Andres Freund <andres@anarazel.de> writes:
> Right, but why don't we allow for more tests in a group, and then use a
> default max_connections to limit concurrency? Having larger groups is
> advantageous wrt test runtime - it reduces the number of artificial
> serialization point where the slowest test slows things down. Obviously
> there's still a few groups that are needed for test interdependency
> management, but that's comparatively rare. We have have plenty groups
> that are just broken up to stay below max_concurrent_tests.
Meh. That would also greatly increase the scope for hard-to-reproduce
conflicts between concurrent tests. I'm not especially excited about
going there.
regards, tom lane