On Mon, Mar 27, 2017 at 12:18:17PM +0200, Dominique d'Humières wrote:
> > > If he added a new option affecting libgfortran, then he should
> > > fix up libgfortran.
> >
> > He didn't add the warning to specifically annoy fortran developers.
> > It is trivial to add seven gcc_fallthrough() or breaks for someone who
> > knows the code and the person who added the warning obviously doesn't.
> The following patch fixes the warnings
>
> --- ../_clean/libgfortran/io/list_read.c 2017-03-25 20:42:40.000000000
> +0100
> +++ libgfortran/io/list_read.c 2017-03-27 12:06:10.000000000 +0200
> @@ -51,7 +51,8 @@ typedef unsigned char uchar;
> #define CASE_DIGITS case '0': case '1': case '2': case '3': case '4': \
> case '5': case '6': case '7': case '8': case '9'
>
> -#define CASE_SEPARATORS case ' ': case ',': case '/': case '\n': \
> +#define CASE_SEPARATORS /* Fall through. */ \
> + case ' ': case ',': case '/': case '\n': \
> case '\t': case '\r': case ';'
>
> /* This macro assumes that we're operating on a variable. */
>
> Indeed before applying this patch, someone will have to check that the
> warnings do not occur because of missing breaks.
>
> Note that putting /* Fall through. */ before the use of the macro
> CASE_SEPARATORS does not work. Is it a (known) bug?
Yes, it's known that the "falls through" comments don't work if they're
preceding a macro.
Marek
