Re: strange row count estimates with conditions on multiple column

> Tomas Vondra <tv@fuzzy.cz> writes:
>> I've been thinking about this and I think it might be improved. If I
>> understand the logic corretly, it says 'use half of the histogram bin
>> size'. But the value
>
>> #define DEFAULT_RANGE_INEQ_SEL 0.005
>
>> says it's always 0.5%, which is not not true if STATISTICS TARGET is not
>> 100. This could actually yield 10x more precise estimates when the
>> STATISTICS TARGET is set to 1000.
>
> Huh?  The default estimates are completely unrelated to the size of the
> histogram, and certainly unrelated to the default size of the
> histogram.  We use those estimates when we don't have relevant stats.
> It's pure wishful thinking to suppose that changing the statistics
> target would have any impact on what the estimate ought to be in such
> a case.

Ooops, sorry for the crazy gibberish I've posted earlier. I thought those
default estimates work a somehow different and haven't checked that in the
code. The proposed 'optimization' obviously does not make any sense.

regards
Tomas