The client in the RPC tutorial for Bunny isn't super performant (maybe not a hard requirement for RPC) because it creates new consumers over and over again, but it also has a race condition bug, where messages with "wrong" correlation_id would be discarded.
On 13 May 2014 at 17:55:02, Michael Klishin ([hidden email]) wrote:
> > Can you submit a PR to rabbitmq-tutorials? I will propagate
> version to rabbitmq.com(http://rabbitmq.com) after that.
Also note that there were 2 or so RPC tutorial iterations before
we've settled on the current version. The one we have today has higher
throughput than the original one.
We definitely want to have a correct/safe version in the tutorials
but can also provide one or even multiple alternative implementations in rabbitmq-tutorials. Have you
tried running any benchmarks to see how they compare?