Re: Attribute-values in separate table

> There are two schools of thought as to whether the addition of a new
> attribute should or should not require additional data definition.
>
> Adding a row to a table is data manipulation. altering a table to add a new
> column is data definition. Creating a new table is likewise data
> definition.
>
> I'm of the school of thought that data in a database is managed better when
> the right to create and alter data definitions is severely restricted,
> while the right to manipulate data is extended to the applications that
> interact with the database.
>
> The other school of thought is that the superior flexibility of defining new
> attributes without altering any database definitions overrules the data
> management advantages in prohibiting the same. I can't defend this school
> of thought (apparently yours) because, in my view, it leads inexorably to
> undocumented and therefore unusable data.
>
> Hope this helps.

Good point, David.
Although here I've found a lot of small attributes only containing Y/N- values, for which nobody knows what they're standing for. When creating new records, they are always filled with the default. It's a beautiful example that adding rows does not lead to more documented use.

I do not belong to any school, I'm only open-minded. I think I look to data the wrong way. IfI see records with many attributes, mostly all 'N', except for the column HasWheels, I get confused.
When I only see one record telling (123, HasWheels,Y), it points me directly to the fact that furniture 123 has something special: it has wheels!

Is adding a column not forcing the application to stop as it will lock the entire table, or am I wrong?

Anja. Received on Sat Sep 01 2007 - 09:03:42 CEST