Logo
/ application-testing

5 key software testing steps every engineer should perform

By DeviQA on Tue Dec 26 2017 00:00:00 GMT+0000 (Coordinated Universal Time)

Subscribe Now!

Get the latest QA news and tips from DeviQA

Introduction

With the increased focus on quality of the software delivered, testing commences even before code is compiled for the first time. Looking at this, here are some essential steps a software engineer should do:

Testing basic functionality

When testing software, it may not always be an obvious step to check that all controls on all forms, pages, windows, etc. work. Having them accurately linked to whatever logical result is one aspect, checking for when a button presses and starts its process is another. It is essential that a software engineer tests for the basic functions of a software. While test cases may relate to specific inputs and outputs, without checking that everything works (whether or not it needs to be part of a complex test case), the testing cannot be considered complete. If the software has links to external applications, it is also necessary to test that the basic connections work appropriately. This step is essential because it helps separate logical errors in execution of more advanced algorithms that result due to basic functionality errors instead of errors in said algorithms.

Reviewing source code

Some aspects of the code may be faulty and fail to run. Before these are gone through testing in the system, it is helpful to take another look at the source code manually. Performing a walkthrough in one’s mind or notebook to check how the code would execute and whether something would break. This helps correcting possible errors then and there before even running the software. This step is so essential that many organization make review of source code a part of their standard protocol. This step can be improved if the review is done by a peer instead of the developer himself. The second person would be more objective with the review and have a higher chance of identifying errors.

Analyse static code

Having tested the basic functionality and review the source code, it might be time for a live test. However, before that engineers should perform a static code analysis, which means testing code without actually executing it. Static code analysis identifies potential problems with the code that would affect the execution, possibly indirectly. Examples include unused variables, undeclared variables and security vulnerabilities. Several static analysis tools are available to test code for such problems. Many aspects are already built into the development environments used to code. Both of these can be utilized to clear as much errors as are possible before moving towards live testing.

Unit testing

Once everything looks good on paper, it’s time for live testing. Unit testing means testing out a part of the code (such as a class, a method, an algorithm, etc.) separately. Doing so helps narrow down the error to a specific part of the code and also warns users of that part of the code what to expect when running their errors. Unit testing is a necessary part of any software development project and is an industry standard. Under the modern approach of continuous integration, unit tests are performed each time code is updated to check if the change broke any units. If no errors occur, the code is good for deployment and further use. Further processes using the faulty portion of the code may be put on a halt until the unit is fixed.

Performance testing

If software works correctly but takes too long to get things done, it cannot be ranked as high quality. Therefore, another essential step is to test for performance. This means testing the speed of operations in the software. Performance can be measured using automated tools or manually checking how long an activity takes. It has to be tested for a single user and for multiple users operating at the same time. Single user performance is helpful in identifying errors in the responsiveness due to the code itself, including processing of code on the client-side. When testing for errors in functionality, single user testing should be performed. Once single user testing has been completed, load testing can commence. The purpose for this is to check whether the code is able to handle multiple connections between the server and clients and whether this affects data stores. When doing this test, the target is to break the system with the load so that the maximum limit can be identified and it can be assessed how to increase this limit and what kind of costs that would incur.

Conclusion

All these steps are to be performed by the developer before handing in his work to save time with obvious errors caught in software testing or further development processes.