Hi Elle,

The recent update to GEGL USM is consistently crashing on Windows (my test machine is Windows 8.1 64-bit).

Thanks, Partha

Thanks! Would you happen to have a copy of unsharp-mask.c from 2 weeks ago? I was not experiencing any crashes with unsharp mask then.

Thanks again, Partha

On Fri, Dec 6, 2013 at 7:30 PM, Elle Stone wrote:

On 12/06/2013 07:11 PM, Partha Bagchi wrote:

Hi Elle,

The recent update to GEGL USM is consistently crashing on Windows (my test machine is Windows 8.1 64-bit).

Thanks, Partha

Hi Partha,

I don't have a Windows machine for testing. I suppose it's possible that the ui patch for GEGL USM might cause a crash. Did you try reverting just that one commit?

Here are the last three gegl USM-related commits from "git log" (probably your commit numbers aren't the same as my commit numbers):

commit 1a0e01e9dd3d7f689a75f9f456a2cbd942a46562 Author: Elle Stone
Date: Thu Dec 5 13:54:06 2013 -0500

operations: Change ui slider values for unsharp mask and gaussian blur

commit 773561349e27bb1e0dfa4cbe82a8eb4f1ccd999e Author: To Mazars
Date: Thu Dec 5 19:33:19 2013 +0100

po: forgot to add the new gaussian blur implementation

commit b3557301766d1aa3e3a62859968c3c1b898b7e37 Author: Massimo Valentini
Date: Fri Jul 5 20:32:56 2013 +0200

workshop: add a better gaussian-blur implementation, work in progress

- the fir case is missing for small std-dev - the implementation doesn't allow infinite plane input - the ouptput extent doesn't grow anymore

I can send you a copy of gaussian-blur.c from before the commit, if that would help.

Elle

Hi Elle,

Not a problem. It's possible the issue is related to gaussian-blur. However, I did try to revert back the latest changes to the unsharp-mask.c from 2 weeks ago and still got the Windows equivalent of segfault. So, I decided to completely start afresh with the a new git pull of babl, gegl and gimp.

I am still getting crashes. The crash is probably related to GSlice:

0x000000006862110F (0x0000000000000001 0x00000001508B82E0 0x00000001508B82F0 0x0000000000339D95), g_slice_free1() + 0x17F bytes(s) 0x0000000000358B1D (0x0000000000000000 0x0000001500000780 0x00006D8EB17BD0C9 0x00000000030EE5A0), g_type_free_instance() + 0x18D bytes(s)
0x00000000618E166E (0x000000000023CA30 0x00000000031B3450 0x0000000000000098 0x000000000035D0D9), gtk_object_destroy() + 0x10E bytes(s)
0x00000000617F966D (0x0000000000000020 0x000000000000008D 0x000000000023CCB0 0x00000000685E6157), gtk_container_unset_focus_chain() + 0x36D bytes(s)
0x0000000000335F6F (0x0000000000000000 0x00000000685E6157 0x00000000030E1830 0x00000000031759D0), g_closure_invoke() + 0x12F bytes(s) 0x00000000003481CD (0x0000000000000000 0x000000000000004A 0x000000003DE44B80 0x0000000000000030), g_signal_handler_disconnect() + 0x198D bytes(s)

....

I can post the full error set if anyone wishes.

Thanks, Partha

On Sat, Dec 7, 2013 at 5:58 AM, Elle Stone wrote:

On 12/07/2013 05:04 AM, Tobias Ellinghaus wrote:

Am Freitag, 6. Dezember 2013, 20:18:13 schrieb Partha Bagchi:

Thanks! Would you happen to have a copy of unsharp-mask.c from 2 weeks ago?
I was not experiencing any crashes with unsharp mask then.

This is the version from before Elle changed it:

https://git.gnome.org/browse/gegl/plain/operations/common/ unsharp-mask.c?id=08287da809f1ff41501c331ad18e8c615c395b98

Thanks again,

Partha

Tobias

Tobias, Partha, my apologies, it was late and I turned my computer off.

