Re: prolog =and2, =or2

[Jay McCarthy]

These are all bad tests and the implementation is totally wrong.

The only way to interact with a prolog expression is to call it with a
failure continuation and see what it does. For example, it is not the
case that all prolog expressions that succeed once are equal? (meaning
are the same source code) as =succeed.

Here are some real examples for =and2:

(define initial-fk (λ (_) (error 'fk "It failed")))
(test/exn ((=and2 =fail =fail) initial-fk)
"It failed")
(test/exn ((=and2 =succeed =fail) initial-fk)
"It failed")
(test/exn ((=and2 =fail =succeed) initial-fk)
"It failed")
(test/pred ((=and2 =succeed =succeed) initial-fk)
procedure?)
(test/exn (((=and2 =succeed =succeed) initial-fk) 'ignored)
"It failed")

Notice that the only way we "look" at the result of =and2 is by
calling it. Notice in the last case, where it succeeds once, it
returns a procedure? (which is a success continuation) and when we
call that, it fails, which means it succeeded one time.

In contrast, suppose that =succeed-twice is a prolog expression that
has two ways to succeed. Then, this is a test case:

(test/pred ((=and2 =succeed-twice =succeed) initial-fk)
procedure?)
(test/pred (((=and2 =succeed-twice =succeed) initial-fk) 'ignored)
procedure?)
(test/exn ((((=and2 =succeed-twice =succeed) initial-fk) 'ignored) 'ignored)
"It failed")

Jay


On Sun, Dec 15, 2013 at 1:02 PM, Reed Wilson <reedc...@gmail.com> wrote:
> Jay,
>
> So, I can pass all of the test cases below using =succeed and =fail;
> however, I think that it needs to do more than this. ;) I am really lost
> though on how to bridge the gap between just checking for "true" and "false"
> and testing the prolog-expression's result.
>
> (define (=and2 e1 e2)
> (cond
> [(and (equal? e1 =succeed)
> (equal? e2 =succeed))
> =succeed]
> [else
> =fail]))
>
> ;; =and2 tests
> (test (=and2 =succeed =succeed) =succeed) ;; succeeds 1
> (test (=and2 =succeed =fail) =fail) ;; succeeds 0
> (test (=and2 =fail =fail) =fail) ;; succeeds 0
> (test (=and2 =fail =succeed) =fail) ;; succeeds 0
> (test (=and2 =fail (=and2 =succeed =succeed)) =fail) ;; succeeds 0
>
> Any light on this would be very appreciated.
>
> Thanks,
> Reed



--
Jay McCarthy <j...@cs.byu.edu>
Assistant Professor / Brigham Young University
http://faculty.cs.byu.edu/~jay

"The glory of God is Intelligence" - D&C 93

[Jay McCarthy]

No. =fail is a Prolog expression that fails. All Prolog expressions
take a failure continuation that they call when they fail. So, =fail
calls it's failure continuation. In my example, I use a failure
continuation that errors, but it is not correct to say that =fail
throws an error when called.

On Tue, Dec 17, 2013 at 6:23 PM, Willard Hagen <willar...@gmail.com> wrote:
> So if =fail is called it throws an error? What I mean by error is using
> racket's erring system.

>> --
>> You received this message because you are subscribed to the Google Groups
>> "byu-cs-330-fall-2013" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to byu-cs-330-fall-...@googlegroups.com.
>> For more options, visit https://groups.google.com/groups/opt_out.
>
>
>
>
> --
> -Willard

[Willard Hagen]

Thanks.  I have another question.  

The documentation says: 

"A failure-continuation? is a function that accepts a value (which it ignores) and fails; failure may mean aborting the entire Prolog search (meaning it returns nothing), or it may mean backtracking to an earlier point and trying something different (meaning it returns a success-continuation?)."

How do I determine if the search should be aborted or should be backtracked?

--
-Willard

[Jay McCarthy]

That depends on the semantics of the function you're implementing.
When you write failure continuations (as you must when you call prolog
expressions) you have to decide.

[Willard Hagen]

Ah, so the function doesn't have to be called "failure-continuation".  It is just the type of function.

--
-Willard

[Jay McCarthy]

Correct. That name is not used anywhere in the assignment text. 

Sent from my iPhone