Comment by t43562
3 years ago
With python you could choose to use Pypy and make it work for your system - that might speed things up.
Alternatively, you could use smoketests. In my last largish project we grouped the tests into a short run and long run group. This allowed us to develop code and do a set of short tests in a fairly quick cycle and then before merging to the master branch we would do more comprehensive tests or let the CI system do it.
FWIW I got more joy from python by a long way but go isn't that awful except when you're trying to fork a github module to make it do something different.
Oh ya know, I never did consider PyPy. Is it significantly faster than standard CPython? I'm going to look into trying it out. Would be amazing if we could easily get performance improvements with it.
Also, your idea about having a smaller set of quick running tests and then separate smoke tests is actually a great idea. What we've really done in many of the cases where we should have a unit tests, is essentially to write integration tests. Not quite end to end but pretty close and that's where a lot of the slowness comes from.
Appreciate the reply!