11:45
<zcorpan>
TabAtkins: has using Custom Highlight for syntax highlighting in specs been considered? see https://github.com/whatwg/html/issues/12782#issuecomment-5438468433
12:44
<Dominic Farolino>
SharedWorkers can't access any of their owners' WindowProxy objects. Has changing this ever been discussed?
12:48
<Dominic Farolino>
Basically bringing something like service worker clients API to shared workers I guess
13:00
<Ms2ger>
I think we don't really like shared workers anymore
13:09
<TabAtkins>
TabAtkins: has using Custom Highlight for syntax highlighting in specs been considered? see https://github.com/whatwg/html/issues/12782#issuecomment-5438468433

I have thought about it; it would certainly simplify þͤ code a bit too (no more careful tree merging, at least if you're not using line numbers).

It's been a bit since I first thought about it - is Custom Highlights sufficiently widely supported yet?

13:10
<zcorpan>
TabAtkins: Baseline 2025
13:11
<zcorpan>
TabAtkins: also part of interop 2026
13:13
<TabAtkins>
Cool, I'll look into it then
13:17
<TabAtkins>
Lol at þͤ "late dfnEnabled class thrashes þͤ entire document's style just to add a pointer style to definitions".
13:17
<zcorpan>
yes :)
13:20
<bkardell>
TabAtkins: has using Custom Highlight for syntax highlighting in specs been considered? see https://github.com/whatwg/html/issues/12782#issuecomment-5438468433
Funny i was thinking the same thing yesterday while out for a walk and came here today to aak the same question when i remembered
13:20
<Dominic Farolino>
I'm not observing this anywhere.
13:53
<annevk>
Yeah I think we still like shared workers. In fact I recently made dedicated workers inside shared workers work in WebKit. As for your earlier question, I think having clients be a more common concept was considered at some point, but not to the point of anyone designing what that might look like.
15:04
<jarhar>
what do you mean by it would be bound to something in the new page?
15:07
<zcorpan>
TabAtkins: sideshowbarker : thoughts on https://github.com/whatwg/wattsi/issues/169 ?
16:02
<Thomas Steiner>

Hey folks 👋,

You're absolutely right (no pun intended), I am very much open to this sort of feedback and I very much apologize for overdoing it with the agent use (which, to be sure, was still human-reviewed).

I do talk with our privacy folks, and we do have a solution for COS that Chrome is happy with (coupling it to 3rd party cookies), but I am carefully positive that we can come to a place with other vendors which would allow them to implement the API with a strict probing budget (see #73 in our repo). I hope you can chime in.

Sorry again!

Cheers,
Tom

18:10
<smaug>
Basically visually bind it to some container. The common cases is that these cross-page things are something on top of the main page
18:10
<smaug>
I mean chats and so
18:11
<smaug>
So they are almost like PiPs
18:12
<smaug>
Or new window could have some accessor to the pip window so that it could move it inside its own viewport
18:13
<smaug>
But in general handling these widgets as top level pages would simplify implementation (and spec) a lot, I think
19:20
<smaug>
(so, if we go with the new element approach, maybe that should be kind of virtually top level, and get its own session history among other things)