🚀 go-pugleaf

jfw@ksr.com (John F. Woods) writes:

>Losing the compile-time diagnostic is regrettable,
>but note that there is no diagnostic required by the following:

>	#include <string.h>
>		*strchr("hammer me", ' ') = 0;

But this argument uses one botch in the C definition to justify another
proposed botch.  If C had C++-style overloaded functions (or better
yet, polymorphic functions), there would be no problem, because there
could be two versions of strchr:

	char *strchr(char *, int);
	char const *strchr(char const *, int);

and the above error would be caught at compile-time (assuming string
constants were changed to be of type char const *, not char *).  Since
C lacks such features, it must botch strchr's declaration; but that's no
excuse for also botching the semantics of pointer assignment.