I noticed some test failures today where the test is failing the same in all browsers:
https://wpt.fyi/results/css/css-backgrounds/animations/border-image-width-interpolation.html?label=master&label=experimental&aligned&view=subtest&q=firefox%3Afail

https://wpt.fyi/results/css/css-flexbox/animation/flex-basis-interpolation.html?label=master&label=experimental&aligned&view=subtest&q=firefox%3Afail

We don't have these annotated as failures locally, possibly because the failures are caused by a change that we haven't merged yet, I suspect (?). Anyway, it looks like these are depending on us (and other browsers) to fix bug 1805727.

It looks like the failures are in calls to test_no_interpolation with values-that-cannot-be-interpolated-between. I think the test is just exercising bug 1805727, which was a recent csswg resolution that nobody has implemented yet (but apparently is becoming part of interop2023 via these test updates).

The work here probably will be done in bug 1805727, but I filed this bug for any notes that are specifically about the test failures, since there may be a handful and they may be spread out among the various interop2023 components.

Two other interop2023-relevant tests whose failures are this same issue:
https://wpt.fyi/results/css/css-backgrounds/animations/border-image-slice-interpolation.html?label=master&label=experimental&aligned&view=subtest&q=firefox%3Afail
https://wpt.fyi/results/css/css-backgrounds/animations/border-image-source-interpolation.html?label=master&label=experimental&aligned&view=subtest&q=firefox%3Afail

That last one (for border-image-source) has some tests that we fail but other browsers pass. It looks like that's because other browsers do have varying degrees of support for transitioning properties that are entirely discrete values, like border-image-source, I guess?

Boris noticed that we have some of these failures in the various -interpolation tests in https://wpt.fyi/results/css/motion/animation , too.

There are some in css-masking, too:
https://wpt.fyi/results/css/css-masking/animations/mask-image-interpolation.html
https://wpt.fyi/results/css/css-masking/animations/clip-path-interpolation-shape.html
--> adding the corresponding metabug.

(In reply to Daniel Holbert [:dholbert] from comment #0)

It looks like the failures are in calls to test_no_interpolation with values-that-cannot-be-interpolated-between. I think the test is just exercising bug 1805727, which was a recent csswg resolution that nobody has implemented yet (but apparently is becoming part of interop2023 via these test updates).

Update: as discussed in bug 1805727 comment 17 and bug 1805727 comment 23, Chromium folks discovered that the csswg resolution isn't shippable, and it's probably being morphed to be behind some new css keyword, to make it opt-in.

Per https://github.com/web-platform-tests/wpt/issues/39871#issuecomment-1557678903 , I suspect that this means the various *-interpolation.html tests will have their expectations reverted, and we (and other browsers) will start passing all of them (at least, until they're potentially extended to add tests for the new behavior, once we've sorted out a syntax to ask for the new behavior).

So: this bug will hopefully become fixed-via-upstream-test-changes soon.

There's a pull request to address this here:
https://github.com/web-platform-tests/wpt/pull/40177

It's been sitting in a ready-to-merge state for a little while; I just posted a comment to check on the status.

(That pull request is still nearly-merged; it's under review in https://chromium-review.googlesource.com/c/chromium/src/+/4558664 and got some review feedback on Friday.)

The upstream pull request was just merged.

Once we've gotten a wpt.fyi upstream test-run with the updated tests, we should see Firefox passing all subtests on e.g. this test from comment 0 here:
https://wpt.fyi/results/css/css-backgrounds/animations/border-image-width-interpolation.html?label=master&label=experimental&product=chrome&product=firefox&aligned&view=subtest

This is essentially fixed now, via the aforementioned pull request that reverted the expectation-change to these tests.

The tests in comment 0, comment 1, and comment 3 all pass, except for https://wpt.fyi/results/css/css-masking/animations/clip-path-interpolation-shape.html?label=experimental&label=master&aligned which passes in no browsers, I think due to using unsupported syntax (not this transition issue).

The tests in the comment 2 directory mostly all pass now. The remaining failures there are likely property-specific edge cases that are unrelated to this bug here.

Hence, closing this as FIXED upstream via test-changes.