Post

Alex
@alex.bsky.team

tap currently uses the full Relay as a data source rather than Jetstream. Jetstream v2 makes it possible to do stateless replay/snapshotting of historical data so you don’t need to use tap for a heavyweight stateful backfill unless that’s actually what you need. will probably revisit tap after.

Alex
@alex.bsky.team

in other words, it technically complements tap, but it may wind up replacing it for a significant number of tap users, and we’ll do some follow-on work at some point to see how to better position/improve tap for people who still have a need for it instead of jetstream

Alex
@alex.bsky.team

tap is also going to have a significantly smaller presence in the new Jetstream v2 docs we're about to launch as we expect that most people will not want to learn about this distinction out of the gate 😊

nel
@nel.pet

i honestly would be sad to see tap fully go. to me the verifiability of a full sync solution has always been one of the biggest selling points of atproto and its always saddened me a bit how little attention that part gets. it enables not only trustless operation (which i can understand why a lotnof

nel
@nel.pet

people may not fully care about even if i think they should) but equally importantly to me the ability of a syncer to detect when it missed even a single event and bring itself up to speed is a huge thing in terms of reliability. especially in a network like atproto where the reliability of sync

🇨🇦 Samir Al-Battran
@samir.fedica.com

Great! My TAP instance just finished syncing after 6 months of letting it run and you guys do this now ? lol Jokes aside, Well done! I think this is great progress, also a lot cheaper to maintain than TAP