Re: Alternative definition of the GEN type.

[Igor Schein <igor@txc.com>]

On Mon, Dec 13, 2004 at 06:29:44PM +0100, Bill Allombert wrote:
> On Mon, Dec 13, 2004 at 05:53:33PM +0100, Xavier-Francois Roblot wrote:
> > On Mon, 2004-12-13 at 17:34 +0100, Bill Allombert wrote:
> > > Hello PARI developers,
> > > 
> > > Apparently current PARI/GP CVS does not build with gcc 3.x due to
> > > strict aliasing problem. Apparenlty setting -fno-strict-aliasing 
> > > work-around the problem.
> > 
> > Just want to point out that it does compile fine with gcc 3.4.3 without
> > using the flag -fno-strict-aliasing
> 
> It compiles fine, but it does not pass the bench. matker() on inexact
> matrices is broken.  See <http://pari.math.u-bordeaux.fr/buildlog.html>
> for some evidences of this.
> 
> Are you using the latest CVS and the CFLAGS provided by configure 
> (-O3 -fomit-frame-pointer) ?

I am using gcc-3.2 on a 32bit Linux and everything compiles and
benches fine with all defaults.  On 64bit Linux, where I have
gcc-3.2.3, my experience is exactly the same as Philippe's.

Igor