On Fri, 30 Jan 2004, Rajarshi Guha wrote:
> Hi,
> I downloaded the subselect package from CRAN and installed it in the
> system wide R library path. During installation the package compiled the
> fortran sources with no errors.
No _reported_ errors.
> However, when loading the library gives me an error:
>
> > library(subselect)
> Error in dyn.load(x, as.logical(local), as.logical(now)) :
> unable to load shared library
> "/usr/lib/R/library/subselect/libs/subselect.so":
> /usr/lib/R/library/subselect/libs/subselect.so: undefined symbol:
> f_iob
> Error in library(subselect) : .First.lib failed
>
> I would had expected that if it could'nt find a symbol, that would have
> shown up during compilation. But there were no compile time errors. I
All you did was create a shared object. It does not need to be complete.
> also tried R CMD check on the package directory and compilation occurs
> with no errors but I get the errors below:
>
> * checking S3 generic/method consistency ... WARNING
> Error in .tryQuietly({ : Error in library(package, lib.loc = lib.loc,
> character.only = TRUE, verbose = FALSE) :
> .First.lib failed
> Execution halted
>
> (And more such errors)
>
> I have contacted the author of the package and he sent me few postings
> regarding the f_iob symbol and it appears to be a problem resulting from
> a LAPACK compiled with the Intel fortran compiler (ifc). However in this
> case, ifc is not installed on my system and during compilation it uses
> the Rlapack library.
The problem is to do with a Fortran run-time, and AFAICS not that in gcc
3.3.2.
> I installed R from rpm's and am running on Fedora Core 1.
...
> It appears I am missing some library that contains f_iob (though why the
> error did'nt show up at compile time I dont know).
As I said above ....
> Has anybody faced this problem or have any suggestions as to how to fix
> it?
Compile R from the sources: works for me on FC1. My subselect.so does not
have a reference to f_iob.
Or look with nm to see which of the components which made up subselect.so
contains the reference, and check its compilation line.
--
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