databases: giftwrap deletions

[TheAwiteb]

• We should not delete user-authored giftwraps during vanish and delete requests, since "giftwraps are owned by the recipient, not the sender".
• We need to allow the public key to delete giftwraps that p-tag it (we already support this in vanish requests).
• A database test for this would be helpful.

---

Since signing keys are random, relays SHOULD delete kind:1059 events whose p-tag matches the signer of NIP-09 deletions or NIP-62 vanish requests.

https://github.com/nostr-protocol/nips/blob/2ab507db107d0af2aa9a61327151799a1ad2d491/59.md?plain=1#L128

---

To prevent participants from deleting each other's giftwraps, relays should reject giftwrap deletions by author. This matches NIP-59, which already requires relays to allow deletion by the p-tagged user: giftwraps are owned by the recipient, not the sender. Clients should use at least one relay that rejects giftwrap deletions.

https://github.com/concord-protocol/concord/blob/96f0647d774bccfa383673c6f2f3588f7574abaa/01.md?plain=1#L80

[TheAwiteb]

@JSKitty, did I miss anything about giftwraps and Concord?

[TheAwiteb]

@JSKitty Ping

[JSKitty]

This is correct! And is fundamentally what makes Concord work properly:

• By restricting author deletion, Concord members can't sabotage other member activities (which is correct to the NIP-59 Spec anyway).
• By allowing deletion via p-tag, members may create ephemeral keys that are p-tagged and allow the deletion of their own messages, without revealing who they are, nor enabling griefing.

Strfry just (partially?) merged this functionality, as it was not entirely adhering to NIP-59, I suspect more relays and libraries are going to adopt this, and I fully support it, for both spec compliance and Concord compatibility. 🙏