mikedilger on Nostr: > I keep reading this and I try to find an answer to the question "how can I resolve ...
> I keep reading this and I try to find an answer to the question "how can I resolve a URL to the host in less than 500ms at worse if I am CLI like nslookup, with cold start and now context" and the only answer I can find is "ask Damus and wish for the best" am I wrong?
You've lost me. Let me make a few points that might be clarifying:
1) The outbox model is fundamentally about declaring that authors will be publishing to locations of their choice. This is in contrast to an email/inbox model where authors publish to places of the recipient's choice (the recipients SMTP server is an inbox). HTTP, RSS, and some parts of nostr use outbox model. Email and ActivityPub use inbox model. I tried to call the whole problem space "relay rendezvouz" because the fundamental problem is "How can we communicate? Somehow we need to rendezvous at the same relays." But the term never caught on.
2) The mapping from pubkey-to-url includes data about how that pubkey uses that url. We could have done this in a hundred different ways. We could have used DNS TXT records. But we did it as a nostr event. That's not fundamentally necessary for such a system to work, it is just how we did it. And now there is 2 years of code built on that. Fundamentally there just needs to be some system whereby people can lookup what my outbox relays are and what my inbox relays are, I need to be able to change them, and look them up for any pubkey. This could be Pkarr if it is better suited. I'm open minded. But we need to debate the details, usually we do that on the github nips repo.
3) Today is the first day I've seen your name/avatar on Nostr and the first time I've heard of Pkarr so you'll forgive me as I learn. And today is my weekend, not a day I'm terribly active on nostr.
4) Damus is the most famous nostr client that doesn't use the outbox model (unless they have changed recently). So it is odd to hear it as an example of the use of the outbox model.
> Also, if I remember correctly, you had a section discussing the limitation of this approach and it being suited best to social apps, and that for the general case you need a DHT, was that true or me hallucinating memories? If so, did you change your mind?
It is only for events that someone wants to publish for worldwide reading (like a web page). Not for every way of using nostr.
And yes I've always thought that a DHT perhaps made sense for bootstrapping -- finding the kind 10002 events -- rather than just looking at popular relays. But I never learned enough about DHTs to suggest any next step.
You've lost me. Let me make a few points that might be clarifying:
1) The outbox model is fundamentally about declaring that authors will be publishing to locations of their choice. This is in contrast to an email/inbox model where authors publish to places of the recipient's choice (the recipients SMTP server is an inbox). HTTP, RSS, and some parts of nostr use outbox model. Email and ActivityPub use inbox model. I tried to call the whole problem space "relay rendezvouz" because the fundamental problem is "How can we communicate? Somehow we need to rendezvous at the same relays." But the term never caught on.
2) The mapping from pubkey-to-url includes data about how that pubkey uses that url. We could have done this in a hundred different ways. We could have used DNS TXT records. But we did it as a nostr event. That's not fundamentally necessary for such a system to work, it is just how we did it. And now there is 2 years of code built on that. Fundamentally there just needs to be some system whereby people can lookup what my outbox relays are and what my inbox relays are, I need to be able to change them, and look them up for any pubkey. This could be Pkarr if it is better suited. I'm open minded. But we need to debate the details, usually we do that on the github nips repo.
3) Today is the first day I've seen your name/avatar on Nostr and the first time I've heard of Pkarr so you'll forgive me as I learn. And today is my weekend, not a day I'm terribly active on nostr.
4) Damus is the most famous nostr client that doesn't use the outbox model (unless they have changed recently). So it is odd to hear it as an example of the use of the outbox model.
> Also, if I remember correctly, you had a section discussing the limitation of this approach and it being suited best to social apps, and that for the general case you need a DHT, was that true or me hallucinating memories? If so, did you change your mind?
It is only for events that someone wants to publish for worldwide reading (like a web page). Not for every way of using nostr.
And yes I've always thought that a DHT perhaps made sense for bootstrapping -- finding the kind 10002 events -- rather than just looking at popular relays. But I never learned enough about DHTs to suggest any next step.