From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Alexander Korotkov <aekorotkov(at)gmail(dot)com> |
Cc: | Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Statistics and selectivity estimation for ranges |
Date: | 2012-08-15 16:33:18 |
Message-ID: | 7576.1345048398@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Alexander Korotkov <aekorotkov(at)gmail(dot)com> writes:
> Ok, we've to decide if we need "standard" histogram. In some cases it can
> be used for more accurate estimation of < and > operators.
> But I think it is not so important. So, we can replace "standard" histogram
> with histograms of lower and upper bounds?
You should assign a new pg_statistic "kind" value (see pg_statistic.h)
rather than mislabel this as being a standard histogram. However,
there's nothing wrong with a data-type-specific stats collection
function choosing to gather only this type of histogram and not the
standard one.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2012-08-15 16:45:54 | Re: text search: restricting the number of parsed words in headline generation |
Previous Message | Bruce Momjian | 2012-08-15 16:22:50 | Re: sha1, sha2 functions into core? |