On 2011-06-30 09:39, Yeb Havinga wrote:
>
> 9) as remarked up a different thread, more than one clause could be
> pushed down a subquery. The current patch only considers alternative
> paths that have at most one clause pushed down. Is this because of the
> call site of best_inner_subqueryscan, i.e. under one join rel? Would
> it be an idea to consider, for each subquery, only a single
> alternative plan that tries to have all suitable clauses pushed down?
Ehm, please forget this remark, I've mistaken join rel for base rel.
-- Yeb