Path: utzoo!utgpu!watserv1!watmath!att!linac!uwm.edu!zaphod.mps.ohio-state.edu!rpi!sci.ccny.cuny.edu!phri!cmcl2!adm!smoke!gwyn
From: gwyn@smoke.brl.mil (Doug Gwyn)
Newsgroups: comp.bugs.sys5
Subject: Re: Annoying prototypes for execv* functions in V.4 <sys/unistd.h>
Message-ID: <15028@smoke.brl.mil>
Date: 31 Jan 91 17:58:58 GMT
References: <3598@lupine.NCD.COM>
Organization: U.S. Army Ballistic Research Laboratory, APG, MD.
Lines: 21

In article <3598@lupine.NCD.COM> rfg@lupine.ncd.com (Ron Guilmette) writes:
>I'm fairly sure that these declarations should be changed to:
>extern int execv(const char *, const char *const *);
>extern int execve(const char *, const char *const *, const char *const *);
>extern int execvp(const char *, const char *const *);

Unfortunately there is no "right" way to do this.  Pointer-to-const has
special meaning in function parameter declarations, not quite in line
with the meaning in other forms of declaration.  In particular, type
compatibility checking will approve of passing a pointer to non-const
at the "first" level, but not at the second level of referencing.  Thus
the common usage
	char *args[] = { "prog", "arg1", NULL };
	execv( args[0], args );
would cause an "incompatible types for second argument" diagnostic,
were <unistd.h> (what is this "sys/" prefix?) to provide the additional
"const"s that you suggest.  You can legitimately consider this a
technical deficiency in the C standard if you wish; we simply couldn't
figure out an acceptable way to specify what one would really want for
this and therefore just fixed up the spec for the top level of
referencing.