Oracle Database Discussions

Hi all. I searched first, but there's too much noise in the results.

My questions follow; assume they apply to 10gR2 in case it makes a difference.

1. In practice, do you or do you not create "not null" constraints on columns that are part of the primary key?

2. It isn't needed, so if you do, why? (I suspect "habit".)

I can anticipate a reply that using both lets you disable the primary key constraint and still prevent null columns, although I fail to see how that is useful.

3. Finally, if you create both, is there actual execution overhead?

Thanks.