Re: Macintosh / Intel

[Darren Foltinek <darren dot foltinek at shaw dot ca>]

I got GSL to compile by hacking the ieee-utils functionality.

1) Added ieee-utils/fp-darwin-x86.c with no support for runtime 
defining of FP precision, rounding or exceptions (just an empty 
gsl_ieee_set_mode function returning GSL_SUCCESS), suggested by 
Richard Mather. 
2) Modified ieee-utils/fp.c to include fp-darwin-x86.c, switched 
by...
3) the HAVE_DARWNX86_IEEE_INTERFACE that I added to config.h

It was beyond my hacking skills to get configure to generate the 
config.h file with the new DARWINX86 flag.

Certainly not a proper solution, but it works for now...  :-)

Tried yesterday to build a universal library - no success.  Very 
complex.  :-(

- Darren



From: Brian Gough <bjg@network-theory.co.uk>
Date: Thursday, March 2, 2006 12:23 pm
Subject: Re: Macintosh / Intel

> Darren Foltinek writes:
> > We are using the GSL library on Mac/PPC machines, and 
recently 
> > got one of the new Intel-based Macs.
> > 
> > Trying to compile GSL on the Intel Mac, it does not use the 
> correct 
> > ieee-utils/fp-* code and so the compile fails.
> > 
> > This machine has, in it's /usr/include/architecture directory 
> BOTH 
> > a "ppc" subdirectory and an "i386" subdirectory.
> > 
> > The config.guess script correctly identifies the machine as 
> "i386-
> > apple-darwin8.4.1".
> 
> Hello,
> 
> If you change the line
> 
>  #include <architecture/ppc/fp_regs.h> 
> 
> in ieee-utils/fp-darwin.c to
> 
>  #include <architecture/i386/fp_regs.h> 
> 
> does it work?
> 
> -- 
> Brian Gough
> 
> Network Theory Ltd,
> Publishing the GSL Manual --- http://www.network-
> theory.co.uk/gsl/manual/