Args out of range with new auto-composition implementation

Args out of range with new auto-composition implementation

[Juanma Barranquero]

An error I'm getting since the auto-composition changes of a few days ago:

;;; testbug.el ;;;
(column-number-mode 1)

(defun testbug ()
  (interactive)
  (split-window)
  (other-window 1)
  (pop-to-buffer (get-buffer-create "testbug")))
;;;;;;;;;;;;;;;;;

emacs -Q -l testbug.el
C-h H
M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
M-x testbug <RET>

  => "Args out of range: 274, 274"

(and the windows are not correctly painted.)

Setting a breakpoint at data.c:args_out_of_range, you get this backtrace:

Breakpoint 3, args_out_of_range (a1=2192, a2=2192) at data.c:124
124       xsignal2 (Qargs_out_of_range, a1, a2);
(gdb) bt
#0  args_out_of_range (a1=2192, a2=2192) at data.c:124
#1  0x0119556b in validate_interval_range (object=51728900,
begin=0x82dd90, end=0x82dd90, force=0) at textprop.c:154
#2  0x01197467 in Ftext_properties_at (position=2192, object=51728896)
at textprop.c:588
#3  0x01197537 in Fget_text_property (position=2192, prop=47716689,
object=51728900) at textprop.c:609
#4  0x011ed5fc in face_at_buffer_position (w=0x351fe00, pos=274,
region_beg=0, region_end=0, endptr=0x82dec8, limit=276,
    mouse=0) at xfaces.c:6278
#5  0x011a48d1 in font_range (pos=274, limit=0x82df38, w=0x351fe00,
face=0x0, string=47716353) at font.c:3678
#6  0x0119b7e4 in autocmp_chars (cft_element=<value optimized out>,
charpos=274, bytepos=346, limit=276, win=0x351fe00,
    face=0x0, string=47716353) at composite.c:944
#7  0x011a00ca in composition_reseat_it (cmp_it=0x82e014, charpos=274,
bytepos=346, endpos=276, w=0x351fe00, face=0x0,
    string=47716353) at composite.c:1081
#8  0x011705a9 in scan_for_column (endpos=0x82e0c4, goalcol=0x82e0c8,
prevcol=0x0) at indent.c:597
#9  0x01170d85 in current_column () at indent.c:725
#10 0x01053467 in redisplay_window (window=62406148,
just_this_one_p=0) at xdisp.c:13315
#11 0x010566cc in redisplay_window_0 (window=62406148) at xdisp.c:12206
#12 0x01018e71 in internal_condition_case_1 (bfun=0x10566a9
<redisplay_window_0>, arg=62406148, handlers=47699205,
    hfun=0x1023dd2 <redisplay_window_error>) at eval.c:1559
#13 0x0102e0d4 in redisplay_windows (window=274) at xdisp.c:12185
#14 0x0102e046 in redisplay_windows (window=274) at xdisp.c:12179
#15 0x010477cd in redisplay_internal (preserve_echo_area=<value
optimized out>) at xdisp.c:11761
#16 0x0108b159 in read_char (commandflag=1, nmaps=2, maps=0x82fb70,
prev_event=47716353, used_mouse_menu=0x82fc34, end_time=0x0)
    at keyboard.c:2648
#17 0x0108f7aa in read_key_sequence (keybuf=0x82fcd4, bufsize=30,
prompt=47716353, dont_downcase_last=0,
    can_return_switch_frame=1, fix_current_buffer=1) at keyboard.c:9345
#18 0x0109291d in command_loop_1 () at keyboard.c:1620
#19 0x010190f6 in internal_condition_case (bfun=0x109268f
<command_loop_1>, handlers=47780105, hfun=0x10898f6 <cmd_error>)
    at eval.c:1511
#20 0x01088d9b in command_loop_2 () at keyboard.c:1338
#21 0x010191a0 in internal_catch (tag=47776177, func=0x1088d78
<command_loop_2>, arg=47716353) at eval.c:1247
#22 0x0108973b in command_loop () at keyboard.c:1317
#23 0x01089a8f in recursive_edit_1 () at keyboard.c:942
#24 0x01089bfa in Frecursive_edit () at keyboard.c:1004
#25 0x01002c41 in main (argc=4, argv=0xa927a8) at emacs.c:1693

This is on Windows, MinGW 4.3.0, optimized build.

