Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
> thinking that fudging the tuple cost GUCs is going to work seems
> unrealistically optimistic. If you need the optimizer to know
> about this, you need the optimizer to REALLY know about this...
I rule nothing out. If you have a more refined idea, I welcome you
to include in the wiki's "R&D Issues" section. Or describe it here
and I'll add it. Frankly, it seemed no more of a hack than some
aspects of our costing calculations, but it obviously pays to model
it as well as we can. But I will take something which shows any
improvement without getting too nasty, until we have something
better. I see the optimization phase as lasting a while and trying
out many ideas, some of which won't turn out to have been good ones.
We don't use those.
-Kevin