Skip to main content

Posts

Showing posts from June, 2016

Reviews find around 60% of defects. Are you reviewing your stuff?

Be it a formal inspection or a guided informal peer review, reviews on average find around 60% of the defects in the system being built. And even though the evidence in favor of performing reviews is overwhelming, not many organizations practice this simple activity, and its mostly due to lack of management backing. So if you're in a team that doesn't review yet, be sure to propose the reviews as an activity that will surely help improve the quality of the software being built. Benefits of performing reviews: A. Lower cost of defect correction The longer a defect lies in a system, the more costlier it becomes to rectify it. A high severity defect found in production costs a 100 times more to resolve than if it were found during the requirements stage. A low severity defect found in production costs 2 times more to resolve than if it were found during the requirements stage. Since reviews can be performed as early as the requirements stage, the potential cost benefits ar...

A quick guide to writing better test cases

This topic has been touched many times over on the internet; I'll leave out the advise of keeping simple, your test cases and the language of your test cases. Without going into methodologies and techniques used in testing, I'd like to outline a few principles of test case writing that you should bear in mind the next time you're writing test cases. 1. Design scenarios Firstly, don't think of writing test cases.. :) Or anything related to test cases. For the particular application / unit under test (AUT), take time to think through and identify possibilities of: Inputs - All possible inputs that can be given to the AUT. Your inputs can be data or a certain action to be performed, like clicking a button Outputs - All possible outcomes / behaviours of the AUT in response to the input  Trigger points - Conditions that when activated, cause the AUT to perform certain action(s) Alternate flows - When testing workflows, always ensure you identify flows other than t...

3 Common mistakes that testing teams should avoid

A friend and I were recently talking about the kind of common yet critical mistakes that are made quite unknowingly during the test phase of the SDLC. Some of the points mentioned below are the kind of processes mandated by companies for ages! It is not uncommon to see that test managers, along with the test leadership within companies making such blunders. And if you see such practices being followed at your company, let them know that they should be ridding themselves of these common mistakes. 1. Test cases are not traced back to the requirements One of essential artifacts of the testing phase is the Requirement Trace-ability Matrix. What this matrix does is help you link your requirements to your test cases. As a result, you would be aware of the coverage your test cases provide to your requirements. Not having an RTM puts you and your team at the risk of coming up with an incomplete coverage of the requirements  and the possibility of leaking defects. Her...

Attention! These 3 Activities are Slowing Down Your Testing!

These are activities you perform almost daily. You're tired of the monotonous work involved and yet its hard to realize that these are the activities that you must optimize in order to speed up your testing process. You would probably have these at the back of your mind.. Let me help you bring them to the fore! 1. You derive your test cases from a requirement document: We're all faced with this situation each time some piece of code needs to be tested. We look for the requirements, and frame our test scenarios and test cases. Have you ever thought what exactly are you doing by writing test cases? You're simply rewording the requirements! Something simple for example: Requirement = The login page will allow system access to the user by authenticating his/her username and password. The user will be allowed to enter the username and password through text boxes of not more than 20 characters (this does not mean the data type char). Test Case#1 = Verify the presence of ...