 Juanma

bug#874: Args out of range with new auto-composition implementation

[Juanma Barranquero]

;;; testbug.el ;;;
(column-number-mode 1)

(defun testbug ()
  (interactive)
  (split-window)
  (other-window 1)
  (pop-to-buffer (get-buffer-create "testbug")))
;;;;;;;;;;;;;;;;;

emacs -Q -l testbug.el
C-h H
M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
M-x testbug <RET>

  => "Args out of range: 274, 274"

(and the windows are not correctly painted.)

At home's laptop I can reproduce the bug at will; as a weird
datapoint, at work I've only been able to reproduce it when the buffer
is called "*ielm*", even if IELM is *not* loaded.

emacs-devel discussion (including a backtrace):

http://lists.gnu.org/archive/html/emacs-devel/2008-09/msg00045.html

bug#874: Args out of range with new auto-composition implementation

[Juanma Barranquero]

On Wed, Sep 3, 2008 at 18:21, Juanma Barranquero <lekktu@gmail.com> wrote:

> emacs -Q -l testbug.el
> C-h H
> M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
> M-x testbug <RET>
>
>  => "Args out of range: 274, 274"

Curious. The out of range error is caused by this code in
textprop.c:validate_interval_range

      if (!(BUF_BEGV (b) <= XINT (*begin) && XINT (*begin) <= XINT (*end)
	    && XINT (*end) <= BUF_ZV (b)))
	  args_out_of_range (*begin, *end);

At that point,

  BUF_BEGV (b)  = 1
  BUF_ZV (b)    = 1
  XINT (*begin) = 274
  XINT (*end)   = 274

where *begin and *end point to the position in buffer "HELLO", but b
(and hence object, from which b is extracted) points to the just
created buffer "testbug":

(gdb) p object
$1 = 64949764
(gdb) xbuffer
$2 = (struct buffer *) 0x3df0e00
(unsigned char *) 0x3de6880 "testbug"

so it's no wonder it fails.

Re: Args out of range with new auto-composition implementation

[Kenichi Handa]

In article <f7ccd24b0809011639n7e92101ey22dd80c600dce97@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:

> An error I'm getting since the auto-composition changes of a few days ago:
> ;;; testbug.el ;;;
> (column-number-mode 1)

> (defun testbug ()
>   (interactive)
>   (split-window)
>   (other-window 1)
>   (pop-to-buffer (get-buffer-create "testbug")))
> ;;;;;;;;;;;;;;;;;

> emacs -Q -l testbug.el
> C-h H
> M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
> M-x testbug <RET>

>>> "Args out of range: 274, 274"

I can't reproduce that bug with the latest code.  Can you?

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Args out of range with new auto-composition implementation

[Juanma Barranquero]

On Thu, Sep 11, 2008 at 04:07, Kenichi Handa <handa@m17n.org> wrote:

> I can't reproduce that bug with the latest code.  Can you?

Yes.


 Juanma

Re: Args out of range with new auto-composition implementation

[Juanma Barranquero]

On Thu, Sep 11, 2008 at 04:07, Kenichi Handa <handa@m17n.org> wrote:

> I can't reproduce that bug with the latest code.  Can you?

BTW, did you see the additional info at
http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=874#10

   Juanma

Re: Args out of range with new auto-composition implementation

[Ted Zlatanov]

On Thu, 11 Sep 2008 11:07:18 +0900 Kenichi Handa <handa@m17n.org> wrote: 

KH> In article <f7ccd24b0809011639n7e92101ey22dd80c600dce97@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:
>> An error I'm getting since the auto-composition changes of a few days ago:
>> ;;; testbug.el ;;;
>> (column-number-mode 1)

>> (defun testbug ()
>> (interactive)
>> (split-window)
>> (other-window 1)
>> (pop-to-buffer (get-buffer-create "testbug")))
>> ;;;;;;;;;;;;;;;;;

>> emacs -Q -l testbug.el
>> C-h H
>> M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
>> M-x testbug <RET>

>>>> "Args out of range: 274, 274"

KH> I can't reproduce that bug with the latest code.  Can you?

