09:53
<annevk>
wanderview_: ping https://github.com/whatwg/html/pull/1527
09:55
<smaug____>
Does anything define where cursor is put when input element is created
09:55
<smaug____>
data:text/html,<script>window.onload = function() { var i = document.createElement("input"); i.value="foo"; document.body.appendChild(i); i.focus(); }</script>
09:55
<smaug____>
I couldn't immediately see anything in HTML spec, but I could have missed it
09:56
<annevk>
smaug____: selection spec maybe?
09:56
<annevk>
smaug____: all that stuff is horribly bad, just like the impls
09:58
<smaug____>
Looks like https://github.com/w3c/web-platform-tests/blob/master/html/semantics/forms/textfieldselection/selection.html relies on certain behavior
10:02
<smaug____>
http://w3c.github.io/selection-api/ isn't the right one
10:28
<mounir>
annevk: regarding the PR I sent to the fullscreen spec, is this in your queue or are you waiting for foolip to comment?
10:28
<annevk>
mounir: queue, though I might want feedback from others too
10:29
<annevk>
mounir: will hopefully get to it today, had to run various errands plus backlog
10:32
<mounir>
annevk: I left a summary of browser feedback in the PR
10:32
<mounir>
annevk: feel free to ping me when it's on top of your queue
11:02
<annevk>
mounir: https://w3c.github.io/screen-orientation/#dfn-triggered-by-a-user-generated-orientation-change doesn't point anywhere?
11:03
<annevk>
mounir: should probably also add a reference to screen orientation
11:04
<mounir>
annevk: the dfn is in a PR
11:04
<annevk>
mounir: oh, screen orientation hasn't changed yet?
11:04
<mounir>
annevk: sorry, I should have pointed this -- you are CC'd there
11:04
<annevk>
found it I think
11:04
<mounir>
I just need to press merge for the other one to change but I wonted foolip and you to comment first
11:05
<mounir>
annevk: https://github.com/w3c/screen-orientation/pull/90 (to be sure)
11:05
<mounir>
annevk: how do I add the ref in this anolis thing?
11:05
<annevk>
mounir: you'd need to patch whatwg/xref if it's not there
11:07
<annevk>
mounir: both changes look reasonable to me, <code>screen.orientation</code> seems a little sketchy
11:07
<mounir>
annevk: Document's ScreenOrientation instance?
11:07
<mounir>
or s/instance/object/
11:07
<annevk>
mounir: yeah, the latter
11:08
<annevk>
although I guess you'd have to define how you get to that instance from the currently script
11:08
<annevk>
probably through the current global
11:09
<mounir>
annevk: have to leave for a meeting, can you put comments on the PR, will address them this afternoon
11:09
<annevk>
okay
14:58
<smaug____>
jgraham: in wpt, what is the right assertion for something which shouldn't happen? assert_true(false, "this shouldn't happen"); ?
14:58
<jgraham>
smaug____: assert_unreached()
14:59
smaug____
wishes http://testthewebforward.org/docs/#writing-tests worked
14:59
<smaug____>
er, http://testthewebforward.org/docs/
15:00
<smaug____>
click "writing tests" on the right doesn't do anything
15:00
<gsnedders>
smaug____: what browser? WFM here
15:01
<smaug____>
gsnedders: FF and Chrome
15:01
<smaug____>
gsnedders: I'm clicking the link in table of contents
15:02
<gsnedders>
smaug____: ooooh. that doesn't do anything. that's a totally useless TOC given all the sections are in view already…
15:02
<smaug____>
oh, Chrome flashes oddly if the the 'Writing tests' is open below
15:03
<smaug____>
FF doesn't flash but something is scrolled oddly
15:03
<gsnedders>
yeah, stuff is a bit weird
15:07
<gsnedders>
someone really needs to sort a fair bit of that stuff out
15:07
<gsnedders>
and that probably means me doing it at some point…
15:21
<Domenic>
annevk: does it make sense to you how there is no current global unless some JS function (including a built in) is being executed?
15:21
<wanderview_>
JakeA: are you making an agenda again for next week? can we add #839 to the list?
15:21
<Domenic>
So for anything that happens async, there is no current global
15:21
<Domenic>
Maybe we should add that info to the globals/realms/settings objects section
15:21
<annevk>
Domenic: I guess, just wondering how mouse events work then
15:22
<Domenic>
Well they call JS callbacks mostly
15:22
<annevk>
Domenic: maybe they get to an object through hit testing
15:22
<Domenic>
Ah I see
15:22
<Domenic>
Yeah, that seems likely
15:22
<Domenic>
After all how do you know which window's button was clicked
15:24
<annevk>
For orientation it's a little trickier, but maybe that's the active browsing context or some such
15:24
<annevk>
Gotta go for a bit
15:41
<JakeA>
wanderview_: added to https://github.com/slightlyoff/ServiceWorker/issues/932
15:42
<wanderview_>
thanks!
