Jim-

I liked your post; it made me think more clearly on these issues.

Jim Benson wrote:
> Let's use Squeak to make something better. 
> Use Squeak as a bootstrap to
> build a worthy successor. 

This is one thing that keeps nagging me. It is one thing that prevents
me from investing in making Squeak as is better. There are many new/old
ideas out there I woudl like to have available for use such as:
  Prototypes (like Self)
  Universal VM for multiple languages, or at least multiple VMs
  Remote debugging, browsing and inspecting
  Code in a database (something like Les Tyrrell's Oasis)
  Platform-independent C API layer
  Indentational Syntax (like Python) and other syntaxes
  Literate programming (documentation, data, and code mixed together)
  Threaded code generation (or compiled) for speed
  Enhanced Modularity (objects know their "module" and "world" with
other reserved words?)
  Relational programming (i.e. relations between objects as first class
objects vs. having instance variables)
  Integrated communication tools for team programming
  Being a knowledge repository
  Others...

Of all these, for me, the issue of modularity -- and the related fact
that an object is only meaningful in the context of other objects -- is
the deepest, and I think the system should somehow reflect that.
Smalltalk as a language fails to do this (despite its other strengths,
some of which result in part from this failing, i.e. consistent 
browsing and inspecting is easy if there are no boundaries). Seperating
code into loadable "modules" or having "namespaces" is just the surface
of this deeper issue.

These issues always tempt me when I start thinking about making Squeak
better in a production sense (producing essentially at best a slower
VisualWorks with a funky GUI). There were two Camps in the original
Squeak BOF at OOPSLA in 1997. One camp wanted something new; the other
camp wanted a stable open source Smalltalk. To an extent, neither of
these visions have been realized yet by Squeak, and they are even
frustrating each other! I keep hoping the solution is some sort of
really great system everybody agrees is both easier to use for
production and really innovative, but that hasn't happened yet (although
there have been glimpses of it occasionally on this list.) As a
practical matter, at best we will see is probably a family of somewhat
related efforts.

I think there is value in essentially a slower VisualWorks that is open
source, community maintained, and extensible (I'd use it right now for a
couple of things) which is where "Stable Squeak" seems to be heading at
the moment. But you raise the larger issues. Thanks. Reconciling and
balancing these issues into some sort of sensible development directions
is difficult.

-Paul Fernhout
Kurtz-Fernhout Software 
=========================================================
Developers of custom software and educational simulations
Creators of the Garden with Insight(TM) garden simulator
http://www.kurtz-fernhout.com