| 00:10 | <rbyers> | annevk Tobie: chromium bug tracker has an API here: https://chromium.googlesource.com/infra/infra/+/master/appengine/monorail/doc/api.md |
| 00:11 | <rbyers> | annevk: documentation on crafting a URL for pre-filling a new bug form is here: https://www.chromium.org/issue-tracking/creating-a-bug-template-url |
| 00:12 | <rbyers> | That's what we use on chromestatus.com for filing a bug - really simple to use and just one extra click to confirm the bug submission. |
| 00:15 | <rbyers> | Note that you don't need edit-bugs to set arbitrary labels when using the URLs - so you can just auto-generate appropriate links that fill in everything and anyone can click, login with a Google account, and submit. |
| 06:47 | <MikeSmith> | gsnedders: I have no objection to extending the checker to add CSS support, if somebody else wanted to do the work. I have no plans to work on it myself. |
| 08:06 | <annevk> | Domenic: rbyers: is there a more specific component than Blink for URL API issues? https://bugs.chromium.org/p/chromium/issues/detail?id=682143 |
| 08:07 | <annevk> | rbyers: thanks btw |
| 08:10 | <annevk> | Domenic: rbyers: found Blink>Network, oops |
| 08:21 | <hsivonen> | hmm. Travis jobs no longer starting on my pushes to encoding/visualization |
| 08:22 | <hsivonen> | TabAtkins: Thanks. should I have found a mention of the magic move of <link> in bikeshed docs? |
| 08:24 | <annevk> | hsivonen: looking into Travis |
| 08:34 | <annevk> | hsivonen: the narrow Python build stuff is causing massive amount of NFC errors again |
| 08:34 | <annevk> | hsivonen: I suspect that might be why Travis gave up |
| 08:34 | <annevk> | hsivonen: log got too large or something |
| 08:40 | <hsivonen> | annevk: weird. I diffed the output and there wasn't supposed to be a difference |
| 08:41 | <annevk> | hsivonen: maybe because Travis uses a different Python? |
| 08:41 | <annevk> | I'm surprised I never hit this problem with the Python tools I wrote for the Unicode standard |
| 08:42 | <annevk> | Because they also output characters and some of those are > 0xFFFF I think |
| 08:42 | <hsivonen> | annevk: which travis logs are you looking at |
| 08:43 | <annevk> | hsivonen: https://travis-ci.org/whatwg/encoding/builds/192959422 |
| 08:44 | <annevk> | hsivonen: but it seems to have passed, since the NFC stuff is only a warning |
| 08:44 | <annevk> | hsivonen: it does seem that rebooting Travis here and there made things working again though for whatwg/encoding |
| 08:46 | <hsivonen> | annevk: so weird that the change I made interferes in any way with anything other than big5 |
| 08:47 | <annevk> | hsivonen: it might be that you never fixed the NFC issue since it's not fatal |
| 08:47 | <hsivonen> | annevk: good point! |
| 08:48 | <annevk> | MikeSmith: I wonder if we should configure the checker for specs to be fatal for warnings |
| 08:48 | <annevk> | MikeSmith: perhaps with some flag for opt-out if we have any specs today with "legit" warnings |
| 08:48 | <hsivonen> | annevk: well, first we should consider if we *want* this NFC warning to be fatal to this stuff |
| 08:49 | <hsivonen> | fwiw, I just tried using the new visualization for encoding_rs development, and they are much nicer to use than the old ones |
| 08:49 | <hsivonen> | I so regret not making them this nice earlier |
| 08:50 | <annevk> | hsivonen: sure, but in general I wouldn't want non-NFC |
| 08:51 | <hsivonen> | I think at this point, it's necessary to figure out what exactly the NFC violations are |
| 08:53 | <MikeSmith> | annevk: I guess I would need to add a new option to the command-line checker to make it exit non-zero if there are warnings |
| 08:54 | <MikeSmith> | I think right now the way I have it set up it exits non-zero if there are any errors but exits zero if there are only warnings (or no messages at all) |
