🚀 go-pugleaf

In article <9211301427.AA23744@platine.edps.bull.fr> T.CAmbon@frcl.bull.FR (Thierry Cambon) writes:
>
>I had epoch 4.2 compiled on a Zenith Z-Server running SCO 2.0 r3.2 (i386)
>without any problem. when I run epoch on this server with the console display,
>it works fine, but when I use another display (even on a Z-Server running the same
>version of SCO) I get the Xlib warning:
>
>Xlib:  warning, client built for newer rev (11) than server (0)!
>Xlib:  warning, client is protocol rev 0, server is rev 9216!
>
>Then I tried to link temacs with older versions of libX11.a (from SCO 1.1 and SCO 1.0)
>and I got either the same warnings or the following:
>
>Warning: Client built for newer server!
>Warning: Protocol rev. of client does not match server!
>
>
>I used the files config/s-sco.h and config/m-intel386.h in config.h.
>


I'm not sure if you mean that you are running ODT 2.0 - I hope that
the following stuff will help (I'm currently running unix 3.2v4 with
X-Sight 2.2/2.3, TLS001 X11R4 Server and lng311b X11R4 libraries)

config/s-sco.h:
____________________________________________________________________________
#include "s-usg5-3.h"

#define sco
#define X11R4

/* Use termios instead of termio, to turn off C-z.  */

#define HAVE_TCATTR

/* It's possible for Emcs to suspend itself.  */

#undef NOMULTIPLEJOBS

/* The setsid system call exists.  */

#define HAVE_SETSID

/* The rename system call exists.  */

/* #define HAVE_RENAME */

/*
 *	Define HAVE_TIMEVAL if the system supports the BSD style clock values.
 *	Look in <sys/time.h> for a timeval structure.
 */

#define HAVE_TIMEVAL
#define USG_SYS_TIME
#define USE_UTIME
#define HAVE_GETTIMEOFDAY

/*
 *	Define HAVE_PTYS if the system supports pty devices.
 */

/* Some versions of V.3 have this, but not all.  */

#define HAVE_PTYS
#define SYSV_PTYS

#define HAVE_SOCKETS
#define HAVE_SELECT
#define select XSelect
#define BROKEN_FIONREAD

/* Include ptem.h, not sioctl.h.  */

#define NO_SIOCTL_H
#define NEED_PTEM_H

#define POSIX_SIGNALS
#undef sigsetmask
#undef HAVE_SYSV_SIGPAUSE

/* Inhibit macro definition of `signal' in m-intel386.h.  */

#define DONT_DEFINE_SIGNAL

/* We cannot get alloca from -lPW because various other symbols
   in that library conflict with symbols in GCC.  */

#ifdef	__GNUC__
#define alloca(x) __builtin_alloca (x)
#else
#define C_ALLOCA
#undef HAVE_ALLOCA
#define STACK_DIRECTION	-1
#endif /* __GNUC__ */

/* TIOCWINSZ doesn't work on ptys.  */
#define BROKEN_TIOCGWINSZ

#define BSTRING

#undef LIBX11_SYSTEM
#define LIBX11_SYSTEM -lsocket -lmalloc

#define XBACKWARDS
#define HAVE_RANDOM
____________________________________________________________________________

I also changed the position of #define STACK_DIRECTION in m-intel386.h
since it wasn't being defined, except for XENIX.

I recall that adding NEED_PTEM_H breaks process.c and sysdep.c, but it
does work - I just hacked the conditional bits to fix it.

This gives me a dump-able epoch-4.2 for this system - I think that ODT
2.0 should have shared X11R4 libraries which may break things again (I
have never succeeded in linking with libc_s.a). You should still be
able to get a working epoch, but without the pre-loaded lisp.

-- 
Barrie Spence					barrie@calvin.demon.co.uk
						Telephone +44 (0) 506 442304