A 'graveyard' for packages / unloved packages

Hey all,
So, we're one day going to have pyrus / pear2 as 'stable' and the preferred
method for installing packages.

Sadly, some code won't make the cut. An example of this is SQL_Parser.

Apparently, the entire premise is flawed and needs to be completely
re-written - Sebastian gave it a shot, but couldn't make the concept work
(or so I hear - speak up if that's wrong).

IMO: there are two viewpoints here:

   - Maintain! BC! For! Ever!
   - Change/Break everything, god bless unit tests!

Extreme 1:
We never, ever make code we've previously distributed unavailable.

This means that we'll never break someone's application; we avoid upsetting
the 0.04% or whatever it is of PHP4 users, etc.

Extreme 2:
We stampede around, killing of code ("Refactor mercilessly"); screw the
consequences.



So; my question for us all is what's the approach we take when we make the
cutover to pear2?
Do we just do #2; which is the most liberating; or do we take great efforts
to port things forward (most stable)?

What I would like to see is a middle ground. In this world:
 * package 1.0 is dead - no packages remain which are 1.0.
 * PHP 4 is dead too - pyrus is laughing at it's grave
 * You can still get code and install from (somewhere) even if civilisation
has moved on
 * Non stable and PHP4? -> Swap to PHP5 is acceptable BC breakage (current
practice)

To accomplish this, I'd like to introduce a concept or practice of
graveyarding a pear package - shifting it to its own pear channel (till
suggests pearfarm or pearhub); doing the appropriate end of life
announcements, and booting it the heck out of SVN / pear itself.

SQL_Parser is a good example of the kind of package to target - it's unloved
and unworkable at the moment -> it can be graveyarded safely.

The criteria I imagine for graveyarding a package:
 * PHP 4
 * Relatively few downloads
 * Unmaintained
 * package 1.0
 * No one wanting to bring it across to pear2/PHP5 goodness


Additionally, I'd love to get updated reporting on PHP4 usage stats (I think
we're about 2 years old on our opinions). Hopefully, with these stats, this
makes us all relax somewhat about that style of BC break.

Collective thoughts, PEAR?
My intention is to trial this with SQL_Parser, and shifting it off to
another provider by the end of the weekend if no one objects.