Daily report (2026-11-10)

1. Summary

Today work has focused on two things; The ctest extension matters and the Lean problem on the monadic evaluator with multiway calculations. Both have seen great progress, and I can surprisingly say the parsing refactor was not buggy.

The first thing that I did was to test out the parsing refactor of the ctest extension PR. That turned out better than expected because I didn’t have to debug anything. The thing worked in one go.

While doing that, I found I could simplify the core parsing logic in syn’s visitor interface by moving some code to the post-parsing stage. That also went well, so I started testing if resolution worked again.

Testing out the successful cases from two weeks ago, I found that resolution also worked nicely. This does mean that we can finally come back to working on the next revision of the resolution algorithm.

Lean-wise, I have gone through a few iterations of the monadic evaluator. Thus far, I’ve found one idea that seems like it could work. Instead of having an expression register operations, we go one step back during parsing.

The expression representation on which evaluation takes place considers one of an atom or a function applied to some list of operands. This function injection currently also holds a function proper producing a monad over â„•.

This essentially requires that there be a monad in context during evaluation, such that it may perform the operation. The monad is really meant as an extension to the monad type class, where binding is not done during evaluation.

This extension type class is essentially a generic handler that takes a list of monads of its type, and a token (a raw string.) Then it performs whatever operations it sees fit and yields another constant in its type.

This I’ve yet to flesh out so that I don’t literally require the handler to be “registered” at parse-time. It would be best if the function injection on expressions didn’t have any functions within it.

2. Blockers

None.

3. Plan for the week

I expect to continue working on the next revision of the resolution algorithm tomorrow. I think this is finally it, and we should probably be bound to move on to the next stage of the PR by the end of next week.

Lean-wise, work will continue as per the usual.