Design of Test Cases

* Undetectable errors are infinite in quantity and variety, in contrast to detectable errors, which by definition are limited

Although structured coding techniques and coding standards improve the clarity and enable enhancement of the end product but it won’t be wise to assume that the application will always be perfect, i.e., free from any error or bugs. A bug may be there in the application because of incorrect requirement specification, improper design, or because of coding errors. So proper testing of the application should be carried out to enhance the quality and productivity of the application.

* There is still one more bug left in your program

Few points to be remembered during testing are:

• testing should be carried out with a purpose of finding errors rather than to determine what is correct

• extreme cases should be considered

• good test cases should be used

• line by line code walkthrough and inspection should be carried out

• checking number of actual and formal arguments of a function to be in agreement

• unnecessary code should be removed

• proper validation of the fields should be carried out

• testing should be carried out in structured manner starting from the testing of each module to that of the testing of the GUI

• testing and validation are recursive processes and should be carried out at all phases of the software development

• the testing team should be preferably different from that of the coding team as it is easy to find other’s errors than one’s own

* Test to find an error rather than to determine what is correct

There are basically four types of tests, which need to be performed on an application to ensure its productivity and applicability. These are as follows:

• Functional test: Functional test cases specify operating conditions, typical input values and typical expected results. Function tests also test behavior just inside, on, and just beyond the functional boundaries.

• Performance test: Performance tests are designed to verify response time under varying loads, percent of execution time spent in various segments of the program, throughput, primary and secondary memory utilization.

• Stress test: These are designed to overload a system in various ways. Example of stress test include processing more than the allowed number of identifiers or static levels, or disconnecting a communication link.

• Structure test: These are concerned with examining the internal processing logic of a software system. All the routines called and all the paths traveled should be checked properly.

* A successful test is one which find errors

Before the actual implementation is done by the programmer the programmer should be provided a set of test cases so that he can take care of it. This actually ensures that the minimal common bugs are taken care during the implementation itself.

The designer should give a numbered list of test cases for each Form and Report as follows: