I was doing a test build of zipios++ for rawhide in mock and when the build for the documentation was attempted I got an interesting (if unhelpful) error.
It looks like the make doc target uses convert for part of the documentation:
+ make V=1 doc convert doc/images/zipios++.jpg doc/images/zipios++.eps Makefile:808: recipe for target 'doc' failed make: *** [doc] Illegal instruction (core dumped)
Doing this by hand in my f21 system seems to work.
Thanks, Richard
On Wed, Apr 29, 2015 at 21:34:45 -0500, Richard Shaw hobbes1069@gmail.com wrote:
I was doing a test build of zipios++ for rawhide in mock and when the build for the documentation was attempted I got an interesting (if unhelpful) error.
It looks like the make doc target uses convert for part of the documentation:
- make V=1 doc
convert doc/images/zipios++.jpg doc/images/zipios++.eps Makefile:808: recipe for target 'doc' failed make: *** [doc] Illegal instruction (core dumped)
Doing this by hand in my f21 system seems to work.
I tested this and it happens on my AMD machine, but not my intel machine. So its probably using SSE without testing to make sure it's supported. Most likely this can be disabled in the build.
Have you filed a bug for this?
On Thu, Apr 30, 2015 at 3:08 PM, Bruno Wolff III bruno@wolff.to wrote:
On Wed, Apr 29, 2015 at 21:34:45 -0500, Richard Shaw hobbes1069@gmail.com wrote:
I was doing a test build of zipios++ for rawhide in mock and when the build for the documentation was attempted I got an interesting (if unhelpful) error.
It looks like the make doc target uses convert for part of the documentation:
- make V=1 doc
convert doc/images/zipios++.jpg doc/images/zipios++.eps Makefile:808: recipe for target 'doc' failed make: *** [doc] Illegal instruction (core dumped)
Doing this by hand in my f21 system seems to work.
I tested this and it happens on my AMD machine, but not my intel machine. So its probably using SSE without testing to make sure it's supported. Most likely this can be disabled in the build.
Have you filed a bug for this?
Not yet, wanted confirmation first.
Thanks, Richard
On Thu, Apr 30, 2015 at 3:08 PM, Bruno Wolff III bruno@wolff.to wrote:
I tested this and it happens on my AMD machine, but not my intel machine. So its probably using SSE without testing to make sure it's supported. Most likely this can be disabled in the build.
I had some time to think about this... I'm running x86_64 so SSE should be supported, correct?
Thanks, Richard
Bug submitted:
https://bugzilla.redhat.com/show_bug.cgi?id=1217741
Perhaps it just needs to be rebuilt for the c++ ABI breakage?
Thanks, Richard
On 05/01/2015 03:06 PM, Richard Shaw wrote:
Bug submitted:
https://bugzilla.redhat.com/show_bug.cgi?id=1217741
Perhaps it just needs to be rebuilt for the c++ ABI breakage?
I ran into the same thing and pushed http://pkgs.fedoraproject.org/cgit/ImageMagick.git/commit/?id=225d561423c0cf... which should hopefully fix this up.