Relations as Repeating Groups & Namespaces

A relation within that schema is named uniquely within that namespace. The relation is then a sub-namespace where attributes are named uniquely within that namespace.

Within a SQL-RDBMS I could not think of any other clear namespaces below that level. There are undesignated namespaces for values, such as a State attribute providing a scope for the City attribute so if the value of City is "Grand Rapids" that value does not stand on its own, it requires a designation, such as a State of MI, to resolve the meaning of the value.

Repeating Groups...
The root name space (an Oracle schema, for example) has relations as entities and, therefore, it has repeating groups (which relations are) as entities. The secondary namespaces, relations, are now permitted to have repeating groups too, but they are not part of the SQL-92 standard and infrequently used by anyone using a SQL-DBMS.

Other models go one or more levels deeper than the relational model in permitting namespaces below the relational level. Other models are also more consistent in permitting elements within the root name space to be repeating groups (e.g. relations) as well as elements within sub-namespaces and so on.

So, if you look at the root name space as analogous to a relation, you can see that it would be mathematically elegant, at least, to permit the top level name space and sub-name spaces to play by the same rules rather than being decidedly different as they are in a SQL-DBMS.

Did that make any sense? If not, ask questions as I want to write this up in a way that is not dismissed (or would like to know if this point is so stupid it should be dismissed, but be gentle ;-) Thanks! --dawn Received on Thu Jun 03 2004 - 05:16:08 CEST