[Pkg-octave-devel] [RFU] octave3.0 to unstable

[Thomas Weber]

Am Montag, den 25.02.2008, 11:35 +0100 schrieb Rafael Laboissiere:
> * Thomas Weber <thomas.weber.mail at gmail.com> [2008-02-25 11:03]:
>  
> > Am Sonntag, den 24.02.2008, 18:43 +0100 schrieb Rafael Laboissiere:
> > > * Rafael Laboissiere <rafael at debian.org> [2008-02-21 14:15]:
> > >  
> > > I hope that this is not a problem releated to g++ 4.3 but a random failure
> > > in the autobuilder system.  Let us see how it goes for 3.0.0-5 on arm.
> > > 
> > > [1] http://buildd.debian.org/fetch.cgi?&pkg=octave3.0&ver=1%3A3.0.0-4&arch=arm&stamp=1203772565&file=log
> > 
> > The -5 build died with the expected symptom. However, I'm not sure the
> > build was actually using gcc-4.3. 
> 
> Indeed, it was done with 4.2.
>  
> > On my system, with gcc-4.2 and gcc-4.3 installed, gcc points to gcc-4.2.
> > So, I think we must set 
> > 	CXX=g++-4.3
> > explicitely in debian/rules.
> 
> The following code in debian/rules is not working as expected:
> 
>     ifneq "$(findstring $(arch), arm-linux-gnu)" ""
>     c_compiler   = gcc-4.3
>     cxx_compiler = g++-4.3
>     else
>     c_compiler   = gcc
>     cxx_compiler = g++
>     endif
> 
> Do you know how why?

I'm looking into it. I guess there's a bug in debian/rules somewhere,
the definitions and checks done with $(arch) are not consistent.



> > However, Thomas Girard has a problem with omniorb4 on ARM that sounds quite
> > similar to ours (460419). Maybe we should try with gcc-4.1 first?
> 
> Sure, as soon as we figure out how to make the code above work, we can try
> g++-4.1.  Just one question: will g++-4.1 work together well with gfortran?

No idea. But there's gfortran-4.1, so I'll check with this.

	Thomas