just remembered why the interpreter has to be half rewritten.. Currently the objects don't anywhere store whether they're dependent on the input/whether their type is known at all
oh yeah I was gonna integrate that in the function setup so I wouldn't have to deal with it
I'm thinking now whether splitting the execution from the pseudo-stack would be a better idea..
pros: simpler to implement and less dealing with the weirdness of two uses in one cons: adding a built-in takes adding it in two separate parts of code
so later I could execute the function inside another one, e.g. fn("+")(10, 5). Well maybe not for + but for more complicated functions that could be useful
having that not possible is one of my biggest mistakes with SOGL
@ASCII-only well the definition of the function is tt s, meaning that it has to push a string one way or another. The function never pushes anything, so the interpreter, trying to fix that, pushes what the function returned