Re: Give people with crashes more of a clue
24 years, 6 months
Jan Vroonhof
Jan Vroonhof
wmperry(a)aventail.com (William M. Perry) writes:
> GCC_VERSION=`some magic here`
>
> dnl Search for GCC specific build problems we know about
> if test "$GCC" = "yes"; then
> case `uname -s`:`uname -m`:$GCC_VERSION in
If this could be made into a real patch than this is approved and
retroactively recommended for every XEmacs known to man.
Jan
- 1 participants
- 0 comments
Another GPM change...
24 years, 6 months
William M. Perry
William M. Perry
This adds support for GPM on multiple VCs using 'gnuclient', adds a
lisp-level (gpm-enabled-p) function, and reworks the gpm.el code. It also
now hooks delete-device-hook instead of kill-emacs-hook, so that the GPM
stuff gets shut down correctly if you connect to a XEmacs running in X,
then kill the tty connection.
It is also safe now to call
(progn
(gpm-enable nil t)
(gpm-enable nil nil)
(gpm-enable nil nil))
Last round of patches it would explode because of insufficient checking of
the return value of Fget_process. The GPM processes are now named uniquely
for what TTY they are attached to.
Well, lunch is over, time to get back to real work. :(
No patch because they are all longer than the files now. If anybody on
linux could try this out, I would appreciate it. With this, I'm pretty
close to happy using it instead of X on my linux w/frame-buffer support
compiled in. The screen is sure a lot wider :)
-Bill P.
- 1 participants
- 0 comments
Re: Reworked GPM support...
24 years, 6 months
Jan Vroonhof
Jan Vroonhof
wmperry(a)aventail.com (William M. Perry) writes:
> Well, in the process of getting 'gnuclient -nw' to work correctly after you
> start XEmacs in X, I'm proceeding towards utter enlightenment. :)
Nice. Unfortunately I messed up the CVS tree on our linux machine
(that I don't hav easy console access to anyway) so badly when
handapplying the origional patches that I cannot easily test it now. I
hope I find the time to play with the (by then perfect :-)) version
later this week.
Jan
- 1 participants
- 0 comments
Re: Give people with crashes more of a clue
24 years, 6 months
William M. Perry
William M. Perry
Robert Pluim <rpluim(a)bigfoot.com> writes:
> Hrvoje Niksic writes:
> > rpluim(a)bigfoot.com writes:
> >
> > > I've finally cracked, I've had it with skip-syntax-backwards crashes
> > > on gnu.emacs.xemacs. Polite (for once) patch attached. Recommended
> > > for every version of XEmacs in the universe.
> >
> > Good idea, but isn't it pointless to refer people to a PROBLEMS file
> > that's not there anymore? If we aren't already shipping PROBLEMS with
> > XEmacs, maybe we should start.
>
> I did say "If you have access ...". If we're not shipping it, either
> we should ship it, or put the information in something we do ship
> (README? INSTALL? The FAQ? [Hmm, we don't seem to ship the FAQ,
> either. Maybe we should]).
For some of the problems in PROBLEMS, I would vote for configure hacking to
make sure people don't even try to build XEmacs that way unless they know
what they are doing.
GCC_VERSION=`some magic here`
dnl Search for GCC specific build problems we know about
if test "$GCC" = "yes"; then
case `uname -s`:`uname -m`:$GCC_VERSION in
*:sun4*:gcc_281|sun4*egcs_10*)
dnl Don't use -O2 with gcc 2.8.1 and egcs 1.0 under SPARC architectures
dnl without also using `-fno-schedule-insns'.
case "$CFLAGS" in
*-O2*)
case "$CFLAGS" in
*-fno-schedule-insns*) ;;
*)
AC_MSG_WARN(Don't use -O2 with gcc 2.8.1 and egcs 1.0 under SPARC architectures)
AC_MSG_WARN(without also using `-fno-schedule-insns'.)
AC_MSG_ERROR(Aborting due to known problem)
;;
esac
;;
esac
;;
Linux:alpha:egcs_11*)
AC_MSG_WARN(There have been reports of egcs-1.1 not compiling XEmacs correctly on)
AC_MSG_WARN(Alpha Linux. There have also been reports that egcs-1.0.3a is O.K.)
AC_MSG_ERROR(Aborting due to known problem)
;;
*:i*86*:gcc_272)
case "$CFLAGS" in
*-O2*)
case "$CFLAGS" in
*-fno-strength-reduce*) ;;
*)
AC_MSG_WARN(Don't use -O2 with gcc 2.7.2 under Intel/XXX without also using)
AC_MSG_WARN(`-fno-strength-reduce'.)
AC_MSG_ERROR(Aborting due to known problem)
;;
esac
;;
case "$CFLAGS" in
*-fno-caller-saves*) ;;
*)
AC_MSG_WARN(Don't use -O2 with gcc 2.7.2 under Intel/XXX without also using)
AC_MSG_WARN(`-fno-caller-saves'.)
AC_MSG_ERROR(Aborting due to known problem)
;;
esac
;;
esac
;;
.
.
.
esac
fi
-bp
- 2 participants
- 1 comments
Re: Reworked GPM support...
24 years, 6 months
William M. Perry
William M. Perry
Jan Vroonhof <vroonhof(a)math.ethz.ch> writes:
> However if you want to progress from mere coolness to utter enlightenment
> then you should
Well, in the process of getting 'gnuclient -nw' to work correctly after you
start XEmacs in X, I'm proceeding towards utter enlightenment. :)
> 0. make sure the case when we have frames on more than one VC works
This is working now, although the pointer disappears from both VCs when you
have multiple ones going. Very very odd. I'm attaching the latest
gpmevent.c in case you want to play with it. I'm not going to send a diff,
because it is now 90 lines longer than the file itself. :)
> 1. make sure the xterm mouse stuff works.
