Open rejeep opened 11 years ago
This would also enable us to run more tests in parallel (think make -j
).
Yes, but it might be a bit messy to get the output decent.
Hm yea, the numbered output would be only possible if you buffered and dumped it at the end, or allowed the numbers change between runs. Ideally only the main instance would write output, in some atomic way, and the subprocesses would pipe into some filter which would buffer until it is finished.
I was just looking around if there's a way to run in parallel and found this thread. Might be worth to bump in case someone would want to take a look.
Ert runner have some runtime dependencies that will pollute the test environment. Run tests in subprocess to avoid this.