> See https://eklitzke.org/goroutines-nonblocking-io-and-memory-usage for an interesting discussion of the drawbacks of some buffer handling idioms.

Thanks for the link!

It does make sense to use a pool of buffers for the proposed BufferedProtocol when you need to keep thousands of long-open connections.  The current design makes that easy: when BufferedProtocol.get_buffer() is called you either take a buffer from the pool or allocate a temporary new one.

For use-cases like DB connections (asyncpg) a permanently allocated buffer per protocol instance is a good solution too, as usually there's a fairly limited number of open DB connections.