Still not going to work at the C level. I still have high hopes for
xt-mouse.el though. Maybe I'll hack on it sometime this week.
> 3. Make XEmacs support the GPM selection protocol.
Done in the latest round of patches.
-Bill P.
- 1 participants
- 0 comments
Re: applying patches
24 years, 6 months
Adrian Aichner
Adrian Aichner
>>>>> "Andreas" == Andreas Jaeger <aj(a)suse.de> writes:
Andreas> I'll try to find my way through the packages. Updating
Andreas> everything will be the first job - but I'll also try to
Andreas> make new packages and sumos regularly available.
I'm looking forward, thanks for your work!
Gunnar> [...]
Andreas> Andreas
Andreas> --
Andreas> Andreas Jaeger
Andreas> SuSE Labs aj(a)suse.de
Andreas> private aj(a)arthur.rhein-neckar.de
- 1 participants
- 0 comments
Re: applying patches
24 years, 6 months
Andreas Jaeger
Andreas Jaeger
>>>>> Gunnar Evermann writes:
Gunnar> Andreas Jaeger <aj(a)suse.de> writes:
>> >>>>> Jan Vroonhof writes:
>>
Jan> Gunnar Evermann <ge204(a)eng.cam.ac.uk> writes:
>> >> How are patches sent to xemacs-patches handled at the moment?
>>
Jan> Extremely ad-hoc. AFAICS nobody is applying patches to the packages at
Jan> the moment.
Gunnar> I know -- It was more of a rhetorical question, I just wanted to
Gunnar> kick-start the discussion :-)
You've been successfull;-)
>> >> Is anyone applying the patches sent there by people without cvs-write
>> >> access?
>>
Jan> Andy and Michael have applied most them. Normally Steve applies all
Jan> that were left out, just before a new beta releases.
>>
>> can I help out? After considering this for some time, I'm
>> volunteering to help.
>>
>> I should have enough time to keep the CVS archive for the packages
>> uptodate.
Gunnar> I'm not sure whether there is a debate about this going on in
Gunnar> xemacs-review, but FWIW I think this is an excellent idea.
It is going on - and the result is that I'm looking at the packages.
Now I only need to have all infrastructure set up.
Gunnar> I guess it would be easier for everyone if we had a split of
Gunnar> responsibilities like the following:
Gunnar> - 21.1 Vin
Gunnar> - 21.2 Steve
Gunnar> - packages Andreas
Gunnar> Your (Andreas) contribution would be even more significant if you
Gunnar> could not only apply the package patches, but regularly build the new
Gunnar> packages and sumos. We REALLY REALLY need to update them! Do we still
I know :-(
Gunnar> ship that sumo with broken lookup etc.?
I'll try to find my way through the packages. Updating everything
will be the first job - but I'll also try to make new packages and
sumos regularly available.
Gunnar> [...]
Andreas
--
Andreas Jaeger
SuSE Labs aj(a)suse.de
private aj(a)arthur.rhein-neckar.de
- 1 participants
- 0 comments
Re: applying patches
24 years, 6 months
Gunnar Evermann
Gunnar Evermann
Andreas Jaeger <aj(a)suse.de> writes:
> >>>>> Jan Vroonhof writes:
>
> Jan> Gunnar Evermann <ge204(a)eng.cam.ac.uk> writes:
> >> How are patches sent to xemacs-patches handled at the moment?
>
> Jan> Extremely ad-hoc. AFAICS nobody is applying patches to the packages at
> Jan> the moment.
I know -- It was more of a rhetorical question, I just wanted to
kick-start the discussion :-)
> >> Is anyone applying the patches sent there by people without cvs-write
> >> access?
>
> Jan> Andy and Michael have applied most them. Normally Steve applies all
> Jan> that were left out, just before a new beta releases.
>
> can I help out? After considering this for some time, I'm
> volunteering to help.
>
> I should have enough time to keep the CVS archive for the packages
> uptodate.
I'm not sure whether there is a debate about this going on in
xemacs-review, but FWIW I think this is an excellent idea.
I guess it would be easier for everyone if we had a split of
responsibilities like the following:
- 21.1 Vin
- 21.2 Steve
- packages Andreas
Your (Andreas) contribution would be even more significant if you
could not only apply the package patches, but regularly build the new
packages and sumos. We REALLY REALLY need to update them! Do we still
ship that sumo with broken lookup etc.?
> For those who don't know me, let me introduce myself briefly. I'm one
> of the glibc developers and XEmacs is one of my favorite programs.
[...]
I'm sure nobody doubts that you are qualified!
Many thanks for volunteering.
Gunnar
--
Gunnar Evermann
Speech, Vision & Robotics Group
Cambridge University Engineering Department
- 1 participants
- 0 comments
Re: Give people with crashes more of a clue
24 years, 6 months
Hrvoje Niksic
Hrvoje Niksic
rpluim(a)bigfoot.com writes:
> I've finally cracked, I've had it with skip-syntax-backwards crashes
> on gnu.emacs.xemacs. Polite (for once) patch attached. Recommended
> for every version of XEmacs in the universe.
Good idea, but isn't it pointless to refer people to a PROBLEMS file
that's not there anymore? If we aren't already shipping PROBLEMS with
XEmacs, maybe we should start.
- 2 participants
- 1 comments
Hopefully final GPM patch...
24 years, 6 months
William M. Perry
William M. Perry
This incorporates all of ben's recent suggestions. If someone wants to
hack whatever is necessary (dumped-lisp.el?) to get gpm.el loaded before
dump time, the device creation hooking seems to work, and correctly brings
up XEmacs with GPM enabled with no mucking about by the user on the linux
console.
-Bill P.
Index: lisp/gpm.el
===================================================================
*** gpm.el Tue May 5 15:32:27 1998
--- gpm.el Sat Nov 6 19:01:45 1999
***************
*** 0 ****
--- 1,73 ----
+ ;;; gpm.el --- Support the mouse when emacs run on a Linux console.
+
+ ;; Copyright (C) 1999 Free Software Foundation
+
+ ;; Author: William Perry <wmperry(a)gnu.org>
+ ;; Keywords: mouse, terminals
+
+ ;; This file is part of XEmacs.
+
+ ;; XEmacs is free software; you can redistribute it and/or modify
+ ;; it under the terms of the GNU General Public License as published by
+ ;; the Free Software Foundation; either version 2, or (at your option)
+ ;; any later version.
+
+ ;; XEmacs is distributed in the hope that it will be useful,
+ ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
+ ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ ;; GNU General Public License for more details.
+
+ ;; You should have received a copy of the GNU General Public License
+ ;; along with GNU Emacs; see the file COPYING. If not, write to the
+ ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ ;; Boston, MA 02111-1307, USA.
+
+ (defvar gpm-minor-mode nil
+ "Non-nil when gpm-minor-mode is active. Call `gpm-minor-mode' to toggle.")
+
+ (defun gpm-minor-mode (&optional arg)
+ "Toggle GPM mouse mode.
+ With prefix arg, turn GPM mouse mode on if and only if arg is positive."
+ (interactive "P")
+ (cond
+ ((null arg) ; Toggle
+ (if gpm-minor-mode
+ (progn
+ (gpm-enable nil)
+ (setq gpm-minor-mode nil))
+ (if (gpm-enable t)
+ (setq gpm-minor-mode t))))
+ ((> arg 0) ; Turn on
+ (if (gpm-enable t)
+ (setq gpm-minor-mode t)))
+ (t ; Turn off
+ (gpm-enable nil)
+ (setq gpm-minor-mode nil))))
+
+ (defun turn-on-gpm-mouse-tracking ()
+ ;; Enable mouse tracking on linux console
+ (if gpm-minor-mode
+ (gpm-enable t)))
+
+ (defun turn-off-gpm-mouse-tracking ()
+ ;; Disable mouse tracking on linux console
+ (if gpm-minor-mode
+ (gpm-enable nil)))
+
+ (defun gpm-create-device-hook (device)
+ (if (and (not noninteractive) ; Don't want to do this in batch mode
+ (fboundp 'gpm-enable) ; Must have C-level GPM support
+ (eq system-type 'linux) ; Must be running linux
+ (eq (device-type device) 'tty) ; on a tty
+ (equal "linux" (console-tty-terminal-type ; an a linux terminal type
+ (device-console device))))
+ (gpm-minor-mode 5)))
+
+ ;; Restore normal mouse behaviour outside Emacs
+
+ (add-hook 'suspend-hook 'turn-off-gpm-mouse-tracking)
+ (add-hook 'suspend-resume-hook 'turn-on-gpm-mouse-tracking)
+ (add-hook 'kill-emacs-hook 'turn-off-gpm-mouse-tracking)
+ (add-hook 'create-device-hook 'gpm-create-device-hook)
+
+ (provide 'gpm)
Index: lisp/mouse.el
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/lisp/mouse.el,v
retrieving revision 1.19.2.6
diff -u -r1.19.2.6 mouse.el
--- mouse.el 1999/09/02 08:47:55 1.19.2.6
+++ mouse.el 1999/11/07 03:30:16
＠＠ -88,7 +88,9 ＠＠
the X cutbuffer. A mark is pushed, so that the inserted text lies
between point and mark."
