News for hackers. Stuff that matters.
Re:Jetrelay: A high-performance ATproto relay in 500 LOC by phillipseamore on Friday May 16, 2025 @08:12AM (#44004472)
Great read and interesting implementation. I'm hoping for more like this from Alex in the future.
Re:Jetrelay: A high-performance ATproto relay in 500 LOC by remram on Friday May 16, 2025 @09:43AM (#44005453)
JetStream? They literally took the name of another CNCF message-queuing project? https://docs.nats.io/nats-concepts/jetstream
Re:Jetrelay: A high-performance ATproto relay in 500 LOC by wallyqs on Friday May 16, 2025 @02:47PM (#44008658)
Bluesky's project name is 'Jetstream' and NATS.io persistence engine is `JetStream` (camel case), which can be confusing but name not exactly the same.

- NATS JetStream docs: https://docs.nats.io/nats-concepts/jetstream

- Bluesky's 'Introducing Jetstream': https://docs.bsky.app/blog/jetstream

Re:Jetrelay: A high-performance ATproto relay in 500 LOC by tedunangst on Friday May 16, 2025 @05:59PM (#44010147)
Unclear what makes this an atproto relay and not a websocket relay.
Re:Jetrelay: A high-performance ATproto relay in 500 LOC by asayers on Friday May 16, 2025 @09:09PM (#44011251)
The only ATproto-specific part is how clients specify their starting position. They give a timestamp, which refers specifically to the `time_us` field in the messages, which I guess must be an ATproto thing. If jetrelay supported filtering, that would be ATproto-specific too. But you’re right: if you removed backfill support there’d be nothing ATproto- or even websocket-specific about it.