porting and maintaining system data structures per linux source code, which is often behind layers of macros and optimization flags, is a pain. If you miss an implicit data alignment and it goes completely wrong. Currently we do repr(C) and hope it matches the api/abis of the host system, and as the previous bug reports showed, we made bunch of mistakes.
My question: does it worth to use more native APIs/ABIs/data structures via FFI instead of making our own copy?
porting and maintaining system data structures per linux source code, which is often behind layers of macros and optimization flags, is a pain. If you miss an implicit data alignment and it goes completely wrong. Currently we do
repr(C)and hope it matches the api/abis of the host system, and as the previous bug reports showed, we made bunch of mistakes.My question: does it worth to use more native APIs/ABIs/data structures via FFI instead of making our own copy?