R devel - Sep 2005 - configuration and installation of R packages with C/C++ library dependencies.

Dear R-developers, Bioc-developers,

I am working on an R package which provides a R binding to a C library, 
which again depends on two other "non-standard" C++ libraries.
I have this libraries installed on my box of course and I specified the 
library location in the Makevars file.

However I thinking about to make the package available to other users 
and wonder

a) Where I should provide the information which C++ libraries must be 
installed on the computer? README/ DESCRIPTION file?
b) The library location on other users computers will be in a different 
place than specified by me in the Makevars file.  What mechanism of 
finding out the library locations, or prompting the user to specify them 
are integrated in the package installation procedure?
c) Which packages available on CRAN or BioConductor are *good* practice 
examples?


Thanks
Eryk

On Tue, 20 Sep 2005, Witold Eryk Wolski wrote:

Please ask R development questions on the R list and not cross-post.
> Dear R-developers, Bioc-developers,
>
> I am working on an R package which provides a R binding to a C library,
which
> again depends on two other "non-standard" C++ libraries.
> I have this libraries installed on my box of course and I specified the 
> library location in the Makevars file.
>
> However I thinking about to make the package available to other users and 
> wonder
>
> a) Where I should provide the information which C++ libraries must be 
> installed on the computer? README/ DESCRIPTION file?
As stated in `Writing R Extensions': the SystemRequirements field in 
DESCRIPTION or a README file.
> b) The library location on other users computers will be in a different
place
> than specified by me in the Makevars file.  What mechanism of finding out
the
> library locations, or prompting the user to specify them are integrated in 
> the package installation procedure?
As stated in `Writing R Extensions' section 1.2: use a configure file.
> c) Which packages available on CRAN or BioConductor are *good* practice 
> examples?
Several, e.g. RODBC (which is used as an example in `Writing R 
Extensions', section 1.2).


Did we mention this was all in `Writing R Extensions'?

-- 
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