the Spoon module system

Ralph Johnson johnson at cs.uiuc.edu
Thu May 25 11:04:33 UTC 2006


On 5/23/06, Craig Latta <craig at netjam.org> wrote:
>  > It is a very good idea.  A module is defined as all the code that is
>  > needed to run a test suite.  To have a module, you must have tests.  I
>  > like it!
>
>        Cool! I expect that there will be some fiddling with that delineation
> (to deal with conceptual overlap, etc.), but that's more or less the idea.

The more I think about it, the better I like the idea of using tests
to define module boundaries.  A module depends on some other modules,
of course.  The other modules are loaded before the tests are run, so
their code will not be counted among the code of the module.  People
can control the code in a module by changing its tests.  The only
problem would be code that is hard to test with automatic tests, such
as an event loop.  However, it should be easy for a test to
explicitely list methods that should be included even though they are
not called.  Do you have a way of doing this already?

Good code needs tests.  Encouraging people to write tests helps them
to write better code.  So, defining modules as "all the code covered
by a test suit" will make the code better, because it ensures that it
has at least minimal tests.

I also like the idea of implementing a module as the objects (classes
and methods) that are traced during execution.  I was at a workshop a
few weeks back that included both Schemers and Smalltalkers.  One
thing that impressed me was that Schemers think a program is a text
file, while Smalltalkers think a program is a set of objects in
memory.  There was one Lisp guy there, and he was more like
Smalltalkers, because he thought of a program as a set of
S-expressions, and the textual representation of S-expressions in
files was secondary.  Lisp is reflective; Scheme is not.  In a
reflective language, the internal representation of a program has
computational meaning.  So, Spoon is very much in line with the
Smalltalk tradition, but in a way that will seem extreme to
non-Smalltalkers.

Like I said in another message, I think you will have to deal better
with class name clashes because the global variable Smalltalk is
important.  I think this will lead to name spaces in Spoon.  I'll post
another message about that later.

-Ralph



More information about the Squeak-dev mailing list