1:35 a.m.
This bug report will be sent to the XEmacs Development Team,
not to your local site managers!!
Please write in English, because the XEmacs maintainers do not have
translators to read other languages for them.
In XEmacs 21.2 (beta38) "Peisino,Ak(B" [Lucid] (i386-unknown-freebsd4.2,
Mule) of Wed Dec 13 2000 on elvenbow.nc.kyushu-u.ac.jp
configured using `configure --with-mule --without-xim --error-checking=all --debug
--with-xfs --with-widgets=athena --without-canna --site-prefixes=/usr/local
'--cflags=-g -O3 -pipe -Wall -Wno-switch -Wpointer-arith -Winline -Wmissing-prototypes
-Wshadow''
Please describe exactly what actions triggered the bug
and the precise symptoms of the bug:
xemacs -vanilla
M-x make-frame-on-tty
C-x b test (in the X frame)
C-x b test (in the tty frame)
C-x k (in the X frame)
** crash **
It is repeatable. It won't crash when doing C-x k in the tty frame.
Lisp backtrace follows:
# (condition-case ... . error)
# (catch top-level ...)
C backtrace follows:
(gdb) where
#0 0x286323e0 in kill () from /usr/lib/libc.so.4
#1 0x80bb823 in fatal_error_signal (sig=6) at emacs.c:535
#2 0xbfbfffac in ?? ()
#3 0x80be869 in assert_failed (file=0x824d9a7 "eval.c", line=1885,
expr=0x824d508 "abort()") at emacs.c:3214
#4 0x80c12f1 in signal_1 (sig=137300068, data=142340244) at eval.c:1885
#5 0x80c2791 in error (fmt=0x8285dc0 "Marker does not point anywhere")
at eval.c:2052
#6 0x8194e66 in marker_position (marker=141246172) at marker.c:354
#7 0x81afc17 in redisplay_window (window=141753344, skip_selected=0)
at redisplay.c:5896
#8 0x81b098c in redisplay_frame (f=0x8738400, preemption_check=0)
at redisplay.c:6393
#9 0x81b0b6f in redisplay_device (d=0x870a600, automatic=1)
at redisplay.c:6467
#10 0x81b0edd in redisplay_without_hooks () at redisplay.c:6556
#11 0x81b8ee6 in redisplay () at redisplay.c:6613
#12 0x8114eca in Fnext_event (event=141418308, prompt=137299972)
at event-stream.c:2176
#13 0x80a2ce9 in Fcommand_loop_1 () at cmdloop.c:574
#14 0x80a2f96 in command_loop_1 (dummy=137299972) at cmdloop.c:494
#15 0x80cac25 in condition_case_1 (handlers=137300068,
bfun=0x80a2f58 <command_loop_1>, barg=137299972,
hfun=0x80a3008 <cmd_error>, harg=137299972) at eval.c:1651
#16 0x80a30f7 in command_loop_2 (dummy=137299972) at cmdloop.c:256
#17 0x80cab24 in internal_catch (tag=137384396,
func=0x80a30bc <command_loop_2>, arg=137299972, threw=0x0) at eval.c:1317
#18 0x80a256e in initial_command_loop (load_me=137299972) at cmdloop.c:305
#19 0x80bcb9f in xemacs_21_2_b38_i386_unknown_freebsd4_2 (argc=2,
argv=0xbfbff5b4, envp=0xbfbff5c0, restart=0) at emacs.c:2253
#20 0x80bead1 in main (argc=2, argv=0xbfbff5b4, envp=0xbfbff5c0)
at emacs.c:2682
#21 0x80808e9 in _start ()
--
Yoshiaki Kasahara
Computing and Communications Center, Kyushu University
kasahara(a)nc.kyushu-u.ac.jp
12:10 p.m.
Yoshiaki Kasahara <kasahara(a)nc.kyushu-u.ac.jp> writes:
xemacs -vanilla
M-x make-frame-on-tty
C-x b test (in the X frame)
C-x b test (in the tty frame)
C-x k (in the X frame)
** crash **
It is repeatable. It won't crash when doing C-x k in the tty frame.
The problem is that replace-buffer-in-windws doesn't work
correctly when called on X. I can't understand how
window_loop can work under all conditions, so I made a
simpler functions to do it. It does more CONSing, but I
guess it's faster because you don't have to call
Fnext_window every time. Perhaps I should convert some type
of window_loop, too.
2000-12-18 Yoshiki Hayashi <yoshiki(a)xemacs.org>
* buffer.c (Fkill_buffer): Map over all devices.
* window.c (window_loop): Remove UNSHOW_BUFFER code.
(list_windows): New function.
(list_all_windows): Ditto.
(Freplace_buffer_in_windows): Use them.
Index: src/buffer.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/buffer.c,v
retrieving revision 1.36.2.45
diff -u -r1.36.2.45 buffer.c
--- buffer.c 2000/12/18 11:13:37 1.36.2.45
+++ buffer.c 2000/12/18 11:43:04
＠＠ -1259,7 +1259,9 ＠＠
/* #### This is a problem if this buffer is in a dedicated window.
