01:00
<cabanier>
promises promises
01:02
<terinjokes>
i have a joke about promises
07:15
<hober>
Hixie_: just do it a sensible way and if that doesn't match the API fashion of the month, so be it
07:17
<TabAtkins>
hober: That's a great way to get implementors to willfully violate the spec, which is exactly what they've publicly said they'd do if Hixie did try and do a hack around WebIDL.
07:28
<annevk>
If that happens Hixie_ will change the spec, so nobody loses?
07:29
<hober>
annevk: indeed
07:39
<TabAtkins>
Well, Hixie loses, because he has to edit twice, and implementors lose, because they have to violate the spec in an undefined way to get sane behavior; forcing implementors to violate the spec because the author doesn't agree with everyone else also just generally lowers respect without any benefit.
08:09
<annevk>
http://news.netcraft.com/archives/2013/05/13/how-certificate-revocation-doesnt-work-in-practice.html is this still the case?
09:36
<Ms2ger>
annevk, http://wiki.whatwg.org/wiki/TR_strikes_again
09:41
<astearns>
the fullscreen/dialog entry on that page doesn't specify a particular problem, as far as I can tell
09:42
<annevk>
astearns: I think what it's trying to state is that it's very hard to determine what the latest state of a particular piece of layout is
09:42
<annevk>
astearns: e.g. if I were to implement a new box creation algorithm, where do I look?
09:43
<annevk>
astearns: or if I implemented stacking contexts, how do I learn about http://fullscreen.spec.whatwg.org/#new-stacking-layer
09:43
<annevk>
astearns: CSS is still a rather monkey patched landscape
09:44
<astearns>
that's a fair point, but the other links are to specific instances of harm, not general difficulties that might someday result in a problem
09:45
<astearns>
and your point is more about CSS modularization than TR publication
09:48
<annevk>
I don't disagree, I didn't create that page
09:49
<annevk>
It could perhaps be reframed a bit better
09:49
<annevk>
Perhaps we should turn it into a page like http://wiki.whatwg.org/wiki/IETF
09:49
<annevk>
Ms2ger: ^
10:32
hsivonen
hopes setting .innerHTML on annotation-xml doesn't depend on the attributes of the annotation-xml node
10:32
hsivonen
goes read the spec
10:40
<hsivonen>
sigh. innerHTML now depends on the attributes of the context node
10:40
<hsivonen>
hooray for annotation-xml
10:45
<annevk>
Can't we just drop support for annotation-xml?
11:15
<ondras>
sooo
11:15
<ondras>
<input> elements inside a shadow dom
11:15
<ondras>
are they considered when submitting a form?
11:16
<hsivonen>
annevk: I think that would be a bit drastic and rather bait-and-switchy towards the math folks
11:44
<TabAtkins>
ondras: They are not, afaik.
11:49
<ondras>
TabAtkins: okay
11:52
<zcorpan>
hsivonen: can we make the spec more stupid for innerHTML on annotation-xml? is there a use case for innerHTML on annotation-xml that needs to look at the attribute?
12:00
<hsivonen>
zcorpan: I don't know
12:00
<hsivonen>
zcorpan: in principle, if you ever set innerHTML on annotation-xml, it would be weird for it to behave differently from a full-doc parse
12:23
<zcorpan>
hsivonen: that's already the case for foreign stuff due to break-out tags
12:23
<zcorpan>
hsivonen: but i agree with the principle
12:58
<Ms2ger>
zcorpan, do you know about the websocket tests?
12:59
<Ms2ger>
Or jgraham
12:59
<zcorpan>
Ms2ger: yeah
13:01
<Ms2ger>
/websockets/interfaces/WebSocket/close/006.html fails with a lot of complaining dumped to the console running wptserve
13:05
<zcorpan>
Ms2ger: the console complains because the test is closing the connection before it is established
13:05
<zcorpan>
Ms2ger: presto passes the test with complaining in the console
13:07
<zcorpan>
Ms2ger: i guess gecko fails because it sync fires a close event when calling close()
13:07
<Ms2ger>
Might be
13:07
<zcorpan>
or something
13:07
<Ms2ger>
Blink fails too
13:08
<Ms2ger>
Different way, though, maybe
13:08
<zcorpan>
blink seems to set readyState to CLOSED immediately
13:08
<Ms2ger>
Yep
13:09
<Ms2ger>
Oh, no
13:09
<Ms2ger>
Gecko has closed/fired the close even before the timeout
13:11
<Ms2ger>
So I guess I'll file a bug on Gecko
13:12
<hsivonen>
so Web Sockets without TLS had some deployability problems. What caused those to be resolved with a long process with delays instead of the HTTP/2 approach of hiding everything inside TLS?
