[Fwd: Problems with eclipse 5.5]

attached mail follows:

Hello,

my name is Alessandro Dal Palu', a Computer Science PhD student in Udine.
I'm working in collaboration with Agostino Dovier on protein folding using
constraints on finite domains. I was writing some code in Eclipse 5.5,
when I figured out that something was wrong: I tried to restrict myself to
a minimal example and I'm reporting to you the problems that I
encountered.

Using the library 'fd', the following three predicates are producing a
wrong interval propagation:

 :-  [A,B]::[-1..1],A*B #= 0.
 :-  A*B#=0.
 :-  A*B*C#=0. (only C domain is wrong)

while for

 :-  [A,B]::[0..1],A*B #= 0.

there are no problems.


Interesting to note that if playing with some other constraints before the
* operator, then everything is fine (and I get solutions):

 :-[A,B,C]::[-1..1], C#>0, A*B*C#= 0, B*C##0.


...but if I simply swap the C#>0 constraint at the end

 :-[A,B,C]::[-1..1], A*B*C#= 0, B*C##0, C#>0.

then the doamin for C is not correct and the last constraint is producing
a failure, while the predicate has the same solutions as above.

This problem is found also when the domain is unbounded:
 :- C#>0, A*B*C#= 0, B*C##0.


I hope you can help me solving this problem,
thank you in advance for your answer,

all the best,

Alessandro Dal Palu'


 -----------------------------------------------------------------------------
   Alessandro Dal Palu'
   Dip. di Matematica e Informatica, Universita' di Udine
   Via delle Scienze 206, I-33100 Udine (UD), Italy
   Phone: +39 (0432) 558 444
   Fax:   +39 (0432) 558 499
   E-mail: dalpalu@dimi.uniud.it
   home-Page: http://www.dimi.uniud.it/~dalpalu
 -----------------------------------------------------------------------------