Smalltalk on the small end (was: Advice requested: GUI project beginning)

[John Duncan]

> The point is that there's little point to "scripting" in Smalltalk
> when it doesn't really get powerful 'til you build up a rich set of
> functionality in a "Smalltalk Environment."  And *that* is something
> you don't want to respawn repeatedly...

Hmm...  This brings up an interesting conflict to the term "general-purpose
programming language."  I have seen this term used quite often, usually
meaning that a language is turing-complete and allows access to system
structures.  But when it comes down to it, there are no general-purpose
languages.  The market thinks so, but anyone who programs in Smalltalk knows
better.  As Squeak evolves, people often ask, "Can Squeak do this feature of
some other system?"  The Squeak team says, "No, but no-one's stopping you."
If the person goes ahead and does it, something like implementing native
widgets or scripting support for a certain OS, the response, even by the
implementor, is usually, "Why did I care in the first place?"  Simply put,
it's much easier to let environments that are good at one thing do that
thing, rather than trying to make a single environment do everything.  For
the same reason, I don't understand why people still write user interfaces
in C++.  Or even Java for that matter.  The tools are simply not advanced
enough to do it well.  But, you do it in ST, and you have something quick
and elegant.  Similarly, you have people trying to make Smalltalk do hard
realtime systems.  Why?  To me, this is kind of like implementing Linux on
an Atari 800.  It has geek value but there's still a much better tool for
the job.

-John