William Yu wrote:
> There are cases where seqscan will be faster than indexscans. For
> example, your query to retrieve the latest 25 threads -- always faster
> using seqscan. If it was using indexscan, that would explain the 9
> seconds to run because the HD heads would have to jump back & forth from
Oops, I didn't write down my thoughts clearly. I meant to say --
DEPENDING ON THE DISTRIBUTION OF THE VALUES -- the query might be faster
using seqscan.