Matt McClure wrote:
> > >
> > > If vacuum does not alter row oids, then I have another question. How does
> > > postgres re-use oids? I've seen the numbers grow and grow, but despite
> > ^^^^^^^^^^^^^^^^^^^^
> > It doesn't.
>
> Doesn't the fact that postgres never re-uses deleted (and therefore no
> longer in use anywhere) oids create a problem when you reach the upper
> bound? Or is the upper bound on oids so ridiculously high that it
> shouldn't be a concern? Or does postgres have a scheme for increasing
> oids without bound entirely?
We have plans to make using global oid in user tables optional...
>
> In any case, using row oids from one table as values in another table
> won't ever be an issue, right?
Right. But you could also use sequences...
Vadim