Pytest is a very cool (because easy to use) framework for writing automated tests in Python. It is much less verbose than the built-in unittest module, yet has the same (if not more) functionality and is plain great when it comes to detailled error output on failing tests.
As I have just started diving into the world of unittesting and test-driven development, I however found it especially helpful to not have to write all the boilerplate code that you would need with the unittest module. Just create a file that start with test_ and has some functions in it with the same prefix, and pytest will find and execute it automatically when called on any point above. The test output on failure is also much more helpful for debugging.
When writing tests for tools working with the Nuke API however, there is a problem. Since Nuke version 8.x you can use
import nuke directly in a script without having to start Nuke at all, but in prior versions you are required to do that. As my tests are for Nuke 6.x I can not start them the usual way from the commandline likepy.test tests/ , as
import nuke will fail.
My workaround was to open the Nuke GUI manually and start pytest from a script editor like:
import pytest pytest.main(['-vv', 'path/to/tests/'])
As you can imagine, that is not very convenient.Instead I copied the snipped to a python file called runTests.py and created a runTests.bat that calls this script using the Nuke python interpreter like:
"C:\Program Files\Nuke 6.x\Nuke6.x.exe" -t path/to/runTests.py
Nuke is still required to launch on every testrun, but only on your commandline, which is much faster than opening the GUI and you still get the usual colored pytest output 🙂
The setup is a bit hacky, but I found it to speed up my testing workflow quite a bit, and you only have to do it once. Of course your alias could be the python call directly instead, I just found it useful to have a .bat that I could use from outside of a shell.