Re: Incorrect expected rows by ANALYZE - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Incorrect expected rows by ANALYZE
Date
Msg-id 252.1049729999@sss.pgh.pa.us
Whole thread Raw
In response to Re: Incorrect expected rows by ANALYZE  ("scott.marlowe" <scott.marlowe@ihs.com>)
List pgsql-hackers
On Mon, 7 Apr 2003 pgsql@mohawksoft.com wrote:
>> Couldn't there be an ANALYZE FULL option?

It already gets the row count right if you did VACUUM ANALYZE.  I don't
think we need ANALYZE FULL (at least, not for this purpose).

"scott.marlowe" <scott.marlowe@ihs.com> writes:
> There's something close, that is, you can change the statistics target for 
> a column in a table like so:
> ALTER TABLE [ ONLY ] table [ * ]
>     ALTER [ COLUMN ] column SET STATISTICS integer
> which can make it so that more of the rows are analyzed.

I'm not sure though how effective that will be at avoiding the problem
Kenji is complaining of.
        regards, tom lane



pgsql-hackers by date:

Previous
From: "scott.marlowe"
Date:
Subject: Re: more contrib: log rotator
Next
From: Lamar Owen
Date:
Subject: Re: more contrib: log rotator