[Squeakfoundation]A Possible Process/Product for the Squeak Foundation
Fri, 11 May 2001 15:10:20 -0700
>I'm hoping by the weekend to put out a list of projects and processes that=
I can imagine taking place between SqF and SqC in the next 6 months to a=
year. This may seem out of sequence in the chaordic plan, but my hope is=
to load our cache with some of the situations we (at least I) will want to=
deal with, so that the orderly plan goes forward with these in mind.
(Oh, you thought I meant last weekend? ;-)
I want to start by acknowledging John Sarkela for undertaking to test what=
could be a very workable Squeak Foundation release process. And Dave=
Thomas for encouraging him to do so.
What John did was to dismantle the current release, parition it into various=
packages to a certain extent, make some important replacements=
(unrestricted fonts, for instance), and get the whole thing working again. =
You have to pick a fixed point for this, and you can't have a bunch of=
other people participating very easily (though this should change as we=
strengthen the partitioning into real name spaces).
This is what SqC should do at each release, but neither do we have the time=
nor the focus (we've got focus, but it's usually elsewhere). Therefore I=
really think it's something SqF should do. Here's my hope for how Squeak=
releases might get made once SqF is underway:
The base image is nicely built up from modules
Starts as a micro kernel suitable for building simple headless apps.
Adds enough programming environment to have fun on PDA's
(like present majorShrink but less cruft)
Adds a package mechanism
Adds all the stuff in 3.1 as packages corresponding
roughly to the major system categories
All the SqD's (Squeak developers, of which SqC is one one) contribute the
latest version of their work in the form of one or more
packages as phase 1 of a release.
As part of each major release, SqF undertakes the
process of integrating these as best they can,
writing or suggesting more documentation,
and sending tentative releases back to the SqD's in
some sort of alpha, beta, gamma test cycle.
At the end of this cycle, SqF puts out any number of release
images (probably a tiny, and a jumbo at least),
and all the final packages as approved by the SqD's.
Also a set of VMs for each platform of significance.
(Some of these may be assembled elsewhere, but
SqF is the clearinghouse for shoppers)
Each of the SqD's may also maintain and distribute their own
possibly more current versions of special-interest
images (Siren is an existing example of this).
It's not rocket science. In fact I anticipate a certain amount of anti-pink=
backlash about this rather mundane vision. But here's my feeling about=
that. It's much easier to convene an organization around well-specified=
stuff that has to get done, than it is around a bunch of blue-sky research=
ideas. Especially if you have no funding.
This doesn't mean giving up on blue-plane progress. This community is full=
of good people with good ideas for cool directions to go with Squeak. =
Think of the pink plane as soil and fertilizer. If we do our job on the=
pink plane, a thousand blue flowers will bloom there next year.
And if there is a decent open process in place (the above is more open than=
what SqC now does), then the mainline Squeak will follow the blue successes=
faster than it does now. That would be a Good Thing.