Re: Tools for automated normalization ?

>To write such a program would be trivially easy,
>but would it be worth having?

Well, probably not...
It would not be too useful for those kind of direct dependencies as in your example, but I thought it might be useful for detecting transitive dependencies, for example if you would have AB->CDE , B->F , F->E
then an algorithm/program would drop the E from the first dependency and create the three tables
AB->CD , B->F , F->E

Usually, I don't really create a list of dependencies like this when I create a database structure, but just make an ER-model with the domain concepts and their multiplicity relations and if there is some attribute that can occurr many times then I draw a new table, or if there is one attribute that automatically can lead to another then also I draw a new table...

So, I do not really think it is difficult to create a normalized structure through ER-modeling, but just was curious about to some time try using a different approach and focus on the dependencies and then put them into a program that normalizes the structure by detecting transitive dependencies.

Though, maybe this kind of original dependencies (as in my example above) only will occurr in academically and artificially constructed examples...
I mean, if you in a real situation could figure out that B->F and F->E then you would maybe never say that AB->CDE in the first place, but directly rather figure out that AB->CD

/ Tom Received on Thu May 05 2005 - 12:20:44 CEST