Posted: Sun Dec 01, 2002 11:59 pm
Joined: 14 Oct 2002
Posts: 80
It compiles on OS X until it gets to this:
"/bin/sh ../libtool --mode=link gcc -g -O2 -o blenderpublisher.app -L/usr/local/lib/python2.2/config...........................ld: table of contents for archive: ../intern/python/frozen/libfrozen.a is out of date; rerun ranlib(1) (can't load from it)
*** Error code 1"
I ran ranlib -s and -c libfrozen.a then reran make but came up with the same message.
Posted: Mon Dec 02, 2002 4:12 am
Joined: 15 Oct 2002
Posts: 10
Try a "make clean" and rebuild. But my gut feelings tell me the same thing is going to happen...
I'm wondering if libfrozen is being built properly. Or even at all.
_________________
echo '70:68:61:65:74:68:6F:6E:40:6C:69:6E:75:78:2E:75:63:6C:61:2E:65:64:75:' |
perl -npe 's/([^:]+):/chr(hex($1))/ge'
Posted: Mon Dec 02, 2002 7:48 am
Joined: 14 Oct 2002
Posts: 80
I did a make clean and rebuild, with the same results. A libfrozen.a does in fact exist in the python/frozen dir. it is a 396k file, and I'm able to open it with textedit. The first part of libfrozen.a as compiled:
!<arch>
#1/16
1038783478 501 20 100644 1340 `
__.SYMDEF SORTED__PyImport_Inittab
As modified with ranlib -s:
!<arch>
#1/16
1038813716 501 20 100644 1340 `
__.SYMDEF SORTED
And as modified with ranlib -c:
!<arch>
#1/16
1038813787 501 20 100644 1340 `
__.SYMDEF SORTED
Posted: Mon Dec 02, 2002 9:34 am
Joined: 13 Oct 2002
Posts: 939
I would do the following:
make dist clean
foreach i (`find . -name Makefile*`)
rm $i
end
Then do a cvs update (twice)
The first time to restore the Makefile stuff.
The second time to see if there are any files left over in the python dir.
(On my system there are a bunch of .pyo files that I need to remove by hand,
I'm using a seperate directory for the build and its still building some of the python stuff
in the source dir which is sort of a problem if you want to do parallel builds on different systems)
Posted: Wed Dec 04, 2002 6:14 am
Joined: 14 Oct 2002
Posts: 80
I downloaded (cvs co blender) for a clean start, went through the complete build and ended up with the same results:
"/bin/sh ../libtool --mode=link gcc -g -O2 -o blenderpublisher.app -L/usr/local/lib/python2.2/config...........................ld: table of contents for archive: ../intern/python/frozen/libfrozen.a is out of date; rerun ranlib(1) (can't load from it)
*** Error code 1"
The new configure.ac no longer finds the OS X /system......../gl.h and glu.h location. ./configure only seems to know about --build=powerpc the configure.ac has *-apple-darwin*.
Posted: Wed Dec 04, 2002 5:21 pm
Joined: 13 Oct 2002
Posts: 939
Would *darwin* work better maybe?
Posted: Wed Dec 04, 2002 10:51 pm
Joined: 14 Oct 2002
Posts: 80
*darwin* would be great, as long as configure knows about it. Why can't configure find the host?
Posted: Thu Dec 05, 2002 4:12 pm
Joined: 16 Oct 2002
Posts: 49
try replacing config.guess and/or config.sub with the ones from /usr/share/automake-1.6
HTH,
sgefant
Posted: Thu Dec 05, 2002 11:44 pm
Joined: 14 Oct 2002
Posts: 80
| sgefant wrote: |
try replacing config.guess and/or config.sub with the ones from /usr/share/automake-1.6
HTH,
sgefant |
Replacing both seems to work. 
[PowerBookG4 1:40:45pm michaelv] ~/Desktop/blender/build% ../configure
checking build system type... powerpc-apple-darwin6.2
checking host system type... powerpc-apple-darwin6.2
checking target system type... powerpc-apple-darwin6.2
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
ld: table of contents for archive: ../intern/python/frozen/l