[gentoo-science] anyone on the list with ppc hardware?

[gentoo-science] anyone on the list with ppc hardware?

[François Bissey]

Hi all,

I have been trying to get ecls running on my ppc box but the build fails.
It is somewhat sage-on-gentoo related as I would prefer to build maxima 
with ecls as used in standard sage.
See bug #308909 (http://bugs.gentoo.org/show_bug.cgi?id=308909).

I have been looking at it for a while and I don't I have a clue.
The fact that the build goes further with gcc-4.3.4 than gcc-4.4.3 seems
to point the finger to a gcc problem on ppc but I am not sure. 

So is there anyone else with ppc hardware that could test this?

Francois

Re: [gentoo-science] anyone on the list with ppc hardware?

[Francesco Biscani]

On Thu, Mar 18, 2010 at 10:51 AM, François Bissey
<f.r.bissey@massey.ac.nz> wrote:
> Hi all,
>
> I have been trying to get ecls running on my ppc box but the build fails.
> It is somewhat sage-on-gentoo related as I would prefer to build maxima
> with ecls as used in standard sage.
> See bug #308909 (http://bugs.gentoo.org/show_bug.cgi?id=308909).
>
> I have been looking at it for a while and I don't I have a clue.
> The fact that the build goes further with gcc-4.3.4 than gcc-4.4.3 seems
> to point the finger to a gcc problem on ppc but I am not sure.
>
> So is there anyone else with ppc hardware that could test this?
>
> Francois
>
>

Hi Francois,

It might be a long shot, but try to remove the mtune/march and play
around a bit with the optimisation flag.

 F.

Re: [gentoo-science] anyone on the list with ppc hardware?

[François Bissey]

> On Thu, Mar 18, 2010 at 10:51 AM, François Bissey
> 
> <f.r.bissey@massey.ac.nz> wrote:
> > Hi all,
> > 
> > I have been trying to get ecls running on my ppc box but the build fails.
> > It is somewhat sage-on-gentoo related as I would prefer to build maxima
> > with ecls as used in standard sage.
> > See bug #308909 (http://bugs.gentoo.org/show_bug.cgi?id=308909).
> > 
> > I have been looking at it for a while and I don't I have a clue.
> > The fact that the build goes further with gcc-4.3.4 than gcc-4.4.3 seems
> > to point the finger to a gcc problem on ppc but I am not sure.
> > 
> > So is there anyone else with ppc hardware that could test this?
> > 
> > Francois
> 
> Hi Francois,
> 
> It might be a long shot, but try to remove the mtune/march and play
> around a bit with the optimisation flag.
> 
Kind of a good call. For some reasons with ppc I have to add -std=c99
to my CFLAGS if I compile with gcc-4.4.3 - note that it is not necessary
with gcc-4.3.4 on ppc or gcc-4.4.3 on x86. Just adding that got me to 
a further failure to lisp bootstrapping (that I had with gcc-4.3.4).

Removing all other flags however seems to have done the trick, so there
may be one that is dodgy to ecls in the lot.
I will see if building has finished later, cross finger :)
If it does I may take some time to see which flag(s) lead to the failure.
In the meantime I also asked for help from the ppc forum.

Francois

Re: [gentoo-science] anyone on the list with ppc hardware?

[Francesco Biscani]

On Fri, Mar 19, 2010 at 3:18 AM, François Bissey
<f.r.bissey@massey.ac.nz> wrote:
> Kind of a good call. For some reasons with ppc I have to add -std=c99
> to my CFLAGS if I compile with gcc-4.4.3 - note that it is not necessary
> with gcc-4.3.4 on ppc or gcc-4.4.3 on x86. Just adding that got me to
> a further failure to lisp bootstrapping (that I had with gcc-4.3.4).
>
> Removing all other flags however seems to have done the trick, so there
> may be one that is dodgy to ecls in the lot.
> I will see if building has finished later, cross finger :)
> If it does I may take some time to see which flag(s) lead to the failure.
> In the meantime I also asked for help from the ppc forum.

In my case, it was specifically the march/mtune/mcpu flags that
provoked compilation failure with similar error messages to yours
while emerging certain packages from the portage tree (MySQL was one
of them,  IIRC). Removing them and using just -O2 did the trick in my
case.

 F.