dspoon@ncratl.Atlanta.NCR.COM (dspoon) (04/14/89)
We are interested in finding a code coverage analysis tool to perform code coverage analysis on our C++ code, which will be compiled using Zortech's C++ Compiler. Thus the tool must work under MS-DOS. I wonder if code coverage analysis for C++ differs from that for C code...in C, functions are "stand-alone entities", and if you exercise the limits of defined input parameters, you've probably got a "safe" routine or module. But in C++, is it also necessary to ensure that inherited methods perform as expected? Seems like the language (and OOP philosophy) stresses that a tested base class is solid and can be built upon without further need to confirm inherited methods in derived classes. But if you "happen to have" friends, protected data, and other small violations of encapsulation, it does seem you'd want to confirm inherited methods. Another area that I think needs attention on the topic is inline functions...seems like a C++ code coverage analysis tool would need to know if the compiler actually gave you an inline method or not. Inlines can be "covered" as they occur in the code, but !inlines (i.e., converted to functions) fall under the same requirements as the rest of a class's methods. You got any thoughts you'd care to add? -------------------------------David Witherspoon------------------------------- D.Witherspoon@Atlanta.NCR.COM | "It's not the code that counts... NCR Sys Engineering-Retail/Atlanta | it's the attitude!" MY OPINIONS...ALL MINE!!! | - me
mhyman@hsfmsh.UUCP (Marco S. Hyman) (04/15/89)
In article <352@ncratl.Atlanta.NCR.COM> dspoon@ncratl.Atlanta.NCR.COM (dspoon) writes: > We are interested in finding a code coverage analysis tool to perform > code coverage analysis on our C++ code, which will be compiled using > Zortech's C++ Compiler. Thus the tool must work under MS-DOS. Look at a DOS profiler called Inside! (I think that's the name) by a company called Paradigm Systems. I think they are located someplace back east. I haven't used this product but saw it demo-ed at a show and was impressed. The profiler reads the map file and can output line numbers visited or not visited. The output isn't fancy (ascii text scrolling off the screen) but use of the ``lines not visited'' option could be massaged into a test coverage report easily. Since it uses the link map I believe it will work with any compiler that either uses the DOS linker or has the same output. --marc -- //Marco S. Hyman //UUCP: ...!sun!sfsun!hsfmsh!mhyman //Domain: sfsun!hsfmsh!mhyman@sun.com