Alex Pilosov <alex@pilosoft.com> writes:
> Example: Index scan when data in index is stored in approximately same
> order in physical table will be fast. Index scan when index order does not
> match table order will be slow. I don't see a way you can explain _that_
> to optimizer, nor I really want to.
Precisely why you should leave it to the optimizer to figure
these things out for itself. Curiously enough, there is a
physical-order-correlation statistic in pg_statistic now ;-)
regards, tom lane