Sam Willis

@alexgleason this is all so awsome! The last to-do if so me to add some tests to CI, let's leave it open till early next week, I'll need the reminder!

Did you happen to get the error logs from when it crashed? If you could post them (use `debug: 1`) that would help a lot in working out why it...

Hey @mindplay-dk > is this library abandoned or are you still interested? 🙂 Not abandoned, would describe as a slow burn. It's an experiment to look at these ideas, with...

Yep, this is where having transaction boundaries in the log would be useful. Essentially everything within the compacted segment of the log becomes a single transaction.

Also worth noting that the current theory is that most of applications don't really care for transactional consistency. That's why we don't yet have transaction boundaries in the log. They...

Until we have transactional boundary messages, can we have all messages that are part of a compaction have the same LSN offset? I.e. all messages prior to the end of...

Another alternative (similar to @magnetised, but with a permanent tombstone log): - Keep a separate tombstone log that only logs deletes. This can quite small and very efficient to keep,...

@kevin-dp Yes, however on initial fetch (new shape for the client) you do not need to send the tombstone log as they have no history to clean up. The tombstone...

@KyleAMathews I think we need to clarify what guarantees we want to provided. In your initial message: > We'll want the same guarantees [as Kafka] — https://docs.confluent.io/kafka/design/log_compaction.html#compaction-guarantees However Kafka compaction...

Hey @naliferov I've also noticed this, as far as I can see it only happens on a restart, jumping ahead in the sequence. I have a suspicion it is due...