00:37
<hayato>
Ms2ger: Sorry for breaking travis. I fixed. https://github.com/w3c/web-platform-tests/commit/12b780086f4a0d1d6ff5e6d588482ae04624e794
05:50
annevk
signed up for a week of TPAC
08:54
<zcorpan>
hmm https://twitter.com/zcorpan/status/744896721017257984 only 12 votes. maybe people don't understand the question?
08:57
<annevk>
zcorpan: too hard
10:51
<annevk>
JakeA: if we're going to do navigation transitions it'd be great if someone helped out refactoring the navigation algorithm to make it actually match browsers
11:27
<zcorpan>
MikeSmith: getting 502 for https://checker.html5.org/
11:44
<MikeSmith>
zcorpan: yeah was restarting it, should work now
11:44
<zcorpan>
ah ok, yep works now
11:53
<annevk>
hsivonen: thanks for the review, please only merge if there's a generated index with an-up-to-date date next time; or just leave that part to me
11:56
<annevk>
hsivonen: it didn't matter much since I fixed it with the second PR, so just a note for next time
12:12
<hallvors>
zcorpan: it is tricky indeed :)
12:12
<hallvors>
I've guessed something, but don't know
12:15
<hallvors>
annevk: quick XHR question.. #handle-response-end-of-file step 10 seems to guarantee that you'll always get a progress event when readyState is 4. I don't really see implementations doing that..?
12:16
<annevk>
hallvors: yeah, zewt and I determined that would be good, but I guess implementations never really got there
12:16
<annevk>
hallvors: maybe it's time to give up on that especially since XHR isn't really the API-to-be anyway
12:16
<annevk>
hallvors: file an issue?
12:16
<hallvors>
OK.
12:17
<hallvors>
Seems fine to me to drop that requirement, you get load and loadend and readystatechange ..
12:17
<annevk>
Yeah, the idea was that you could just listen to progress and be fine
12:17
<hallvors>
I don't think we really need more events here :)
12:17
<annevk>
So that you'd never get a progress bar that hangs at 80% or so
12:18
<annevk>
But at this point everyone probably learned not to do that...
12:44
<hallvors>
annevk: https://github.com/whatwg/xhr/issues/72
12:48
<annevk>
ta
13:06
<howdoi>
So, anyone can write a proposal on a gist/github/anyLink and it will be counted as stage-0 ?
13:07
<annevk>
howdoi: you need a TC39 member to champion it
13:07
<howdoi>
annevk: and on what factors will the championing depend on? (I know there might be many, but the cruz would be?)
13:08
<annevk>
howdoi: that someone thinks it's good?
13:08
<howdoi>
and how does on make it to TC39 community as a member, I do noticed that there are many who havn't made a proposal or their proposals are still not in stage-4
13:09
<annevk>
howdoi: I'm not sure I follow that question
13:09
<howdoi>
annevk: makes sense :) (Can consider as voting or majority of them supports in via the mailing list or something else)
13:09
<annevk>
howdoi: note that I'm not really doing TC39 stuff
13:09
<howdoi>
in simple terms : Who are eligible to become a TC39 member?
13:10
<annevk>
howdoi: I think anyone who pays USD 10k a year
13:10
<annevk>
howdoi: not sure if the fee remained the same, might be more now?
13:10
<howdoi>
just 10k :D heh heh ok!
13:13
<annevk>
Domenic: playsinline change LGTM, though not entirely sure about ordering, but I'm guessing you thought about that
13:13
<howdoi>
annevk: last (silly) question, say Mr.X is working at a company Y which is a TC39 member, this person can also get into the community via the company?
13:13
<hallvors>
annevk: an improved test for POST and redirects here https://github.com/w3c/web-platform-tests/compare/hallvors/upload-redirect - should be ready for a PR now
13:14
<annevk>
howdoi: sure, everyone that participates is in the community
13:14
<howdoi>
annevk: killer! nice :)
13:15
<annevk>
hallvors: cool, so if you want to do more work, you could generalize the method bit
13:15
<annevk>
hallvors: e.g., PUT needs to be preserved for all redirect types
13:15
<howdoi>
annevk: luckily the company at which I work in an ordinary member, let me see what best I can get from them ;)
13:16
<annevk>
hallvors: POST -> GET happens for 301/302; oh and 303 always goes to GET
13:16
<annevk>
hallvors: 307 and 308 always preserve
13:16
<hallvors>
heh, that should all be tested for sure
13:16
<annevk>
hallvors: Fetch spells this out
13:16
<annevk>
hallvors: yeah, not sure to what length you want to go with this one
13:17
<hallvors>
Not sure if I should extend that test or write a couple of other ones.. might get too complex to be readable at some point.
