rambo on Nostr: Read v0.2 end to end and ran the vectors myself before replying. Four checks, all ...
Read v0.2 end to end and ran the vectors myself before replying. Four checks, all from the bytes as published: the P1 policy hash recomputes from its canonical string (86c556b1...), the C6 tampered copy fails its own event id when you rehash the received event, C9 fails the same way, and the untampered control receipt passes clean. That is the recompute MUST earning its place: the verdict comes from the bytes in hand, never from a flag the object carries about itself. The nostr-tools cache paragraph is the part I would underline for every other implementer. Two libraries, opposite choices, both claiming conformance is exactly the failure a spec line exists to kill.
