🚀 go-pugleaf

Thomas M. Breuel writes:

> ...
> and Prolog may be well suited to writing "meta-interpretors" (sic), 
> ...
> The advantages each of these languages have for each of these domains
> lies in the standard data types, libraries, and programming
> environment available for them. As languages, however, they are not
> very "expressive", lacking variously important features like closures,
> continuations, static type systems, and module systems.
> ...

I could not resist when I saw the quote about Prolog.
If even people with your background (Institut Dalle Molle d'Intelligence
Artificielle Perceptive) think Prolog MAY be suited for meta-interpretors
(ONLY seems implied), something is quite wrong.

Meta-interpreters is the best way to let Prolog look bad.
Source-to-source transformations are made real easy for the same reasons
meta-interpreters are easy, but are much more valuable: it is very easy
to write pre-compilers etc. in Prolog.

When you quote the lacking features, you are missing something too.
All commercial systems have had module systems since long. You might have
tried an old PD version (C-Prolog ?). As for continuations, you will
be able to find a complete implementation of Prolog based on continuations,
and other papers about continuations and Prolog.
The lack of type systems is often referred to as the advantage of dynamic
typing. A lot of effort goes into type checking and type derivation
for performance and correctness reasons too.

Your best argument in favor of Prolog is the quote:
> ...
> The
> quickest, cheapest, and most practical way to get a programming job
> done is usually not by writing the shortest program, but by writing
> clear, unsurprising, reusable, and debuggable code.
> ...

In IMHO, far to many people criticise languages and start rolling their
own-is-better (TM :-) language before having gone to the limit to really
find out about what exists.

Andre' Marien
bimandre@cs.kuleuven.ac.be