[R-SIG-Mac] lme4 fails to load in R GUI 2.8.0
Prof Brian Ripley
ripley at stats.ox.ac.uk
Sat Dec 13 17:09:21 CET 2008
On Sat, 13 Dec 2008, Christopher David Desjardins wrote:
> It was suggested to update my lme4 package to the lastest version at CRAN.
> After installing it I get the following error:
>
>> library(lme4)
> Error in dyn.load(file, DLLpath = DLLpath, ...) :
> function 'cholmod_start' not provided by package 'Matrix'
> Error: package/namespace load failed for 'lme4'
>
> Searching on the internet it seems someone had the same error
with an unreleased R-forge version, I believe, but it is basic good
practice to give a URL for such comments.
> and downgrade their Matrix from *-17 to *-16 except that I can't find a
> binary for 2.8 for Matrix anywhere.
This is a known problem (as of a couple of days ago when -17 was
released). The binary of lme4 needs to be rebuilt -- I have already
suggested that a new version be released urgently so that this happens
automatically. AFAIK Simon is in Japan for another week, so most likely
this will not happen unless the Matrix/lme4 authors do help out.
[More verbosely, the new version of Matrix has changed the C-level
entrypoint names and requires that lme4 be re-compiled against the new
version of Matrix. This was not announced, and apparently unknown to the
Matrix authors.]
The only option I can offer is to install lme4 from the sources. Because
the CRAN R build has special features (static Fortran, particular BLAS
struture, special location for libgcc_s ...), even if I or someone else
builds lme4 for you it is unlikely to work on your machine. (Again, I've
asked for this to be documented so others can produce distributable
binary packages.)
To everyone else using binary packages I suggest not updating Matrix until
you see an lme update offered.
(Copying Uwe Ligges because the same thing will happen on Windows: and
bumping the lme4 number will be needed to ensure that update.packages()
gets the new version.)
>
> Thanks,
> Chris
>
>
>
>
> On Nov 19, 2008, at 2:28 PM, Simon Urbanek wrote:
>
>>
>> On Nov 19, 2008, at 15:07 , Christopher David Desjardins wrote:
>>
>>> Thanks. What I am still a bit confused about is why it loads in Aquamacs
>>> and from Terminal but not from R.app GUI. This still seems like a bit of
>>> a mystery to me.
>>>
>>
>> Because you have set different library paths in each. Check .libPaths() as
>> I was suggesting in the first e-mail.
>>
>> Cheers
>> S
>>
>>
>>>
>>> On Nov 19, 2008, at 8:34 AM, Simon Urbanek wrote:
>>>
>>>>
>>>> On Nov 18, 2008, at 22:30 , Christopher David Desjardins wrote:
>>>>
>>>>> Am I correct in thinking then that's it this old version of lme4 that I
>>>>> have installed? This version is from 14 October 2007 but I need this
>>>>> specific version for a course I'm taking.
>>>>
>>>> Yes, it appears so. You cannot use such an old version (binary) with the
>>>> current R. For the course it should not matter which version you are
>>>> using.
>>>> If you feel brave, you can try to compile the old version from sources in
>>>> the new R (download the source .tar.gz file and use R CMD INSTALL to
>>>> install it). It may not work because old versions are not designed to
>>>> work in new R versions, but you can try your luck if you really can't
>>>> live with the new version.
>>>>
>>>> Cheers,
>>>> S
>>>>
>>>>
>>>>>
>>>>> On Nov 18, 2008, at 8:23 PM, Simon Urbanek wrote:
>>>>>
>>>>>> You are picking up a very ancient binary for R 2.5.0 (as you can see
>>>>>> from the error). Check your .libPath() for old package binaries and
>>>>>> remove/reinstall them.
>>>>>>
>>>>>> Cheers,
>>>>>> S
>>>>>>
>>>>>> On Nov 18, 2008, at 8:37 PM, Christopher David Desjardins wrote:
>>>>>>
>>>>>>> When I try to load lme4 on R (2.8.0) in the R GUI for Mac OS X, I get
>>>>>>> the following warnings and errors ...
>>>>>>>
>>>>>>>> library(lme4)
>>>>>>> Loading required package: Matrix
>>>>>>> Loading required package: lattice
>>>>>>>
>>>>>>> Attaching package: 'Matrix'
>>>>>>>
>>>>>>>
>>>>>>> The following object(s) are masked from package:stats :
>>>>>>>
>>>>>>> xtabs
>>>>>>>
>>>>>>>
>>>>>>> The following object(s) are masked from package:base :
>>>>>>>
>>>>>>> colMeans,
>>>>>>> colSums,
>>>>>>> rcond,
>>>>>>> rowMeans,
>>>>>>> rowSums
>>>>>>>
>>>>>>> Error in dyn.load(file, DLLpath = DLLpath, ...) :
>>>>>>> unable to load shared library
>>>>>>> '/Library/Frameworks/R.framework/Resources/library/lme4/libs/i386/lme4.so':
>>>>>>> dlopen(/Library/Frameworks/R.framework/Resources/library/lme4/libs/i386/lme4.so,
>>>>>>> 6): Library not loaded:
>>>>>>> /Library/Frameworks/R.framework/Versions/2.5/Resources/lib/libRlapack.dylib
>>>>>>> Referenced from:
>>>>>>> /Library/Frameworks/R.framework/Resources/library/lme4/libs/i386/lme4.so
>>>>>>> Reason: image not found
>>>>>>> In addition: Warning messages:
>>>>>>> 1: In namespaceImportFrom(self, asNamespace(ns)) :
>>>>>>> replacing previous import: cov2cor
>>>>>>> 2: In namespaceImportFrom(self, asNamespace(ns)) :
>>>>>>> replacing previous import: update
>>>>>>> 3: In namespaceImportFrom(self, asNamespace(ns)) :
>>>>>>> replacing previous import: xtabs
>>>>>>> Error: package/namespace load failed for 'lme4'
>>>>>>>
>>>>>>> However, I can properly load lme4 from Aquamacs via ESS and from
>>>>>>> Terminal. I get the same warnings just not the dyn.load() error.
>>>>>>>
>>>>>>> Does anyone know what I need to do?
>>>>>>>
>>>>>>> Thanks,
>>>>>>> Chris
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> R-SIG-Mac mailing list
>>>>>>> R-SIG-Mac at stat.math.ethz.ch
>>>>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>
>
> _______________________________________________
> R-SIG-Mac mailing list
> R-SIG-Mac at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
--
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-SIG-Mac
mailing list