13:12
<hsivonen>
(just wondering about what were all the things that went wrong with Web Sockets)
13:13
<hsivonen>
(and whether the thing that caused delays could have been avoided with a TLS-only policy)
13:13
<hsivonen>
or did the concerns that lead to long delays apply to wss, too?
13:15
<zcorpan>
hsivonen: what delays do you mean?
13:16
<zcorpan>
hsivonen: was it intermediaries hanging on to the response until it thinks it has a full response or some such?
13:17
<hsivonen>
zcorpan: delays in the process of getting the spec done
13:20
<zcorpan>
hsivonen: ah. so what deployability problem do you mean?
13:20
<zcorpan>
hsivonen: i recall there being different success rates on different ports
13:20
<Ms2ger>
zcorpan, (r? https://critic.hoppipolla.co.uk/r/2542 https://critic.hoppipolla.co.uk/r/2556)
13:22
<hsivonen>
zcorpan: didn't some intermediaries mess up the upgrade?
13:26
<zcorpan>
hsivonen: i don't recall the details
13:28
<zcorpan>
Ms2ger: https://critic.hoppipolla.co.uk/r/2542 was 100% 1 issue
13:29
<Ms2ger>
Dammit
13:29
<Ms2ger>
The dashboard really needs to distinguish those
13:35
<annevk>
hsivonen: fair, though perhaps a usage survey could help make our case?
13:35
<annevk>
hsivonen: though I guess once you write the code to pass around some extra state, it's not too bad either way
13:36
<annevk>
hsivonen: I think Hixie_ thought requiring TLS was too hard on authors
13:36
<annevk>
hsivonen: but don't take my word for it
13:37
<annevk>
hsivonen: that was before http://tools.ietf.org/html/rfc7258 and therefore before everyone opened their eyes to what some had been saying for a long time
13:41
<annevk>
es-discuss is somewhat bad at not feeding trolls. And worse at not answering actual feedback. Wrong S/N ratio :-/
13:45
<zenparsing>
annevk yeah - i think it used to be better a couple of years ago
13:46
<zenparsing>
partly it's the release cycle - features for ES6 are either done, or so late that no one really wants to debate them
13:47
<zenparsing>
but i've noticed that ES7 features aren't really being discussed there, which might be a bad sign
14:45
<zcorpan>
at least csswg doesn't subscribe to forking whatwg specs
14:55
<Hixie_>
hober: APIs being inconsistent is even worse than being bad, IMHO.
14:56
<Hixie_>
hober: but it definitely does make me reluctant about using promises in general
14:57
<caitp>
bad apis and inconsistent apis are equally bad
14:57
<Hixie_>
not imho
14:58
<Hixie_>
a bad api that's consistent is easier to use than a variety of good apis all of which have different philosophies
14:59
<caitp>
the point is, you don't want to come up with bad apis, you want to come up with good apis which are adopted
14:59
<caitp>
being bad isn't going to help adoption
15:00
<Hixie_>
adopted by whom? implements or authors?
15:00
<Ms2ger>
The Promise fanclub
15:01
<zenparsing>
Hixie_ i'm sympathetic to your position wrt promise-returning functions throwing, but...
