Installing R in Fedora Core 4 - R-help

Fri, Aug 19, 2005 8:07 AM #

-----Original Message-----

On Fri, 8/19, Gavin Simpson wrote:

Peter Dalgaard has noted, on the R-Devel list (sorry I can't provide the
link to the mail - the link from the R site to the mail archives wasn't
working when I tried), that there are problems with the R rpm from
Fedora Extras, including a strange printing bug. I believe Peter now
thinks this is a bug in R (seem to have deleted that post - doh) exposed
by the compilation flags used by the maintainer of the Fedora Extras
rpm.

If you want and rpm to install, then Martyn Plummer provides R binaries
for Red Hat / Fedora systems that are available from CRAN e.g: for FC4
http://www.stats.bris.ac.uk/R/bin/linux/redhat/fc4/

Martyn has also made these available via a yum-compatible repository, so
the benefits you note of auto-notification of updates etc. apply here as
well.

HTH

G
-----End Original Message-----

The last time I tried to use the Martyn Plummer RPMs they were compiled
without enabling shared libraries and I ended up compiling R myself so
that I could use JGR. Messages describing the problem with the version
actually on the Extras CD were my other unstated reason for describing
yum instead of downloading the CD. I haven't heard there is a problem
with Fedora's new RPMs but that doesn't prove that there aren't any. I
fully agree with Martyn Plummer's readme notice that describes Fedora
Core as bleeding edge technology not to be trusted for production use.
Fedora Core describes itself that way.

Gavin Simpson

Fri, Aug 19, 2005 8:55 AM #

On Fri, 2005-08-19 at 11:07 -0400, White, Charles E WRAIR-Wash DC wrote:

Do you mean the shared library libR.so? If so, the README in the FC3
section indicates that these rpms now include the functionality you
require. The absence of a README in the FC4 section means it is not
clear from there if these rpms are also compiled with the shared
library.

Extras continues to use the compiler flags that cause the problems
previously described (I haven't confirmed that this is still the case
mind you).

Although FC4 is bleeding edge, I've had very few problems with in on my
new laptop or my home desktop - after initial gcc v4.0.0 and gfortran
teething troubles that is.

HTH

G

%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%
Gavin Simpson                     [T] +44 (0)20 7679 5522
ENSIS Research Fellow             [F] +44 (0)20 7679 7565
ENSIS Ltd. & ECRC                 [E] gavin.simpsonATNOSPAMucl.ac.uk
UCL Department of Geography       [W] http://www.ucl.ac.uk/~ucfagls/cv/
26 Bedford Way                    [W] http://www.ucl.ac.uk/~ucfagls/
London.  WC1H 0AP.
%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%~%

Brian Ripley

Fri, Aug 19, 2005 10:19 AM #

On Fri, 19 Aug 2005, Gavin Simpson wrote:

They almost certainly are, for 2.1.1, since they are compiled from the 
same spec file.

It is confirmed not the case for R-devel, but is for 2.1.1 which is 
what the RPMs are for.

Although gcc 4.0.1 has helped, there are dozens of ongoing problems which 
seems squarely laid at the door of gfortran, as well as an appreciable 
performance loss.  On AMD64 there are subtle incompatibility issues 
between gfortran and everything else (g77, gcc3 and gcc4) that make mixing 
C and Fortran code tricky at best.

Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595