(interactive)
- (if (not (console-on-window-system-p))
+ (if (and (not (console-on-window-system-p))
+ (and (featurep 'gpm)
+ (not gpm-minor-mode)))
(yank)
(push-mark)
(if (region-active-p)
Index: src/console-tty.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/console-tty.c,v
retrieving revision 1.18.2.5
diff -u -r1.18.2.5 console-tty.c
--- console-tty.c 1999/10/24 03:48:31 1.18.2.5
+++ console-tty.c 1999/11/07 03:30:17
＠＠ -38,9 +38,6 ＠＠
#ifdef FILE_CODING
#include "file-coding.h"
#endif
-#ifdef HAVE_GPM
-#include "gpmevent.h"
-#endif
DEFINE_CONSOLE_TYPE (tty);
DECLARE_IMAGE_INSTANTIATOR_FORMAT (nothing);
＠＠ -129,10 +126,6 ＠＠
#endif /* MULE */
tty_con->terminal_type = terminal_type;
tty_con->controlling_process = controlling_process;
-
-#ifdef HAVE_GPM
- connect_to_gpm (con);
-#endif
if (NILP (CONSOLE_NAME (con)))
CONSOLE_NAME (con) = Ffile_name_nondirectory (tty);
Index: src/console-tty.h
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/console-tty.h,v
retrieving revision 1.8.2.2
diff -u -r1.8.2.2 console-tty.h
--- console-tty.h 1999/02/17 20:03:57 1.8.2.2
+++ console-tty.h 1999/11/07 03:30:18
＠＠ -41,9 +41,6 ＠＠
struct tty_console
{
int infd, outfd;
-#ifdef HAVE_GPM
- int mouse_fd;
-#endif
Lisp_Object instream, outstream;
Lisp_Object terminal_type;
Lisp_Object controlling_process;
＠＠ -203,9 +200,6 ＠＠
unsigned int is_stdio :1;
};
-#ifdef HAVE_GPM
-#define CONSOLE_TTY_MOUSE_FD(c) (CONSOLE_TTY_DATA (c)->mouse_fd)
-#endif
#define CONSOLE_TTY_DATA(c) CONSOLE_TYPE_DATA (c, tty)
#define CONSOLE_TTY_CURSOR_X(c) (CONSOLE_TTY_DATA (c)->cursor_x)
#define CONSOLE_TTY_CURSOR_Y(c) (CONSOLE_TTY_DATA (c)->cursor_y)
Index: src/device-tty.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/device-tty.c,v
retrieving revision 1.9
diff -u -r1.9 device-tty.c
--- device-tty.c 1998/05/27 20:42:00 1.9
+++ device-tty.c 1999/11/07 03:30:18
＠＠ -38,10 +38,6 ＠＠
#include "syssignal.h" /* for SIGWINCH */
-#ifdef HAVE_GPM
-#include <gpm.h>
-#endif
-
#include <errno.h>
Lisp_Object Qinit_pre_tty_win, Qinit_post_tty_win;
＠＠ -155,15 +151,6 ＠＠
CONSOLE_TTY_DATA (con)->width = width;
CONSOLE_TTY_DATA (con)->height = height;
-#ifdef HAVE_GPM
- /* We need to tell GPM how big our screen is now
- ** I am pretty sure the GPM library will get incredibly confused
- ** if you try to connect to more than one mouse-capable device,
- ** so I don't think it will cause any more damage in that case.
- */
- gpm_mx = width;
- gpm_my = height;
-#endif
for (tail = DEVICE_FRAME_LIST (d);
!NILP (tail);
tail = XCDR (tail))
Index: src/emacs.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/emacs.c,v
retrieving revision 1.82.2.32
diff -u -r1.82.2.32 emacs.c
--- emacs.c 1999/11/02 02:41:45 1.82.2.32
+++ emacs.c 1999/11/07 03:30:20
＠＠ -1088,6 +1088,10 ＠＠
syms_of_eldap ();
#endif
+#ifdef HAVE_GPM
+ syms_of_gpmevent ();
+#endif
+
/* Now create the subtypes for the types that have them.
We do this before the vars_*() because more symbols
may get initialized here. */
＠＠ -1481,6 +1485,10 ＠＠
#ifdef HAVE_LDAP
vars_of_eldap ();
+#endif
+
+#ifdef HAVE_GPM
+ vars_of_gpmevent ();
#endif
/* Now initialize any specifier variables. We do this later
Index: src/event-Xt.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/event-Xt.c,v
retrieving revision 1.41.2.14
diff -u -r1.41.2.14 event-Xt.c
--- event-Xt.c 1999/10/10 11:45:15 1.41.2.14
+++ event-Xt.c 1999/11/07 03:30:22
＠＠ -2034,9 +2034,6 ＠＠
{
Lisp_Object console;
int infd;
-#ifdef HAVE_GPM
- int mousefd;
-#endif
if (CONSOLE_X_P (con))
return; /* X consoles are automatically selected for when we
＠＠ -2044,22 +2041,6 ＠＠
infd = event_stream_unixoid_select_console (con);
XSETCONSOLE (console, con);
select_filedesc (infd, console);
-#ifdef HAVE_GPM
- /* On a stream device (ie: noninteractive), bad things can happen. */
- if (EQ (CONSOLE_TYPE (con), Qtty)) {
- mousefd = CONSOLE_TTY_MOUSE_FD (con);
- /* We check filedesc_to_what_closure[fd] here because if you run
- ** XEmacs from a TTY, it will fire up GPM, select the mouse fd, then
- ** if you run gnuattach to connect to another TTY, it will fire up
- ** GPM again, and try to reselect the mouse fd. GPM uses the same
- ** fd for every connection apparently, and select_filedesc will
- ** fail its assertion if we try to select it twice.
- */
- if ((mousefd >= 0) && !filedesc_to_what_closure[mousefd]) {
- select_filedesc (mousefd, console);
- }
- }
-#endif
}
static void
＠＠ -2067,9 +2048,6 ＠＠
{
Lisp_Object console;
int infd;
-#ifdef HAVE_GPM
- int mousefd;
-#endif
if (CONSOLE_X_P (con))
return; /* X consoles are automatically selected for when we
＠＠ -2077,15 +2055,6 ＠＠
infd = event_stream_unixoid_unselect_console (con);
XSETCONSOLE (console, con);
unselect_filedesc (infd);
-#ifdef HAVE_GPM
- /* On a stream device (ie: noninteractive), bad things can happen. */
- if (EQ (CONSOLE_TYPE (con), Qtty)) {
- mousefd = CONSOLE_TTY_MOUSE_FD (con);
- if (mousefd >= 0) {
- unselect_filedesc (mousefd);
- }
- }
-#endif
}
/* read an event from a tty, if one is available. Returns non-zero
Index: src/event-unixoid.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/event-unixoid.c,v
retrieving revision 1.9.2.1
diff -u -r1.9.2.1 event-unixoid.c
--- event-unixoid.c 1999/03/07 15:25:17 1.9.2.1
+++ event-unixoid.c 1999/11/07 03:30:27
＠＠ -41,10 +41,6 ＠＠
#include "sysproc.h" /* select stuff */
#include "systime.h"
-#ifdef HAVE_GPM
-#include "gpmevent.h"
-#endif
-
/* Mask of bits indicating the descriptors that we wait for input on.
