See the comment in NEWS. We know MacOS X does not currently build *on
some installations* unless vecLib is installed (see R-admin.texi).
On Thu, 6 Feb 2003, Roger Koenker wrote:
> I've been running a binary installation (kindly provided by Jan de
Leeuw)
> on mac G4, but was inspired this morning to try to install 1.7.0 on the G4
I presume you mean R-devel? 1.7.0 is not released, and this will be
addressed before release (in a couple of months).
> as well as on my desktop solaris box.
>
> This worked smoothly on my solaris system but on the G4 the
> make dies shortly after compiling lapack with:
>
> /usr/bin/libtool: internal link edit command failed
> make[4]: *** [libRlapack.dylib] Error 1
> make[3]: *** [R] Error 2
> make[2]: *** [R] Error 1
> make[1]: *** [R] Error 1
> make: *** [R] Error 1
>
> The output from configure seemed ok...but there were lots of messages of
the following
> sort that I thought might also suggest that something was wrong.
>
> /usr/include/gcc/darwin/2.95.2/g++/../machine/limits.h:7: #error
architecture not supported
> In file included from /usr/include/sys/signal.h:70,
> from /usr/include/signal.h:62,
> from ../../../src/include/Defn.h:60,
> from sockconn.c:32:
> /usr/include/machine/signal.h:31: #error architecture not supported
> In file included from /usr/include/setjmp.h:25,
> from ../../../src/include/Defn.h:61,
> from sockconn.c:32:
I am told that is not the current gcc compiler for MacOS X (and it is way
old for gcc), and I suspect nothing before gcc 3.1 is going to work.
R-devel needs the ability to make dynamic libraries including Fortran
code, and MacOS X seems to have troubles doing that. Using vecLib this is
avoided, and it seems to work out of the box, and without it seems to need
a recent compiler (but Jan de Leeuw has built it).
--
Brian D. Ripley, ripley@stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595