I also substituted "Gaussian Blur" for USM in my mind, totally different things (there's a blur in USM, but that's beside the point).

Anyway, anyone have any ideas why a change in the ui slider values might cause a crash in Windows USM? Could it be because the minimum for the top slider is 0.2 instead of 0.0?

gegl_chant_double_ui (std_dev, _("Std. Dev."), 0.2, 300.0, 0.55, 0.2, 10.0, 3.0, _("Standard deviation (spatial scale factor)"))

gegl_chant_double_ui (scale, _("Scale"), 0.0, 300.0, 4.0, 0.0, 10.0, 3.0, _("Scale, strength of effect"))

Elle, feeling very foolish for having mixed up GB and USM.

Sorry, that should be "a new git clone" and not "a new git pull".

Thanks, Partha

On Sat, Dec 7, 2013 at 11:01 AM, Partha Bagchi wrote:

Hi Elle,

Not a problem. It's possible the issue is related to gaussian-blur. However, I did try to revert back the latest changes to the unsharp-mask.c from 2 weeks ago and still got the Windows equivalent of segfault. So, I decided to completely start afresh with the a new git pull of babl, gegl and gimp.

I am still getting crashes. The crash is probably related to GSlice:

0x000000006862110F (0x0000000000000001 0x00000001508B82E0 0x00000001508B82F0 0x0000000000339D95), g_slice_free1() + 0x17F bytes(s) 0x0000000000358B1D (0x0000000000000000 0x0000001500000780 0x00006D8EB17BD0C9 0x00000000030EE5A0), g_type_free_instance() + 0x18D bytes(s)
0x00000000618E166E (0x000000000023CA30 0x00000000031B3450 0x0000000000000098 0x000000000035D0D9), gtk_object_destroy() + 0x10E bytes(s)
0x00000000617F966D (0x0000000000000020 0x000000000000008D 0x000000000023CCB0 0x00000000685E6157), gtk_container_unset_focus_chain() + 0x36D bytes(s)
0x0000000000335F6F (0x0000000000000000 0x00000000685E6157 0x00000000030E1830 0x00000000031759D0), g_closure_invoke() + 0x12F bytes(s) 0x00000000003481CD (0x0000000000000000 0x000000000000004A 0x000000003DE44B80 0x0000000000000030), g_signal_handler_disconnect() + 0x198D bytes(s)

....

I can post the full error set if anyone wishes.

Thanks, Partha

On Sat, Dec 7, 2013 at 5:58 AM, Elle Stone

wrote:

On 12/07/2013 05:04 AM, Tobias Ellinghaus wrote:

Am Freitag, 6. Dezember 2013, 20:18:13 schrieb Partha Bagchi:

Thanks! Would you happen to have a copy of unsharp-mask.c from 2 weeks ago?
I was not experiencing any crashes with unsharp mask then.

This is the version from before Elle changed it:

https://git.gnome.org/browse/gegl/plain/operations/common/ unsharp-mask.c?id=08287da809f1ff41501c331ad18e8c615c395b98

Thanks again,

Partha

Tobias

Tobias, Partha, my apologies, it was late and I turned my computer off.

I also substituted "Gaussian Blur" for USM in my mind, totally different things (there's a blur in USM, but that's beside the point).

Anyway, anyone have any ideas why a change in the ui slider values might cause a crash in Windows USM? Could it be because the minimum for the top slider is 0.2 instead of 0.0?

gegl_chant_double_ui (std_dev, _("Std. Dev."), 0.2, 300.0, 0.55, 0.2, 10.0, 3.0, _("Standard deviation (spatial scale factor)"))

gegl_chant_double_ui (scale, _("Scale"), 0.0, 300.0, 4.0, 0.0, 10.0, 3.0, _("Scale, strength of effect"))

Elle, feeling very foolish for having mixed up GB and USM.