This was a good read!  I completely agree that focussing on the
social structures is the best way to move forward, and also that we
should pay attention to what others are coming up with.  There are
plenty of tools around, and we can make anything that we need anyway.

Along these lines, we should think hard about what our actual
communities are doing and desire.  In particular, please notice that
a lot of Squeak users are not tracking the "standard" Squeak
release at this point, but have started with 3.0 (or earlier!) and
gone on their own way.  THIS IS NOT BAD.  Or at least, we cannot
afford to simply say all these guys suck and need to change pronto.
We need to come up with a community structure that supports our
current disparate members, even if we want to change this over time.


One of your ideas I find particularly helpful is to consider that
putting
a package into any individual distribution requires some amount of
extra work.  Each distribution will have its own standards, all of
which are more than zero.

This does not mean, however, that people *must* be responsible.  It is
simply that they must be responsible if they want their code to go into
the particular distribution they are interested in.  There is no need to
put guilt trips on people.  Instead, let us set up standards, and then
rely on people's own interests about whether it is worthwhile meeting
those standards or not.  Keep in mind that people *do* have pressure to
get things into the centralized distribution(s) even if they are
completely self-interested.  Anyone who has tried to maintain patches
while the upstream source is being modified, can tell you all about how
un-pleasant that is.

Another interesting aspect of having multiple distributions is that it
removes some pressure from the gatekeepers in the "standard"
distribution.  If they think of themselves as making "the" Squeak, then
they are under a ton of pressure to make every decision right, and this
inevitable slows down the acceptance rate of good-looking patches.  If
they think of themselves as making *one* good Squeak distribution, then
the standards are different.  It becomes easier to allow for mistakes,
because users who cannot tolerate the mistakes will be using a different
distribution anyway.

I picture this multiple distributions approach as being an excellent way
to both support our current community and to support any efforts to
migrate that community in desired directions.  It would be nice, by
itself, to go to some Squeak site and then be able to tour around the
various sub-communities.  And there are useful tools for migration being
talked about: let packages be shared among communities, let them be
added to distributions when they have met the distribution's policies,
and do let distributions have policies attached to them.


One aspect that bears some investigation is these "workflow" tools.  I
would be ecstatic to have a good Squeaky bug tracker, but the ToxicFarms
guys seem to think workflow tools are better than mere bug trackers.  I
know nothing about them so cannot comment.



On a small matter, it seems reasonable to have more than one
image per distribution, so let's not *quite* identify images and
distributions
one-to-one.  A classic example is 3.6 basic versus 3.6 full: these both
should draw from the same set uf packages.


Finally, your "distribution" seems exactly what I mean by a "package
universe", but I'm not sure if it is exactly what Linux people use it
as.  For whatever it's worth.  I'm perfectly fine with "distribution" so
long as we keep it straight.


Lex


PS -- let's not forget to update the web sites with any updated
community viewpoint we come up with after these discussions.  And of
course, these thoughts should actually be run by some of the
non-squeak-dev Squeakers at some point!!