| From: | Greg Stark <gsstark(at)mit(dot)edu> |
|---|---|
| To: | Neil Conway <neilc(at)samurai(dot)com> |
| Cc: | Greg Stark <gsstark(at)mit(dot)edu>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, "Jim C(dot) Nasby" <decibel(at)decibel(dot)org>, Christopher Petrilli <petrilli(at)gmail(dot)com>, Ying Lu <ying_lu(at)cs(dot)concordia(dot)ca>, pgsql-general(at)postgresql(dot)org, pgsql-performance(at)postgresql(dot)org |
| Subject: | Re: [PERFORM] "Hash index" vs. "b-tree index" (PostgreSQL |
| Date: | 2005-05-11 13:59:05 |
| Message-ID: | 87d5rxrhhy.fsf@stark.xeocode.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general pgsql-performance |
Neil Conway <neilc(at)samurai(dot)com> writes:
> I'm posting mainly because I wasn't sure what to do to avoid false positives in
> the case of hash collisions. In the hash AM code it is somewhat awkward to
> fetch the pointed-to heap tuple and recheck the scankey.[1] I just did the
> first thing that came to mind -- I marked all the hash AM opclasses as "lossy",
> so the index qual is rechecked. This works, but suggestions for a better way to
> do things would be welcome.
I would have thought that would be the only way worth considering.
Consider for example a query involving two or more hash indexes and the new
bitmap indexscan plan. You don't want to fetch the tuples if you can eliminate
them using one of the other indexes.
--
greg
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Csaba Nagy | 2005-05-11 14:01:17 | Re: Fixing a too long column value in a before insert |
| Previous Message | Tom Lane | 2005-05-11 13:57:07 | Re: [PERFORM] "Hash index" vs. "b-tree index" (PostgreSQL |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | John A Meinel | 2005-05-11 14:04:47 | Re: Optimizer wrongly picks Nested Loop Left Join |
| Previous Message | Tom Lane | 2005-05-11 13:57:07 | Re: [PERFORM] "Hash index" vs. "b-tree index" (PostgreSQL |