Information Technology Reference
In-Depth Information
the application was not “foolproof.” One's QA team must be
conscious of this fact. If the application is intended for such users,
they should make an additional effort to test beyond the require-
ment or solution specification. In fact, code coverage-type tests are
better suited for such environments. Between the two types, the
computer savvy users are more forgiving when it comes to com-
plaining about problems. Having a user educated about the com-
plexity of the engineering involved in the product helps in adjusting
his or her expectations.
Development cycle phase
: The extent and type of testing also
depends on the phase of the application.
New applications that are being tested for the first time will
require more effort and more cycles of testing even if they are
relatively simple.
Applications in corrective maintenance (user-reported defects
are being fixed) require testing efforts that depend on the
criticality and severity of the fixes being made. Because of the
prior testing experience with the application at the time of its
full release, estimating this effort can be easier. Regression
testing should be accounted for in one's estimations.
When functional enhancements are being made in a product
release cycle, additional time should be allocated for creating
new test cases for the enhancements. The need for regression
testing still continues.
Applications that are in adaptive maintenance (auxiliary devel-
opment for supporting additional OS platforms and tools) need
testers who are experts in the respective platforms.
Dealing with Poor Specifications
The approach that quality is about meeting specifications is not an issue
by itself. The problem sometimes lies with the quality of the specifications.
Specifications are, like models, abstractions of an often complex and messy
reality. They are necessarily incomplete. This means that one can meet
specifications yet fail in the real world. For example, an application's
specification may say nothing about performance requirements or the
support of multiple Web browsers.
QA often faces such situations. Should QA include such tests? QA may
decide to ignore the issue of multiple browsers, or it may want to add
such tests to the test suite. The right way to handle this is to ensur e,
through upstream processes, that such a requirement is added to the
specifications. Such problems with specifications can be discovered early