| 08:54 | <MikeSmith> | but it is doable |
| 08:54 | <annevk> | We should probably have it for Bikeshed too since I just noticed again with URL that I almost landed something with Bikeshed errors |
| 08:54 | <hsivonen> | annevk: the PUA warnings we shouldn't fix by changing the visualizations |
| 08:54 | <annevk> | Bikeshed LINK ERRORS that is, which are not fatal |
| 08:55 | <hsivonen> | remaining NFC is all CJK |
| 08:55 | <hsivonen> | I wonder if it's about compatibility ideographs or something |
| 08:55 | hsivonen | looks up the definition of NFC |
| 08:55 | <annevk> | hsivonen: a small worry I have is that some warnings are okay while others are not, but I guess we can check the logs whenever something sensitive is being touched |
| 09:02 | <hsivonen> | annevk: so. the first case I happened to look at is the ångström sign... |
| 09:05 | <hsivonen> | annevk: the next one is a compatibility ideograph |
| 09:08 | <hsivonen> | annevk: so I only solved the non-NFCness of combining characters but not singletons |
| 09:09 | <hsivonen> | still trying to find spec text saying that compatibility ideographs are singletons |
| 09:11 | <hsivonen> | still failing to find spec text, but Wikipedia and the W3C i18n wiki suggest that compatibility ideographs are singletons |
| 09:12 | <hsivonen> | annevk: considering the use cases for these tables, I don't think we should normalize away singleton distinctions for clipboard and search operations |
| 09:13 | <hsivonen> | for PUA, there's even the use case of seeing if they match something interesting in system fonts |
| 09:13 | <hsivonen> | e.g. on Ubuntu, PUA on the last row of gbk gets radical glyphs, which is an interesting finding |
| 09:14 | <hsivonen> | (hmm. should Radical Glyphs be added to WHATWG band names?) |
| 09:14 | <annevk> | (yes) |
| 09:18 | <hsivonen> | (done) |
| 09:55 | <tobie> | Thanks rbyers. tracking here: https://github.com/heycam/webidl/issues/267 |
| 10:43 | <nox> | jgraham, hsivonen: Around? Have some questions about HTML and script.onerror. |
| 10:48 | <jgraham> | nox: I'm here, but I don't know how much I know about this |
| 10:49 | <nox> | jgraham: https://github.com/w3c/web-platform-tests/blob/master/html/semantics/scripting-1/the-script-element/script-onerror-insertion-point-2.html |
| 10:49 | <nox> | jgraham: I think that test doesn't agree with the spec. |
| 10:50 | <nox> | Because at this point, the only way to make .write do nothing is to use https://html.spec.whatwg.org/multipage/webappapis.html#document.write():ignore-destructive-writes-counter and it is not incremented when parsing a script with an invalid src attribute. |
| 10:53 | <Ms2ger> | Is the intention of the test that .write() does nothing? |
| 10:53 | <Ms2ger> | I thought it was that .write() replaces the page |
| 10:56 | <jgraham> | nox: (it seems like bz was involved with the bug, so my prior is strong for the test being correct) |
| 10:57 | <jgraham> | write shouldn't do nothing. The only thing that isn't clear to me is whether it's certain that error will fire after the document finishes loading, which seems to be required for the pass condition |
| 10:57 | <nox> | Ms2ger: If it replaces the page, then all the elements are removed from the document, |
| 10:57 | <nox> | and thus document.write won't write in the parent. |
| 10:57 | <jgraham> | It doesn't write in the parent? |
| 10:58 | <Ms2ger> | Are we looking at the same test? |
| 10:58 | <nox> | Err oh right, misread the support file. |
| 10:59 | <nox> | jgraham: It can't be fired after the document finished loading AFAICT. |
| 10:59 | <nox> | jgraham: I'm pretty sure this is about the counter I linked. |
| 11:00 | <yoav> | annevk: Friendly ping RE https://github.com/whatwg/fetch/pull/442 :) |
| 11:02 | <annevk> | yoav: it's blocked on the test |
