On Tue, Aug 9, 2011 at 6:24 PM, Simon Urbanek
<simon.urba...@r-project.org> wrote:
> It actually works ;) I'm using it for testing on my RForge.net machine and 
> yes, it's Debian - everything just works there :).
>
> But back to the original question. First a minor detail, don't set 
> environment variables use configure variables instead. Second, don't build in 
> the source directory, always create an object directory. Third, r_arch is 
> simply a name you set for the architecture, it has no meaning other than that 
> it's a label.
>
> So now to the real stuff. If you want 32-bit build, you'll need 32-bit 
> runtime of everything important in your system and the multilib compilers. In 
> Debian (and thus likely in Ubuntu too) that can be achieved by something like
>
> sudo apt-get install  ia32-libs-dev lib32readline6-dev lib32ncurses5-dev 
> lib32icu-dev gcc-multilib gfortran-multilib
>
> Then you can build both 64-bit and 32-bit R, the difference will be in the 
> all compiler flags -- for 64-bit you'll use -m64 (or nothing since it's the 
> default) and for 32-bit you'll use -m32.
>
> So roughly something like
>
> tar fxz R-2.13.1.tar.gz
> mkdir obj-32
> cd obj-32
> ../R-2.13.1/configure r_arch=i386 CC='gcc -std=gnu99 -m32' CXX='g++ -m32' 
> FC='gfortran -m32' F77='gfortran -m32'
> make -j24 && sudo make install rhome=/usr/local/R/2.13
> cd ..
> mkdir obj-64
> cd obj-64
> ../R-2.13.1/configure r_arch=amd64
> make -j24 && sudo make install rhome=/usr/local/R/2.13
>
> That will leave you with multi-arch R that you can run with
> R --arch=i386 # 32-bit
> R --arch=amd64 # 64-bit
> Packages will be also built as multi-libs. Good luck :)
> [BTW the rhome=... setting is entirely optional, I just like to keep my R 
> versions organized…]
>
> Cheers,
> Simon
>

Thanks Simon!  Confirm that these instructions work.  ia32-libs-dev
was not available for Ubuntu Natty, so I installed ia32-libs instead,
and the compilation works!

-- Vinh

______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel

Reply via email to