What happens if a use-case fails use-case testing?

Question

I've not been able to find this question anywhere on the Internet, and since i'm working on my exam project at the moment, I would hope for this to get answered.

What happens to a use-case if it fails the boss/EBP/size tests? Do you throw it out, rewrite it, does it become a non-functional requirement?

Or does the testing simply prove whether or not a use-case has high business/risk/architectural value for the rest of the project? And you should down prioritize a use-case which failed tests (if only just one test)?

Thanks for your time. Craig Larman's book doesn't really go into depth about this, as far as I know.

Answer 1

I assume you refer to http://www.craiglarman.com/wiki/downloads/applying_uml/larman-ch6-applying-evolutionary-use-cases.pdf .

does it become a non-functional requirement?

In the vast majority of cases, use cases represent functional requirements, thus it's pretty unlikely that you refactor them into nonfunctional ones.

If you can rewrite them and they're likely to pass the test, then probably you should. If you can't achieve a rewrite, try to find a real value in it; you might find the information useful in a subsequent modeling step. If you don't see that happening, put it to the trash...

And you should down prioritize a use-case which failed tests (if only just one test)?

If you stick with the example of Move Piece on Game Board that fails the size test, you'd probably be better of putting that in an activity diagram or including it into the documentation of the corresponding use case that this step actually belongs to.

Referring to the boss test mentioned by Larman, the "Log in" could become a precondition to other use cases. You might also join several to a use case with more measurable value. As the author mentions, the test may fail but the use case can still be of a value (see Reasonable Violations).

So at least this is what common sense suggests. There might be rules in RUP (= don't know that) which suggest otherwise.

Answer 2

If a usecase under consideration fails Boss Test (means this usecase is not valuable for boss)

Then it might stand as a Common Sub Function (as it is a sub function of all major other usecases

If you ask me i dont formalize any name for assigning Priority to usecases in order to consider them for a proper testing.

These are just names assigned to Priority Levels.