[R] Error compiling R.2.8.1 with gcc 4.4 on Mac OS 10.5.6

[Prof Brian Ripley]

On Mon, 22 Dec 2008, Peter Dalgaard wrote:

> Mike Lawrence wrote:
>> Hi all,
>> I've encountered a build error with the latest R source (2.8.1). This is a
>> relatively fresh install of OS Leopard (10.5.6), latest developer tools
>> installed, gcc/g++/gfortran version 4.4 installed (via
>> http://hpc.sourceforge.net/, after which I updated the gcc & g++ symlinks 
>> to
>> link to the 4.4 versions; gfortran used the 4.4 version without updating 
>> the
>> symlink).
>> 
>> Ultimately I wanted to instalI pnmath, so as per a previous thread (
>> http://www.nabble.com/Parallel-R-tt18173953.html#a18196319) I built with:
>> 
>> LIBS=-lgomp ./configure --with-blas='-framework vecLib'
>> make -j4
>> 
>> The configure runs without a hitch, but make fails, throwing an error
>> seemingly related to qdCocoa:
>> making qdCocoa.d from qdCocoa.m
>> <built-in>:0: internal compiler error: Abort trap
>> 
>> Below is the output of configure, followed by the output of make (error is
>> in the last 10 lines). Any suggestions to fix this would be greatly
>> appreciated.
>
> Ouch. As you probably realize, this is very Mac-specific, and the actual bug 
> is in GCC and has nothing to do with R.
>
> I don't think  there's any obvious way to fix this, but there might be a 
> temporary workaround. I'd consider making the .d file by hand, so that make 
> doesn't even try to build it. You can probably rather easily create one based 
> on one of the other .d files in the same directory (.d files are usually 
> overkill, it is probably not necessary to get it completely right.)
>
> Next question is whether this GCC can handle .m files at all....

It can: at least the released versions can provided the ObjC language is 
installed.  Most likely if making the .d file is an error then actual 
compilation will be too.

Note that this is a non-Apple build of a non-FSF version of a non-released 
version of gcc. I think the workaround is to use Apple compilers for 
Mac-specific code. So either use --with-aqua with Apple compilers, or 
--without-aqua with others.  From memory (I am not on my Mac right now), 
the latter will not build the quartz code at all: certainly 
src/library/grDevices/src/Makefile suggests so.

I'd suggest reporting Mac-specific problems on R-sig-mac, and problems 
with building R on experimental systems (including unreleased compilers) 
to the system's developers or perhaps R-devel.

>
>
>
>> making qdBitmap.d from qdBitmap.c
>> making qdPDF.d from qdPDF.c
>> making qdCocoa.d from qdCocoa.m
>> <built-in>:0: internal compiler error: Abort trap
>> Please submit a full bug report,
>> with preprocessed source if appropriate.
>> See <http://gcc.gnu.org/bugs.html> for instructions.
>> make[4]: *** [qdCocoa.d] Error 1
>> make[4]: *** Waiting for unfinished jobs....
>> make[3]: *** [all] Error 1
>> make[2]: *** [R] Error 1
>> make[1]: *** [R] Error 1
>> make: *** [R] Error 1
>> 
>> 
>> 
>
>
> --
>   O__  ---- Peter Dalgaard             Øster Farimagsgade 5, Entr.B
>  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
> (*) \(*) -- University of Copenhagen   Denmark      Ph:  (+45) 35327918
> ~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)              FAX: (+45) 35327907
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

-- 
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