On 11/02/2016 03:53 PM, Dave Howorth wrote:

On Wed, 2 Nov 2016 21:12:08 +0100 Urs Beyerle wrote:

...

Hi,

Since a few day(?) we can no longer open ps or eps files with okular and evince. gv still works. We see it on 42.1 and 13.2, but not 13.1. Open any ps or eps file with okular gives an empty page and

okular(6579)/okular (Spectre) GSRendererThread::run: Generated image does not match wanted size: [0x0] vs requested [1120x1449] QImage::scaled: Image is a null image invalidaccess -7

evince just crashes with

invalidaccess -7 Segmentation fault

Can someone else reproduce this problem?

I never heard of a .ps or .eps file. Where are they coming from? Can you post a link to one or both so as to test the software I and others may have? --doug -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Thu, 3 Nov 2016 00:52:33 +0100 "Carlos E. R." wrote:

eps is encapsulated postscript.

Or expanded polystyrene :) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 11/02/2016 10:39 PM, jdd wrote:

Le 02/11/2016 à 21:12, Urs Beyerle a écrit :

...

Hi,

Since a few day(?) we can no longer open ps or eps files with okular and evince. gv still works. We see it on 42.1 and 13.2, but not 13.1. Open any ps or eps file with okular gives an empty page and

okular juste works here (openSUSE 42.1)

which ghostscript version do you have? rpm -q ghostscript In my case (42.1) with ghostscript-9.15-8.1 I see the problem. Old ghostscript-9.15-5.1 and okular works. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Nov 2, 2016, at 20:26, Dave Howorth wrote:

On Wed, 2 Nov 2016 23:00:50 +0100 Urs Beyerle wrote:

...

which ghostscript version do you have? rpm -q ghostscript

ghostscript-9.15-8.1.x86_64

...

In my case (42.1) with ghostscript-9.15-8.1 I see the problem. Old ghostscript-9.15-5.1 and okular

What does ps2pdf do when you go to covert to PDF? If that doesn't work, could be a problem with Ghostscript that is being tickled by your PS files. I saw that a lot when I worked for a publisher. Can we see your PS file? I want to know if it's also only for some or for all. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 11/03/2016 08:01 AM, jdd wrote:

may be it's only on some pdf?

I have no problem reading this:

http://dodin.org/owncloud/index.php/s/txPFbRaBVEbySew

The problem are not pdf's. It's only .ps and .eps files. I see it with all ps and eps files. Example the "standard tiger.eps": http://www.xm1math.net/doculatex/insertionimages/tiger.eps Or just take a simple hello_world.ps %!PS % hello_world.psurs /Courier 20 selectfont 72 500 moveto (Hello world!) show showpage download here: http://iacweb.ethz.ch/staff/beyerleu/example_files/hello_world.ps ps2pdf still works without a problem. Also old "gv" has no problem with ps and eps files. Which is somehow strange. Cheers, Urs -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 11/03/2016 12:38 PM, Tom Kacvinsky wrote:

...

On Nov 3, 2016, at 04:05:19, Urs Beyerle wrote:

On 11/03/2016 08:01 AM, jdd wrote:

...

may be it's only on some pdf?

I have no problem reading this:

http://dodin.org/owncloud/index.php/s/txPFbRaBVEbySew

The problem are not pdf's. It's only .ps and .eps files. I see it with all ps and eps files.

Example the "standard tiger.eps": http://www.xm1math.net/doculatex/insertionimages/tiger.eps

Or just take a simple hello_world.ps

%!PS % hello_world.psurs /Courier 20 selectfont 72 500 moveto (Hello world!) show showpage

download here: http://iacweb.ethz.ch/staff/beyerleu/example_files/hello_world.ps

ps2pdf still works without a problem. Also old "gv" has no problem with ps and eps files. Which is somehow strange.

Using Leap 42.1, ghostscript-9.15-8.1.x86_64, tiger.eps views fine in gv-3.7.4-5.2.x86_64.

Using ghostscript-9.15-11.1.x86_64 also displays without error in gv.

So I can't be of use in helping you diagnose this problem. If I could reproduce the crash, I might be of some (limited) use.

