sending and and or as an argument links their behavior

[Gregory Marton]

Hi folks,

I'm wondering what to make of this behavior, seen in both guile 1.8.1 and 
1.8.3, discovered by my colleague L. Brown Westrick, cc:ed above:

guile> (define (apply2 f a b) (f a b))
guile> (apply2 or #t #f)
#t
guile> (apply2 and #t #f)
#t
[ #f expected ]


guile> (quit)
[because we're dealing with order here, it's useful to get a clean start.
 it turns out that this doesn't reset.  You will keep getting #t for
 (apply2 and #t #f) and (apply2 and #f #t), though (apply2 and #f #f) will
 be correct.]


guile> (define (apply2 f a b) (f a b))
guile> (apply2 and #t #f)
#f
guile> (apply2 or #t #f)
#f
[ #t expected ]

(and #t #f) and (or #t #f) are not sensitive to this -- one has to pass 
them to another function and invoke them thus.  I know these are not normal 
procedures, in that (procedure? and) ===> #f  but this still seems wrong.

I've attached a new test file boolean.test that tests this, as well as 
other standard things we expect of booleans which all pass.  I am in the 
process of getting signed off w.r.t. copyright.

Thanks,
Grem

--
------ __@   Gregory A. Marton                http://csail.mit.edu/~gremio/
--- _`\<,_                                                                .
-- (*)/ (*)              The perfect is the enemy of the good.
~~~~~~~~~~~~~~~~-~~~~~~~~_~~~_~~~~~v~~~~^^^^~~~~~--~~~~~~~~~~~~~~~++~~~~~~~

boolean.test
Description: Text document