[R] Re: R-1.1.0 on Alpha?

Prof Brian D Ripley ripley at stats.ox.ac.uk
Thu Jul 13 09:00:00 CEST 2000


On Thu, 13 Jul 2000, Ray Brownrigg wrote:

> I asked:
> > Has anyone succeeded in compiling R-1.1.0 on a Digital (Compaq) Alpha
> > running OSF4.0F (or E)?  Everything seems to compile OK (except the
> > frequent "Warning:  Unresolved:" at link time - this threw me off the
> > scent for a while!), but then at run time I get:
> > 
> > Fatal error: The X11 shared library could not be loaded.
> >   The error was dlopen: cannot load ....../R-1.1.0/bin/R_X11.so 
> > 
> Well I've finally solved it - it was an 'rpath' problem!
> 
> For completeness, here are the changes I had to make to config.site to
> get R running on Digital Unix 4.0F, using f77 and gcc.  Systems with g77
> may not have some of these problems.
> 
> 1) Add:
> CPPFLAGS="-I/usr/local/include -I/usr/X11R6/include"
> otherwise there are problems compiling tcltk

That is intended to be fixed already in R-patched: could you try that?
(It sets PKG_CPPFLAGS in src/library/tcltk/src/Makefile. Whether this is
needed seems to depend on the Tcl version.)

> 2) Add:
> LIBS="-L../../.. -L/usr/local/lib"
> to include a 'private' copy of readline.a, otherwise I get the following
> when linking R.bin:
> ld:
> Object file format error in:
> /usr/local/lib/libreadline.a(tilde.o):local_is_complete:
>  iaux(21) > iauxMax(14) for obj /usr/local/lib/libreadline.a(tilde.o)
> fort: Severe: Failed while trying to link. 
> 
> As far as I can remember, my 'private' libreadline.a has the -g option
> specified, whereas the system one does not.  [So this may not be a
> problem in other systems.]

Looks like a compiler-mix problem to me, as you hint.  Let us assume that
is specific to you.

> 3) Add:
> MAINLDFLAGS="-Wl,-rpath -Wl,/usr/X11R6/lib"
> This is the key to setting appropriate shared library paths in the main
> program.  Under R-1.0.0 this flag was known as LDFLAGS.

Um.  The X11 flags are found from xmkmf. My Solaris box has

X_LIBS =  -L/usr/openwin/lib -R/usr/openwin/lib -lX11

for example, and R_X11.so is built with $(X_LIBS).  So is the -rpath
missing in X_LIBS, or is there something else wrong?  Can you take a 
look at the command used to build R_X11.so (in src/unix/X11)?
My understanding is that setting the rpath on the dynamic library
(not the main executable) should suffice, but perhaps DU-4.0 is different
from everything else.

The simplest way out is to add /usr/X11R6/lib to R_LD_LIBRARY_PATH
in the R front-end script, I guess.  It might be desirable to do the
equivalent of that on all platforms.

-- 
Brian D. Ripley,                  ripley at 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 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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