Re: TODO idea - implicit constraints across child tables with a common column as primary key (but obviously not a shared index)

Gregory Stark <stark@enterprisedb.com> writes:
> The main data from the statistics that's of interest here are the extreme
> values of the histogram. If we're not interested in any values in that range
> then we can exclude the partition entirely.

Except that there is *no* guarantee that the histogram includes the
extreme values --- to promise that would require ANALYZE to scan every
table row.
        regards, tom lane