On Wednesday 25 February 2004 22:37, Tom Lane wrote:
> I wrote:
> > try writing
> > WHERE 'ABCDE' >= pr_min AND 'ABCDE' <= pr_max
> > AND pr_min < (SELECT pr_min FROM table
> > WHERE pr_min > 'ABCDE'
> > ORDER BY pr_min LIMIT 1)
> > The idea here is to add an upper bound on pr_min to the index scan
> > conditions, so that the scan can stop short of the end of the index.
>
> Argh, got that backwards. What you are missing is a *lower* bound on
> pr_min, and the index scan will therefore run from the start of the
> index up to pr_min = 'ABCDE'. So reverse the sense of the added test:
>
> AND pr_min >= (SELECT pr_min FROM table
> WHERE pr_min <= 'ABCDE'
> ORDER BY pr_min DESC LIMIT 1)
Aha! I was trying something like that, but couldn't get it quite right and it
was getting too late for me to see clearly.
Thanks Tom, I'll have a play with this later today.
-- Richard Huxton Archonet Ltd