Re: Stable releases

[Neil Jerram]

address@hidden (Ludovic Courtès) writes:

> I share your concerns about having a really stable series, where new
> releases can be made with minimal overhead.
>
> That said, I'm afraid adopting a strict stable policy might have
> undesirable side-effects.  In particular, it might be the case that
> either users would end up always using the unstable series (because they
> don't want to wait for one year to get some new tiny feature), or we
> would end up creating new stable series so often that they'd be really
> unstable (I think the former is more or less what happens with Debian).

Well we have always had a strict stable policy until very recently, so
there should already be evidence one way or the other.  I don't have
any numbers, but I am pretty sure (anecdotally) that we have had most
users sticking to the stable releases, and a smaller number going
unstable by using either CVS or the nightly snapshots.

That sounds fine to me.  The point is that people know what their
choice means and so set their expectations accordingly.  Under my
proposal - i.e. strict stable policy + unstable releases - the only
thing that would change is that it would be easier for the more
experimental users to get at the unstable code.

If your main concern is getting new stuff out to the users who want to
experiment with it, I would have thought that making unstable releases
would meet that concern.  Does it?  As I asked before: is there some
way that we can specify when or how we would make an unstable release,
that would give you enough confidence about new stuff being made
available?

If we went with your preference - i.e. allowing Scheme-level and
certain C-level enhancements into 1.8.x - I suspect that before long
we would be asked to reinvent the concept of a strictly stable 1.8.1.x
series.  We'd then end up in the same state as I'm proposing, but with
less obvious numbering.

Regards,
     Neil