In the process of decomposition to normalize a relation. If I reach the point where all attributes in a relation depend on the primary key, can I assume that they will all depend entirely on the different candidate keys? If that is not a case can you please give me an example of a case where all attributes depend on the primary key, but some of them depend on the part of other candidate keys.
I'm starting learning databases
Surrogate primary IDs make an example really easy:
(row_id PK, student_id, course_id, student_name)
where row_id
and (student_id, course_id)
are candidate keys and student_id -> student_name
. Of course the row_id
trivially determines any other attributes if it's an auto-incremented number.