bug#33036: Bug with the procedure nil? inside a specific code

[calcium]

Order of content:
1] --- Setup information
2] --- what is the procedure nil?
3] --- snippet of code that trigger the bug
4] --- snippet of code that doesn't trigger the bug
5] --- Summary and end

1] ---

Setup :

$ ./config.guess
x86_64-pc-linux-gnu

$ guile --version
guile: warning: failed to install locale
guile (GNU Guile) 2.2.3
etc ...

-> Guile was installed in a foreign Gnu/Linux distribution via Guix.

2] ---

The procedure nil? is not listed in the procedure index :
https://www.gnu.org/software/guile/docs/docs-2.0/guile-ref/Procedure-Index.html

There is a page that speak about nil :
https://www.gnu.org/software/guile/manual/html_node/Nil.html
However it doesn’t mention the nil? Procedure

The only documentation that I found about nil? is this one :

(procedure-documentation nil?)
=> "- Scheme Procedure: nil? x\n     Return `#t' iff X is nil, else
return `#f'."

(procedure-documentation null?)
=> "- Scheme Procedure: null? x\n     Return `#t' iff X is the empty
list, else `#f'."

The procedure documentation is not the same for nil? and null?
They seem nonetheless to be the same.

(null? #nil)
=> #t
(null? ‘())
=> #t

(nil? #nil)
=> #t
(nil? '())
=> #t

3] ---

Here is the bug that I found :

;;; -START- code with the bug -START- ;;;

(define (strange lst)
  (let loop ((lst lst)
             (is-empty '()))
    (cond ((nil? lst)
           (if (nil? is-empty) 'works
             (list 'should-not-occur is-empty)))
          (else
           (loop (cdr lst)
                 is-empty)))))

(strange '())
=> (should-not-occur ())

(strange #nil)
=> (should-not-occur ())

;;; -END- code with the bug -END- ;;;

4] ---

And here follows 3 version that doesn’t trigger this bug.

;;; in this one we remove the call to the loop in the else clause
;;; -START- code that works (1) -START- ;;;

(define (not-strange-1 lst)
  (let loop ((lst lst)
             (is-empty '()))
    (cond ((nil? lst)
           (if (nil? is-empty) 'works
             (list 'should-not-occur is-empty)))
          (else
           'no-more-bug))))

(not-strange-1 '())
=> works
(not-strange-1 #nil)
=> works

;;; -END- code that works (1) -END- ;;;

;;;  we change the ‘() of is-empty to #nil
;;; -START- code that works (2) -START- ;;;

(define (not-strange-2 lst)
  (let loop ((lst lst)
             (is-empty #nil))
    (cond ((nil? lst)
           (if (nil? is-empty) 'works
             (list 'should-not-occur is-empty)))
          (else
           (loop (cdr lst)
                 is-empty)))))

(not-strange-2 '())
=> works
(not-strange-2 #nil)
=> works

;;; -END- code that works (2) -END- ;;;

;;;  if we use null? instead of nil? at the critical point (nil? is-empty)
;;; -START- code that works (3) -START- ;;;

(define (not-strange-3 lst)
  (let loop ((lst lst)
             (is-empty '()))
    (cond ((nil? lst)
           (if (null? is-empty) 'works
             (list 'should-not-occur is-empty)))
          (else
           (loop (cdr lst)
                 is-empty)))))

(not-strange-3 '())
=> works
(not-strange-3 #nil)
=> works

;;; -END- code that works (3) -END- ;;;

5] ---

So in summary :

1) removing the call to the loop in the else clause remove the bug.
This is interesting because we don’t enter the else clause

2) using the #nil instead of ‘() did change the outcome.
But except for this case, it doesn't seem to change the outcome to use
#nil or ‘()

3) using null? Instead of nil? did change the outcome.
This is similar to point 2. except that we change the procedure instead
of the value.


I think that there is 2 way to look at this bug:

The first one is to consider that it is a bug because nil? and null?
should give the same answer.

The second one is to consider that nil? should only answer #t when we
use it with #nil, and otherwise give #f (this include ‘() ).


 ---

If you require any further information or did not understand what I
said, feel free to contact me at address@hidden