Re: master 388a874 2/4: Do interactive mode tagging for man.el

[Eli Zaretskii]

> From: Stefan Kangas <stefan@marxist.se>
> Date: Fri, 19 Feb 2021 20:24:49 -0600
> Cc: emacs-devel@gnu.org
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Stefan, did you consider the fact that woman.el defines a different
> > mode, but uses some commands from man.el?  (I didn't try invoking the
> > commands you tagged, so maybe there's no problem.)
> 
> Thanks for catching this.

So I think the procedure for tagging should include a mandatory
grepping of all the Lisp files to see if the mode or its features are
referenced or used anywhere else.

> So if we want to keep the tagging, I see two possible workarounds:
> 
> 1) Tag the man-mode commands with both `Man-mode' and `woman-mode'.
> 
> 2) Introduce a new mode, `man-shared-mode', inheriting from
>    special-mode, that both `Man-mode' and `woman-mode' can inherit
>    from.  Then tag the commands using that.
> 
> Both options have their pros and cons.

There's a 3rd option:

 3) Put a 'completion-predicate' property on the relevant woman.el
    commands, so that command-completion-default-include-p does TRT
    with them even though the modes don't match.

I will withhold my opinion for now, and let Lars and others speak up.
This is basically the first time we bump into the issues that were
discussed here already, but sounded theoretical back then.  We now
need to fix such an issue with the tagging in practice, and we might
as well invest some thinking before we decide which way to go.
Because whatever we decide to do now, and the considerations to go
into that decision, will be most probably taken as the canonical
solution for such situations.