Designing the Data Structure › Data Model Retransformation › Basics of Retransformation

Basics of Retransformation

CA Gen retains complete knowledge of how each component of the data model is implemented in the data structure. Each object on the list (except for indexes) is the direct implementation of an object or objects in the data model. CA Gen can easily keep track of what has or has not been implemented. CA Gen-executed consistency check refreshes the Data Structure List and the Data Store List to eliminate the errors.

Note: A change to the DSD name for an implemented entity type or attribute does not cause a consistency check error and so will never result in a change to the name of the corresponding table or column.

As system development progresses and the data model changes, you may need to transform specific pieces of the data model. These changes may need to be implemented into all of the possible DBMSs, only a few, or only one. This is achieved through incremental retransformation.

These situations may require incremental retransformation:

• Adding a new entity type, relationship, or attribute to the data model
• Changing an existing entity type, relationship, or attribute

When new objects are added to the data model, their implementations must be added to the Data Structure List for the targeted DBMS as well.

When an existing data model object is changed, its implementation on the Data Structure List must be deleted. CA Gen then puts the object on the list of unimplemented data model objects, and it may then be added as if it were a new data model object.

The following list details the rules of retransformation for multiple target DBMSs:

• If you want to make sure the data model is transformed for all of the targeted databases, use the synchronization of the data model to the technical design.
• If you need to update only a single DBMS with changes from the data model, use specialize technical design for current DBMS.
• If you have only a single target DBMS, use synchronization of the data model to the technical design. If there is only a concern about a single DBMS, this will always keep all of the DBMSs synchronized.

An object deleted from the data model requires no retransformation action as the object and its dependent objects are immediately deleted from the Data Structure List.

For example, deleting an identifying attribute in the data model will cause the following to be deleted:

• Corresponding column in the table
• Column in the unique column
• Foreign key columns in dependent tables to be deleted
• Foreign key columns in the indexes to be deleted