[R] executable R script under xp (to avoid migration toward Matlab or C++)

[Barry Rowlingson]

On Mon, Feb 15, 2010 at 10:37 AM, PtitBleu <ptit_bleu at yahoo.fr> wrote:
>
> Hello,
>
> I discovered R two years ago and thanks to the R-community I managed to
> write some scripts to analyze my data stored in mysql databases.
> The only problem is that I am the only one using R in the lab. Colleagues
> mainly use Matlab (but not with mysql, only with text files) but regularly
> come to me to get data treated with R-scripts !!!.
>
> To allow the use of my scripts by other people, my boss asked me to make
> executables (.exe) with my scripts or to pay someone (I'm only end-user and
> not a computer scientist) to translate them into matlab langage (and then
> into .exe) or into C++ meaning abandoning R. And I don't want to.
>
> So is it possible to make executables from R scripts ? It seems it is not
> the case but I hope I missed a way to do it.

If your boss isn't dumb enough to insist on binary executables and
just wants some processing programs that can be delivered to the
matlab guys, then you can create "standalone" R scripts - easy with
"littler":

http://dirk.eddelbuettel.com/code/littler.html

You still need R and any required packages installed on the target
machine, but the "user" doesn't need to know R at all. Just run the R
script, and feed it whatever arguments it needs. You can even build
simple GUIs with the tcltk package (or complex GUIs with RGtk?).

Your colleagues can also look at the R scripts and modify them if they
don't quite do what they want them to do. Then these guys and girls
will be learning R. You can't do that with a binary .exe file. Big
wins all round.

Barry


-- 
blog: http://geospaced.blogspot.com/
web: http://www.maths.lancs.ac.uk/~rowlings
web: http://www.rowlingson.com/
twitter: http://twitter.com/geospacedman
pics: http://www.flickr.com/photos/spacedman