Context Navigation

Changes between Initial Version and Version 1 of Ticket #37156, comment 2

initial	v1
3	3	Sigh. This is essentially a myth, at least these days. Unless you're repeatedly crossing the boundary in a hot-loop — which, don't! — overhead is essentially negligible.
4	4
5		Redis-py's async client is no faster than the sync client (indeed it seems slightly slower, but I need to verify that before stating definitively — it's negligible too though). Throughput is limited by pool size. And do enable pipelining!
	5	Redis-py's async client is no faster than the sync client (indeed it seems slightly slower, but I need to verify that before stating definitively — it's negligible too though). Similar to the way psycopg's async cursors serialise queries, throughput is limited by pool size. And do enable pipelining!
6	6
7	7	I shall put writing this up on the backlog, but it's not at all clear that the argument to remove (particularly) `sync_to_async` calls throughout is/remains sound.