I wrote about my problems building R-0.63.3 on a Digital Alpha running OSF
V4.0D. Thanks to the suggestion Jonathan.Yuen at tvs.slu.se, I tried using the
DEC CC compiler instead of gcc, and it worked just fine. (Seeing that I
needed gmake to build, it never occured to me to try DEC CC. It seems that
you cannot combine DEC F77 with gcc; I do not have g77 to try.)
Below is a set of instructions which happily builds R on my Alphas.
Paul Szabo - System Manager // School of Mathematics and Statistics
psz at maths.usyd.edu.au // University of Sydney, NSW 2006, Australia
---
#!/bin/ksh -
set -x
V=R-0.63.3
gzcat $V.tgz | tar xf -
cd $V
: 'Versions (for the record):'
: '% uname -a'
: 'OSF1 turin.maths.usyd.edu.au V4.0 878 alpha'
: '% sizer -v '
: 'Digital UNIX V4.0D (Rev. 878); Tue Dec 15 07:43:31 EST 1998 '
:
: 'Set values in config.site:'
: 'use DEC C and F77 compilers with appropriate options, GNU make.'
: '(Cannot use gcc with DEC F77; we do not have g77)'
: '(Need GNU make, DEC make fails in several respects)'
cp -ip config.site config.site.bak <&- >&- 2>&-
cat <<EOF >> config.site
CC=/usr/bin/cc
CFLAGS='-g3 -O'
FC=/usr/bin/f77
FFLAGS='-g3 -O'
DLLFLAGS=-nofor_main
MAKE=gmake
EOF
./configure
:
: 'Fix some bugs:'
:
: 'Will get compiler warnings:'
: ' In this statement, the expression "j=++j%k" modifies the
variable "j" more'
: ' than once without an intervening sequence point. This behavior is
undefined.'
: 'in src/main/subassign.c, line 928 and src/main/subset.c, line 376.'
: 'What to do about them (do they matter)?'
:
: 'Fix a line in plot.c (look in PLATFORMS file)'
perl -i.bak -pe 's/(if \(FINITE.asp. .. asp > 0)(\) \{)$/$1 \&\&
asp != NA_REAL$2/' src/main/plot.c
:
: 'Reportedly there are bugs in src/graphics/graphics.c : add the line'
: ' extern double pow(double, double); '
: 'to Mathlib.h ??'
:
: 'No dependency: only remake when needed'
perl -i.bak -pe 's/::$/:/' src/unix/Makefile
:
: 'Get rid of pesky "ARGUMENT '"''"'
__ignored__" message'
: '(This bug is in DEC /bin/sh, not in R: use ksh instead)'
perl -i.bak -pe 's:/bin/sh:/bin/ksh:' src/scripts/R.sh
:
: '(I like to make my installations simple... can R adopt this scheme?)'
: 'Install man pages in RHOME, with symlinks from /usr/local/man/man1'
perl -i.bak -pe 's/mandir/rhome/ unless m/cd/; s/^\t.(.cd ..mandir)/\t-$1/;
print "\t for f in R.1 Rdconv.1; do rm -f \$\$f; \$(LN_S)
\$(rhome)/man1/\$\$f \$\$f; done; \\\n" if m/in Rd2txt/' doc/Makefile
:
: 'Install with symlinks from /usr/local/bin'
perl -i.bak -pe 's:^.*cat .* ..bindir./R$:\techo Not doing /usr/local/bin/R
yet:; s: ..bindir./R : : if m/chmod/; s/^.*(for f in )(Rd2txt )/\t-$1R $2/;
s:^(.*)..INSTALL. (.*) (..bindir.):$1\$(LN_S) \$(rhome)/$2 $3:;' Makefile
gmake
: 'To install the lot, need to do as root:
mkdir /usr/local/lib/R
chown psz /usr/local/lib/R
On the machine with /usr/local do (as psz):
gmake install
and finally again as root:
chown -R bin.bin /usr/local/lib/R
chmod -R go=rX /usr/local/lib/R
ln -s /usr/local/lib/R/man1/R.1 /usr/local/man/man1
ln -s /usr/local/lib/R/man1/Rdconv.1 /usr/local/man/man1
ln -s Rdconv.1 /usr/local/man/man1/Rd2txt.1
ln -s Rdconv.1 /usr/local/man/man1/Rdindex.1
ln -s Rdconv.1 /usr/local/man/man1/Sd2Rd.1
ln -s /usr/local/lib/R/bin/R /usr/local/bin
ln -s /usr/local/lib/R/bin/Rd2txt /usr/local/bin
ln -s /usr/local/lib/R/bin/Rdconv /usr/local/bin
ln -s /usr/local/lib/R/bin/Rdindex /usr/local/bin
ln -s /usr/local/lib/R/bin/Sd2Rd /usr/local/bin
'
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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 at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
