Talk:Second normal form/Archive 1

It bothers me that we don't use a single example between the articles 1NF to 4NF, thus allowing people to really see the progressive differences. But that's not too bad.

What really bothers me is that the example in this article goes from 1NF to 3NF and the 3NF article starts at 1NF and goes to 3NF. This really makes the 2NF-3NF distinction hard to grasp. All we need to do to fix this article's example is add a column, right? -- General Wesc 02:29, 3 September 2006 (UTC)

What does "irreducibly" mean in that sentence? It doesn't make sense to me.--80.227.100.62 07:35, 31 October 2006 (UTC)

I agree that the examples don't make clear what the difference is between 2NF and 3NF (and since I came here to try and find out for myself, I'm not really qualified to try and fix the examples).

Hard to understand?

I must say that I read this article multiple times and even after that I had to resort to finding notes from my college Database lectures. These are the points I would like to make:

1) At the beginning of the article there is a definition of 2NF which reads "irreducibly dependent" - what does it mean? Perhaps it would be better to explaint this somewhere. (Irreducibly dependent on the key means that the attribute may not depend only on some strict subset of the key.)

2) As already pointed out, the difference between 2NF and 3NF should be explained either here or in the Third normal form article. (The difference is that non-key attributes dependent solely on other non-key attributes are allowed in 2NF but not in 3NF -- 2NF eliminates only dependencies on strict subsets of the key.)

3) What happens if there is more than one candidate key? The article talks about primary key but as far as I know, the definition of 2NF works with all the keys -- this means that every occurence of "primary key" should be replaced by "each candidate key" (or something more appropriate). Maybe we could put in another example which has more than one candidate key? Also, this kind of example could show how 2NF (or even 3NF, BCNF ...) fails to resolve some particular kinds of redundancy.

Michal Burger 13:28, 5 January 2007 (UTC)

---

I have tried to address some of your concerns in the latest revised version of the article. --Nabav 11:16, 16 February 2007 (UTC)

A better definition

I think this definition is better because the surrent is unclear if, in 2NF the data should depend on the ENTIRE key or NONE of it.

test each table for partial dependencies on a concatenated key. This means that for a table that has a concatenated primary key, each column in the table that is not part of the primary key must depend upon the entire concatenated key for its existence. If any column only depends upon one part of the concatenated key(and not the entire key), then we say that the entire table has failed Second Normal Form and we must create another table to rectify the failure. —The preceding unsigned comment was added by Shreyasjoshis (talk • contribs) 18:03, 8 May 2007 (UTC).

Being dependent on none of it isn't a possibility. That possibility is excluded by the definition of "candidate key".

Also note that we cannot bring primary keys into the definition. 2NF is defined with reference to candidate keys.

However, I take your basic point that a less technical but still reliable definition would be desirable; so I have gone ahead and added one. --Nabav 22:22, 9 May 2007 (UTC)