13:17
<annevk>
hallvors: does this pass in any browser?
13:17
<hallvors>
yes
13:18
<hallvors>
once I gave up expecting a progress event in readystate 4 it passes nicely in Firefox and Chrome (on a Mac), fails in Safari which appears to not do 307 correctly
13:18
<annevk>
hallvors: cool, LGTM then
13:18
<annevk>
hallvors: could maybe ping cdumez in the eventual PR/merge for the Safari thing
13:19
<annevk>
hallvors: sure they'd like to fix
13:19
<hallvors>
I'll report a bug on testing those other methods..because I should be doing webcompat bug analysis right now..
13:20
<annevk>
yeah, back to staring at cross-origin function wrappers
13:21
<annevk>
they are exactly as exciting as they sound
13:21
<hallvors>
just say that in a deep, sexy voice and see how exciting you can make them :-p
13:32
<hallvors>
annevk: Travis has found some console use in a file I haven't touched.. https://github.com/w3c/web-platform-tests/pull/3210
13:32
<hallvors>
wot?
13:33
<Domenic>
annevk: ordering?
13:33
<annevk>
Domenic: relative to other attributes
13:33
<annevk>
hallvors: dunno
13:33
<Domenic>
annevk: ah yeah I just kind of was like... seems kind of like autoplay, let's put it there. But in IDL it should probably be last since that in theory has webcompat impacts
13:41
<Ms2ger>
hallvors, should be fixed on master
13:41
<hallvors>
Ms2ger: this afternoon? I thought I had pulled recently.
13:42
<hallvors>
hm.. maybe I pulled on another laptop
13:42
<hallvors>
#-]
13:42
<Ms2ger>
hallvors, https://github.com/w3c/web-platform-tests/commit/12b780086f4a0d1d6ff5e6d588482ae04624e794
13:44
<hallvors>
Ms2ger: thanks, will rebase..
15:01
<annevk>
Domenic: I don't really need the [[Wrapped]] stuff anymore?
15:01
<Domenic>
annevk: yeah I think CrossOriginFunctionWrapper can go away
15:01
<annevk>
Domenic: it's never become entirely clear to me how ECMAScript does anonymous functions, but what you suggested works for me
15:12
<annevk>
Domenic: thanks, still not entirely pleased with the IDL mismatch, but it's a lot better than a security hole
15:16
<annevk>
Domenic: playsinline fixup LGTM
15:16
<Domenic>
\o/
15:16
<annevk>
Domenic: yay for avoiding the negations
15:29
<hsivonen>
annevk: ok. sorry. I won't merge going forward. I thought it was part of having it assigned to me.
15:29
<annevk>
hsivonen: yeah I should have clarified
15:29
<annevk>
hsivonen: all is good now
15:30
<annevk>
hsivonen: I assume you saw all the new tests?
15:52
<hsivonen>
annevk: I didn't see new tests yet.
16:28
<annevk>
hsivonen: see new open issues from Richard
16:35
<annevk>
Domenic: I have some internal emails here suggesting Worklets need a ton of work
16:35
<annevk>
Domenic: planning on shifting through tomorrow
16:36
<Domenic>
annevk: sounds interesting... not sure it's something i want to tackle, but I guess I can help.
16:36
<annevk>
Domenic: was wondering if you had taken a peek yet
16:36
<Domenic>
annevk: not really looked, no
16:36
<annevk>
Domenic: okay, if I end up filing a bunch of things tomorrow I'll let you know
17:10
<annevk>
Navigation 😭
19:39
<mathiasbynens>
Domenic: old discussion about those, ehm, “best practices”: https://github.com/mathiasbynens/mothereff.in/issues/29
19:40
<Domenic>
mathiasbynens: I can't find the rationale there? Some people in that thread thinks it looks bad? The spec explicitly has examples that end in non-alphanumerics
19:46
<mathiasbynens>
seems like it was about aesthetics
19:46
<mathiasbynens>
= subjective, ofc
19:48
<mathiasbynens>
it’s not a bad idea to be strict and warn on things that might cause confusion in tools like these, imho
19:48
<mathiasbynens>
people who want to use non-alpha can and will do so
19:49
<caitp>
if you can't have custom elements named <\u1f922>, what is the point
20:14
<mathiasbynens>
caitp: <foo-�> (U+FFFD) is valid though, which results in U+1F922 anyway
20:19
<caitp>
yeah, but but you can't have <�-foo>
20:19
<caitp>
which is just terrible!
