Replies (23)

Bfgreen's avatar
Bfgreen _@briangreen.net 7 months ago
The interface is superb, but I'm not too sure of the results. There were 141 Dead npubs found, all with zero notes/zaps and no activity "never" - and I know several that I have interactied with in the last week. image
1 replies ↓
⚡️DEFAUXKING⚡️'s avatar
⚡️DEFAUXKING⚡️ Defauxking@primal.net 7 months ago
Like I said, it doesn’t work well enough to just unfollow all yet 🤷🏻 I’ve noticed it too with others I just can’t figure out what the common issue is with those that always show inactive like @utxo the webmaster 🧑‍💻 I’ve tried to run it back multiple ways: But my only guess is the ones that are active but show otherwise run exclusively on paid/private relays. And maybe some on relays from certain countries I can’t tap into it seems If someone could advise in any way, I’m all ears.
3 replies ↓
utxo the webmaster 🧑‍💻's avatar
utxo the webmaster 🧑‍💻 _@utxo.one 7 months ago
You need to use an outbox model, so check for activity based on the relays each individual user set in their relaylist, you should probably cache a map of npub -> relay i do use exclusively small relays, i bet you will find that @fiatjaf also shows as inactive often
1 replies ↓
⚡️DEFAUXKING⚡️'s avatar
⚡️DEFAUXKING⚡️ Defauxking@primal.net 7 months ago
Well it seems to have performed well. The only npub that proved to be an exception is @Alan ₿ but other than that it seems entirely accurate Sorry Alan, but I don’t know why I can’t prove you’ve done anything in 140days. Even though I know you just posted. Any ideas? image
⚡️DEFAUXKING⚡️'s avatar
⚡️DEFAUXKING⚡️ Defauxking@primal.net 7 months ago
Well, I went ahead and tried to update it with what you said, but it seems that it is just as slow somehow when utilizing the cluster of relays. "Clustered Outbox Analysis Running Discovers user relays, clusters users by common relays, then batch queries each relay cluster for maximum speed and accuracy Checking: Relay clustering + batch queries" Have any more ideas of what it might be doing inefficiently?
1 replies ↓
⚡️DEFAUXKING⚡️'s avatar
⚡️DEFAUXKING⚡️ Defauxking@primal.net 7 months ago
The optimization works by: Phase 1: Discover relays for each user (NIP-65, NIP-02, activity-based) Phase 2: Create relay clusters Map<relay, Set<pubkey>> Phase 3: For each relay cluster, make 3 optimized batch queries Phase 4: Distribute results back to individual users with deduplication
1 replies ↓
utxo the webmaster 🧑‍💻's avatar
utxo the webmaster 🧑‍💻 _@utxo.one 7 months ago
the discovery should probably just happen once a day on the backend job, then each user is just served a cache based map on their existing list. basically your backend should know 99% of all npub's outbox relays without checking for each user. then a user comes with a follow list, you should be able to use that to return a cluster map nearly instantly then req all latest notes limit 1 per author to every relay at once, a browser can handle easily 50 websockets
↑