Robert Haas <robertmhaas@gmail.com> writes:
> On Sun, Dec 12, 2010 at 9:16 PM, Tomas Vondra <tv@fuzzy.cz> wrote:
>> The proposed solution is based on contingency tables, built for selected
>> groups of columns (not for each possible group). And the contingency
>> table gives you the ability to estimate the probabilities needed to
>> compute the selectivity. Or am I missing something?
> Well, I'm not real familiar with contingency tables, but it seems like
> you could end up needing to store a huge amount of data to get any
> benefit out of it, in some cases.
The reason that this wasn't done years ago is precisely that nobody's
figured out how to do it with a tolerable amount of stats data and a
tolerable amount of processing time (both at ANALYZE time and during
query planning). It's not hard to see what we'd ideally like to do;
it's getting from there to something useful in production that's hard.
regards, tom lane