Lucas Kent

This sounds like a problem! I need to verify if this does actually affect all map types in the protocol. But assuming for a second that it does, I wonder...

Thinking about the ways in which the protocol is actually used I think that replacing IndexMap with Vec might actually be the best for performance. Since the kafka protocol very...

thats a good idea! In case you or someone else is interested in implementing this the steps to implement this would look like: 1. Add a field to `HighLevelHitbox` called...

https://github.com/shotover/shotover-proxy/issues/1537 is a prereq

IIRC there is a bit of a mismatch between the design of tokio and io-uring beyond just the implementation being immature. I also dont know of any serious projects that...

I've completed the planned the swap to sending Vec instead of Message for KafkaSinkCluster and found no noticeable improvement to benchmarks or flamegraph allocation. So I think that my guess...

This work is effectively done, any remaining related work is tracked on https://github.com/shotover/shotover-proxy/issues/1507

I'm not really sure what this PR is doing, but it might be easier to read if you resolve the merge conflicts? (merge conflicts also seem to be preventing me...

fwiw I think that in general using iterators to iterate over the contents of the records is a good way to avoid allocations, but we might need to rewrite most...

~~looks like its impossible to get the clustered tests running on mac with docker desktop: https://docs.docker.com/desktop/networking/#per-container-ip-addressing-is-not-possible This leaves running shotover tests in a VM or on a cloud instance.~~