Need to undedicate any windows of this buffer first (and delete them?)
*/
- Freplace_buffer_in_windows (buf, Qnil, Qnil);
+ GCPRO1 (buf);
+ Freplace_buffer_in_windows (buf, Qnil, Qall);
+ UNGCPRO;
font_lock_buffer_was_killed (b);
Index: src/window.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/window.c,v
retrieving revision 1.41.2.68
diff -u -r1.41.2.68 window.c
--- window.c 2000/12/18 11:13:43 1.41.2.68
+++ window.c 2000/12/18 11:43:17
＠＠ -2132,7 +2132,7 ＠＠
By default, only the windows in the selected frame are considered.
The optional argument WHICH-FRAMES changes this behavior:
WHICH-FRAMES = `visible' means search windows on all visible frames.
-WHICH-FRAMES = 0 means search windows on all visible and iconified frames.
+WHICH-FRAMES = 0 means search windows on all visible and iconified frames.
WHICH-FRAMES = t means search windows on all frames including invisible frames.
WHICH-FRAMES = a frame means search only windows on that frame.
Anything else means restrict to the selected frame.
＠＠ -2278,7 +2278,7 ＠＠
By default, only the windows in the selected frame are considered.
The optional argument WHICH-FRAMES changes this behavior:
WHICH-FRAMES = `visible' means search windows on all visible frames.
-WHICH-FRAMES = 0 means search windows on all visible and iconified frames.
+WHICH-FRAMES = 0 means search windows on all visible and iconified frames.
WHICH-FRAMES = t means search windows on all frames including invisible frames.
WHICH-FRAMES = a frame means search only windows on that frame.
Anything else means restrict to the selected frame.
＠＠ -2481,7 +2481,7 ＠＠
By default, only the windows in the selected frame are considered.
The optional argument WHICH-FRAMES changes this behavior:
WHICH-FRAMES = `visible' means search windows on all visible frames.
-WHICH-FRAMES = 0 means search windows on all visible and iconified frames.
+WHICH-FRAMES = 0 means search windows on all visible and iconified frames.
WHICH-FRAMES = t means search windows on all frames including invisible frames.
WHICH-FRAMES = a frame means search only windows on that frame.
Anything else means restrict to the selected frame.
＠＠ -2536,7 +2536,6 ＠＠
DELETE_OTHER_WINDOWS, /* Arg is window not to delete */
DELETE_BUFFER_WINDOWS, /* Arg is buffer */
GET_LARGEST_WINDOW,
- UNSHOW_BUFFER, /* Arg is buffer */
GET_BUFFER_WINDOW_COUNT, /* Arg is buffer */
GET_BUFFER_MRU_WINDOW /* Arg is buffer */
};
＠＠ -2549,7 +2548,7 ＠＠
int dedicated_too,
Lisp_Object which_devices)
{
- /* This function can GC if type == DELETE_BUFFER_WINDOWS or UNSHOW_BUFFER */
+ /* This function can GC if type == DELETE_BUFFER_WINDOWS */
Lisp_Object w;
Lisp_Object best_window = Qnil;
Lisp_Object next_window;
＠＠ -2621,7 +2620,6 ＠＠
for (;;)
{
struct window *p = XWINDOW (w);
- struct frame *w_frame = XFRAME (WINDOW_FRAME (p));
/* Pick the next window now, since some operations will delete
the current window. */
＠＠ -2771,57 +2769,6 ＠＠
break;
}
- case UNSHOW_BUFFER:
- {
- if (EQ (p->buffer, obj))
- {
- /* Find another buffer to show in this window. */
- Lisp_Object another_buffer =
- Fother_buffer (obj, Qnil, Qnil);
- if (NILP (another_buffer))
- another_buffer
- = Fget_buffer_create (QSscratch);
- /* If this window is dedicated, and in a frame
- of its own, kill the frame. */
- if (EQ (w, FRAME_ROOT_WINDOW (w_frame))
- && !NILP (p->dedicated)
- && other_visible_frames (w_frame))
- {
- /* Skip the other windows on this frame.