| 11:02 | <annevk> | yoav: tests get merged first |
| 11:02 | <jgraham> | nox: So, the pass condition seems to be that the iframe document is entirely replaced by the string "text" |
| 11:02 | <nox> | AH |
| 11:02 | <nox> | Ok. |
| 11:03 | <yoav> | annevk: missed your comment on the test PR. Addressing |
| 11:04 | <annevk> | Cool, about to take a break, will have time later I think |
| 11:04 | <nox> | jgraham: Ok, next question. :) |
| 11:05 | <jgraham> | … |
| 11:05 | <nox> | jgraham: What is https://github.com/w3c/web-platform-tests/blob/master/old-tests/submission/Opera/script_scheduling/119.html about? Isn't there a race condition there? |
| 11:06 | <nox> | AFAIK the script when included in the document becomes an asap in order script, and there is nothing that forbids these to be executed before DOMContentLoaded is fired, right? |
| 11:07 | <jgraham> | nox: I'll read it, but I'm about 98% sure from what you just said that the spec changed after this test was written |
| 11:08 | <nox> | jgraham: Ok. |
| 11:10 | <jgraham> | (although gecko seems to pass the test reliably) |
| 11:11 | <nox> | Yeah. |
| 11:11 | <jgraham> | (also Chrome) |
| 11:12 | <nox> | Yeah all of them, I just don't understand why it's supposed to work. |
| 11:15 | <nox> | jgraham: https://html.spec.whatwg.org/multipage/syntax.html#the-end Is step 5 supposed to wait for step 4 to finish? |
| 11:16 | <hsivonen> | nox: I take it that jgraham already covered what you wanted to ask me? |
| 11:16 | <nox> | hsivonen: Yep, thanks. |
| 11:16 | <nox> | Err, is step 5 supposed to be run after the event is fired, or after a task has been queued to fire the event? |
| 11:17 | <jgraham> | nox: After the event is fired |
| 11:17 | <nox> | Ok. |
| 11:17 | <jgraham> | It doesn't say "queue a task to" |
| 11:17 | <nox> | jgraham: "Queue a task to run the following substeps:" |
| 11:17 | <jgraham> | Oh, I can't read |
| 11:18 | <jgraham> | OK, it does |
| 11:18 | <jgraham> | But the reaon this test should pass is that the script takes 1s to load and it's assumed that 1s is enough time to finish parsing the document and dispatch a DOMContentLoaded event |
| 11:19 | <jgraham> | If servo doesn't manage that, you probably want to do some perf work :) |
| 11:19 | <nox> | Mmmh. |
| 11:19 | nox | compiles in release mode. |
| 11:20 | <jgraham> | If you need to increase the delay a bit for the test to be stable in debug mode that's OK, but I think that it's a reasonable test in general |
| 11:21 | <nox> | jgraham: Another question, |
| 11:21 | <nox> | jgraham: why do we call t.done() in the DOMContentLoaded event handler? |
| 11:22 | <jgraham> | hah |
| 11:22 | <jgraham> | That is totally broken |
| 11:22 | <nox> | Ok. :) |
| 11:22 | <nox> | Will reindent and fix that. |
| 11:22 | <nox> | The indentation in that file is post-modern art. |
| 11:24 | Ms2ger | blames jgraham |
| 11:25 | jgraham | blames emacs or hallvord |
| 11:32 | <smaug> | foolip: you might know, why blink has different CSS for dialog::backdrop than what the spec has |
| 11:33 | <smaug> | https://cs.chromium.org/chromium/src/third_party/WebKit/Source/core/css/html.css?q=html.css&sq=package:chromium&dr&l=1090 vs https://html.spec.whatwg.org/#flow-content-3 |
| 11:34 | <nox> | jgraham: I wonder if the trickle even matters. |
| 11:34 | <nox> | jgraham: "the end" should execute just afterwards the </script>, and the parser never gets suspended, so supposedly no event had the opportunity to be processed, right? |
| 11:37 | <nox> | Oh but the asap in order script can get executed between the moment DOMContentLoaded is queued and the moment it is fired, I guess. |
| 11:37 | <jgraham> | Yeah |
| 11:38 | <nox> | jgraham: I guess I have a bigger problem given trickle(d5) didn't help. :) |
