What’s our job as testers? Reporting bugs, right?
When I first started reading about Session-Based Test Management, I was intrigued by the session sheet. At the top, there’s a bunch of metadata about the session—the charter, the coverage areas, who did the testing, when they started, how long it took, and how much time was spent on testing versus interruptions to testing. Then there’s the meat of the session sheet, a more-or-less free-form section of test notes, which include activities, observations, questions, musings, ideas for new coverage, newly recognized risks, and so forth. Following that, there’s a list of bugs. The very last section, at the bottom of the sheet, sets out issues.
“What’s an issue?” I asked James. “That’s all the stuff that’s worth reporrting that isn’t a bug,” he replied. Hmmm. “For example,” he went on, “if you’re not sure that something is a bug, and you don’t want to commit it to the bug-tracking system as a bug, you can report it as an issue. Say you need more information to understand something better; that’s an issue. Or you realize that while you’ve been testing on one operating system, there might be other supported operating systems that you should be testing on. That’s an issue, too.”
That was good enough as far as it went, but I still didn’t quite get the idea in a comprehensive way. The information in the “test notes” section of session sheet is worth reporting too. What distinguishes issues from all that other stuff, such that “Issues” has its own section on parallel with “Bugs”?
Parallelism saves the day. In the Rapid Software Testing class, we teach that a bug is anything that threatens the value of the product. (Less formally, we also say that a bug is something that bugs somebody… who matters.) At one point, a definition came to me: if a bug is anything that threatens the value of the product, an issue is anything that threatens the value of our testing. In our usual way, I transpected on this with James, and we now say that an issue is anything that threatens the value of the project, and in particular the test effort. Less formally and more focused on testing, an issue is anything that slows testing down or makes it harder. If testing is about making invisible problems visible, then an issue is anything that gives problems more time or more opportunities to hide.
When believe that we we see a bug, it’s because there’s an oracle at work. Oracles—those principles or mechanisms by which we recognize a problem—are heuristic. A heuristic helps us to solve a problem or make a decision quickly and inexpensively, but heuristics aren’t guaranteed to work. As such, oracles are fallible. Sometimes it’s pretty clear to us that we’re seeing a bug in a product, the program seems to crash in the middle of doing something. Yet even that could be wrong; maybe something else running on the same machine crashed, and took our program down with it. A little investigation shows that the product crashes in the same place twice more. At that point, we should have no compunction reporting what we’ve seen as a bug in the product.
An issue may be clear, or it may be something more general and less specific. A few examples of issues:
- As you’re testing, you see behaviour in the new version of the product that’s inconsistent with the old version. The Consistency with History oracle tells you that you might be seeing a problem here, yet one could make the case that either behaviour is reasonable. The specification that you’re working from is silent or ambiguous on the subject. So maybe you’ve got a bug, but for sure you have an issue.
- While reviewing the architecture for the system, you realize that there’s a load balancer in the production environment, but not in the test environment. You’ve never heard anyone talk about that, and you’re not aware of any plans to set one up. Maybe it’s time to identify that as an issue.
- You sit with a programmer for a few minutes while she sketches out the structure of a particular module to help identify test ideas. You copy the diagram, and take notes. At the end of the meeting, you ask her to look the diagram over, and she agrees that that’s exactly what she meant. You reflect on it for a while, and add some more test ideas. You take the diagram to another programmer, one who works for her, and he points at part of the diagram and says, “Wait a second—that’s not a persistent link; that’s stateless.” You’ve found disagreement between two people making a claim. Since the code hasn’t been built for that feature yet, you can’t log it as a bug in the product, but you can identify it as an issue.
- As you’re testing the application, a message dialog appears. There’s no text in the dialog; just a red X. You dismiss the dialog, and everything seems fine. It seems not to happen again that day. The next day, it happens once more, in a different place. Try as you might, you can’t replicate it. Maybe you can report it as an intermittent bug, but you can definitely record it as an issue.
- A steady pattern of broken builds means that you wait from 10:00am until the problem is fixed—typically at least an hour, and often three or four hours. Before you’re asked, “Why is testing taking so long?” or “Why didn’t you find that bug?” report an issue.
- You’ve been testing a new feature, and there are lots of bugs. 80% of your session time is being spent on investigating the bugs and logging them. This has a big impact on your test coverage; you only got through a small subset of the test ideas that were suggested by the session’s charter. The bugs that you’ve logged are important and you can expect to be thanked for that, but you’re concerned that managers might not recognize the impact they’ve had on test coverage. Raise an issue.
- You’re a tester in an outsourced test lab in India. Your manager, under a good deal of pressure himself, instructs you to run through the list of 200 test cases that has been provided to him by the clueless North American telecom company, and to get everything done within three days. With practically every test you perform, you see risk. All the tests pass, if you follow them to the letter, but the least little experimentation shows that the application shows frightening instability if you deviate from the test steps. Still your boss insists that your mission is to finish the test cases. He’s made it clear that, for the next three days, he doesn’t want to hear anything from you except the number of tests that you’ve run per day. Do your best to finish them on schedule, but sneak a moment here and there to identify risks (consider a Moleskine notebook or an ASCII text file). When you’re done, hand him your list of bugs—and in email, send him your list of issues.
- You’re a tester in a small development shop that provides customizable software for big banks. You have concerns about security, and you quickly read up on the subject. What you read is enough to convince you that you’re not going to get up to speed soon enough to test effectively for security problems. That’s an issue.
- As a new tester in a company, you’ve noticed that the team is organized such that small groups of people tend to work in their own little silos. You can point to a list of a dozen high-severity bugs that appear to have been the result of this lack of communication. You can see the cost of these twelve bugs and the risk that there are more lurking. You recognize that you’re not responsible for managing the project, yet it might be a good idea to raise the issue to those who do.
Those are just a few examples. I’m sure you can come up with many, many more without breaking a sweat.
Teams might handle issues in different ways. You might like to collect an issues list, and put on a Big Visible Chart somewhere. Someone might become responsible for collecting and managing the issues submitted on index cards. Some issues might end up as a separate category in the bug tracking system (but watch out for that; out of sight, out of mind). Still others might get put onto the project’s risk list.
Some issues might get handled by management action. Some issues might get addressed by a straightforward conversation just after tomorrow morning’s daily standup. Someone might take personal responsibility for sorting out the issue; other issues might require input and effort from several people. And, alas, some issues might linger and fester.
When issues linger, it’s important not to let them linger without them being noticed. After all, an issue may have a terrible bug hiding behind it, or it may slow you down just enough to prevent you from finding a problem as soon as you can. Issues don’t merely present risk; they have a nasty habit of amplifying risks that are already there.
So, as testers, it’s our responsibility to report bugs. Even more importantly, it’s our responsibility to raise awareness of risk, by reporting those things that delay or interfere with our capacity to find bugs as quickly as possible: issues.