On 8 Nov 2012, at 10:23, Akim Demaille wrote: > Le 8 nov. 2012 à 10:12, Hans Aberg a écrit : > >>> fprintf (stderr, file_name != NULL ? "%s:%d: " : " ", >>> file_name, line_number); >> >> It is a dynamic check which cannot be verified statically, so the clang >> warning is perhaps inappropriate as default. > > No, it's really static: it can perfectly see what both possible formats are, > so it may, indeed, warn that in one of the cases the arguments are not used. > I would bet that this warning would be addressed if GCC made similar > warnings. But in the current state of affairs, I don't feel like lobbying to > get patches applied in this piece of code.
Since it is legal C, there should be no default diagnostics, only if there has been a flag set, which probably GCC can do.
