What are some of the guidelines for test execution? Why
are these guidelines important?
There are eight axioms for software validation testing:
- Tests should be used to show the presence of errors, not their absence.
- Testers must know when to stop testing
- Unplanned, non-reusable, throw-away test cases should be avoided
unless a throw-away program is shat you are developing.
- Know the definition of the expected output/result: compare actual
vs. expected results of each test.
- Test cases must be developed to give the desired output conditions.
Determine the input required to give a pre-designed set of outputs.
- Apart from integration testing, the program should not be tested
by the developing organization.
- Test cases must be developed for invalid/unexpected as well as valid/expected
input conditions.
- Number of undiscovered errors is directly proportional to number
of errors already discovered.
Knowing these axioms, test can developed:
- Requirements-based tests based on black-box strategy: user requests
are tested without having to know the internal code or design specs.
Tests are based on requirements docs but developed using functional
design
specs.
- Function-based tests based on white-box strategy: tests designed
by using functional design specs.
- Internal-based tests based on white-box strategy: tests designed
by using functional design specs.
Finally, any test-execution guideline should meet the following basic
criteria:
- Define results: the expected output should be pre-defined. Otherwise, “the
eyes see what they want to see.”
- Be repeatable: each time the same test case is run, it should produce
identical results (on the same hardware/software config). If results
are to the contrary, the failure should be recreated by the developers
when they debug.
More Software Testing information:
top of page
|
