Bruno Haible wrote: > Jim Meyering wrote: >> I wrote these changes some time ago, and >> am going to push them shortly. >> >> +#undef UNUSED_PARAM >> +#ifdef __linux__ >> +# define UNUSED_PARAM _UNUSED_PARAMETER_ >> +#else >> +# define UNUSED_PARAM >> +#endif >> + >> ... >> >> +#undef UNUSED_PARAM >> +#ifdef GETFSTYP >> +# define UNUSED_PARAM >> +#else >> +# define UNUSED_PARAM _UNUSED_PARAMETER_ >> +#endif >> + > > Now this gets really ugly. Not only because a macro with the same name is > being > undefined and redefined repeatedly, but also because the #if conditions for > this macro must be the same as the #if conditions inside the function's body. > When they get out of sync, the warning will reappear. > > The other way to silence this warning is a statement > (void)param; > in the function's body. (Or was there some problem with that? I just tested > gcc 2.95.3 to 4.3.2, and in all versions such a void cast makes the warning > disappear.) > > Bottom line: Whenever the _UNUSED_PARAMETER_ macro cannot be used > unconditionally, I would use a cast to void.
I have to agree. In that case _UNUSED_PARAMETER_ is really misnamed, and would have been better as _MAYBE_UNUSED_. For the case where you know parameters will be unused due to a #define for e.g. I think it's better to cast to (void) in the function body as you say. cheers, Pádraig.
