ed29b3101dab6a2dec52/win7_ie_8.0.png |
| 09:20 | <zcorpan> | (edge 13: https://www.browserstack.com/screenshots/9b8c537e3590fbe42688ed29b3101dab6a2dec52/win10_edge_13.0.png ) |
| 09:22 | <Ms2ger> | So, @font-face |
| 09:22 | <Ms2ger> | What's the consensus on whether to sniff or not? |
| 09:23 | <zcorpan> | Ms2ger: back when we added that in presto, i think everybody completely ignored content-type (and there was no correct mime type for fonts anyway) |
| 09:23 | <zcorpan> | not sure if that has changed since |
| 09:24 | <zcorpan> | well now there are official mime types |
| 09:24 | <Ms2ger> | https://mimesniff.spec.whatwg.org/#sniffing-in-a-font-context says something about XML types |
| 09:26 | <Ms2ger> | GPHemsley, ^ |
| 09:31 | <zcorpan> | https://cs.chromium.org/chromium/src/third_party/WebKit/Source/platform/MIMETypeRegistry.cpp?dr=C&sq=package:chromium&rcl=1477451763&l=122 - not sure if that gets called from @font-face though (or just link preload stuff?) |
| 09:32 | <mathiasbynens> | nox: https://babeljs.io/docs/plugins/preset-stage-0/ |
| 09:32 | <Ms2ger> | TextCaseInsensitive, huh |
| 09:35 | <zcorpan> | Ms2ger: i suppose that's a bug |
| 09:36 | <Ms2ger> | XCTO: nosniff does not seem to apply to fonts in Gecko |
| 09:41 | <Ms2ger> | I can't follow the Chrome code |
| 09:50 | <annevk> | Ms2ger: I think Chrome might only apply it to script/style and Gecko might do that too at the moment |
| 09:51 | <annevk> | Ms2ger: note that MIMESniff doesn't define XCTO |
| 09:53 | <Ms2ger> | Fetch does, kinda |
| 09:53 | <Ms2ger> | Which I assume you know too :) |
| 09:53 | <annevk> | Yeah, Fetch does |
| 09:54 | <annevk> | Has one open issue about maybe reducing the scope (no longer letting it apply to images) |
| 10:05 | <Ms2ger> | I wonder if it would be easier to move that out of fetch and just have a "don't sniff me" flag on the response |
| 10:05 | <Ms2ger> | Easier to miss, I guess |
| 10:09 | <annevk> | Ms2ger: wouldn't even work for script/style |
| 10:09 | <Ms2ger> | Oh? |
| 10:10 | <annevk> | Ms2ger: I guess you could add logic there to start using that |
| 10:10 | <annevk> | Ms2ger: but currently neither really sniffs |
| 10:12 | <Ms2ger> | It's really "enforce-my-content-type-for-real" more than "nosniff" |
| 10:13 | <annevk> | "enforce-my-content-type-for-real-if-you-happen-to-be-a-style-or-script-context |
| 10:13 | <annevk> | " |
| 10:15 | <Ms2ger> | Uhuh |
| 10:17 | <Ms2ger> | There's a reason I wanted a spec before getting another impl... |
| 11:46 | <littledan__> | why is the HTML spec seemingly accepting of other, non-JS scripting languages? |
| 11:47 | <littledan__> | with wording like, " Scripts (typically in JavaScript)" |
| 11:47 | <littledan__> | why not just say, JS is the language here for HTML |
| 11:55 | <Ms2ger> | littledan__, if you file an issue, it'll probably be removed |
| 12:18 | <littledan__> | Filed at https://github.com/whatwg/html/issues/1978 ; maybe I will make a PR |
| 12:50 | <annevk> | foolip_sheriff: I created a word-diff but it's hardly readable |
| 12:51 | <annevk> | foolip_sheriff: and a more normal diff isn't much better due to the different wrapping |
| 13:22 | <annevk> | TabAtkins: bikeshed/Shepherd still has the old https://w3c.github.io/ServiceWorker/spec/service_worker/ links |
| 13:50 | <JonathanNeal> | Well, that was fun. Hope everyone is well. |
| 14:32 | <foolip> | annevk: have to run now, but did you not find git diff --word-diff -w readable in the terminal at least? |
| 16:09 | <TabAtkins> | tobie: What do you mean by "hidden inputs"? Algorithms should ideally declare all their inputs visibly. |
| 16:16 | <tobie> | TabAtkins: agreed. They're semi-globals (i.e. shared between a number of algorithms). I'm planning to refactor them at some point, but it's a significant amount of not super useful work for now. |
| 16:17 | <tobie> | TabAtkins: sticking them within in Ignore Vars works for now but has the downsides that: |
| 16:17 | <tobie> | TabAtkins: 1) it affects all algorithms |
| 16:18 | <tobie> | TabAtkins: 2) it's not clear which algorithm rely on them being in Ignore Vars |
| 16:18 | <TabAtkins> | No need to refactor, just call them out with a quick sentence in the algo intro saying "This also depends on the global values |foo| and |bar|" |
| 16:18 | <tobie> | TabAtkins: as I said, though, this is somewhat of an edge case (large spec moved to BS) |
