upon closer inspection, aliasing issues with converting three-operand instructions into two-operand ones might not be so bad. the only complicated case is a <- b op a with op not commutative, so just integer division, but i already need to push push pop pop because intel's stupid. nothing clever to be done here
=> More informations about this toot | More toots from 40whirl@tech.lgbt
@40whirl what is this work for
=> More informations about this toot | More toots from hipsterelectron@circumstances.run
@hipsterelectron minimal effort backend because i don't want to learn how to use llvm
=> More informations about this toot | More toots from 40whirl@tech.lgbt
@40whirl awesome
=> More informations about this toot | More toots from hipsterelectron@circumstances.run
@40whirl i was looking at cranelift
=> More informations about this toot | More toots from hipsterelectron@circumstances.run
@hipsterelectron ooh, i get the impression that it's quite popular these days
=> More informations about this toot | More toots from 40whirl@tech.lgbt
@40whirl API looks nice. but i was also thinking of doing an interpreter or VM so i could virtualize or distribute execution and packaging is slightly easier
=> More informations about this toot | More toots from hipsterelectron@circumstances.run
@hipsterelectron @40whirl yeah, the internals are also very nice to work with in my limited experience.
=> More informations about this toot | More toots from mei@donotsta.re
@mei @40whirl oh that's neat !!
=> More informations about this toot | More toots from hipsterelectron@circumstances.run This content has been proxied by September (3851b).Proxy Information
text/gemini