Posted by Sven Gothel on Jul 04, 2012; 3:06am
URL: https://forum.jogamp.org/My-Application-crashed-tp4025350p4025404.html

On 07/03/2012 10:08 PM, gouessej [via jogamp] wrote:
> A statistical evidence is not a formal evidence. I had such crashes because of
> problems of synchronization after having solved the problem with text
> renderers, I did not reproduce it systematically, it was not reproducible on
> some machines. I run JOGL applications successfully even under Windows XP.

I second that.
Due to complexity (Component space of an OpenGL Java application)
evidence from one platform is inconclusive.

It wouldn't be the 1st time that one (OpenGL) driver behaves very generous
to the application, i.e. tolerates pulled resources etc.,
where other driver are less friendly and simply crash *SIGSEGV*.

This is one reason we have to validate the JogAmp stack on our supported platforms
using a wide range of unit tests. Even this evidence only indicates
whether our software is sound, but is ofc not a proof.

A combination of unit tests which describe the proof (or attack - as you wish)
tested w/ 2-3 drivers and a (partial) code review may be best practice here.

~Sven