These work as follows:
＠＠ -83,12 +79,6 ＠＠
Lisp_Object console;
XSETCONSOLE (console, con);
-
-#ifdef HAVE_GPM
- if (fd == CONSOLE_TTY_MOUSE_FD (con)) {
- return handle_gpm_read (event,con,fd);
- }
-#endif
nread = read (fd, &ch, 1);
if (nread <= 0)
Index: src/frame-tty.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/frame-tty.c,v
retrieving revision 1.8.2.1
diff -u -r1.8.2.1 frame-tty.c
--- frame-tty.c 1998/12/05 16:55:48 1.8.2.1
+++ frame-tty.c 1999/11/07 03:30:27
＠＠ -33,10 +33,6 ＠＠
#include "events.h"
-#ifdef HAVE_GPM
-#include <gpm.h>
-#endif
-
/* Default properties to use when creating frames. */
Lisp_Object Vdefault_tty_frame_plist;
＠＠ -86,33 +82,6 ＠＠
call1 (Qinit_post_tty_win, FRAME_CONSOLE (f));
}
-#ifdef HAVE_GPM
-static int
-tty_get_mouse_position (struct device *d, Lisp_Object *frame, int *x, int *y)
-{
- Gpm_Event ev;
- int num_buttons;
-
- num_buttons = Gpm_GetSnapshot(&ev);
- *x = ev.x;
- *y = ev.y;
- *frame = DEVICE_SELECTED_FRAME (d);
- return (1);
-}
-
-static void
-tty_set_mouse_position (struct window *w, int x, int y)
-{
- /* XXX
- I couldn't find any GPM functions that set the mouse position.
- Mr. Perry had left this function empty; that must be why.
- karlheg
- */
-}
-
-#endif
-
-
/* Change from withdrawn state to mapped state. */
static void
tty_make_frame_visible (struct frame *f)
＠＠ -223,10 +192,6 ＠＠
CONSOLE_HAS_METHOD (tty, init_frame_1);
CONSOLE_HAS_METHOD (tty, init_frame_3);
CONSOLE_HAS_METHOD (tty, after_init_frame);
-#ifdef HAVE_GPM
- CONSOLE_HAS_METHOD (tty, get_mouse_position);
- CONSOLE_HAS_METHOD (tty, set_mouse_position);
-#endif
CONSOLE_HAS_METHOD (tty, make_frame_visible);
CONSOLE_HAS_METHOD (tty, make_frame_invisible);
CONSOLE_HAS_METHOD (tty, frame_visible_p);
Index: src/gpmevent.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/gpmevent.c,v
retrieving revision 1.5
diff -u -r1.5 gpmevent.c
--- gpmevent.c 1998/03/31 20:11:47 1.5
+++ gpmevent.c 1999/11/07 03:30:28
＠＠ -1,5 +1,28 ＠＠
-/* William Perry 1997 */
+/* GPM functions
+ Copyright (C) 1997 William M. Perry <wmperry(a)gnu.org>
+ Copyright (C) 1999 Free Software Foundation, Inc.
+
+This file is part of XEmacs.
+
+XEmacs is free software; you can redistribute it and/or modify it
+under the terms of the GNU General Public License as published by the
+Free Software Foundation; either version 2, or (at your option) any
+later version.
+
+XEmacs is distributed in the hope that it will be useful, but WITHOUT
+ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with XEmacs; see the file COPYING. If not, write to
+the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+Boston, MA 02111-1307, USA. */
+/* Synched up with: Not in FSF. */
+
+/* Authors: William Perry */
+
#include <config.h>
#include "lisp.h"
#include "console.h"
＠＠ -8,11 +31,18 ＠＠
#include "events.h"
#include "events-mod.h"
#include "sysdep.h"
+#include "commands.h"
+#include "lstream.h"
+#include "process.h"
#ifdef HAVE_GPM
#include "gpmevent.h"
#include <gpm.h>
+/* This is commented out in commands.h, I'm not sure why */
+extern Lisp_Object Vunread_command_events;
+extern Lisp_Object Vunread_command_event;
+
#if (!defined(__linux__)) /* possible under xterm */
#define KG_SHIFT 0
#define KG_CTRL 2
＠＠ -20,95 +50,414 ＠＠
#else
#include <linux/keyboard.h>
#endif
+
+static int (*orig_event_pending_p) (int);
+static void (*orig_next_event_cb) (struct Lisp_Event *);
+
+static Lisp_Object gpm_event_queue;
+static Lisp_Object gpm_event_queue_tail;
+
+static int
+get_process_infd (struct Lisp_Process *p)
+{
+ Lisp_Object instr, outstr;
+ get_process_streams (p, &instr, &outstr);
+ assert (!NILP (instr));
+ return filedesc_stream_fd (XLSTREAM (instr));
+}
+
+DEFUN ("receive-gpm-event", Freceive_gpm_event, 0, 2, 0, /*
+Run GPM_GetEvent().
+This function is the process handler for the GPM connection.