I noticed something very similar in CVS Gnus with CVS Emacs from
yesterday.  See http://thread.gmane.org/gmane.emacs.gnus.general/67353

For now I've wrapped the problem with a condition-case in Gnus...
Definitely not the right solution, but I can enter groups now :)

Ted

Re: Args out of range with new auto-composition implementation

[Kenichi Handa]

In article <f7ccd24b0809101921o1a8eb478qd5c867c79b0bcdee@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:

> On Thu, Sep 11, 2008 at 04:07, Kenichi Handa <handa@m17n.org> wrote:
> > I can't reproduce that bug with the latest code.  Can you?

> BTW, did you see the additional info at
> http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=874#10

Yes.  The only possibility that happens is that w->buffer is
different from current_buffer.  I've thought that when
current_column is called, the selected window is that same
as the window showing the current buffer.  But, it seems
that we can't assume that.  Perhaps that is the reason why
it uses buffer-display-table of the current buffer instead
of window-display-table as in compute_motion().

Anyway, I've just installed a workaround of not handling
automatic composition if the current buffer is not
associated with a window.

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Args out of range with new auto-composition implementation

[Juanma Barranquero]

On Fri, Sep 12, 2008 at 08:32, Kenichi Handa <handa@m17n.org> wrote:

> Anyway, I've just installed a workaround of not handling
> automatic composition if the current buffer is not
> associated with a window.

That fixes the problem. Thanks.

 Juanma

bug#874: marked as done (Args out of range with new auto-composition implementation)

[Emacs bug Tracking System]

[-- Attachment #1: Type: text/plain, Size: 908 bytes --]


Your message dated Fri, 12 Sep 2008 10:56:39 +0200
with message-id <f7ccd24b0809120156o45ab375dhd2c6097e1c375890@mail.gmail.com>
and subject line Re: Args out of range with new auto-composition implementation
has caused the Emacs bug report #874,
regarding Args out of range with new auto-composition implementation
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact don@donarmstrong.com
immediately.)


-- 
874: http://emacsbugs.donarmstrong.com/cgi-bin/bugreport.cgi?bug=874
Emacs Bug Tracking System
Contact don@donarmstrong.com with problems

[-- Attachment #2: Type: message/rfc822, Size: 2899 bytes --]

From: "Juanma Barranquero" <lekktu@gmail.com>
To: "Emacs Bug Tracker" <submit@emacsbugs.donarmstrong.com>
Subject: Args out of range with new auto-composition implementation
Date: Wed, 3 Sep 2008 18:21:51 +0200
Message-ID: <f7ccd24b0809030921q643779e8t28aae7d9cf1b6d44@mail.gmail.com>

;;; testbug.el ;;;
(column-number-mode 1)

(defun testbug ()
  (interactive)
  (split-window)
  (other-window 1)
  (pop-to-buffer (get-buffer-create "testbug")))
;;;;;;;;;;;;;;;;;

emacs -Q -l testbug.el
C-h H
M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
M-x testbug <RET>

  => "Args out of range: 274, 274"

(and the windows are not correctly painted.)

At home's laptop I can reproduce the bug at will; as a weird
datapoint, at work I've only been able to reproduce it when the buffer
is called "*ielm*", even if IELM is *not* loaded.

emacs-devel discussion (including a backtrace):

http://lists.gnu.org/archive/html/emacs-devel/2008-09/msg00045.html



[-- Attachment #3: Type: message/rfc822, Size: 2818 bytes --]

From: "Juanma Barranquero" <lekktu@gmail.com>
To: "Kenichi Handa" <handa@m17n.org>
Cc: emacs-devel@gnu.org, 874-done@emacsbugs.donarmstrong.com
Subject: Re: Args out of range with new auto-composition implementation
Date: Fri, 12 Sep 2008 10:56:39 +0200
Message-ID: <f7ccd24b0809120156o45ab375dhd2c6097e1c375890@mail.gmail.com>

On Fri, Sep 12, 2008 at 08:32, Kenichi Handa <handa@m17n.org> wrote:

> Anyway, I've just installed a workaround of not handling
> automatic composition if the current buffer is not
> associated with a window.

That fixes the problem. Thanks.

