Re: query plan ignoring check constraints

On Mon, 2005-06-20 at 14:40 -0400, Alvaro Herrera wrote:
> On Mon, Jun 20, 2005 at 10:35:24AM -0700, Josh Berkus wrote:
> > Rohit,
> >
> > > Now, when I do a select on the basetable with a range of ids, it looks up
> > > each subtable that inherits from the base table and using an indexed scan
> > > searches for values in the range. It does it even for subtables whose check
> > > constraint completely rules out the possibility of it containing any such
> > > row . Should not check constraint act as the first filter? The index should
> > > ideally be scanned only when the check constraint is passed by the search
> > > criteria but surprisingly it did not happen. The explain analyze showed
> > > cost for index scans of subtables that cannot contain rows matching the
> > > search criteria.
> >
> > This is called "range partitioning". We're working on it. You're welcome to
> > join the Bizgres project where most of the discussion on this feature takes
> > place:
>
> Why are you discussing development there? I can see in the archives
> that people are talking about changing page format, semantics of tuple
> info bits, and it's not getting to some "people that matters."

IRC, telephone, private mail and face-to-face have also been used to
discuss development...

Hackers has been used to discuss how to implement the ideas raised in
other forums.

Best Regards, Simon Riggs