Re: [Fink-users] matplotlib-py27 Problem
On 7/06/13 04:22, Alexander Hansen wrote: [] I'm not familiar enough with matplotlib's build system to suggest an _actual_ fix at this stage. ;-) In the matplotlib setupext.py file, one sees basedir = defaultdict(lambda: ['/usr/local', '/usr'], { # execptions to the ['/usr/local', '/usr'] defaults 'win32' : ['win32_static',], 'darwin' : ['/usr', '/usr/X11', '/opt/X11', '/sw/lib/freetype219', '/sw',], 'sunos5' : [os.getenv('MPLIB_BASE') or '/usr/local',], 'gnu0' : ['/usr'], 'aix5' : ['/usr/local'], }) This comes directly from the Fink matplotlib-py.patch file. It could be worthwhile to try changing the order in the 'darwin' section. -- Martin -- How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
On 6/7/13 12:13 AM, Martin Costabel wrote: On 7/06/13 04:22, Alexander Hansen wrote: [] I'm not familiar enough with matplotlib's build system to suggest an _actual_ fix at this stage. ;-) In the matplotlib setupext.py file, one sees basedir = defaultdict(lambda: ['/usr/local', '/usr'], { # execptions to the ['/usr/local', '/usr'] defaults 'win32' : ['win32_static',], 'darwin' : ['/usr', '/usr/X11', '/opt/X11', '/sw/lib/freetype219', '/sw',], 'sunos5' : [os.getenv('MPLIB_BASE') or '/usr/local',], 'gnu0' : ['/usr'], 'aix5' : ['/usr/local'], }) This comes directly from the Fink matplotlib-py.patch file. It could be worthwhile to try changing the order in the 'darwin' section. (maintainer cc'ed) It was indeed worthwhile.:-) Putting the Fink paths at the head of the lists in the .patch appears to solve the problem. If there are no objections I can commit a modified .info and .patch with this fix plus the addition of -MD flags and a fink-package-precedence scan. -- Alexander Hansen, Ph.D. Fink User Liaison My package updates: http://finkakh.wordpress.com/ -- How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Has anyone managed to fix this problem yet? Should I contact upstream, or has someone already done so? This makes matplotlib unusable for me. Thanks, John On 26May 2013, at 12:24, Alexander Hansen wrote: Fink only supports manual intervention via changes to the package description (.info) and specified patch files. I tried to do something comparatively simple and use Fink's flag-sort tool to change the order of compiler flags, but that appears possibly to change the order _too_ much and produces a failure later on in the build: flag-sort -r clang -L/sw/lib -bundle -L/sw/lib/python2.7/config -lpython2.7 -L/sw/lib -L/sw/lib/freetype219/lib -L/sw/lib/fontconfig2/lib -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 build/temp.macosx-10.8-x86_64-2.7/src/_ttconv.o build/temp.macosx-10.8-x86_64-2.7/ttconv/pprdrv_tt.o build/temp.macosx-10.8-x86_64-2.7/ttconv/pprdrv_tt2.o build/temp.macosx-10.8-x86_64-2.7/ttconv/ttutil.o -L/usr/lib -L/usr/X11/lib -L/opt/X11/lib -L/sw/lib/freetype219/lib -L/sw/lib -L/sw/lib -o build/lib.macosx-10.8-x86_64-2.7/matplotlib/ttconv.so Undefined symbols for architecture x86_64: std::basic_stringbufchar, std::char_traitschar, std::allocatorchar ::str() const, referenced from: get_pdf_charprocs(char const*, std::vectorint, std::allocatorint , TTDictionaryCallback) in pprdrv_tt.o ... On 5/25/13 8:49 PM, John K. Parejko wrote: Hello, I'm also having this exact problem for matplotlib-py26. Is there a solution available? It sounds like patching the matplotlib build script to change the directory ordering would fix it. Could this be done, or is there a way to do it manually? Thank you, John On 6Apr 2013, at 05:31, Martin Costabel wrote: On 4/04/13 11:55, Sean Lake wrote: Hello all, I've run into a problem in matplotlib. I'm using the GTKAgg backend and the matplotlib.pyplot.show() command after labeling the axes with LaTeX formatted labels. I run into multiple copies of this error: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Traceback (most recent call last): [] File /sw/lib/python2.7/site-packages/matplotlib/texmanager.py, line 510, in get_grey X = read_png(os.path.join(self.texcache, pngfile)) RuntimeError: _image_module::readpng: png_create_read_struct failed I think what happens is that when you build matplotlib-py27, it picks up an include file png.h from libpng-1.2.41. This may be /usr/X11/include/png.h or perhaps /usr/local/include/png.h. Did you mention what version of MacOSX you are running? If it is 10.6, you may indeed have /usr/X11/include/png.h of version 1.2.41. Try grep define PNG_LIBPNG_VER_STRING /usr/X11/include/png.h This might be considered a bug in the matplotlib-py27 package, in that it includes /usr/X11/include before /sw/include/libpng15. I see the following relevant compiler line during the build phase of matplotlib-py27 (on 10.8.3) gcc -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -fwrapv -O3 -Wall -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 -DPY_ARRAY_UNIQUE_SYMBOL=MPL_ARRAY_API -DPYCXX_ISO_CPP_LIB=1 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I/sw/include/libpng15 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I. -I/sw/lib/python2.7/site-packages/numpy/core/include -I. -I/sw/include -I/sw/include/python2.7 -c src/_png.cpp -o build/temp.macosx-10.8-x86_64-2.7/src/_png.o This is bad ordering of include directories. -- Martin -- Alexander Hansen, Ph.D. Fink User Liaison My package updates: http://finkakh.wordpress.com/ -- How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
