fd_all_different/1

[Claude Certain]

Hello,
I'm in a really big trouble.
Before digging more into what seems to be a bug, I want your advice:
After computing fd constraints in a rather simple way but in a really time
consumming process (12-15 hours of elapsed time), I obtain a resulting list
with duplicated elements despite the fact that the un-instantiated list has
at the very beginning the fd_all_different symbolic constraint applyed to
it.
All elements of the list are then instantiated with a fd_labelling applyed
to it.
The domain of these elements is 0..35 .
Could you see some reasons justifying this behavior. I have no warning
messages displayed during the computation.
Thank you for your help.
Claude Certain

OS: Windows XP Pro
gprolog-1.2.16