Whenever we have failures in a try push, we would, after fixing them, ideally only rerun the subset of tests that failed. But this is hard to get right, and most devs, I suspect, just re-run the whole try push. Is it within MozReview's ambitions to do this automatically? It already has the ability to do a try push, and even seems to know how many tests have succeeded and failed. When I submit a follow-up review, could it run (only) the failed tests again automatically?

This is a good idea, at least, providing some easy way to do it, even if we don't want to automatically rerun them.

We've talked a little bit about providing good defaults for try jobs in Bug 1159869. Using information on previously failed jobs makes good sense. We've discussed reimporting pull requests from GitHub if new revisions are pushed there. This is a very similar use case, I don't think it would require that much additional work to support.

MozReview is now obsolete. Please use Phabricator instead. Closing this bug.