No, nothing has happened. I wouldn't necessarily expect that upstream will understand the subtleties here. I've got 10.6 available now, and a workaround in mind that may help until this can actually be fixed properly. I'll test that out and report back. On 6/6/13 3:28 PM, John K. Parejko wrote: Has anyone managed to fix this problem yet? Should I contact upstream, or has someone already done so? This makes matplotlib unusable for me. Thanks, John On 26May 2013, at 12:24, Alexander Hansen wrote: Fink only supports manual intervention via changes to the package description (.info) and specified patch files. I tried to do something comparatively simple and use Fink's flag-sort tool to change the order of compiler flags, but that appears possibly to change the order _too_ much and produces a failure later on in the build: flag-sort -r clang -L/sw/lib -bundle -L/sw/lib/python2.7/config -lpython2.7 -L/sw/lib -L/sw/lib/freetype219/lib -L/sw/lib/fontconfig2/lib -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 build/temp.macosx-10.8-x86_64-2.7/src/_ttconv.o build/temp.macosx-10.8-x86_64-2.7/ttconv/pprdrv_tt.o build/temp.macosx-10.8-x86_64-2.7/ttconv/pprdrv_tt2.o build/temp.macosx-10.8-x86_64-2.7/ttconv/ttutil.o -L/usr/lib -L/usr/X11/lib -L/opt/X11/lib -L/sw/lib/freetype219/lib -L/sw/lib -L/sw/lib -o build/lib.macosx-10.8-x86_64-2.7/matplotlib/ttconv.so Undefined symbols for architecture x86_64: std::basic_stringbufchar, std::char_traitschar, std::allocatorchar ::str() const, referenced from: get_pdf_charprocs(char const*, std::vectorint, std::allocatorint , TTDictionaryCallback) in pprdrv_tt.o ... On 5/25/13 8:49 PM, John K. Parejko wrote: Hello, I'm also having this exact problem for matplotlib-py26. Is there a solution available? It sounds like patching the matplotlib build script to change the directory ordering would fix it. Could this be done, or is there a way to do it manually? Thank you, John On 6Apr 2013, at 05:31, Martin Costabel wrote: On 4/04/13 11:55, Sean Lake wrote: Hello all, I've run into a problem in matplotlib. I'm using the GTKAgg backend and the matplotlib.pyplot.show() command after labeling the axes with LaTeX formatted labels. I run into multiple copies of this error: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Traceback (most recent call last): [] File /sw/lib/python2.7/site-packages/matplotlib/texmanager.py, line 510, in get_grey X = read_png(os.path.join(self.texcache, pngfile)) RuntimeError: _image_module::readpng: png_create_read_struct failed I think what happens is that when you build matplotlib-py27, it picks up an include file png.h from libpng-1.2.41. This may be /usr/X11/include/png.h or perhaps /usr/local/include/png.h. Did you mention what version of MacOSX you are running? If it is 10.6, you may indeed have /usr/X11/include/png.h of version 1.2.41. Try grep define PNG_LIBPNG_VER_STRING /usr/X11/include/png.h This might be considered a bug in the matplotlib-py27 package, in that it includes /usr/X11/include before /sw/include/libpng15. I see the following relevant compiler line during the build phase of matplotlib-py27 (on 10.8.3) gcc -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -fwrapv -O3 -Wall -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 -DPY_ARRAY_UNIQUE_SYMBOL=MPL_ARRAY_API -DPYCXX_ISO_CPP_LIB=1 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I/sw/include/libpng15 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I. -I/sw/lib/python2.7/site-packages/numpy/core/include -I. -I/sw/include -I/sw/include/python2.7 -c src/_png.cpp -o build/temp.macosx-10.8-x86_64-2.7/src/_png.o This is bad ordering of include directories. -- Martin -- How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
On 6/6/13 3:49 PM, Alexander Hansen wrote: No, nothing has happened. I wouldn't necessarily expect that upstream will understand the subtleties here. I've got 10.6 available now, and a workaround in mind that may help until this can actually be fixed properly. I'll test that out and report back. Oh, and it would be nice to have a simple test case that I can run so that I can see if I've actually done anything. -- Alexander Hansen, Ph.D. Fink User Liaison My package updates: http://finkakh.wordpress.com/ -- How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
On 6/6/13 3:51 PM, Alexander Hansen wrote: On 6/6/13 3:49 PM, Alexander Hansen wrote: No, nothing has happened. I wouldn't necessarily expect that upstream will understand the subtleties here. I've got 10.6 available now, and a workaround in mind that may help until this can actually be fixed properly. I'll test that out and report back. Oh, and it would be nice to have a simple test case that I can run so that I can see if I've actually done anything. Meh. I'm on a fresh install and the build tree is huge. My idea was to try renaming /usr/X11/include/png.h and /usr/X11/include/pngconf.h temporarily, and then rebuild matplotlib-py26. -- Alexander Hansen, Ph.D. Fink User Liaison My package updates: http://finkakh.wordpress.com/ -- How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
