On Sun, Nov 18, 2007 at 09:59:41AM -0800, David Fetter wrote:
> Ordering is never guaranteed without an ORDER BY, except in the time
> between a CLUSTER and the first write operation after it.
It's my understanding that with the new "seqscan piggy-backing" in 8.3
even this will go. I'm not sure if this would affect the behaviour of
GIN indexes, but I'd doubt it.
A quick question for the OP; if the results from these queries are going
to be displayed for human consumption, does it matter much if they're in
perfect sorted order? For computer algorithms it generally matters much
more, but people are generally a bit more flexible.
Sam