Open csuwildcat opened 1 year ago
What happens if those nodes go away or you want to rotate nodes out of that set?
That must be handled by users migrating where they keep their DID-related events. There are no way it will be possible to distribute the data to all relays, but users (clients) must be able to migrate history between relays as they disappear or retire.
Adding a pointer to my response here for any observer who may view the thread: https://github.com/TBD54566975/did-nostr/issues/2#issuecomment-1446101709
@csuwildcat this is a good point. Perhaps we could separate the work into two parts:
I ask the above question because I am wondering how reliable, branch-safe resolution would work if there is no strongly, globally retained, independently verifiable, singular state you can boot to. If all relays are not required to retain events, does this just effectively reduce to a user declaring a few trusted nodes that are willing to treat these particular messages as forever-held? What happens if those nodes go away or you want to rotate nodes out of that set?
Conversely, if all nodes are required to propagate and globally retain all DID-related events for all DIDs, what prevents me from generating a billion DIDs in a minute on my computer and streaming them to a bunch of nodes to unlimitedly bloat the DID-space of the network to the point they can't possibly all be holding it?