Robert Haas <robertmhaas@gmail.com> writes:
> We don't actually have a patch for GTT at this point; Noah is at least
> the second person to threaten to write one, but nobody's actually done
> it yet to my knowledge.
IMO, the main reason that's been let slide for nine years is that there
wasn't a particularly strong use-case for temp tables implemented the
spec's way. Worse: according to the 2003 thread, there were in fact no
major RDBMS players that hewed closely to the spec's semantics (though
possibly that's changed by now); which made the "it's standard" argument
far too weak to justify doing anything either. Now that there's a
realistic use-case in hot standby scenarios, I think we can expect that
something will get done within the foreseeable future. At least for the
GLOBAL case --- I concur that there's nothing on the horizon suggesting
we'll have spec-style LOCAL temp tables.
> Maybe the right thing to do here is nothing. I think to some degree
> we are arguing about what color to paint an imaginary bikeshed. If at
> some point we support GTTs using the syntax CREATE GLOBAL TEMPORARY
> TABLE, then there is going to be a compatibility break.
If we can foresee that this will happen, warning about it in advance
seems like a good idea. See for comparison our handling of the "=>"
operator business.
> What we are
> arguing about is whether to pull that compatibility break forward into
> 9.2, or wait and let it break in the release where it has to break;
Uh, no, Simon's original patch pulled the compatibility break forward,
which was what I objected to. But a WARNING won't break applications,
and it does provide some notice, even though I admit that not everybody
will be helped.
regards, tom lane