| 16:23 | <tobie> | TabAtkins: well, except that these variables are used across the spec, outside of algorithms, to mean different things, so you really have to point to the right instance of that (which means adding a hook, etc.) |
| 16:26 | <tobie> | TabAtkins: iirc it was the |A| in https://heycam.github.io/webidl/#es-maplike |
| 16:27 | <tobie> | TabAtkins: so maybe it's actually pretty trivial to fix your way |
| 17:21 | <annevk> | TabAtkins: should I file a bug on Shepherd indexing the wrong Service Workers URL? |
| 17:22 | <annevk> | TabAtkins: or is bikeshed simply not synced yet? |
| 17:41 | <annevk> | Sigh, did Aryeh close all outstanding WPT PRs again? |
| 17:43 | <annevk> | Oh, https://trac.webkit.org/changeset/207162 |
| 20:46 | <TabAtkins> | annevk: Fixed the Shepherd url. |
| 21:00 | <KiChjang> | when should you use XHR and when should you use the Fetch API?> |
| 21:02 | <TabAtkins> | fetch() doesn't yet let you cancel your request (waiting on cancelable promises or equivalent). I think that's probably about the only reason you'd want to still use XHR. |
| 21:06 | <KiChjang> | so in the future, XHR would be deprecated in favour of fetch? |
| 21:06 | <KiChjang> | (when cancelable promises becomes production-ready) |
| 21:06 | <KiChjang> | i believe Domenic is working hard on that |
| 21:11 | <Domenic> | What does "deprecated" mean to you, KiChjang? :) |
| 21:11 | <KiChjang> | Domenic, ah, good point |
| 21:12 | <KiChjang> | i was thinking that maybe XHR would be removed from the spec entirely, or just have a stub implementation of fetch |
| 21:12 | <Domenic> | that seems bad |
| 21:12 | <TabAtkins> | That would break a lot of legacy websites for no real gain |
| 21:13 | <KiChjang> | then again, this doesn't happen if implementors don't actually implement it |
| 21:13 | <KiChjang> | the web isn't quite receptive to code refactorings :/ |
| 22:08 | <MikeSmith> | anybody know, is the “new URLParser” in WebKit implemented from the URL spec? |
| 22:14 | <MikeSmith> | I guess it must be, from looking at, e.g., https://bugs.webkit.org/show_bug.cgi?id=160811#c10 |
| 22:14 | <MikeSmith> | which mentions “the spec” |
| 22:16 | <MikeSmith> | but since it also says “more standards-compliant URL parser”, curious to know in what ways it doesn’t conform to the spec and why |
| 22:25 | <Domenic> | When my Safari testing machine finishes installing the new macOS I can try running the web platform tests |
| 22:25 | <Domenic> | I'm like 4 hours in right now, gonna leave it overnight and if that fails then take it to tech support |
| 22:27 | <MikeSmith> | wow |
| 22:29 | <MikeSmith> | in the past when I had problems with MacOS installs taking that long, it seemed to be due to it tripping over stuff installed by MacPorts |
| 22:29 | <MikeSmith> | and maybe homebrew too |
| 22:29 | <MikeSmith> | but I no longer have that problem |
| 22:30 | <MikeSmith> | or at least didn’t with the Sierra install |
| 22:30 | <MikeSmith> | though I no longer have any MacPorts packages installed |
| 22:31 | <MikeSmith> | now just have homebrew ones |
| 22:32 | <MikeSmith> | anyway about the Webkit URLparser differences, I guess I am more interested in hearing from them in what cases they intentionally made it non-conformant, and why |
| 22:32 | <MikeSmith> | pinged hober and sam on twitter to ask |
| 22:32 | <Domenic> | There are a couple issues on whatwg/url where they are actively still discussing |
| 22:32 | <Domenic> | So presumably those |
| 22:32 | <MikeSmith> | ah OK |
| 22:32 | <Domenic> | maybe more though, dunno |
| 22:32 | MikeSmith | looks at the issue tracker |
| 22:33 | <MikeSmith> | ah I see https://github.com/achristensen07 comments |
| 22:34 | <MikeSmith> | only open one I see is https://github.com/whatwg/url/issues/150 |
| 22:34 | <MikeSmith> | unicode fragments |
| 23:11 | <MikeSmith> | ah about URL parsing I see https://github.com/whatwg/url/issues/87#issuecomment-256444393 is open still too and actively in discussion |
| 23:19 | <Domenic> | Today I learned I don't understand what document.close() even does |
| 23:26 | <MikeSmith> | Domenic: how so? |
| 23:27 | <Domenic> | I mean I read the spec steps and I don't really understand what effects they would have |
| 23:47 | <KiChjang> | Domenic, are there any JS libraries out there that actually utilizes document.open()? |
| 23:47 | <Domenic> | who knows |
