Merge upstream changes from Warp's core-foundation-rs fork

[alokedesai]

Hey servo folks,
Sending a PR to merge Warp's changes to core-foundation-rs upstream. Note, these changes are purely additive, and shouldn't have an affect on any existing core-foundation-rs APIs:

Changes

1. Adds get_matrix to CTFont, matching CTFontGetMatrix.
2. Adds get_typographic_bounds to CTRun, matching CTRunGetTypographicBounds
3. Adds a patch for a panic in extract_number_for_key that occurs in Ventura. The underlying issue is an issue within the Core Framework library on Ventura--but this patch allows this library to not panic on Ventura if someone tries to access any of the font traits.

[alokedesai]

cc @madsmtm--tagging you in case you'd be interested in reviewing. I think the changes here (notably the change to fix a crash on Ventura) would be pretty valuable to be merged upstream. Thanks!

[jrmuizel]

Can you clean up the commit history and avoid the formatting changes?

[jrmuizel]

Can you also give some example code or add a test case for reproducing the Ventura failure?

[madsmtm]

Don't know much about CoreText, but probably looks fine to me

[bors-servo]

☔ The latest upstream changes (presumably b009c87) made this pull request unmergeable. Please resolve the merge conflicts.

[vorporeal]

In the name of finally getting (at least some of) these changes merged upstream, I've sent out #674 and #675 for the bits adding new functions.

In terms of the Ventura fix, I can't think of an easy way to write a test for it. FWIW, this has been live in Warp (used by ~100k macOS users) for a year now with no indication of it causing any issues. I can send out a separate PR for that as well, or we can keep it in our own fork; either works for us.

[nicoburns]

I can send out a separate PR for that as well, or we can keep it in our own fork; either works for us.

I definitely think we will want this fix upstream. A separate PR might make sense at this point seeing as it has been done for the other changes.

In terms of the Ventura fix, I can't think of an easy way to write a test for it.

I think the best test for this would be a usage of the API that triggers the bug without the fix if you have a standalone reproduction (perhaps using a particular system font). But unless anyone else strongly objects, I think it would be better to land this without a test than not at all if you are unable to provide that. A link to the original bug in warp that lead to this fix would also potentially be helpful.

I'm sorry that progress on reviewing/merging this has been so slow. The Servo project is still getting back up to speed after a long hiatus and is still quite behind on maintenance of libraries. Hopefully this will improve over the next few months.

[vorporeal]

Yeah we (@alokedesai or I) can put together a separate PR for the extract_number_for_key fix.

No need to apologize for latency here - there was an explicit request to clean up the commit history and provide some example/test code for the extract_number_for_key issue, and nobody from our side followed up. 😅

[vorporeal]

Hey all, finally got around to sending out one last PR for the Ventura crash fix portion of this PR: #735

As mentioned in the PR description, we attempted ~6 months ago to switch to the published version of this crate (in case Apple had fixed the underlying issue in a newer release) but immediately had to roll back that release of Warp due to it still being an issue in the wild.

[jdm]

I'm fine with these changes but it would be nice to clean up the commit history as previously requested.

[vorporeal]

With the PR you just approved, we'll be able to close this out - I've already merged separate PRs for the other bits.

(Feel free to close this one out now and we can focus on getting #735 merged.)

[jdm]

Cool, let's do that!