On Fri, Dec 18, 2015 at 12:52 PM, Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> Mind you, I don't think "inference specification" is very good
> terminology, but what's there right now is just wrong.
It doesn't appear in the documentation. The term "inference
specification" only appears where it's necessary to precisely describe
the input to unique index inference.
--
Peter Geoghegan