R help - Oct 2004 - building an R package : where and how should my fortran library be loaded ?

Hi ,

I'm currently trying to make available 
a few fortran subroutines and R functions 
(which make interface to these subroutines),
as an R package.
I'm doing it under linux with R 1.9.0 
(but hope to do it for windows too).
I have trouble for loading my fortran code.
Here are the steps involved :

1) In R: 
                                      # Create tree for package
                                        # and include my R functions in it
R> package.skeleton(name="Geneland",
                 list=c("mcmcFmodel",
                   "simFmodel",
                   "rdiscr",
                   "tessel.post"),
                 path="/home/guillot/projets/flux/package/",
                 force=T)

                                        # copy my fortran code in the tree
R>    system("cp /home/guillot/projets/flux/package/fortran/*.f   
/home/guillot/projets/flux/package/Geneland/src/")

R> q()

2) In a Unix shell
$ R CMD build Geneland
$ R CMD check Geneland
$ R CMD build Geneland
$ R CMD INSTALL Geneland

3) Back to R
R> library("Geneland")

So far, it's OK, pure R functions work fine.
But  R functions calling  some Fortran via .Fortran
do not work, I get an error message like :
Error in .Fortran("mcmc", ...
Fortran function name not in load table

If I make 
R> dyn.load("/usr/lib/R/library/Geneland/libs/Geneland.so")

then it works .
I was thinking that my Geneland.so was loaded automatically
by the command library("Geneland")
Obviously, it's not.
Where and how should the library loaded ?


Gilles

_____________________________________________________________________
Gilles GUILLOT
INRA -D??partement Math??matiques et Informatique Appliqu??es

Unit?? de Mixte de Recherche INRA - INAPG - ENGREF
Institut National Agronomique de Paris-Grignon
16 rue Claude Bernard
75231 Paris cedex 5

Aile Claude Bernard
Niveau cours +3 ??tages
tel : +33 (0)1 44 08 72 71
fax : +33 (0)1 44 08 16 66
http://www.inapg.fr/ens_rech/mathinfo/personnel/guillot/welcome.html
______________________________________________________________________
-- 
_____________________________________________________________________
Gilles GUILLOT
INRA -D??partement Math??matiques et Informatique Appliqu??es

Unit?? de Mixte de Recherche INRA - INAPG - ENGREF
Institut National Agronomique de Paris-Grignon
16 rue Claude Bernard
75231 Paris cedex 5

Aile Claude Bernard
Niveau cours +3 ??tages
tel : +33 (0)1 44 08 72 71
fax : +33 (0)1 44 08 16 66
http://www.inapg.fr/ens_rech/mathinfo/personnel/guillot/welcome.html

On Fri, 22 Oct 2004, Gilles GUILLOT wrote:
> I'm currently trying to make available 
> a few fortran subroutines and R functions 
> (which make interface to these subroutines),
> as an R package.
> I'm doing it under linux with R 1.9.0 
Please update, as few of us have such an old system in use.
> (but hope to do it for windows too).
> I have trouble for loading my fortran code.
[...]
> I was thinking that my Geneland.so was loaded automatically
> by the command library("Geneland")
`Thinking'?  `Hoping', perhaps, but where did you read that?
> Obviously, it's not.
> Where and how should the library loaded ?
By your R code, via library.dynam.  A common `spell' is for package foo to 
include in foo/src/zzz.R

.First.lib <- function(libpath, pkgname)  
    library.dynam("foo", pkgname, libpath)

and it described in `Writing R Extensions'.  That also explains the 
different rules for packages with namespaces.

-- 
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 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595