Mail Archives: djgpp/1997/10/31/15:18:59
On Tue, 28 Oct 1997, Eli Zaretskii wrote:
>
> On Sun, 26 Oct 1997, Peter Berdeklis wrote:
>
> > > Compiler gives warning, that
> > > "comparison is always 0 due to limited range of data type"
> > > (no matter with or without '-funsigned-char' flag)
> >
> > This appears to be a bug in DJGPP. Compiled under gcc/SGI there is no
> > complaint.
>
> Since when printing a warning is deemed a bug in the compiler? The
> warning here is perfectly valid, since it pinpoints a real bug in the
> code (which assumed '\xFD' is an unsigned quantity), so gcc did a real
> service here.
>
> I don't know much about SGI, but it could be that the default
> character type there is different from x86. Or it could be a bug in
> the SGI version of gcc.
I apologize, Eli's right of course.
The SGI has -funsigned-char as default. Also, since I created my test
program as C++ code the unsigned char affects the '\xfd', because under
C++ character constants are chars, not ints as for C. When I performed
the same test under DJGPP, using C++ and -funsigned-char, the warning
disappeared.
---------------
Peter Berdeklis
Dept. of Physics, Univ. of Toronto
- Raw text -