Skip to content

package installation issues

11 messages · Steve Lianoglou, Michael Barnett, David Winsemius +1 more

#
Hi Michael,
On Nov 12, 2009, at 12:34 PM, Michael Barnett wrote:

            
While I can't speak to using rrgobi, I definitely use igraph everyday,  
and I'm using R 2.10

Could you elaborate on the error you were getting when trying to  
"run" (load?) igraph?

-steve

--
Steve Lianoglou
Graduate Student: Computational Systems Biology
   |  Memorial Sloan-Kettering Cancer Center
   |  Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact
#
I apologize for not being more precise. I'm pasting the printout I get  
below:

*****
 > install.packages("igraph")
trying URL 'http://lib.stat.cmu.edu/R/CRAN/bin/macosx/leopard/contrib/2.10/igraph_0.5.2-2.tgz'
Content type 'application/x-gzip' length 3312372 bytes (3.2 Mb)
opened URL
==================================================
downloaded 3.2 Mb


The downloaded packages are in
	/var/folders/7J/7JpbFNySGACoH3teAShOLE+++TM/-Tmp-//RtmpXG1NQC/ 
downloaded_packages
 > library(igraph)
Error in dyn.load(file, DLLpath = DLLpath, ...) :
   unable to load shared library '/Library/Frameworks/R.framework/ 
Resources/library/igraph/libs/x86_64/igraph.so':
   dlopen(/Library/Frameworks/R.framework/Resources/library/igraph/ 
libs/x86_64/igraph.so, 10): Symbol not found: ___gmpz_clear
   Referenced from: /Library/Frameworks/R.framework/Resources/library/ 
igraph/libs/x86_64/igraph.so
   Expected in: dynamic lookup

Error : .onLoad failed in 'loadNamespace' for 'igraph'
Error: package/namespace load failed for 'igraph'
******

What's different between my set-up and yours, Steve? I'm on a 13"  
MacBook 2.4Ghz 4GB RAM running OS X 10.5.8

-Michael
On Nov 12, 2009, at 12:38 PM, Steve Lianoglou wrote:

            
#
Heh ...
On Nov 12, 2009, at 12:45 PM, Michael Barnett wrote:

            
... that's really funny. I just tried to re-install igraph like you  
did, and now I got the same error :-)

Do you have XCode installed? Installing the package from source seems  
to do the trick:

install.packages('igraph', type='source')
... lots of compile info ...

Now loading igraph works (I had to restart R ... I guess having tried  
to load the original compiled package hose the workspace, somehow).

I typically install most packages from source, anyway, so I guess this  
is why I didn't see this before.

Does this work for you?

-steve

btw - I'm on a mac pro w/ snow leopard, not that I'm sure that matters  
now ...

--
Steve Lianoglou
Graduate Student: Computational Systems Biology
   |  Memorial Sloan-Kettering Cancer Center
   |  Weill Medical College of Cornell University
Contact Info: http://cbio.mskcc.org/~lianos/contact
12 days later
#
On Nov 24, 2009, at 3:41 PM, Michael Barnett wrote:

            
The default for install.packages might be for binary installation.  
(What does getOption("pkgType") report? If you took that effort to  
install Xcode, then you must have been thinking of building from  
"source".

?install.packages
#
On Nov 24, 2009, at 15:41 , Michael Barnett wrote:

            
The error is on my side (Fortran disappeared for some builds on the  
Leopard build machine), you don't really need Xcode at all - please  
just re-install igraph using

install.packages("igraph",,"http://r.urbanek.info")

(CRAN mirrors should pick up the updated binary within a day or two  
but the above will work immediately).

In fact there are several other packages affected, so I'd recommend  
update.packages(checkBuilt=TRUE) to all users (in a day or two or use  
the above master repository).

Thanks,
Simon
#
On Nov 24, 2009, at 5:34 PM, Michael Barnett wrote:

            
Ooops -- I'll need to have a look - it is an issue in the 64-bit  
binary (I was checking 32-bit and that was fine) .. I'll keep you  
posted. In the meantime you can try

install.packages("igraph",type='source')

Thanks,
Simon
#
Ok, I found the problem -- it's GMP. We have only 32-bit GMP on the  
build machine so igraph is trying to use it but doesn't get it for 64- 
bit. So this is really not related to the Fortran issue after all.  
I'll put 64-bit GMP on the CRAN machine tomorrow and keep you posted.  
If you don't need 64-bit just use 32-bit R in the meantime (R --arch  
i386) -- that will work.

Thanks,
Simon
On Nov 24, 2009, at 6:37 PM, Simon Urbanek wrote:

            
1 day later
#
64-bit GMP is now on the CRAN machine and I have re-compiled igraph so  
it should be fine (install from r.urbanek.info or wait for CRAN to  
synchronize). I'm checking other packages that may depend on GMP as  
well and will update them when seen.

Thanks,
Simon
On Nov 24, 2009, at 8:49 PM, Simon Urbanek wrote: