David Fetter <david@fetter.org> writes:
> On Fri, Aug 06, 2010 at 03:00:35PM -0400, Tom Lane wrote:
>> Perhaps run through pg_class after restart and flush anything marked
>> relistemp? Although the ideal solution, probably, would be for temp
>> tables to not have persistent catalog entries in the first place.
> For the upcoming global temp tables, which are visible in other
> sessions, how would this work?
Well, that's a totally different matter. Those would certainly have
persistent entries, at least for the "master" version. I don't think
anyone's really figured out what the best implementation looks like;
but maybe there would be per-backend "child" versions that could act
just like the current kind of temp table (and again would ideally not
have any persistent catalog entries).
regards, tom lane