Well, having foolishly participated in the recent little brouhaha (no
pun intended) over this issue, I would like to wholeheartedly concur
with Roger here.  Also, I think this gets back to addressing Andrew's
original issue of forking.

And yes, it would be great fun to debate where to draw the SBP line.
I could certainly see reasonable arguments for places other than
Squeak 2.8.

				-Dean


-----Original Message-----
From: Roger Vossler [mailto:rvossler at qwest.net]
Sent: Friday, November 01, 2002 6:49 PM
To: squeak-dev at lists.squeakfoundation.org
Subject: Re: An uncomfortable question


Hi Gang,

About 18 months ago, I pointed out to this list that modules, or 
components, are
a complex and very difficult problem. I was ignored. Recently, a number 
of folks
on this list have come to the hot conclusion that modules, or 
components, are a
complex and very difficult problem. What a surprise. So, just this 
once, I'm going
to say, "I told you so." :-)

Before anyone gets serious about modules, or components, they must deal 
with
issues of architecture and interfaces before they get down into the 
design and
implementation details of modularity, namespaces, interdependencies, et 
al.
While there have been a few references to "architecture" and 
"interfaces" on the
list over the past 18 months, the focus has been on design and 
implementation,
which puts the cart squarely before the horse.

Having said all of this, where do we go from here? I have a few 
suggestions:

1. Squeak 3.3 is a dead duck which should be set aside as: (1) a source 
of ideas
for any future work on modules, (2) a monument to ambition over common 
sense
engineering, and/or (3) a mechanism for torturing graduate students 
(re: thesis
advisor to PhD candidate says, "Here. Take this code and make it work".)

2. Squeak 3.3 should be salvaged for anything that can be backported to 
3.2. Some
of this has already happened.

3. Continue the Squeak development stream starting with Squeak3.2-4956, 
which
appears to be the last "stable" version. Add the Squeak Map and related 
stuff
(Vainsencher, et al). Add the VI4 stuff (Hannan, et al). Package the 
results of
all this as Squeak 4.0 and declare victory.

4. Squeak 4.0 should become the Squeak upper bound while Squeak 2.8, or
Stable Squeak, should become the Squeak lower bound. Somewhere between 
the
upper bound and the lower bound, define a Squeak Base Platform (SBP). 
The SBP
becomes a thoroughly debugged and documented platform from which the 
fans
of modules can work as well as a platform for other folks who might 
want to use
Squeak for something useful. We can have lots of fun fighting over 
where to draw
the SBP baseline.

I am serious about this. Squeak 2.8 is more than enough for me to 
handle. Squeak 3.2
is about three (3) times the size of Squeak 2.8. Whatever happened to 
that "exquisite
personal computing environment that one person can comprehend"? With or 
without
modules, nobody is capable of fighting their way through hundreds of 
thousands of
lines of code. Apparently, this fact has torpedoed the various module 
efforts to date.

Well, this is more than enough for now.

Cheers, Roger.....