Re: Summary and next steps for (package-initialize)

[Radon Rosborough]

> > I think we should instead explore the possibility that
> > package-initialize will be called only in startup.el.
>
> I think this is unlikely to work well.

Agreed. The only way I can see this possibly working is if we
introduce a second init-file, which IMO is way too much overkill for
supporting a single package out of the thousands in Emacs.

> I think a better long-term ideal situation is either:
>
> - call package-initialize before reading ~/.emacs (this will require
>   finding some way to handle the case where the user wants to change
>   package.el's configuration in his .emacs).
> - call package-initialize automatically when needed, by having a more
>   declarative way to configure packages in ~/.emacs.
>
> The first approach looks easier from where I sit, but that's just
> a gut feeling.

There's also my proposal. From where *I* sit, that proposal looks like
the easiest, both for the end user and implementation-wise. Let's make
sure to do a comparison before discarding it.

> In the short term, we can tweak the current approach so that
> package--ensure-init-file is called in fewer cases.  E.g. only call it
> during package installation rather than at Emacs startup.

If that can be done right away, please let's do that right away. But
we can't stop there; having Emacs modify the init-file in this way is
still fundamentally broken.