15:01
<zenparsing>
Hixie_ in the end, as users move to async functions, it will matter less and less
15:02
<Hixie_>
or more and more :-)
15:02
<caitp>
in the fuuuuuuuuture
15:03
<zenparsing>
less and less: async function f() { await gimmePromise(x) } doesn't really matter if gimmePromise throws or rejects
15:05
<caitp>
from a programme correctness standpoint, you do want the early error, otherwise there's a chance you might not notice
15:06
<Hixie_>
zenparsing: i think that behaviour for async is dumb too
15:07
<Hixie_>
zenparsing: i fully expect authors to never check for these error conditions, which is going to lead to all kinds of bugs when their rejection handlers get called for reasons unrelated to what they're expecting
15:07
<Hixie_>
just like they never check them today, except today it's fine because it just aborts script execution
15:10
<caitp>
but even if async did make it better, who knows when a platform with the async sugar will exist, why design for a hypothetical future platform
15:17
<zenparsing>
Hixie_ once async functions are in, manually wiring up rejection handlers will be the exception i think
15:47
<annevk>
hsivonen: https://twitter.com/sleevi_/status/509723775349182464
16:16
<annevk>
Wut. Twitter is serving me JSON
16:16
<annevk>
"Something is technically wrong." no shit
16:16
<annevk>
This is why you don't do content negotiation
16:20
<jgraham>
annevk: Sounds like an improvement
16:50
<annevk>
I wonder there's a jabber folder with mimesniff.spec.whatwg.org in it on my WHATWG DreamHost account
16:52
<Ms2ger>
What's jabber?
17:06
<annevk>
Is there a way to set a default user for a remote host with SSH?
17:07
<annevk>
Or enable another user by playing with the .ssh directory?
17:21
<JonathanNeal>
Sorry I’m late to the party, but does Microsoft’s touch events spec cover pressure? Is it Apple Watch ready?
17:26
<caitp>
the big questions
17:31
<annevk>
JonathanNeal: you mean pointer events?
17:31
<JonathanNeal>
pointer events, yes.
17:31
<JonathanNeal>
caitp: thoughts?
17:31
<annevk>
JonathanNeal: https://dvcs.w3.org/hg/pointerevents/raw-file/tip/pointerEvents.html#widl-PointerEventInit-pressure
17:32
<annevk>
(I don't actually know, but searching for pressure in that draft...)
17:34
<JonathanNeal>
annevk: wow, i didn’t think it would be so obvious! I’m sorry I made you google that for me.
17:35
annevk
finds http://stackoverflow.com/questions/10197559/ssh-configuration-override-the-default-username
17:36
<JonathanNeal>
This confirms that Pointer Events is a brilliant spec. Is Chrome still set to ignore it? Sorry if I missed how that resolved.
17:41
<JonathanNeal>
^ https://code.google.com/p/chromium/issues/detail?id=162757
19:24
<mounir>
annevk: ping
19:25
<mounir>
foolip: ping
19:33
<caitp>
wow, blink's svg rendering works really nicely compared to geckos, when dynamically inserting hundreds of paths
19:33
<caitp>
good job on the jank-free side of things :>
19:43
<annevk>
mounir: email or ask and wait until tomorrow
19:47
<mounir>
annevk: I can wait :)
19:58
<smaug____>
caitp: yup, we have some very well known perf bugs
19:58
<smaug____>
in svg
19:58
<smaug____>
unfortunately
19:58
<caitp>
not a criticism, I am just impressed with the blink behaviour here
20:00
<smaug____>
it is a fair criticism.
20:01
<smaug____>
caitp: if you have a minimal testcase, feel free to file a bug ;)
20:03
<jmrenner_>
Hey does anyone know why a cross-domain XHR with withCredentials false still requires CORS headers?
20:47
<odinho>
caitp: I know that at least one of presto's SVG experts (fs) has been working a lot on SVG in blink since Opera moved to chromium. Though I'd guess many of the remove-jank low level changes are most responsible for any recent smoothness changes.
22:14
<smaug____>
I don't have account to modify http://wiki.whatwg.org/wiki/TR_strikes_again, but it could say there is a constant flow of questions on IRC regarding obsolete TR/ specs, and one needs to all the time (daily or weekly) point to the up-to-date specs.
22:16
<smaug____>
one case from today http://logs.glob.uno/?c=content#c239087
23:29
<Domenic>
Oh Glenn Adams.
23:37
<smaug____>
indeed
23:44
<jamesr_>
Delivery to the following recipient failed permanently:
23:44
<jamesr_>
public-w3cprocess⊙wo
23:44
<jamesr_>
oh noes! guess that post won't happen
23:44
jamesr_
will survive