| 11:38 | <zcorpan> | annevk: what is intended to go into branch-snapshots/ ? with hsivonen's PR it's only visualization/ in that directory, the rest is under commit-snapshots |
| 11:38 | <jgraham> | nox: Hurrah, testing worked! :) |
| 11:40 | <zcorpan> | annevk: never mind, it makes sense to me now |
| 12:07 | <foolip> | smaug: I have noticed this too while working on Fullscreen. It includes some of https://fullscreen.spec.whatwg.org/#user-agent-level-style-sheet-defaults |
| 12:07 | <nox> | jgraham: I was just... waiting for asap scripts to process the deferred ones. |
| 12:08 | <nox> | Obviously that wasn't going to make that test pass. |
| 12:08 | <smaug> | foolip: ah, ok |
| 12:08 | <foolip> | smaug: I thought I added a TODO somewhere, but maybe I didn't after determining the difference shouldn't be observable. It is possible that something is still observably wrong here. |
| 12:09 | <foolip> | let me know if you find it :) |
| 12:10 | <jgraham> | nox: :) |
| 12:24 | <zcorpan> | hsivonen: "Private Use, code point follows immediately the code point of previous pointer" and "Four bytes in UTF-8, code point follows immediately the code point of previous pointer" appear to be unused |
| 12:25 | <hsivonen> | shouldn't be |
| 12:25 | hsivonen | looks |
| 12:26 | <hsivonen> | zcorpan: for the former, see bottom right of https://encoding.spec.whatwg.org/branch-snapshots/visualization/gb18030.html |
| 12:26 | <hsivonen> | zcorpan: for the latter, see 5 down and 5 right from the first mapped cell at https://encoding.spec.whatwg.org/branch-snapshots/visualization/big5.html |
| 12:28 | <zcorpan> | hsivonen: ah. i grep'ed for the same order of the classes as they were in index.html |
| 12:31 | <zcorpan> | hsivonen: did you check color contrast between text and background? "Private Use, code point follows immediately the code point of previous pointer" fails in http://webaim.org/resources/contrastchecker/ |
| 12:31 | <hsivonen> | zcorpan: I did not |
| 12:32 | <hsivonen> | zcorpan: would white text on that bg pass? |
| 12:32 | <zcorpan> | hsivonen: passes AA, fails AAA |
| 12:33 | <hsivonen> | zcorpan: I'll flip the text color then |
| 12:34 | <hsivonen> | zcorpan: done |
| 12:35 | <zcorpan> | 👍 |
| 13:06 | <quarky_> | Are any of these users bots? |
| 13:06 | <quarky_> | ... |
| 13:06 | <quarky_> | this is a dead chat isn't it |
| 13:06 | <quarky_> | hahahahaha |
| 13:06 | <quarky_> | probably not |
| 13:07 | <quarky_> | What's really sad is I came to ask a question but now I am talking to myself. |
| 13:07 | <No_se> | Just ask |
| 13:07 | <quarky_> | no se |
| 13:07 | <quarky_> | how many of you ppl here have read the https://html.spec.whatwg.org/print.pdf |
| 13:09 | <quarky_> | also |
| 13:09 | <No_se> | I am not taking part in user surveys. You probably should go and ask a real question. So what bothers you? |
| 13:10 | <quarky_> | is the w3c affiliated in any form with whatwg |
| 13:10 | <quarky_> | html |
| 13:10 | <quarky_> | its different than i remember it |
| 13:10 | <quarky_> | so many more parameters you can use |
| 13:10 | <quarky_> | the syntax hasn't change has it? |
| 13:11 | <annevk> | quarky_: W3C isn't really affiliated, but there's overlap in the groups of people |
| 13:11 | <annevk> | quarky_: syntax has definitely changed due to the addition of new elements (if we're talking relative to HTML4) |
| 13:11 | <quarky_> | okay ty @annevk |
| 15:21 | <annevk> | jgraham: MikeSmith: for https://github.com/w3c/web-platform-tests/pull/4556 no OWNERS were identified |
| 15:21 | MikeSmith | wakes up |
| 15:23 | <gsnedders> | annevk: I presume tobie's bot didn't run for whatever reason |
| 15:24 | <MikeSmith> | yeah seems so |
| 15:24 | <MikeSmith> | given the fetch/OWNERS file is still there |
| 15:25 | <gsnedders> | didn't get any labels either |
| 15:25 | <MikeSmith> | ah |
| 15:25 | <MikeSmith> | yeah so definitely something borked tehre |
