Today work has focused on two things; The ctest extension matters and the Lean book problems. Both tasks have seen very slow progress because I took handwritten notes for most of the day.
The ctest extension stuff I have not gone through any other changes beyond the filtering logic, which had a pending refactor. I also implemented a cheaper iterator over items in an item container and decoupled some logic into a macro.
Beyond that, though, work has only consisted of taking notes about the approach to parsing I took with the tuple-based state monad. Being handwritten, it’s slower, but I would dare say the logic seems solid. Testing will tell.
Lean-wise, I also didn’t get a lot of work done because I tried again to take handwritten notes. I was planning on abstracting away all operations from the evaluator into an untyped representation for functions.
The idea here was to perform evaluation based off of the tokens used between operands or before a delimited list of operands. The tokens proper would be handed off to a handler extending the monad interface for interpretation.
Along the way, and once I got back to the computer, I realized most of this approach was not feasible. The actual idea I had was a tad bit crazier than what I explained above, and it really wasn’t hard to realize its shortcomings.
I would still like to adscribe no meaning to an operation and to merely support both infix as well as prefix notation for functions in the arithmetic language. The infix notation would be assumed to apply to only two operands.
The prefix notation is the more complex part because it would have to be delimited such that the untyped operation had a known set of arguments to pass to the monadic handler. This I’m working on right now.
Getting an ergonomic API going when the known set of operations is unknown seems like one though endeavor.
None.
The ctest extension matters I really do hope to push forward tomorrow — Handwritten notes have consumed a lot of time today, and I didn’t even start testing, much less debugging the parsing implementation.
Lean-wise, I would also like to test out more stuff in front of the computer because a lot of the ideas I have on paper are garbage.