[R] s_copy

Roger Koenker roger at ysidro.econ.uiuc.edu
Tue Aug 24 20:33:40 CEST 1999

A brief report on the s_copy problem described below.

It turns out that s_copy at least in my version of redhat linux is part of
the f2c library.  There is a routine scopy in the blas, but this is different,
I believe.  I managed to get things to work by putting -lf2c on the 

	R SHLIB -o rq.so *.f

line.  It seems that one can also accomplish this by setting, 

	FFLAGS = -lf2c

in Makeconf, but this is probably a bad idea.  If someone has a good
suggestion on how to do this for purposes of making a package, I would
be grateful.

On a related note, I would like to make a plea for a more inclusive version
of the blas, and lapack in general.  It would be nice to have the level 3
blas available, and even better if the rest of lapack were also there.

url:	http://www.econ.uiuc.edu		Roger Koenker	
email	roger at ysidro.econ.uiuc.edu		Department of Economics
vox: 	217-333-4558				University of Illinois
fax:   	217-244-6678				Champaign, IL 61820

On Tue, 24 Aug 1999, Prof Brian D Ripley wrote:

> On Mon, 23 Aug 1999, Roger Koenker wrote:
> > 
> > 
> > I'm trying to port some routines from S to R.  The code includes some
> > fortran stuff that calls lapack.  In the past in similar situations, but
> > without any lapack calls, I've had success using simply, as in the 
> > V&R R-complements,
> > 
> >         R SHLIB -o rq.so *.f
> > 
> > then in R
> > 
> >         dyn.load("rq.so")
> > 
> > But now when the code is called from R, R crashes and I get a message
> > 
> >         undefined symbol:  s_copy 
> > 
> > This reference is evidently resolved in the Splus linking, but I vaguely
> > recall problems like this in ancient versions of Splus.
> > 
> > I'm running V63.2 of R for i586/linux.
> > 
> > If someone could provide some diagnosis, or suggestions for further testing,
> > I would be most grateful.
> I had exactly that (and two others) on early versions of the mclust port.
> If I recall correctly, it means you need to load in more of LAPACK,
> as I believe s_copy is a LAPACK utility routine: it is certainly
> in the S executable to link against.
> If you use 0.65.0 (due for release on Friday) it will tell you at
> dyn.load time about missing symbols. I do recommend an upgrade.
> -- 
> 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