R-beta: R-0.62.1 under Digital Unix

Douglas Bates (bates@stat.wisc.edu)
30 Jun 1998 09:50:05 -0500


To: "Mark A. Beaumont" <M.Beaumont@ucl.ac.uk>
Subject: R-beta: R-0.62.1 under Digital Unix
From: Douglas Bates <bates@stat.wisc.edu>
Date: 30 Jun 1998 09:50:05 -0500
In-Reply-To: "Mark A. Beaumont"'s message of Tue, 30 Jun 1998 15:21:24 +0100

I have tried making R-0.62.1 on a DEC Alpha with Digital Unix V4.0A.
The immediate problem you are having has to do with the install
program.  The GNU version of install allows you to specify the -d
option to create a directory path.  The standard Digital Unix version
of install does not have that option.  That is what causes the
complaint about -d.

I got around this by using the GNU version of install under Digital
Unix.  If you don't want to do that you may be able replace the lines
in the install target of $RHOME/Makefile.2nd that are of the form
        $(INSTALL) -d $(bindir)
        ...
by
        mkdir -p $(bindir)
        ...
Only do this for those calls to $(INSTALL) that use the -d option.

Even when you finish the compilation successfully there may be
problems with the handling of IEEE floating point exceptions under
Digital Unix.  It appears that these are not handled according to the
IEEE specifications and some code that is run in the initialization of
R causes it to abort on a floating point exception.  This happens
before the first prompt.

We were able to get R to run on a DEC Alpha running Linux but not on a
DEC Alpha running Digital Unix.

"Mark A. Beaumont" <M.Beaumont@ucl.ac.uk> writes:

> I am trying to make the R-0.62.1 distribution on a DEC alpha running
> Digital Unix 3.2d.

> I have run into a few problems.....
> 
> (1) the Makefile in src/graphics need to have INCLUDES= -I../include
> not -I ../include
> 
> (2) I have the standard DEC fortran 90 compiler which needs to link R.binary
> with a -nofor_main flag. The appropriate slot in config.site appears to be
> ignored, which means Makeconf itself has to be edited. However in Makeconf
> LDFLAGS appears to be ignored too, so one just has to make do with
> LDCMD=f90 -nofor_main
> 
> (3) The last problem has completely stumped me. The make goes OK until
> .
> .
> .
> R binary installed
> make[3]: Leaving directory `/usr/local/R-0.62.1/src/main'
> make[3]: Entering directory `/usr/local/R-0.62.1/src/library'
> make[4]: Entering directory `/usr/local/R-0.62.1/src/library/base'
> Building R base package
> make[4]: Leaving directory `/usr/local/R-0.62.1/src/library/base'
> make[4]: Entering directory `/usr/local/R-0.62.1/src/library/profile'
> building system startup profile
> make[4]: Leaving directory `/usr/local/R-0.62.1/src/library/profile'
> make[3]: Leaving directory `/usr/local/R-0.62.1/src/library'
> make[3]: Entering directory `/usr/local/R-0.62.1/src/library'
> Building application packages..
> install: cannot open -d
> install: cannot open -d
> install: cannot open -d
> make[3]: *** [libs] Error 3
> make[3]: Leaving directory `/usr/local/R-0.62.1/src/library'
> make[2]: *** [R] Error 2
> make[2]: Leaving directory `/usr/local/R-0.62.1/src'
> make[1]: *** [R] Error 2
> make[1]: Leaving directory `/usr/local/R-0.62.1'
> make: *** [all] Error 2

-- 
Douglas Bates                            bates@stat.wisc.edu
Statistics Department                    608/262-2598
University of Wisconsin - Madison        http://www.stat.wisc.edu/~bates/
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._