Planning the Method Hierarchy Oracle Database Tips by Donald Burleson

While the coupling of data with behaviors is a revolutionary concept for database management, the failure to properly plan the implementation can be a disaster.  In order to achieve the benefits of using methods, a careful decomposition of the processes must take place.  As we discussed, there are three types of processes, those that are independent of the database classes, and those that are attached to a base-level database class and those that are attached to aggregate database classes.  In addition, any method may have other methods nested within its structure, while at the same time participating as a sub-method within another method (Figure 8.1)

Figure 8.1 - The recursive many-to-many nature of method nesting.

Here we see that the check_credit() method is composed of the sub-processes, check_payment_history() and check_credit_references() methods.  At the same time, check_credit() participates in the place_order() and the hire_employee() methods.  Given this huge array of choices, where do we begin developing the methods?  There is a sequence of events that must take place to achieve this mapping of methods to classes:

1. Before the mapping of methods to classes can begin, you should have already created the following analysis and design documents:

1. A set of fully decomposed data flow diagrams for your system - This will be used as the specification for all of the methods.  From the DFD we will gather the method names, the input and output values, and the breakdown of nested methods. 

2. A entity/relationship diagram for the system - This diagram will be used to identify the base classes for the system.

3. An aggregate object diagram - This diagram is used to associate the higher level processes with the methods that will be attached to these classes.

2. From the DFD, create a prototype for every process on the data flow diagram.  This will formally state the input and output parameters for each process on all levels of the DFD.

3. Identify and prototype all standalone functions in the system.

4. Map the prototypes to the entities.

While this mapping of processes to methods may seem to be a straightforward approach, there are many new concepts that a traditional analyst my not be familiar with using.  For the purpose of illustration, we will use the running example from our order processing system that was discussed in earlier chapters and use as our mapping example the sub-processes that are contained within the fill_order process.

 

---