[PPL-devel] Regression test failures on OpenBSD

[Abramo Bagnara]

Il 14/07/2012 20:40, Edd Barrett ha scritto:
> On Sat, Jul 14, 2012 at 11:25:53AM +0200, Abramo Bagnara wrote:
>> Il 04/07/2012 12:27, Edd Barrett ha scritto:
>>> On Tue, Jul 03, 2012 at 03:23:33PM +0200, Roberto Bagnara wrote:
>>>> thanks to you for the report.  Could you please send to
>>>> ppl-devel at cs.unipr.it the config.log file generated at
>>>> configure time, along with the output of `make check' in
>>>> `/usr/ports/pobj/ppl-1.0/ppl-1.0/tests/Concrete_Expression'
>>>> once you have added `PPL_NOISY_TESTS=yes' to the environment?
>>>
>>> See attached the gzipped config.log and regression test failures in
>>> verbose mode.
>>
>> We have changed the test to produce more output to help to diagnose what
>> you're observing.
>>
>> Can you retry the above with current git head?
> 
> I applied the patch you just comitted and re-ran the tests with noisy output.

I've just tried with a VirtualBox image where I've installed a pristine
OpenBSD 5.1 amd64, gmp-5.0.5 and I'm unable to reproduce what you're
observing: both polyhedron1 and polyhedron2 succeeds.

Might you have something installed that cause the miscompilation or a
bogus library?


-- 
Abramo Bagnara

BUGSENG srl - http://bugseng.com
mailto:abramo.bagnara at bugseng.com