DEV Community


Dropping local test runtime and scope with pytest

toonarmycaptain profile image toonarmycaptain ・2 min read

I love pytest.

I have a project that I use as something of a learning platform (and that I don't want to break because it's at least occasionally used in production by my wife..). Most recently I've used it to learn how to abstract a database layer, implement a SQLite database, and now I'm similarly implementing SQLAlchemy.

Database tests can take a...while to run.
The cross product of a test suite and a variety of database backends via pytest.mark.parametrize takes...longer.

The other day I googled "pytest only run specific parameters" and happened across the python -m pytest -k some_string invocation.

Now, I see that the documentation refers to strings in quotation marks,1 but I missed that, and my incovation python -m pytest -k sqlalchemy works just fine cut my test runtime down to 1/6th the full suite, running only tests specific to my SQLiteSQLAlchemyDatabase object in, and my common tests that run against every database backend, but only the parameters using my empty_sqlite_sqlalchemy_database fixture.2

So while I'm working on this backend, and assuming I'm not changing the database interface, I can run the full suite of tests, but only on the part of the code I'm working on. This way, I can ensure that the implementation works (or is getting closer to working), and that it's not breaking any thing that relies on it (because my common interface tests are running, and that one test for another component I would have forgotten that uses this code) as I go, without waiting for the entirety of the project's tests to run!

This is not only saving me time waiting, but means I'm not just running the tests more often, but receiving the feedback much closer to the writing of the code. You know, I'm actually checking the results of the tests before writing much/any subsequent code, that generally relies on the former actually working.

Discussion (0)

Editor guide