RE: [External] : Re: Moving packages out of core to ELPA

[Drew Adams <drew.adams@oracle.com>]

> OK, then I suppose staging in lisp/obsolete/ makes sense.
>
> I do worry about the connotations of obsolete meaning
> "stop using", "does not work at all" or "soon to vanish".
> In fact IDLWAVE does still work and is useful for a small
> set of users, it just does more harm than good in core, IMO.

FWIW, I agree.  The label "obsolete" doesn't
correspond to the description.

I think of a longstanding, likely hardly used,
standard library such as `completion.el', for
example.  That code still works fine.

Following the criteria described, that could
well end up being sent to lisp/obsolete/.

But (IMO) it's food for thought.  The approach
could perhaps be picked up and revitalized,
maybe even combined with some other existing
in-buffer completion library - either within
Emacs or in some 3rd-party code.

IOW, it's an example (IMO) of code that could
be mined or thought about, to do something
new or in a new way.  And again, it still just
works.

I wouldn't object to it (& similar) libraries
being sent to a different directory, such as
"obsolete".  It's just the name "obsolete"
that I think is unfortunate for what's been
described here (no development support etc.). 

> One idea would be to create another category for packages "on their way
> out" of core.  I'm sure there's a better name, but lisp/noncore comes
> to mind.  The intended difference being that obsolete packages will
> likely disappear entirely from the project, whereas "noncore" packages
> will be migrated to ELPA, where users can find them once they have left
> core for good.

Something like that, yes, +1.  Some name that
doesn't have the connotation that the code's
been replaced by something better etc.  The
code's simply been removed from maintenance,
active development, and automatic loading.

 Eli> the package fails to load automatically

Failing to load automatically is _fine_ for
such a library.  Calling that out in NEWS is
also fine.

Calling the library "obsolete" and calling
the move "deprecation" or "obsoleting" is
misleading.  It pretty much guarantees that
folks won't try it or even look at it.

If we keep it around then a possible reason
for doing that should be that we think
there's some merit/interest in what it does
or tries to do, or in the code techniques
it uses.

This is the kind of thing that's sometimes
called "more" or "also interesting" or
"see also".  A name such as those can
invite serendipitous exploring and perusal,
as opposed to stigmatizing the library as
uninteresting, nonworking, or depassee -
"obsolete".

(Just one opinion.)