[R] Potential for R to conflict with other softwares

[Peter Dalgaard]

Duncan Murdoch <murdoch at stats.uwo.ca> writes:

> On 11/3/2005 9:11 AM, Soukup, Mat wrote:
> > Hi.
> > 
> > After some time, my collegues at the Food and Drug Adminstration have
> > finally acknowledged R as a powerful statistical computing environment.
> > However, in order to comply with the Office of Information and Technology
> > standards there are a couple of questions about whether R could interfere
> > with other software. As I'm more of a driver of the R software and not a
> > mechanic, I was hoping for the insight of the many great useRs. Below is a
> > list of 5 proposed questions to which I value any comment.
> > 
> > Thank you for your time,
> > 
> > -Mat
> > 
> > 
> 
> These answers are about the Windows version only, but from the 
> questions, I think that's what you were looking for.  They apply to all 
> versions since 1.6.x at least (though the earlier ones would have put 
> fewer entries into the registry, they put them in the same places).
> 
> > 1. Does R have high resolution graphics?
> 
> Yes, but I don't think I get the point of this question.  How would that 
> interfere with other software?

Device drivers! We have seen cases where the FPU control word had to
be reset (DM will know this more precisely than me, I think). We do
have code in place to catch that particular issue though. 

If you want to be really sure, I believe that batch runs using the
postscript() or pdf() drivers would be immune to such issues (right?).

(BTW, does the FDA trust SAS for Windows? I've seen a weird thing or
two happening to the display there...)

> > 
> > 2. Does R have .dll files, or other executables which are not located in the
> > R software directory tree?
> 
> No, it installs everything below R_HOME.

We do rely on MCVCRT.DLL and some other system DLLs though, and I
think this can get modified by other software.

> > 
> > 3. Does R modify the Windows registry in a non-obvious way, i.e. other than
> > defining itself and what extensions to associate with R, and what are those
> > extensions?
> 
> I think all of its modifications would count as obvious.  They are 
> mainly below HKLM/Software/R-core or HKCU/Software/R-core (where the 
> file locations are recorded); additionally file associations are set up 
> for .Rdata files (which are called RWorkspace files there), and an 
> uninstall entry is made.
> > 
> > 4. Does R add macros to any part of MS Office?
> 
> No.
> > 
> > 5. Can you anticipate any other way in which installing and using R could
> > disrupt the operation of another software?
> 
> No, not really.  Maybe users will become addicted to it?  ;-)
> 
> Duncan Murdoch
> >  
> > 
> > ***********************************************************************
> > Mat Soukup, Ph.D.
> > Food and Drug Administration
> > 10903 New Hampshire Ave. 
> > BLDG 22 RM 5329
> > Silver Spring, MD 20993-0002
> > Phone: 301.796.1005
> > ***********************************************************************

(& I think Mat is owed a big "Thank you" for his efforts).

-- 
   O__  ---- Peter Dalgaard             Øster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark          Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)                  FAX: (+45) 35327907