🚀 go-pugleaf

In article <3880@pasteur.Berkeley.Edu> faustus@ic.Berkeley.EDU (Wayne A. Christopher) writes:
>In article <350@piring.cwi.nl>, guido@cwi.nl (Guido van Rossum) writes:
>> Remember that trees are often used as a way to implement other
>> abstractions: associative arrays, sorted sets with easy insert/
>> retrieve/delete operations, etc.  It might be more appropriate to
>> add such "higher-level" data types to a language than pointers.
>
>The way you would add such data types to a language is to build them out
>of the primitives available, which are pointers (in the case of C), or
>arrays (in the case of APL).  Actually adding them as primitives would
>be a very bad idea -- you could never add enough to make everybody happy.
>
>The sorts of things I use pointers for that I couldn't use arrays for
>are parse trees and graphs (both directed and undirected).

In this particular case, however, I believe Guido speaks from personal
experience.  The language project he was formerly associated with, the ABC
language, developed by Lambert Meertens at the CWI, is designed for novice
users (a replacement for BASIC and the like).  It has exactly five data
types: numbers, texts (strings), compounds (records without field names), 
lists (which are sorted), and tables (associative arrays).  This seems to
be a fairly complete and useful set, particularly for things like graphs
and parse trees.  (I once wrote the LALR sets-of-states construction
algorithm in ABC; it was about two pages of code).

I've never been able to figure out why we think in computer science that
students should have to build their tools before they use them.  Sort of
like having to use an anvil and forge to build your hammers and
screwdrivers before you can learn carpentry.  Now really; wouldn't it make
much more sense to give people a few useful tools to start out with, and
let them get down to solving their problems that much sooner?

--tim budd, oregon state university  (budd@cs.orst.edu)