Bug#839992: libmath-quaternion-perl: autopkgtest failure: not ok 48 - rotation_angle works

[Santiago Vila]

severity 839992 important
retitle 839992 libmath-quaternion-perl: FTBFS randomly (not ok 48 - rotation_angle works)
thanks

This actually happens when the package is being built (see attach).

On Sat, 8 Oct 2016, gregor herrmann wrote:

> On Fri, 07 Oct 2016 10:54:55 +0300, Niko Tyni wrote:
> 
> >   not ok 48 - rotation_angle works
> >   #   Failed test 'rotation_angle works'
> >   #   at t/001_basic.t line 357.
> >   not ok 55 - rotation_axis works
> >   #   Failed test 'rotation_axis works'
> >   #   at t/001_basic.t line 378.
> 
> > It works for me on current sid. This is probably [rt.cpan.org #93159]:
> > the test suite has random inputs which sometimes cause failures.
> 
> I can reproduce the failure from CPAN RT#93159 by running
> t/001_basic.t in a loop, and the failure is not surprising:
> 
> my $exponent = int rand 100;
> my $q1manual = $q1;
> for (1..($exponent-1))  { $q1manual = Math::Quaternion::multiply($q1,$q1manual); }
> 
> 'int rand 100' returns results between 0 and 99, and with 0 and 1,
> the for loop is not run.

So it fails deliberately once every 50 times?

> Ok, after applying the proposed patch from CPAN RT, I don't get this
> problem anymore, and I now also get the failures we're seeing on
> ci.d.n. But it takes quite some time to trigger them; possible
> cuplrit in the code:
> 
> 353: my $theta = rand(0.25*$pi);
> 
> which could be 0.
> 
> 
> So far for today :)

Would be possible to apply the patch for stretch?

Thanks.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: libmath-quaternion-perl_0.07-1_amd64-20170125T112508Z.gz
Type: application/gzip
Size: 6260 bytes
Desc: 
URL: <http://lists.alioth.debian.org/pipermail/pkg-perl-maintainers/attachments/20170126/aecc61c0/attachment.bin>