Implementation - Software Engineering: A Hands-On Approach - page 181

A major strength of top-down integration is that major design flaws show up
early. The modules of a product can be divided into two groups: logic modules and
operational modules. Logic modules incorporate the high-level decision making
flow of the product, and are generally situated close to the root of the module
interconnection diagram. From Fig. 9.5 , it is reasonable to expect that modules a,
b, c, d, and perhaps g and j are logic modules. The operational modules perform
the actual operations of the product, such as getLineFromTerminal or measu-
reReactorTemperature. These modules are generally found at the lower levels,
close to the leaves of the module interconnection diagram. In Fig. 9.5 , modules e,
f, h, i, k, l, and m may be operational modules. By coding and testing the logic
modules before the operational modules, top-down integration will exploit any
major design faults early in the development process.
9.10.1.2 Disadvantages
The main disadvantage of top-down integration is that potentially reusable modules
may not be adequately tested. These are the operational modules, which implement
fundamental functions and therefore, are likely candidates for use in future projects.
Unfortunately, these modules are generally the lower-level modules and as such, are
often not tested as thoroughly as the upper-level modules. For example, if there are
184 modules, the root module will be tested 184 times, but the last module integrated
will be tested only once. This makes the reuse of operational modules a much more
risky proposition.
9.10.2 Bottom-Up Integration
In bottom-up integration, if module mAbove calls module mBelow, then mBelow
is implemented/integrated before mAbove. Returning to Fig. 9.5 one possible
ordering is l, m, h, i, j, k, e, f, g, b, c, d, and a. In order to have the product coded
and tested by a team, a better bottom-up ordering would be as follows: h, e and
b are given to one programmer and i, f and c to another. The third programmer
starts with l, m, j, k and g, and then implements d and integrates his/her work with
that of the second programmer. Finally, when b, c, and d have been successfully
integrated, a can be implemented/integrated.
9.10.2.1 Advantages
The operational modules are more thoroughly tested when using a bottom-up
strategy. Bottom-up integration also provides fault-isolation, as does top-down
integration.