On 4/6/13 8:43 AM, Daniel Macks wrote: On Sat, 6 Apr 2013 13:00:55 +0200, Peter Dyballa peter_dyba...@freenet.de wrote: Am 06.04.2013 um 11:31 schrieb Martin Costabel: -I/usr/include -I/usr/X11/include -I/opt/X11/include If it's that simple – and I'm starting to believe that – then it would be useful to patch the make system to have -idirafter /usr/X11/include -idirafter /opt/X11/include instead. In that case GCC searches /usr/X11/include (and MacPorts' /opt/X11/include) definitely last for C header files so that a foreign png.h file won't be used because Fink's one is found first. This can easily be checked by adding -H to CFLAGS of CPPFLAGS (because then all C header used for compilation are listed). Even better is to use -MD (often enabled by autoconf/libtool's --enable-dependency-tracking). Instead of relying on one's eye to catch the wrong path (and having it still build just wrong), one can then put a fink-package-precedence command in the CompileScript to cause the use of wrong libpng (and others) to be a fatal error. The GNOME2 suite of packages have been doing this for years. dan -- Daniel Macks dma...@netspace.org By adding -MD to CPPFLAGS, adding SetCXXFLAGS: -MD, and putting fink-package-precedence --depfile-ext='\.d' build in the CompileScript, I found the following (10.6/i386, stock X11) Looking for incorrect headers in 33 dependency files... build/temp.macosx-10.6-i386-2.7/src/backend_agg.d uses /usr/X11/include/ft2build.h build/temp.macosx-10.6-i386-2.7/src/_png.d uses /usr/X11/include/png.h build/temp.macosx-10.6-i386-2.7/src/_png.d uses /usr/X11/include/pngconf.h Please fix build process to get consistent use of fink's headers. However, on 10.8, I got simply: Looking for incorrect headers in 33 dependency files... build/temp.macosx-10.8-x86_64-2.7/src/backend_agg.d uses /usr/X11/include/ft2build.h Please fix build process to get consistent use of fink's headers. -- Alexander Hansen, Ph.D. Fink User Liaison My package updates: http://finkakh.wordpress.com/ -- How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
On 6/6/13 4:01 PM, Alexander Hansen wrote: On 6/6/13 3:51 PM, Alexander Hansen wrote: On 6/6/13 3:49 PM, Alexander Hansen wrote: No, nothing has happened. I wouldn't necessarily expect that upstream will understand the subtleties here. I've got 10.6 available now, and a workaround in mind that may help until this can actually be fixed properly. I'll test that out and report back. Oh, and it would be nice to have a simple test case that I can run so that I can see if I've actually done anything. Meh. I'm on a fresh install and the build tree is huge. My idea was to try renaming /usr/X11/include/png.h and /usr/X11/include/pngconf.h temporarily, and then rebuild matplotlib-py26. Also /usr/X11/include/ft2build.h I'm not familiar enough with matplotlib's build system to suggest an _actual_ fix at this stage. ;-) -- Alexander Hansen, Ph.D. Fink User Liaison My package updates: http://finkakh.wordpress.com/ -- How ServiceNow helps IT people transform IT departments: 1. A cloud service to automate IT design, transition and operations 2. Dashboards that offer high-level views of enterprise services 3. A single system of record for all IT processes http://p.sf.net/sfu/servicenow-d2d-j ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Fink only supports manual intervention via changes to the package description (.info) and specified patch files. I tried to do something comparatively simple and use Fink's flag-sort tool to change the order of compiler flags, but that appears possibly to change the order _too_ much and produces a failure later on in the build: flag-sort -r clang -L/sw/lib -bundle -L/sw/lib/python2.7/config -lpython2.7 -L/sw/lib -L/sw/lib/freetype219/lib -L/sw/lib/fontconfig2/lib -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 build/temp.macosx-10.8-x86_64-2.7/src/_ttconv.o build/temp.macosx-10.8-x86_64-2.7/ttconv/pprdrv_tt.o build/temp.macosx-10.8-x86_64-2.7/ttconv/pprdrv_tt2.o build/temp.macosx-10.8-x86_64-2.7/ttconv/ttutil.o -L/usr/lib -L/usr/X11/lib -L/opt/X11/lib -L/sw/lib/freetype219/lib -L/sw/lib -L/sw/lib -o build/lib.macosx-10.8-x86_64-2.7/matplotlib/ttconv.so Undefined symbols for architecture x86_64: std::basic_stringbufchar, std::char_traitschar, std::allocatorchar ::str() const, referenced from: get_pdf_charprocs(char const*, std::vectorint, std::allocatorint , TTDictionaryCallback) in pprdrv_tt.o ... On 5/25/13 8:49 PM, John K. Parejko wrote: Hello, I'm also having this exact problem for matplotlib-py26. Is there a solution available? It sounds like patching the matplotlib build script to change the directory ordering would fix it. Could this be done, or is there a way to do it manually? Thank you, John On 6Apr 2013, at 05:31, Martin Costabel wrote: On 4/04/13 11:55, Sean Lake wrote: Hello all, I've run into a problem in matplotlib. I'm using the GTKAgg backend and the matplotlib.pyplot.show() command after