Chris Rathman wrote:
It's a solution to my immediate problem, but I'm still at a loss as to
why my original attempt enters into a wait. In combining some ideas
that you gave me, with my original attempts, I think I see where I've
been going astray. It would appear that binding a top level variable
within a choice statement causes my grief. In piecing together the
behavior, I wrote 3 versions of the list generator. The first (ListA)
generates the infinite choices similar to your solution with no binding
involved. The second (ListB) tries to bind the variable that is sent to
the function. The third (ListC) is what I originally started with - it
has the same problem as ListB.
So, I guess my question is: Is it not possible to bind a top level
variable within Choice? I'm speculating that the choice doesn't allow a
bind, but rather goes into a wait until such time as the top level space
gets bound. But since the top level is waiting on the choice to
succeed, you go into a wait state.
The problem is not binding the global variable. The problem is deciding
whether the local computation space has failed.
local X in
{Browse c2#{SolveN 1 fun {$} {ListC 1|2|3|X} end}}
end
Notice that in this case you will get different answers depending on the
value to which X is bound outside. So, Space.ask (which in used in Solve
(http://www2.info.ucl.ac.be/people/PVR/ds/booksuppl.oz)) does need to
wait in order to decide whether the space at hand has failed or not.
Condiser, this case:
declare X={FD.int 1#3}
proc {Script _}
choice X=0 [] X=4 end
end
{Browse {SearchAll Script}}
In this particular case, you don't need to wait for the total
determination of X since in each case Space.ask will find that the space
has failed. We can say that the set of solutions does not depend on the
value to which X is bound outside because there is no solution (given
the partial determination of X and the values to which X is bound in the
local computation spaces).
Luis
_________________________________________________________________________________
mozart-users mailing list
[email protected]
http://www.mozart-oz.org/mailman/listinfo/mozart-users
