Paul Eggert wrote:
> > #include "unitypes.h"
> >
> > +/* Get bool. */
> > +#include <stdbool.h>
>
> In examples like these would it make sense to do the following instead?
>
> #if !@HAVE_C_BOOL@ && !defined __cplusplus
> #include <stdbool.h>
> #endif
No, this would not work:
* @HAVE_C_BOOL@ tests a property of the C compiler that has built
libunistring. It is unrelated to the compiler that uses <unistr.h>
after it has been installed in public locations.
* As we have just discovered (re Sun C++), 'defined __cplusplus' is not
a guarantee that 'true' is defined correctly. So, even in C++ mode,
it may be a win to include <stdbool.h>. Remember, here we are in a
context where gl_C_BOOL / AC_C_BOOL has not been invoked.
Bruno
