run the full CPython test suite (regrtest) inside a subinterpreter
create an issue for each failure
There are also possible issues that could arise when using a stdlib module in two different subinterpreters, which the test suite doesn't surface. It would be nice to have a strategy for covering that case.
It might also be reasonable (in the long term) to add a test for each stdlib modules that checks subinterpreter support.
would it be a good idea to run each test twice, in their own subinterpreters simultaneously (or at least as simultaneously as the current implementation of subinterpreters allows)?
This could be done with the following:
There are also possible issues that could arise when using a stdlib module in two different subinterpreters, which the test suite doesn't surface. It would be nice to have a strategy for covering that case.
It might also be reasonable (in the long term) to add a test for each stdlib modules that checks subinterpreter support.