[Friday, March 17, 12:45]

> I have a question regarding the homework.  Is it true that for this relation
> no two authors can have the same name?  For example, Ewe Miller and Ewe
> Miller are father and son with the same name, and have both published books.
> If this were true my solution would be different from what I have:
> FD1 ISBN --> Publisher
> FD2 ISBN --> Title
> FD3 Author --> No
> FD4 ISBN, Author --> Title, Publisher

We assume for this exercise that they have not published a book together.
I have excluded the case that an author with the same name appears twice
in the list of authors of the same book.
This was probably too restrictive, but I didn't think about such a case.

> In other words if this were true FD3 and FD4 would be false and I would
> decompose this table.

Your FD3 is wrong even if we don't allow this.
FD3 means that if an author is e.g. first author of one book, he/she must also
be first author of all other books to which he/she contributed.
This is not true. For instance, the UML books by Booch, Jacobson, and
Rumbaugh have a different author sequence for each of the three books
(so that each one becomes first author).

I cannot comment on the other FDs, but please try to avoid implied FDs.