Lisa Dusseault wrote:
> Lisp may make better programmers, but that doesn't necessarily mean it makes
> better programs.
>
> To take an example from Shaolin Wooden Man (I love martial arts movies,
> particularly Jackie Chan), you can become a better fighter by walking around
> in lead shoes & carrying water up stairs in buckets with holes that make you
> run to get to the top before all the water runs out. But when the training
> ends, you lose those restraints and take the best tools you have, and the
> strength or technique you've built up is there for you.
I think this analogy is very broken. Lisp doesn't make better programmers by
artificially handicapping them and forcing them to work around / through these
handicaps; rather, just the opposite. It provides an extremely cogent and
high-level language environment for doing programming in the most abstract sense
possible. It teaches programmers to think abstractly about problems and the
task of programming itself.
When I think about algorithms, I tend to think in something like Scheme with
some more overt ZF / ZFC set theory thrown in. I really believe Lisp / Scheme
is the best tool out there for algorithm design, and certainly the best for
general metaprogramming tasks. Unf., it's *never* the best choice for a
commercial endeavor, for entirely nontechnical reasons. James' Ariba comment is
sadly par for the course... I've been involved in three projects where the
initial work was done in a Lisp but which were forced to migrate to C/C++ for
business reasons. Seen many, many others. The general attitude among
technology marketers, investors, and so on seems to be that if you're writing in
a Lisp, you're doing a "science fair project" rather than building a business.
jb
This archive was generated by hypermail 2b29 : Sun Apr 29 2001 - 20:26:19 PDT