Re: Review remove {join, from}_collapse_limit, add enable_join_ordering

On Thursday 16 July 2009 17:27:39 Greg Stark wrote:
> On Thu, Jul 16, 2009 at 4:16 PM, Tom Lane<tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > However, I do observe that this seems a sufficient counterexample
> > against the theory that we can just remove the collapse limits and let
> > GEQO save us on very complex queries. On my machine, the example query
> > takes about 22 seconds to plan using CVS HEAD w/ all default settings.
> > If I set both collapse_limit variables to very high values (I used 999),
> > it takes ... um ... not sure; I gave up waiting after half an hour.
> What's the point of GEQO if it doesn't guarantee to produce the
> optimal plana and *also* doesn't guarantee to produce some plan, any
> plan, within some reasonable amount of time? Either we need to fix
> that or else I don't see what it's buying us over our regular planner
> which also might not produce a plan within a reasonable amount of time
> but at least if it does it'll be the right plan.
Well, I could not find a plan where it errored out with the old limits. So one
could argue its just not adapted.
Although I also could not find a single case where geqo was relevantly faster
with the default settings even if it was used.
The default settings currently make it relatively hard to trigger geqo at all.

Andres