@supersat LLVM IR is target dependent, you have different data layouts, different intrinsics for different architectures, different features supported by a specific code generator. So the scenario you describe might work (more likely w/o optimizations) though it's not guaranteed.
@sgoldfed And the price you pay for integration with the frontends LLVM support via wasm is that you either have no control on the architecture or you have to do translation from wasm assembly to your VM which is a layering violation and could regress tons of work LLVM does in codegen.
@chfast@gluk64@zksync Clang is an LLVM-based compiler, but LLVM IR is not a target-independent representation. Obviously no one added EVM target to Clang yet ;) And with i256 native ints in mind, here be dragons I would say.
@KevlinHenney @cleancpp @tvaneerd In my experience the meaning of ‘get’ is a ro/rw access to some capacity of an object. It might or might not perform computations and might or might not produce side effects. %)
Really GREAT alternative to ifdefs for a multi-platform #cpp application or for an application that supports different features depending on some compile-time known parameters. I hope we will be liberated from the preprocessor one day.
https://t.co/vi4NFjHJk2