R devel - Oct 2010 - Rcmd build 32bit-only package on win7 64 bit platform?

Is it possible to suppress compiling for the x64 arch when doing 'Rcmd
build <mypkg>? Such that the pre-compiled package for binary
distribution only contains a 'lib/i386/mypkg.dll'.

For CHECK (and also INSTALL) there is a '--no-multiarch' argument, but
I didn't find something similar for build.

Otherwise, I suppose, I could
- handle the problem in Makevars or Makefile or
- build the shlib manually and put it at the correct place or
- provide a placeholder library with C (my code is Delphi, 64bit
should/will come next year)
- (build inside a XP32bit VM)

Are there any other packages which are 32bit only (for now)?

Thanks for any advise,
Hans-Peter

On 18.10.2010 22:01, Hans-Peter Suter wrote:
> Is it possible to suppress compiling for the x64 arch when doing 'Rcmd
> build<mypkg>?

R CMD build <mypkg> does not compile anything except for creating the 
vignettes. Is this the case?
If so: During the pre-release cycle of R-2.12.0 we have not got any 
report telling us this is a desirable feature.

> Such that the pre-compiled package for binary
> distribution only contains a 'lib/i386/mypkg.dll'.
>
> For CHECK (and also INSTALL) there is a '--no-multiarch' argument,
but
> I didn't find something similar for build.

For now, you can use an only 32-bit installation of R in order to build. 
Passing down a flag such as --no-multiarch will be considered.

> Otherwise, I suppose, I could
> - handle the problem in Makevars or Makefile or
> - build the shlib manually and put it at the correct place or
> - provide a placeholder library with C (my code is Delphi, 64bit
> should/will come next year)
> - (build inside a XP32bit VM)
>
> Are there any other packages which are 32bit only (for now)?
Yes, http://cran.r-project.org/bin/windows/contrib/2.12/@ReadMe tells us:

Packages
     Cairo, RBrownie, RSvgDevice, RSVGTipsDevice, SNPassoc, coxphf, eco,
     gcmrec, magnets, proj4, ripa, sampfling, survrec, yaml
are only available for 32-bit installations.

Best,
Uwe Ligges
> Thanks for any advise,
> Hans-Peter
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel