Better Software - November 2007 - (Page 31)

WAW—The application allows you to pop bubble wrap, but questions why you would, and it’s very sneaky, too. WLL—I continue popping. WHW—I pop all the bubbles. I assume the bubble wrap screen will be replaced by something to indicate that I am finished popping bubbles. No, nothing happens. WLL—I click the Quit button, and a meter and count appear indicating that I popped seventy-seven bubbles. I didn’t record the exact number, so I’ll assume it is correct. After a short delay, another window appears inviting me to either play again or to enter three email addresses to which I can send information about this program. WAW—An alternate charter is “investigate the ‘send this to some friends’ function.” I’ll need at least three email addresses to test it. My understanding of the bubble wrap program has expanded. My session to-do list contains “start by not clicking ‘Click to pop!’ icon.” I will try that now. Since I have discovered the program counts the bubbles, there may be something to check there. Edges are often causes of errors, so I might try just popping edge bubbles. WC—I choose to play again. WLL—The initial screen appears with a fresh piece of bubble wrap. I deliberately ignore the “Click to pop!” icon and pop ten half-hidden edge bubbles, and then I cross “start by not clicking ‘Click to pop!’ icon” off my to-do list. WHW—I quit after popping ten bubbles, so I expect the count to be ten. No, the count is eighty-eight. OK, so the count must not clear between bubble wrap sheets. The total is still curious, though, because seventy-seven popped bubbles plus ten popped bubbles equals eighty-seven not eighty-eight. I seem to have found a pattern of failure but should investigate and understand the failure more completely before I report it. The “send this to some friends” box appears quickly after the count is displayed, and it blocks the count and cannot be moved. This may be worth reporting also. WHW—I want to stop rather than play again or send emails. I expect a Stop option, but there isn’t one, nor is there a visible Help option. I could investigate different keystrokes to see if one stops the game or displays Help, but I close the browser window. This may also be worth reporting as a defect. WAW—This application simulates popping bubble wrap. A user can pop any unpopped bubble he chooses or quit at any time. After quitting, a user can email up to three friends, play again, or close the browser window to finish. Apart from a single button, there are no alternate flows during play. The email function was not tested. We’ve discovered a failure pattern in bubble counts that needs further investigation. There are also possible usability bugs with the screen: obscuring the “send this to some friends” box and the lack of a Stop option. This is the end of my first test. TEST 2 WLL—Start the virtual bubble pop application. WHW—I don’t pop any bubbles and click Quit. I expect the popped-bubble count to be zero. No, the bubble count is one. I close the browser window to finish. WAW—This off-by-one bug provides a failure model for the “eighty-seven count that was eighty-eight” behavior I observed. Defect: The count of bubbles popped is one bubble too high for every sheet of bubbles used. This application is quite limited in its functionality, but I seem to have investigated the main function. This is the end of my second test. Note that I use WAW entries to enter initial understandings, consolidate learning, and note defects. TEST 3 I’ll leave it to you to test the email function that sends information about this program to three of your friends. It has a surprising twist. Frequently Asked Questions How can I use the framework? You can use the framework to guide your thinking while you are exploring. My example above is in that style. You also can use it at a higher level to summarize your testing—with less detail. The acronyms serve as milestones to show the main direction you follow in your test. www.StickyMinds.com People reading your test log or trying to reproduce or vary the test can follow the milestones to get a high-level view of your testing. What is the most effective way to use the framework? Tests focused on failure—where models are implemented or interact incorrectly—are the quickest way to measure quality. Originally called error guessing, this skill is often learned subconsciously by testers who recognize patterns of failure from other applications or shared experiences and then search for examples of those defect patterns. Taxonomies or lists of possible defects can suggest areas to investigate for failures. We also can investigate defect clusters. Finding one failure may then suggest others. One reason ET is so efficient at finding defects is that the results of one test can be used to direct subsequent tests. This is important in investigating the defect clusters found in software and shifting our focus from areas where no or few defects are being found to other, more “fertile” areas. Can you learn patterns of failure? Until recently, this skill appeared to be intuitive, one that only testing experience could provide. Now it seems that the basics of this skill can be taught. Part of Alan Jorgenson’s 1999 doctoral thesis at the Florida Institute of Technology identifies four main failure models. Defects occurred when: • An application accepted incorrect input (e.g., a date of birth with a five-digit year, 11/11/19900) • An application stored incorrect information (e.g., storing date of joining company as date of birth) • An application created incorrect output (e.g., trying to display a percentage with only two digits, which probably would display 100 percent incorrectly as 00) • An application performed an incorrect computation based on input or stored data (e.g., trying to perform a percentage calculation when one number is a letter) Based on those failure models, an analysis of approximately 10,000 bugs 31 NOVEMBER 2007 BETTER SOFTWARE http://www.StickyMinds.com

Table of Contents Feed for the Digital Edition of Better Software - November 2007

Better Software - November 2007 Contents MarkYour Calendar Technically Speaking What’s Happening @StickyMinds.com Code Craft Test Connection Management Chronicles The Measure of a Management System Behind the Scenes A Story About User Stories and Test-Driven Development Product Announcements The Last Word Ad Index

Better Software - November 2007

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.