16:24
<mounir>
annevk: is anolis using a pinned version of xref
16:52
<annevk>
mounir: no, just your checked out copy
17:00
<mounir>
annevk: figured it out
17:00
<mounir>
annevk: html.json and html-generated.json ... is confusing :)
17:01
<annevk>
👍
17:01
<mounir>
annevk: sent a PR for you to make fun of me
17:01
<mounir>
:)
17:41
<annevk>
Domenic: a Window always has a browsing context?
17:42
<Domenic>
annevk: seems like. Lots of places assume that.
17:44
<annevk>
Domenic: so basically window.document starts returning null when you discard?
17:44
<annevk>
That doesn't happen, hmm
17:45
<Domenic>
annevk: is that implied by my PR? Hmm.
17:46
<annevk>
I guess the Window can still hold onto a document and a browsing context even though the document no longer keeps track of them
17:46
<annevk>
Domenic: I'm not sure, just trying to think it through
17:47
<annevk>
Domenic: well, concept-document-window which window.document uses does assume the document has a browsing context
17:47
<annevk>
Domenic: or did you change that?
17:48
<annevk>
Nope
17:48
<Domenic>
Seems possibly broken, hmm.
17:49
<annevk>
I left a comment
17:49
<Domenic>
The spec has never been good about distinguishing the document's window and the window's document :(
17:49
<annevk>
Yeah and this clearly shows why that's been a problem
17:49
<Domenic>
Unfortunate how this PR keeps growing but I don't see any good intermediate steps
17:50
<annevk>
I think this PR is amazing though in the amount of things it fixes around lifetime
17:51
<Domenic>
Still all feels a bit on shaky ground, but glad you like it :)
18:08
<annevk>
Fetch now has 26 unique contributors
18:08
<annevk>
This GitHub thing ain't so bad
18:11
<jsbell>
Speaking of fetch: anyone recall the process taken to determine that adding 'fetch' to the global namespace was safe? HTTPArchive search, "canary" (etc) builds, ... anything else?
18:13
<annevk>
jsbell: I think we just tried
18:13
<jsbell>
context: https://wicg.github.io/entries-api/#issue-e6c0a36d - I'd like to dropping [NoInterfaceObject] from some de facto standard types. At the moment, Gecko is implementing with [NoInterfaceObject] and Edge will prefix them (*sigh*).
18:13
<annevk>
jsbell: because everyone wanted it to work
19:01
<smaug____>
annevk: Domenic: do you think someone will object fixing https://github.com/whatwg/html/issues/1568 ?
19:01
<smaug____>
since I'm about to add some wpt tests for the behavior browsers have
19:16
<annevk>
smaug____: seems unlikely
19:47
<Domenic>
smaug____: if you want to suggest spec text (not in full PR form, we can do that, just the words you'd suggest) that'd be even better :)
19:48
<smaug____>
Domenic: sure. I need to fix this one gecko bug first, but I should look at the activation stuff some more
19:48
<Domenic>
\o/
19:48
<smaug____>
it seems to have quite some issues still in the spec
19:48
<Domenic>
I believe you :(
19:48
<Domenic>
Having to work around the giant hole that is hit testing doesn't help
19:48
<smaug____>
in other words, I think we may want to switch to the setup browsers have
19:48
<smaug____>
and not what has been in the spec for a long time
19:49
<Domenic>
oh dear, hmm ok
19:49
<Domenic>
the big mismatch i know of is the default actions for click events
19:49
<smaug____>
yes
19:49
<smaug____>
we need to create event path first and operate on that both before and after event dispatch
19:50
<smaug____>
spec has different model
19:51
<Domenic>
Well it has pre-click and post-click activation steps... but I imagine the particulars are not right?
19:52
<smaug____>
well, it stores just one element during pre-click, but it needs to actually have the whole event path during post-click
19:52
<Domenic>
I see, interesting
19:52
<smaug____>
and I think also the synthetic click stuff is wrong, since it should also operate on event path, not just the target
19:53
<smaug____>
(I think. That is something I haven't tested on various browsers yet)
21:30
<nobso>
Hi @all How is the CSS Grid Layout support in the modern browsers over all? Is there a polyfill that i can rely on ?
21:32
<nobso>
TabAtkins, do you have any comments on that?
21:50
<smaug____>
Domenic: FYI, forms can be nested and there is some black magic in event handling code (related to https://github.com/whatwg/html/issues/1533#issuecomment-233771168)
21:51
<smaug____>
(but that is actually an argument to not add similar black magic for toggle)
21:51
<smaug____>
I'd need to check what other browsers do there though
21:51
<smaug____>
with submit
21:51
<smaug____>
and reset
21:52
<Domenic>
Right, I meant conformant, such that we should worry about that case when designing bubbling
21:52
<Domenic>
*conformantly
21:57
<smaug____>
I don't recall spec saying anything about nested forms and submit event
21:58
<smaug____>
bad me that I didn't file a spec bug 8 years ago, https://bugzilla.mozilla.org/show_bug.cgi?id=428135
21:58
<smaug____>
(but apparently webkit had similar behavior then too)