Once we have a Lisp port of #101 to address #61 , we might want to extend bitc to include modular arithmetic on n-bit-width natural numbers, and then make the natural-number backend use bitc. (Indeed, we could consider extending bitc first and porting to it directly.) Then we could continue to consider bitc as the single source of truth regarding what VampIR APIs are available to, and potentially used by, Geb, and the bitc backend itself would have knowledge of the whole program available for potential optimizations.
Once we have a Lisp port of #101 to address #61 , we might want to extend
bitcto include modular arithmetic onn-bit-width natural numbers, and then make the natural-number backend usebitc. (Indeed, we could consider extendingbitcfirst and porting to it directly.) Then we could continue to considerbitcas the single source of truth regarding what VampIR APIs are available to, and potentially used by, Geb, and thebitcbackend itself would have knowledge of the whole program available for potential optimizations.See this comment to #105 for some details.