- Wednesday: I tried to work out some details of how Seed should work.
- Thursday: I made an effort to work through some of the logic underlying Seed's operations.
- Friday: I played Nova Drift, stewed in the heat, and considered which features of Seed to de-prioritize.
- Saturday: I did some quick work on Seed while I was writing the post, focusing on designing the "fork" command.
- Sunday: I started implementing the "fork" command.
- Monday: I wrote bit more code for Seed, and got back to the work on Jedi; I think it's mostly done, but the tests are failing in some configurations on Windows for reasons I don't understand yet.
Next week, I'm going to take it easier because we're traveling again. I've been thinking about where I want to take some projects.
Like, it's conceivable that Seed and pip could end up in a state where the negative space of the two of them working together could be filled by some other PEP 517/518 build system than Poetry. It's not that I'm in a hurry to replace Poetry, but I have been noticing ways in which it's not a precise fit for my workflow. And if Seed and pip are eventually going to implement some of the same features anyway... Note that Seed isn't well-placed to compete with most Python packaging solutions, since it's a general-purpose tool for project lifecycle management, built on top of a specific revision control system.
I was also thinking about MonFree, and the role that grind should play in game design.