[R-SIG-Mac] compiling R 2.9.0 failed on Mac OS 10.5.6

[Simon Urbanek]

On Apr 24, 2009, at 3:36 PM, Martin Renner wrote:

> I'm trying to compile R 2.9.0 from source on a MacBookPro (1.83 GHz  
> Intel Core Duo, latest system software). I'm using the gfortran  
> version from the R 2.9.0 GUI binary image and the following:
>
> configure --disable-nls --with-aqua=no --enable-R-framework=no -- 
> with-x=no
> make
>

The above works for me (modulo the modules issue*) on a clean system,  
so I suspect your setup. In fact given the mismatches in the sources  
below I'd suggest you check that you really have the correct 2.9.0  
sources and not some mix of old and new sources.  More see inline below.

Cheers,
Simon


> This worked fine with R-2.8.1 but fails with the current version  
> (see below). Using configure with defaults gives the same result. I  
> have fink installed. Renaming /sw did not fix things either but  
> resulted in a different failure (?). I don't know whether this is  
> specific to my machine. I wondering whether there are any known  
> conflicts with fink packages? Am I missing something obvious? Any  
> pointers how to fix this would be much appreciated.
>
> Martin
>
>
> Martin Renner					
> current address:
> Alaska Maritime NWR
> Homer, AK 99603, USA
>
>
>
>
>
>
> R is now configured for i386-apple-darwin9.6.0
>
>  Source directory:          /Users/martin/src/R-2.9.0
>  Installation directory:    /usr/local
>
>  C compiler:                gcc -std=gnu99  -g -O2
>  Fortran 77 compiler:       gfortran  -g -O2
>
>  C++ compiler:              g++  -g -O2
>  Fortran 90/95 compiler:    gfortran -g -O2
>  Obj-C compiler:	     gcc -g -O2
>
>  Interfaces supported:      tcltk
>  External libraries:        readline
>  Additional capabilities:   PNG, JPEG, TIFF, iconv, cairo, ICU
>  Options enabled:           shared BLAS, Java
>
>  Recommended packages:      yes
>
>
> [snip...]
>
> gcc -std=gnu99 -I/Users/martin/src/R-2.9.0/src/extra/zlib -I/Users/ 
> martin/src/R-2.9.0/src/extra/bzip2 -I/Users/martin/src/R-2.9.0/src/ 
> extra/pcre  -I. -I../../src/include -I/Users/martin/src/R-2.9.0/src/ 
> include -I/sw/include -I/usr/local/include -DHAVE_CONFIG_H    -g -O2  
> -c /Users/martin/src/R-2.9.0/src/main/attrib.c -o attrib.o
> /Users/martin/src/R-2.9.0/src/main/attrib.c: In function ‘do_class’:
> /Users/martin/src/R-2.9.0/src/main/attrib.c:531: warning: implicit  
> declaration of function ‘S3Class’
> /Users/martin/src/R-2.9.0/src/main/attrib.c:531: warning: assignment  
> makes pointer from integer without a cast
> /Users/martin/src/R-2.9.0/src/main/attrib.c: At top level:
> /Users/martin/src/R-2.9.0/src/main/attrib.c:616: error: conflicting  
> types for ‘S3Class’
> /Users/martin/src/R-2.9.0/src/main/attrib.c:531: error: previous  
> implicit declaration of ‘S3Class’ was here
> /Users/martin/src/R-2.9.0/src/main/attrib.c: In function  
> ‘dimnamesgets1’:
> /Users/martin/src/R-2.9.0/src/main/attrib.c:854: warning: implicit  
> declaration of function ‘asCharacterFactor’
> /Users/martin/src/R-2.9.0/src/main/attrib.c:854: warning: return  
> makes pointer from integer without a cast
> make[3]: *** [attrib.o] Error 1
> make[3]: *** Waiting for unfinished jobs....
> make[2]: *** [R] Error 2
> make[1]: *** [R] Error 1
> make: *** [R] Error 1
>

I cannot reproduce this, either, but it is really strange since  
S3Class defined in the headers. Are you sure you have the correct  
sources?


> and without fink:
>
> making util.d from /Users/martin/src/R-2.9.0/src/main/util.c
> /Users/martin/src/R-2.9.0/src/main/util.c:1572:28: error: unicode/ 
> utypes.h: No such file or directory
> /Users/martin/src/R-2.9.0/src/main/util.c:1573:26: error: unicode/ 
> ucol.h: No such file or directory
> /Users/martin/src/R-2.9.0/src/main/util.c:1574:26: error: unicode/ 
> uloc.h: No such file or directory
> /Users/martin/src/R-2.9.0/src/main/util.c:1575:27: error: unicode/ 
> uiter.h: No such file or directory
> make[3]: *** [util.d] Error 1
> make[2]: *** [R] Error 2
> make[1]: *** [R] Error 1
> make: *** [R] Error 1
>

--without-ICU may help here (the line numbers don't match with the R  
2.9.0 source, though...). For some reason it seems as if R can detect  
non-Apple ICU on your system but it doesn't quite work. Apple's ICU  
comes without headers, so we work around that, but only if there is no  
other ICU. Please send me the config.log so I can try to find out why.


* - this was discussed on R-devel, probably the easiest work-around is
echo R: > ~/Makefile