Marcel --

I believe that you have put your finger on the main issue. What we really
want is to grow our understanding and ability to create "art" on the
computer. We want Squeak to change in as many positive directions as
possible, and we're very happy to have such a dynamic and vital group of
colleagues who want to contribute.
     Science works because it allows two very different levels of discourse
to flourish. The first has to be free and nondogmatic in order to let new
ideas be put forth into the public domain and not be crushed by religons
formed about the old. The second level is how ideas should be discussed and
argued about, and this is much tougher about what it considers to be
reasonable forms of discourse. No process has been more fruitful for the
human race over the last several hundred years.
     In applying this to Squeak, it is clear that we must have a very
different structure to what we call the image. We have to allow different
ideas to coexist without hurting each other, as we learn which are the most
fruitful. If the interpreter acts as the second level of science -- in that
it is the tough process about "reasonable argument" -- then a great form of
modularity could be a way to get a free and nondogmatic first level to hold
many kinds of ideas.
     The first good approach to this in a Smalltalk was done by Ira
Goldstein and Danny Bobrow at Xerox PARC in the late 70s and early 80s, and
has been mentioned slightly on this list in the past. We have  been
thinking about ways to do this in an even stronger way and, as you
mentioned, Dan, Ted and the others have started the process of including a
strong modularity mechanism for Squeak. As you noted, this is tremendously
important for the continued growth and change of Squeak.
     In short, it is really important for all to help find the strongest
ways for Squeakers to have their cake and to eat it too!

Cheers to all,

Alan

-------

At 12:44 AM -0800 7/14/99, Marcel Weiher wrote:
>> >If its' utter garbage and crashes the image, then it wont last a
>day. People
>> >will just remove the code.
>>
>> Well, removing code tends to be a touch more difficult than adding.
>
>That's the problem.
>
>I'd suggest (as others have before) that a modularity mechanism is
>urgently needed that allows us to integrate code into images
>non-destructively.
>
>I know that this is a "meta problem" and solving it will distract
>from solving real problems that may be more important now.  However,
>I believe it is an investment that'll be amortized very quickly as it
>rationalizes the development process and takes pressure off squeak
>central.
>
>I *think* that Dan's modularity mechanism provides most but not
>quite all of the mechanism needed.  I am quite willing and eager to
>do the work of extending this to the modularity mechanism I think is
>needed, but I would need significant help.  Even just playing with
>the existing mechanism I managed to crash Squeak in an instant
>without understanding why.
>
>Marcel