This is more of an RFC than a bug report per se.

https://github.com/w3c/web-platform-tests/commit/c867b9c6d192fd98d6cbe69298dddaf9b54ca7c1 was imported into Blink in https://chromium-review.googlesource.com/c/610122. However, not all tests are passing, and some can take a bit too long to run, which led to wpt-importer marking the test with [ Timeout ] in that CL. Consequently, the tests's accompanying -expected.txt file was not rebaselined.

The time it takes for the test to run varies though, so when it did finish running the mismatch between the test expectations and the test's new results would lead to a failure. As far as I can see, a sheriff tried to work around it with https://chromium-review.googlesource.com/610897 by marking it with [ Failure ] instead of rebaselining it.

Earlier today, another sheriff changed the expectations to [ Pass Failure Timeout ] with https://chromium-review.googlesource.com/612060 to attempt to cover all possible results (I think the Pass bit is wrong by the way). I then rebaselined the test and changed the expectations again in https://chromium-review.googlesource.com/c/612164.

What Philip and I were wondering is whether automation could have helped here: if it wasn't for the timeouts in the initial import, wpt-importer would have rebaselined the test; should it have tried doing that once the test started failing instead of timing out in more recent imports?