[Fixed it with everyone's help and learned something useful along the way.  
Here's how I worked it out, mainly for my records.  Bryan]

Alright, let me try to pull some ideas from the different answers together 
here.  When I did:

install.packages("rgl", repos = "http://r-forge.r-project.org/";, type = 
"source", configure.args = "--no-cocoa")

because rgl has it's own configure script, it detected that I am on x86_64 and 
did not install the i386 version.  For building my own package I'd like to use 
both arch's and so when I tried to build it, it saw there were no i386 
resources and stopped.  I think this much is right (please correct me if not).  
This is the main problem.

Now, after reading R-inst, I thought the solution would be to do this:

install.packages("rgl", repos = "http://r-forge.r-project.org/";, type = 
"source", configure.args = c("--no-cocoa", "--arch=i386", "--libs-only"))

But that does not work, it does not recognize any of the configure.args (tried 
each separately).  I was thinking this might be because rgl's configure script 
overrode these.  So, using Rob's idea, I moved to one of the tempdirs of these 
unsuccessful installations, and tried it from the command line:

R CMD INSTALL  --libs-only rgl_0.93.924.tar.gz

but this doesn't give me the needed framework either, probably still 
auto-detecting the arch.  After several variations and re-reading R-inst, 

R --arch=i386 CMD INSTALL  --libs-only rgl_0.93.924.tar.gz

worked, in which I discovered that the flags have to be in the right place (!). 
 In particular, --arch must be with R and not CMD.  I'm sure I knew this once.

Thanks for all your patience and help.  I seem to be running.  Bryan

On Feb 5, 2013, at 11:09 AM, Prof Brian Ripley <[email protected]> wrote:

