Re: summary: lilypond, lambda, and local-eval

[David Kastrup <dak@gnu.org>]

Andy Wingo <wingo@pobox.com> writes:

> What are the meanings of these expressions:

I found it amusing to see what my definitions using
with-current-continuation will produce here.

>   ;; Toplevel
>   (local-eval '(define foo 42) (the-environment))

guile> (my-eval '(define foo 42) (my-env))
guile> foo
42

>   ;; Lexical, tail context
>   (local-eval '(define foo 42) (let ((x 100)) (the-environment)))

guile> (my-eval '(define foo 42) (let ((x 100)) (my-env)))

Backtrace:
In standard input:
   1: 0* [my-eval (define foo 42) ...
   1: 1*  (let* ((x 100)) (#<continuation 1401 @ 9253970> (define foo 42)))
   1: 2   (begin #<continuation 1581 @ 9252000>)
   1: 3   [#<continuation 1401 @ 9253970> ...
   1: 4*   (define foo 42)

standard input:1:11: In procedure memoization in expression (define foo 42):
standard input:1:11: In file "standard input", line 0: Bad define placement (define foo 42).
ABORT: (syntax-error)
guile> 

>   ;; Lexical, tail context -- but with a definition
>   (local-eval '(begin (define foo 42) foo) (let ((x 100)) (the-environment)))

guile> (my-eval '(begin (define foo 42) foo) (let ((x 100)) (my-env)))

Backtrace:
In standard input:
   6: 0* [my-eval (begin (define foo 42) foo) ...
   6: 1*  (let* ((x 100)) (#<continuation 1401 @ 9219b38> (begin # foo)))
   6: 2   (begin #<continuation 1581 @ 9259a80>)
   6: 3   [#<continuation 1401 @ 9219b38> ...
   6: 4*   (begin (define foo 42) foo)
   6: 5*   (define foo 42)

standard input:6:18: In procedure memoization in expression (define foo 42):
standard input:6:18: In file "standard input", line 5: Bad define placement (define foo 42).
ABORT: (syntax-error)
guile> 

>   ;; Lexical, tail context -- but with a definition, and nested reference
>   (local-eval '(begin (define foo 42) (bar))
>               (let ((x 100)) (define (bar) foo) (the-environment)))

guile> (my-eval '(begin (define foo 42) (bar)) (let ((x 100)) (define (bar) foo) (my-env)))

Backtrace:
In standard input:
  13: 0* [my-eval (begin (define foo 42) (bar)) ...
  13: 1*  (let* ((x 100)) (letrec (#) (# #)))
In unknown file:
   ?: 2   (letrec ((bar #)) (#<continuation 1401 @ 925e7c0> (begin # #)))
In standard input:
    ...
  13: 3   [#<continuation 1401 @ 925e7c0> ...
  13: 4*   (begin (define foo 42) (bar))
  13: 5*   (define foo 42)

standard input:13:18: In procedure memoization in expression (define foo 42):
standard input:13:18: In file "standard input", line 12: Bad define placement (define foo 42).
ABORT: (syntax-error)

>   ;; Lexical, not a definition context
>   (local-eval '(define foo 42) (let ((x 100)) not-a-definition (the-environment)))

guile> (my-eval '(define foo 42) (let ((x 100)) "hello" (my-env)))

Backtrace:
In standard input:
  12: 0* [my-eval (define foo 42) ...
  12: 1*  (let* ((x 100)) "hello" (#<continuation 1401 @ 9253970> (define foo 42)))
  12: 2   (begin #<continuation 1581 @ 9252000>)
  12: 3   [#<continuation 1401 @ 9253970> ...
  12: 4*   (define foo 42)

standard input:12:11: In procedure memoization in expression (define foo 42):
standard input:12:11: In file "standard input", line 11: Bad define placement (define foo 42).
ABORT: (syntax-error)
guile> 

> What about this one:
>
>   ;; Keeping in mind that `or' expands to (let ((t ...)) (if t t ...)),
>   ;; hygienically
>   (local-eval 't '(let ((t 42)) (or #f (the-environment)))) 

Assuming that the second quote mark is a typo.

guile> (my-eval 't (let ((t 42)) (or #f (my-env))))
42
guile> 

Now of course, the continuation based approach that just hijacks the
expander and jumps in and out of it is not really a measure of how
things should work.  But it makes clear that (the-environment) is a bit
of a chimera: it captures content at a level conceptually relevant for
(define), but returns a value and has to be placed accordingly, like in
a function call or at the providing side of a binding construct.

If those different syntactic aspects prove to be too hard to conciliate,
it might help to look at the kind of interface that some other chimeras
like call-with-values or call-with-current-continuation have taken.

-- 
David Kastrup