Oh ok, my bad, now I've figured out how it should work. > I had a look at the -o option, but I'm not sure how to use it. For example, > this is the only way I can get gcov to run: > > $ gcov tests/tests.cpp -o tests/tests-tests You have to run gcov in the same directory as the .cpp file, otherwise it gives an error message about being unable to find it (even with a path.) This works: $ cd tests; gcov tests.cpp -o tests-tests Likewise for other files: $ cd tests; gcov blah.cpp -o tests-blah > I'm also a bit confused about why I should supply a .cpp filename instead of a > program name. How does it know which program to run? If I run "gcov > string.cpp" how does gcov know that it needs to run some other program with > certain parameters to actually execute the code in that file? Now I also realise I'm supposed to run the program normally *first* to generate the .gcda files, *then* gcov is used to convert them into human-readable text files with the commands above. Now it seems to work as advertised :-) Cheers, Adam.