- There might be one, the minibuffer! */
- if (! EQ (w, last_window))
- while (w_frame == XFRAME (WINDOW_FRAME
- (XWINDOW (next_window))))
- {
- /* As we go, check for the end of the
- loop. We mustn't start going
- around a second time. */
- if (EQ (next_window, last_window))
- {
- last_window = w;
- break;
- }
- next_window = Fnext_window (next_window,
- mini ? Qt : Qnil,
- frame_arg, Qt);
- }
- /* Now we can safely delete the frame. */
- delete_frame_internal (XFRAME (WINDOW_FRAME (p)),
- 0, 0, 0);
- }
- else
- {
- /* Otherwise show a different buffer in the
- window. */
- p->dedicated = Qnil;
- Fset_window_buffer (w, another_buffer, Qnil);
- if (EQ (w, Fselected_window (Qnil)))
- Fset_buffer (p->buffer);
- }
- }
- break;
- }
-
default:
abort ();
}
＠＠ -3099,6 +3046,63 ＠＠
return Qnil;
}
+static Lisp_Object
+list_windows (struct window *w, Lisp_Object value)
+{
+ for (;;)
+ {
+ if (!NILP (w->hchild))
+ value = list_windows (XWINDOW (w->hchild), value);
+ else if (!NILP (w->vchild))
+ value = list_windows (XWINDOW (w->vchild), value);
+ else
+ {
+ Lisp_Object window;
+ XSETWINDOW (window, w);
+ value = Fcons (window, value);
+ }
+ if (NILP (w->next))
+ break;
+ w = XWINDOW (w->next);
+ }
+ return value;
+}
+
+static Lisp_Object
+list_all_windows (Lisp_Object frame_spec, Lisp_Object device_spec)
+{
+ Lisp_Object devcons, concons;
+ Lisp_Object retval = Qnil;
+
+ DEVICE_LOOP_NO_BREAK (devcons, concons)
+ {
+ Lisp_Object frame_list, the_window;
+ Lisp_Object device, tail;
+
+ device = XCAR (devcons);
+ frame_list = DEVICE_FRAME_LIST (XDEVICE (device));
+
+ LIST_LOOP (tail, frame_list)
+ {
+ if ((NILP (frame_spec)
+ && !EQ (XCAR (tail), DEVICE_SELECTED_FRAME (XDEVICE (device))))
+ || (EQ (frame_spec, Qvisible)
+ && !FRAME_VISIBLE_P (XFRAME (XCAR (tail))))
+ || (FRAMEP (frame_spec)
+ && !EQ (frame_spec, XCAR (tail)))
+ || (!NILP (frame_spec)
+ && !device_matches_device_spec (device,
+ NILP (device_spec) ?
+ Vselected_console :
+ device_spec)))
+ continue;
+ the_window = FRAME_ROOT_WINDOW (XFRAME (XCAR (tail)));
+ retval = list_windows (XWINDOW (the_window), retval);
+ }
+ }
+ return Fnreverse (retval);
+}
+
DEFUN ("replace-buffer-in-windows", Freplace_buffer_in_windows, 1, 3,
"bReplace buffer in windows: ", /*
Replace BUFFER with some other buffer in all windows showing it.
＠＠ -3124,18 +3128,46 ＠＠
(buffer, which_frames, which_devices))
{
/* This function can GC */
- buffer = Fget_buffer (buffer);
- CHECK_BUFFER (buffer);
+ Lisp_Object window_list;
+ Lisp_Object tail;
+ struct gcpro gcpro1, gcpro2;
- /* WHICH-FRAMES values t and nil mean the opposite of what
- window_loop expects. */
if (EQ (which_frames, Qnil))
which_frames = Qt;
else if (EQ (which_frames, Qt))
which_frames = Qnil;
+ window_list = list_all_windows (which_frames, which_devices);
- /* Ignore dedicated windows. */
- window_loop (UNSHOW_BUFFER, buffer, 0, which_frames, 0, which_devices);
+ buffer = Fget_buffer (buffer);
+ CHECK_BUFFER (buffer);
+
+ GCPRO2 (window_list, buffer);
+ LIST_LOOP (tail, window_list)
+ {
+ Lisp_Object window = XCAR (tail);
+ if (!MINI_WINDOW_P (XWINDOW (window))
+ && EQ (XWINDOW (window)->buffer, buffer))
+ {
+ Lisp_Object another_buffer = Fother_buffer (buffer, Qnil, Qnil);
+ Lisp_Object frame = WINDOW_FRAME (XWINDOW (window));
+ if (NILP (another_buffer))
+ another_buffer = Fget_buffer_create (QSscratch);
+ if (!NILP (XWINDOW (window)->dedicated)
+ && EQ (window,
+ FRAME_ROOT_WINDOW (XFRAME (frame)))
+ && other_visible_frames (XFRAME (frame)))
+ {
+ delete_frame_internal (XFRAME (frame), 0, 0, 0); /* GC */
+ }
+ else
+ {
+ Fset_window_buffer (window, another_buffer, Qnil);
+ if (EQ (window, Fselected_window (Qnil)))
+ Fset_buffer (XWINDOW (window)->buffer);
+ }
+ }
+ }
+ UNGCPRO;
return Qnil;
}
--
Yoshiki Hayashi
6:04 a.m.
On 18 Dec 2000 21:10:58 +0900,
Yoshiki Hayashi <yoshiki(a)xemacs.org> said:
2000-12-18 Yoshiki Hayashi <yoshiki(a)xemacs.org>
* buffer.c (Fkill_buffer): Map over all devices.
* window.c (window_loop): Remove UNSHOW_BUFFER code.
