Conversation

Notices

1. lamp (lamp@kitty.haus)'s status on Sunday, 14-Jul-2024 05:25:04 JST lamp
   One of the flaws of #ActivityPub is the inability to guarantee consistent timelines, and how peers have to queue and repeatedly retry publishing when other peers go down, until it gives up and the message is lost.
   
   What I would do is assign a consecutive integer to each activity, so it is obvious when events are missing. This eliminates the need to buffer massive job queues, as the other peers can just ask for all events after or between indexes to synchronize their copies of your feeds.
   • lamp (lamp@kitty.haus)'s status on Sunday, 14-Jul-2024 08:23:21 JST lamp

     • gentooP
     @gentooP idk? it's down? any instance that's down is gonna miss stuff.