Tyler Ward <tjw19(at)columbia(dot)edu> writes:
> I'm using postgres 7.3.x (the default installed on Fedora Core 1) and
> I'm running into a serious
> bug in the insert performance.
Could we see EXPLAIN ANALYZE output from all three cases? I'd have
expected "INSERT ... SELECT foo" to generate the same plan as
"SELECT foo", but it sounds like you're getting something different.
regards, tom lane