by Anindya Sen

Created on: January 29, 2007   Last Updated: November 22, 2007

Software evaluation and testing

The software verification and validation tests are done to check the performance of a software product.

Software verification is the process of checking to ensure that the developed product accepts input and produces output within the limited scope of its declared specifications.

Software Validation checks for repeatability. The output is repeatable for any number of times that the same test is repeated. Validation is a check to determine that the system always give the same output for the same set of input.

There are several test methods to evaluate a software program or a module. The methods primarily adopted are, black box or functional testing, white box testing, module testing, unit testing, regression testing, user specification test, alpha test, and beta test. Hazard detection, fault tree determination are used to do fault analysis of the system. There are more methods. A brief description of each method follows.

Black box/functional test
A person without the internal knowledge of the box should check that the box meets the output functional specifications for a given specified input. This should be done by a non-programmer. Hazard detection,

White Box test
This testing should be done such that the internal knowledge of the box is tested as well as the output functional specification. This is done by the programmer.

Module test
This is testing at the program code level to find out if each pre-defined functional module accepts the specified inputs and produces the specified outputs without any error.

Unit test
This testing method is code level checking for open ended loops or ill terminated conditions, memory leaks. (An utility called bounds checker is routinely used.)

Regression test
Every time a new function is added it is necessary to repeat the test set for the older or previous functionalities, in addition to the tests for the new function. This is to make sure that there is no error propagation due to a software modification. Regression test can be time consuming but it makes sure that the software model behaves properly after a software change.

User-specification test
User-specification test is done to make sure that product developed has built in functions, which adheres to the guidelines of the user specified functions.

Alpha test
This is the complete testing of the first total build of the software, at the vendors own site.

Beta test
This is the complete testing of the first total build of the software, at a selected user

• 1
• 2
• Next Page >>