Mac OS X bug in FD solver?

Dear bug-prolog list,

I posted an example to the users-prolog list and it seems that it's  
probably a bug, so I'm posting some more details here.

I'm using Mac OS X.3 (on a G4 system) and I've tried this in both the  
1.2.16 release (as downloaded and installed via fink) and also the  
current 1.2.18 release (downloaded from  
ftp://ftp.inria.fr/INRIA/Projects/contraintes/gprolog/unstable/gprolog 
-20040608.tgz).

I've had a look at the configuration / install for the most recent  
version and it all seems fine to the untrained eye. The configuration  
process recognizes the host as Darwin powerpc BSD etc. The output from  
uname -a and gprolog --version is as follows:

	scottc$ uname -a
	Darwin Conrads-Computer.local 7.4.0 Darwin Kernel Version 7.4.0: Wed  
May 12 16:58:24 PDT 2004; root:xnu/xnu-517.7.7.obj~7/RELEASE_PPC  Power  
Macintosh powerpc

	scottc$ gprolog --version
	Prolog top-Level (GNU Prolog) 1.2.18
	By Daniel Diaz
	Copyright (C) 1999-2004 Daniel Diaz
	GNU Prolog comes with ABSOLUTELY NO WARRANTY.
	You may redistribute copies of GNU Prolog
	under the terms of the GNU General Public License.
	For more information about these matters, see the files named COPYING.

The smallest test case I've found for the problem is (as I posted in  
users-prolog):

	| ?- A#=B,fd_domain([A,B],0,1),fd_labeling([A,B]).

	A = 0
	B = 0 ? a

	A = 1
	B = 1

	yes

That looks correct to me, but if I re-order the query as follows it  
doesn't seem right:

	| ?- fd_domain([A,B],0,1),A#=B,fd_labeling([A,B]).

	A = 0
	B = 0 ? a

	A = 1
	B = 0

	yes

I'm happy to have a go at debugging this (though I'd need a hint of  
where to start) or at providing information for someone else to work on  
it.

Thanks,

// Conrad