| 15:26 | <MikeSmith> | the admin UI at https://github.com/w3c/web-platform-tests/settings/hooks isn’t showing any red |
| 15:27 | <MikeSmith> | I wonder is tobie bot https://wpt-ganesh.herokuapp.com/github-hook |
| 15:27 | <MikeSmith> | if so there are no indications there of any failures |
| 15:28 | <tobie> | what's up? |
| 15:29 | <gsnedders> | tobie: https://github.com/w3c/web-platform-tests/pull/4556 |
| 15:29 | <gsnedders> | tobie: has no labels, nor owners comment |
| 15:30 | <tobie> | looking at the logs |
| 15:32 | <tobie> | weird |
| 15:33 | <tobie> | gsnedders: https://gist.github.com/tobie/45bf422175a78d536e1ea608b602fd38 |
| 15:35 | <annevk> | moz://a ugh |
| 15:36 | <annevk> | I guess at least with https://github.com/whatwg/url/pull/185 it won't parse into moz://a/ |
| 15:36 | <annevk> | mozillai |
| 15:37 | <annevk> | I suppose I should add a test specifically for that input |
| 15:37 | <annevk> | Or example |
| 15:38 | <tobie> | gsnedders: something didn't work on that occasion. |
| 15:39 | <tobie> | gsnedders: I don't know what as my logging is limited |
| 15:39 | <tobie> | gsnedders: can you ping me if you see more issues? |
| 15:40 | <gsnedders> | tobie: idk, annevk noticed it aboved, I just passed it on :) |
| 15:44 | <tobie> | gsnedders: found it |
| 15:44 | <tobie> | https://www.irccloud.com/pastebin/bhDj5y90/ |
| 15:46 | <jgraham> | annevk: Beautiful, isn't it |
| 15:46 | <annevk> | jgraham: very classy |
| 15:47 | <annevk> | Oh and all the protocol references. It's just the scheme folks... |
| 15:52 | <nox> | annevk: Ah ah. |
| 15:52 | <nox> | annevk: The video is self-trolling, in case you didn't see. |
| 15:53 | <nox> | annevk: https://twitter.com/IAmNotCyril/status/821714771514900480 |
| 15:53 | <nox> | "Me in front of the new logo" |
| 15:55 | <annevk> | heh |
| 15:56 | <annevk> | Oh well, we also landed a search box in <select> today, so I'm still good |
| 15:56 | <nox> | annevk: And I refactored script loading without making any more tests fail, woo. |
| 16:02 | <TabAtkins> | hsivonen: I mean, link in body works just fine. I only move it for validation purposes. |
| 16:11 | <MikeSmith> | the checker allows link in body these days |
| 16:12 | <MikeSmith> | well not for all cases but in line with what the spec now allows |
| 16:12 | <MikeSmith> | and what the spec allows for link has changed significantly over the last several months |
| 16:20 | <JakeA> | Anyone running Windows, have Chrome Canary, & not mind running a test that'll 100% the CPU for up to 10 mins? |
| 16:20 | <boogyman> | win 8.1 |
| 16:21 | <JakeA> | boogyman: that'll do! http://output.jsbin.com/jirinuf/quiet - the results will appear in the console. It's finished when it logs out 10 values |
| 16:23 | <boogyman> | only using 15 - 40% cpu |
| 16:24 | <JakeA> | probably single thread |
| 16:29 | <boogyman> | completed, do you want the output data? |
| 16:30 | <MikeSmith> | No_se: is it OK if I close https://github.com/validator/validator/issues/446? |
| 16:31 | <JakeA> | boogyman: please! Just the final console log with all the numbers |
| 16:32 | <boogyman> | there are two "final" logs, one is an array of arrays (paired numbers), the other is a single log with the second of the paired numbers delimited with new lines |
| 16:32 | <boogyman> | delimited by* |
| 16:32 | <JakeA> | boogyman: the newline one please! |
| 16:33 | <boogyman> | send in msg to avoid clutter here |
| 16:33 | <No_se> | MikeSmith: Yes, of course. I was wrong, not the validator. |
| 16:33 | <MikeSmith> | well you were not wrong |
| 16:34 | <MikeSmith> | I still wish I could improve the checker message for that case, but that is a bigger issue |
| 16:34 | <No_se> | I found it in the spec and: Whether I was wrong or not is not part of the github issue :-) |
| 16:35 | <MikeSmith> | OK, fair enough |
