A redisplay bug in 22.0.99?

A redisplay bug in 22.0.99?

[Juanma Barranquero]

With 22.0.99 on Windows:

emacs -Q

C-x d etc/images/ <RET>
C-n C-o       => display README
C-5 C-n C-o   => display cancel.pbm

cancel.pbm is shown as an empty box in the lower window, which is not
correctly redrawn (it has leftover pixels from the previous contents).

Does anyone else see this?

             Juanma

RE: A redisplay bug in 22.0.99?

[Drew Adams]

>> Does anyone else see this?

I see it, in GNU Emacs 22.0.96.1 (i386-mingw-nt5.1.2600)
 of 2007-03-21 on LENNART-69DE564
Windowing system distributor `Microsoft Corp.', version 5.1.2600
configured using `configure --with-gcc (3.4) --cflags -Ic:/g/include'

Re: A redisplay bug in 22.0.99?

[Jason Rumney]

Juanma Barranquero wrote:
> cancel.pbm is shown as an empty box in the lower window, which is not
> correctly redrawn (it has leftover pixels from the previous contents).
>
> Does anyone else see this?

Not displaying pgm images correctly is caused by the following recent 
change:

2007-03-18  Chong Yidong  <cyd@stupidchicken.com>

    * image.c (pbm_load): Signal error for invalid image size.

I'm not sure whether the garbaged display after an image error is w32 
specific or not. Before release it is probably best to at least verify 
that we are not drawing from uninitialised memory in this case, as that 
could cause a crash.

Re: A redisplay bug in 22.0.99?

[Johan Bockgård]

Jason Rumney <jasonr@gnu.org> writes:

> I'm not sure whether the garbaged display after an image error is w32
> specific or not.

I see the problem with GNU Emacs 22.1.50.5 (x86_64-unknown-linux-gnu,
X toolkit) of 2007-05-11 on muon

   emacs -Q etc/images/cancel.pbm

Instead of the correct image I see an image of a mouse cursor (arrow,
I, or busy cursor).

-- 
Johan Bockgård

Re: A redisplay bug in 22.0.99?

[Eli Zaretskii]

> Date: Fri, 11 May 2007 16:41:52 +0200
> From: "Juanma Barranquero" <lekktu@gmail.com>
> 
> emacs -Q
> 
> C-x d etc/images/ <RET>
> C-n C-o       => display README
> C-5 C-n C-o   => display cancel.pbm
> 
> cancel.pbm is shown as an empty box in the lower window, which is not
> correctly redrawn (it has leftover pixels from the previous contents).

Could it be that cancel.pbm is corrupted?  Can you display it using
some other program?

Re: A redisplay bug in 22.0.99?

[Juanma Barranquero]

On 5/11/07, Eli Zaretskii <eliz@gnu.org> wrote:

> Could it be that cancel.pbm is corrupted?

If so, it is corrupted on the CVS repository; I just checked it out anew.

> Can you display it using some other program?

I don't have any other program able to display .pbm images (that I know of).

Do you see the problem?

             Juanma

Re: A redisplay bug in 22.0.99?

[David Kastrup]

"Juanma Barranquero" <lekktu@gmail.com> writes:

> On 5/11/07, Eli Zaretskii <eliz@gnu.org> wrote:
>
>> Could it be that cancel.pbm is corrupted?
>
> If so, it is corrupted on the CVS repository; I just checked it out
> anew.

You are a Windows user, so it could also be a CVS/line ending/binary
problem not seen on other platforms.

> I don't have any other program able to display .pbm images (that I know of).
>
> Do you see the problem?

Not on my current 22.0.91 and I doubt there have been changes since then.

-- 
David Kastrup

Re: A redisplay bug in 22.0.99?

[Jason Rumney]

David Kastrup wrote:
> Not on my current 22.0.91 and I doubt there have been changes since then.
When was 22.0.91 current? We are up to 22.0.99 now, and the change 
causing this problem was made on 18 March.

Re: A redisplay bug in 22.0.99?

[Eli Zaretskii]

> Cc: "Eli Zaretskii" <eliz@gnu.org>, emacs-devel@gnu.org
> From: David Kastrup <dak@gnu.org>
> Date: Fri, 11 May 2007 18:06:16 +0200
> 
> You are a Windows user, so it could also be a CVS/line ending/binary
> problem not seen on other platforms.

No, it isn't: I checkout all files with -kb, and I still see the
problem.

> > Do you see the problem?
> 
> Not on my current 22.0.91 and I doubt there have been changes since then.

In 22.0.91 I don't see the problem on Windows as well.

Re: A redisplay bug in 22.0.99?

[Eli Zaretskii]

> Date: Fri, 11 May 2007 17:59:09 +0200
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: emacs-devel@gnu.org
> 
> Do you see the problem?

Yes.  But other *.pbm files display okay.

And Gimp seems to display cancel.pbm allright, so the file is fine.

Re: A redisplay bug in 22.0.99?

