I refactored the FlatEx evaluation code a bit to separate the logic from the ignored number tracking to make changes to the operation implementation like #41 easier.
This should not impact performance of expressions with less then usize::BITS nodes. Large expressions could get a bit faster because I implemented a simple bit vector logic for [usize] instead of using [bool]. I wrote rather extensive tests to make shure this works correctly. I did not conduct clean performance benchmarks yet.
I refactored the
FlatExevaluation code a bit to separate the logic from the ignored number tracking to make changes to the operation implementation like #41 easier.This should not impact performance of expressions with less then
usize::BITSnodes. Large expressions could get a bit faster because I implemented a simple bit vector logic for[usize]instead of using[bool]. I wrote rather extensive tests to make shure this works correctly. I did not conduct clean performance benchmarks yet.