Post any defects you find in the released or beta versions of the ImageMagick software here. Include the ImageMagick version, OS, and any command-line required to reproduce the problem. Got a patch for a bug? Post it here.
-rw-rw-r-- 1 mi wheel 16654 Oct 5 17:44 work/ImageMagick-6.3.6/coders/exr.c
-rw-rw-r-- 1 mi wheel 1719 Oct 30 12:57 work/ImageMagick-6.3.6/coders/exr.la
-rw-rw-r-- 1 mi wheel 2314 Oct 30 12:57 work/ImageMagick-6.3.6/coders/.libs/exr.a
-rw-rw-r-- 1 mi wheel 36 Oct 30 12:57 work/ImageMagick-6.3.6/coders/.libs/exr.exp
lrwxrwxr-x 1 mi wheel 9 Oct 30 12:57 work/ImageMagick-6.3.6/coders/.libs/exr.la -> ../exr.la
-rw-rw-r-- 1 mi wheel 1132 Oct 30 12:57 work/ImageMagick-6.3.6/coders/.libs/exr.lai
-rwxrwxr-x 1 mi wheel 6125 Oct 30 12:57 work/ImageMagick-6.3.6/coders/.libs/exr.so
If you disable EXR support, the module is compiled but it is just a stub. Type
identify -list format
Under the EXR tag you see a mode '---' which means this format can neither be read or written. We have a mechanism to not create a module if support is disabled but we thought showing a user that ImageMagick is capable of supporting the format is more informative. What is your preference?
magick wrote:We have a mechanism to not create a module if support is disabled but we thought showing a user that ImageMagick is capable of supporting the format is more informative. What is your preference?
Uhm, I would think, that in the case of it being a stub, the use of an external module can be avoided, even if other modules are enabled and other formats are compiled as modules.
My preference would be for all modules to be treated consistently. Disabling FPX, for example, prevents fpx-module from getting created (as one would expect). But EXR is always built -- as a real module or as a stub...
We are going for consistency for the next point release. Like FPX, and TIFF, if EXR is disabled, the exr.so module will not be built. Will have a patch in ImageMagick SVN within a day or two. Thanks.
magick wrote:Under the EXR tag you see a mode '---' which means this format can neither be read or written.
Wait a minute. If that is "just a stub", why is it linked with OpenEXR's libraries (and all the others)? Search the ldd's output below for "libIlm" (for Industrial Light and Magick):
Check coders/exr.c and you will see that no OpenEXR code is called unless HasOPENEXR is defined (and HasOPENEXR is not defined unless OPENEXR support is enabled at configure time). No matter though, we patched ImageMagick 6.3.6-5 so that the EXR module is not compiled or installed if OPENEXR is not enabled at configure time.
magick wrote:No matter though, we patched ImageMagick 6.3.6-5 so that the EXR module is not compiled or installed if OPENEXR is not enabled at configure time.
Trying to build the newly released 6.3.6-5 (with OpenEXR configured)
Before you ask, the errors quoted above strike regardless of whether OpenEXR is enabled. Here is the log of the build with OpenEXR disabled -- same linkage error.
I must, once again, express serious doubts, that ImageMagick releases are subjected to automated tests before being, uhm, released
magick wrote:Oopsies. Will have a fix momentarily. Thanks.
Noticed the re-rolled 6.3.6-5 release on the FTP-site (a bad thing, BTW -- you should've published a patch or released a 6.3.6-6). Downloaded and tried rebuilding... Same failure.
"Oopsies" indeed. Here is the new log, if you care.
It is fairly clear, that you are lacking the resources to perform not only the post-build tests, but even to fully rebuild the software before publishing a release.
I can offer an account on my FreeBSD/amd64 box. It has a static-IP and is almost always online. Please, reply privately with the public SSH-key and the desired /etc/passwd entry.
Curious. Register/UnregisterEXRImage() is conditionally compiled only if HasOPENEXR is defined suggesting if configure fails to detect the OpenEXR library or if OpenEXR is disabled these methods will not be called. In coders/Makefile.am we conditionally compile coders/exr.c only if HasOPENEXR is defined. If you spot the flaw in this logic let us know.
magick wrote:Curious. Register/UnregisterEXRImage() is conditionally compiled only if HasOPENEXR is defined suggesting if configure fails to detect the OpenEXR library or if OpenEXR is disabled these methods will not be called. In coders/Makefile.am we conditionally compile coders/exr.c only if HasOPENEXR is defined. If you spot the flaw in this logic let us know.
Yes, indeed, there is no problem any more with OpenEXR disabled.
The problem strikes when OpenEXR support is enabled.
Something is still problematic, and I'm still waiting for a confirmation, that it does, indeed, work on systems other than FreeBSD.
Hi folks,
I'd just like to say that you people kick ass and your work is much appreciated
(both programming and bug-hunting!!)
thanks!!
Best regards
/H
ps giving the developers axxess to a FreeBSD-boxen sounds like a great idea, easiest with root on a jail with full developer package installed maybe
Hero wrote:ps giving the developers axxess to a FreeBSD-boxen sounds like a great idea, easiest with root on a jail with full developer package installed maybe
Yes. My offer of a login-access to my box (aldan.algebra.com) still stands...