I got really into improving Structured Data's tests from the standpoint of mutation testing, so I think most of the few surving mutants are some combination of "Oh, that's interesting", "Oh, that's... interesting", and "I don't understand why that commit didn't fix these problems".
I was thinking about this earlier, and I think I might be getting into a situation, again, where I'm neglecting some work in favor of lower-priority work that's (relatively) easier and more enjoyable. In this case, the fact that I'm trying to improve test coverage, when some of the desired behavior isn't actually specified in a well-justified way. And it's not specified in a well-justified way, because I don't have use cases yet that would motivate any particular behavior. I think the only way around this is to get people, including me, using this code outside of the context of its own test suite.
I'm thinking the way to go for that is to find problems on Rosetta Code or something, that lend themselves well to the strengths of the library. Rough priorities for doing this:
- Update the changelog as needed, and cut a new release.
- Put up a repo for this stuff, probably on GitHub, and start working on problems in it.
- If needed, get to work on the alternative function decorator idea.
Besides working on that, and unrelated to Structured Data, I want to have a nice hard-copy reference for the writing aids I'm working on, so I'm also figuring out what I want to have for that.
Okay, pretty loopy right now, definitely need to get ready for bed.