A Full Software Development Lifecycle Testing Project - Software Testing: Testing Across the Entire Software Development Life Cycle - page 221

fi le of all 75 locations, and a completed certifi cate fi le of all 1,267 certifi cates issued
last month. A total of 1,542 test data records would be manually built using the brute
force approach. An example of the equivalence class approach would be to start by
creating 20 certifi cates records representative of different ages, different sexes, dif-
ferent instructors, and different classroom locations. Then create only the 10 or so
instructor records needed to print the 20 certifi cate records. Then create only the 5
classroom location records needed to print the 20 certifi cate records. A total of 35
test data records would be manually built for fi rst testing using the equivalence class
approach. Once testing is begun, the number of certifi cate records can be increased,
the number of associated instructor records can be increased, and the number of
associated classroom location records can be increased to meet the specifi c volume
and variety testing needs. Lower volume, wider variety of records are needed fi rst
for functional testing. Higher volume, fairly similar records are needed second for
performance testing.

13.5.3 Functional Testing

Chapter 7 identifi es six functional testing techniques classifi ed as white box testing,
that is, testing with the source code available. These techniques are

1.
2.
3.
4.
5.
6.

Statement coverage technique
Branch (Single condition) coverage technique
Compound condition coverage technique
Path coverage technique
Loop coverage technique
Intuition and experience

As the developers write and debug the DCPS screen programs, the test team encour-
ages the developers to strive for 100% coverage of their code using each of the fi rst fi ve
techniques. The more coverage the developers achieve, the fewer defects are expected
to be discovered later by the testers. The intuition and experience techniques include

1.
2.
3.

dates
zero length anything
buffer overfl ow

and any aspects of the chosen database management system known to be trouble-
some. Dates will play a major role in the DCPS from a data entry standpoint (student
birthdate, date class completed) and from a record management standpoint (comple-
tion records entered today, certifi cate records printed this week, instructors with
certifi cations about to expire, and so forth). Because most data entry fi elds are re-
quired, the “zero length anything activity” will apply more to data records and data
fi les than to data fi elds. Buffer overfl ow may not be a signifi cant area of risk for the
DCPS because the overwhelming majority of process activity is single thread, that
is, one user performing one task at a time on one workstation.