the huntrods zone - teaching

Creating and Documenting Test Plan

A good test plan should list the following test conditions, together with expected and ACTUAL results. I will use a temperature conversion program as my example:
  • normal data (i.e. input temperature = 100.0) - at least 5 relevent examples
  • abnormal data (i.e. input temperature = 'q') - one example of each type of abnormal data
  • limit condition data (i.e. absolute zero, 0 or 32, 100 or 212 - for temperatures) - as many as required to define limits
For each data set, expected as well as actual output should be reported. Screen shots are not necessary.

The idea here is that good testing tests many normal data conditions, as many bad data inputs as possible, and all limit conditions. You need to provide a representative sample, not an exhaustive test suite.

Remember, a test plan is just telling the user what you did to test the programs: how you know they are working. "The program ran" is not an acceptable test plan. Tell the user *exactly* what you expected for output, and what you actually saw. This can be done even if the program accepts no input. For programs that produce a lot of output, selected items may be reported. You don't have to calculate all the primes from 1 to 10000, for example - just a few. You can also reference a book for a figure or table of required output, if appropriate. is a small program illustrating good test plan practices.

copyright © Richard S. Huntrods
all rights reserved

last updated: