[R] Problems loading Lapack library

Prof Brian Ripley ripley at stats.ox.ac.uk
Wed Sep 8 21:07:02 CEST 2004


Haven't you sent this before?

d_sign is part of the Fortran runtime support.  Since you are running 
sunperf, you should not have been building libRlapack at all.

Please don't try to outsmart R, but let it decide what configuration to
try.  Once that works, feel free to try to improve it.  My guess is that
your use of sunperf (and *PLEASE* read what the R-admin manual says about
LAPACK_LIBS) is the problem, probably because such an ancient sunperf is
broken (I never managed to get it to work prior to SunPro 6).


On Wed, 8 Sep 2004, Ed Hughes wrote:

> I have just installed R 1.9.1 on an old Sun Sparc
> with the following specs:
> version
> Machine hardware:   sun4u
> OS version:         5.6
> Processor type:     sparc
> Hardware:           SUNW,Ultra-4
> The following components are installed on your system:
> Sun Visual WorkShop C++ 3.0
>          Sun WorkShop Compiler C 4.2 
>          Sun WorkShop Compiler C++ 4.2 
>          Sun WorkShop Tools.h++ 7.0 
>          Sun WorkShop Tools.h++ 6.0.4 
>          Sun WorkShop Visual 2.0 
>          Sun WorkShop IPE 4.0 
>          Sun WorkShop CodeManager 2.0 
>          Sun WorkShop Distributed Make 2.0 
>          Sun WorkShop FileMerge 3.0 
>          Sun WorkShop FreezePoint 2.0 
>          Sun WorkShop Maketool 2.0 
>          Sun WorkShop VersionTool 2.0 
>          Sun WorkShop Dbx 4.0 
>          Sun WorkShop Performance Analyzer 4.0 
>          Sun WorkShop LoopTool 2.1 
>          Sun WorkShop LockLint 2.1 
>          Sun WorkShop Thread Analyzer 1.2 
>          Sun WorkShop XEmacs 20.00 
> Sun WorkShop Professional C 3.0
>          Sun WorkShop Compiler C 4.2 
>          Sun WorkShop IPE 4.0 
>          Sun WorkShop Dbx 4.0 
>          Sun WorkShop Performance Analyzer 4.0 
>          Sun WorkShop IPE 4.0 
> Sun WorkShop Compiler FORTRAN 77 4.2 
> Sun WorkShop Compiler Fortran 90 1.2 
> Sun Performance Library 1.2 
> 
> Here are the lines I changed in config.site:
> 
> R_PAPERSIZE=letter
> CC="cc -xarch=v8plusa"
> CFLAGS="-xO3 -dalign"
> F77="f90 -xarch=v8plusa"
> FFLAGS="-xO3 -dalign"
> MAIN_LDFLAGS="-xarch=v8plusa"
> LDFLAGS="-L/opt/SUNWspro/lib -L/usr/local/lib"
> CXX="CC -xarch=v8plusa"
> CXXFLAGS="-xO3 -dalign"
> BLAS_LIBS="-xlic_lib=sunperf"
> LAPACK_LIBS="-xlic_lib=sunperf"
> 
> The configure and make steps finished apparently normally,
> but "make check" failed.  I found that R would start, and
> do elementary things, but anything involving linear algebra
> failed:  for example, I tried the "lm-tests.R" file in the
> tests directory, with the following result:
> 
>   . . . . . . 
> 
> Type 'q()' to quit R.
> 
>  > source("tests/lm-tests.R")
> Error in La.chol2inv(x, size) : lapack routines cannot be loaded
> In addition: Warning message: 
> unable to load shared library "/disk4/home4/ehug/newSoft/R-1.9.1/modules/lapack.so":
>    ld.so.1: /disk4/home4/ehug/newSoft/R-1.9.1/bin/R.bin: fatal: relocation 
>    error: file /disk4/home4/ehug/newSoft/R-1.9.1/bin/libRlapack.so: symbol d_sign: 
>    referenced symbol not found 
> 
> I don't think "d_sign" is a symbol in Lapack.  Similar things happen 
> with any attempt to use functions that involve Lapack.  
> Can anyone give me some clues about what's happening?  Any help
> would be appreciated.
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
> 
> 

-- 
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




More information about the R-help mailing list