[Squeakfoundation]An architecture for sustainable Squeaking
Sat, 02 Jun 2001 18:06:32 +0200
You're right, modularity is not everything, but it's a big step in the
right direction. If you want to check out some good reference sources,
start out by reading
"An Overview of Modular Smalltalk" by Alan Wirfs-Brock & Brian Wilkerson, and
"Orwell - A Configuration Management System for Team Programming" by Dave
Thomas & Kent Johnson
both in the OOPSLA '88 proceedings.
Alan Wirfs-Brock has done some very important work in this direction. You
can find more of his papers at http://www.instantiations.com/sts/pubs.htm
More recent work can be found in Gamma et al.'s Paper on TeamStreams,
presented at XP 2000 (I'm not sure whether there's a web link to it), and
(of course <grin>) "Mastering ENVY/Developer" by Pelrine, Knight and Cho,
available at any of the rare bookstores that still sell Smalltalk books.
Most of the work I'm doing for Stable Squeak is involved in implementing
these and other ideas aimed at managing modularity in Squeak. I'm on a roll
now, and may be able to show something next week.
Of course, full namespace support - including method-level namespaces -
would help a lot of things ;-)
On the other hand, code quality can't be learned from a book. I benefited a
lot from having had access to world-class Smalltalkers who would whack me
upside the head if my code was crappy. It became a matter of pride for me
to write clean code. If you're alone, SUnit and full-strength SmallLint
help a lot.
At 16:43 02.06.2001 , firstname.lastname@example.org wrote:
>Hi, Joseph. Glad you decided to stay on this list.
>Joseph Pelrine <email@example.com> wrote:
> > Quoting you - ("Sorry, newcomer X, but your code's just too shaky, we'll
> > have to pass on that for now")
> > Please start by reading the numerous writings on modularity in Smalltalk
> > before posting things like this.
>Modularity is only part of the story here. I just posted an example
>describing problems integrating a goodie I wrote. This Goodie isn't even
>worthy of a module, it's just a few changes needed to several places
>that belong in different modules. So if the goodie isn't accepted into
>the various modules, it's dead as a nail. Getting accepted requires
>gatekeepers that can say "Well, Daniel, that thing just isn't getting
>into my module without SUnit tests and removing the MethodSubstitute
>specific stuff from PluggableListMorph".
>Modularity is a neccessary but insufficient condition to letting people
>contribute things effectively.
>(As your keep-off-my-mind-meld posts show you very well know...)
> > Joseph
>Squeakfoundation mailing list
- Joseph Pelrine [ | ]
Smalltalk - scene and not herd!