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

"Tom Lane" <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:

> Gregory Stark <stark(at)enterprisedb(dot)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.

That's why I said:

a subsequent VACUUM ANALYZE could mark the resulting statistics as
"authoritative"

Not just plain analyze.

There's another issue here too. One of the other motivations is to be able to
put read-only tables on read-only media. To do that would require freezing
every tuple which would at the very least involve looking at every tuple. (It
would also involve waiting until all tuples are freezable too.)

So there's a natural step in which to gather these authoritative statistics
anyways.

--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com