Le 29/01/2012 21:53, Thorsten Froehlich a écrit :
> On 29.01.12 12:11, Le_Forgeron wrote:
>> Le 28/01/2012 22:09, Thorsten Froehlich nous fit lire :
>>> Why would a crash in Unix-specific code have any impact on
>>> Windows-specific code?
>>
>> Not directly, but a similar error might have been copied&pasted... or
>> served as a common design (by same authors or like, so it was an error
>> on unix, it might also on others.. I just do not know their ways of
>> handling the scaling (if any) of large pictures)
> 
> Well, xcb_io.c is a system library and not POV-specific, so it cannot be
> a POV-Raz bug. Googling suggests the assertion failure is actually a
> race condition in some ATI binary-only graphics drivers...

I was talking about the missing .reserve() (and the loop of
initialisation) of m_PxCount in TakeOver (when the crash happened on the
second frame, it's because m_PxCount is indeed an empty vector, so it
does not like Povray accessing any of its element).

Ambiguous quote from previous messages are my fault, all my fault, all
my big fault.

Alain did check the windows port, seems fine. That's a good news.

what about the Mac port ? (or is it unix based now ?)


The xcb_io.c issue is yet another issue, but I cannot reproduce it (I'm
on Ubuntu with a nvidia card). I could use a VirtualMachine (VM) to get
an OpenSuse, but that won't create any Ati card.

Post a reply to this message