[R-SIG-Mac] R 2.1.1 tiger compile problem

Bill Northcott w.northcott at unsw.edu.au
Mon Sep 26 14:17:36 CEST 2005


On 22/09/2005, at 8:00 PM, Joel Bremson wrote:
> I'm trying to compile R-2.1.1 on a Tiger/G5 - Darwin 8.2.0 - XCode  
> 2.1.
> The gcc is:
> powerpc-apple-darwin8-gcc-4.0.0 (GCC) 4.0.0 (Apple Computer, Inc.  
> build
> 5026)
>
> The gfortran is:
> GNU Fortran 95 (GCC 4.0.0 (Apple Computer, Inc. build 5202))
>
> My configure line is:
> ./configure F77=gfortran FLIBS=/usr/lib/libgfortran.a
>
> The make runs for a long while and then crashes out with this output:
> gfortran -fno-common -g -O2 -c cmplxblas.f -o cmplxblas.lo
> gcc -dynamiclib -L/sw/lib -L/usr/local/lib -install_name
> /Library/Frameworks/R.framework/Version

Your problem is not the compiler.  It is the configure command that  
is wrong.  There is no problem building R with gfortran and gcc-4.   
The F77 define is probably necessary, but you should not need to add  
standard Fortran libraries.

The recommended configure command is
./configure --with-blas='-framework vecLib' --with-lapack --with-aqua
see http://cran.r-project.org/bin/macosx/RMacOSX-FAQ.html

BLAS and LAPACK are part of the Accelerate framework in MacOS X.   
Because you have omitted the '--with-blas' and '--with-lapack'  
options R is trying to build its own LAPACK which AFAIK is not  
supported on MacOS X.  Hence your problems.

Bill Northcott



More information about the R-SIG-Mac mailing list