danielv at netvision.net.il
Mon Dec 2 17:51:43 UTC 2002
I agree with this point and the wish for a higher abstraction (maybe one
that allows one to instantiate a complete systems of interoperating
objects), but I think patterns books in a way would still be useful.
Once, because I think the most important aspect of the patterns books is
that they present a very condensed source of design problems and their
applicability. Not the solutions matter, but the problems. I think I
didn't understand the tradeoffs implicit in reuse via inheritence, for
example, until I read the GoF book. So assuming most programmers aren't
born with this knowledge installed, awareness is already 90% of the
distance to better code. Whether the book then points you at schemas for
manual generation of a family of good solutions, or to a library solving
it that's perfect and tested is a matter of degree. The large risk
(coders that are amateur designers thinking they're inventing the wheel
- for the first time!) is solved.
Twice, because an important thing that makes artefacts live is that they
be modular. Pattern languages aren't the answer, they are a crutch we
should use to make this understanding concrete and widespread. Whether
pattern books will still be useful as teaching aids when the knowledge
is common is an open question.
Christopher Alexanders books on patterns, including his early "Notes on
the synthesis of form" address these topics and are very interesting and
Marcel Weiher <marcel at metaobject.com> wrote:
> In an ideal world / programming language,
> there would be no patterns books.
> Marcel Weiher Metaobject Software Technologies
> marcel at metaobject.com www.metaobject.com
> Metaprogramming for the Graphic Arts. HOM, IDEAs, MetaAd etc.
More information about the Squeak-dev