[R] rgdal installation with two versions of GDAL
Prof Brian Ripley
ripley at stats.ox.ac.uk
Wed Apr 22 17:43:56 CEST 2015
On 22/04/2015 05:43, Rob Skelly wrote:
> Hello,
>
> I am using GDAL 2.0 for most of my work, but rgdal depends on GDAL < 2. I
> have built and installed GDAL 1.11.2 in /opt/gdal-1.11.2, and rgdal
> compiles and installs into R, using the following command:
>
> sudo R CMD INSTALL
> --configure-args="--with-gdal-config=/opt/gdal.1.11.2/bin/gdal-config"
> rgdal-0.9-2.tar.gz
>
> However when R attempts to load rgdal at the end of the installation, it
> fails with the error,
>
> Error in dyn.load(file, DLLpath = DLLpath, ...) :
> unable to load shared object
> '/usr/local/lib/R/site-library/rgdal/libs/rgdal.so':
> /usr/local/lib/R/site-library/rgdal/libs/rgdal.so: undefined symbol:
> _ZN10OGRPolygon7addRingEP13OGRLinearRing
>
> The addRing method on OGRPolygon seems to be a relic of GDAL 1.11.2 and no
> longer exists in 2.0, so R is loading libgdal from /usr/local/lib, not
> /opt/gdal-1.11.2/lib. I have confirmed this by temporarily moving the
> 1.11.2 libs into /usr/local, where it works fine.
You can *really* confirm where things are found via R CMD ldd ... see
the manual.
> So, the question is, how to I convince R to use the new library search
> path? I'm on xubuntu, and R is installed using apt.
You either use the ld options when building rgdal.so (e.g. for your OS
-Wl,-rpath=/opt/gdal-1.11.2/lib) or you link statically. The latter is
easier and safer ... just build a static GDAL.
Many projects set -R/-rpath flags in their config scripts: it is
something you could suggest to the GDAL maintainers (and finding them is
part of libtool which GDAL uses).
But (see the posting guide) the generic question belonged on R-devel and
questions about rgdal on R-sig-geo.
> Thanks,
> Rob
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
--
Brian D. Ripley, ripley at stats.ox.ac.uk
Emeritus Professor of Applied Statistics, University of Oxford
1 South Parks Road, Oxford OX1 3TG, UK
More information about the R-help
mailing list