On Sat, May 27, 2006 at 11:52:40AM -0400, Tom Lane wrote:
> Martijn van Oosterhout <kleptog@svana.org> writes:
> > How about the suggestion of using a sequential index scan like the
> > recent changes to VACUUM in the case that there are no regular index
> > quals?
>
> Nonstarter (hint: the solution we found for VACUUM assumes there can
> be only one).
Bummer, I was envisioning allowing index AMs to have another access
method, the unordered sequential scan. Just like we consider a random
access of a real table to be more expensive than a seq scan, and index
scan that seeks a lot would be more expensive that a sequential scan.
So if you know you're going to scan most of an index, scanning
sequentially would be cheaper...
Ah well,
Have a nice day,
--
Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/
> From each according to his ability. To each according to his ability to litigate.