@soatok i still have not found any reason why RFC 9421 is any better than the draft that isn't just "its an rfc therefore it's better". from what little i can see it just seems like overcomplication for no good reason (for fedi use cases anyway)
later http signature drafts set the key algorithm to "hs2019" which means "ignore the algorithm field", you could just as well run ed25519 over them (which i support and plan to implement on my own software)
@kopper@not-brain.d.on-t.work