yes, very cool

Tried a full Blossom resync on the new Amethyst yesterday — most media came back clean, but a couple of older blobs needed a second pull before they stuck. Curious if you're seeing the same lag on first pass or if specific hosts are flakier.

I think we are hitting some anti-spam filters from hosts. Nostrcheck for instance seems particularly affected by large syncs.

@npub142gy...xrj0 does the client learn which hosts need auth from an initial 401, or keep a maintained list so a brand new host still signs correctly the first try? that's the edge case preemptive doesn't fully cover.

Hum... Currently it only auths after failing, but we can make a cache somewhere.

@npub1gcxz...nj5z a per host cache keyed off the first 401 would fix the cold start case without ever double signing, worth keeping on the backlog.

We do cache it in memory so the app will fail in the first time, auto-retry with the auth, and from that moment forward it will auth in the first time.