On Sep 25, 2004, at 11:34 PM, Andreas Raab wrote:

> Hi Avi,
>
> I hope I haven't wrecked total havoc in the repository ;-) But I 
> wanted the approach to give a real workout to see what happens. See 
> http://people.squeakfoundation.org/article/39.html#1 for my analysis.

Yes, I think that was exactly the right thing to do.  That was the 
point of this experiment - to discover the problems like those you 
outline.  It's no fun if it works right off the bat :).

In the long run, I think all the problems you bring up are solvable 
within the framework of MC - essentially, I think problem #1 
(serialization) and problem #3 (dynamic modification) are the same 
problem, in that if you can serialize changes and particularly 
initializations you should be able to do anything a changeset can do.  
In the short term, though, I think it's also useful to continue the 
experiment even if it *can't* do everything the update stream can yet - 
so when we go live again, maybe we should hold off on Diego's look 
changes :).  And hopefully I can do something about problems 1 and 2 in 
the meantime.

Avi