[R-SIG-Mac] New intel based laptops

[Jan de Leeuw]

For those who also want to dabble in this:

Intel ifort and Intel icc do compile R, but cannot link
bundles. You can hack around this by using the bundled gcc
compiler driver for linking, but then you also
have to tell it where the Intel libraries are. You could
probably also get around this by using configure flags and/or
environmental variables. And possibly by using the -X flag
to pass icc arguments directly to the static linker ld. But
none of this is nice.

GNU gcc and gfortran from CVS have the same problem,
basically. They don't know about --bundle, so linking
of packages with shared libraries fails. But at least
you can build both gcc and gfortran on MacIntel from
here.

Apple gcc and gfortran from CVS have more problems,
because you cannot build a native gfortran from
the current Apple CVS. You cannot build it using the
bundled gcc, and not with the Intel compilers
either.

So what's left: use the bundled gcc from apple to
build gfortran from GNU-CVS. These two together build and link
R just fine -- it just doesn't pass make check yet.

On Jan 19, 2006, at 13:55 , Simon Urbanek wrote:

> On Jan 19, 2006, at 10:13 AM, Cox, Stephen wrote:
>
>> Folks - please forgive a rather uninformed post from a poor soul
>> who bailed out on the mac os several years ago...  I am considering
>> the purchase of a new MacBook Pro; however, I rely heavily on R for
>> my teaching and research.  I am assuming that the current binary of
>> v. 2.2.1 will work through the Rosetta engine, but - anyone have a
>> guess as to how long it will be before a native binary is available
>> (I am no prorgammer, and it would take me months to try to figure
>> out how to compile, especially in uncharted warters)?  Anyone know
>> of any problems I should be aware of?
>>
>
> There is an unofficial, experimental R 2.2.1 release for MacIntels
> available already (see the CRAN OS X page at the bottom - Jan kindly
> posted the direct link). The R.app GUI is in fact shipped as
> universal binary since 2.2.1.
>
> The reason for the Intel build being experimental is that it doesn't
> pass make check. The compilers (both Apple gcc and Intel) still seem
> to have bugs that cause them to miscompile code.
>
> However, there is a light at the end of the tunnel - just now I was
> able to compile R for MacIntel that is much closer to passing make
> check ("only" one part of one check failed), so I expect to have more
> serious binary sometime next week.
>
> Cheers,
> Simon
>
> _______________________________________________
> R-SIG-Mac mailing list
> R-SIG-Mac at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
>

===
Jan de Leeuw; Distinguished Professor and Chair, UCLA Department of  
Statistics;
Editor: Journal of Multivariate Analysis, Journal of Statistical  
Software
US mail: 8125 Math Sciences Bldg, Box 951554, Los Angeles, CA 90095-1554
phone (310)-825-9550;  fax (310)-206-5658;  email: deleeuw at stat.ucla.edu
.mac: jdeleeuw ++++++  aim: deleeuwjan ++++++ skype: j_deleeuw
homepages: http://gifi.stat.ucla.edu ++++++ http://www.cuddyvalley.org
   
------------------------------------------------------------------------ 
-------------------------
           No matter where you go, there you are. --- Buckaroo Banzai
                    http://gifi.stat.ucla.edu/sounds/nomatter.au