[R] R-help Digest, Vol 79, Issue 9

Ista Zahn istazahn at gmail.com
Wed Sep 9 14:05:44 CEST 2009


> From: "Carlos Alzola" <calzola at cox.net>
> To: <r-help at r-project.org>
> Date: Tue, 8 Sep 2009 23:19:11 -0400
> Subject: [R] SAS vs. R in web application
> Good evening,
>
> I have been asked to investigate the pros and cons of using SAS vs. R in a web application. Either SAS or R would be the engine used to make some very simple calculations and to produce graphs, preferably in png format.
>
> The advantages of R are pretty obvious as there would be no licensing issues. The only drawback I can see is that when calling it in batch (using R CMD BATCH), a DOS window appears. Thus I have some basic questions:
>
> a) Is it possible to have R operate in the background without the DOS window appear? How?
> b) Is it correct that there will be no licensing issues?
> c) What would be an efficient way to run it? I am thinking of having R running in the client's local machine and upload the results to a central server.

I don't think you want to do this. Much better to have the R process
run on the server. My advice is to use the apache2 web server on a
unix-alike system, along with the RApache module
(http://biostat.mc.vanderbilt.edu/rapache/). There are other ways to
use  R in a web page, but in my experience RApache is easier to set
up/use and is more  robust. My website at http://yourpsyche.org uses
RAapche, and I've been very pleased with it.

>
> If using SAS, would the model described in c) above be the best way to design it, or would it be better to upload the raw data to the server and have SAS perform the calculations there. Would this option require a multi-user SAS license? (I know, I should check with SAS Institute, but I thought I'd ask anyway. Someone in the list may have done something similar).
>

No experience with SAS, but again I think you're better off  uploading
the data and performing calculations server-side, regardless of the
program you use for the calculations.

> Thanks in advance for any suggestions.
>
> Carlos Alzola




More information about the R-help mailing list