bug#73596: 29.4; empty (let) does not signal an error

bug#73596: 29.4; empty (let) does not signal an error

[Ulrich Mueller]

Not sure if the following is an error or the intended behavior.

IIUC empty (let) should signal an error, and it did so in Emacs 27.2.
However, it returns nil now.

Apparently the new behavior was introduced by commit d41a4ad4ae6f.

bug#73596: 29.4; empty (let) does not signal an error

[Eli Zaretskii]

Ping!  Stefan, any comments?

> Cc: Stefan Monnier <monnier@iro.umontreal.ca>
> From: Ulrich Mueller <ulm@gentoo.org>
> Date: Wed, 02 Oct 2024 14:54:00 +0200
> 
> Not sure if the following is an error or the intended behavior.
> 
> IIUC empty (let) should signal an error, and it did so in Emacs 27.2.
> However, it returns nil now.
> 
> Apparently the new behavior was introduced by commit d41a4ad4ae6f.
> 
> 
> 
> 

bug#73596: 29.4; empty (let) does not signal an error

[Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors]

>> Not sure if the following is an error or the intended behavior.
>> IIUC empty (let) should signal an error, and it did so in Emacs 27.2.
>> However, it returns nil now.

The compiler emits a warning when encountering a `let` without a body,
which I think is the most important aspect.

Returning nil (when the user decides to run the code despite the
warning) seems just as good as signaling an error.  Just don't rely on it.


        Stefan

bug#73596: 29.4; empty (let) does not signal an error

[Ulrich Mueller]

>>>>> On Sat, 12 Oct 2024, Stefan Monnier wrote:

>>> Not sure if the following is an error or the intended behavior.
>>> IIUC empty (let) should signal an error, and it did so in Emacs 27.2.
>>> However, it returns nil now.

> The compiler emits a warning when encountering a `let` without a body,
> which I think is the most important aspect.

> Returning nil (when the user decides to run the code despite the
> warning) seems just as good as signaling an error.  Just don't rely on it.

(func-arity 'let)  ⇒  (1 . unevalled)

Shouldn't this be updated to return (0 . unevalled) when empty (let)
is allowed?

bug#73596: 29.4; empty (let) does not signal an error

[Andreas Schwab]

On Okt 12 2024, Ulrich Mueller wrote:

> (func-arity 'let)  ⇒  (1 . unevalled)
>
> Shouldn't this be updated to return (0 . unevalled) when empty (let)
> is allowed?

$ emacs -batch -eval '(let)'

Error: wrong-number-of-arguments (let 0)
  (let)
  eval((let) t)
  command-line-1(("-eval" "(let)"))
  command-line()
  normal-top-level()
Wrong number of arguments: let, 0

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

bug#73596: 29.4; empty (let) does not signal an error

[Ulrich Mueller]

>>>>> On Sat, 12 Oct 2024, Andreas Schwab wrote:

> On Okt 12 2024, Ulrich Mueller wrote:
>> (func-arity 'let)  ⇒  (1 . unevalled)
>> 
>> Shouldn't this be updated to return (0 . unevalled) when empty (let)
>> is allowed?

> $ emacs -batch -eval '(let)'

> Error: wrong-number-of-arguments (let 0)
>   (let)
>   eval((let) t)
>   command-line-1(("-eval" "(let)"))
>   command-line()
>   normal-top-level()
> Wrong number of arguments: let, 0

That's even stranger then, because it seems to depend on how Emacs
is invoked:

$ echo "(princ (let))" >foo.el
$ emacs -Q -batch -L . -l foo.el
nil

$ emacs -Q
M-: (let) RET
⇒ nil

bug#73596: 29.4; empty (let) does not signal an error

[Andreas Schwab]

On Okt 12 2024, Ulrich Mueller wrote:

> $ emacs -Q
> M-: (let) RET
> ⇒ nil

Don't use eval-expression.  It does not evaluate the form, but something
else.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

bug#73596: 29.4; empty (let) does not signal an error

[Ulrich Mueller]

>>>>> On Sat, 12 Oct 2024, Andreas Schwab wrote:

> On Okt 12 2024, Ulrich Mueller wrote:
>> $ emacs -Q
>> M-: (let) RET
>> ⇒ nil

> Don't use eval-expression.  It does not evaluate the form, but something
> else.

Well, it's the same with eval-last-sexp. And behavior has definitely
changed, with Emacs 29.4 (as well as 31.0.50) I see a nil result, while
with Emacs 27.2 I get:

Debugger entered--Lisp error: (wrong-number-of-arguments let 0)
  (let)
  (progn (let))
  eval((progn (let)) t)
  elisp--eval-last-sexp(nil)
  eval-last-sexp(nil)
  funcall-interactively(eval-last-sexp nil)
  call-interactively(eval-last-sexp nil nil)
  command-execute(eval-last-sexp)

bug#73596: 29.4; empty (let) does not signal an error

[Andreas Schwab]

On Okt 12 2024, Ulrich Mueller wrote:

>>>>>> On Sat, 12 Oct 2024, Andreas Schwab wrote:
>
>> On Okt 12 2024, Ulrich Mueller wrote:
>>> $ emacs -Q
>>> M-: (let) RET
>>> ⇒ nil
>
>> Don't use eval-expression.  It does not evaluate the form, but something
>> else.
>
> Well, it's the same with eval-last-sexp.

That's because does the same.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

bug#73596: 29.4; empty (let) does not signal an error

[Michael Heerdegen via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Andreas Schwab <schwab@linux-m68k.org> writes:

> >> Don't use eval-expression.  It does not evaluate the form, but something
> >> else.
> >
> > Well, it's the same with eval-last-sexp.
>
> That's because does the same.

At the end it's due to (macroexpand-all '(let ())) -> (let () nil).

Michael.

bug#73596: 29.4; empty (let) does not signal an error

[Eli Zaretskii]

Ping! What should we do with this bug report?  Close it?

> From: Andreas Schwab <schwab@linux-m68k.org>
> Cc: Stefan Monnier <monnier@iro.umontreal.ca>,  Eli Zaretskii
>  <eliz@gnu.org>,  73596@debbugs.gnu.org
> Date: Sat, 12 Oct 2024 18:32:48 +0200
> 
> On Okt 12 2024, Ulrich Mueller wrote:
> 
> >>>>>> On Sat, 12 Oct 2024, Andreas Schwab wrote:
> >
> >> On Okt 12 2024, Ulrich Mueller wrote:
> >>> $ emacs -Q
> >>> M-: (let) RET
> >>> ⇒ nil
> >
> >> Don't use eval-expression.  It does not evaluate the form, but something
> >> else.
> >
> > Well, it's the same with eval-last-sexp.
> 
> That's because does the same.
> 
> -- 
> Andreas Schwab, schwab@linux-m68k.org
> GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
> "And now for something completely different."
> 

bug#73596: 29.4; empty (let) does not signal an error

[Andreas Schwab]

If there is a bug then it is that eval-expression and eval-last-sexp
call macroexpand-all.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."