4.3 Discrepancies between Pre-ID and
Post-ID
A detailed analysis of the source code enables a qualitative
evaluation of the discrepancy between the Pre-ID and Post-ID
software units. Table 6 presents the analysis of the 14 designed
classes, which includes the nine adapted classes and the five
added classes based on a concern and coupling analysis. This
qualitative analysis is based on the observations performed by an
evaluator reading both Pre-ID and source code.
Concerns are defined with respect to the purposes of a class, for
example interface, serialization, undo/redo control, or data
computation. Coupling defines relationships with other classes,
through internal objects or method calls. This qualitative analysis
confirmed the presence of a design drift: Some concerns and
coupling accounted for in the Pre-ID were not found in the Post-
ID, and similarly, some concerns and coupling not present in the
Pre-ID were found in the Post-ID.
The discrepancy evaluation (Delta) between the Pre-ID and Post-
ID is based on the following two-level scale. Minor Delta means
that the discrepancy shows only trivial changes between the Pre-
ID and Post-ID artifacts. Major Delta means that significant
functionalities were added or removed between the Pre-ID and
Post-ID artifacts.
For example, in Table 6, the second class, named CMainWindow,
which is an adapted (redesigned from the existing library) class,
had a Pre-ID of good quality but was badly implemented as
evaluated by the expert. Major modifications were found between