Well, I do get this when viewing gv (with return code 0). Perhaps it is this that is throwing off evince. tkacvinsky@hypatia:~/Downloads> gv ISV-Infos-troubles-du-sommeil.pdf Error: /invalidaccess in --.locksafe-- Operand stack: --dict:1181/1684(ro)(G)-- getenv --nostringval-- Execution stack: %interp_exit .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- --nostringval-- --nostringval-- false 1 %stopped_push 1951 1 3 %oparray_pop 1950 1 3 %oparray_pop 1934 1 3 %oparray_pop 1820 1 3 %oparray_pop --nostringval-- %errorexec_pop .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- 1960 0 3 %oparray_pop 1958 0 3 %oparray_pop --nostringval-- Dictionary stack: --dict:1181/1684(ro)(G)-- --dict:0/20(G)-- --dict:79/200(L)-- Current allocation mode is local Current file position is 1899 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 11/03/2016 12:57 PM, Tom Kacvinsky wrote:

...

On Nov 3, 2016, at 12:50:49, Tom Kacvinsky wrote:

On 11/03/2016 12:38 PM, Tom Kacvinsky wrote:

...

...

On Nov 3, 2016, at 04:05:19, Urs Beyerle wrote:

On 11/03/2016 08:01 AM, jdd wrote:

...

may be it's only on some pdf?

I have no problem reading this:

http://dodin.org/owncloud/index.php/s/txPFbRaBVEbySew

The problem are not pdf's. It's only .ps and .eps files. I see it with all ps and eps files.

Example the "standard tiger.eps": http://www.xm1math.net/doculatex/insertionimages/tiger.eps

Or just take a simple hello_world.ps

%!PS % hello_world.psurs /Courier 20 selectfont 72 500 moveto (Hello world!) show showpage

download here: http://iacweb.ethz.ch/staff/beyerleu/example_files/hello_world.ps

ps2pdf still works without a problem. Also old "gv" has no problem with ps and eps files. Which is somehow strange.

Using Leap 42.1, ghostscript-9.15-8.1.x86_64, tiger.eps views fine in gv-3.7.4-5.2.x86_64.

Using ghostscript-9.15-11.1.x86_64 also displays without error in gv.

So I can't be of use in helping you diagnose this problem. If I could reproduce the crash, I might be of some (limited) use.

Well, I do get this when viewing gv (with return code 0). Perhaps it is this that is throwing off evince.

tkacvinsky@hypatia:~/Downloads> gv ISV-Infos-troubles-du-sommeil.pdf Error: /invalidaccess in --.locksafe-- Operand stack: --dict:1181/1684(ro)(G)-- getenv --nostringval-- Execution stack: %interp_exit .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- --nostringval-- --nostringval-- false 1 %stopped_push 1951 1 3 %oparray_pop 1950 1 3 %oparray_pop 1934 1 3 %oparray_pop 1820 1 3 %oparray_pop --nostringval-- %errorexec_pop .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- 1960 0 3 %oparray_pop 1958 0 3 %oparray_pop --nostringval-- Dictionary stack: --dict:1181/1684(ro)(G)-- --dict:0/20(G)-- --dict:79/200(L)-- Current allocation mode is local Current file position is 1899

I just built Ghostscript 9.20 from source and this fixed the PostScript interpreter error I saw. So perhaps it is time to update the GhostScript shipped with Leap. I can work with the package maintainer if he/she desires it.

I am no longer convinced that this is a ghostscript/gv problem, at least for evince. I am using evince-3.14.2-2.8.x86_64. Here is the relevant tidbit that makes me think this: Thread 1 "evince" received signal SIGSEGV, Segmentation fault. INT_cairo_surface_set_device_scale (surface=0x0, x_scale=1, y_scale=1) at cairo-surface.c:1712 1712 if (unlikely (surface->status)) (gdb) p surface $1 = (cairo_surface_t *) 0x0 We are de-referencing a NULL pointer, so bad values are being passed into the cairo library. Why this doesn't happen with PDF files but does for PS and EPS, I don't know. Looking into issues with okular now, I can reproduce the problem there and I believe that this is a gv problem. Looking into to it now -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On 03/11/2016 20:41, Tom Kacvinsky wrote:

...

On Nov 3, 2016, at 13:50:42, Tom Kacvinsky wrote: <snip/>

...