> On 05/02/2013 16:07, Bryan Hanson wrote:
>> Thanks, I'll try that in a short while.  But I'm still not sure exactly sure 
>> how this issue arose, I only work with the 64 bit version (or so I thought). 
>>  Bryan
> 
> But the R you installed from CRAN is biarch and hence tries to install 
> packages for both i386 and x86_64.  If you only want 64-bit packages, use R 
> CMD INSTALL --no-multiarch.
> 
>> 
>> On Feb 5, 2013, at 11:01 AM, Prof Brian Ripley <[email protected]> wrote:
>> 
>>> On 05/02/2013 15:56, Bryan Hanson wrote:
>>>> Is that something I can control?  I've never had to worry about that 
>>>> before, never run into it.  Clearly I'm near the edge of my expertise.  
>>>> Thanks, Bryan
>>> 
>>> See the --merge-multiarch argument to INSTALL, or 'Writing R Extensions'.
>>> 
>>> Since rgl has a configure script, you only installed it for x64_64.
>>> 
>>>> On Feb 5, 2013, at 10:53 AM, Robert J Goedman <[email protected]> wrote:
>>>> 
>>>>> Bryan,
>>>>> 
>>>>> It looks like it's mixing packages compiled for 32 and 64 bit.
>>>>> 
>>>>> Regards,
>>>>> Rob J. Goedman
>>>>> [email protected]
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> On Feb 5, 2013, at 7:46 AM, Bryan Hanson <[email protected]> wrote:
>>>>> 
>>>>>> Thanks Rob, that's a very good trick to know.
>>>>>> 
>>>>>> Duncan's suggestion about using configure.args="--disable-cocoa" worked 
>>>>>> for me - I have an rgl installation that works interactively.  Now 
>>>>>> however, when I try to install a source package that uses rgl, I get the 
>>>>>> following (also sent to Duncan separately).  Does this ring any bells?  
>>>>>> I'm not sure what the install process is doing with dependent packages 
>>>>>> that is different, but obviously it doesn't like something.  Thank you.  
>>>>>> Bryan
>>>>>> 
>>>>>> * installing *source* package ‘ChemoSpec’ ...
>>>>>> ** R
>>>>>> ** data
>>>>>> ** inst
>>>>>> ** byte-compile and prepare package for lazy loading
>>>>>> Warning: replacing previous import ‘panel.lines’ when loading ‘lattice’
>>>>>> ** help
>>>>>> *** installing help indices
>>>>>> ** building package indices
>>>>>> ** installing vignettes
>>>>>>   ‘ChemoSpec.Rnw’
>>>>>> ** testing if installed package can be loaded
>>>>>> *** arch - i386
>>>>>> ********
>>>>>> ********  Loading Bryan's .Rprofile
>>>>>> ********
>>>>>> Loading required package: brew
>>>>>> ********
>>>>>> ********  Completed Bryan's .Rprofile GET TO WORK!
>>>>>> ********
>>>>>> Warning: replacing previous import ‘panel.lines’ when loading ‘lattice’
>>>>>> Error : .onLoad failed in loadNamespace() for 'rgl', details:
>>>>>>  call: library.dynam(dll, pkg, lib.loc = lib)
>>>>>>  error: shared object ‘rgl.so’ not found
>>>>>> Error: loading failed
>>>>>> Execution halted
>>>>>> *** arch - x86_64
>>>>>> ********
>>>>>> ********  Loading Bryan's .Rprofile
>>>>>> ********
>>>>>> Loading required package: brew
>>>>>> ********
>>>>>> ********  Completed Bryan's .Rprofile GET TO WORK!
>>>>>> ********
>>>>>> Warning: replacing previous import ‘panel.lines’ when loading ‘lattice’
>>>>>> ERROR: loading failed for ‘i386’
>>>>>> * removing ‘/Users/bryanhanson/Library/R/3.0/library/ChemoSpec’
>>>>>> * restoring previous ‘/Users/bryanhanson/Library/R/3.0/library/ChemoSpec’
>>>>>> 
>>>>>> 
>>>>>> On Feb 5, 2013, at 10:35 AM, Robert J Goedman <[email protected]> wrote:
>>>>>> 
>>>>>>> Bryan,
>>>>>>> 
>>>>>>> I've seen that issue a couple of times as well. Trying to install rgl 
>>>>>>> 0.93.934 from R-Forge using R.app, I get (in the R.app console):
>>>>>>> 
>>>>>>> ...
>>>>>>> mv AGL/* .
>>>>>>> MAKINGAGL=TRUE /Library/Frameworks/R.framework/Resources/bin/R CMD 
>>>>>>> SHLIB  -o aglrgl.so *.cpp *.c \
>>>>>>>                 osx/*.mm osx/*.cpp
>>>>>>> mv: rename AGL/* to ./*: No such file or directory
>>>>>>> make: [aglrgl.so] Error 1 (ignored)
>>>>>>> g++ -arch x86_64 -I/Library/Frameworks/R.framework/Resources/include 
>>>>>>> -I/Library/Frameworks/R.framework/Resources/include/x86_64 -DNDEBUG 
>>>>>>> -DHAVE_PNG_H -DRGL_USE_COCOA -Iext -I/usr/local/include   -g -O2 -fPIC  
>>>>>>> -g -O2  -c x11gui.cpp -o x11gui.o
>>>>>>> g++ -arch x86_64 -I/Library/Frameworks/R.framework/Resources/include 
>>>>>>> -I/Library/Frameworks/R.framework/Resources/include/x86_64 -DNDEBUG 
>>>>>>> -DHAVE_PNG_H -DRGL_USE_COCOA -Iext -I/usr/local/include   -g -O2 -fPIC  
>>>>>>> -g -O2  -c x11lib.cpp -o x11lib.o
>>>>>>> g++ -arch x86_64 -I/Library/Frameworks/R.framework/Resources/include 
>>>>>>> -I/Library/Frameworks/R.framework/Resources/include/x86_64 -DNDEBUG 
>>>>>>> -DHAVE_PNG_H -DRGL_USE_COCOA -Iext -I/usr/local/include   -fPIC   -c 