labeling the axes with LaTeX formatted labels. I run into multiple copies of this error: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Traceback (most recent call last): [] File /sw/lib/python2.7/site-packages/matplotlib/texmanager.py, line 510, in get_grey X = read_png(os.path.join(self.texcache, pngfile)) RuntimeError: _image_module::readpng: png_create_read_struct failed I think what happens is that when you build matplotlib-py27, it picks up an include file png.h from libpng-1.2.41. This may be /usr/X11/include/png.h or perhaps /usr/local/include/png.h. Did you mention what version of MacOSX you are running? If it is 10.6, you may indeed have /usr/X11/include/png.h of version 1.2.41. Try grep define PNG_LIBPNG_VER_STRING /usr/X11/include/png.h This might be considered a bug in the matplotlib-py27 package, in that it includes /usr/X11/include before /sw/include/libpng15. I see the following relevant compiler line during the build phase of matplotlib-py27 (on 10.8.3) gcc -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -fwrapv -O3 -Wall -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 -DPY_ARRAY_UNIQUE_SYMBOL=MPL_ARRAY_API -DPYCXX_ISO_CPP_LIB=1 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I/sw/include/libpng15 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I. -I/sw/lib/python2.7/site-packages/numpy/core/include -I. -I/sw/include -I/sw/include/python2.7 -c src/_png.cpp -o build/temp.macosx-10.8-x86_64-2.7/src/_png.o This is bad ordering of include directories. -- Martin -- Alexander Hansen, Ph.D. Fink User Liaison My package updates: http://finkakh.wordpress.com/ -- Try New Relic Now We'll Send You this Cool Shirt New Relic is the only SaaS-based application performance monitoring service that delivers powerful full stack analytics. Optimize and monitor your browser, app, servers with just a few lines of code. Try New Relic and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Hello, I'm also having this exact problem for matplotlib-py26. Is there a solution available? It sounds like patching the matplotlib build script to change the directory ordering would fix it. Could this be done, or is there a way to do it manually? Thank you, John On 6Apr 2013, at 05:31, Martin Costabel wrote: On 4/04/13 11:55, Sean Lake wrote: Hello all, I've run into a problem in matplotlib. I'm using the GTKAgg backend and the matplotlib.pyplot.show() command after labeling the axes with LaTeX formatted labels. I run into multiple copies of this error: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Traceback (most recent call last): [] File /sw/lib/python2.7/site-packages/matplotlib/texmanager.py, line 510, in get_grey X = read_png(os.path.join(self.texcache, pngfile)) RuntimeError: _image_module::readpng: png_create_read_struct failed I think what happens is that when you build matplotlib-py27, it picks up an include file png.h from libpng-1.2.41. This may be /usr/X11/include/png.h or perhaps /usr/local/include/png.h. Did you mention what version of MacOSX you are running? If it is 10.6, you may indeed have /usr/X11/include/png.h of version 1.2.41. Try grep define PNG_LIBPNG_VER_STRING /usr/X11/include/png.h This might be considered a bug in the matplotlib-py27 package, in that it includes /usr/X11/include before /sw/include/libpng15. I see the following relevant compiler line during the build phase of matplotlib-py27 (on 10.8.3) gcc -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -fwrapv -O3 -Wall -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 -DPY_ARRAY_UNIQUE_SYMBOL=MPL_ARRAY_API -DPYCXX_ISO_CPP_LIB=1 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I/sw/include/libpng15 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I. -I/sw/lib/python2.7/site-packages/numpy/core/include -I. -I/sw/include -I/sw/include/python2.7 -c src/_png.cpp -o build/temp.macosx-10.8-x86_64-2.7/src/_png.o This is bad ordering of include directories. -- Martin -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users -- Try New Relic Now We'll Send You this Cool Shirt New Relic is the only SaaS-based application performance monitoring service that delivers powerful full stack analytics. Optimize and monitor your browser, app, servers with just a few lines of code. Try New Relic and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
On 4/04/13 11:55, Sean Lake wrote: Hello all, I've run into a problem in matplotlib. I'm using the GTKAgg backend and the matplotlib.pyplot.show() command after labeling the axes with LaTeX formatted labels. I run into multiple copies of this error: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Traceback (most recent call last): [] File /sw/lib/python2.7/site-packages/matplotlib/texmanager.py, line 510, in get_grey X = read_png(os.path.join(self.texcache, pngfile)) RuntimeError: _image_module::readpng: png_create_read_struct failed I think what happens is that when you build matplotlib-py27, it picks up an include file png.h from libpng-1.2.41. This may be /usr/X11/include/png.h or perhaps /usr/local/include/png.h. Did you mention what version of MacOSX you are running? If it is 10.6, you may indeed have /usr/X11/include/png.h of version 1.2.41. Try grep define PNG_LIBPNG_VER_STRING /usr/X11/include/png.h This might be considered a bug in the matplotlib-py27 package, in that it includes /usr/X11/include before /sw/include/libpng15. I see the following relevant compiler line during the build phase of matplotlib-py27 (on 10.8.3) gcc -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -fwrapv -O3 -Wall -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 -DPY_ARRAY_UNIQUE_SYMBOL=MPL_ARRAY_API -DPYCXX_ISO_CPP_LIB=1 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I/sw/include/libpng15 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I. -I/sw/lib/python2.7/site-packages/numpy/core/include -I. -I/sw/include -I/sw/include/python2.7 -c src/_png.cpp -o build/temp.macosx-10.8-x86_64-2.7/src/_png.o This is bad ordering of include directories. -- Martin -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Yep, 10.6 and the result of the command you sent is: #define PNG_LIBPNG_VER_STRING 1.2.41 Thanks, Sean On Apr 6, 2013, at 2:31, Martin Costabel costa...@wanadoo.fr wrote: On 4/04/13 11:55, Sean Lake wrote: Hello all, I've run into a problem in matplotlib. I'm using the GTKAgg backend and the matplotlib.pyplot.show() command after labeling the axes with LaTeX formatted labels. I run into multiple copies of this error: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Traceback (most recent call last): [] File /sw/lib/python2.7/site-packages/matplotlib/texmanager.py, line 510, in get_grey X = read_png(os.path.join(self.texcache, pngfile)) RuntimeError: _image_module::readpng: png_create_read_struct failed I think what happens is that when you build matplotlib-py27, it picks up an include file png.h from libpng-1.2.41. This may be /usr/X11/include/png.h or perhaps /usr/local/include/png.h. Did you mention what version of MacOSX you are running? If it is 10.6, you may indeed have /usr/X11/include/png.h of version 1.2.41. Try grep define PNG_LIBPNG_VER_STRING /usr/X11/include/png.h This might be considered a bug in the matplotlib-py27 package, in that it includes /usr/X11/include before /sw/include/libpng15. I see the following relevant compiler line during the build phase of matplotlib-py27 (on 10.8.3) gcc -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -fwrapv -O3 -Wall -I/sw/include/pygtk-2.0 -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/fontconfig2/include -DNSIG=32 -DPY_ARRAY_UNIQUE_SYMBOL=MPL_ARRAY_API -DPYCXX_ISO_CPP_LIB=1 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I/sw/include/libpng15 -I/usr/include -I/usr/X11/include -I/opt/X11/include -I/sw/lib/freetype219/include -I/sw/include -I. -I/sw/lib/python2.7/site-packages/numpy/core/include -I. -I/sw/include -I/sw/include/python2.7 -c src/_png.cpp -o build/temp.macosx-10.8-x86_64-2.7/src/_png.o This is bad ordering of include directories. -- Martin -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Am 06.04.2013 um 11:31 schrieb Martin Costabel: -I/usr/include -I/usr/X11/include -I/opt/X11/include If it's that simple – and I'm starting to believe that – then it would be useful to patch the make system to have -idirafter /usr/X11/include -idirafter /opt/X11/include instead. In that case GCC searches /usr/X11/include (and MacPorts' /opt/X11/include) definitely last for C header files so that a foreign png.h file won't be used because Fink's one is found first. This can easily be checked by adding -H to CFLAGS of CPPFLAGS (because then all C header used for compilation are listed). -- Greetings Pete If men could get pregnant, abortion would be a sacrament. – Florynce Kennedy -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
On Sat, 6 Apr 2013 13:00:55 +0200, Peter Dyballa peter_dyba...@freenet.de wrote: Am 06.04.2013 um 11:31 schrieb Martin Costabel: -I/usr/include -I/usr/X11/include -I/opt/X11/include If it's that simple – and I'm starting to believe that – then it would be useful to patch the make system to have -idirafter /usr/X11/include -idirafter /opt/X11/include instead. In that case GCC searches /usr/X11/include (and MacPorts' /opt/X11/include) definitely last for C header files so that a foreign png.h file won't be used because Fink's one is found first. This can easily be checked by adding -H to CFLAGS of CPPFLAGS (because then all C header used for compilation are listed). Even better is to use -MD (often enabled by autoconf/libtool's --enable-dependency-tracking). Instead of relying on one's eye to catch the wrong path (and having it still build just wrong), one can then put a fink-package-precedence command in the CompileScript to cause the use of wrong libpng (and others) to be a fatal error. The GNOME2 suite of packages have been doing this for years. dan -- Daniel Macks dma...@netspace.org -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Am 06.04.2013 um 17:43 schrieb Daniel Macks: Even better is to use -MD (often enabled by autoconf/libtool's --enable-dependency-tracking). Instead of relying on one's eye to catch the wrong path (and having it still build just wrong), one can then put a fink-package-precedence command in the CompileScript to cause the use of wrong libpng (and others) to be a fatal error. The GNOME2 suite of packages have been doing this for years. A combination, protect check, is best. If one wouldn't want that package, one wouldn't invoke 'fink install …'. -- Greetings Pete Cyanide is so poisonous that one drop of it on a dog's tongue will kill the strongest man. -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
