00:59
<MikeSmith>
Domenic: thanks for the heads-up, banned
01:18
<kochi>
MikeSmith: thanks, it seems my request was accepted.
01:35
<MikeSmith>
kochi: super
09:07
<jgraham>
Anyone know what the pref name is in Chrome to allow popups? (i.e. the one that gets written to the Preferences file)
09:41
<jgraham>
(found it)
12:32
<mnot1>
Sitting in the IETF Security Area meeting, where an NSA person is presenting about a WebSockets vulnerability
12:32
<mnot1>
https://www.ietf.org/proceedings/96/slides/slides-96-saag-1.pdf
12:32
<mnot1>
basically, timing attack against the handshake
12:34
<mnot1>
annevk: put the mitigation in fetch (if it pans out)?
12:34
<annevk>
mnot1: port blocking is part of Fetch
12:35
<mnot1>
he wants to delay the handshake so you can't tell between a listening and non-listening port
12:35
<annevk>
mnot1: oh I see
12:35
<mnot1>
looking for his draft...
12:35
<annevk>
mnot1: since WebSocket gets its own connection there's a little more room for nastyness?
12:36
<mnot1>
y
12:36
<annevk>
Because in principle you can do this with <img>
12:37
<mnot1>
webworkers + websockets makes it faster… or could you do webworkers + <img>? hmm
12:38
<annevk>
can't, can do importScripts() or fetch()
12:40
<nox>
https://github.com/w3c/IndexedDB/issues/83 Does this make sense to anyone?
12:41
<annevk>
nox: does
12:45
<smaug____>
is registerElement still defined in some spec?
12:45
<smaug____>
or was that removed?
12:46
<annevk>
smaug____: renamed to customElements.define()
12:46
<annevk>
smaug____: HTML Standard
12:46
<smaug____>
HTML?
12:46
<botie>
HTML is, like, different, only browsers use HTML for the most part
12:47
<smaug____>
not CustomElements spec which has interface CustomElementsRegistry
12:47
<annevk>
smaug____: CustomElements spec is a copy that is typically out-of-date as I understand it
12:47
<smaug____>
uh
12:47
<smaug____>
ok, now I've been looking at wrong spec
12:48
<smaug____>
why CustomElements are in HTML spec?
12:48
<smaug____>
wouldn't DOM be more natural?
12:48
<smaug____>
though, doesn't matter much
12:48
<annevk>
smaug____: because it's custom HTML elements mostly and they integrate with the HTML parser and such
12:48
<annevk>
smaug____: there's a couple of hooks in DOM too to queue the callbacks
12:48
<annevk>
smaug____: and to define the new state for elements
12:49
<annevk>
smaug____: and DOM defines createElement() still of course
12:49
<annevk>
smaug____: I think we mostly went with HTML since you can only subclass HTMLElement
12:49
<annevk>
(and the HTML parser)
12:49
<smaug____>
I see
12:50
<smaug____>
hmm, I wonder why Element can't be subclassed
13:00
<annevk>
smaug____: I have forgotten, probably because you could never serialize it and therefore it would not be that useful
13:11
<annevk>
smaug____: so yeah, https://w3c.github.io/webcomponents/spec/custom/ is a month out-of-date
13:11
<smaug____>
ok, thanks
13:11
<annevk>
smaug____: I landed changes for custom elements three hours ago in HTML
13:11
<smaug____>
https://w3c.github.io/webcomponents/spec/custom/ could perhaps have a warning it being out-of-date most of the time
13:11
<smaug____>
(no one reads the Abstract ;) )
13:13
<annevk>
Domenic: ^^
13:35
<nox>
annevk: Thanks btw. :)
13:36
<nox>
annevk: I died inside when i found this problem.
13:36
<annevk>
nox: standards being ignorant about threads and having issues with parallelism is fairly common still
13:37
<annevk>
nox: we only really started thinking in terms of event loops after Hixie wrote one down
13:37
<nox>
annevk: I know, but it's the first time it happens in a spec I'm actually implementing. :P
13:37
<annevk>
hehe
13:47
<wanderview>
I wish spec's were written in markdown. It would be a lot easier to review changes
13:47
<nox>
And a lot harder to edit IMO.
13:48
<wanderview>
or at least fricking wrap lines to a reasonable width
13:51
<annevk>
smaug____: does it seem reasonable to you to fix one part of <iframe>'s problems first https://github.com/w3c/webcomponents/issues/145#issuecomment-234258706?
13:51
<annevk>
wanderview: 100 columns is not reasonable?