(list_windows): New function.
(list_all_windows): Ditto.
(Freplace_buffer_in_windows): Use them.
I updated my xemacs source using cvs update and built it today, but
the crash persists. I believe your patches are applied in my source
tree.
(gdb) where
#0 0x286323e0 in kill () from /usr/lib/libc.so.4
#1 0x80bb82f in fatal_error_signal (sig=6) at emacs.c:535
#2 0xbfbfffac in ?? ()
#3 0x80be875 in assert_failed (file=0x824dae7 "eval.c", line=1885,
expr=0x824d648 "abort()") at emacs.c:3214
#4 0x80c12fd in signal_1 (sig=137300068, data=140504844) at eval.c:1885
#5 0x80c279d in error (fmt=0x8285f00 "Marker does not point anywhere")
at eval.c:2052
#6 0x8194e86 in marker_position (marker=141242076) at marker.c:354
#7 0x81afc7f in redisplay_window (window=141749248, skip_selected=0)
at redisplay.c:5896
#8 0x81b09f4 in redisplay_frame (f=0x8738400, preemption_check=0)
at redisplay.c:6393
#9 0x81b0bd7 in redisplay_device (d=0x870a600, automatic=1)
at redisplay.c:6467
#10 0x81b0f45 in redisplay_without_hooks () at redisplay.c:6556
#11 0x81b8f4e in redisplay () at redisplay.c:6613
#12 0x8114ed6 in Fnext_event (event=141418076, prompt=137299972)
at event-stream.c:2176
#13 0x80a2cf5 in Fcommand_loop_1 () at cmdloop.c:574
#14 0x80a2fa2 in command_loop_1 (dummy=137299972) at cmdloop.c:494
#15 0x80cac31 in condition_case_1 (handlers=137300068,
bfun=0x80a2f64 <command_loop_1>, barg=137299972,
hfun=0x80a3014 <cmd_error>, harg=137299972) at eval.c:1651
#16 0x80a3103 in command_loop_2 (dummy=137299972) at cmdloop.c:256
#17 0x80cab30 in internal_catch (tag=137384396,
func=0x80a30c8 <command_loop_2>, arg=137299972, threw=0x0) at eval.c:1317
#18 0x80a257a in initial_command_loop (load_me=137299972) at cmdloop.c:305
#19 0x80bcbab in xemacs_21_2_b38_i386_unknown_freebsd4_2 (argc=2,
argv=0xbfbff5a8, envp=0xbfbff5b4, restart=0) at emacs.c:2253
#20 0x80beadd in main (argc=2, argv=0xbfbff5a8, envp=0xbfbff5b4)
at emacs.c:2682
#21 0x80808e9 in _start ()
--
Yoshiaki Kasahara
Computing and Communications Center, Kyushu University
kasahara(a)nc.kyushu-u.ac.jp
3:37 p.m.
Yoshiaki Kasahara <kasahara(a)nc.kyushu-u.ac.jp> writes:
> 2000-12-18 Yoshiki Hayashi <yoshiki(a)xemacs.org>
>
> * buffer.c (Fkill_buffer): Map over all devices.
> * window.c (window_loop): Remove UNSHOW_BUFFER code.
> (list_windows): New function.
> (list_all_windows): Ditto.
> (Freplace_buffer_in_windows): Use them.
I updated my xemacs source using cvs update and built it today, but
the crash persists. I believe your patches are applied in my source
tree.
Did you manually applied my patch? It's not in the CVS tree
yet. I usually wait one day before commiting my patch so
that others have chances to review it before it goes in.
--
Yoshiki Hayashi
7:18 a.m.
On 20 Dec 2000 00:37:05 +0900,
Yoshiki Hayashi <yoshiki(a)xemacs.org> said:
Did you manually applied my patch? It's not in the CVS tree
yet. I usually wait one day before commiting my patch so
that others have chances to review it before it goes in.
Ouch. I'm sorry I misunderstood. I just applied your patches
manually and the problem seemed to be fixed. I'll report if something
goes wrong.
--
Yoshiaki Kasahara
Computing and Communications Center, Kyushu University
kasahara(a)nc.kyushu-u.ac.jp
11:40 a.m.
Hello,
On 18 Dec 2000 21:10:58 +0900,
Yoshiki Hayashi <yoshiki(a)xemacs.org> said:
2000-12-18 Yoshiki Hayashi <yoshiki(a)xemacs.org>
* buffer.c (Fkill_buffer): Map over all devices.
* window.c (window_loop): Remove UNSHOW_BUFFER code.
(list_windows): New function.
(list_all_windows): Ditto.
(Freplace_buffer_in_windows): Use them.
I have yet another problem related to X and tty frames. It is not
caused by your fix, because I have some difficulties using Mew caused
by this problem for a couple of months. I feel that it may be a
similar problem, so I post it here.
