From owner-freebsd-current  Wed Mar 26 16:54:42 1997
Return-Path: <owner-current>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id QAA18006
          for current-outgoing; Wed, 26 Mar 1997 16:54:42 -0800 (PST)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id QAA18000
          for <current@FreeBSD.org>; Wed, 26 Mar 1997 16:54:38 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.5/8.6.9) id LAA00483; Thu, 27 Mar 1997 11:50:21 +1100
Date: Thu, 27 Mar 1997 11:50:21 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199703270050.LAA00483@godzilla.zeta.org.au>
To: bde@zeta.org.au, imp@village.org
Subject: Re: make world fails on libmytinfo
Cc: current@FreeBSD.org, helbig@MX.BA-Stuttgart.De
Sender: owner-current@FreeBSD.org
X-Loop: FreeBSD.org
Precedence: bulk

>: >  /* for quit.c */
>: >  extern int sys_nerr;
>: >! #if __FreeBSD__ != 2 && __FreeBSD__ != 3
>: >  extern char *sys_errlist[];
>: >  #endif
>: >  extern char *prg_name;
>: 
>: It is just harmful.  Const'ness of sys_errlist has nothing to do with
>: FreeBSD.
>: 
>: BTW, sys_nerr is also const in FreeBSD, but the type mismatch is only
>: warned about if gcc is invoked with -pedantic.
>
>And it is even esier to fix.  Most modern systems have errno.h.  Just
>include that.  Since it is in our CVS tree, that seems will be
>acceptible.

errno.h is the problem, not the solution.  It pollutes the application
namespace by declaring sys_nerr and sys_errlist.  libmytinfo is chummy
with the implementation in using (*) these variables instead of the public
interface (strerror()), but not chummy enough to get the declarations
right (except using an ifdef hack).

(*) libmytinfo doesn't actually use these variables.  It just miscdeclares
them.  It claims to use them ``/* for quit.c */'', but quit.c isn't in
FreeBSD.

Bruce