Page 289 - Software and Systems Requirements Engineering in Practice

R
ç
T
E

P
I

A
I
S
N
Q
E
A
P
U
H
N
T
M
E
ç
C
H
4
E
$
E
D
ç
V
O
P
E
L
ç ç # # H A P T E R ç ç ç 2 2 A P I D ç $ E V E L O P M E N T ç 4E C H N I Q U E S ç ç
SMALL NUMBER OF REPRESENTATIVE TEST CASES TO BE USED TO INTERNALLY VERIFY
THE MODEL OR PROTOTYPE UNTIL THE NEXT OPPORTUNITY TO PRESENT THE RESULTS
TO THE DOMAIN EXPERT 5NLIKE THE COMMON USE OF TEST FIRST DEVELOPMENT
WE DON T WANT TO PRODUCE A TEST SUITE THAT VERIFIES THE QUALITY OF ANY
PRODUCED SOFTWARE /UR GOALS ARE LIMITED TO HAVING AN ORACLE THAT
UNAMBIGUOUSLY VERIFIES IF THE IDEAS OR SYSTEMS WE ARE EXPERIMENTING
WITH WILL SATISFY THE REQUIREMENTS GIVEN BY THE DOMAIN EXPERTS
/BVIOUSLY THE QUALITY OF THE FEEDBACK COLLECTED FROM THE
VERIFICATION AGAINST SUCH TESTS IS DETERMINED BY THEIR
REPRESENTATIVENESS 7E CONSIDER THE TESTS TO BE REPRESENTATIVE IF THEY
MODEL THE BEHAVIOR IN SITUATIONS WHERE THE REQUIREMENTS ARE NOT
WELL UNDERSTOOD !LSO THE PREDEFINED TESTS SHOULD BE SUFFICIENTLY
SPECIFIC AS TO MAKE THE VERIFIED SYSTEM UNAMBIGUOUS 4HIS LEVEL OF
DETAIL IS THE RESPONSIBILITY OF THE DOMAIN EXPERTS AND THE PROTOTYPERS
USUALLY DON T HAVE SUFFICIENT DOMAIN EXPERTISE TO FORMULATE SUCH
DOMAIN SPECIFIC QUESTIONS 4HIS APPROACH IS ONLY USED AS AN INTERIM
STEP AND ITS RESULTS MUST BE VALIDATED WITH THE DOMAIN EXPERTS
IN THE FORM OF A PROTOTYPE BEFORE BEING CONSIDERED hACCEPTED v
4HE SET OF TESTS THAT SATISFIES OUR GOALS HAS THE FOLLOWING
CHARACTERISTICS IT IS REPRESENTATIVE OF THE IMPORTANT REQUIREMENTS OF
THE SYSTEM AND IT IS UNAMBIGUOUS ! PROTOTYPE MODELS THE TARGET
SYSTEM WHILE THE SET OF TESTS MODELS THE EXPECTATIONS OF THE SYSTEM S
IMPACT ON ITS ENVIRONMENT
-ODIFICATIONõ/PTIMIZATION
! PRIMARY OPTIMIZATION GOAL FOR OUR PROTOTYPING APPROACH IS TO
EMPHASIZE THE COLLABORATIVE ASPECTS OF INTERACTION AMONG THE
STAKEHOLDERS AND TO MAXIMIZE THE EMERGENT SHARED UNDERSTANDING
7E ASSUME THAT THE PROTOTYPING EFFORT IS TAKING PLACE EARLY IN THE
DEVELOPMENT PROJECT FOR CERTAIN FEATURES THAT ARE UNKNOWN SUCH THAT
THE TOTAL COMPLEXITY CAN T BE ACCURATELY ESTIMATED OR PLANNED
(OWEVER EARLY DETERMINATION OF STABLE REQUIREMENTS IS A CRITICAL ISSUE
THAT WILL ALLOW SUBSEQUENT OPTIMIZATION ON DEVELOPMENT EFFORTS 7E
BELIEVE THAT LIMITED EARLY EFFORT AIMED AT FOSTERING THIS SHARED
UNDERSTANDING AND TRUST ELIMINATES WASTED EFFORT AND REDUCES THE
RISK INHERENT IN DEVELOPMENT LATER ON
4HE PROTOTYPING PROCESS IS CENTERED ON ARTIFACTS THAT
UNAMBIGUOUSLY REPRESENT SOME ASPECT OF THE PLANNED SYSTEM OR
FEATURE )T IS RARELY THE CASE IN PRACTICE THAT THE ISSUES OF INTEREST FOR
EVALUATION ARE VERY EASY TO INTEGRATE INTO ONE ARTIFACT PRIMARILY
BECAUSE THEIR INTERPLAY IS UNKNOWN AT THAT TIME #REATION OF MULTIPLE
ARTIFACTS FORCES SOME OVERHEAD IN CREATION AND UPDATES BUT IT PROVIDES
THE NECESSARY DEGREE OF FREEDOM FOR THE INDIVIDUAL ARTIFACTS TO EVOLVE
TOWARD THE BEST REPRESENTATION OF THE DESIRED ASPECT OF THE SYSTEM
7ITH MULTIPLE ARTIFACTS IN CONCURRENT EVOLUTION AND ANALYSIS THE
FEEDBACK LOOP ON EACH ARTIFACT IS SHORTENED AND IS DECOUPLED FROM
ANY PROBLEMS THAT MAY SLOW THE PROGRESS FOR SOME OTHER ARTIFACT