Re: [Pythonmac-SIG] which fortran for pythonmac.org binaries?

2007-06-02 Thread Rob J Goedman
Agreed, the compiler can compile binaries for multiple architectures  
(we use that all the time). On that same website ( http:// 
r.research.att.com/ ):


"Building universal R is done by compliling two R binaries and  
setting r_arch parameter to ppc and i386 respectively, along with the  
proper compiler flags. Those two builds can then be installed into  
the same framework location, R install process merges them  
correspondingly."


The issue is in the link step not the compile phase.

Rob


On Jun 1, 2007, at 6:05 PM, Robert Kern wrote:


Christopher Barker wrote:

Rob J Goedman wrote:

Hope this helps a bit,


not much, but I don't need to get  how R works.

It seems there is still no Fortran compiler for OS-X that build
universal binary libs -- too bad.


The gfortran binary that Paul pointed you to does make Universal  
binaries. I
don't know why you think it doesn't. If you want an explicit  
statement to that

effect:

  http://r.research.att.com/tools/

If you want a demonstration:

[fitpack]$ gfortran -arch i386 -arch ppc -c bispev.f
[fitpack]$ file bispev.o
bispev.o: Mach-O universal binary with 2 architectures
bispev.o (for architecture i386):   Mach-O object i386
bispev.o (for architecture ppc):Mach-O object ppc

--
Robert Kern

"I have come to believe that the whole world is an enigma, a  
harmless enigma
 that is made terrible by our own mad attempt to interpret it as  
though it had

 an underlying truth."
  -- Umberto Eco

___
Pythonmac-SIG maillist  -  [email protected]
http://mail.python.org/mailman/listinfo/pythonmac-sig


On Jun 1, 2007, at 3:10 PM, Rob J Goedman wrote:
I don't think that is implied on Simon's R website. The context of  
the statement is building universal

packages for the R statistical software system.

Within that context the compiler inter-works with the latest set of  
Xcode tools to build a package (if the
package includes Fortran in addition to C/C++). These packages are  
then installed in R's library and after
loading into R, methods in the package can be called from R. I  
think it is the install step of both packages

and R/R.app that makes sure the matching architecture is chosen.

Most packages are available in binary, universal form (and include  
both binaries), but installing a package

from source will pass the correct architecture flag.

Hope this helps a bit,
Rob



___
Pythonmac-SIG maillist  -  [email protected]
http://mail.python.org/mailman/listinfo/pythonmac-sig


Re: [Pythonmac-SIG] which fortran for pythonmac.org binaries?

2007-06-02 Thread Christopher Barker
Rob J Goedman wrote:
> "Building universal R is done by compliling two R binaries and setting 
> r_arch parameter to ppc and i386 respectively, along with the proper 
> compiler flags. Those two builds can then be installed into the same 
> framework location, R install process merges them correspondingly."

Which looks a whole lot to me like you have to build the Intel and PPC 
binaries separate, then the R installer selects the right one for you 
when you install. This is a smart installer, NOT a universal binary.

However, this isn't an R list. Universal python requires Universal 
binary libs (if you want a Universal install and/or want to build 
universal py2app bundles), rather than a smart installer.

>>   http://r.research.att.com/tools/

great, thanks. That wasn't the page previously linked, and I didn't find 
that statement anywhere.

So does that mean we can build Universal binaries of Scipy now?

And back to the original question -- is the binary at python mac (only 
2.4 at my last look) Universal?

-Chris


-- 
Christopher Barker, Ph.D.
Oceanographer

NOAA/OR&R/HAZMAT (206) 526-6959   voice
7600 Sand Point Way NE   (206) 526-6329   fax
Seattle, WA  98115   (206) 526-6317   main reception
___
Pythonmac-SIG maillist  -  [email protected]
http://mail.python.org/mailman/listinfo/pythonmac-sig