Hardware Reference
In-Depth Information
objective. The
states and transitions
approach considers a TC as a set of transition
patterns among states. The
contractual
approach defines TC as a contract since the
outcomes of pre-defined conditions are fully defined. Finally, there are several
other
definitions
stemming from various contexts. Table 3 lists examples of definitions in
each category. The implications of this variability are discussed next.
5 Discussion
The TC serves as the backbone of testing processes, and is a fundamental unit for test
planning, execution, monitoring and control. It is also used as a common metric in
quantifying testing effort, scope and status. Furthermore, there is a growing quest to
automate TC generation, execution and management. Nonetheless and quite interest-
ingly, there is no consensus regarding the formal definition of a TC.
From the papers reviewed for this work it is evident that the TC concept is fre-
quently used in various contexts, yet infrequently formally defined (Table 4). Please
note that numbers in Tables 4 and 5 do not add up because papers could be classified
to more than one sub-topic.
Table 4.
TC-related papers, definitions and contexts used
Topic
Cost/ROI
Estimations
Manage-
ment
UML/MBT
OO/SOA
Automation/
Generation
Metrics
Total
TCs Papers
Reviewed
86
69
46
25
44
267
Formal TC
Definition
11 (13%)
19 (28%)
26 (57%)
4 (16%)
14 (32%)
38 (14%)
Table 4 shows that 267 reviewed papers referring to TCs covered five different
topics, yet only in 38 papers (14%) a formal definition of TC was attempted, particu-
larly in studies focusing on OO related issues and TC automation and Management. It
is thus valid to wonder why only 14% of authors bothered to formally define the cen-
tral concept of their work in spite of heavily using this term (some mention TC more
than a hundred). Thus, in the 38 papers where TC was defined, various definitions
were employed representing all four definition categories: input-process-output-
objective, states & transitions, contract, and other. It is thus interesting to examine
whether there is an association between the definition category used and the specific
context (Table 5). For example, it could be expected that works in the UML/MBT/OO
context would use states & transitions definitions that stem from the OO world. This,
however, could not be substantiated by the present literature review, as those few
authors who have used the TC definition in their OO-related work chose definitions
from all categories (Table 5). Moreover, no author has articulated the reasons for
choosing one definition or another. As seen in Table 5, authors using TCs in the con-
text of OO/MBT/UML more frequently used the input-process-output-objective
(termed hereinafter process-based for brevity sake) definitions rather than the more
naturally related states & transitions definitions, which turn out as the most popular
definition category. Evidently, no correlation could be deduced between the definition
category and the context, possibly attesting to the arbitrary choice of the former.
