"Dr.R.F.Wolpert" <wolpert_at_uark.edu> writes:
> >>Also sprach Walter Moreira:
> >Moreover, how big is the scheme community? I think it is small or at least
> >very disperse. This looks very strange for me. If you look at Python, for
> >example, with a clear and simple semantic, close to scheme; why do you think
> >it has a incredible success? Why scheme isn't so successful? Has anybody an
> >answer?
>
> Perhaps an O'Reilly book on scheme would help -- there isn't much in terms of
> readable, enjoyable introductory printed material around. And nice books like
> Harvey and Wright's _Simply Scheme_ need a non-standard startup script,
> which doesn't go down well with all flavours of scheme. (And it isn't in a
> popular series.)
>
> And that might be another problem -- how to choose "your" scheme... there is
> just one Perl and one Python. But we choose between STk, Bigloo, gambit, scm,
> etc. etc. (no particular order, or omission), but I can't use my bigloo script
> without
> some work with stk, or vice versa, for example (and both come from
> "kaolin.unice.fr")
> We all like diversity to some degree, but students and beginners don't.
>
> Having a prominent scheme in Linux, for example, would help too: the gimp uses
> scheme. This should be made better known. (And is another reason for a good
> O'Reilly
> book.)
>
> Let's ask O'Reilly to have a Scheme book!
One thing I've noticed about scheme, and lisp in general, is that
almost all of the books are impractical; on one extreme, there are
R5RS and CLtL, which are semi-incomprehensible to people with years of
non-lisp programming, and on the other are "intro to computer science
using scheme" type books, which tend to be aimed too low for people
who want to use scheme for scripting and whatnot. There are also a
number of books which cover various computer science theory issues
using scheme or lisp, but they tend to be pedantic and impractical for
the uses to which python, perl, tcl (bleh), and the like are applied.
I think it would be wonderful to have an O'Reilly Scheme "cookbook"
with examples of practical scripts and applications, and whatnot. I
also particularly like STk because it tends to have a good set of
macros and the like to make what I consider sensible programming
easy-- a lot of scheme books emphasize using tail-recursion for
iteration, while I find that to be an interesting theory point, but in
practice is an obfuscation. Also, closures, while an interstingly
"deep" subject when doing esoteric things, are almost never explained
in a sensible way--- when talking to a C/perl/... programmer, they can
be described as functions which can reference "more global" variables
in a "smarter" dynamic way, and that's sufficient for most of the
uses. Likewise, lambda expressions are often described via the lambda
calculus, while for practical uses, the main time I ever use them is
for defining functions, including anonymous ones.
I haven't used python enough to compare,
but I found that STk blows tcl/tk out of the water, and is comprable
with but different from perl, perl/tk, and java.
I also find that CLOS and its ilk are *much* better in many regards
than the C++/java/... approaches, since it has RTTI and multiple
polymorphism as "builtins," and is much better about being a good
combination of simple to use but very configurable (in deciding how to
match parameters, etc.).
Another book company to consider, by the way, is Coriolis-- they claim
to have an explicit goal of providing the kinds of books that O'Reilly
has the reputation for. New Riders has published a few that I like, as
well.
So, who's going to write this? After this email, I'm probably
considered too opinionated ( ;-) ), and anyway, I'm kinda flakey.
shrug
- M
--
Mark "Monty" Montague | monty_at_gg.caltech.edu | I don't do Windows(tm)
I'm dubious about any company whose assets can be destroyed by rm -rf
<URL:http://www.gg.caltech.edu/~monty/monty.shtml>
X-PGP-Fingerprint: E4 EA 6D B1 82 46 DB A1 B0 FF 60 B9 F9 5D 5C F7
Received on Thu Nov 30 2000 - 01:30:33 CET