Page 266 - Software and Systems Requirements Engineering in Practice

E
M
R
U
I
S

T
E
N
Q
E
R
T
A
P
2
E

ç

4
E
ç
E
M
N
G
I
S
T
T
V
E
I
$
R
Y
S
3
N
ç
H
ç ç # # H A P T E R ç ç ç 2 E Q U I R E M E N T S $ R I V E N ç 3 Y S T E M ç 4E S T I N G ç ç
2EVIEWINGõ-ODELS
7E HAVE FOUND THAT IT IS EASIER FOR DOMAIN EXPERTS AND OTHER
STAKEHOLDERS TO REVIEW MODELS SUCH AS USE CASE DIAGRAMS THAN TO
REVIEW TEST SPECIFICATIONS ! USE CASE MODEL IS WRITTEN FROM THE
PERSPECTIVE OF THE END USER AND THIS IS THE PERSPECTIVE UNDERSTOOD
BEST BY THE PRODUCT STAKEHOLDERS 2EVIEWING THE USE CASE MODELS NOT
ONLY FINDS GAPS IN THE TESTING BUT ALSO DISCOVERS GAPS IN THE
REQUIREMENTS SO THE MODELS HAVE MULTIPLE BENEFITS
4HE ACTIVITY DIAGRAMS THAT DESCRIBE THE SCENARIOS SPECIFY THE
DESIRED TESTS IN THE SAME WAY AS A TEST SUITE BUT THE DIAGRAMS ARE
EASIER TO REVIEW THAN SAY PAGES OF TEST SPECIFICATIONS WRITTEN IN
PLAIN TEXT 4HERE ARE FEWER ACTIVITY DIAGRAMS THAN TEST CASES SINCE AN
ACTIVITY DIAGRAM CAN REPRESENT SEVERAL TEST CASES BECAUSE OF THE
DECISION POINTS AND DATA VARIATIONS DESCRIBED IN THE ACTIVITY
DIAGRAMS
7E ALSO HAVE OBSERVED THAT THE DEVELOPMENT OF USE CASE MODELS
FORCES REQUIREMENTS TO BE TESTABLE AND DOCUMENTS AMBIGUITIES IN
REQUIREMENTS 7HEN REQUIREMENTS ARE NOT TESTABLE OR AMBIGUOUS
EITHER THE REQUIREMENT IS REWRITTEN OR THE USE CASE DESCRIPTION
DOCUMENTS A TESTABLE INTERPRETATION OF THE REQUIREMENT
)MPROVEDõ4ESTõ#OVERAGE
'ENERATING TESTS FROM ACTIVITY DIAGRAMS ALLOWS US TO SPECIFY A BASIS
FOR TEST COVERAGE 7E CAN CHOOSE TO CREATE TESTS FOR EVERY PATH IN THE
ACTIVITY DIAGRAM OR EVERY TRANSITION IN THE DIAGRAM AND WE CAN BE
ASSURED THAT WE HAVE CREATED TEST CASES FOR EACH OF THESE CASES -ISSING
TEST CASES ARE ONLY A RESULT OF INCOMPLETE MODELS AND THE MODELS
PROVIDE TRANSPARENCY TO WHAT IS TESTED AND WHAT IS NOT TESTED
4RACINGõTOõ2EQUIREMENTS
3YSTEM TESTS ARE REQUIRED TO BE TRACED BACK TO REQUIREMENTS IN ORDER
TO DEMONSTRATE VERIFICATION OF REQUIREMENTS 7HEN USE CASE MODELS
ARE USED AS THE BASIS FOR MODEL BASED TESTING THE MAPPING BACK TO
REQUIREMENTS IS MORE NATURAL THAN WITH OTHER TYPES OF MODELS SUCH AS
STATE MACHINES WHICH TYPICALLY ARE ASSOCIATED WITH IMPLEMENTATION
COMPONENTS
3INCE USE CASES CAN BE EASILY ASSOCIATED WITH REQUIREMENTS THE
GENERATED TESTS CAN ALSO BE MORE EASILY ASSOCIATED WITH REQUIREMENTS
7HEN TESTS ARE GENERATED IT IS ALSO POSSIBLE TO AUTOMATICALLY GENERATE
TRACES BACK TO THE ASSOCIATED REQUIREMENTS
3TARTõ%ARLYõINõTHEõ$EVELOPMENTõ,IFEõ#YCLE
&OR MOST OF THE PROJECTS WE WORK ON THE USE CASE MODELS FOR AUTOMATED
TESTING ARE DEVELOPED DURING THE SYSTEM TEST PHASE OF THE PROJECT
(OWEVER WE RECOMMEND BEGINNING THE DEVELOPMENT OF THE USE CASE