From owner-freebsd-current  Fri Apr  7 05:38:47 1995
Return-Path: current-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id FAA05295
          for current-outgoing; Fri, 7 Apr 1995 05:38:47 -0700
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.34])
          by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id FAA05284
          for <current@FreeBSD.org>; Fri, 7 Apr 1995 05:37:58 -0700
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id WAA25471; Fri, 7 Apr 1995 22:35:35 +1000
Date: Fri, 7 Apr 1995 22:35:35 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199504071235.WAA25471@godzilla.zeta.org.au>
To: current@FreeBSD.org, pete@pelican.com
Subject: Re: more ctype woes
Sender: current-owner@FreeBSD.org
Precedence: bulk

>In reference to my problem with ctype - defining _USE_CTYPE_CLIBRARY_
>fixed the compile problems and gave rise to:
>gcc -o geod geod.o geod_setup.o geod_forwd.o geod_invrs.o emess.o libproj.a -lm
>geod.o: Undefined symbol `___isctype' referenced from text segment
>gmake: *** [geod] Error 1

>i.e. the relevant entry (there is a reference in libc.so to isctype.so
>but not ___isctype) isn't in the C library...

I committed fixes for this a few minutes ago.

___isctype wasn't defined because libc/locale/nomacros.c didn't understand
the convolutions in <ctype.h>.  It needed to define _USE_CTYPE_CLIBRARY_
to stop <ctype.h> from declaring the inline versions and it shouldn't have
tested either _USE_CTYPE_INLINE_ or _USE_CTYPE_MACROS_ to decide whether
to generate code.

I replaced _USE_CTYPE_CLIBRARY_ and _ANSI_LIBRARY by
_EXTERNALIZE_CTYPE_INLINES_ and another layer of macros for toupper()
and tolower().  Now all the ctype "functions" are implemented as
macros (which usually expand to inline functions) so that both the
library and applications can #undef them to work with the functions.

Bruce