Two macros and a wrapper function

Hi all,

There are two macros I've written here:

http://paste.lisp.org/display/91129

which bind annonymous search functions to dynamically generated names using
(setf (fdefinition NAME) (lambda (x) ...)) forms.

The problem I'm experiencing happens at load time, i.e., although the macros
compile fine on their own (and expand as intended) they seem to be preventing
the function that wraps them from compiling at all. Hence the error I get is
this:

(defun make-indexed-slot-query-tools (pclass psubclasses indexed-slots)
(if (null psubclasses)
(%make-single-query-tools (list pclass) indexed-slots t)
(progn (%make-single-query-tools (append pclass psubclasses) indexed-slots)
(%make-multi-query-tools (append pclass psubclasses) indexed-slots))))

=> value INDEXED-SLOTS is not of the expected type LIST.
[Condition of type TYPE-ERROR]

Any suggestions as to why this is happening very much appreciated.

Regards,

Seb
--
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap

Sebastian Tennant <sebyte@smolny.plus.com> writes:

> Hi all,
>
> There are two macros I've written here:
>
> http://paste.lisp.org/display/91129
>
> which bind annonymous search functions to dynamically generated names using
> (setf (fdefinition NAME) (lambda (x) ...)) forms.
>

> ;;; In this simple example we can see how the arguments to wrapper
> ;;; function FOOW are appended to form a single list which is then
> ;;; passed to the macro FOO.

Delete this comment, it is misleading. Nothing gets appended when
passed to FOO.


> (defmacro foo (arg)
> `(mapcar (lambda (x) (1+ (1+ x))) ,arg))
> ;;; => FOO

There is no reason to make it a macro. Use defun here:

(defun foo (arg)
(mapcar (lambda (x) (1+ (1+ x))) arg))


> (defun foow (a b)
> (foo (append a b)))
> ;;; => FOOW



You need revise the way arguments are passed to macros: they ARE NOT evaluated.

C/USER[248]> (defmacro foo (arg)
(format t "arg = ~S~%" arg)
(let ((result `(mapcar (lambda (x) (1+ (1+ x))) ,arg)))
(format t "result = ~S~%" result)
result))
FOO
C/USER[249]> (defun foow (a b)
(foo (append a b)))
arg = (APPEND A B)
result = (MAPCAR (LAMBDA (X) (1+ (1+ X))) (APPEND A B))
FOOW
C/USER[250]> (foow '(1) '(2))
(3 4)
C/USER[251]>


> The problem I'm experiencing happens at load time, i.e., although the macros
> compile fine on their own (and expand as intended)

Why do you lie to us? They do not expand as intended!

C/USER[254]> (macroexpand '(%make-single-query-tools (list pclass) indexed-slots t))

*** - MAPC: A proper list must not end with INDEXED-SLOTS



> they seem to be preventing
> the function that wraps them from compiling at all. Hence the error I get is
> this:
>
> (defun make-indexed-slot-query-tools (pclass psubclasses indexed-slots)
> (if (null psubclasses)
> (%make-single-query-tools (list pclass) indexed-slots t)
> (progn (%make-single-query-tools (append pclass psubclasses) indexed-slots)
> (%make-multi-query-tools (append pclass psubclasses) indexed-slots))))
>
> => value INDEXED-SLOTS is not of the expected type LIST.
> [Condition of type TYPE-ERROR]
>
> Any suggestions as to why this is happening very much appreciated.

Indeed, INDEXED-SLOTS is not a list, it is a symbol.

Try to write something like:

(defun make-indexed-slot-query-tools (pclass psubclasses indexed-slots)
(if (null psubclasses)
(%make-single-query-tools (list pclass)
(a b c d) t)
(progn (%make-single-query-tools (append pclass psubclasses)
(a b c d))
(%make-multi-query-tools (append pclass psubclasses)
(a b c d)))))

instead.


--
__Pascal Bourguignon__

Quoth pjb@informatimago.com (Pascal J. Bourguignon):
> Delete this comment, it is misleading. Nothing gets appended when passed to
> FOO. [...] You need to revise the way arguments are passed to macros: they
> ARE NOT evaluated.

That's what I firmly believed, but then my example of foo and foow made me
doubt myself. Your version of foo makes things a lot clearer and the penny has
finally dropped. '(append a b)' happens to satisfy mapcar's final argument type
requirement (a list) whereas the symbol 'indexed-slots' doesn't.

Thanks a lot.

Seb
--
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap

Quoth Sebastian Tennant <sebyte@smolny.plus.com>:
> That's what I firmly believed, but then my example of foo and foow made me
> doubt myself. Your version of foo makes things a lot clearer and the penny has
> finally dropped. '(append a b)' happens to satisfy mapcar's final argument type
> requirement (a list) whereas the symbol 'indexed-slots' doesn't.

No need to correct my flawed explanation above.

Two pennies have dropped. The first one got stuck half but a second penny came
along and dislodged it. I fully understand how it all ties together now.

Seb
--
Emacs' AlsaPlayer - Music Without Jolts
Lightweight, full-featured and mindful of your idyllic happiness.
http://home.gna.org/eap