Re: Functional Dependencies > Uniqueness Constraints


Jon Heggland wrote:

To sum up: A good RDBMS should be able to infer FDs from keys /
uniqueness constraints (and inclusion dependencies), and reason about
them; in particular, to infer FDs (and thus keys) for arbitrary relation
expressions, e.g. in order to perform semantic optimisation.

On the other hand, in general the implication problem for INDs and FDs
is undecidable.

-- Jan Hidders

.