+*/
+ (process, string))
+{
+ Gpm_Event ev;
+ int modifiers = 0;
+ int type = -1;
+ int button = 1;
+ Lisp_Object fake_event;
+ Lisp_Process *proc;
+ struct Lisp_Event *event = NULL;
+ struct gcpro gcpro1;
+ static int num_events;
+
+ CHECK_PROCESS (process);
+
+ if (!Gpm_GetEvent(&ev))
+ {
+ warn_when_safe (Qnil, Qcritical, "Gpm_GetEvent failed");
+ return(Qzero);
+ }
+
+ GCPRO1(fake_event);
+
+ num_events++;
+
+ fake_event = Fmake_event (Qnil, Qnil);
+ event = XEVENT(fake_event);
+
+ event->timestamp = 0;
+ event->channel = Fselected_frame (Qnil); /* CONSOLE_SELECTED_FRAME (con); */
+
+ /* Whow, wouldn't named defines be NICE!?!?! */
+ modifiers = 0;
+
+ if (ev.modifiers & 1) modifiers |= MOD_SHIFT;
+ if (ev.modifiers & 2) modifiers |= MOD_META;
+ if (ev.modifiers & 4) modifiers |= MOD_CONTROL;
+ if (ev.modifiers & 8) modifiers |= MOD_META;
+
+ if (ev.buttons & GPM_B_LEFT)
+ {
+ button = 1;
+ }
+ else if (ev.buttons & GPM_B_MIDDLE)
+ {
+ button = 2;
+ }
+ else if (ev.buttons & GPM_B_RIGHT)
+ {
+ button = 3;
+ }
+
+ switch (GPM_BARE_EVENTS(ev.type)) {
+ case GPM_DOWN:
+ case GPM_UP:
+ event->event_type =
+ (ev.type & GPM_DOWN) ? button_press_event : button_release_event;
+ event->event.button.x = ev.x;
+ event->event.button.y = ev.y;
+ event->event.button.button = button;
+ event->event.button.modifiers = modifiers;
+ break;
+ case GPM_MOVE:
+ case GPM_DRAG:
+ event->event_type = pointer_motion_event;
+ event->event.motion.x = ev.x;
+ event->event.motion.y = ev.y;
+ event->event.motion.modifiers = modifiers;
+ default:
+ /* This will never happen */
+ break;
+ }
+
+ /* Handle the event */
+ enqueue_event (fake_event, &gpm_event_queue, &gpm_event_queue_tail);
+
+ GPM_DRAWPOINTER(&ev);
+ UNGCPRO;
+
+ return (Qzero);
+}
-int
-handle_gpm_read (struct Lisp_Event *event, struct console *con, int fd)
+static void turn_off_gpm (void)
{
- Gpm_Event ev;
- int modifiers = 0;
- int type = -1;
- int button = 1;
-
- if (!Gpm_GetEvent(&ev))
- return 0;
-
- event->timestamp = 0;
- event->channel = CONSOLE_SELECTED_FRAME (con);
-
- /* Whow, wouldn't named defines be NICE!?!?! */
- modifiers = 0;
-
- if (ev.modifiers & 1) modifiers |= MOD_SHIFT;
- if (ev.modifiers & 2) modifiers |= MOD_META;
- if (ev.modifiers & 4) modifiers |= MOD_CONTROL;
- if (ev.modifiers & 8) modifiers |= MOD_META;
-
- if (ev.type & GPM_DOWN)
- type = GPM_DOWN;
- else if (ev.type & GPM_UP)
- type = GPM_UP;
- else if (ev.type & GPM_MOVE) {
- type = GPM_MOVE;
- GPM_DRAWPOINTER(&ev);
- }
-
- if (ev.buttons & GPM_B_LEFT)
- button = 1;
- else if (ev.buttons & GPM_B_MIDDLE)
- button = 2;
- else if (ev.buttons & GPM_B_RIGHT)
- button = 3;
-
- switch (type) {
- case GPM_DOWN:
- case GPM_UP:
- event->event_type =
- type == GPM_DOWN ? button_press_event : button_release_event;
- event->event.button.x = ev.x;
- event->event.button.y = ev.y;
- event->event.button.button = button;
- event->event.button.modifiers = modifiers;
- break;
- case GPM_MOVE:
- event->event_type = pointer_motion_event;
- event->event.motion.x = ev.x;
- event->event.motion.y = ev.y;
- event->event.motion.modifiers = modifiers;
- default:
- return 0;
- }
- return 1;
-}
-
-void
-connect_to_gpm (struct console *con)
-{
- /* Only do this if we are running after dumping and really interactive */
- if (!noninteractive && initialized) {
- /* We really only want to do this on a TTY */
- CONSOLE_TTY_MOUSE_FD (con) = -1;
- if (EQ (CONSOLE_TYPE (con), Qtty)) {
- Gpm_Connect conn;
- int rval;
-
- conn.eventMask = GPM_DOWN|GPM_UP|GPM_MOVE;
- conn.defaultMask = GPM_MOVE;
- conn.minMod = 0;
- conn.maxMod = ((1<<KG_SHIFT)|(1<<KG_ALT)|(1<<KG_CTRL));
-
- rval = Gpm_Open (&conn, 0);
- switch (rval) {
- case -1: /* General failure */
- break;
- case -2: /* We are running under an XTerm */
+ Fdelete_process (build_string ("gpm"));
Gpm_Close();
- break;
- default:
- set_descriptor_non_blocking (gpm_fd);
- CONSOLE_TTY_MOUSE_FD (con) = gpm_fd;
- }
- }
- }
+
+ /* Remove the TTY mouse methods */
+ /* This is breaking abstraction a bit, but... */
+ tty_console_methods->get_mouse_position_method = NULL;
+ tty_console_methods->set_mouse_position_method = NULL;
+ tty_console_methods->own_selection_method = NULL;
+ tty_console_methods->get_foreign_selection_method = NULL;
+}
+
+#ifdef TIOCLINUX
+static Lisp_Object
+tty_get_foreign_selection (Lisp_Object selection_symbol, Lisp_Object target_type)
+{
+ /* This function can GC */
+ struct device *d = decode_device (Qnil);
+ int fd = DEVICE_INFD (d);
+ char c = 3;
+ Lisp_Object output_stream;
+ Lisp_Object terminal_stream;
+ Lisp_Object output_string;
+ struct gcpro gcpro1,gcpro2,gcpro3;
+
+ GCPRO3(output_stream,terminal_stream,output_string);
+
+ /* The ioctl() to paste actually puts things in the input queue of
+ ** the virtual console, so we need to trap that data, since we are
+ ** supposed to return the actual string selection from this
+ ** function.
