Re: generalizing the planner knobs

Rod Taylor <pg@rbt.ca> writes:

> > In the extreme, no amount of added intelligence in the optimizer is going to
> > help it come up with any sane selectivity estimate for something like 
> > 
> >   WHERE radius_authenticate(user) = 'OK'
> 
> Why not?
> 
> The missing capability in this case is to be able to provide or generate
> (self learning?) statistics for a function that describe a typical result
> and the cost of getting that result.

Ok, try "WHERE radius_authenticate(user, (select ...), ?)"

The point is that you can improve the estimates the planner gets. But you can
never make them omniscient. There will always be cases where the user knows
his data more than the planner. And those hints are still valid when a new
optimizer has new plans available.

This is different from hints that tell the planner what plan to use. Every
situation where the predicted cost is inaccurate despite accurate estimates
represents a fixable bug in the optimizer's cost model. When a new version of
the optimizer is available with a more accurate cost model or new available
plans those kinds of hints will only get in the way.

-- 
greg