Okay, so, something about today hit me hard and I'm pretty low on energy, so I'm just going to write up the current state of the task runner's tests, and figure out if there's anything to salvage, or if I should just write now tests and delete these.
So far as the generator-provided tests go, theres:
- With nothing special done, the CLI exits with status code 0. This does not correspond to the behavior I want; the task runner will require a configuration file, so it should search for one and error out. This gives us two tests: expect an error with no configuration file, and expect minimal output and success given a minimal configuration file.
- The base command accepts a --debug flag. I haven't decided whether that fits into the design or not. I'll either drop it, or make it parameter to the test above.
- The final generated test runs the subcommand with various arguments and makes assertions about the resulting render calls. While I want tests against the subcommand, I'm not sure how I feel about the specific things that these tests look for.
I think my plan is to sketch out the code that I want to try to test first, see if there are any dependencies I should work on first, then write tests based on the sketches and work on getting the tests to pass.
Since I'm kind of out of it right now, I think the smart course of action is to take a break for a few days and work on other things in the meantime. For now, I'm going to wind down a little earlier than usual, because I'm tired and I might have a rough day ahead of me.