John Naylor <john(dot)naylor(at)enterprisedb(dot)com> writes:
> A customer reported a planning regression from 11.5 to 12.9. After
> bisecting, it seems the fix for the thread subject here, commit
> f230614da28, broke partition pruning in some cases.
[ facepalm... ] I didn't consider the possibility that
find_typmod_coercion_function would find something ...
so we're getting an actual call of bpchar(), not just
a RelabelType.
regards, tom lane