On 5/04/13 01:16, Sean Lake wrote: I was hoping that maybe tetex had a non-specific dependency on libpng and would compile with the newer one. Looking at the show-deps for tetex-base lists libpng3, so I would agree that's where the problem is. My guess is that the easiest fix would be to update matplotlib to depend on texlive instead of tetex. What's more, the last line of the rerror is the read_png function call in the texmanager.py file, so, yeah. Matplotlib depends on tetex-base, but tetex-base is also provided by texlive-base, so you can get rid of tetex and install texlive{,-base,-texmf} in any case. -- Martin -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Am 04.04.2013 um 16:27 schrieb Kurt Schwehr: otool output would be good to know. What version of macosx and matplotlib are you using here? Kurt, can this be the cause of the problem? /sw/fink/10.4/stable/main/finkinfo/sci/matplotlib-basemap-data-py.info: libgeos2 /sw/fink/10.4/stable/main/finkinfo/sci/matplotlib-basemap-py.info: libgeos3.3.1 -- Mit friedvollen Grüßen Pete No man was ever taken to hell by a woman unless he already had a ticket in his pocket, or at least had been fooling around with timetables. – Archie Goodwin -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Am 04.04.2013 um 11:55 schrieb Sean Lake: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Sean, one possible try to find a potential culprit is this script. First the version for (t)csh: foreach Lib ( `find /sw/lib /usr/X11/lib -type f \( -name *.dylib -a \! -name libpng* \)` ) otool -L $Lib | grep png echo $Lib end And now the version for (ba)sh: for Lib in `find /sw/lib /usr/X11/lib -type f \( -name *.dylib -a \! -name libpng* \)` ; do otool -L $Lib | grep png echo $Lib done If you want you can change grep png to grep png12 and you'll have less hits. If your X11 distribution in not in /usr/X11 you'll need to correct /usr/X11/lib and similarly /sw/lib. The output will be a list à la: /sw/lib/graphviz/tcl/libgdtclft.0.dylib /sw/lib/libpng12.0.dylib (compatibility version 50.0.0, current version 50.0.0) /sw/lib/graphviz/tcl/libtcldot.0.dylib /sw/lib/libpng12.0.dylib (compatibility version 50.0.0, current version 50.0.0) /sw/lib/graphviz/tcl/libtcldot_builtin.0.dylib /sw/lib/libpng15.15.dylib (compatibility version 26.0.0, current version 26.0.0) – which is a biz puzzling… -- Greetings Pete A lot of us are working harder than we want, at things we don't like to do. Why? ...In order to afford the sort of existence we don't care to live. – Bradford Angier -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Well, it isn't (just) tetex. I've switched to texlive, and the error is still there. The output of the script follows. The only mention of libpng12 is in libcairo, under x11. /sw/lib/libpng14.14.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/graphviz/tcl/libgdtclft.0.dylib /sw/lib/libpng14.14.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/graphviz/tcl/libtcldot.0.dylib /sw/lib/libpng14.14.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/graphviz/tcl/libtcldot_builtin.0.dylib /sw/lib/libpng15.15.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libcairo.2.dylib /sw/lib/libpng14.14.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libfltk_images.1.1.dylib /sw/lib/libpng15.15.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libgd.2.dylib /sw/lib/libpng15.15.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libming.1.4.1.dylib /sw/lib/libpng15.15.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libplot.2.dylib /sw/lib/libpng15.15.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libplotter.2.dylib /sw/lib/libpng15.15.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libpoppler.34.dylib /sw/lib/libpng14.14.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libwebkit-1.0.2.dylib /sw/lib/libpng14.14.dylib (compatibility version 26.0.0, current version 26.0.0) /sw/lib/libwmf-0.2.7.1.0.dylib /usr/X11/lib/libpng12.0.dylib (compatibility version 47.0.0, current version 47.0.0) /usr/X11/lib/libcairo.2.dylib On Fri, Apr 5, 2013 at 1:07 PM, Peter Dyballa peter_dyba...