[R] Porting Custom Packages from Linux to Windows
marc_schwartz at me.com
Wed Oct 28 02:40:18 CET 2009
On Oct 27, 2009, at 8:32 PM, Duncan Murdoch wrote:
> Marc Schwartz wrote:
>> On Oct 27, 2009, at 6:20 PM, David Lubbers wrote:
>>> On my linux workstation I have a nice, very little project in a
>>> directory called "mystuff"
>>> I installed "mystuff" into site-library with R CMD build, R CMD
>>> and R CMD install. This was all with the file mystuff.tar.gz.
>>> Now I want to install "mystuff" on my windows workstation. Can I
>>> the same file mystuff.tar.gz. Do I need to do something
>>> different with
>>> R CMD build?
>>> Ideally somebody has a tutorial for porting from linux to windows
>>> Thanks for your time.
>> The package source tar file is not directly installable on Windows.
> That's not true. You just need to install the tools, and then it's
> similar to Linux. The tools are available from http://www.murdoch-sutherland.com/Rtools
> Duncan Murdoch
Right, which I point to below Duncan. I was referring to a typical R
install on Windows, where your tools would not normally be available.
>> You need to create a .zip 'binary' package file.
>> There is a cross-building document here by Rossini and Yan:
>> and the associated makefile here:
>> I have not looked at them in detail relative to their applicability
>> to current versions of R, as they are a rather dated (circa 2003).
>> So caveat emptor.
>> If you might consider just directly building the packages on your
>> Windows system, Duncan provides an easy Windows installation that
>> contains the requisite tools (C and FORTRAN compilers, vanilla
>> Perl, etc.) here:
>> along with others that are referenced in the R-Admin Manual
>> Windows Toolset Appendix:
>> would enable you to build your package on Windows, including any C
>> or FORTRAN code that you might have in the package.
>> Marc Schwartz
>> R-help at r-project.org mailing list
>> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
More information about the R-help