bug#19693: Further information

[Stefan Monnier]

> I'm sorry if it's not really a bug. After some testing I understood that
> functions (something-mode ARG) when called from Lisp, enable
> corresponding mode if ARG is omitted, if it's NIL, and even if it's not
> NIL (some people use (something-mode t) to enable the such a
> actually). So, it's quite strange: almost any argument or its absence
> enables the mode, but to disable it one must use negative number. It
> would be more logical if T would enable the mode and NIL would disable
> it.

This is the same for all minor-modes.  The reason is partly historical
and partly pragmatic.  The situation is basically:
- there are 3 different cases: enable, disable, toggle (so just using
  nil and t doesn't cut it).
- when called from Elisp the most common case (by far) is `enable' and
  in many cases it's very handy to be able to skip the argument
  (typically (add-hook 'foo-mode-hook #'bar-mode) instead of (add-hook
  'foo-mode-hook (lambda () (bar-mode <something>)))).

This means that (optional) nil has to mean "enable".  The rest doesn't
matter much.  I chose `toggle' for the toggling case, and the
"negative/positive" was chosen before I got involved in Emacs.


        Stefan