Re: [AD] Allegro in Debian

[Tobias Hansen <tobias.han@xxxxxxxxxx>]

On 08.07.2012 03:21, Peter Wang wrote:
> On Sat, 07 Jul 2012 10:59:45 +0200, Tobias Hansen <tobias.han@xxxxxxxxxx> wrote:
>> Thanks for addressing the issues so quickly.
>>
>> Am 07.07.2012 04:58, schrieb Peter Wang:
>>> On Sat, 07 Jul 2012 01:48:11 +0200, Tobias Hansen <tobias.han@xxxxxxxxxx> wrote:
>>>>
>>>> Four Allegro 5 tests are failing in a i386 chroot on a amd64 host (with
>>>> --no-display). Can someone test if this happens generally on i386?
>>>>
>>>> FAIL test hl thick-50 [sw] - hash=62819701
>>>> FAIL test hl thick-50 clip [sw] - hash=a601764e
>>>> FAIL test hl nolight [sw] - hash=9889e6b5
>>>> FAIL test hl nolight clip [sw] - hash=85600dc7
>>>
>>> Works on http://alleg.de/tests.py
>>>
>>> It's probably just a small floating point difference.  You can send the
>>> failing test output.  Run in the build/tests directory:
>>>
>>>     ./test_driver -s ../../tests/test_prim.ini
>>>
>>
>> The output is attached. If it's a floating point difference, would you
>> make sure that they are compared at the correct precision to avoid false
>> positives?
> 
> There is an acceptable one pixel difference at (581,196) in the
> al_draw_spline call.  Can you try this attached patch?
> 
> Peter
> 

That fixes it, thanks!

Tobias