weird compilation effects with progn + defclass + macro

[Jan Igger]

;; Code:

(require 'closer-mop) ;; 1

(defmacro gen () ;; 2
  (let ((type (sb-mop:slot-definition-type
               (first (sb-mop:class-slots (find-class 'A))))))
    (unless (equal type 'boolean) (error "No type specified for the slot."))))

(progn ;; 3
  (defclass A () ((x :type boolean))) ;; (x :type boolean) -> (x)
  (c2mop:ensure-finalized (find-class 'A))
  (gen))

#|
1. eval-defun expressions 1 and 2
2. compile-defun expression 3
3. observe error: "no class named common-lisp-user::a" (which is unexpected)
4. compile-defun again, error is gone
5. remove type specification in class A
6. compile-defun expression 3, observe no error (should be error)
7. compile-defun expression 3 again, observe error "No type specified for the slot"
unlike compile-defun, eval-defun produces expected behavior.
What's happening here?

SBCL 2.0.5

Linux archbox 5.7.9-arch1-1 #1 SMP PREEMPT Thu, 16 Jul 2020 19:34:49 +0000 x86_64 GNU/Linux

(:SLYNK :QUICKLISP :ASDF3.3 :ASDF3.2 :ASDF3.1 :ASDF3 :ASDF2 :ASDF :OS-UNIX
 :NON-BASE-CHARS-EXIST-P :ASDF-UNICODE :X86-64 :GENCGC :64-BIT :ANSI-CL
 :COMMON-LISP :ELF :IEEE-FLOATING-POINT :LINUX :LITTLE-ENDIAN
 :PACKAGE-LOCAL-NICKNAMES :SB-CORE-COMPRESSION :SB-LDB :SB-PACKAGE-LOCKS
 :SB-THREAD :SB-UNICODE :SBCL :UNIX)
|#

[Stas Boukarev (stassats)]

You are defining the class and expanding the macro at different times, so each time you edit your form the macro gets the previous definition.

[Jan Igger (le-perv)]

In retrospect this seems such an obvious thing. Thanks for pointing it out. Going to use (eval '(gen)) now.

[Jan Igger (le-perv)]

In case someone stumbles into this issue, I found a much cleaner solution than a straightlaced eval: eval-when. It can shift evaluation to compile-time. Like this:

(progn ;; 3
  (eval-when (:compile-toplevel :load-toplevel :execute)
    (defclass A () ((x :type boolean))) ;; (x :type boolean) -> (x)
    (c2mop:ensure-finalized (find-class 'A)))
  (gen))