Re: Better estimates of index correlation

On Mon, Mar 14, 2011 at 10:25 AM, Alvaro Herrera
<alvherre@commandprompt.com> wrote:
> Excerpts from Robert Haas's message of lun mar 14 11:18:24 -0300 2011:
>> On Mon, Mar 14, 2011 at 10:09 AM, Alvaro Herrera
>> <alvherre@commandprompt.com> wrote:
>
>> > It sure would be nice to be able to do it only during the last scan.
>>
>> Does it really matter?  What Tom was describing sounded embarassingly cheap.
>
> Well, you only do multiple passes for tables that are really large, so
> it's precisely there that you want to save the extra overhead of having
> to do it multiple times.

Right but if the overhead is 0.0001% then who cares?  I have no idea
what the real number is but it doesn't seem unrealistic to me that it
could be on that order.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company