bug#11782: 24.1; closure is not self-quoting

bug#11782: 24.1; closure is not self-quoting

[Christopher Schmidt]

I am not exactly sure if this is a bug, a feature or something that is
just undefined.

    (funcall (eval (eval '(let ((var)) (lambda ())))))

works fine, whereas

    (funcall (eval (eval '(let ((lex-var)) (lambda ())) t)))

does not work.

Backtrace:

    Debugger entered--Lisp error: (void-function closure)
      (closure ((lex-var) t) nil)
      eval((closure ((lex-var) t) nil))
      (funcall (eval (eval (quote (let ((lex-var)) (lambda nil))) t)))
      eval((funcall (eval (eval (quote (let ((lex-var)) (lambda nil))) t))))

I am on GNU Emacs 24.1.1 (x86_64-unknown-linux-gnu, GTK+ Version
2.24.10).

        Christopher

bug#11782: 24.1; closure is not self-quoting

[Stefan Monnier]

>     (funcall (eval (eval '(let ((lex-var)) (lambda ())) t)))
> does not work.

Note that (eval (eval '(cons 1 2))) does not work either.
So I don't consider it a bug.


        Stefan

bug#11782: 24.1; closure is not self-quoting

[Christopher Schmidt]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

Thanks for your answer.

>>     (funcall (eval (eval '(let ((lex-var)) (lambda ())) t)))
>> does not work.
>
> Note that (eval (eval '(cons 1 2))) does not work either.
> So I don't consider it a bug.

That's true, of course.  Lambdas are an exception in Emacs Lisp and
CL, though.

    (info "(elisp)Lambda Expressions")

    In Emacs Lisp, such a list is valid as an expression--it evaluates
    to itself.  But its main use is not to be evaluated as an
    expression, but to be called as a function.

I stumbled upon this issue a few times already.  So did others.

    http://stackoverflow.com/questions/7913207/how-does-the-definition-for-apply-partially-work-in-emacs-24

IMO the info node I referenced above should mention this.

        Christopher

bug#11782: 24.1; closure is not self-quoting

[Christopher Schmidt]

Ping?

(I think the issue I describe is a severe bug.)

        Christopher

bug#11782: 24.1; closure is not self-quoting

[Stefan Monnier]

> (I think the issue I describe is a severe bug.)

Could you expand on why you think it's severe?


        Stefan

bug#11782: 24.1; closure is not self-quoting

[Christopher Schmidt]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

Hi Stefan,

> Could you expand on why you think it's severe?

I think this issue is severe because the behaviour is odd and follows
neither other mature Lispy languages nor the reference manual.

    (info "(elisp)Lambda Expressions")

    In Emacs Lisp, such a list is valid as an expression--it evaluates
    to itself.

As I said 11 month ago, I hit on this issue a few times already.  So did
others.  In most cases I had a lambda form that was eval'ed twice due to
macro expansion.

    ;; -*- lexical-binding: t -*-

    (defmacro a ()
      (let ((l (lambda () )))
        `(funcall ,l)))

    (a)

I realise there is a technical need for lambda's to evaluate to closures
in a lexical environment.  Are there any technical reasons for lexical
closures to not self-quote?

BTW I am perfectly fine with just fixing the docs.

        Christopher

bug#11782: 24.1; closure is not self-quoting

[Stefan Monnier]

>> Could you expand on why you think it's severe?
> I think this issue is severe because the behaviour is odd and follows
> neither other mature Lispy languages nor the reference manual.

I see.  So we just disagree on what is considered severe (I expected
something like "this and that common idiom found in those packages fails").

    In Emacs Lisp, such a list is valid as an expression--it evaluates
    to itself.

Clearly this doc needs to be fixed since it confuses lambda expressions
and lambda values and while this conflation mostly worked in dynamic
binding, it does not work any more in the presence of closures.  I just
installed the patch below.


        Stefan


--- doc/lispref/functions.texi	2013-01-01 09:11:05 +0000
+++ doc/lispref/functions.texi	2013-05-29 00:16:58 +0000
@@ -196,9 +196,8 @@
 @end example
 
 @noindent
-In Emacs Lisp, such a list is valid as an expression---it evaluates to
-itself.  But its main use is not to be evaluated as an expression, but
-to be called as a function.
+In Emacs Lisp, such a list is a valid expression which evaluates to
+a function object.
 
   A lambda expression, by itself, has no name; it is an @dfn{anonymous
 function}.  Although lambda expressions can be used this way

bug#11782: 24.1; closure is not self-quoting

[Christopher Schmidt]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
>     In Emacs Lisp, such a list is valid as an expression--it evaluates
>     to itself.
>
> Clearly this doc needs to be fixed since it confuses lambda
> expressions and lambda values and while this conflation mostly worked
> in dynamic binding, it does not work any more in the presence of
> closures.  I just installed the patch below.

Thanks a lot!

        Christopher