Another issue here is what is meant by 'package' in this context.

I have packages from elpa, melpa, etc in ~/.emacs.d/elpa. That directory
also includes development versions of packages I'm working on, installed
via package-install-file. Then there's all the bits and pieces of elisp
that I've either written myself or found online. These are stored either in
~/.emacs.d/ or a subdirectory.

I also keep the emacs sources in .emacs.d/trunk, as I'm using the latest
development version.

Everything in that directory is backed up in a private bitbucket
repository, set to ignore .emacs.d/trunk and .emacs.d/elpa.

From Stefan's comment it sounds as if I should pull everything out of
.emacs.d except init and elpa. But it's convenient for me to have emacs in
its own contained world on my computer. Is there a risk of breaking
something with this approach?

Tyler


On Mon, Oct 21, 2013 at 2:09 PM, Josh <josh@foxtail.org> wrote:

> On Mon, Oct 21, 2013 at 6:19 AM, Tyler Smith <tyler.smith@mail.mcgill.ca>
> wrote:
> > On Mon, Oct 21, 2013 at 8:58 AM, Stefan Monnier <
> monnier@iro.umontreal.ca>
> > wrote:
> >>
> >>
> >> Still, we have the problem with ~/.emacs.d that many users put their
> >> elisp packages in there and then add it to their load-path.
> >>
> > That's exactly what I do. What is wrong with this, and what is the
> preferred
> > alternative?
>
> That's a good question:
>
>     package-user-dir is a variable defined in `package.el'.
>     Its value is "~/.emacs.d/elpa"
>
>       This variable may be risky if used as a file-local variable.
>
>     Documentation:
>     Directory containing the user's Emacs Lisp packages.
>     The directory name should be absolute.
>     Apart from this directory, Emacs also looks for system-wide
>     packages in `package-directory-list'.
>