Nathan Bossart <nathandbossart@gmail.com> writes:
> This sort of thing comes up enough that perhaps we should add a
> better-supported way to deal with GUCs that depend on each other...
Yeah, a GUC check hook that tries to inspect the value of some
other GUC is generally going to create more problems than it
solves; we've learned that the hard way in the past. We have
your patch to remove one instance of that on the CF queue:
https://www.postgresql.org/message-id/flat/ZnMr2k-Nk5vj7T7H@nathan
But that fix only works because those GUCs are PGC_POSTMASTER
and so we can perform a consistency check on them after GUC setup.
I'm not sure what a more general consistency check mechanism would
look like, but it would have to act at some other point than the
check_hooks do.
regards, tom lane