bug#33375: 27.0.50; Nested function definitions

["Andreas Röhler" <andreas.roehler@easy-emacs.de>]

On 14.11.2018 16:40, Eli Zaretskii wrote:
>> From: Andreas Röhler <andreas.roehler@easy-emacs.de>
>> Date: Wed, 14 Nov 2018 11:06:52 +0100
>>
>> (defun foo1 ()
>>     "[Some docu]
>>
>> (defun other-foo "
>>     (interactive)
>>     ;; defining another function inside
>>     (defun foo2 ()
>>       (interactive)
>>       (message "%s"   "Foo2")
>> ;; another syntactically correct function just badly indented
>> (defun foo3 ()
>>     (interactive)
>>     (message "%s" "Foo3"))
>>         )
>>     (message "%s" "Foo1"))
>>
>> ;;;;
>> Results of evaluation:
>>
>> (foo1) ==> "Foo1"
>> (foo2) ==> foo3 ?
>> (foo3) ==> "Foo3"
> 
> Why do you think there's anything wrong here?  AFAICT, each evaluation
> returns the last sexp evaluated in the respective function.  foo1 and
> foo3 have a call to 'message' as the last sexp, so they return the
> message string, but foo2's last sexp evaluates a defun form, so you
> get the result of that, which is the function foo3.  IMO, this is
> expected and correct behavior.
> 

Basically these forms were made to display a beginning-of-defun bug.

See tests in previous posts.

What puzzles me when evaluating:


(setq a (foo2))
a ==> foo3
(call-interactively 'foo2) ==> foo3

But M-x foo2 RET messages "Foo2"