nostr:npub10npj3gydmv40m70ehemmal6vsdyfl7tewgvz043g54p0x23y0s8qzztl5h From a system engineering pespective, if we can all agree on the data structure of Nostr events, why do we have to have the same API? For example, a GraphQL API over a relay can be much more expressive than current filters API. Does it really make things more complex? Not necessarily. Mainstream web systems are request/response oriented and streaming API is added later. Nostr is the other way around. It does not have request/response model for its events.
i can definitely agree with that, the filters are pretty crappy
they are very cheap to implement though
i haven't thought very much about how i would do it better, to be honest, but all the optional fields is a mess for parsing... it was probably the biggest job in my rewrite of the JSON parsing for my relay (i completely wrote it myself, took me about 3 weeks, and it's way faster and more efficient than any json parsing library could ever be)
No replies yet.
