> On 2016-10-28 13:02, Quentin Rameau wrote: > > On Fri, Oct 28, 2016 at 12:45:14PM +0000, Ali H. Fardan wrote: > >> actually, imo, I think > >> usage() should return success. > > Surely not. > > The call to usage() is made when wrong options have been passed to > > the tool, you wouldn't return “no error” code when there actually > > has been an error. > > How could a script test if 'foo' is compiled correctly and runs > without any missing shared libraries? if such a thing existed it will > return 1> else, 0 is returned, the program still works fine, you just > didn't give it the right arguments. You're clearly on the wrong path for what you want to test here.
A script wouldn't test on runtime if there has been some compile-time errors. That's not for the tool to tell you if it has been correctly built or not, its behaviour is expected to be built correctly. If not, that's your fault (or the packager you trust do to it for you, actually again your responsability).
