Better Software - April 2008 - (Page 19)

Test Connection “Wait. That’s too many numbers,” she said, clearly frustrated. I said, “The only things that work in this place are the people. I really do appreciate that you’re struggling here.” She smiled. “What about A434-300?” She tried it. “Nope. Six numbers is right, but it can’t start with a letter.” I looked all over the item. There was a number on it—but only four digits. “What about dropping the A?” I suggested. “434-300?” She tried backspacing, but that displayed only equal signs in the text field. Shift-backspace allowed her to backspace over the whole field, and she typed in the number. “Ah, that works. Shelf bracket. A dollar ninety-seven.” That was a lot of work for a twodollar part, and the system didn’t do very much to help. Why are there two numbers on the bin label? Why is neither number identified? Why does the software accept neither and reject both? Why isn’t the acceptable number printed on the label? When I mistype something in the Google search window that doesn’t match an entry in its database, Google offers a plausible suggestion— a close-but-not-exact match. Couldn’t the store’s system have tried a database search based on the numeric portion of the number? Couldn’t it have crossreferenced one catalog with another to offer at least a choice or a guess as to what the product might be? Meanwhile, the clerk had an idea about the correct number format, but not how to deal with an incorrect number. Had management realized this as a training issue? I’ll guess that when the developers for this product designed, built, and tested it, they thought in terms of confirming that it worked. I’ll guess that when they tested it, they used all kinds of boundary testing and field validation checks to make sure that the software accepted properly formatted numbers and rejected improperly formatted numbers. Those things are important. But I’ll also guess that they tested the software in isolation from the environment in which it was intended to operate, and that they looked for functional correctness with- out testing the product in the field and asking, “Is there a problem here?” And finally, I’ll guess that, to this day, they don’t know about the frustrations that customers and cashiers alike are having with the system. Systems and software aren’t just about correctness; they’re also about solving problems for people. One principle of the context-driven software testing movement says that if the problem isn’t solved, the product doesn’t work. Testers: Could we find more bugs—and more important bugs—in our systems by observing something other than the software itself? And managers: What problems could we prevent by letting our testers see more than some screens and some specifications? {end} How would you answer the questions above? What additional questions should we be asking? Follow the link on the StickyMinds.com homepage to join the conversation. PNSQC 2008 Collaborative Quality World-class quality does not happen in a vacuum. Agile-inspired collaboration spans levels, disciplines, and industries. 2008 keynote speakers Ron Jeffries and Chet Hendrickson will address Quality Dynamics in Agile Software Development. October 13 -15, 2008 Portland, OR www.pnsqc.org www.StickyMinds.com APRIL 2008 BETTER SOFTWARE 19 http://www.StickyMinds.com http://www.pnsqc.org http://www.pnsqc.org http://www.StickyMinds.com

Table of Contents Feed for the Digital Edition of Better Software - April 2008

Better Software - April 2008 Contents Mark Your Calendar Contributions eLightenment Technology Speaking - A Change Would Do You Good Code Craft - A "D" In Programming, Part 1 Test Connection - Learning the Hardware Lessons Management Chronicles - The Art of Persuading Management Cover Story - Incremental and Iterative Development Developers...Start Your Engines Where Do I Go From Here Product Announcements 10 Things You Might Not Know About... The Last Word - Software Quality and the Prisoner's Dilemma Ad Index

Better Software - April 2008

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.