Skip to content

library problem (leopard v. tiger)

3 messages · Andrew Beckerman, Simon Urbanek

#
Dear all -

10.5.1, R 2.6.1.

I am trying to get a biocinductor package called GeneticsPed to work  
with leopard and R2.6.1.  It worked on Tiger with 2.5.

I have tried to install via Bioconductor, via the binary as a  
download, and as a source file.

Before documenting the behaviours, please note, that I have the latest  
Xcode installed, and also the latest GCC and gfortran installed  
through HPC.

1) Bioconductor can't find it, and that's ok, because it only exists  
now in the Devel version.

2) When I try and load the installed binary, I get the following error  
(note that it appeast to be built against 2.5?).

 > library(GeneticsPed)
Loading required package: gdata

Attaching package: 'gdata'


	The following object(s) are masked from package:Hmisc :

	 combine,
	 reorder.factor

Error in dyn.load(file, ...) :
   unable to load shared library '/Library/Frameworks/R.framework/ 
Resources/library/GeneticsPed/libs/i386/GeneticsPed.so':
   dlopen(/Library/Frameworks/R.framework/Resources/library/ 
GeneticsPed/libs/i386/GeneticsPed.so, 6): Library not loaded: /Library/ 
Frameworks/R.framework/Versions/2.5/Resources/lib/libgfortran.2.dylib
   Referenced from: /Library/Frameworks/R.framework/Resources/library/ 
GeneticsPed/libs/i386/GeneticsPed.so
   Reason: image not found
Error: package/namespace load failed for 'GeneticsPed'
 >

3) When I try and install via source, I get the following error  
(indicating use of the 10.4SDK?):

/Library/Frameworks/R.framework/Resources/library
* Installing *source* package 'GeneticsPed' ...
** libs
** arch - i386
g++ -arch i386 -isysroot /Developer/SDKs/MacOSX10.4u.sdk -mmacosx- 
version-min=10.4 -no-cpp-precomp -I/Library/Frameworks/R.framework/ 
Resources/include -I/Library/Frameworks/R.framework/Resources/include/ 
i386  -msse3    -fPIC  -g -O2 -march=nocona -c ainverse.cc -o ainverse.o
g++ -arch i386 -isysroot /Developer/SDKs/MacOSX10.4u.sdk -mmacosx- 
version-min=10.4 -no-cpp-precomp -I/Library/Frameworks/R.framework/ 
Resources/include -I/Library/Frameworks/R.framework/Resources/include/ 
i386  -msse3    -fPIC  -g -O2 -march=nocona -c eibd.cc -o eibd.o
gfortran -arch i386  -isysroot /Developer/SDKs/MacOSX10.4u.sdk   - 
fPIC  -g -O2 -march=nocona -c ggmatmult.f -o ggmatmult.o
** arch - ppc
g++ -arch ppc -isysroot /Developer/SDKs/MacOSX10.4u.sdk -mmacosx- 
version-min=10.4 -no-cpp-precomp -I/Library/Frameworks/R.framework/ 
Resources/include -I/Library/Frameworks/R.framework/Resources/include/ 
ppc  -I/usr/local/include    -fPIC  -g -O2 -c ainverse.cc -o ainverse.o
f951: error: unrecognized command line option "-arch"
make: *** [ggmatmult.o] Error 1
chmod: /Library/Frameworks/R.framework/Resources/library/GeneticsPed/ 
libs/i386/*: No such file or directory
g++ -arch ppc -isysroot /Developer/SDKs/MacOSX10.4u.sdk -mmacosx- 
version-min=10.4 -no-cpp-precomp -I/Library/Frameworks/R.framework/ 
Resources/include -I/Library/Frameworks/R.framework/Resources/include/ 
ppc  -I/usr/local/include    -fPIC  -g -O2 -c eibd.cc -o eibd.o
gfortran -arch ppc  -isysroot /Developer/SDKs/MacOSX10.4u.sdk   -fPIC   
-g -O2 -c ggmatmult.f -o ggmatmult.o
** Removing '/Library/Frameworks/R.framework/Resources/library/ 
GeneticsPed'
** Restoring previous '/Library/Frameworks/R.framework/Resources/ 
library/GeneticsPed'
f951: error: unrecognized command line option "-arch"
make: *** [ggmatmult.o] Error 1
chmod: /Library/Frameworks/R.framework/Resources/library/GeneticsPed/ 
libs/ppc/*: No such file or directory
ERROR: compilation failed for package 'GeneticsPed'

Any advice or obvious comments on what to do?

Cheers
andrew
---------------------------------------------------------------------------------
Dr. Andrew Beckerman
Department of Animal and Plant Sciences, University of Sheffield,
Alfred Denny Building, Western Bank, Sheffield S10 2TN, UK
ph +44 (0)114 222 0026; fx +44 (0)114 222 0002
http://www.beckslab.staff.shef.ac.uk/
1 day later
#
Andrew,
On Jan 7, 2008, at 10:57 AM, Andrew Beckerman wrote:

            
HPC fortran doesn't work (it has several problems but the main one is  
that it is not universal). You'll have to use a universal  fortran  
compiler and we strongly recommend the use of the one we supply. (See  
also http://r.research.att.com/tools/ for details).
Are you sure it's not a problem on your end pulling packages from the  
wrong place (see .libPaths())? It is the most common mistake to forget  
to wipe old 2.5 packages from custom lib paths.

I'm not building BioC packages anymore, but I have checked the binary  
from http://www.bioconductor.org/packages/2.2/bioc/bin/macosx/universal/contrib/2.7/GeneticsPed_1.1.0.tgz
and that works just fine:
i386/GeneticsPed.so:
	GeneticsPed.so (compatibility version 0.0.0, current version 0.0.0)
	/Library/Frameworks/R.framework/Versions/2.6/Resources/lib/ 
libgfortran.2.dylib (compatibility version 3.0.0, current version 3.0.0)

ppc/GeneticsPed.so:
	GeneticsPed.so (compatibility version 0.0.0, current version 0.0.0)
	/Library/Frameworks/R.framework/Versions/2.6/Resources/lib/ 
libgfortran.2.dylib (compatibility version 3.0.0, current version 3.0.0)
Yes, the R you are using is a Tiger build, so it makes sure everything  
stays Tiger-compatible.
Problem with HPC fortran - please use the GNU Fortran that comes with R.

Cheers,
Simon
#
Simon -

Many thanks for the advice.  I've re-installed 2.6.1 with the GNU  
fortran with it.  GeneticsPed 1.1.0 installed and loads fine now.   
Don't quite know what happened before.

Do I need to remove the HPC fortran/GCC stuff now?  Is that prudent?

Cheers
andrew


---------------------------------------------------------------------------------
Dr. Andrew Beckerman
Department of Animal and Plant Sciences, University of Sheffield,
Alfred Denny Building, Western Bank, Sheffield S10 2TN, UK
ph +44 (0)114 222 0026; fx +44 (0)114 222 0002
http://www.beckslab.staff.shef.ac.uk/
----------------------------------------------------------------------------------
On 8 Jan 2008, at 15:59, Simon Urbanek wrote: