Re: OUTER JOIN performance regression remains in 8.3beta4

Alvaro Herrera <alvherre@commandprompt.com> writes:
> Would it be a good idea to keep removing redundant clauses and rethink
> the preference for clauseful joins, going forward?

No --- it would create an exponential growth in planning time for large
join problems, while not actually buying anything in the typical case.

It's possible that we could do something along the lines of inserting
"dummy" join conditions, to allow particular join paths to be explored,
without generating any clause that actually requires work at runtime.
I'm not convinced this complication is needed though; at least not if
the only thing it's good for is this rather specialized optimization
rule.

            regards, tom lane