7 replies
White-box testing is defined by IEEE as structural testing.
Basically, you're attempting to test the execution of the code. Some types of white-box testing are: branch testing, path testing, and statement testing. Each of these testing types are designed to test specific structural aspects of a program.
I don't know how "manual" you can be with this type of testing. I'm assuming that "manual" means by way of a user-interface. Some types of White-box testing can be accomplished by writing a program to test and validate functions and their returns; but, usually it is done by tracing through the code and validating the contents of variables and the return codes of functions.
Of course, it's pretty difficult test each and every possible execution path in a program. So, you'll probably want to narrow it down to some specific common usage scenarious (or use-cases, if you have them).
by jkonair
by realgem
With this testing, you are not worried about testing the requirements so much as the function of the module (of course the module must support some requirement).
Your tests will be based on the intended module function. What must it do? Base your test cases on those functions.
You have to decide how complete your testing will be. Are you going to test every possible path through the code? Every trivial case (empty files/arrays) or extreme case (more than peak data, etc.)
Once youknow this, it's just a matter of running through your checklist. If it's a batch module you're testing, then you may have to set up relevant test data each time you want to run a test. If possible, I would suggest setting up your data once, then perform a backup. Prior to each subsequent round of testing, restore from the backup. That way, you're running the same test each time. Hopefully, you don't consider the "backup utility" as a tool - I think you were referring to a testing tool likeWinRunner.
by jkonair
by limbo
It can be done.
One ancient method was to output to a log the excuted line numbers.
In procedural programming, to can do the same. Just output the procedure names whenver executed.
Now here's the nasty bit, what use is a huge log of procedure names or line numbers? You are likely to need a lot of extra effort to create the log in first place, and even more so to interpret the results.
Wouldn't it be simplier to have a Robot do this? Investigate Rational Robot from at www.Rational.com.
If this is out of your price range, investigate other tool vendors. Otherwise, you will be spending a HUGE amount of time covering this field, an untimately find that anything you create in this instance will be useless when you have to repeat it in the next project.
by jkonair
by jkonair
This discussion is closed to new replies.
Trending Web Development posts
Sort by Trending