view src/s/aix4-1.h @ 802:19dfb459d51a

[xemacs-hg @ 2002-04-03 10:47:37 by ben] fix tty problem et al internals/internals.texi: Add section on correctly merging a branch back into the trunk. console-tty.c, eval.c, event-unixoid.c, file-coding.c, file-coding.h, lisp.h, print.c, sysdep.c: Fix data corruption error in print.c from print_depth becoming negative. Borrow primitives internal_bind_int, internal_bind_lisp_object from my stderr-proc ws, soon to be integrated; use them to properly bind print_depth et al. First fix for TTY problem. The basic problem is I switched things so that the TTY I/O is filtered through a coding system, for the support of kterm and such, that speak JIS or similar encodings. (#### I ***swear*** I had this working way back in 19.12.) Anyway, this introduced buffering issues, in which instead of one char being read, it tried to read 1024 chars. I tried setting the stdin descriptor non-blocking, but it doesn't appear to work on Cygwin. (#### Andy, do you know anything about this?) So I fixed it elsewhere. If you get weirdness on the TTY, look in console-tty.c and see how it gets the coding system; maybe there's a way to change it (and if not, there should be!). Also fix warning in sysdep.c.
author ben
date Wed, 03 Apr 2002 10:47:52 +0000
parents 3ecd8885ac67
children
line wrap: on
line source

/* Synched up with: FSF 19.31. */

#define AIX4_1

#include "aix4.h"

#ifdef __GNUC__
#undef _NO_PROTO
#endif

/* For AIX, it turns out compiling emacs under AIX 3.2.4 REQUIRES "cc -g"
   because "cc -O" crashes.  Under AIX 3.2.5, "cc -O" is required because
   "cc -g" crashes. Go figure.  --floppy@merlin.mit.edu.
   4.1 seems to need -g again. -- larry@vaquita.mitra.com.  */
/* David Edelsohn <dje@watson.ibm.com> says that this actually depends
   on the version of XLC, which can't be predicted from the system version.
   What a mess!  */
#ifndef __GNUC__
#undef C_DEBUG_SWITCH
#undef C_OPTIMIZE_SWITCH
#define C_DEBUG_SWITCH "-g"
#endif

/* The X internationalization stuff is still broken in AIX 4.1, so
   don't #undef X11R5_INHIBIT_I18N
   It still causes shift, ctrl, and alt to resend the last character,
   if it was a control character like tab, enter, backspace, or ESC.
   Bill_Mann @ PraxisInt.com   */
/* #undef X11R5_INHIBIT_I18N */