[R-sig-Geo] Installation of rgdal from CRAN on MACOSX does not work

Agustin Lobo alobolistas at gmail.com
Wed Jan 12 11:05:14 CET 2011 

Thanks,

it has worked with no problems on 4 Macs, both Snow Leopard and Leopard.

Agus


2011/1/10 Roger Bivand <Roger.Bivand at nhh.no>:
> On Sat, 8 Jan 2011, Roger Bivand wrote:
>
>> On Thu, 30 Dec 2010, Roger Bivand wrote:
>>
>>> On Wed, 29 Dec 2010, Agustin Lobo wrote:
>>>
>>>> Roger,
>>>>
>>>> But
>>>> http://www.kyngchaos.com/files/software/frameworks/rgdal-0.6.29-1.zip
>>>> installs with no problems on the same machine and on the same R, would
>>>> that not
>>>> discard the possibility of an
>>>>>
>>>>> issue of your R installation,
>>>>> possibly a multi-architecture issue
>>>>
>>>> as you suggest?
>>>>
>>>> Please note I'm not personally in a hurry any more as we can work with
>>>> the rgdal from kyngchaos.
>>>
>>> I will have access to an OSX laptop in a little over a week, and will try
>>> out the CRAN rgdal binaries. Just one report of installation failure isn't
>>> enough. I think that maybe someone on R-sig-mac might know why rgdal.so is
>>> looking for libR.dylib; I don't use OSX, and don't have easy access to such
>>> a machine, which is why I asked for others to comment.
>>
>> On a Leopard 10.5.8, I see the problem reported by Agus for the CRAN-extra
>> OSX binary rgdal package and freshly installed R 2.12.1 (libR.dylib image
>> not found) in the 32-bit R-GUI, but no problem when running R from the
>> command line. A different problem (mach-o, but wrong architecture) occurs in
>> 64-bit systems, both run from the command line and from R-GUI. Could anyone
>> who wants to follow this up please post to R-sig-Mac, which is the correct
>> list for OSX issues?
>>
>> Conclusion: on Leopard, the CRAN extras rgdal binary can be used in the
>> 32-bit R application run from the command line, but not otherwise. R 2.12.1
>> installs the 64-bit GUI into the OSX applications toolbar - one needs to
>> open a terminal and type "R", checking that the length of a pointer is 4, to
>> run the version that works.
>
> Prof. Ripley has updated the OSX CRAN extras rgdal binaries to rgdal 0.6.33,
> and this new version is confirmed to work under Leopard both 32 and 64 bit,
> both GUI and terminal. So the simplest way to install rgdal for OSX for the
> basic set of GDAL and OGR drivers is once again:
>
> setRepositories(ind=1:2)
> install.packages("rgdal")
>
> Roger
>
>>
>> Roger
>>
>> PS. The symptoms are:
>>
>> R64, R-GUI 64
>>
>>> library(rgdal)
>>
>> Loading required package: sp
>> Error in dyn.load(file, DLLpath = DLLpath, ...) :
>>  unable to load shared object
>> '/Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/x86_64/rgdal.so':
>>
>>
>> dlopen(/Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/x86_64/rgdal.so,
>> 6): no suitable image found.  Did find:
>>
>>
>> /Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/x86_64/rgdal.so:
>> mach-o, but wrong architecture
>> Error: package/namespace load failed for 'rgdal'
>>
>>
>>
>> R-GUI 32
>>
>>> library(rgdal)
>>
>> Loading required package: sp
>> Error in dyn.load(file, DLLpath = DLLpath, ...) :
>>  unable to load shared object
>> '/Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i386/rgdal.so':
>>
>>
>> dlopen(/Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i386/rgdal.so,
>> 6): Library not loaded: libR.dylib
>>  Referenced from:
>> /Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i386/rgdal.so
>>  Reason: image not found
>> Error: package/namespace load failed for 'rgdal'
>>
>>
>>>
>>> Roger
>>>
>>>> I just want to report the error to help
>>>> having an rgdal installation for MACOSX as
>>>> simple as for Windows.
>>>> I understand installing from source will always be a bit more
>>>> difficult on MACOSX than on linux as there are some basic tools and
>>>> libraries that are missing on regular MAC systems that must be
>>>> previously installed, but this is not my concern now as I'm focusing
>>>> on installations from binaries.
>>>>
>>>> Thanks!
>>>>
>>>> Agus
>>>>
>>>>
>>>>
>>>> 2010/12/29 Roger Bivand <Roger.Bivand at nhh.no>:
>>>>>
>>>>> On Wed, 29 Dec 2010, Agustin Lobo wrote:
>>>>>
>>>>>> While the installation from
>>>>>> http://www.kyngchaos.com/files/software/frameworks/rgdal-0.6.29-1.zip
>>>>>> works fine
>>>>>> the installation of rgdal from CRAN on MACOSX does not work, despite
>>>>>> my previous message that this issue was solved as well
>>>>>
>>>>> Please do read the error message:
>>>>>
>>>>>>  unable to load shared object
>>>>>>
>>>>>
>>>>> '/Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i$
>>>>>>
>>>>>
>>>>> dlopen(/Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal$
>>>>>>
>>>>>> 6): Library not loaded: libR.dylib
>>>>>>  Referenced from:
>>>>>>
>>>>>
>>>>> /Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i3$
>>>>>>
>>>>>>  Reason: image not found
>>>>>
>>>>> The image that is not found is libR.dylib, the main R dylib against