+ */
+
+ /* I really hate doing this, but it doesn't seem to cause any
+ ** problems, and it makes the Lstream_read stuff further down
+ ** error out correctly instead of trying to indefinitely read from
+ ** the console.
+ **
+ ** There is no set_descriptor_blocking() function call, but in my
+ ** testing under linux, it has not proved fatal to leave the
+ ** descriptor in non-blocking mode.
+ **
+ ** William Perry Nov 5, 1999
+ */
+ set_descriptor_non_blocking (fd);
+
+ /* We need two streams, one for reading from the selected device,
+ ** and one to write the data into. There is no writable version
+ ** of the lisp-string lstream, so we make do with a resizing
+ ** buffer stream, and make a string out of it after we are
+ ** done.
+ */
+ output_stream = make_resizing_buffer_output_stream ();
+ terminal_stream = make_filedesc_input_stream (fd, 0, -1, LSTR_BLOCKED_OK);
+ output_string = Qnil;
+
+ /* #### We should arguably use a specbind() and an unwind routine here,
+ ** #### but I don't care that much right now.
+ */
+ if (NILP (output_stream) || NILP (terminal_stream))
+ {
+ /* Should we signal an error here? */
+ goto out;
+ }
+
+ if (ioctl (fd, TIOCLINUX, &c) < 0)
+ {
+ /* Could not get the selection - eek */
+ UNGCPRO;
+ return (Qnil);
+ }
+
+ while (1)
+ {
+ Bufbyte tempbuf[1024]; /* some random amount */
+ int i;
+ int size_in_bytes = Lstream_read (XLSTREAM (terminal_stream),
+ tempbuf, sizeof (tempbuf));
+
+ if (size_in_bytes <= 0)
+ {
+ /* end of the stream */
+ break;
+ }
+
+ /* convert CR->LF */
+ for (i = 0; i < size_in_bytes; i++)
+ {
+ if (tempbuf[i] == '\r')
+ {
+ tempbuf[i] = '\n';
+ }
+ }
+
+ Lstream_write (XLSTREAM (output_stream), tempbuf, size_in_bytes);
+ }
+
+ Lstream_flush (XLSTREAM (output_stream));
+
+ output_string = make_string (resizing_buffer_stream_ptr (XLSTREAM (output_stream)),
+ Lstream_byte_count (XLSTREAM (output_stream)));
+
+ Lstream_delete (XLSTREAM (output_stream));
+ Lstream_delete (XLSTREAM (terminal_stream));
+
+ out:
+ UNGCPRO;
+ return (output_string);
+}
+
+static Lisp_Object
+tty_selection_exists_p (Lisp_Object selection)
+{
+ return (Qt);
+}
+#endif /* TIOCLINUX */
+
+#if 0
+static Lisp_Object
+tty_own_selection (Lisp_Object selection_name, Lisp_Object selection_value)
+{
+ /* There is no way to do this cleanly - the GPM selection
+ ** 'protocol' (actually the TIOCLINUX ioctl) requires a start and
+ ** end position on the _screen_, not a string to stick in there.
+ ** Lame.
+ **
+ ** William Perry Nov 4, 1999
+ */
+}
+#endif
+
+/* This function appears to work once in a blue moon. I'm not sure
+** exactly why either. *sigh*
+** William Perry Nov 4, 1999
+*/
+static int
+tty_get_mouse_position (struct device *d, Lisp_Object *frame, int *x, int *y)
+{
+ Gpm_Event ev;
+ int num_buttons;
+
+ memset(&ev,'\0',sizeof(ev));
+ num_buttons = Gpm_GetSnapshot(&ev);
+
+ if (!num_buttons)
+ {
+ /* This means there are events pending... */
+ return(-1);
+ }
+ *x = ev.x;
+ *y = ev.y;
+ *frame = DEVICE_SELECTED_FRAME (d);
+ return (1);
+}
+
+static void
+tty_set_mouse_position (struct window *w, int x, int y)
+{
+ /*
+ #### I couldn't find any GPM functions that set the mouse position.
+ #### Mr. Perry had left this function empty; that must be why.
+ #### karlheg
+ */
+}
+
+static int gpm_event_pending_p (int user_p)
+{
+ Lisp_Object event;
+
+ EVENT_CHAIN_LOOP (event, gpm_event_queue)
+ {
+ if (!user_p || command_event_p (event))
+ {
+ return (1);
+ }
+ }
+ return (orig_event_pending_p (user_p));
+}
+
+static void gpm_next_event_cb (struct Lisp_Event *event)
+{
+ /* #### It would be nice to preserve some sort of ordering of the
+ ** #### different types of events, but that would be quite a bit
+ ** #### of work, and would more than likely break the abstraction
+ ** #### between the other event loops and this one.