I am no longer convinced that this is a ghostscript/gv problem, at least for evince. I am using evince-3.14.2-2.8.x86_64.

Here is the relevant tidbit that makes me think this:

Thread 1 "evince" received signal SIGSEGV, Segmentation fault. INT_cairo_surface_set_device_scale (surface=0x0, x_scale=1, y_scale=1) at cairo-surface.c:1712 1712 if (unlikely (surface->status)) (gdb) p surface $1 = (cairo_surface_t *) 0x0

We are de-referencing a NULL pointer, so bad values are being passed into the cairo library. Why this doesn't happen with PDF files but does for PS and EPS, I don't know.

Looking into issues with okular now, I can reproduce the problem there and I believe that this is a gv problem. Looking into to it now. Yes, the okular problem is definitely Ghostscript based. I had to rebuild Ghostscript 9.20 so that libgs.so - needed by spectre, the PS rendering library used by okular - was built. Once I did that and set LD_LIBRARY_PATH to locate this newly built library, okular did the right thing and displayed tiger.eps

It turns out the problem with evince is also Ghostscript based, unlike my previous suspicion that it wasn't. evince depends on libgs. I found this by looking at /proc/<evince pid>/maps. It is an indirect dependency but I haven't found exactly what requires libgs.

In the end, I really think Leap should get an updated version of Ghostscript.

Tom

You need to file a bug and add this information. Regards Dave P -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Nov 4, 2016, at 06:47:12, Tom Kacvinsky wrote:

...

On Nov 4, 2016, at 03:05:49, Dave Plater wrote:

On 03/11/2016 20:41, Tom Kacvinsky wrote:

...

...

On Nov 3, 2016, at 13:50:42, Tom Kacvinsky wrote: <snip/>

...

I am no longer convinced that this is a ghostscript/gv problem, at least for evince. I am using evince-3.14.2-2.8.x86_64.

Here is the relevant tidbit that makes me think this:

Thread 1 "evince" received signal SIGSEGV, Segmentation fault. INT_cairo_surface_set_device_scale (surface=0x0, x_scale=1, y_scale=1) at cairo-surface.c:1712 1712 if (unlikely (surface->status)) (gdb) p surface $1 = (cairo_surface_t *) 0x0

We are de-referencing a NULL pointer, so bad values are being passed into the cairo library. Why this doesn't happen with PDF files but does for PS and EPS, I don't know.

Looking into issues with okular now, I can reproduce the problem there and I believe that this is a gv problem. Looking into to it now. Yes, the okular problem is definitely Ghostscript based. I had to rebuild Ghostscript 9.20 so that libgs.so - needed by spectre, the PS rendering library used by okular - was built. Once I did that and set LD_LIBRARY_PATH to locate this newly built library, okular did the right thing and displayed tiger.eps

It turns out the problem with evince is also Ghostscript based, unlike my previous suspicion that it wasn't. evince depends on libgs. I found this by looking at /proc/<evince pid>/maps. It is an indirect dependency but I haven't found exactly what requires libgs.

In the end, I really think Leap should get an updated version of Ghostscript.

Tom

You need to file a bug and add this information.

Done.

https://bugzilla.opensuse.org/show_bug.cgi?id=1008506

Known bug, marked as duplicate. Fix until official release comes out is this: You can easily fix it yourself though until an update is released: Open the file /usr/share/ghostscript/9.15/Resource/Init/gs_init.ps in a text editor (as root), and change line#2024 to: systemdict /getenv {pop //false} .forceput (i.e. replace "put" with ".forceput") -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

On Nov 4, 2016, at 09:38:35, Urs Beyerle wrote:

On 11/04/2016 12:58 PM, Tom Kacvinsky wrote:

...

Known bug, marked as duplicate. Fix until official release comes out is this:

You can easily fix it yourself though until an update is released: Open the file /usr/share/ghostscript/9.15/Resource/Init/gs_init.ps in a text editor (as root), and change line#2024 to: systemdict /getenv {pop //false} .forceput (i.e. replace "put" with ".forceput")

works like a charm. Many thanks for the quick fix

I received an email today that the fix is being pushed through OBS, so it should be available soon. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org