CSIS 370 – Lecture 5
Testing Computer Software: Chapter 1
Key Points:
- Don't waste time testing a program that doesn't even do simple things correctly
- Boundary tests are the most powerful
- Can't test every combination, so you need to pick carefully
- The reason the program is in testing is that it probably doesn't work
- About 1 in 3 attempts to fix a bug either doesn't fix it, or creates another problem!
- The more you know about computer programming, the more internal boundaries
you can anticipate and test for.
Page 9: Example of a non-obvious internal boundary condition – the values 127 and -128
based on the possibility that the data is stored in a single byte.
Common programmer/project manager response to some bug reports: "That's not a bug",
or "no one would do that", or "it will never happen". Why do you think they engage in
this behavior?
What does book suggest doing about this? Do you think it would be effective?
Find test cases that make the bug appear in situations that are so likely to occur that no
one would put up with it.
Measure of best tester: one who gets the most bugs fixed. This requires humility and
craftiness – don't want to antagonize the developers, but you do want to convince them
that the bugs need to be fixed.
Discuss Triangle Example.
Get students to develop test cases. What's a test case?
- list of input values, and the expected result. If you don’t know
what you expect, how can you know if the software is right or
wrong?
positive testing – test case that shows software responds as we expect
it to when we supply good data
negative testing – test case that shows that software behaves
reasonably when we give it bad data.
How many test cases do we have just for this simple program? Could
we test all possible combinations of the 3 input values?
[Book example: 39,601 combinations of numbers only!]
