Jeff Davis <pgsql(at)j-davis(dot)com> writes:
> We can already keep stats for a functional index. Is there a reason we
> can't keep stats for a multi-column index?
The questions that need to be answered are (1) what stats are you gonna
collect, and (2) exactly what are you going to do with them when you
have 'em?
All the previous discussions have stalled on the question of how to
avoid trying to collect stats about an exponentially large number of
column combinations; we've never even reached the question of what
stats we'd actually want given that a particular combination has been
determined to be interesting. Perhaps that's a trivial question,
but it's been a mighty long time since I took statistics ...
regards, tom lane