Re: benchmarking the query planner

Simon Riggs <simon@2ndQuadrant.com> writes:
> On Thu, 2008-12-11 at 17:45 -0500, Tom Lane wrote:
>> Simon Riggs <simon@2ndQuadrant.com> writes:
>>> I would like it even more if there was a data type specific default.
>>> Currently we have a special case for boolean, but that's it.
>> 
>> No, we don't (or if we do I'd be interested to know where). 

> Your commit, selfuncs.c, 7 Jul.

As with Robert's pointer, that's about coping with missing stats,
not about determining what stats to collect.

>  ... neither of those were ones I was thinking about. I see 3 main classes:
> * data with small number of distinct values (e.g. boolean, smallint)
> * data with many distinct values
> * data with where every value is typically unique (e.g. text)

These three categories are already dealt with in an entirely
type-independent fashion by the heuristics in compute_scalar_stats.
I think it's quite appropriate to drive them off the number of observed
values, not guesses about what a particular datatype is used for.
        regards, tom lane