if there is no delete in #nostr, there also should not be replaceable events deleting old versions
my relay doesn't delete
it has delete but nobody sends them out and no clients have a button for it
a delete event will anyway be just a tombstone, meaning it should show "event deleted" in the interface but may or may not actually delete it, but should not return it to anyone except maybe the owners of the relay
deleting old events, especially replaceable ones, is something that does need to happen eventually... there is only so much storage on a relay...
i am currently working on a connection to an external database storage and implementing the relay local database as a cache instead of primary store is a big part of the work that will be done from march to april
i'm not really a big fan of the data store we will be using, but after this project is done and out the door maybe someone will see the sense in sponsoring me to bring #pnyxdb to production - this is a distributed database that has moderately weak consistency and can keep together a cluster of around 200 nodes, which is the biggest distributed database cluster size ever
https://github.com/technicolor-research/pnyxdb
i think actually that the protocol of #pnyxdb could turn out to be the solution for an even larger cluster replication strategy for #nostr - it builds on web of trust and has a broadcast mechanism and a quorum system that is subjectively computed based on the web of trust
i just learned that #pnyx is the name of a hill around athens that has been a historical gathering place for democratic assemblies