On my IRIX system (SGI compilers, gcc not installed) ./configure runs fine,
but there is a problem with make (output below) that can be solved by
changing one line in src/modules/lapack/Makefile. If this line is changed,
make runs without problems. 'make check' still has a problem
(sh[14]: /usr/sbin/perl: arg list too long), but I assume this is
"cosmetic".
Manfred
----------- The patch:
msm4:/usr/local/src/R-1.3.1/src/modules/lapack% diff -u Makefile.o Makefile
--- Makefile.o Sat Dec 8 19:23:32 2001
+++ Makefile Sat Dec 8 19:27:26 2001
@@ -19,7 +19,7 @@
SOURCES_BLAS = blas2.f cmplxblas.f
DEPENDS = $(SOURCES_C:.c=.d)
-OBJECTS = $(SOURCES_C:.c=.lo) $(SOURCES_F:.f=.lo) \
+OBJECTS = $(SOURCES_C:.c=.lo) $(SOURCES_F:.f=.lo)
# $(SOURCES_BLAS:.f=.lo)
HEADERS = Lapack.h
------------ make stops with the following error:
"cmplx.f", line 19283: warning(2290): actual argument is incompatible
with
dummy argument
CALL ZLARFB( SIDE, TRANS, 'Forward', 'Columnwise',
MI, NI,
^
don't know how to make # (bu42).
*** Error code 1 (bu21)
*** Error code 1 (bu21)
*** Error code 1 (bu21)
*** Error code 1 (bu21)
------------ configure:
R is now configured for mips-sgi-irix6.5
Source directory: .
Installation directory: /usr/local
C compiler: cc -OPT:IEEE_NaN_inf=ON -g
C++ compiler: CC -OPT:IEEE_NaN_inf=ON -g
FORTRAN compiler: f77 -OPT:IEEE_NaN_inf=ON -g
X11 support: yes
Gnome support: no
Tcl/Tk support: no
R profiling support: yes
R as a shared library: no
configure: warning: you cannot build DVI versions of the R manuals
configure: warning: you cannot build info versions of the R manuals
configure: warning: you cannot build PDF versions of the R manuals
------------- make check problem:
msm4:/usr/local/src/R-1.3.1% make check
collecting examples for package `base' ...
>>> Building/Updating help pages for package `base'
Formats: text example
sh[14]: /usr/sbin/perl: arg list too long
*** Error code 1 (bu21)
*** Error code 1 (bu21)
*** Error code 1 (bu21)
*** Error code 1 (bu21)
*** Error code 1 (bu21)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: maba.vcf
Type: text/x-vcard
Size: 356 bytes
Desc: Card for Manfred W. Baumstark
Url :
https://stat.ethz.ch/pipermail/r-help/attachments/20011208/46c037f0/maba.vcf
On Sat, 8 Dec 2001, Manfred W. Baumstark wrote:> On my IRIX system (SGI compilers, gcc not installed) ./configure runs fine, > but there is a problem with make (output below) that can be solved by > changing one line in src/modules/lapack/Makefile. If this line is changed, > make runs without problems. 'make check' still has a problem > (sh[14]: /usr/sbin/perl: arg list too long), but I assume this is > "cosmetic". > > Manfred > > ----------- The patch: > msm4:/usr/local/src/R-1.3.1/src/modules/lapack% diff -u Makefile.o Makefile > --- Makefile.o Sat Dec 8 19:23:32 2001 > +++ Makefile Sat Dec 8 19:27:26 2001 > @@ -19,7 +19,7 @@ > SOURCES_BLAS = blas2.f cmplxblas.f > > DEPENDS = $(SOURCES_C:.c=.d) > -OBJECTS = $(SOURCES_C:.c=.lo) $(SOURCES_F:.f=.lo) \ > +OBJECTS = $(SOURCES_C:.c=.lo) $(SOURCES_F:.f=.lo) > # $(SOURCES_BLAS:.f=.lo) > HEADERS = Lapack.hWell, that Makefile is not part of the sources! Makefile.in in the near-release 1.4.0 ha OBJECTS = $(SOURCES_C:.c=.lo) $(SOURCES_F:.f=.lo) \ @USE_EXTERNAL_BLAS_FALSE@ $(SOURCES_BLAS:.f=.lo) and on my system Makefile has OBJECTS = $(SOURCES_C:.c=.lo) $(SOURCES_F:.f=.lo) \ # $(SOURCES_BLAS:.f=.lo) as does 1.3.1. What make is this? As R-admin.texi says To compile @R{}, you will most likely find it easiest to use GNU @command{make}. On Solaris 2.6/7/8 in particular, you need a version of GNU @command{make} different from 3.77; 3.79 works fine, as does the Sun @command{make}. I suspect your make is broken.> ------------ make stops with the following error: > "cmplx.f", line 19283: warning(2290): actual argument is incompatible with > dummy argument > CALL ZLARFB( SIDE, TRANS, 'Forward', 'Columnwise', MI, NI, > CALL ZLARFB( SIDE, TRANS, 'Forward', 'Columnwise', MI, NI,That's standard LAPACK 3 code, so the bug report needs to go to LAPACK. I think your compiler is incorrect, but 'F', 'C' might work. BDR -- 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 272860 (secr) Oxford OX1 3TG, UK Fax: +44 1865 272595 -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- 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 _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
...> Do you have means to find out whether this is a known bug in SGI make? >Yes, I will do this.> R is trying not to depend on the GNU tools if possible ...This was exactly my motivation to post. I was aware of the tread giving several advices how to build R under IRIX. Anyway I have shown were the problem is, and gave a workaround how you can build R on an SGI without gnu tools. It worked for me on my own O2 and on an Origin in the computing center were I`m not root (but can use gigabytes of RAM :-)). I did also look more deeply into the failure of 'make check': If I run the test individually in directory tests/ 'make test-Specific' and most others I tried are ok, 'make test-Examples' gives the error: sh[14]: /usr/sbin/perl: arg list too long *** Error code 1 (bu21) Manfred -------------- next part -------------- A non-text attachment was scrubbed... Name: maba.vcf Type: text/x-vcard Size: 356 bytes Desc: Card for Manfred W. Baumstark Url : https://stat.ethz.ch/pipermail/r-help/attachments/20011209/cdd7a041/maba.vcf