I just noticed that when I had some X frames and a tty frame which
were displaying a BUFFER, and did (delete-windows-on BUFFER) on the
tty frame, sometimes the window of the BUFFER on the tty frame wasn't
deleted.
It is a bit difficult to reproduce. Still I don't find a simple
method. I'm using a couple of virtual consoles on fvwm2 window
manager. The problem usually happens when one of X frames is on a
hidden virtual console or iconified, and the hidden frame and the tty
frame both have a window of the BUFFER.
It doesn't always happen. Once it happens, it is easy to repeat.
--
Yoshiaki Kasahara
Computing and Communications Center, Kyushu University
kasahara(a)nc.kyushu-u.ac.jp
4:10 a.m.
Yoshiaki Kasahara <kasahara(a)nc.kyushu-u.ac.jp> writes:
I just noticed that when I had some X frames and a tty frame which
were displaying a BUFFER, and did (delete-windows-on BUFFER) on the
tty frame, sometimes the window of the BUFFER on the tty frame wasn't
^^^
You mean, X.
deleted.
I tested
(delete-windows-on "*tmp*" nil 'all)
on an X frame and a window on TTY wasn't deleted.
Note that you need non-nil and non-t value for WHICH-DEVICES
to map over every device.
I can repeat it 100%. I'll write a similar patch and test
cases.
--
Yoshiki Hayashi
5:14 a.m.
On 20 Dec 2000 13:10:06 +0900,
Yoshiki Hayashi <yoshiki(a)xemacs.org> said:
Yoshiaki Kasahara <kasahara(a)nc.kyushu-u.ac.jp> writes:
> I just noticed that when I had some X frames and a tty frame which
> were displaying a BUFFER, and did (delete-windows-on BUFFER) on the
> tty frame, sometimes the window of the BUFFER on the tty frame wasn't
^^^
You mean, X.
No, I really meant tty...
I just found a simple method to produce the problem.
xemacs -vanilla (make an X frame #1)
C-x 5 2 (to create another X frame #2)
C-x b test (on the X frame #1)
C-x b test (on the X frame #2)
M-x make-frame-on-tty (to create a tty frame)
the frame automatically display the buffer "test"
M-: (delete-windows-on "test") (on the tty frame)
The buffer "test" on the tty frame won't disappear. Additional
argument ("nil 'all") didn't help.
Actually iconify or hiding a frame is not really needed. The problem
seems not to occur when I'm using only one X frame and one tty frame.
I need two X frames.
I'm not sure if it is the same problem as you have found.
--Y.Kasahara
7:15 a.m.
Yoshiaki Kasahara <kasahara(a)nc.kyushu-u.ac.jp> writes:
> > tty frame, sometimes the window of the BUFFER on the tty
frame wasn't
> ^^^
> You mean, X.
No, I really meant tty...
I just found a simple method to produce the problem.
xemacs -vanilla (make an X frame #1)
C-x 5 2 (to create another X frame #2)
C-x b test (on the X frame #1)
C-x b test (on the X frame #2)
M-x make-frame-on-tty (to create a tty frame)
the frame automatically display the buffer "test"
M-: (delete-windows-on "test") (on the tty frame)
The buffer "test" on the tty frame won't disappear. Additional
argument ("nil 'all") didn't help.
I can't reproduce it. However, I wrote a patch to fix the
problems I found. Could you test and see whether this makes
any difference?
This patch is difficult to read because I moved list_windows
and list_all_windows upward. But the idea is same as my
previous patch.
2000-12-20 Yoshiki Hayashi <yoshiki(a)xemacs.org>
* window.c (window_loop): Remove DELTETE_BUFFER_WINDOWS_CODE.
* window.c (Fdelete_windows_on): Rewrite to use list_all_windows.
Index: window.c
===================================================================
RCS file: /usr/CVSroot/XEmacs/xemacs/src/window.c,v
retrieving revision 1.41.2.69
diff -u -r1.41.2.69 window.c
--- window.c 2000/12/20 04:50:48 1.41.2.69
+++ window.c 2000/12/20 06:58:19
＠＠ -2534,7 +2534,6 ＠＠
GET_BUFFER_WINDOW, /* Arg is buffer */
GET_LRU_WINDOW, /* Arg is t for full-width windows only */
DELETE_OTHER_WINDOWS, /* Arg is window not to delete */
- DELETE_BUFFER_WINDOWS, /* Arg is buffer */
GET_LARGEST_WINDOW,
GET_BUFFER_WINDOW_COUNT, /* Arg is buffer */
GET_BUFFER_MRU_WINDOW /* Arg is buffer */
＠＠ -2548,7 +2547,6 ＠＠
int dedicated_too,
Lisp_Object which_devices)
{
- /* This function can GC if type == DELETE_BUFFER_WINDOWS */
Lisp_Object w;
Lisp_Object best_window = Qnil;
Lisp_Object next_window;
＠＠ -2697,59 +2695,6 ＠＠
break;
}
- case DELETE_BUFFER_WINDOWS:
- {
- if (EQ (p->buffer, obj))
- {
- struct frame *f = XFRAME (WINDOW_FRAME (p));
-
- /* If this window is dedicated, and in a frame
- of its own, kill the frame. */
- if (EQ (w, FRAME_ROOT_WINDOW (f))
- && !NILP (p->dedicated)
- && other_visible_frames (f))
- {
- /* Skip the other windows on this frame.
