Why isn't Smalltalk more popular?

Les Tyrrell tyrrell at canis.uiuc.edu
Wed Jul 14 16:14:37 UTC 1999


Scott Norman wrote:

> There seems to be a lot of material on the Net about how to use Smalltalk, the 
benefits of Smalltalk, the history of Smalltalk, etc.  But there doesn't seem to 
be much of anything about the specific failures of C++ and Java relative to 
Smalltalk.  Nor does there seem to be anything explaining why, given Smalltalk's 
obvious superiority, it is not widely used or even known about.  Has anything 
much on these subjects been written and posted to the Net?

-----

For years and years and years there has been plenty of debate in the newsgroups
about the relative merits of any given language/environment.  There have always
been certain classes of people for whom there was only one figure of merit,
the holy word "performance", ill-defined as it may be.  Typically they meant
it in the sense of crunching lots of numbers real fast, or drawing lots of
polygons fast, that sort of thing.  Nothing else mattered.  Sometimes this
was a legitimate concern, as their work truly did depend upon doing those things
as fast as possible, and did choose the correct language for doing them. Other
times, their work did not depend upon doing these sorts of things, but still
they felt that "performance" was the one true god.

Smalltalkers, for a good long time, have maintained that a really good use of
CPU time is for the MACHINE to do the drudge work, and have the programming
environment be as dynamically fluid and accessible as possible, for their figure
of merit was not "performance" in the sense of number crunching but instead in
the sense of getting at a given problem's throat and killing it.  Everything
that gets in the way of this process is considered to be crufty and without
merit.

As it turns out, for the things that I have done in Smalltalk, it has always
performed admirably in the second sense, and within acceptable bounds for the
first.  I have never seen a system that achieved the first that also achieved
the second.  One of my long, long, over-the-horizon dreams for Smalltalk is
that it be able to support "exceptional performance" in both senses, and I
believe that there are ways to augment the environment to do just that.

That's my take on the "performance" issue.

les





More information about the Squeak-dev mailing list