+ */
+
+ if (!NILP (gpm_event_queue))
+ {
+ Lisp_Object queued_event = dequeue_event (&gpm_event_queue, &gpm_event_queue_tail);
+ *event = *(XEVENT (queued_event));
+ return;
+ }
+
+ return (orig_next_event_cb (event));
+}
+
+static void hook_event_callbacks_once (void)
+{
+ static int hooker;
+
+ if (!hooker)
+ {
+ orig_event_pending_p = event_stream->event_pending_p;
+ orig_next_event_cb = event_stream->next_event_cb;
+ event_stream->event_pending_p = gpm_event_pending_p;
+ event_stream->next_event_cb = gpm_next_event_cb;
+ hooker = 1;
+ }
}
+DEFUN ("gpm-enable", Fgpm_enable, 0, 1, 0, /*
+Toggle accepting of GPM mouse events.
+*/
+ (arg))
+{
+ Gpm_Connect conn;
+ int rval;
+ Lisp_Object gpm_process;
+ Lisp_Object gpm_filter;
+
+ if (NILP (arg))
+ {
+ turn_off_gpm ();
+ return (Qnil);
+ }
+
+ if (noninteractive)
+ {
+ error ("Can't connect to GPM in batch mode.");
+ }
+
+ hook_event_callbacks_once ();
+
+ /* DANGER DANGER.
+ ** Though shalt not call (gpm-enable t) after we have already
+ ** started, or stuff blows up.
+ */
+ if (!NILP (Fget_process ( build_string("gpm"))))
+ {
+ error ("GPM already enabled.");
+ }
+
+ /* Install the mouse position methods for the TTY console type */
+ CONSOLE_HAS_METHOD (tty, get_mouse_position);
+ CONSOLE_HAS_METHOD (tty, set_mouse_position);
+
+#ifdef TIOCLINUX
+ CONSOLE_HAS_METHOD (tty, get_foreign_selection);
+ CONSOLE_HAS_METHOD (tty, selection_exists_p);
+#if 0
+ CONSOLE_HAS_METHOD (tty, own_selection);
#endif
+#endif /* TIOCLINUX */
+
+ conn.eventMask = GPM_DOWN|GPM_UP|GPM_MOVE|GPM_DRAG;
+ conn.defaultMask = GPM_MOVE;
+ conn.minMod = 0;
+ conn.maxMod = ((1<<KG_SHIFT)|(1<<KG_ALT)|(1<<KG_CTRL));
+
+ rval = Gpm_Open (&conn, 0);
+
+ switch (rval) {
+ case -1: /* General failure */
+ break;
+ case -2: /* We are running under an XTerm */
+ Gpm_Close();
+ break;
+ default:
+ /* Is this really necessary? */
+ set_descriptor_non_blocking (gpm_fd);
+ gpm_process = connect_to_file_descriptor (build_string ("gpm"), Qnil,
+ make_int (gpm_fd),
+ make_int (gpm_fd));
+
+ if (!NILP (gpm_process))
+ {
+ rval = 0;
+ Fprocess_kill_without_query (gpm_process, Qnil);
+ XSETSUBR (gpm_filter, &SFreceive_gpm_event);
+ set_process_filter (gpm_process, gpm_filter, 1);
+ }
+ else
+ {
+ Gpm_Close();
+ rval = -1;
+ }
+ }
+
+ return(rval ? Qnil : Qt);
+}
+
+void vars_of_gpmevent (void)
+{
+ gpm_event_queue = Qnil;
+ gpm_event_queue_tail = Qnil;
+ staticpro (&gpm_event_queue);
+ staticpro (&gpm_event_queue_tail);
+ pdump_wire (&gpm_event_queue);
+ pdump_wire (&gpm_event_queue_tail);
+}
+
+void syms_of_gpmevent (void)
+{
+ DEFSUBR (Freceive_gpm_event);
+ DEFSUBR (Fgpm_enable);
+}
+
+#endif /* HAVE_GPM */
Index: src/gpmevent.h
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/gpmevent.h,v
retrieving revision 1.3
diff -u -r1.3 gpmevent.h
--- gpmevent.h 1998/03/31 20:11:47 1.3
+++ gpmevent.h 1999/11/07 03:30:28
＠＠ -1,7 +0,0 ＠＠
-#ifndef _HAVE_GPM
-#define _HAVE_GPM
-
-int handle_gpm_read (struct Lisp_Event *event, struct console *con, int fd);
-void connect_to_gpm (struct console *con);
-
-#endif
Index: src/symsinit.h
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/symsinit.h,v
retrieving revision 1.31.2.13
diff -u -r1.31.2.13 symsinit.h
--- symsinit.h 1999/10/30 03:00:51 1.31.2.13
+++ symsinit.h 1999/11/07 03:30:28
＠＠ -148,6 +148,7 ＠＠
void syms_of_window (void);
void syms_of_xselect (void);
void syms_of_eldap (void);
+void syms_of_gpmevent (void);
/* Initialize the console types (dump-time but for reinit_). */
＠＠ -368,6 +369,7 ＠＠
void vars_of_xselect (void);
void reinit_vars_of_xselect (void);
void vars_of_eldap (void);
+void vars_of_gpmevent (void);
/* Initialize specifier variables (dump-time only). */
cvs server: Diffing src/m
cvs server: Diffing src/s
- 2 participants
- 1 comments