- There might be one, the minibuffer! */
- if (! EQ (w, last_window))
- while (f == XFRAME (WINDOW_FRAME
- (XWINDOW (next_window))))
- {
- /* As we go, check for the end of the
- loop. We mustn't start going
- around a second time. */
- if (EQ (next_window, last_window))
- {
- last_window = w;
- break;
- }
- next_window = Fnext_window (next_window,
- mini ? Qt : Qnil,
- frame_arg, Qt);
- }
- /* Now we can safely delete the frame. */
- Fdelete_frame (WINDOW_FRAME (p), Qnil);
- }
- else
- /* If we're deleting the buffer displayed in
- the only window on the frame, find a new
- buffer to display there. */
- if (NILP (p->parent))
- {
- Lisp_Object new_buffer;
- new_buffer = Fother_buffer (obj, Qnil, Qnil);
- if (NILP (new_buffer))
- new_buffer = Fget_buffer_create (QSscratch);
- Fset_window_buffer (w, new_buffer, Qnil);
- if (EQ (w, Fselected_window (Qnil)))
- Fset_buffer (p->buffer);
- }
- else
- Fdelete_window (w, Qnil);
- }
- break;
- }
-
case GET_LARGEST_WINDOW:
{
/* Ignore dedicated windows and minibuffers. */
＠＠ -2813,6 +2758,63 ＠＠
#endif
+static Lisp_Object
+list_windows (struct window *w, Lisp_Object value)
+{
+ for (;;)
+ {
+ if (!NILP (w->hchild))
+ value = list_windows (XWINDOW (w->hchild), value);
+ else if (!NILP (w->vchild))
+ value = list_windows (XWINDOW (w->vchild), value);
+ else
+ {
+ Lisp_Object window;
+ XSETWINDOW (window, w);
+ value = Fcons (window, value);
+ }
+ if (NILP (w->next))
+ break;
+ w = XWINDOW (w->next);
+ }
+ return value;
+}
+
+static Lisp_Object
+list_all_windows (Lisp_Object frame_spec, Lisp_Object device_spec)
+{
+ Lisp_Object devcons, concons;
+ Lisp_Object retval = Qnil;
+
+ DEVICE_LOOP_NO_BREAK (devcons, concons)
+ {
+ Lisp_Object frame_list, the_window;
+ Lisp_Object device, tail;
+
+ device = XCAR (devcons);
+ frame_list = DEVICE_FRAME_LIST (XDEVICE (device));
+
+ LIST_LOOP (tail, frame_list)
+ {
+ if ((NILP (frame_spec)
+ && !EQ (XCAR (tail), DEVICE_SELECTED_FRAME (XDEVICE (device))))
+ || (EQ (frame_spec, Qvisible)
+ && !FRAME_VISIBLE_P (XFRAME (XCAR (tail))))
+ || (FRAMEP (frame_spec)
+ && !EQ (frame_spec, XCAR (tail)))
+ || (!NILP (frame_spec)
+ && !device_matches_device_spec (device,
+ NILP (device_spec) ?
+ Vselected_console :
+ device_spec)))
+ continue;
+ the_window = FRAME_ROOT_WINDOW (XFRAME (XCAR (tail)));
+ retval = list_windows (XWINDOW (the_window), retval);
+ }
+ }
+ return Fnreverse (retval);
+}
+
DEFUN ("get-lru-window", Fget_lru_window, 0, 2, 0, /*
Return the window least recently selected or used for display.
