Re: Friendly discussion about (package-initialize)

[Mark Oteiza <mvoteiza@udel.edu>]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> Sure, I think it's reasonable for Emacs to provide special support for
>> packages which are built in. But there's such a thing as going too
>> far.  And I personally think that you've gone too far in providing
>> special support when that support actively makes it *more* difficult
>> to swap out an alternative implementation.
>
> You wrote lots and lots of lines of text just to complain about the
> addition of a single "(package-initialize)"

Not every user of package.el needs (package-initialize) in their init
file.  Not every user needs it at the beginning of the file, either.

> This said, the main motivation for calling package--ensure-init-file
> from package-initialize was to fix existing user's config where they had
> packages installed yet their .emacs didn't call package-initialize, so
> they had trouble configuring their packages.  One might argue that this
> situation is now mostly fixed and we could change tactic: only call
> package--ensure-init-file when the user installs a package.

The existing behaviour:

- subverts `package-load-list' settings
- breaks setting of package-archives that would normally happen before
  calling `package-initialize'
- writes init.el twice for some reason?
- does nothing to solve the disparity between packages and customize,
  which is the topic that opened the can of worms in the first place
- is basically undocumented

I can think of one other instance of a fragile init file lisp parser in
Emacs, and that one only exists for political reasons.