[R] choosing g/g++ compiler executable in R CMD SHLIB

Dirk Eddelbuettel edd at debian.org
Sat Nov 17 18:53:23 CET 2001

  "Faheem" == Faheem Mitha <faheem at email.unc.edu> writes:
  Faheem>  Dear People,
  Faheem> After considerable effort, I managed to install gcc-3.0 on my
  Faheem> Debian system. However, Debian keeps gcc-2.95 around as the default

Provided you told apt about testing or unstable, all you need is 

     $ apt-get update; apt-get install gcc-3.0

What is wrong with that?

  Faheem> compiler, and gcc, g++ point to the 2.95 version. The 3.0 versions

Depends. Some architectures (hppa) only have the 3.0* branch of gcc.

  Faheem> need to be explicitly invoked as gcc-3.0, g++-3.0 etc. How can I
  Faheem> get R CMD SHLIB to invoke gcc-3.0 instead of gcc and g++-3.0
  Faheem> instead of g++? I would like this to be done by default but only
  Faheem> for R. That is, I would like everything except R CMD SHLIB to call
  Faheem> the 2.95 versions by default, but have R CMD SHLIB to call the 3.0
  Faheem> versions by default.

One full-blown way is to re-configure and build R with these.  An easy is to
(provided you told apt about source archives)

     $ apt-get source r-base
     $ cd r-base-1.3.1
     [ go off and edit debian/rules; just uncomment my stanza about using
       gcc/g++/g77-3.0 on i386 in lines 62-68 in the r-base_1.3.1-5 package
       currently in http://incoming.debian.org ] 
     [ you also want to change debian/changelog to make the revision, say,
       1.3.1-5local to protect it from being overwritten by the one from the
       archive ] 
     $ fakeroot dpkg-buildpackage
     $ su -c "dpkg -i ../r-base_1.3.1-5local_i386.deb"

and you should be all set.

A more hackish way is to simply define CC=gcc-3.0, CXX=g++-3.0, FC=g77-3.0
and to export those. You will end up mixing gcc-2.95 and gcc-3.0, not sure if
that works.
  Faheem> I didn't see this issue addressed in "Writing R Extensions". Is
  Faheem> there a way to do this?

It is way too distribution-specific to be covered there, and almost too
specific for this list.

Hope this helps. Bug me privately if you need more help.

Cheers, Dirk

Better to have an approximate answer to the right question 
than a precise answer to the wrong question. -- John Tukey
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list