mistral was able to synthesize pretty good program. it follows interface definition very well, and possible improvements are probably simply question about interface definition rather than limitation of used llm.
so what happens when one publishes an event? nobody can verify its signature because its detached.
however one can publish another event which references to the previous event and include signature to that event.
:DDD
are you really going to specify functionality of every web application that will ever exist in a github repository? because this is what you are doing with the nips repository. this is your scaling solution for the next decentralized web?
im just saying any number of event kinds is not gonna be enough due to collisions.
instead we should use more unique namespacing for specifying data structures.
search data rows can probably be compressed quite a lot.
for example if we have 20 000 tokens, that would equal to 20 000 bits per row. however if only 5 tokens activate, this only takes 5 bits + position data for these bits to store.
imo nostr should adapt full unrestricted filtering of events similar to any object db, such as mongodb:
https://www.mongodb.com/docs/manual/tutorial/query-documents/
bu..but performance.
yep,
thats a tradeoff between what applications can be built and what are the requirements for running relays.
most clients anyway do 10x too much queries based on the ui rendering times observed.