On Oct 5, 2008, at 1:11 AM, Peter Eisentraut wrote:
> I don't think we should overload syntax choices with optimization
> hints. We don't really know why or how people will be using this
> syntax, and labeling it from the start as "will have unusual
> performance behavior" isn't a good sell.
>
> As a precedent, consider the JOIN syntax, which is obviously
> redundant and in its first implementation contained an implicit
> optimization hint with regard to join order that later had to be
> done away with because it confused users (I think). The CTE case
> is quite similar, and maybe the GUC answer of old could apply here
> as well. But I think by default we should abide by SQL's
> declarative approach of "Tell me what you want and I'll execute it
> any way I like."
Agreed. It's already horrible that we suggest people use OFFSET 0,
only because we don't want to define formal optimizer hints (and
that's *exactly* what OFFSET 0 is).
--
Decibel!, aka Jim C. Nasby, Database Architect decibel@decibel.org
Give your computer some brain candy! www.distributed.net Team #1828