| From: | Gregory Stark <stark(at)enterprisedb(dot)com> |
|---|---|
| To: | "Brian Hurt" <bhurt(at)janestcapital(dot)com> |
| Cc: | <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Sorting Improvements for 8.4 |
| Date: | 2007-12-22 13:15:26 |
| Message-ID: | 87lk7mrhu9.fsf@oxford.xeocode.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
"Brian Hurt" <bhurt(at)janestcapital(dot)com> writes:
> 3) It's possible to perform the sort lazily. You have the initial O(N) pass
> over the list, but then each block is only O(log N) cost. If it's likely that
> only the first part of the result is needed, then much of the work can be
> avoided.
Now that's a *fascinating* idea. I'm having trouble coming up with a really
killer use case for it since the bounded heap sort takes care of many cases
where it would seem to apply. But it seems rally promising.
--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com
Ask me about EnterpriseDB's RemoteDBA services!
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Martijn van Oosterhout | 2007-12-22 14:13:20 | Re: binary decode |
| Previous Message | Bruce Momjian | 2007-12-22 12:21:51 | Re: Negative LIMIT and OFFSET? |