> I've wondered at times if PG isn't deliberately leaving this problem Since you know about the implementation(I haven't looked closely), won't it need a compl

> I've wondered at times if PG isn't deliberately leaving this problem Since you know about the implementation(I haven't looked closely), won't it need a complete overhaul to take closures out of equation? Or implement some sort of closure serialization(on-disk maybe)? Is there a third option where pg just bangs at the keyboard for a couple of hours, and done?