Ignore erroneous files that don't contain the required test

[yaelbh]

Suppose that we have two test files: relevant_tests.py and irrelevant_with_bad_import.py. relevant_tests.py is the only file containing a specific test named test_interesting. In addition, irrelevant_with_bad_import.py contains a bad import.

If I run

stestr run -n relevant_tests.py

then everything's fine.

If I run

stestr run - test_interesting

then I get an error because of the bad import of irrelevant_with_bad_import.py, and no test is executed.

I understand that the reason for the difference is that the latter triggers a discovery mechanism, which inspects all the tests files, including the one with the bad import. Unlike the first example, which from the beginning directs to one specific file.

However, in my opinion, erroneous test files that don't contain the required test should not trigger errors. And, even if yes, this should not prevent the test from running normally in the other, correct test file.