| 16:38 | <No_se> | The other validator linked from whatwg "CSE…" found the issue as well, and even shows a useful hint. |
| 16:41 | <No_se> | I don't know how the validator generates the messages, but the mentioned messages shows a possible approach (if performance is not a problem): 'The "ul" element cannot be contained in a "p" element. The end tag for "p" was found on line 289, column 2 (this error may also be caused by a stray or misplaced end tag).' |
| 16:42 | <No_se> | So, an extra run searching for closing </p> when implicitly closing it |
| 17:10 | <MikeSmith> | hmm yeah that wording is pretty much how we also ideally should be reporting it in the Nu checker |
| 17:27 | <MikeSmith> | botie, inform zcorpan, for the style[type] should the checker warn if the value is anything other than text/css? or maybe even warn if style[type] is specified at all, warn: “It is not necessary to specify the type attribute for the style element if the contents of the element are a CSS stylesheet.”? |
| 17:27 | <botie> | will do |
| 17:28 | <tantek> | I have a historical question about video/audio elements, what were the reasons against (because someone must have proposed it) a media element (in addition or instead), e.g. so you could do <media><source src="...mp4"/><source src="...mp3"/> </media> (actual use-case for music videos where you want to provide an audio-only track as an alternative) |
| 17:35 | <annevk> | tantek: we did not want another <object> |
| 17:36 | <annevk> | tantek: the UI and API are different, so therefore distinct elements |
| 17:36 | <tobie> | gsnedders, annevk: fixed wpt-bot issues |
| 17:38 | <MikeSmith> | tantek: https://lists.w3.org/Archives/Public/public-whatwg-archive/2007Feb/thread.html#msg487 may have some elaboration on the rationale |
| 17:38 | <MikeSmith> | or other threads in the months before that |
| 17:39 | <MikeSmith> | first few months of 2007 |
| 17:39 | <MikeSmith> | or maybe some at the end of 2006 |
| 17:39 | <MikeSmith> | hey <video> is nearly 10 years old I guess |
| 17:43 | <MikeSmith> | more from Hixie at https://lists.w3.org/Archives/Public/public-whatwg-archive/2007Mar/0587.html |
| 17:45 | <MikeSmith> | also https://lists.w3.org/Archives/Public/public-whatwg-archive/2007Mar/0558.html |
| 17:45 | <MikeSmith> | ah yeah, that one, https://lists.w3.org/Archives/Public/public-whatwg-archive/2007Mar/0558.html from Hixie |
| 17:46 | <MikeSmith> | see the section “ON HANDLING OTHER MEDIA FORMATS” |
| 17:47 | <MikeSmith> | and also a bit of earlier feedback from Hixie at https://lists.w3.org/Archives/Public/public-whatwg-archive/2007Mar/0437.html |
| 17:50 | <tantek> | thanks MikeSmith annevk |
| 17:55 | <MikeSmith> | tantek: fwiw when I want to do archaeology on stuff, I start at https://platform.html5.org/history/ to find out the date when it was added to the spec, then just start looking back through https://lists.w3.org/Archives/Public/public-whatwg-archive/ in the months around that date |
| 17:55 | tantek | cringes at any attempt at email archeology |
| 17:56 | <MikeSmith> | haha |
| 17:56 | <MikeSmith> | tantek: just pretend the e-mail archives are a different kind of wiki |
| 17:57 | <tantek> | MikeSmith: no, they are like badly organized wiki-diffs at best. no "current" page |
| 17:57 | <tantek> | MikeSmith, annevk the bit I don't get is that we already have various media API abstractions that both <audio> and <video> support. Almost implies that a generic <media> element could (still) work |
| 18:07 | <botie> | zcorpan, at 2017-01-18 17:27 UTC, MikeSmith said: for the style[type] should the checker warn if the value is anything other than text/css? or maybe even warn if style[type] is specified at all, warn: “It is not necessary to specify the type attribute for the style element if the contents of the element |
