bug#44341: 27.1; define-minor-mode generates inaccurate docstring

[Eli Zaretskii <eliz@gnu.org>]

> From: Philipp Stephani <p.stephani2@gmail.com>
> Date: Mon, 2 Nov 2020 13:28:06 +0100
> Cc: 44341@debbugs.gnu.org, Thibault Polge <thibault@thb.lt>
> 
> Am So., 1. Nov. 2020 um 16:30 Uhr schrieb Stefan Monnier
> <monnier@iro.umontreal.ca>:
> 
> > >     disable ; Should disable (as a random symbol)
> > >     disable ; Again
> > >     "What?" ; Same.
> >
> > These should be considered as errors.  Whether we catch them and signal
> > an error or silently do something else is not particular important
> > to me.
> 
> We should definitely signal an error here. A form such as (my-mode
> 'enable) actually disabling the mode is very confusing.

Signaling an error would be an incompatible change.  Someone who has

  (my-mode 'disable)

in their Lisp code will complain that it makes perfect sense.

I object to making incompatible changes in this area; let's fix the
problems wrt documentation, but it's too late to introduce
incompatible changes into this stuff, which is used all over, in Emacs
and elsewhere.  Wed already had the first bug report about such
incompatible changes, less than a day after it was pushed.  I'm quite
certain that incompatible change was unintended, but here you propose
to make it quite intentionally, and that would be a serious mistake,
IMO.