Googling will give you a lot of good info, as will browsing Borders and/or Barnes & Nobles. However, there are a few good rules of thumb I've seen after supporting a test team for a year:
- Test the design first - make sure the design is sane. This includes the application, the database, the physical setup, the user interface, etc.
- Nail down the specs - if you don't know what you're testing to, you don't know if you passed or not.
- A corollary to the last point is that you shouldn't accept a spec if you can't test it. Remember, it's your responsibility to make sure that your code meets the spec. If you can't test a feature, you can't discharge your responsibility.
- Don't test application functionality, test user functionality. Remember, the users don't care if you have a neat-o-keen sorting procedure. They just want to know that their data returns in the order they asked for / were expecting.
- Test both coding units and logical units. You have to make sure that a given subroutine works, but you also have to make sure that a given feature works, too.
- Don't just test functionality - you want to test under different scenarios. Light load, heavy load, different types of users, etc.
- Automate your testing as much as possible. Ideally, you'll be running thousands of tests every time you change. It's a lot easier to have the computer do the grunt work for you.
A lot of these suggestions assume that you have a rational codebase. For example, if you cannot extract a given subroutine from its context and test it, you very well could have a problem.
Also, your testing design
------
We are the carpenters and bricklayers of the Information Age. The idea is a little like C++ templates, except not quite so brain-meltingly complicated. -- TheDamian, Exegesis 6 Please remember that I'm crufty and crochety. All opinions are purely mine and all code is untested, unless otherwise specified. | [reply] |
