On Wed, Apr 25, 2012 at 1:31 PM, Simon Riggs <simon@2ndquadrant.com> wrote:
> However, it is a fallacy that this is a good solution for using temp
> tables on HS. I think the wish to enhance Oracle compatibility is
> making some wishful thinking happen with regard to how useful this is
> going to be. We need to spend just as much time considering the
> utility of our work as we do spending time on the quality of the
> implementation, otherwise its just well-implemented shelfware.
Well, like I say, if you want to use locally-defined temp tables on
HS, you have to somehow solve the problem of catalog entries, and
nothing in your email looks like a proposal for how to do that. I've
come up with one design, which I sketched in my original response, but
it relies on creating some new system catalogs that are themselves
GTTs, and it's also hideously complicated. If you or anyone else can
come up with a better design, great, but so far no one has.
It's no skin off my neck if this project gets done in way that
bypasses the need for GTTs; I just don't have a credible proposal for
how to do that, and Noah stated that he doesn't either.
I do agree that what Noah's proposing to implement is shooting at a
pretty narrow target, but I don't think it's so narrow that we
wouldn't commit it if he's willing to do the work to implement it.
All of the infrastructure that he's proposing to create seems to me to
have plausible other uses, so even if the immediate feature doesn't
bring a lot of benefit there's every reason to suppose that it will
pave the way for further improvements down the line.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
