R devel - May 2008 - OS_type

We've now implemented an OS_type field in a package's DESCRIPTION file, 
which if set has to match .Platform$OS.type (possible values "unix"
and
"windows").  CRAN has set up some overrides and added this field to 
packages known to be Unix-only or Windows-only, but authors should do so 
for themselves in future.

Unsuitable packages will not be offered by available.packages(), will not 
be installed by R CMD INSTALL, and R CMD check will select --no-install.

This happens in R-devel and R-patched, and is intended to simplify 
automated installation and checking.


-- 
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

On Mon, May 05, 2008 at 11:56:47AM +0100, Prof Brian Ripley
wrote:
> We've now implemented an OS_type field in a package's DESCRIPTION
file,
> which if set has to match .Platform$OS.type (possible values
"unix" and
> "windows").  CRAN has set up some overrides and added this field
to
> packages known to be Unix-only or Windows-only, but authors should do so  
> for themselves in future.
>
> Unsuitable packages will not be offered by available.packages(), will not 
> be installed by R CMD INSTALL, and R CMD check will select --no-install.
>
> This happens in R-devel and R-patched, and is intended to simplify  
> automated installation and checking.
Good idea.  

Just yesterday, someone was just wondering on the r-sig-finance list
why his Linux box was bringing in RBloomberg and thusly RDCOMClient
when retrieving the default task view set for Finance.  As maintainer
of the task view, I had no good answer other than the default of
howling at Achim and calling it a ctv feature.  

But yes, it really was a design issue and this should fix it. So thanks.

Dirk

-- 
Three out of two people have difficulties with fractions.