07.01.2013 Views

Lecture Notes in Computer Science 3472

Lecture Notes in Computer Science 3472

Lecture Notes in Computer Science 3472

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Summary<br />

14 Tools for Test Case Generation 433<br />

AutoL<strong>in</strong>k is an (<strong>in</strong>dustrial strength) test generator to generate (readable) TTCN<br />

test suites from SDL specifications. The test suite generation is guided by test<br />

purposes that can be supplied by the user, or also generated fully automatically.<br />

Unfortunately, a theoretical underp<strong>in</strong>n<strong>in</strong>g of the algorithms used <strong>in</strong> AutoL<strong>in</strong>k<br />

was not present <strong>in</strong> the papers we studied. Fortunately, it turned out to be possible<br />

to reverse eng<strong>in</strong>eer the conformance relation def<strong>in</strong>ition for AutoL<strong>in</strong>k [Gog01].<br />

AutoL<strong>in</strong>k has been used <strong>in</strong> a number of case studies.<br />

Related Papers<br />

• Tool Overview: [KGHS98, SEG00]<br />

• Related Tools: [GSDH97]<br />

• Underly<strong>in</strong>g Theory: [Gog01]<br />

14.3 Comparison<br />

Many aspects can be addressed when compar<strong>in</strong>g tools. Below we name just a<br />

few, grouped by separat<strong>in</strong>g theoretical aspects from more practical ones.<br />

• Theoretical aspects<br />

– Are the test generation algorithms based on a sound theory? How do<br />

these theories relate to each other?<br />

– Which error-detect<strong>in</strong>g power can be achieved theoretically?<br />

– What is the time/space complexity of the underly<strong>in</strong>g algorithms?<br />

– Is the theory suited for compositional issues? Can models be congruently<br />

composed?<br />

– Is the theory suited for distributed issues? Is it possible to generate<br />

several distributed testers or is only a monolithic one possible?<br />

– How is data handled by the formalisms? Is the theory restricted to simple<br />

sets of actions or is there support for complex/symbolic data, e.g. <strong>in</strong>f<strong>in</strong>ite<br />

doma<strong>in</strong>s? How is this handled?<br />

– Is there a notion of time? Is it possible to guarantee time constra<strong>in</strong>ts<br />

dur<strong>in</strong>g the test execution (which is necessary for real time systems)?<br />

– Can it deal with nondeterm<strong>in</strong>istic SUTs, or only with determ<strong>in</strong>istic ones?<br />

– Can it deal with non nondeterm<strong>in</strong>istic specifications?<br />

• Practical aspects<br />

– Which error-detect<strong>in</strong>g power can be achieved practically (case studies)?<br />

– Is it only possible to generate test suites or also to execute them on a<br />

real SUT?<br />

– How user-friendly is the tool? Is there a GUI facilitat<strong>in</strong>g the usage? Are<br />

graphical models used (e.g. UML)?<br />

– Which are the supported test case specifications?<br />

– How difficult is it to create a suitable <strong>in</strong>put (e.g. def<strong>in</strong><strong>in</strong>g test purposes)?<br />

Are many parameters needed and does the tool help <strong>in</strong> sett<strong>in</strong>g them?

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!