Hi, we use Rails/Cucumber/RSpec, and run CircleCI on pull request branches. Our full test suite takes about 25 minutes to run on CircleCI.
If we have a few failing tests, we then do a subsequent git commit, it would be great if the first set of test that get run are the ones that failed last time. Then, if these fail, it stops, and if they pass, it then goes on to run the full suite.
That way we could avoid waiting 25 minutes only to find the same tests failed again.
That reruns an entire failed step. Usually all the specs are in one step, so I suppose people could split it up. I’m not familiar enough to know if that would impair it’s ability to parallelize the specs itself or not. I’m also interested in figuring out how we might be able to do this, because it would save a lot of time and CPU credits.