>Bruno,
>
>A cascaded update describes one of the ways for handling a change to the PK of a table. The other two are restricted and ignored.
>
>In Realtional theory it is necessary to describe what must happen when the vlaue of a PK changes as it can happen depending on the design. The use of meaningless PKs (surrogate PK) simply eliminates the concern for this as the PK never changes.

That is how I understood it...
Just wanted to confirm my thoughts on them not beeing required when using surrogate PKs.

Is a unique serial number (which can never be re-used) entered by a user, considered a surrogate key or are surrogate keys, by definition, strictly generated by increments of a field in a table?

You said "...PK changes as it can happen depending on the design."...that is a requirement in my case. However, is it considered good design? and if not, are there cases where it wood be considered as a good design?

Thanks,