[alexandria-devel] Testing with Lisps other than SBCL, patches

[Luís Oliveira]

On Fri, Aug 1, 2008 at 7:06 PM, Nikodemus Siivola
<nikodemus at random-state.net> wrote:
>> Sun Jul 27 04:01:39 WEST 2008  Luis Oliveira <loliveira at common-lisp.net>
>>  * Mark expected failures for Allegro CL: TYPE=.[234]
>
> Applied all except this. Are these ACL bugs or thinko's in TYPE= or the tests?

ACL gives me these answers:

  (subtypep '(or cons null) 'list) => T, T
  (subtypep 'list '(or cons null)) => NIL, T

All other Lisps seem to return T for that second form. At first sight,
that didn't seem like a wrong answer, since given

  (defclass my-list () ())
  (defclass my-cons (my-list) ())
  (defclass my-null (my-list) ())

then (subtypep 'my-list '(or my-cons my-null)) will return NIL.

Looking a bit further though, the hyperspec says that "[t]he types
cons and null form an exhaustive partition of the type list". and the
hyperspec's glossary entry for "type equivalent (of two types X and
Y)" says: "having the same elements; that is, X is a subtype of Y and
Y is a subtype of X."

So I guess it's an ACL bug.

-- 
Luís Oliveira
http://student.dei.uc.pt/~lmoliv/