Amit Kapila <amit.kapila16@gmail.com> writes:
> Yeah, this message looks better than the current one. However, when I
> tried to do what HINT says, it doesn't initiate replication. It gives
> me the below error:
> postgres=# Alter subscription sub1 refresh publication;
> ERROR: ALTER SUBSCRIPTION ... REFRESH is not allowed for disabled subscriptions
Geez ... is there *anything* that's not broken about this message?
I'm beginning to question the entire premise here. That is,
rather than tweaking this message until it's within hailing
distance of sanity, why do we allow the no-connect case at all?
It's completely obvious that nobody uses this option, or we'd
already have heard complaints about the advice being a lie.
What's the real-world use case?
regards, tom lane