>>>>> which
>>>>> rgdal.so is linked. So this is still an issue of your R installation,
>>>>> possibly a multi-architecture issue (is libR.dylib the i386 version,
>>>>> guessing - is libR64.dylib the 64-bit Intel version?).
>>>>>
>>>>> Could any OSX users who are using CRAN extra rgdal please help Agus to
>>>>> resolve this - it could be that local path modifications are hiding
>>>>> libR.dylib, or that R was installed 64-bit only, and the INSTALL_opts=
>>>>> argument to install.packages() was not set correctly? I assume that the
>>>>> CRAN
>>>>> extras OSX rgdal binary is built against a standard R OSX Intel
>>>>> install,
>>>>> which is multi-architecture.
>>>>>
>>>>> For a fully updated rgdal, the best combination remains installing the
>>>>> full
>>>>> GDAL framework from Kyngchaos, the full tools collection:
>>>>>
>>>>> http://r.research.att.com/tools/
>>>>> http://cran.r-project.org/bin/macosx/tools
>>>>>
>>>>> and install from CRAN source. For a lab, the instructor would rather
>>>>> build a
>>>>> local OSX rgdal binary package using the tools collection on one
>>>>> development
>>>>> machine with all the correct path variables set (R CMD check rgdal; R
>>>>> CMD
>>>>> INSTALL --build rgdal), paying attention to multi-architectures, and
>>>>> distribute to lab machines from there. This means that the lab
>>>>> instructor
>>>>> can keep rgdal updated if need be, as both Kyngchaos and CRAN extras
>>>>> rgdal
>>>>> OSX binary packages typically lag CRAN source.
>>>>>
>>>>> Please also note that there are very many very old GDAL binaries out
>>>>> there
>>>>> that users pick up and apply, missing all the subsequent bug-fixes and
>>>>> improvements - GDAL 1.8.0 release candidate 1 is now out, so we are
>>>>> looking
>>>>> at an upgrade to 1.8.0 during January, which Kyngchaos will most likely
>>>>> pick
>>>>> up quickly, probably with a current rgdal binary.
>>>>>
>>>>> Roger
>>>>>
>>>>>
>>>>>> (my confusion was a consequence of mixing both installations in the
>>>>>> same session):
>>>>>>
>>>>>>> setRepositories(ind=1:2)
>>>>>>> install.packages("rgdal")
>>>>>>
>>>>>> --- Please select a CRAN mirror for use in this session ---
>>>>>> probando la URL
>>>>>>
>>>>>>
>>>>>> 'http://www.stats.ox.ac.uk/pub/RWin/bin/macosx/leopard/contrib/2.12/rgdal_0.6-31.tgz'
>>>>>> Content type 'application/x-gzip' length 10203689 bytes (9.7 Mb)
>>>>>> URL abierta
>>>>>> ==================================================
>>>>>> downloaded 9.7 Mb
>>>>>>
>>>>>>
>>>>>> The downloaded packages are in
>>>>>>
>>>>>>
>>>>>>  /var/folders/iR/iRxOvaG9F10viNoa7WceNU+++TI/-Tmp-//RtmpqIhqn9/downloaded_packages
>>>>>>>
>>>>>>> require(rgdal)
>>>>>>
>>>>>> Loading required package: rgdal
>>>>>> Loading required package: sp
>>>>>> Error in dyn.load(file, DLLpath = DLLpath, ...) :
>>>>>>  unable to load shared object
>>>>>>
>>>>>>
>>>>>> '/Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i386/rgdal.so':
>>>>>>
>>>>>>
>>>>>>  dlopen(/Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i386/rgdal.so,
>>>>>> 6): Library not loaded: libR.dylib
>>>>>>  Referenced from:
>>>>>>
>>>>>>
>>>>>> /Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i386/rgdal.so
>>>>>>  Reason: image not found
>>>>>>>
>>>>>>> sessionInfo()
>>>>>>
>>>>>> R version 2.12.1 (2010-12-16)
>>>>>> Platform: i386-apple-darwin9.8.0/i386 (32-bit)
>>>>>>
>>>>>> locale:
>>>>>> [1] es_ES.UTF-8/es_ES.UTF-8/C/C/es_ES.UTF-8/es_ES.UTF-8
>>>>>>
>>>>>> attached base packages:
>>>>>> [1] stats     graphics  grDevices utils     datasets  methods   base
>>>>>>
>>>>>> other attached packages:
>>>>>> [1] sp_0.9-76
>>>>>>
>>>>>> loaded via a namespace (and not attached):
>>>>>> [1] grid_2.12.1     lattice_0.19-13 tools_2.12.1
>>>>>>
>>>>>> Note that rgdal.so is where it should be:
>>>>>>
>>>>>> imac-de-agustin-lobo:~ agustinlobo$ ls -l
>>>>>>
>>>>>>
>>>>>> /Library/Frameworks/R.framework/Versions/2.12/Resources/library/rgdal/libs/i386/
>>>>>> total 15088
>>>>>> -rwxr-xr-x  1 agustinlobo  admin  7722168  4 dic 15:57 rgdal.so
>>>>>>
>>>>>> This error is not solved by setting the GDAL path.
>>>>>>
>>>>>> Agus
>>>>>>
>>>>>
>>>>> --
>>>>> Roger Bivand
>>>>> Economic Geography Section, Department of Economics, Norwegian School
>>>>> of
>>>>> Economics and Business Administration, Helleveien 30, N-5045 Bergen,
>>>>> Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
>>>>> e-mail: Roger.Bivand at nhh.no
>>>>>
>>>>> _______________________________________________
>>>>> R-sig-Geo mailing list
>>>>> R-sig-Geo at r-project.org
>>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>>>>>
>>>>
>>>
>>>
>>
>>
>
> --
> Roger Bivand
> Economic Geography Section, Department of Economics, Norwegian School of
> Economics and Business Administration, Helleveien 30, N-5045 Bergen,
> Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
> e-mail: Roger.Bivand at nhh.no
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>

More information about the R-sig-Geo mailing list