Re: [igraph] 1 test fails on ppc64 gcc-4.3

[Gabor Csardi]

I have to admit that this test is stochastic, too.
It might fail once in every ten thousand runs. 
It is quite amazing that you run into both of these,
in fact i'm afraid that something is wrong here.

Anyway, here is another patch.
I wouldn't swear that this is enough, but it is unlikely that the 
check fails again anyway. I would be better to have a ppc64 machine
to see what actually goes wrong....

Hopefully there are no more stochastic tests in igraph....

Thanks,
Gabor

On Tue, Feb 26, 2008 at 01:37:04PM -0500, Neal Becker wrote:
> On Tuesday 26 February 2008, Gabor Csardi wrote:
> > Neal, thanks.
> >
> > Hmmm, a tip: that test is a bit stochastic and fails about
> > once in every ten thousand runs. Could you just run the
> > same thing again to see whether this is the problem.
> >
> > If it is something else, then I really need the
> > output of the failed test, otherwise it is impossible
> > to tell.
> >
> > Gabor
> >
> > On Tue, Feb 26, 2008 at 09:39:22AM -0500, Neal Becker wrote:
> > > After that last patch (stdlib.h) I can pass checks on  igraph-0.5 x86_64
> > > gcc-4.3.  But ppc64 fails.  I don't have direct access to this machine,
> > > all I can send you is the build log:
> > >
> 
> 
> I added this patch, but now I have a new failure, this time on i386:
> 
>  35: Erdos-Renyi model (igraph_erdos_renyi_game):    FAILED 
> (structure_generators.at:36)
>  
> Any ideas?

-- 
Csardi Gabor <address@hidden>    UNIL DGM

ER.patch
Description: Text Data