@freenet.de wrote: Am 04.04.2013 um 11:55 schrieb Sean Lake: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Sean, one possible try to find a potential culprit is this script. First the version for (t)csh: foreach Lib ( `find /sw/lib /usr/X11/lib -type f \( -name *.dylib -a \! -name libpng* \)` ) otool -L $Lib | grep png echo $Lib end And now the version for (ba)sh: for Lib in `find /sw/lib /usr/X11/lib -type f \( -name *.dylib -a \! -name libpng* \)` ; do otool -L $Lib | grep png echo $Lib done If you want you can change grep png to grep png12 and you'll have less hits. If your X11 distribution in not in /usr/X11 you'll need to correct /usr/X11/lib and similarly /sw/lib. The output will be a list à la: /sw/lib/graphviz/tcl/libgdtclft.0.dylib /sw/lib/libpng12.0.dylib (compatibility version 50.0.0, current version 50.0.0) /sw/lib/graphviz/tcl/libtcldot.0.dylib /sw/lib/libpng12.0.dylib (compatibility version 50.0.0, current version 50.0.0) /sw/lib/graphviz/tcl/libtcldot_builtin.0.dylib /sw/lib/libpng15.15.dylib (compatibility version 26.0.0, current version 26.0.0) – which is a biz puzzling… -- Greetings Pete A lot of us are working harder than we want, at things we don't like to do. Why? ...In order to afford the sort of existence we don't care to live. – Bradford Angier -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Am 05.04.2013 um 23:31 schrieb Sean Lake: The only mention of libpng12 is in libcairo, under x11. And which libraries is matplotlib using? Does it have plug-ins? dpkg -L matplotlib-py27 should give you a list of all files in the package. Do some file names end in .so? Which are the dylibs in the package, which the binaries? With otool -L you can check the dylibs and the binaries which dylibs they need. Do have in your shell's environment a variable set that contains DYLIB in its name? env | grep DYLIB should reveal this. Among the X11 binaries I have only /usr/X11/bin/xcursorgen that uses /usr/X11/lib/libpng12.0.dylib (compatibility version 42.0.0, current version 42.0.0). Among the Fink binaries I have /sw/bin/dillo, /sw/bin/mencoder, /sw/bin/mplayer, /sw/bin/pamrgbatopng, /sw/bin/pamrgbatopng, /sw/bin/pngtopnm, /sw/bin/pnmtopng that use libpng12. Libcairo (Fink's one) is used by /sw/bin/gkrellm(-2.3.2), /sw/bin/pango-view, /sw/bin/vim. With this you can check binaries if they use libpng or libcairo. First (t)csh: foreach Bin ( `ls -l /usr/X11/bin/* /usr/X11/bin/* | egrep -e '^-' | awk '{print $NF}'` ) otool -L $Bin | egrep 'png|cairo' echo $Bin end and now (ba)sh: for Bin in `ls -l /sw/bin/* /usr/X11/bin/* | egrep -e '^-' | awk '{print $NF}'` ; do otool -L $Bin | egrep 'png|cairo' echo $Bin done Again you can adjust png to libpng12 and the paths in Fink and X11. To grep for exactly X11's libcairo you should change cairo to X11/lib/libcairo or whatever the X11 string element in the library's path name is. -- Greetings Pete To be is to do. – I. Kant To do is to be. – A. Sartre Yabba-Dabba-Doo! – F. Flintstone -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
[Fink-users] matplotlib-py27 Problem
Hello all, I've run into a problem in matplotlib. I'm using the GTKAgg backend and the matplotlib.pyplot.show() command after labeling the axes with LaTeX formatted labels. I run into multiple copies of this error: libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 Traceback (most recent call last): File /sw/lib/python2.7/site-packages/matplotlib/backends/backend_gtk.py, line 436, in expose_event self._render_figure(self._pixmap, w, h) File /sw/lib/python2.7/site-packages/matplotlib/backends/backend_gtkagg.py, line 84, in _render_figure FigureCanvasAgg.draw(self) File /sw/lib/python2.7/site-packages/matplotlib/backends/backend_agg.py, line 440, in draw self.figure.draw(self.renderer) File /sw/lib/python2.7/site-packages/matplotlib/artist.py, line 54, in draw_wrapper draw(artist, renderer, *args, **kwargs) File /sw/lib/python2.7/site-packages/matplotlib/figure.py, line 1006, in draw func(*args) File /sw/lib/python2.7/site-packages/matplotlib/artist.py, line 54, in draw_wrapper draw(artist, renderer, *args, **kwargs) File /sw/lib/python2.7/site-packages/matplotlib/axes.py, line 2086, in draw a.draw(renderer) File /sw/lib/python2.7/site-packages/matplotlib/artist.py, line 54, in draw_wrapper draw(artist, renderer, *args, **kwargs) File /sw/lib/python2.7/site-packages/matplotlib/axis.py, line 1055, in draw tick.draw(renderer) File /sw/lib/python2.7/site-packages/matplotlib/artist.py, line 54, in draw_wrapper draw(artist, renderer, *args, **kwargs) File /sw/lib/python2.7/site-packages/matplotlib/axis.py, line 240, in draw self.label1.draw(renderer) File /sw/lib/python2.7/site-packages/matplotlib/artist.py, line 54, in draw_wrapper draw(artist, renderer, *args, **kwargs) File /sw/lib/python2.7/site-packages/matplotlib/text.py, line 576, in draw self._fontproperties, angle) File /sw/lib/python2.7/site-packages/matplotlib/backends/backend_agg.py, line 227, in draw_tex Z = texmanager.get_grey(s, size, self.dpi) File /sw/lib/python2.7/site-packages/matplotlib/texmanager.py, line 510, in get_grey X = read_png(os.path.join(self.texcache, pngfile)) RuntimeError: _image_module::readpng: png_create_read_struct failed Rebuilding and reinstalling matplotlib didn't help. So, I went a little deeper and rebuilt reinstalled gtk+2-dev, gtk+2-shlibs, glib2-dev, glib2-shlibs, and then rebuilding matplotlib-py27 didn't help. Thanks, Sean -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Am 04.04.2013 um 11:55 schrieb Sean Lake: Rebuilding and reinstalling matplotlib didn't help. So, I went a little deeper and rebuilt reinstalled gtk+2-dev, gtk+2-shlibs, glib2-dev, glib2-shlibs, and then rebuilding matplotlib-py27 didn't help. The reason for the report libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 is not matplotlib. It already has a dependency to libpng15 (see /sw/fink/10.4/stable/main/finkinfo/sci/matplotlib-py.info). Could be some (old?) component of matplotlib uses that very old libpng-1.2.41 – tetex is such a candidate… One option might be to check with 'otool -L a dylib file' all the the dylibs that matplotlib uses directly and indirectly by using a dylib that needs other dylibs… Enough Work for today! -- Greetings Pete Klingon function calls do not have 'parameters' - they have 'arguments' - and they ALWAYS WIN THEM. -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Rebuilding and reinstalling the tetex packages didn't help. Sean On Apr 4, 2013, at 7:27, Kurt Schwehr kurtschw...