| 18:07 | <botie> | are a CSS stylesheet.”? |
| 18:09 | <MikeSmith> | tantek: IMHO from the fact that elements can share common API abstractions it does not follow that the markup mechanism for them should be a single element |
| 18:09 | <KevinMarks> | I still want a playlist element |
| 18:09 | <tantek> | MikeSmith, hence the caveats "almost" "implies" "could" work |
| 18:10 | <MikeSmith> | I think the pattern of subclassing or subtyping elements with a type attribute or whatever is a serious design error with all kinds of bad effects for authors |
| 18:10 | <tantek> | MikeSmith, like everything else, to be justified by real world use-cases. |
| 18:10 | <annevk> | tantek: they share infrastructure, but eg video forms a box, audio does not |
| 18:10 | <MikeSmith> | sure |
| 18:10 | <tantek> | annevk <audio controls> forms a box |
| 18:11 | <KevinMarks> | I know there are scars with SMIL , but it is a huge use case |
| 18:11 | <zcorpan> | the difference is <video> is able to render poster frame and has a place to draw video frames and subtitles |
| 18:14 | <tantek> | MikeSmith some of those emails - wow so much history of having to demonstrate how flawed all the old HTML+Time / SMIL stuff was |
| 18:14 | <tantek> | (and unadopted) |
| 18:14 | <zcorpan> | MikeSmith: i guess warning for not-text/css is useful. not sure about warning for type=text/css (or text/javascript for script)... had we discussed this for script? |
| 18:14 | <annevk> | tantek: sure but not for a moving image |
| 18:15 | <tantek> | why does that matter? |
| 18:15 | <annevk> | tantek: anyway, time to debate was April 2007 or so |
| 18:15 | <tantek> | also zcorpan, your poster frame distinction is ironic, given the well established use-case of songs being published to YouTube with nothing but *static* album art as the "video" |
| 18:15 | <tantek> | that is essentially, audio with a "poster" as it were |
| 18:16 | <zcorpan> | tantek: so use <video> |
| 18:16 | <MikeSmith> | tantek: yeah about SMIL in hindsight it seems like it should not have been necessary to explain but I think that was because the SMIL guy objected to creating something new |
| 18:16 | <tantek> | zcorpan wait are you saying <video poster="...jpg" src="...mp3" /> would work?!? |
| 18:17 | <tantek> | MikeSmith, these kinds of analyses are important to document, because in general, the bias of not creating something new (attempt re-use first) is correct |
| 18:17 | <zcorpan> | tantek: hmm ... not sure. but that's not what i said :-) |
| 18:17 | <tantek> | however, SMIL essentially failed incubation :P |
| 18:18 | <MikeSmith> | zcorpan: about warnings for type values for script I don’t remember having a discussion any time in the last few years at least |
| 18:18 | <tantek> | zcorpan, in that case, I think it's worth a test case. I see no reason why it shouldn't work. |
| 18:21 | <zcorpan> | tantek: http://software.hixie.ch/utilities/js/live-dom-viewer/saved/4816 seems like it works. i thought maybe the poster would disappear when it starts playing, but it doesn't in opera or firefox at least |
| 18:21 | <tantek> | that is awesome! |
| 18:21 | <tantek> | totally worth a blog post |
| 18:21 | <zcorpan> | same in safari tp |
| 18:23 | <zcorpan> | MikeSmith: https://github.com/whatwg/html/pull/443#issuecomment-168973862 |
| 18:23 | <tantek> | zcorpan++ nice choice of music too |
| 18:24 | <zcorpan> | tantek: heh, first result for "sample mp3" |
| 18:24 | <MikeSmith> | zcorpan: oh https://github.com/whatwg/html/pull/443#issuecomment-168973862 if the spec says “Authors should prefer omitting the attribute” I guess that is sufficient to justify the checker emitting a warning for it then |
| 18:25 | <zcorpan> | MikeSmith: yeah |
| 18:25 | <MikeSmith> | man, now reading “@sideshowbarker and I discussed this in #whatwg and we decided it's at least worth trying” |
