I agree heartily. I would greatly love to see equal (ok, *way more*
;)) effort put into producing tests for the base classes, as producing
documentation (e.g., class comments).

I do confess, however, although having read Kent Beck's testing paper, to
not having a very good sense of how to go about devising unit tests. I'm
hoping that both the Squeak Applications & Community book's chapter on XP,
and the book, Extreme Programming Installed (advance draft available on
the xprogramming.org site) will prove enlightening on these matters.

Not to mention the ANSI unit tests that Camp Smalltalk folk (applause
for Ralph Johnson) are producing.

To the end of getting more testing, I again urge the inclusion of the
SUnit framework into the main distribution. While I strongly (and have
used) the two versions which antedate the Camp Smalltalk effort, I must
push in favor of the Camp Smalltalk variant:

	http://wiki.cs.uiuc.edu/CampSmalltalk/SUnit+Camp+Smalltalk
and
	http://ansi-st-tests.sourceforge.net/SUnit.html
		(Where the code may be gotten.)

as it is being developed simultaneous for every Smalltalk under the sun,
and a few under the moon, and which, I believe, the ANSI tests are
targeted.

Cheers,
Bijan Parsia.

On Sat, 3 Jun 2000, Ralph Johnson wrote:

> It is much easier to integrate packages when they have tests.
> File them in, run their tests, and if the tests work then
> you know they work together.  Unless the tests were too weak.
> Tests aren't perfect, but they help a lot.
> 
> -Ralph
> 
>