Pull request: https://github.com/WebKit/WebKit/pull/16813

Committed 267023@main (e225f383b875): <https://commits.webkit.org/267023@main>

Reviewed commits have been landed. Closing PR #16813 and removing active labels.

<rdar://problem/114057284>

Isn't this needed everywhere? Is there something wrong with the implementation on other ports?

(In reply to Michael Catanzaro from comment #4)
> Isn't this needed everywhere? Is there something wrong with the
> implementation on other ports?

The current code is specific to the CoreFoundation runloop implementation. Similar changes for GTK's run loop API would be needed on that platform (and likewise for Windows).

Can you give me some pointers to what needs implemented, please? I skimmed through all recent commits looking for platform-specific code and didn't find anything. And since the implementation is built unconditionally, we already know it doesn't use any platform-specific methods of RunLoop.

GTK/WPE will have an issue where idle callbacks will be scheduled at too-high priority, which will need to be fixed, but that shouldn't stop the feature from working.

(In reply to Michael Catanzaro from comment #6)
> Can you give me some pointers to what needs implemented, please? I skimmed
> through all recent commits looking for platform-specific code and didn't
> find anything. And since the implementation is built unconditionally, we
> already know it doesn't use any platform-specific methods of RunLoop.
> 
> GTK/WPE will have an issue where idle callbacks will be scheduled at
> too-high priority, which will need to be fixed, but that shouldn't stop the
> feature from working.

You need to implement these functions: RunLoopObserver::schedule, RunLoopObserver::invalidate, and RunLoopObserver::isScheduled():

https://github.com/WebKit/WebKit/blob/main/Source/WebCore/platform/RunLoopObserver.cpp

Thanks!

Unfortunately this looks hard, so I've created bug #260478 to track it.