＠＠ -3030,6 +3032,10 ＠＠
(buffer, which_frames, which_devices))
{
/* This function can GC */
+ Lisp_Object window_list;
+ Lisp_Object tail;
+ struct gcpro gcpro1, gcpro2;
+
buffer = Fget_buffer (buffer);
CHECK_BUFFER (buffer);
＠＠ -3040,67 +3046,43 ＠＠
else if (EQ (which_frames, Qt))
which_frames = Qnil;
+ window_list = list_all_windows (which_frames, which_devices);
/* Ignore dedicated windows. */
- window_loop (DELETE_BUFFER_WINDOWS, buffer, 0,
- which_frames, 0, which_devices);
- return Qnil;
-}
-
-static Lisp_Object
-list_windows (struct window *w, Lisp_Object value)
-{
- for (;;)
- {
- if (!NILP (w->hchild))
- value = list_windows (XWINDOW (w->hchild), value);
- else if (!NILP (w->vchild))
- value = list_windows (XWINDOW (w->vchild), value);
- else
- {
- Lisp_Object window;
- XSETWINDOW (window, w);
- value = Fcons (window, value);
- }
- if (NILP (w->next))
- break;
- w = XWINDOW (w->next);
- }
- return value;
-}
-
-static Lisp_Object
-list_all_windows (Lisp_Object frame_spec, Lisp_Object device_spec)
-{
- Lisp_Object devcons, concons;
- Lisp_Object retval = Qnil;
-
- DEVICE_LOOP_NO_BREAK (devcons, concons)
+ GCPRO2 (window_list, buffer);
+ LIST_LOOP (tail, window_list)
{
- Lisp_Object frame_list, the_window;
- Lisp_Object device, tail;
-
- device = XCAR (devcons);
- frame_list = DEVICE_FRAME_LIST (XDEVICE (device));
-
- LIST_LOOP (tail, frame_list)
+ Lisp_Object window = XCAR (tail);
+ if (!MINI_WINDOW_P (XWINDOW (window))
+ && EQ (XWINDOW (window)->buffer, buffer))
{
- if ((NILP (frame_spec)
- && !EQ (XCAR (tail), DEVICE_SELECTED_FRAME (XDEVICE (device))))
- || (EQ (frame_spec, Qvisible)
- && !FRAME_VISIBLE_P (XFRAME (XCAR (tail))))
- || (FRAMEP (frame_spec)
- && !EQ (frame_spec, XCAR (tail)))
- || (!NILP (frame_spec)
- && !device_matches_device_spec (device,
- NILP (device_spec) ?
- Vselected_console :
- device_spec)))
- continue;
- the_window = FRAME_ROOT_WINDOW (XFRAME (XCAR (tail)));
- retval = list_windows (XWINDOW (the_window), retval);
+ Lisp_Object frame = WINDOW_FRAME (XWINDOW (window));
+ if (!NILP (XWINDOW (window)->dedicated)
+ && EQ (window, FRAME_ROOT_WINDOW (XFRAME (frame)))
+ && other_visible_frames (XFRAME (frame)))
+ {
+ Fdelete_frame (frame, Qnil); /* GC */
+ }
+ else
+ {
+ /* If we're deleting the buffer displayed in the only
+ window on the frame, find a new buffer to display
+ there. */
+ if (NILP (XWINDOW (window)->parent))
+ {
+ Lisp_Object another_buffer = Fother_buffer (buffer, Qnil, Qnil);
+ if (NILP (another_buffer))
+ another_buffer = Fget_buffer_create (QSscratch);
+ Fset_window_buffer (window, another_buffer, Qnil);
+ if (EQ (window, Fselected_window (Qnil)))
+ Fset_buffer (XWINDOW (window)->buffer);
+ }
+ else
+ Fdelete_window (window, Qnil);
+ }
}
}
- return Fnreverse (retval);
+ UNGCPRO;
+ return Qnil;
}
DEFUN ("replace-buffer-in-windows", Freplace_buffer_in_windows, 1, 3,
--
Yoshiki Hayashi
8:05 a.m.
On 20 Dec 2000 16:15:38 +0900,
Yoshiki Hayashi <yoshiki(a)xemacs.org> said:
I can't reproduce it. However, I wrote a patch to fix the
problems I found. Could you test and see whether this makes
any difference?
Yes. After your patch was applied, (delete-windows-on "test") on a
tty frame deleted the "test" window on the tty frame, and
(delete-windows-on "test" nil 'all) deleted all the windows including
both X and tty frames.
I didn't see any problems while a simple check using Mew on both X and
tty frames. I'll use this binary for a while to make sure.
Well, I noticed that the behavior of (delete-windows-on "test" t) is
slightly changed, though. Before the patch, it deleted a "test"
window only on the tty frame, but now it also deletes a window on one
of X frames (not both). When I do it on a X frame, a tty window is
deleted, too.
--Y.Kasahara
8:34 a.m.
Yoshiaki Kasahara <kasahara(a)nc.kyushu-u.ac.jp> writes:
Yes. After your patch was applied, (delete-windows-on
"test") on a
tty frame deleted the "test" window on the tty frame, and
(delete-windows-on "test" nil 'all) deleted all the windows including
both X and tty frames.
Thanks for confirmation.
Well, I noticed that the behavior of (delete-windows-on
"test" t) is
slightly changed, though. Before the patch, it deleted a "test"
window only on the tty frame, but now it also deletes a window on one
of X frames (not both). When I do it on a X frame, a tty window is
deleted, too.
Every device has its own selected-frame. I thought it would
be better to delete every selected-frame but it breaks
backward compatibility. Here's the patch to preserve old
behavior. Please apply this on top of my previous patch.
Probably I'll apply my two patches tomorrow to the CVS
repository.
