On 13-11-03 2:40 PM, Adam Clark wrote:> I apologize for what is likely a stupid question - I HAVE looked through
> "Writing R Extensions" and HAVE gone through existing help
requests, but
> have not yet found an answer. I'm sorry if I've missed something
obvious.
>
> I am planning on building a package in the near future. I have a number of
> C programs that I have written to interface with R (compiled using the R
> SHLIB, run using the .C convention).
>
> I now run a Linux system, but seem to remember that for Mac and Windows
> implementations of R, Rtools is required separately in order to run
> compiled C code that interfaces directly with R.
>
> I am wondering whether any packages that I write using such C code will
> require end users of PC and Mac to have Rtools installed, or whether there
> is a way to build the package such that this is not the case.
>
> I ask this because I would like the package to be relatively painlessly
> portable, and know many users who have had trouble installing Rtools on
> their Mac or PC.
Windows users require Rtools to develop packages containing C code, but
not to run them, as long as someone is producing a "binary" package
for
them. If you send your package to CRAN that will happen automatically.
If you don't, you'll need to find some other way to produce the
binaries (e.g. WinBuilder).
Mac users are similar, but don't use Rtools, they use a different
collection of tools.
So you should be fine: if you write portable C and send your package to
CRAN, you don't need to worry.
Duncan Murdoch