Roberto Cornacchia <rcorna@tin.it> writes:
>>>> 1/disbursion is a lower bound on the number of values, but it wouldn't
>>>> be a good estimate unless you had reason to think that the values were
>>>> pretty evenly distributed.
> Thank you, Tom and Bruce.
> This is not a good news for us :-(. In any case, is 1/disbursion the
> best estimate we can have by now, even if not optimal?
I don't have a better idea right at the moment. I'm open to the idea
that VACUUM should compute more or different statistics, though ---
as long as it doesn't slow things down too much. (How much is too much
would probably depend on how much win the new stats would provide for
normal query-planning. For example, I'd resist making two passes over
the table during VACUUM ANALYZE, but I wouldn't rule it out completely;
you could sell me on it if the advantages were great enough.)
Hey, you guys are the researchers ... give us a better approach to
keeping table statistics ;-)
regards, tom lane