I've refined my ideas for Seed a bit. I currently have a bunch of interesting ideas for limitations of high-level concepts, that I'm going to need to see how it all shakes out in practice.
I just tossed together the very, very basics of the wrapper, just exposing all of the commands that I'm pretty sure need to be passed through as-is. I don't want to try to push through this right now; I've been getting to bed earlier lately, which is probably a good thing.
Anyway, I'm thinking when I have the time to try to develop this stuff, I'll try to use the workflows I have in mind manually.
One idea that I think I might look into later, that wouldn't fit into the base workflows, would be making a repository that's some transformation applied to another one. I'm not enthusiastic about this idea. The intent is to have some kind of workaround in case tools like Poetry end up supporting many more repo formats, but the same set of use-cases. (Specifically, not having an obvious way to deal with generated code.) My intuition is that shoving a repo into this mold would on average cause more problems than it would solve. Maybe instead there could be the possibility of creating a "build" branch that gets generated code added to auto-generated patches in a clone, then puts the artifacts back in the source repo. This would be a somewhat involved process, but most of it could be automated.
Anyway, I really need to wrap up. Good night.