Closed Fuuzetsu closed 3 years ago
Thank you for the suggestion!
One common problem with this approach is:flaky tests are not rerun. If your tests are fully determistic, this is not a problem, of course.
Depending on your setup, the errors are often in the build log. Would be interesting to hear what others think.
Right, obviously this sucks for flaky tests, hence the opt-in. We don't have flaky tests and if re-running the test, it should just re-use the result.
I will submit something today.
We are running tests in multiple crates, creating the test report with something like
cargo2junit
and showing it in the CI. But if the crate just fails, we can't see the (failing) output and handle the failure reporting ourselves.Would be cool to unconditionally produce test output even if it failed.
Behind some flag of course.