Skip to content

Compilation of R-0.62.1 on DEC Alpha

5 messages · Douglas Bates, Peter Dalgaard

Original
Douglas Bates
# 
I have compiled R-0.62.1 on a DEC Alpha system that is running Digital
Unix 4.0.  I had to change some settings in Makeconf from those
generated by configure.  I changed the LDCMD to gcc and the LIBS to
include -lFutil and -lfor.  (I wonder if they realized how close their
name for the Fortran utilities is to the word "futile".)  I also had
to force it to use the GNU version of the install program as the DEC
version is not as fully featured.

I get a compilation but it creates a floating point exception as soon
as you try to run it.
 $ /unsup/R-0.62.1/bin/R

 R : Copyright 1998, The R Development Core Team
 Version 0.62.1  (June 15, 1998)

 R is free software and comes with ABSOLUTELY NO WARRANTY.
 You are welcome to redistribute it under certain conditions.
 Type    "?license" or "?licence" for distribution details.

 R is a collaborative project with many contributors.
 Type    "?contributors" for a list.

 Type    "demo()" for some demos, "help()" for on-line help, or
	 "help.start()" for a HTML browser interface to help.

 Floating exception(coredump)
 $ 

Has anyone been successful building R-0.62.1 (and running it) on this
platform?

Builds on Solaris/SPARC and Solaris/Intel and Debian GNU/Linux Intel
went smoothly.  I may try to see if I can get a person running Debian
GNU/Linux on an Alpha to try a build there.
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Peter Dalgaard
# 
Douglas Bates <bates@stat.wisc.edu> writes:
..
There are two things to this: What on earth is causing a floating
exception at that stage (it might actually be normal)? and why is it
dumping core? The latter sounds as if you might need to put in some
option to force IEEE behaviour.
Douglas Bates
# 
Peter Dalgaard BSA <p.dalgaard@biostat.ku.dk> writes:
Is there anything done before the first prompt to initialize the IEEE
NaN, Inf and -Inf?

I do have the -ieee_with_no_inexact flag being passed to gcc.  The end
of the configure output shows

 R is now configured for alpha-dec-osf4.0

   Source directory:         .
   Installation directory:   /usr/local
   C compiler:               gcc -ieee_with_inexact -g -O2
   FORTRAN compiler:         f77 
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Douglas Bates
# 
I see I contradicted myself.

Douglas Bates <bates@cs.wisc.edu> writes:
xxx
...
Strike the "_no"
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Peter Dalgaard
# 
Douglas Bates <bates@stat.wisc.edu> writes:
Never mind. The strange thing is that it isn't behaving according to
IEEE specs, because according to those, default behaviour on fp
exceptions is to set a flag and continue. 

I bet the exception is happening in InitArithmetic() in
src/main/arithmetic.c which does some division by zero *before*
setting a signal handler for SIGFPE. Tried running it under the
debugger? (or debugging the core file?)