WHATWG
2025-05-13
plaintext logs
prev next
source on github
08:31
<annevk>
If you have tests impacted by partitioned BroadcastChannel, what would be the best replacement at this point in time? The stash feature? Is there a client-side wrapper for that available today?
08:50
<Noam Rosenthal>
stash was more about retaining shared state than message-passing, no?
09:00
<zcorpan>
annevk: https://github.com/web-platform-tests/wpt/tree/master/common/dispatcher#sendreceive-message-passing-apis
09:02
<zcorpan>
or maybe there's a replacement for those? https://github.com/web-platform-tests/rfcs/pull/90
09:04
<zcorpan>
or https://github.com/web-platform-tests/rfcs/pull/98
09:07
<zcorpan>
Docs here https://web-platform-tests.org/writing-tests/channels.html
09:13
<Noam Rosenthal>
ooh I needed these a while back when we were porting prerender out of BroadcastChannel.
09:20
<annevk>
zcorpan: thanks, that's very helpful. I'm surprised the cross-origin-opener-policy tests haven't yet been rewritten to make use of those, but I suppose I can take a look at that.
09:23
<keithamus>
data:text/html,<!DOCTYPE%20html><script>document.addEventListener('keypress',console.log);</script>

Chrome & Firefox don't fire a keypress event for the Escape key. WebKit seems to. I'm curious if there's a spec detail that represents this, or if this is something known?

09:27
<zcorpan>
keithamus: WebKit doesn't fire it when in fullscreen https://software.hixie.ch/utilities/js/live-dom-viewer/saved/13781
09:28
<keithamus>
interesting.
09:28
<zcorpan>
Also see https://github.com/whatwg/fullscreen/issues/231
09:28
<keithamus>

https://w3c.github.io/uievents/#event-type-keypress seems to suggest it shouldn't fire because it doesn't produce a printable character:

only if that key normally produces a character value

09:31
<zcorpan>
Ah, then it seems like a bug in webkit. But keydown/keyup are still interesting wrt fullscreen
11:18
<annevk>
zcorpan: how does executing functions in random globals from https://web-platform-tests.org/writing-tests/channels.html work? It's also not clear to me where RemoteGlobal's postMessage might end up or how they can reply. As far as I can tell this isn't actually used by anyone.
12:25
<zcorpan>
annevk: https://github.com/web-platform-tests/rfcs/blob/master/rfcs/remote_channel.md cc jgraham
13:35
<annevk>
zcorpan: thanks. I'm going to use dispatcher.js for now. That seems to suffice and is already integrated to some extent.
14:10
<annevk>
https://github.com/web-platform-tests/wpt/pull/52494 is the PR, in case anyone is curious.