Martin Robinson

I'm fairly sure that the binaries we produce right now run just fine on ARM Macs. While a native binary might be nice, is it worth maintaining an entirely new...

It feels like the right thing to do is not to have two builds, but just have proper support for building "fat binaries" on MacOS.

> @mrobinson do we have data on how often script blocks on layout? It seems that modulo layout queries it shouldn't? All layout is done synchronously with script. Previously, animations...

I think we'll need to keep the changes from https://github.com/servo/servo/commit/29f617056418eae88d396de82ddaf744766e1774 judging by the test results. I've added those back.

I think this is ready for review. The one failure is due to the new test expectation that has now landed in main.

> Are the plans to re-apply (any of) these changes upstream? We're trying to only revert changes that don't make sense to push upstream. In this case most of the...

> Should I add a more specific test? I think if tests start passing when the full feature is implemented, it's good enough. Thanks!

Clicking into the test runs, it looks like the timeouts might be related to GStreamer not behaving properly on the CI: ``` ▶ TIMEOUT [expected OK] /html/semantics/embedded-content/media-elements/track/track-element/track-mode-not-changed-by-new-track.html │ │ ALSA...

> I was reaching out to CI for feedback, because I am running on a Mac M2 and I read on Zulip that this might give flaky results (which it...

The next step is probably to update `string_cache` to use phf 0.11. The repository is here: http://github.com/servo/string_cache.