Closed
Description
I want to teach test-run to collect test timings and report several largest ones after testing. This way we can spot tests that're near to limits (at least in some environments) and reduce them.
My criterias:
- Collect timings always (not under an option, locally and in CI both).
- Per-test × per-configuration statistics.
Wishful properties:
- Not only on Linux, but on Mac OS as well.
- Distiguish several runs of the same test (see The same Test object may be used several times #255).
What better to do with so called 'long_run' tests? Maybe mark them explicitly in the output?