Michael Loftis <mloftis@wgops.com> writes:
> Realistically the system should choos *ANY* index over a sequential
> table scan.
Sorry, I do not accept that. You might as well say that we should
rip out any attempt at cost estimation, and instead put in two or
three lines of brain-dead heuristics. If it were that simple we'd
all be using MySQL ;-)
> Above a fairly low number of records any indexed query
> should be much faster than a seqscan.
Isn't that exactly backwards?
regards, tom lane