- Wednesday: I investigated browser-based Python stuff a little more, and found more bugs. (I found even more today; we'll see what comes of them.) I also did some work on the punq fork.
- Thursday: I rewrote my AST visitor code to use eval instead.
- Friday: I got the eval code working. The resulting code is significantly slower under test, I think because of the greater amount of work required by this approach. I think it'll be worth it, though.
- Saturday: I set up profiling for all of the test suites, and determined that I can mitigate the slowdown using an LRU cache, so I'm doing that. I started pondering refinements and changes to the rough interface that I... "designed" is the wrong word.
- Sunday: I made some progress on the Mythic CLI, and found other things that I'd like to do differently.
- Monday: I wasn't up for much, so I just got the test suite to basically pass again.
Next week, I'm working on actually planning this stuff, both in terms of what to refactor, and what I'm going to implement. I've got a big list of stuff that I'm going to flesh out with more and more information, including my plans for localization, and what I want to implement, from each of the seven systems.
One other thing that I apparently have to work out if I ever want to upgrade pip again for these projects, is how to replicate my old constraints-based flow, which is totally not how constraints files were meant to be used. The cleanest way is probably devpi, but... effort. The quickstart tells me to set up supervisord, and the first time I tried that, it didn't feel quick at all. I'll try again later, I guess. Or, I can try reimplementing URL constraints myself. PRs welcome, and all.