 Juanma

Re: Args out of range with new auto-composition implementation

[Kenichi Handa]

In article <863ak6wxjq.fsf@lifelogs.com>, Ted Zlatanov <tzz@lifelogs.com> writes:

> On Thu, 11 Sep 2008 11:07:18 +0900 Kenichi Handa <handa@m17n.org> wrote: 
KH> In article <f7ccd24b0809011639n7e92101ey22dd80c600dce97@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:
>>> An error I'm getting since the auto-composition changes of a few days ago:
>>> ;;; testbug.el ;;;
>>> (column-number-mode 1)

>>> (defun testbug ()
>>> (interactive)
>>> (split-window)
>>> (other-window 1)
>>> (pop-to-buffer (get-buffer-create "testbug")))
>>> ;;;;;;;;;;;;;;;;;

>>> emacs -Q -l testbug.el
>>> C-h H
>>> M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
>>> M-x testbug <RET>

>>>>> "Args out of range: 274, 274"

KH> I can't reproduce that bug with the latest code.  Can you?

> I noticed something very similar in CVS Gnus with CVS Emacs from
> yesterday.  See http://thread.gmane.org/gmane.emacs.gnus.general/67353

> For now I've wrapped the problem with a condition-case in Gnus...
> Definitely not the right solution, but I can enter groups now :)

Do you still see the same problem with the latest code when
you disable that wrapping?

---
Kenichi Handa
handa@ni.aist.go.jp

Re: Args out of range with new auto-composition implementation

[Ted Zlatanov]

On Wed, 24 Sep 2008 15:07:53 +0900 Kenichi Handa <handa@m17n.org> wrote: 

KH> In article <863ak6wxjq.fsf@lifelogs.com>, Ted Zlatanov <tzz@lifelogs.com> writes:
>> On Thu, 11 Sep 2008 11:07:18 +0900 Kenichi Handa <handa@m17n.org> wrote: 
KH> In article <f7ccd24b0809011639n7e92101ey22dd80c600dce97@mail.gmail.com>, "Juanma Barranquero" <lekktu@gmail.com> writes:
>>>> An error I'm getting since the auto-composition changes of a few days ago:
>>>> ;;; testbug.el ;;;
>>>> (column-number-mode 1)

>>>> (defun testbug ()
>>>> (interactive)
>>>> (split-window)
>>>> (other-window 1)
>>>> (pop-to-buffer (get-buffer-create "testbug")))
>>>> ;;;;;;;;;;;;;;;;;

>>>> emacs -Q -l testbug.el
>>>> C-h H
>>>> M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
>>>> M-x testbug <RET>

>>>>>> "Args out of range: 274, 274"

KH> I can't reproduce that bug with the latest code.  Can you?

>> I noticed something very similar in CVS Gnus with CVS Emacs from
>> yesterday.  See http://thread.gmane.org/gmane.emacs.gnus.general/67353

>> For now I've wrapped the problem with a condition-case in Gnus...
>> Definitely not the right solution, but I can enter groups now :)

KH> Do you still see the same problem with the latest code when
KH> you disable that wrapping?

Yes.  Look at gnus-sum.el:gnus-summary-insert-line and remove the
condition-case wrapper to see the error.

Ted

Re: Args out of range with new auto-composition implementation

[Kenichi Handa]

In article <86myhfrpkq.fsf@lifelogs.com>, Ted Zlatanov <tzz@lifelogs.com> writes:

>>>>> emacs -Q -l testbug.el
>>>>> C-h H
>>>>> M-: (goto-char 276) <RET>   ;; or move the cursor to any auto-composed char
>>>>> M-x testbug <RET>

>>>>>>> "Args out of range: 274, 274"

KH> I can't reproduce that bug with the latest code.  Can you?

>>> I noticed something very similar in CVS Gnus with CVS Emacs from
>>> yesterday.  See http://thread.gmane.org/gmane.emacs.gnus.general/67353

>>> For now I've wrapped the problem with a condition-case in Gnus...
>>> Definitely not the right solution, but I can enter groups now :)

KH> Do you still see the same problem with the latest code when
KH> you disable that wrapping?

> Yes.  Look at gnus-sum.el:gnus-summary-insert-line and remove the
> condition-case wrapper to see the error.

Please do C-u - M-x global-auto-composition-mode RET and try
again.  If you still see "Args out of range" error, I think
it's not related to the changes for auto-composition.

---
Kenichi Handa
handa@ni.aist.go.jp