Alex Pilosov <alex(at)pilosoft(dot)com> writes:
> Are there any plans to implement some way to force optimizer to decide on
> index scan vs table scan _for a specific table_, and to tell optimizer to
> use a particular index when operating on a table?
No. The game plan is to make the optimizer smarter, not to spend time
on nonstandard cruft like that. Besides, "optimization hints" have a
way of becoming pessimization hints over time, as your data shifts and
the hints don't.
I'm about ready to call for alpha testing on the improved pg_statistic
code, BTW. If you've got time you might see what the current CVS tip
thinks about your query.
regards, tom lane