Page 267 - Software and Systems Requirements Engineering in Practice

ç ç 3 O F T W A R E ç ç 3 Y S T E M S ç 2 E Q U I R E M E N T S ç % N G I N E E R I N G ç ) N ç 0 R A C T I C E

MODELS AS EARLY AS POSSIBLE IN THE PROJECT SO THAT THE TESTING TEAM CAN
EFFECT THE DEFINITION OF REQUIREMENTS EARLY IN THE PROCESS TO ENSURE
THAT TESTABLE REQUIREMENTS ARE SPECIFIED
!S THE USE CASE MODELING OCCURS EARLIER IN THE DEVELOPMENT PROCESS
WE EXPECT TO SEE MANY IMPROVEMENTS IN SYSTEM TESTING 4HE FOLLOWING
IMPROVEMENTS ARE EXPECTED
v )MPROVEMENT IN THE OVERALL QUALITY OF GENERAL CUSTOMER USE
CASES IN GRANULARITY AND VARIATIONS
v 2EDUCTION IN THE OVERALL VALIDATION EFFORT FOR SYSTEM TESTS
v )MPROVED COMMUNICATION TO ALL STAKEHOLDERS ABOUT
REQUIREMENTS ISSUES THAT RESULT FROM THE DEVELOPMENT OF THE
TESTING USE CASES
v &INDING DEFECTS AS EARLY AS POSSIBLE BECAUSE YOU CAN DISCUSS
INCONSISTENCIES AND IMPACTS TO THE WORKFLOW MODEL WHEN IT IS
DEFINED AND RELEASED

)MPROVEDõ%FFICIENCY
)N TRADITIONAL TESTING THE CHALLENGE IS HOW TO EFFICIENTLY CREATE THE
TESTS )N MODEL BASED TESTING BY CONTRAST THE PROBLEM IS OFTEN HOW TO
NOT GENERATE TOO MANY TESTS 4HE TEST GENERATION TOOLS THAT WE USE
HAVE FEATURES TO PRUNE THE NUMBER OF TEST PATHS THROUGH THE MODEL TO
REASONABLE NUMBERS WHILE STILL COVERING ALL ACTIVITIES OR ALL TRANSITIONS
THROUGH THE MODEL )T IS EASIER TO CREATE A LARGE NUMBER OF TESTS USING
THIS TECHNIQUE THAN USING TRADITIONAL TESTING METHODS ;/STRAND ET AL
= ;"ALCER ET AL = 4HIS CREATES THE NEED FOR SOME SORT OF
PRIORITIZATION SCHEME TO DECIDE WHICH OF THE GENERATED TESTS HAVE
HIGHER PRIORITY AND SHOULD BE EXECUTED FIRST !LTHOUGH WE ARE
CURRENTLY ABLE TO MANAGE THE NUMBER OF TESTS WE WORK WITH IT WOULD
BE DESIRABLE TO BE ABLE TO ASSIGN PRIORITIES TO THE TESTS -ULTIPLE FACTORS
AFFECT PRIORITIES FOR EXAMPLE GIVING PRIORITY TO RECENTLY CHANGED
CODE GIVING PRIORITY TO SCENARIOS THAT ARE MORE LIKELY TO BE EXECUTED
BY USERS GIVING PRIORITY FOR SCENARIOS WITH MORE SERIOUS CONSEQUENCES
IF THEY FAIL *UST AS WE CAN MODEL THE TEST SCENARIOS IT IS POSSIBLE TO ADD
FEATURES TO PERFORM RISK BASED TESTING AS WELL
7E HAVE USED ACTIVITY DIAGRAMS FOR DESCRIBING USE CASE DIAGRAMS
AND WE HAVE EXPLORED USING SEQUENCE DIAGRAMS TO DESCRIBE TEST
SCENARIOS 4HIS WORKS ADEQUATELY BUT WE WOULD ALSO LIKE TO BE ABLE TO
USE 5-, SEMANTICS WITH COMBINED FRAGMENTS TO REPRESENT
CONDITIONAL EXECUTION OF SEQUENCES )N SOME CASES SEQUENCE DIAGRAMS
MAY BE A MORE EFFICIENT REPRESENTATION OF SCENARIOS