Great to see an initiative like this in testing! A few questions after a glance:
In your blog post you said:
The really great thing about pytest is that it really makes unit testing easy and quick. To shorten you feedback loops I recommend you to clearly distinguish between unit tests and ROS node or integration tests.
After reading the blog, I didn’t quite get how pytest makes unit testing “easy and quick”. I mean it looks like it does, but did you find it easier and quicker than unittest / nosetests that ROS is already integrated with? I’m simply curious, as ROS allows to run unit testing by nosetests as well as integrate it into Node testing.
ROS seems to pick up the test data very well. However, I noticed that failing tests still don’t trigger a failure when running rostests (but the log/terminal output shows the error). I would appreciate any input on why this is the case.
Not sure what you mean by “ROS seems to pick up the test data very well”. But I’d agree the test result is not the easiest to read. In your particular case, have you tried catkin_test_results?