[Rd] Re: [R] Newbie questions

Duncan Temple Lang duncan@research.bell-labs.com
Tue, 31 Oct 2000 14:57:36 -0500 (EST)

 -->>> Moved to r-devel from r-help. <<<---

> Date: Tue, 31 Oct 2000 19:36:40 +0000
> From: Zsombor Cseres-Gergely <z.cseres-gergely@ucl.ac.uk>
> Cc: r-help@stat.math.ethz.ch
> References: <Pine.GSO.4.21.0010301459460.14728-100000@atlas.biostat.washington.edu> <6rd7ghhpum.fsf@franz.stat.wisc.edu>
> Sender: owner-r-help@stat.math.ethz.ch
> Precedence: bulk
> On Mon, Oct 30, 2000 at 05:27:29PM -0600, Douglas Bates wrote:
> > You may want to try "make check" after configuring and compiling R.
> > That should use the postscript graphics device driver rather than
> > anything related to X.  We would be very interested in hearing if that
> > succeeds.
> OK, one good and a bad news. Good news is that R runs with option --gui=none.
> It does demo()-s that does not involve graphics. 

That's good to hear.

> Bad news is that make check fails saying: 
> Fatal error: The X11 shared library could not be loaded.
>   The error was /home/zs/R-1.1.1/bin/R_X11.so: undefined symbol: R_GlobalEnv

Just a quick thought....

Rather than being a problem with X11, it looks like it may be a
problem associated with dlopen() when the R_X11.so library is
explicitly loaded during the initialization. Basically, that `library'
(R_X11.so) has to resolve all of the symbols to which it refers, and
many of them are in the R executable (R.bin). Perhaps the linking and
loading mechanisms in Hurd do not export these symbols from the
application in a way that shared libraries can find them.  Or it may
be as simple as the flag passed to dlopen() not being RTLD_GLOBAL
(although that shouldn't matter).

A mechanism that Luke and I chatted about that could tidy up some of
these things is to link both R.bin and also R_X11.so against libR.so.
So R.bin would consist of a single routine main() that initialized R
and run the mainloop.  As long as this doesn't introduce a noticeable
performance penalty, it might be worth doing to simplify setups like
this one and reduce the dependency on the non-standard behaviour of



Duncan Temple Lang                duncan@research.bell-labs.com
Bell Labs, Lucent Technologies    office: (908)582-3217
700 Mountain Avenue, Room 2C-259  fax:    (908)582-3340
Murray Hill, NJ  07974-2070       

      "Languages shape the way we think, and determine what 
       we can think about."        
                                      Benjamin Whorf
r-devel 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-devel-request@stat.math.ethz.ch