>>>>>>> osx/osxgui.mm -o osx/osxgui.o
>>>>>>> osx/osxgui.mm:9:2: error: #error Cocoa backend reguires FreeType font 
>>>>>>> reneder
>>>>>>> make[1]: *** [osx/osxgui.o] Error 1
>>>>>>> make: *** [aglrgl.so] Error 1
>>>>>>> 
>>>>>>> The downloaded source packages are in
>>>>>>>         
>>>>>>> ‘/private/var/folders/fl/jchvcqm92rd7bjh6hzkd1cqc0000gn/T/RtmpgyhrHG/downloaded_packages’
>>>>>>> ERROR: compilation failed for package ‘rgl’
>>>>>>> * removing ‘/Users/rob/Library/R/3.0/library/rgl’
>>>>>>> * restoring previous ‘/Users/rob/Library/R/3.0/library/rgl’
>>>>>>>> 
>>>>>>> 
>>>>>>> On my system the following trick works. Make sure to keep a copy of the 
>>>>>>> RGL package you want to install (typically the version is in the 
>>>>>>> directory containing the downloaded packages, as printed at the end of 
>>>>>>> the install process in the R.app console). Notice that these are 
>>>>>>> removed if you quit R.app! So just keep R.app running.
>>>>>>> 
>>>>>>> I then use, in a terminal window:
>>>>>>> 
>>>>>>> cd 
>>>>>>> /private/var/folders/fl/jchvcqm92rd7bjh6hzkd1cqc0000gn/T/RtmpgyhrHG/downloaded_packages
>>>>>>> R_ARCH=/X86_64 R CMD INSTALL rgl_0.93.924.tar.gz
>>>>>>> 
>>>>>>> Towards the end of the build/install I get:
>>>>>>> 
>>>>>>> ...
>>>>>>> installing to /Users/rob/Library/R/3.0/library/rgl/libs/X86_64
>>>>>>> ** R
>>>>>>> ** demo
>>>>>>> ** inst
>>>>>>> ** preparing package for lazy loading
>>>>>>> ** help
>>>>>>> *** installing help indices
>>>>>>> ** building package indices
>>>>>>> ** testing if installed package can be loaded
>>>>>>> Warning in rgl.init(initValue) : RGL: unable to open X11 display
>>>>>>> Warning in fun(libname, pkgname) : error in rgl_init
>>>>>>> 
>>>>>>> * DONE (rgl)
>>>>>>> 
>>>>>>> These last 2 warning messages only show up if XQuartz.app is not 
>>>>>>> running.
>>>>>>> 
>>>>>>> Testing in R.app:
>>>>>>> 
>>>>>>> library(rgl)
>>>>>>> demo(rgl)
>>>>>>> 
>>>>>>> works (at least on my system!).
>>>>>>> 
>>>>>>> Regards,
>>>>>>> Rob J. Goedman
>>>>>>> [email protected]
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> On Feb 5, 2013, at 5:25 AM, Duncan Murdoch <[email protected]> 
>>>>>>> wrote:
>>>>>>> 
>>>>>>>> On 13-02-05 8:07 AM, Bryan Hanson wrote:
>>>>>>>>> Thanks for taking the time to look at this carefully Duncan.  I don't 
>>>>>>>>> see the warning you mention, and I don't see the string R_pretty0 in 
>>>>>>>>> the log either.  I'm appending right here in case there is something 
>>>>>>>>> else you want to look at.  I restarted my computer but the result is 
>>>>>>>>> the same.  I'm on OSX OS X 10.8.2 (12C60). Thanks, Bryan
>>>>>>>> 
>>>>>>>> Okay, now you're getting a different error than you described before. 
>>>>>>>> This one I think you can fix:  you need Freetype, and rgl needs to be 
>>>>>>>> able to find it when you do the install.  Normally it does that using 
>>>>>>>> the freetype-config utility, but you're not finding that:
>>>>>>>> 
>>>>>>>> [deletions]
>>>>>>>>> checking for freetype-config... no
>>>>>>>> 
>>>>>>>> I thought (from online readings) that Freetype should be available in 
>>>>>>>> a standard OSX install, so I'm not sure why it's not working for you.  
>>>>>>>> You can disable the Cocoa (native) support by configuring with the 
>>>>>>>> configure option
>>>>>>>> 
>>>>>>>> install.packages("rgl", ..., configure.args="--disable-cocoa")
>>>>>>>> 
>>>>>>>> and then you should be able to get the install to complete, but it 
>>>>>>>> might be better to figure out why freetype-config isn't working for 
>>>>>>>> you.
>>>>>>>> 
>>>>>>>> Duncan Murdoch
>>>>>>>> 
>>>>>>>> _______________________________________________
>>>>>>>> R-SIG-Mac mailing list
>>>>>>>> [email protected]
>>>>>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>>>>>>> 
>>>>>> 
>>>>> 
>>>> 
>>>> 
>>>>    [[alternative HTML version deleted]]
>>>> 
>>>> 
>>>> 
>>>> _______________________________________________
>>>> R-SIG-Mac mailing list
>>>> [email protected]
>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>>>> 
>>> 
>>> 
>>> --
>>> Brian D. Ripley,                  [email protected]
>>> 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
>>> 
>>> _______________________________________________
>>> R-SIG-Mac mailing list
>>> [email protected]
>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>> 
> 
> 
> -- 
> Brian D. Ripley,                  [email protected]
> 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

_______________________________________________
R-SIG-Mac mailing list
[email protected]
https://stat.ethz.ch/mailman/listinfo/r-sig-mac

Reply via email to