i don’t think high throughput and efficient processing of private nostr data is compatible with the async signer model.
in nostrdb we have ndb_add_key which pushes private keys to the ingester threads so that they can efficiently unwrap giftwraps with optimal performance.
I can’t imagine how this would work with a remote signer doing decryption.
Login to reply
Replies (6)
Definitely a topic worth more discussion!
But a nsec outside of the signer renders any signer device garbage by definition...
Do you mean bunkers? Because it works quite well with nip07 and nip55 signers. It can work with bunkers (nip46) too, but you have to load (and decrypt) messages by time. Thankfully most feeds are by time.
With local relays bunkers also work quite fast, not as fast as nip 55
We are talking about many orders of magnitude difference of performance. With giftwrap you would have to have multiple rounds trips just for a single wrap (wrap, seal). This is a lot different than processing thousands in parallel on the scale of milliseconds.
If we are going to be DoS resistant it needs to be crazy fast.
We had to do this to for processing rss feeds and posting them to nostr. Really don't like it but no alternative other than making the user go through a bunch of drafts and sign them from client.