[Andreas Schwab]

Eli Zaretskii <eliz@gnu.org> writes:

>> Date: Fri, 11 May 2007 16:41:52 +0200
>> From: "Juanma Barranquero" <lekktu@gmail.com>
>> 
>> emacs -Q
>> 
>> C-x d etc/images/ <RET>
>> C-n C-o       => display README
>> C-5 C-n C-o   => display cancel.pbm
>> 
>> cancel.pbm is shown as an empty box in the lower window, which is not
>> correctly redrawn (it has leftover pixels from the previous contents).
>
> Could it be that cancel.pbm is corrupted? 

The problem appears to be that this is actually a PGM.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
PGP key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: A redisplay bug in 22.0.99?

[James Cloos]

>>>>> "Andreas" == Andreas Schwab <schwab@suse.de> writes:

Andreas> The problem appears to be that this is actually a PGM.

I thought that was the (entirety) of the problem as well, but simply
renameing the file to cancel.pgm does not let emacs display it
correctly.

Running pnmnoraw on it, and saving that to cancel.pgm does fix the
file so that emacs can display it.

The file is a correct P5 raw PGM file.  It starts out with:

,----[ head -4 /cvs-src/emacs/etc/images/cancel.pbm ]
| P5
| # CREATOR: The GIMP's PNM Filter Version 1.0
| 24 24
| 255
`----

and the rest of the files is 576 octets, exactly correct for a 24x24
pixel P5 with maxval=255.

Removing the comment does not make a difference.

So emacs must have a bug displaying raw (P5 magic) PGM files that is
not tripped with ascii (P2 magic) PGM files.

Converting it to a proper PBM (P4 magic) also works.  Just run
pgmtopbm from the netpbm package on it.

-JimC
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Re: A redisplay bug in 22.0.99?

[Glenn Morris]

James Cloos wrote:

>>>>>> "Andreas" == Andreas Schwab <schwab@suse.de> writes:
>
> Andreas> The problem appears to be that this is actually a PGM.
>
> I thought that was the (entirety) of the problem as well, but simply
> renameing the file to cancel.pgm does not let emacs display it
> correctly.


Perhaps something like this.


*** image.c	18 Mar 2007 23:09:57 -0000	1.65
--- image.c	11 May 2007 22:37:36 -0000
***************
*** 5742,5750 ****
      }
    else
      {
!       if (raw_p && (p + 3 * height * width > end))
  	{
  	  x_destroy_x_image (ximg);
  	  image_error ("Invalid image size in image `%s'",
  		       img->spec, Qnil);
  	  goto error;
--- 5742,5751 ----
      }
    else
      {
!       if (raw_p && (p + (type == PBM_GRAY ? 1 : 3) * height * width > end))
  	{
  	  x_destroy_x_image (ximg);
+ 	  x_clear_image (f, img);
  	  image_error ("Invalid image size in image `%s'",
  		       img->spec, Qnil);
  	  goto error;

Re: A redisplay bug in 22.0.99?

[James Cloos]

>>>>> "Glenn" == Glenn Morris <rgm@gnu.org> writes:

Glenn> Perhaps something like this.

,----
| 
| *** image.c	18 Mar 2007 23:09:57 -0000	1.65
| --- image.c	11 May 2007 22:37:36 -0000
| ***************
| *** 5742,5750 ****
|       }
|     else
|       {
| !       if (raw_p && (p + 3 * height * width > end))
|   	{
|   	  x_destroy_x_image (ximg);
|   	  image_error ("Invalid image size in image `%s'",
|   		       img->spec, Qnil);
|   	  goto error;
| --- 5742,5751 ----
|       }
|     else
|       {
| !       if (raw_p && (p + (type == PBM_GRAY ? 1 : 3) * height * width > end))
|   	{
|   	  x_destroy_x_image (ximg);
| + 	  x_clear_image (f, img);
|   	  image_error ("Invalid image size in image `%s'",
|   		       img->spec, Qnil);
|   	  goto error;
`----

Yes, that does look like a necessary fix.

-JimC

Re: A redisplay bug in 22.0.99?

[Glenn Morris]

Glenn Morris wrote:

> Perhaps something like this.

Alternatively, since the check is not made for PBM_MONO, maybe it
should just not be made for PBM_GRAY? Alternatively alternatively,
maybe PBM_MONO should be checked in the same way?

It was your change originally, can you comment?

Re: A redisplay bug in 22.0.99?

[Chong Yidong]

Glenn Morris <rgm@gnu.org> writes:

> Glenn Morris wrote:
>
>> Perhaps something like this.
>
> Alternatively, since the check is not made for PBM_MONO, maybe it
> should just not be made for PBM_GRAY? Alternatively alternatively,
> maybe PBM_MONO should be checked in the same way?
>
> It was your change originally, can you comment?

I don't know the pbm format very well; any patch that fixes this bug
is fine by me, as long as the original bug remains fixed:

http://lists.gnu.org/archive/html/emacs-pretest-bug/2007-03/msg00335.html

That bug actually caused Emacs to crash, rather than refusing to
display an image.