Schema Versioning in OLE DB Scanner
When scanning a database catalog that is already in a repository database (identified by identical catalog names), the scanner versions the schema at the lowest granularity of change. For example, take the following example data model of an Authors table with two columns.
| Name | Data type | Size |
|---|---|---|
| au_lname | Varchar | 20 |
| au_fname | Varchar | 40 |
Changing the data type of the au_lname column results in a new version of the column object and a relationship to the new data type object.
In general, adding or removing an element of a relationship collection requires that you create a new version of the origin object. For example, adding a column to the table results in a new version of the table with the new column object added to the elements collection. Relationships to existing columns are propagated.
Removing a column from the table results in a new version of the table with the column object removed from elements collection. Relationships to existing columns are propagated.
| Name | Data type | Size |
|---|---|---|
| au_lname | Varchar | 40 |
| au_mname | Varchar | 10 |
The following diagram shows the model for the revised table schema.
