Subdirectory tests lists the various tests (as subdirectories) that can be
run per distro (they are not to be don't run directly, rather they are run
inside containers per distro).
Subdirectory distros lists the Linux distros to test (as subdirectories).
Inside each subdirectory is test-settings.env file which sets:
- IMAGE = a reference to the image for podman(/docer) to pull
- TESTS = the tests to run (in
testssubdirectory)
There is also an optional status subdirectory per distro which may contain
status files for each test. The contents are PASS or FAIL.
Run ./polydistrotest for a fun time. Any distro-test pair which has a status file
will be skipped (but the status reported). Delete the
distros/<distro-id>/status/<test-name> files to run tests.
You can see what tests have passed or failed by doing
grep FAIL distros/*/status/*
grep PASS distros/*/status/*
See which tests are run for which distros:
grep TESTS -R distros/
Pip install tests are skipped for distros that do not have a Python 3.6+ package in the default distribution:
- Ubuntu 16.04 and Debian stretch (9) and have Python 3.5
- Debian jessie (8) has Python 3.4
- Scientific Linux 6 seems to only have Python 2
- RHEL 7 (Universal Base Image) has a Python 3 backport for subscription customers only
Install tests for system installed conda (yum-conda) is only for distros that (as of Aug 2020) have a system package for conda:
- Fedora 31, 32