On Wed, Nov 3, 2021 at 12:22 AM Daniel Latypov <dlatypov@xxxxxxxxxx> wrote: > > This formalizes the checks KUnit maintainers have been running (or in > other cases: forgetting to run). > > This script also runs them all in parallel to minimize friction (pytype > can be fairly slow, but not slower than running kunit.py). > > Example output: > $ ./tools/testing/kunit/run_checks.py > Waiting on 4 checks (kunit_tool_test.py, kunit smoke test, pytype, mypy)... > kunit_tool_test.py: PASSED > mypy: PASSED > pytype: PASSED > kunit smoke test: PASSED > > On failure or timeout (5 minutes), it'll dump out the stdout/stderr. > E.g. adding in a type-checking error: > mypy: FAILED > > kunit.py:54: error: Name 'nonexistent_function' is not defined > > Found 1 error in 1 file (checked 8 source files) > > mypy and pytype are two Python type-checkers and must be installed. > This file treats them as optional and will mark them as SKIPPED if not > installed. > > This tool also runs `kunit.py run --kunitconfig=lib/kunit` to run > KUnit's own KUnit tests and to verify KUnit kernel code and kunit.py > play nicely together. > > It uses --build_dir=kunit_run_checks so as not to clobber the default > build_dir, which helps make it faster by reducing the need to rebuild, > esp. if you're been passing in --arch instead of using UML. > > Signed-off-by: Daniel Latypov <dlatypov@xxxxxxxxxx> > Reviewed-by: David Gow <davidgow@xxxxxxxxxx> Reviewed-by: Brendan Higgins <brendanhiggins@xxxxxxxxxx>