13:51
<annevk>
wanderview: it might be time to get a bigger monitor
13:51
<wanderview>
annevk: https://github.com/slightlyoff/ServiceWorker/commit/5c0ecaeb423d04df7429bfaa2e5fbde9e42038e1
13:52
<annevk>
wanderview: ooh, but you left that comment on a change to HTML
13:52
<wanderview>
line 714 of that commit is a hell of lot bigger than 100 chars
13:52
<annevk>
HTML commits don't have that
13:52
<wanderview>
annevk: how can I interpret the change to html without reviewing that other commit it depends on?
13:52
<annevk>
Sure, your comment was just not very specific
13:59
<annevk>
wanderview: note that the client message queue bit landed in SW so you could read it in the spec
14:29
<mounir>
annevk: xref's html.py takes ages and ends up with exceptions, it is worth battling with it?
14:36
<annevk>
mounir: exceptions are basically terms that need to be removed, but you can leave a comment and I can take care of it I suppose
15:10
<mounir>
annevk: I've removed them
15:10
<mounir>
annevk: updated the 3 CLs -- whenever you have time ;)
15:10
<annevk>
mounir: ta, maybe tonight, but might be tomorrow morning
17:50
<annevk>
mounir: landed whatwg/fullscreen and whatwg/xref
17:50
<annevk>
mounir: thanks a lot for doing all of the work required, that's great
17:52
<annevk>
mounir: it seems I can even merge the screen orientation PR, but I'll leave that up to you
18:27
<smaug____>
does anyone happen to know when blink dispatches click
18:27
<smaug____>
after mousedown/up
18:28
<smaug____>
oh oh, it has some very specific hack for <button>s
19:07
<Domenic>
;_;
19:07
<Domenic>
Why all the sudden UI event infrastructure investigations, smaug____? I mean, it's good for improving the spec and such, but why would you do this to yourself.
19:13
<smaug____>
Domenic: just fixing a bug
19:13
<smaug____>
well, not a but
19:13
<smaug____>
<button> handling in Gecko is per HTML4
19:13
<smaug____>
s/but/bug/
19:13
<smaug____>
but since other browsers have other behavior, need to change the behavior
19:13
<Domenic>
I see
19:13
<Domenic>
I wonder what HTML4 said about buttons... /me goes to look
19:14
<Domenic>
So far it's not clear that they can be clicked https://www.w3.org/TR/html4/interact/forms.html#h-17.5
19:14
<smaug____>
<button> in HTML4 is just a way to have more complicated rendering for the button
19:14
<smaug____>
or that is at least one interpretation
19:15
<smaug____>
so in Gecko events inside <button> go always to the <button>
19:15
<smaug____>
but now, I'm trying to understand how click even works in blink
19:15
<Domenic>
We got a "when activated" in https://www.w3.org/TR/html4/interact/forms.html#submit-button which seems to acknowledge the existence of clicking
19:16
<smaug____>
I thought it click would be dispatched to common ancestor of mousedown and mouseup
19:16
<smaug____>
in blink
19:16
<smaug____>
but apparently it is very different
19:16
<smaug____>
in some cases common ancestor
19:16
<smaug____>
in other cases... it doesn't happen at all
19:16
<Domenic>
rbyers might be able to help
19:19
<smaug____>
yeah, I tried to ping him
19:28
<smaug____>
ok, found the code
19:52
<rbyers>
smaug____: Sorry for the delay, I'm on vacation. I thought it was usually common ancestor - at least that's the code I know.
19:53
<smaug____>
rbyers: it is not that. there is also something about interactivecontent or such
19:53
<rbyers>
Dtapuska@ and Mustaq@ probably know more. You could ask on input-dev⊙co or file a bug for anything you see that's not to spec.
19:55
<rbyers>
Input team considers any spec / blink mismatch to be a chromium bug, even if the right fix ends up being to change the spec...
19:55
<smaug____>
rbyers: I found the relevant code in blink. I was just surprised since it has been discussed before the click should happen on common ancestor
19:56
<smaug____>
but there are cases when click doesn't happen at all in blink after mousedown/up
19:56
<smaug____>
this is per element basis
19:56
<smaug____>
form elements, <a> and some others
19:56
<smaug____>
they all affect to click dispatching in blink apparently
19:57
<rbyers>
Hmm. If its not too breaking, we can probably get rid of any such special cases. IIRC the click we fire for touch had no such special cases.
19:58
<rbyers>
So the inconsistency is weird, probably just another sign of mouse code being so crusty and unmaintained for so long.
19:59
<rbyers>
Then again, web compat in this area is super brittle - every time we try to clean something up we are surprised by unexpected breakage. Still, we're not afraid to try..
20:01
<smaug____>
rbyers: there is some comment in the code about IEism, so it might be quite breaking change
20:01
<smaug____>
I should look at the blame to see when it was added
20:01
<smaug____>
rbyers: go back to your vacation!
20:02
<rbyers>
Best test is probably Edge. If Edge doesn't do that, we can almost certainly change.
20:02
<rbyers>
Thanks, TTYL :-)