incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sam Ruby <>
Subject Re: Starting from the other end
Date Sat, 17 Oct 2015 17:23:09 GMT
On Thu, Oct 15, 2015 at 2:55 PM, Ted Dunning <> wrote:
> We have just had a few interminable threads regarding what is wrong with
> incubator and jumping directly to implementation attempts for changes.
> That isn't the way I would code something complex so it seems to me that by
> analogy we might want to start from the other end and start with what we
> all agree on.
> Building on that agreement and common ground, I think it will be much, much
> easier to determine where we have weaknesses, bottlenecks or limitations
> and also to suggest improvements.
> To do this, I am going to start a google doc to describe WHAT the incubator
> is supposed to do (easier to edit than a wiki) and once that document
> stabilizes a bit will move it to the wiki.
> That document should be able to let us move to similar efforts to describe
> HOW the incubator currently works.  I would hope that one side effect of
> this second stage would be to pull together current documentation on the
> current incubator.
> At that point, we can point out specific improvements and start the real
> work of making things better.
> So here we go. The first step (What Apache Incubator Does) can be edited at

I guess somebody has to go first.

My first impression: this description focus on how and not why, and it
focuses on current (and therefore potentially legacy) processes.
That's not to say it isn't useful, but another view should, in my
opinion, be prepared first or at least in parallel.

I'm growing increasingly enamored of the approach Bertrand (and
others) are advocating: focusing on what the incubator ideally
produces, namely a regular stream of Establish resolutions each
accompanied by documentation assessing to the maturity of the proposed

If that becomes the accepted goal, then as a Director I only need to
care about the end results, and as a mentor, I only need to care about
the community or communities that I am working with.  Beyond that,
different communities could be free to experiment with different

The end result is that the documentation that you (Ted) are producing,
and for that matter the documentation exists on the Incubator web
site, amounts to current best practices.  Which is a good thing, but
perhaps should be considered less binding.

We already have what amounts to three paths: there is the full
Incubation process; there is is IP clearance, and there is straight to
TLP.  We can view these as three points on a spectrum.  Having a well
defined exit criteria and only applying the parts of the process that
were needed to address deficiencies might have helped with projects
like Subversion.

- Sam Ruby

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message