Re: [PATCH] Only signal package.el warnings when needed

[Eli Zaretskii <eliz@gnu.org>]

> From: Radon Rosborough <radon.neon@gmail.com>
> Date: Tue, 15 Jan 2019 11:43:39 -0700
> Cc: emacs-devel <emacs-devel@gnu.org>
> 
> > From: Eli Zaretskii <eliz@gnu.org>
> > Date: Jan 15, 2019, 10:19 AM
> >
> > I'm saying that we should hear the complaints first and devise the
> > solution only after that.
> 
> If we wait until Emacs 27.2 to fix the complaints, then it will
> already be too late to do anything useful, since everyone will have
> already suffered and changed their init-files.

That was the situation before the recent changes, and we still made
those changes.  So I think your fears are unjustified.

> I view the current situation as quite unacceptable: the most prominent
> result on Google when you search for the warning message suggests the
> following:
> 
> (unless package--initialized (package-initialize t))
> 
> That means when Emacs 27 is released we're going to see this code show
> up in everybody's init-files (upgrade -> "what's this weird warning?"
> -> Google -> copy/paste -> "oh good, it's gone now"). I don't think
> that's acceptable, because `package--initialized' is an internal
> variable which should not be referenced in user code.

So maybe the right solution is to make that variable public instead.

> Even if we fix the problem in Emacs 27.2, that won't help to get these
> references to `package--initialize' out of people's init-files.
> Deciding not to merge this patch or something similar will create a
> massive amount of technical debt in places we can't touch (everyone's
> Emacs configurations).

I think "massive amount" is a significant exaggeration.

Anyway, there's no way around some of this when we are trying to solve
such problems.  That there's some fallout cannot be used as an
ultimate argument in favor or against some change.