> A couple years ago, I saw Alan Wirfs-Brock give a demo of an enhanced Visual
> Smalltalk. The development image and the execution image were separated
> (namspace also).  This had a couple of benefits, the most obvious to me was
> not having to build a runtime image iteratively to figure out that you were
> inadvertently using development methods.

Search for Modular Smalltalk- this was first proposed about 12 years ago
as a means of providing a fully declarative Smalltalk.  Everything about
a 'program' was declared, not the result of manipulating the environment.
Alan Wirfs-Brock presented it in an OOPSLA paper, roughly 1986.  There
are references to Tektronics tech reports, indicating that there was an
implementation effort there ( I would assume with Tektronics Smalltalk as
a starting point ? ), and there was a Canadian graduate student who had
apparently implemented it for a doctoral dissertation.

I've mentioned all this before... are there any search engines running
on the mail archives?

les