Re: pg_upgade vs config

I wrote:
> It occurs to me that a back-patchable workaround for this would be to
> make get_loadable_libraries sort the library names in order by length
> (and I guess we might as well sort same-length names alphabetically).
> This would for example guarantee that hstore_plpython is probed after
> both hstore and plpython. Admittedly, this is a kluge of the first
> water. But I see no prospect of back-patching any real fix, and it
> would definitely be better if pg_upgrade didn't fail on these modules.

I've tested the attached and verified that it allows pg_upgrade'ing
of the hstore_plpython regression DB --- or, if I reverse the sort
order, that it reproducibly fails. I propose back-patching this
at least as far as 9.5, where the transform modules came in. It might
be a good idea to go all the way back, just so that the behavior is
predictable.

regards, tom lane