On Thu, 2002-05-02 at 16:52, Tom Lane wrote:
> Hannu Krosing <hannu@tm.ee> writes:
> > Is "PROC array slot number" something internal to postgres ?
>
> Yes.
>
> If we used PID then we'd eventually have 64K (or whatever the range of
> PIDs is on your platform) different pg_temp_nnn entries cluttering
> pg_namespace.
Should they not be cleaned up at backend exit even when they are in
range 1..MaxBackends ?
> But we only need MaxBackends different entries at any one
> time. So the correct nnn value is 1..MaxBackends. BackendId meets the
> need perfectly.
----------
Hannu