20:19
<caitp>
emojis are people too
20:22
<TabAtkins>
#NotAllEmojis
20:47
<zcorpan>
https://twitter.com/jcdsvg/status/745337593210904577 is interesting. we (browsers) should be communicating better that removing the doctype is bad. or something, i dunno
20:50
<gsnedders>
zcorpan: *is* it all bad if browsers interoperate?
20:50
<zcorpan>
they don't
20:51
<gsnedders>
they don't currently, but if browsers converge, does quirks mode continue to be bad?
20:51
<gsnedders>
because we are seeing convergence, slowly
20:52
<zcorpan>
some of the quirks are pretty confusing
20:52
<zcorpan>
like https://quirks.spec.whatwg.org/#the-tables-inherit-color-from-body-quirk
20:53
<gsnedders>
is it still inheritence when it goes from child to parent? ;P
20:53
<zcorpan>
it's not child to parent?
20:55
<zcorpan>
i guess it could be a sibling or a descendant of a sibling to body though
20:55
<gsnedders>
zcorpan: oh, wait
20:55
<gsnedders>
zcorpan: I thought that meant tbody
21:00
<zcorpan>
also the syntax quirks only working for some properties and not others is confusing. color: 5a5a5a working but color: 5e5e5e not working is confusing. them working in @supports but not CSS.supports() is confusing. IDs and class names being case-insensitive can introduce bugs.
21:00
<zcorpan>
etc
21:03
<gsnedders>
(I'm just playing devil's advocate here, given interop does get rid a lot of the tradition arguments)
21:03
<zcorpan>
sure
21:04
<zcorpan>
a bit like the value of validation when html parsers interoperate
21:05
<MikeSmith>
zcorpan: https://github.com/validator/validator/issues/310
21:06
<MikeSmith>
zcorpan: there are many sites using <!--> with IE conditional comments
21:06
<MikeSmith>
due to guidance like https://en.wikipedia.org/wiki/Conditional_comment#Downlevel-revealed_conditional_comment
21:07
<MikeSmith>
and in https://www.sitepoint.com/web-foundations/internet-explorer-conditional-comments/
21:07
<zcorpan>
MikeSmith: thx. need to think a bit how to fix that
21:08
<MikeSmith>
yeah, in checking things like Alexa top 100 sites, many or most of them are using that
21:08
<gsnedders>
MikeSmith: pretty sure everything I work on still uses conditional comments, FWIW
21:09
<MikeSmith>
well not all IE conditional comments use that
21:09
<gsnedders>
plenty of things still care about older versions of IE, for which they're the sanest way to target them
21:09
<MikeSmith>
I am not sure why some sites use that while others do not
21:10
<MikeSmith>
I would think we might be able to allow <!--> at end of a comment while still disallowing things like <!-- foo <!-- bar -->
21:11
<MikeSmith>
(that is, where the nested <!-- is clearly in the midst of some comment and not at the end)
21:20
<zcorpan>
in related news, 36,813 pages of 496,558 pages in httparchive are in quirks mode
21:21
<zcorpan>
7.4%
21:22
<zcorpan>
MikeSmith: also allow <!-- <!---> ?
21:23
<MikeSmith>
zcorpan: probably could just as easily allow that too
21:23
<MikeSmith>
though I am not sure anybody actually does that
21:23
<MikeSmith>
the docs all specifically say just <!--> at the end
21:24
<zcorpan>
yeah
21:26
<zcorpan>
the syntax currently disallows comment data ending with <! or <!-, so if we allow <!-- <!--> but not the other we'd still need to disallow comment data ending with <!-.
21:28
<zcorpan>
maybe only change <!--> and not any more dashes
21:43
<zcorpan>
filed an issue
21:58
<zcorpan>
63,518 resources in httparchive have conditional comments with <!-->
22:01
<zcorpan>
1 has conditional comment with <!--->. 28 have <!---> *somewhere*