Andrew Dunstan <and...@dunslane.net> writes: > Regardless of this problem, I think we should simply call > set_replication_conf unconditionally in init(). Replication connections > are now allowed by default on Unix, this would just bring Windows nodes > into line with that.
Yeah, I was thinking along the same lines yesterday. The fact that pre-commit testing failed to note the problem is exactly because we have this random difference between what works by default on Unix and what works by default on Windows. Let's close that gap before it bites us again. There's still the issue of these tests overriding postgresql.conf entries made by init(), but maybe we can live with that? regards, tom lane