On 29 Sep 2009, at 23:21, Stefan Bruda wrote: > In any event, here is how the thing fails to build: > > gcc -fno-strict-aliasing -mno-fused-madd -DNDEBUG -g -fwrapv -O3 - > Wall -Wstrict-prototypes -Ibuild/codegen/ -Ilibffi-src/include -I/ > sw64/include/python2.5 -c libffi-src/src/types.c -o build/ > temp.macosx-10.5-i386-2.5/libffi-src/src/types.o -I/sw64/include - > DMACOSX -DAPPLE_RUNTIME -no-cpp-precomp -Wno-long-double -g -Wall - > Wstrict-prototypes -Wmissing-prototypes -Wformat=2 -W -Wshadow - > Wpointer-arith -Wmissing-declarations -Wnested-externs -Wno-long- > long -Wno-import
This is very puzzling : also on 64-bit Fink, Core2Duo, OS 10.5.8, I get instead _ in an apparently succesfull (!...) build of pyobjc- py25-2.0+svn2008-4 : gcc -fno-strict-aliasing -mno-fused-madd -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -I/sw64/include -I/sw64/include/python2. 5 -c libffi-src/types.c -o build/temp.macosx-10.5-i386-2.5/libffi-src/ types.o -DPyObjC_STRICT_DEBUGGING -DMACOSX -no-cpp-precomp -W no-long-double -g -fexceptions -O1 -Wall -Wstrict-prototypes -Wmissing- prototypes -Wformat=2 -W -Wshadow -Wpointer-arith -Wmissing- declarations -Wnested-externs -Wno-long-long -Wno-import -I/sw64/ include/libxml2 -I/sw64/include -Ibuild/codegen/ -Ilibffi-src/incl ude -Ilibffi-src/powerpc Note all the discrepancies ! In particular : 1) I get this additional (and apparently "wrong") flag for "-Ilibffi-src/powerpc" [recall a couple of lines before there is > creating build/temp.macosx-10.5-i386-2.5/libffi-src > creating build/temp.macosx-10.5-i386-2.5/libffi-src/powerpc > creating build/temp.macosx-10.5-i386-2.5/libffi-src/x86 ] So I would rather have expected "-Ilibffi-src/x86" in addition... On the other hand, it is YOU who is getting "Unsupported MacOS X CPU type" So your build seems possibly even more correct than mine .. (But this stupid pkg has no configure, and no config.log .. no way to know what happened w/o studying a whole new build system...) 2) On the other hand, I get the flag "-I/sw64/include/libxml2", which you don't, and (as well on 32bit) otool shows the pkg depends on libxml2 and libiconv; those deps and bdeps are apparently missing. It is not completely fake ; nm confirms that %p/lib/python2.5/site-packages/PyObjC/objc/_objc.so does use symbols from libxml2. On the other hand %p/lib/python2.5/site-packages/PyObjC/objc/test/*.so all depend according to otool on both libiconv and libxml2, but none of them uses any symbol from either. 3) The ordering of the flags is changed. Is there someone knowing this type of build-system, and who has an idea what might be going on ? But please, Stefan, to get clear on the trivial items first : I'm getting > # fink list -it libiconv libxml2 > Information about 9232 packages read in 2 seconds. > i libiconv 1.12-3 Character set conversion library > i libiconv-bin 1.12-3 Executables for libiconv package > i libiconv-dev 1.12-3 Developer files for libiconv package > i libxml2 2.6.32-1 XML parsing library, version 2 > i libxml2-bin 2.6.32-1 XML parsing library, version 2 > i libxml2-py25 2.6.32-1 Python bindings for libxml2 library > i libxml2-py26 2.6.32-1 Python bindings for libxml2 library > i libxml2-shlibs 2.6.32-1 XML parsing library, version 2 What do you get from this command ? Thanks a lot for those very helpful reports ! JF Mertens PS : Didn't even try if my above "apparently succesfull builds" were usable or not; clearly the above questions about the build itself don't depend on that. ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf _______________________________________________ Fink-users mailing list Fink-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/fink-users
