I've tested pa with rage and it does indeed work without issues just by changing the binary name, so maybe there's no reason to limit the options. I'd leave it as an undocumented option though, age would still be recommended as a reference implementation with a more active development (and it's still faster AFAIU - str4d/rage#57). Passage providesPASSAGE_AGE option to change the binary name, but I think that solution is worse: in addition to being a redundant configuration option that can just be inferred from what's installed, age spec doesn't specify API for CLIs, so there's no guarantee that any CLI that's going to implement age protocol is going to be interoperable in terms of interface with age(1) the way rage is.
lgtm! i don't see why we shouldn't fall back to rage, especially given that rage aims to be interoperable with age. if rage ever changes this stance, maybe we reconsider!
I've tested pa with rage and it does indeed work without issues just by changing the binary name, so maybe there's no reason to limit the options. I'd leave it as an undocumented option though, age would still be recommended as a reference implementation with a more active development (and it's still faster AFAIU - str4d/rage#57). Passage provides
PASSAGE_AGE
option to change the binary name, but I think that solution is worse: in addition to being a redundant configuration option that can just be inferred from what's installed, age spec doesn't specify API for CLIs, so there's no guarantee that any CLI that's going to implement age protocol is going to be interoperable in terms of interface with age(1) the way rage is.