failed build with current mercurial sources

[Ben Abbott]

On Mar 8, 2008, at 5:08 AM, Jaroslav Hajek wrote:

>>
>> I have access to an older Fedora 4 system with gcc 4.0.2 - I'll try  
>> to
>> compile Octave there and reproduce the segfault.
>>
> I compiled Octave on a Fedora 4 Linux box with gcc/gfortran 4.0.2, but
> the relevant qr test passes OK. Unless someone confirms the problem on
> a different platform, I'm inclined to attribute it to a bug in
> gfortran 4.0.1.

hmmm ... since I'm compiling with gfortran4.2.2, I assume our mean a  
bug in gcc 4.0.1, or in gfortran 4.2.2?

There was quite a bit of discussion regarding fortran compilers and  
lapack on the Mac during the month of December. I recall Thomas  
Treichl, JWE, JF Mertens, and others being involved on various threads.

The discussion took place mainly on bug-octave, but in a few different  
threads. Since Mertens does not frequent the list, I searched on him  
and turned up a pair of threads.

	http://www.nabble.com/forum/Search.jtp?forum=1895&local=y&query=jf+mertens+fortran

I also found

	http://www.nabble.com/Solution-to-MacOSX-on-Intel--tc14206463.html#a14215121

	http://www.nabble.com/octave-doesn%27t-build-DSO%27s-with--fPIC-tc14193717.html#a14195452

There were additional discussions on Fink's mail-list, but I don't  
expect to find any additional information there.

For the moment, I'll try building (1) with no optimization and -fPIC,  
and (2) with no explicit options at all.

I'm just shooting at superficial targets here, so if anyone has some  
other ideas toss them up and I'll take a shot at them as well ;-)

Ben