Require specific versions of test suite's Python dependencies?
|Reported by:||JustinTArthur||Owned by:||nobody|
|Severity:||Normal||Keywords:||testing, test suite, continuous integration, ci, dependencies, pip, pypi|
|Has patch:||no||Needs documentation:||no|
|Needs tests:||no||Patch needs improvement:||no|
Currently, there are several tests that depend on 3rd party libraries that we haven't assimilated into the source code directly. These tests are simply skipped/passed when a developer or tester's environment is missing one of these libraries.
The question is:
Do we want to specify the preferred versions of these dependencies in test/requirements/*.txt?
Do we want to have the CI environment(s) use those text files for the automated testing we do? (they currently don't according to timo)
I vote for at least adjusting those text files to refer to the same dependency versions as our CI environments are currently using for the sake of giving contributors a consistent regression testing process (their passing-on-local test less likely to fail-on-continuous-integration).
We may also have a day when a developer's local test suite goes red, not because Django has regressed, but because one of the test's dependencies has changed in API or behavior. This could be desired behavior? I don't know.