@yahoo.com wrote: -user list (I'm not on it) Pete, Thanks for responding to Sean! Sean, otool output would be good to know. What version of macosx and matplotlib are you using here? I can see that you're using python 2.7 (which is a good thing). -kurt On Apr 4, 2013, at 4:08 AM, Peter Dyballa peter_dyba...@freenet.de wrote: Am 04.04.2013 um 11:55 schrieb Sean Lake: Rebuilding and reinstalling matplotlib didn't help. So, I went a little deeper and rebuilt reinstalled gtk+2-dev, gtk+2-shlibs, glib2-dev, glib2-shlibs, and then rebuilding matplotlib-py27 didn't help. The reason for the report libpng warning: Application built with libpng-1.2.41 but running with 1.5.10 is not matplotlib. It already has a dependency to libpng15 (see /sw/fink/10.4/stable/main/finkinfo/sci/matplotlib-py.info). Could be some (old?) component of matplotlib uses that very old libpng-1.2.41 – tetex is such a candidate… One option might be to check with 'otool -L a dylib file' all the the dylibs that matplotlib uses directly and indirectly by using a dylib that needs other dylibs… Enough Work for today! -- Greetings Pete Klingon function calls do not have 'parameters' - they have 'arguments' - and they ALWAYS WIN THEM. -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
Am 04.04.2013 um 23:54 schrieb Sean Lake: Rebuilding and reinstalling the tetex packages didn't help. Sean, this can't change anything because teTeX is old and dead and therefore will stay with that old version of libpng. (At least it is now using a more secure version of it.) What I wanted to express, is if your matplotlib is using components from teTeX then it's likely that this teTeX component is noticing the younger libpng version and might start complaining. It mus be something with graphics, it must be something that exists since, say, ten years or more and hasn't changed much recently (is unsupported?) because it is still using that old interface. Another way to find the possible cause of the report is to check the INFO files of the packages on which matplotlib depends. And then the INFO files of these packages to see on which packages these depend. There must be some such dependancy to libpng-1.2.41. (I am not using matplotlib.) -- Greetings Pete Increase the size of your bike by at least *five* inches! -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users
Re: [Fink-users] matplotlib-py27 Problem
I was hoping that maybe tetex had a non-specific dependency on libpng and would compile with the newer one. Looking at the show-deps for tetex-base lists libpng3, so I would agree that's where the problem is. My guess is that the easiest fix would be to update matplotlib to depend on texlive instead of tetex. What's more, the last line of the rerror is the read_png function call in the texmanager.py file, so, yeah. Thanks, Sean On Apr 4, 2013, at 15:27, Peter Dyballa peter_dyba...@freenet.de wrote: Am 04.04.2013 um 23:54 schrieb Sean Lake: Rebuilding and reinstalling the tetex packages didn't help. Sean, this can't change anything because teTeX is old and dead and therefore will stay with that old version of libpng. (At least it is now using a more secure version of it.) What I wanted to express, is if your matplotlib is using components from teTeX then it's likely that this teTeX component is noticing the younger libpng version and might start complaining. It mus be something with graphics, it must be something that exists since, say, ten years or more and hasn't changed much recently (is unsupported?) because it is still using that old interface. Another way to find the possible cause of the report is to check the INFO files of the packages on which matplotlib depends. And then the INFO files of these packages to see on which packages these depend. There must be some such dependancy to libpng-1.2.41. (I am not using matplotlib.) -- Greetings Pete Increase the size of your bike by at least *five* inches! -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users -- Minimize network downtime and maximize team effectiveness. Reduce network management and security costs.Learn how to hire the most talented Cisco Certified professionals. Visit the Employer Resources Portal http://www.cisco.com/web/learning/employer_resources/index.html ___ Fink-users mailing list Fink-users@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.macosx.fink.user Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-users