"John.Cowan" <xxxxxx@reutershealth.com> writes:

> I assume you mean "must not affect the behavior, but may affect the
> performance" (= speed).  Unfortunately, you have to be intimately
> familiar with your CL implementation to know when declarations are
> useless, when they are critical to performance, and when they actually
> make things worse.

Yes, but I don't see any way around that problem.

It should be possible to guarantee that declarations never make
anything worse...