IE: Getting test cases from bugs that have been found.... I'm betting many folks do this and I wanted to see if anyone knows of any research/information or has comments.
I'm at a new company and have been tasked to "test the software". The system is complex and involves host(MF), PC, and Palm/PPC components. Currently they have some test plans (spreadsheets) that the support, and one part time tester, go through each release. They also have a bug database.
My plan is to continue with what they have now and at the same time develop better tests and automate everything.
So, here's the "reverse test case engineering" thing...this could actually be something that is out there but hasn't been formally researched, developed or anything. This is also a good way to ramp up in a new QA position. I'm taking the bugs in the database (that I can understand so far) and writing test cases for them. I know that some groups/companies write test cases for every bug that is found and this is basically what I'm doing except it's kinda "3rd party" or "black box" in the fact that I have no details on the exact procedure that came up with the bug (since the bug DB doesn't usually have beginner/newbie/joe shmoe instructions; rather comments that only an expert or developer could work with). So I just use the outcome/results of the bug and figure out how to test for it.
Example: A doctor's tries adding diagnosis from the results of a procedure/test he preformed and the screen locks up. So I start trying this and find it's only a particular way this is done (or a particular diagnosis) that makes this happen. So instead of just testing exactly the way the bug showed up and finding that it doesn't work, I end up creating multiple test case that cover more code.
So anyway, anyone have any comments on this? I'm thinking that many times when "a test case is written for every bug" it is usually so specific (because of the type of bug, because of the bud description, or just because) that it will only catch the one particular thing so this type of test case development is not very relevant.
"Imagination is more valuable than knowledge." (AE)
Re: reverse black box test case engineering and de
That's kind of the concept of Unit testing in XP.
"Everytime you find a bug, write an automated test that detects the bug so that it will always be caught by the automated test suite"
Nice concept in theory [img]/images/graemlins/smile.gif[/img]
But besides that, using defect reports as a source of test ideas is generally a good idea. Here's a suggestion though: instead of trying to create a test for every single defect (that could be a HUGE amount of tests, depending on how many bugs have been found) try looking at the types of bugs that have been found. See if you can group them together (create you own defect taxonomy) and then use that to tailor some tests around known defect areas in your project.