--- window.c.patched Wed Dec 20 17:23:24 2000
+++ window.c Wed Dec 20 17:28:40 2000
＠＠ -2790,6 +2790,7 ＠＠
{
Lisp_Object frame_list, the_window;
Lisp_Object device, tail;
+ struct frame * the_selected_frame = selected_frame ();
device = XCAR (devcons);
frame_list = DEVICE_FRAME_LIST (XDEVICE (device));
＠＠ -2797,7 +2798,7 ＠＠
LIST_LOOP (tail, frame_list)
{
if ((NILP (frame_spec)
- && !EQ (XCAR (tail), DEVICE_SELECTED_FRAME (XDEVICE (device))))
+ && XFRAME (XCAR (tail)) != the_selected_frame)
|| (EQ (frame_spec, Qvisible)
&& !FRAME_VISIBLE_P (XFRAME (XCAR (tail))))
|| (FRAMEP (frame_spec)
--
Yoshiki Hayashi
3:22 p.m.
New subject: erroneous 'delete-windows-on' behaviour (Re: [crash] delete a buffer shown in both X and tty frames)
On 20 Dec 2000 16:15:38 +0900,
Yoshiki Hayashi <yoshiki(a)xemacs.org> said:
2000-12-20 Yoshiki Hayashi <yoshiki(a)xemacs.org>
* window.c (window_loop): Remove DELTETE_BUFFER_WINDOWS_CODE.
* window.c (Fdelete_windows_on): Rewrite to use list_all_windows.
I believe this patch is not applied yet, because of Martin's comment
about the implementation.
Mew uses 'delete-windows-on' to reconfigure the placement of buffers,
and the bug sometimes bites me while I'm reading emails using Mew from
a remote terminal through a tty frame. I'm really frustrated because
I can't recover from the problem until I return to the console or
start another xemacs.
Is it possible to fix the bug before the release of 21.4?
--
Yoshiaki Kasahara
Computing and Communications Center, Kyushu University
kasahara(a)nc.kyushu-u.ac.jp
3:58 a.m.
>>>> "Y" == Yoshiki Hayashi
<yoshiki(a)xemacs.org> writes:
Y> Yoshiaki Kasahara <kasahara(a)nc.kyushu-u.ac.jp> writes:
> xemacs -vanilla
> M-x make-frame-on-tty
> C-x b test (in the X frame)
> C-x b test (in the tty frame)
> C-x k (in the X frame)
> ** crash **
>
> It is repeatable. It won't crash when doing C-x k in the tty frame.
Y> The problem is that replace-buffer-in-windows doesn't work
Y> correctly when called on X. I can't understand how
Y> window_loop can work under all conditions, so I made a
Y> simpler functions to do it. It does more CONSing, but I
Y> guess it's faster because you don't have to call
Y> Fnext_window every time. Perhaps I should convert some type
Y> of window_loop, too.
I recently rewrote much of this code (to fix some old crashes!) and it
is possible I introduced new ones.
I also had trouble understanding window_loop(). But I think we should
either replace it completely or fix it.
Y> 2000-12-18 Yoshiki Hayashi <yoshiki(a)xemacs.org>
Y> * buffer.c (Fkill_buffer): Map over all devices.
Y> * window.c (window_loop): Remove UNSHOW_BUFFER code.
Y> (list_windows): New function.
Y> (list_all_windows): Ditto.
Y> (Freplace_buffer_in_windows): Use them.
Y> Index: src/buffer.c
Y> ===================================================================
Y> RCS file: /usr/CVSroot/XEmacs/xemacs/src/buffer.c,v
Y> retrieving revision 1.36.2.45
Y> diff -u -r1.36.2.45 buffer.c
Y> --- buffer.c 2000/12/18 11:13:37 1.36.2.45
Y> +++ buffer.c 2000/12/18 11:43:04
Y> ＠＠ -1259,7 +1259,9 ＠＠
Y> /* #### This is a problem if this buffer is in a dedicated window.
Y> Need to undedicate any windows of this buffer first (and delete them?)
Y> */
Y> - Freplace_buffer_in_windows (buf, Qnil, Qnil);
Y> + GCPRO1 (buf);
Y> + Freplace_buffer_in_windows (buf, Qnil, Qall);
Y> + UNGCPRO;
Can you prove GCPROing is necessary here? Perhaps
Freplace_buffer_in_windows() should be called before
delete_from_buffer_alist(), which otherwise GCPROtects buf? The code
has tried to prove that `buf' is a live buffer here.
Y> +static Lisp_Object
Y> +list_windows (struct window *w, Lisp_Object value)
I don't like the idea of consing so much here. Admittedly, it's not
that important here, since buffers don't get killed very often. But
as Yoshiki knows, I would like XEmacs to be efficient when there are
many buffers. I would like lisp programmers to feel free to create
and destroy temporary buffers frequently.
Perhaps you could write a clever WINDOW_LOOP macro? Alternatively,
use a function like map_windows, except extended to handle frame_spec
and device_spec?
8653
days inactive
8743
days old
12 comments
3 participants
participants (3)
-
Martin Buchholz -
Yoshiaki Kasahara -
Yoshiki Hayashi