Skip to content

rgdal on Mac OS X 3.9

3 messages · Jonathan Boyd Thayn, Roger Bivand, Scott Mitchell

Original
Roger Bivand
#
On Thu, 7 Sep 2006, Jonathan Boyd Thayn wrote:

            

      
      
      
    

        
Could an OSX user please reply with help on how to find GDAL? Could 
someone also comment on the viability of this OS version?

We have had user reports of success building rgdal from source on OSX, but 
I don't think the specific OS version(s) used are known - could OSX users 
please reply to this thread so that the list archives have a body of 
experience about this, please? The same applies to the way(s) in which 
GDAL and PROJ.4 were installed.

Roger

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Scott Mitchell
    

    

      
      #
    

  


  

      
On 8-Sep-06, at 03:12, Roger Bivand wrote:

        

            

      
      
      
    

        
Depends on how you installed GDAL.  There are Frameworks versions  
available, there's the fink package manager, and there's the option  
to compile from source yourself.  Probably more options are available  
too.  How did you do it, Jonathan?

If you compiled and installed yourself, it's probably in the default  
install location of /usr/local/bin/gdal-config, in which case you  
probably just need to add /usr/local/bin to your PATH in the shell  
you are using.  For example, at the shell prompt, before starting R,  
type
export PATH=$PATH:/usr/local/bin
to add that directory to the end of your path specification.
As long as the install can find the gdal-config program, that program  
should be able to tell the build system where to find the libraries  
and headers.

If you installed GDAL another way, you still need to figure out where  
it has been installed - if it is a framework install, it's probably  
in a subdirectory under /Library/Frameworks.  If you can provide more  
details of your install we can probably figure it out together.

            

      
      
      
    

        
If I understand correctly, the binary downloads work only for OS  
10.4.4 and up, but it should be possible to compile R from source on  
10.3.9 -- again, that's IF I understand correctly.

            

      
      
      
    

        
My answer to this got more complicated the more I delved into it.  So  
I've decided to document the full "transcript" here, in case it helps  
others (or points out mistakes in my interpretations):

The machine I happen to be using at the moment appears to have R2.3.0  
installed from the official Mac binary download site, with various  
geo-related libraries added afterwards, including rgdal 0.4-3.  I  
haven't done updates on this machine for quite a while, and don't  
know precisely what OS version was used when I did the installs, but  
it would have been 10.4.something.  It had GDAL 1.3.0 installed from  
source into /usr/local, PROJ 4.4.x also from source in /usr/local.   
That USED to work (so I've left the above text to document a  
different combination of versions that used to work), but now I have  
GDAL 1.3.1, and in my attempt to load the rgdal library I run into  
trouble because it can't find a symbol.  So I'm using this  
opportunity to update my R.

To try to correspond to the original poster's situation a little more  
closely, I'm compiling R from source instead of using the binaries.   
Note, however, that I AM using  OS 10.4.7 on a G5 machine with XCode  
2.2.1 developer tools.  The first FORTRAN compiler in my path is the  
GNU g77, version 3.4.3, provided by Fink (fink.sf.net).

./configure without any arguments provided the following summary:

R is now configured for powerpc-apple-darwin8.7.0

   Source directory:          .
   Installation directory:    /Library/Frameworks

   C compiler:                gcc  -g -O2 -std=gnu99
   Fortran 77 compiler:       gfortran  -g -O2

   C++ compiler:              g++  -g -O2
   Fortran 90/95 compiler:    gfortran -g -O2

   Interfaces supported:      X11, aqua, tcltk
   External libraries:        readline, BLAS(generic), LAPACK(in blas)
   Additional capabilities:   PNG, JPEG, iconv, MBCS, NLS
   Options enabled:           framework, R profiling

   Recommended packages:      yes

Running make after that worked fine, but then since it upgraded the  
existing R installation, my old extra packages are still in there.  I  
thought that I could make this work by reinstalling the packages that  
use GDAL, and I came close, but ran into a new problem.  If I  
interpret correctly, when I first built R, it was for PPC only.  Now  
R tries to build universal binaries that can work on either PPC or  
Intel chips, and I do have a build system that should handle this,  
BUT some of my R packages are still leftover from the first build,  
and I'm getting errors when it tries to build rgdal.

So, to fix that, I did:

sudo rm -rf /Library/Frameworks/R.framework
cd ~/src/R-2.3.1 (wherever your R source is, I had already done the  
"make" step)
sudo make install

Then, from an R session started by root (you could also use R CMD  
INSTALL):

install.packages(c("rgdal", "sp"), dependencies=TRUE, type="source")

and now the rgdal library successfully compiles and loads.

To put back some of the other packages I used to have installed, I  
then did:

install.packages("ctv")
library("ctv")
install.views("Spatial")

That mostly worked, but complained about a couple of dependencies not  
working, which is because some packages are only available as source  
packages for Macs.  The packages with errors were rgdal and VR.   
Since I've already installed rgdal, I ignore that one, but for VR:

install.packages("VR", dependencies=TRUE, type="source")

And all looks good.

Hopefully all that might help someone trying to do something similar  
in the future, even if it does diverge a bit from this initial post.

            

      
      
      
    

        
Cheers,
Scott

            

      
      
      
    

        
----
Scott Mitchell, Assistant Professor, Carleton University
Department of Geography & Environmental Studies, Loeb B443A
& Geomatics and Landscape Ecology Research Laboratory, Nesbitt 340
Mailing:  Loeb B349, 1125 Colonel By Dr., Ottawa, ON K1S 5B6 Canada
+1-613-520-2600 x2695 Fax: +1-613-520-4301 Scott_Mitchell at carleton.ca