11:13 a.m.
xemacs -vanilla
(functionp 'loop)
=> t
(loop for x from 0 to 3
do (princ x))
=> 0123nil
(functionp 'loop)
=> nil
--
John Paul Wallington
11:54 a.m.
>>>> "John" == John Paul Wallington
<jpw(a)shootybangbang.com> writes:
John> xemacs -vanilla
John> (functionp 'loop)
John> => t
John> (loop for x from 0 to 3
John> do (princ x))
John> => 0123nil
John> (functionp 'loop)
John> => nil
Hi John,
while I can't explain this behavior, I can suggest a workaround:
(fboundp 'loop)
works fine before and after loading the macro.
Here's an extended test case:
(boundp 'loop)
nil
(fboundp 'loop)
t
(functionp 'loop)
t
(loop for x from 0 to 3
do (princ x))
0123nil
(boundp 'loop)
nil
(fboundp 'loop)
t
(functionp 'loop)
nil
John> --
John> John Paul Wallington
--
Adrian Aichner
mailto:adrian＠xemacs.org
http://www.xemacs.org/
12:23 p.m.
while I can't explain this behavior, I can suggest a workaround:
(fboundp 'loop)
works fine before and after loading the macro.
Thanks, that's a good workaround. It's fixed in GNU Emacs, but is
written in elisp. Do you know why `functionp' is a builtin in XEmacs?
--
John Paul Wallington
2:47 a.m.
>>>> "jpw" == John Paul Wallington
<jpw(a)shootybangbang.com> writes:
jpw> Sorry, I wasn't paying attention. That isn't a workaround:
jpw> `functionp' is supposed to return nil for a macro.
Right, but the question is "what should functionp return for an
autoload?"
1. t (naively)
2. nil (it _is_ a function, but the semantics are different from what
we normally mean by function)
3. autoload it and try again
Other ideas?
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
My nostalgia for Icon makes me forget about any of the bad things. I don't
have much nostalgia for Perl, so its faults I remember. Scott Gilbert c.l.py
3:53 a.m.
Right, but the question is "what should functionp return for an
autoload?"
1. t (naively)
2. nil (it _is_ a function, but the semantics are different from what
we normally mean by function)
3. autoload it and try again
Other ideas?
It should return nil for autoloaded macros. Here is how GNU Emacs
does it:
(defun functionp (object)
"Non-nil iff OBJECT is a type of object that can be called as a function."
(or (and (symbolp object) (fboundp object)
(condition-case nil
(setq object (indirect-function object))
(error nil))
(eq (car-safe object) 'autoload)
(not (car-safe (cdr-safe (cdr-safe (cdr-safe (cdr-safe object)))))))
(subrp object) (byte-code-function-p object)
(eq (car-safe object) 'lambda)))
--
John Paul Wallington
5:24 a.m.
"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
>>>>> "jpw" == John Paul Wallington
<jpw(a)shootybangbang.com> writes:
jpw> Sorry, I wasn't paying attention. That isn't a workaround:
jpw> `functionp' is supposed to return nil for a macro.
Right, but the question is "what should functionp return for an
autoload?"
Return T because an autoloaded function is still a function. Our
FUNCTIONP docstring even defines what a function is -- "an object that
can be applied to arguments, using for example `funcall' or `apply'."
Of course, when we know that what we're going to autoload is not a
function, we might as well return NIL immediately. And that's all
Paul is asking for. A patch might look like this (minimally tested):
2002-07-08 Hrvoje Niksic <hniksic(a)xemacs.org>
* eval.c (Ffunctionp): Return nil for autoloaded macros and
keymaps.
--- src/eval.c.orig Mon Jul 8 06:16:34 2002
+++ src/eval.c Mon Jul 8 06:19:38 2002
＠＠ -3607,13 +3607,18 ＠＠
if (SYMBOLP (object))
object = indirect_function (object, 0);
- return
- (SUBRP (object) ||
- COMPILED_FUNCTIONP (object) ||
- (CONSP (object) &&
- (EQ (XCAR (object), Qlambda) ||
- EQ (XCAR (object), Qautoload))))
- ? Qt : Qnil;
+ if (COMPILED_FUNCTIONP (object) || SUBRP (object))
+ return Qt;
+ if (CONSP (object))
+ {
+ Lisp_Object car = XCAR (object);
+ if (EQ (car, Qlambda))
+ return Qt;
+ if (EQ (car, Qautoload)
+ && NILP (Fcar_safe (Fcdr_safe (Fcdr_safe (Fcdr_safe (XCDR (object)))))))
+ return Qt;
+ }
+ return Qnil;
}
static Lisp_Object
6:22 a.m.
Hrvoje Niksic <hniksic(a)xemacs.org> wrote:
"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
>>>>>> "jpw" == John Paul Wallington
<jpw(a)shootybangbang.com> writes:
>
> jpw> Sorry, I wasn't paying attention. That isn't a workaround:
> jpw> `functionp' is supposed to return nil for a macro.
>
> Right, but the question is "what should functionp return for an
> autoload?"
Return T because an autoloaded function is still a function. Our
FUNCTIONP docstring even defines what a function is -- "an object that
can be applied to arguments, using for example `funcall' or `apply'."
Of course, when we know that what we're going to autoload is not a
function, we might as well return NIL immediately. And that's all
Paul is asking for. A patch might look like this (minimally tested):
[patch elided]
Thanks Hrvoje!
Compilation says:
eval.c: In function `Ffunctionp':
eval.c:3840: warning: implicit declaration of function `Fcdr_safe'
Minimal testing:
xemacs -vanilla
(functionp 'loop)
=> nil
Hooray!
--
John Paul Wallington
7:51 a.m.
>>>> "jpw" == John Paul Wallington
<jpw(a)shootybangbang.com> writes:
jpw> eval.c: In function `Ffunctionp':
jpw> eval.c:3840: warning: implicit declaration of function `Fcdr_safe'
Add a line
EXFUN (Fcdr_safe, 1);
to lisp.h.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
My nostalgia for Icon makes me forget about any of the bad things. I don't
have much nostalgia for Perl, so its faults I remember. Scott Gilbert c.l.py
7955
days inactive
7957
days old
10 comments
4 participants
participants (4)
-
Adrian Aichner -
Hrvoje Niksic -
John Paul Wallington -
Stephen J. Turnbull