Page 1 of 1
6.9.4-2 regression
Posted: 2016-06-06T06:56:24-07:00
by argv[0]
Code: Select all
$ convert -version
Version: ImageMagick 6.9.4-7 Q16 amd64 2016-06-06 http://www.imagemagick.org
Copyright: Copyright (C) 1999-2016 ImageMagick Studio LLC
License: http://www.imagemagick.org/script/license.php
Features: Cipher DPC HDRI OpenMP
Delegates (built-in): bzlib cairo fftw fontconfig fpx freetype gvc jbig jng jpeg lcms lqr openexr pangocairo png rsvg tiff webp wmf x xml
$ convert -depth 8 -quality 95 -size 256x256 fractal: -crop 240x240+8+8 +repage im-6.9.4-2-regression.png
As a result, I got the following image:
It seems the changeset
52dae14 has made the cropping operation broken. Reverting this change gives the correct image
Re: 6.9.4-2 regression
Posted: 2016-06-06T08:10:23-07:00
by magick
We had to remove SyncImagePixelCache() call from AcquireAuthenticCacheView() because the method was not properly handling any exceptions. We tried your command on our system with ImageMagick 6.9.4-8 beta and it worked properly. Can you download
http://magick.imagemagick.org/download/ ... -8.tar.bz2 and let us know if you still experience the problem. If so, report back here and let us know which OS you are using and what release of libpng. It would be helpful if we can reproduce the problem to generate a robust patch to fix it.
Re: 6.9.4-2 regression
Posted: 2016-06-06T09:32:52-07:00
by fmw42
It works fine for me under IM 6.9.4.2 and 6.9.4.7 Q16 Mac OSX with 2 threads.
If you are using IM 6.9.4.7, then I am not sure why the result is named with 6.9.4-2.
Re: 6.9.4-2 regression
Posted: 2016-06-06T09:58:33-07:00
by argv[0]
6.9.4-8 beta has the same problem for me.
magick wrote:If so, report back here and let us know which OS you are using and what release of libpng.
OS:
Code: Select all
$ uname -srm
FreeBSD 9.3-STABLE amd64
libpng:
But any other formats (PPM, TIFF, JPEG) has the same result.
Also, I tried to change number of OpenMP threads using OMP_NUM_THREADS environment variable. OMP_NUM_THREADS=1 gives correct image, but an any other value - wrong one. So, looks like it's OpenMP related issue.
Re: 6.9.4-2 regression
Posted: 2016-06-06T10:05:24-07:00
by fmw42
You can also try:
Code: Select all
convert -limit thread 1 -depth 8 -quality 95 -size 256x256 fractal: -crop 240x240+8+8 +repage im-6.9.4-2-regression.png