difference between fixture and yield_fixture in pytest

The only difference is in readability. I think (though I'm not 100% sure) the underlying behavior is identical (i.e. the cleanup after the yield statement is run as a finalizer). I always prefer using yield fixtures for cleanup, since it's more readable.

If you're using pytest <3.0, you'll still need to use pytest.yield_fixture to get that behavior. But if you're able to use pytest 3.0+, pytest.yield_fixture is deprecated and you can use pytest.fixture to get the same yield_fixture behavior.

Here are the explanatory docs:

Since pytest-3.0, fixtures using the normal fixture decorator can use a yield statement to provide fixture values and execute teardown code, exactly like yield_fixture in previous versions.

Marking functions as yield_fixture is still supported, but deprecated and should not be used in new code.

addfinalizer has two key differences over yield:

1. It is possible to register multiple finalizer functions.
2. Finalizers will always be called regardless if the fixture setup code raises an exception. This is handy to properly close all resources created by a fixture even if one of them fails to be created/acquired

From the pytest docs