05:27
<annevk>
Hmm, no event discussion ugh
06:50
<MikeSmith>
oh man
06:51
<MikeSmith>
I was going to post a link to https://twitter.com/w3c/status/723167929395605509 here for the lulz
06:52
<MikeSmith>
but I see Domenic and hsivonen are already way ahead of me and have favorited it
06:52
<annevk>
I saw that, hsivonen make a point of retweeting those events
06:52
<annevk>
mkaes*
06:52
<annevk>
makes*
08:27
<Ms2ger>
annevk, yt?
08:28
<annevk>
Ms2ger: yeah
08:28
<annevk>
Ms2ger: fixing non-dispatchy things
08:28
<Ms2ger>
annevk, so origins are shared and mutable now, right?
08:28
<annevk>
Ms2ger: partially mutable, yes
08:28
<annevk>
Ms2ger: and yes, they are shared freely
08:29
<Ms2ger>
About that "partially" bit... It can't change opaque <-> tuple, or if it's a tuple, change scheme/host/port, correct?
08:29
<annevk>
Ms2ger: if it's tuple, it can change domain, that's it
08:29
<Ms2ger>
Okay, great
08:30
<annevk>
scheme/host/port never change, opaque never changes, they never change from opaque to tuple
08:30
<Ms2ger>
That's what I thought, thanks
08:30
<annevk>
If you file an issue I can add a note underneath that details these invariants
08:30
<annevk>
at some point
08:32
<Ms2ger>
https://github.com/whatwg/html/issues/1093
09:36
<jochen__>
what's the implication of a "relevant mutation" of an img element?
09:36
<annevk>
zcorpan should know
09:38
<annevk>
jochen__: so I think that referrerpolicy should probably be part of relevant mutations
09:39
<zcorpan>
jochen__: it runs "update the image data"
09:39
<annevk>
jochen__: it seems that the eventual request happens asynchronously at a later point so can be updated
09:48
<jochen__>
ok
09:48
<jochen__>
i guess we should have a test for that
10:03
<Ms2ger>
Yes
13:03
<Ms2ger>
zcorpan, fyi https://github.com/whatwg/html/issues/1099
13:50
<annevk>
smaug____: so I guess I'll wait a bit with defining events until that issue is sorted
13:51
<annevk>
smaug____: it's a little annoying feedback from the others is slow since everything else seems to depend on it
13:51
<annevk>
smaug____: e.g., I created a patch for <script> in HTML, but the "load" event it dispatches needs to be scoped...
13:52
<smaug____>
annevk: "that issue"?
13:52
<smaug____>
event dispatching in shadow DOM?
13:52
<annevk>
smaug____: https://github.com/w3c/webcomponents/issues/485
13:52
<annevk>
smaug____: yeah
13:55
<smaug____>
annevk: I wonder...this iframe + session history.
13:56
<smaug____>
if the top level page is navigated back() and then forward(), I assume the iframe in shadow dom will get a new session history
13:56
<smaug____>
(bfcache is different since that keeps the iframe alive)
13:57
<annevk>
smaug____: I'm not sure I know enough about it, but that sounds right if there's no cache
14:00
<smaug____>
and random note, I _think_ having separate session history for iframes should be easy to implement in Gecko. In xul:browser (which internally uses the same code as html:iframe), we support disablehistory attribute. If not set, session history is created for the contained browsing context (and its descendants)
14:22
<wanderview>
JakeA: does chrome return workers as Client objects from matchAll() yet? I wonder if that will confuse sites when we start doing that
14:30
<annevk>
wanderview: not working on streams just yet?
14:30
<wanderview>
annevk: I started this week, but then Matt Gaunt wrote a bug that has been ruining my day for a few days now
14:33
<annevk>
😢
14:36
<gsnedders>
Uh, how am I meant to animate a changing SVG path nowadays?
14:36
<gsnedders>
I presume with JS?
14:37
<JakeA>
wanderview: hm, not sure. It won't show up unless you pass an option to matchAll though right?
14:37
<wanderview>
thats true
14:37
<wanderview>
ok, that seems safe
14:38
<JakeA>
https://slightlyoff.github.io/ServiceWorker/spec/service_worker/#enumdef-clients-clienttype
14:38
<wanderview>
yea... I forgot about that
14:39
<wanderview>
JakeA: the whole "order by most recently focused" thing might be interesting with workers in the mix
14:39
<JakeA>
wanderview: hmm, good point, we need to define that
14:40
<wanderview>
JakeA: like workers count as never being focused... so always at the end
14:40
<JakeA>
That's probably good enough tbh
14:41
<wanderview>
I'll write an issue
16:20
<wanderview>
JakeA: https://github.com/slightlyoff/ServiceWorker/issues/887
16:21
<JakeA>
Taa!
16:22
<TabAtkins>
gsnedders: The 'd' property is in the process of being specced/implemented, so you can animate that.
16:22
<TabAtkins>
Now if you want to animate it *today*, SMIL still works everywhere but IE.
16:22
<TabAtkins>
But if you want IE support, yeah, JS is the way to go.
18:24
<Mek>
hmm, to properly specify Link rel=serviceworker support in http response headers I'd like to somehow express that: for subresource requests the request was made from a secure context, and for non-subresource requests the resulting document/worker will be a secure context. For the subresource case I can just check if the requests client is a secure context. For workers the requests client is the new worker, so that same check should work. But for
18:24
<Mek>
navigations the settings object for which I care if it is a secure context doesn't exist yet until well after the fetch is complete...
20:47
<wanderview>
Mek: err... why would we allow it on http response headers at all? regardless of how it was initiated it could have be mitm'd, right?
20:49
<Mek>
wanderview: the regular registration checks already ensure that various origins match, are all secure origins, http/https only etc.
20:49
<Mek>
so I meant http the protocol, not http vs https
20:49
<wanderview>
oh, sorry
20:51
<gsnedders>
TabAtkins: this freakin' sucks
23:54
<Mek>
woot, I filed html issue 1111 :P (okay, not sure why that number amuses me)