I think it was earlier today that I hammered out the tests and formatting of my failure messages. They're pretty detailed and fairly consistent. They could maybe do with numeric codes for lookup later, but that's not needed right now, and kind of needs documentation first.
Anyway, it's time to get some more coverage. I'm kind of looking at the uncovered lines and thinking "Well, what would happen if I didn't give it any actions to take? That's a valid test."
Let's see what happens.
Well, it worked, but I decided it needed a few behavior changes, so I modified the relevant templates a little. Things are in a good place now. There are just a few lines of code left to test, but they're... kind of a doozy.
I'd gotten this far, see, without writing any legitimate Actions. To finish things up, I need actual Actionss that I can invoke from test code.
Let's see what I've got in mind...
- Delete directory tree (cross-platform)
- Run command, with optional statuses to not abort on.
- Using run command, create a virtual environment
- From a virtual environment, running a command within that virtual environment
These tasks should bring in the last of the prototype code. Once I have this functionality exposed via the API module, I can try running this code against its own repo and see if I actually get the kind of speedup I expected. Either way, I'll hopefully get some more useful profiling data.
For now, I want to wind down. There'll be plenty of time to get this working later. Maybe tomorrow, maybe in a few days, maybe in a week. I'll get there when I get there.