| 18:25 | <zcorpan> | MikeSmith: i guess the spec should have consistent advice for <style> |
| 18:26 | <MikeSmith> | long term loss of short term memory |
| 18:26 | <MikeSmith> | zcorpan: yes |
| 18:26 | <tantek> | zcorpan, my #indieweb sources say that works however if you put the src="...mp3" on a <source> element inside it appears to not work |
| 18:26 | <MikeSmith> | will raise a PR for it |
| 18:26 | <zcorpan> | 6 months is not short term memory though :-D |
| 18:27 | <zcorpan> | tantek: i seriously doubt it without proof |
| 18:27 | <zcorpan> | gotta run though |
| 18:27 | <tantek> | zcorpan, yup, bug coming |
| 18:27 | <MikeSmith> | zcorpan: true! my memory sucks even worse than I remembered 😂 |
| 18:27 | <tantek> | also http://software.hixie.ch/utilities/js/live-dom-viewer/?saved=4816 seems useful enough to be added as a test to webplatform tests |
| 18:27 | <tantek> | is there a simple pipeline for Live DOM Viewer -> Web Platform Tests ? |
| 18:29 | <MikeSmith> | oh wow a spec issue raised by sof https://github.com/whatwg/fetch/issues/459 Don’t seem to see that very often (dunno why) |
| 18:31 | <tantek> | correction, works with <source> tag also. |
| 18:31 | <tantek> | (amazing what things work when you go to actually file a bug ;) ) |
| 18:32 | <annevk> | MikeSmith: you have to ask 😛 |
| 18:32 | <MikeSmith> | hah I knew there was some secret |
| 18:43 | <annevk> | JakeA: https://github.com/w3c/web-platform-tests/pull/4505 is there a way to return a network error from a service worker? |
| 18:45 | <JakeA> | annevk: I don't think so. Fetch rejects on network error |
| 18:46 | <JakeA> | annevk: I guess you could respondWith Response.error() - I've never tried that |
| 18:47 | <annevk> | JakeA: ah yes, that is the way |
| 18:47 | <JakeA> | annevk: https://fetch.spec.whatwg.org/#http-fetch 3.3.3 handles that |
| 18:48 | <annevk> | yeah, ta |
| 18:48 | <annevk> | I forgot we defined Response.error() |
| 18:48 | <annevk> | Or I did, I guess |
| 18:53 | <annevk> | Hmm, didn't get to non-special-URL hosts today nor to revamping whatwg.org |
| 18:54 | <annevk> | And several other things it seems |
| 18:54 | <annevk> | I was kinda on a streak with completing my daily todo list |
| 18:57 | <jyasskin> | The [SameObject] wording appears to ban using it on nullable attributes: "The [SameObject] extended attribute must not be used on anything other than a read only attribute whose type is an interface type or object." Was that intentional? |
| 18:58 | <annevk> | jyasskin: I think so, because it had to match [NewObject], but there are open issues about changing [SameObject] into something like [Cached] |
| 18:58 | <jyasskin> | annevk: Thx |
| 18:59 | <jyasskin> | https://github.com/heycam/webidl/issues/212 ++ |
| 19:48 | <tobie> | jyasskin: sent notes your way |
| 20:52 | smaug | wonders if it is really expected that dialog.showModal(); dialog.removeAttribute("show"); dialog.showModal(); add the element twice to the stack |
| 23:57 | <jyasskin> | Dumb question for Domenic and foolip re https://github.com/WICG/webusb/pull/67: What makes an interface a mixin? |
| 23:57 | tantek | briefly misread webusb as websub |
| 23:57 | <Domenic> | jyasskin: right now, convention + militant policing of usage. In the future, dedicated syntax |
| 23:57 | <tantek> | https://www.w3.org/TR/websub/ |
| 23:58 | <Domenic> | jyasskin: i.e., if your interface is marked [NoInterfaceObject] and always appears on the right-hand side of an `implements` statement, it is a mixin. |
| 23:59 | <jyasskin> | Domenic: Ah, ok. So only [NoInterfaceObject] things will be able to be `implements`ed, and it'll be impossible to extend the members of something that's implemented. And I guess that ability to extend the members is why not to implement non-mixins? |