"Jonathan S. Katz" <jk...@postgresql.org> writes: > I would say that’s the exact reason why it should be documented, > so it’s clear that it was by design.
Seems reasonable. Maybe something in 8.7.4 enum implementation details?
regards, tom lane
