| Dirk Mittler 2004-08-24, 3:59 am |
| In June I posted some rather amateurish messages in which my goal was to
test the consistency of attributed, unbound variables. I wrote that it would
be better for this test to fail, than it would be for the test to succeed
when the attributes try to unify their variable with different values.
Further, I wasn't sure, and am still not sure if as a result of CLP, the
attributes of one variable can end up comparing other, bound variables.
I did learn from some support from this newsgroup, and from a recommended
book, that code snippets which I had posted were not adequate at the time.
Still in June I concluded my exercise with some modest code that was
supposed to be better. But because I did not know that Google was indexing
my messages, I did not decide to post my final code. Here it is:
lousy_attrib(X) :-
freeze(X, fail).
good_attrib(X) :-
freeze(X, (X = 5)).
consistent(X) :- nonvar(X).
consistent(X) :- var(X),
\+ attvar(X);
get_attrs(X, Attributes),
\+ \+ test_attrs(Attributes).
test_attrs([]).
test_attrs(att(_, Value, More)) :-
callable(Value),
call(Value),
test_attrs(More).
test_attrs(att(_, Value, More)) :-
\+ callable(Value),
test_attrs(More).
test_attrs(att(_, Value, More)) :-
\+ callable(Value),
test_attrs(More).
consistent_l([Head | Tail]) :- is_list(Head),
consistent_l(Head),
consistent_l(Tail).
consistent_l([Head | Tail]) :- \+ is_list(Head),
consistent(Head),
consistent_l(Tail).
consistent_l([]).
consistent_e(E) :- nonvar(E),
E =.. L,
consistent_l(L).
consistent_e(E) :- var(E),
consistent(E).
And here is a sample session:
1 ?- consistent(X).
X = _G271
Yes
2 ?- good_attrib(X), consistent(X).
X = _G360{freeze = user: (_G360=5)}
Yes
3 ?- X = 5, consistent(X).
X = 5
Yes
4 ?- lousy_attrib(X), consistent(X).
No
5 ?- X = A + B, consistent_e(X).
X = _G342+_G343
A = _G342
B = _G343
Yes
6 ?- X = A + B, lousy_attrib(B), consistent_e(X).
No
7 ?-
Dirk
|