Martin, > Furthermore, your idea of using lisp for everything sounds quite > dangerous to me. It sounds like: I don't really know what I'm going > to do, but I'm going to use lisp.
Well I don't know why it sounds "dangerous". I can't claim that I know "how" to achieve all that I want to achieve in the crystal idea. However, I do have experience in developing and using large, complex programs in Lisp (e.g Axiom). I also authored a large lisp program that was both a semantic network and a rule-based program called KROPS that was used as the knowledge representation language for IBM's expert system called FAME. I want to use this kind of technology to provide intelligence behind Axiom. KROPS depends fundamentally on the mirror image that program representation == data representation. I don't know how to write such self-modifying learning programs in C, Haskell, or any other language. But self-modification is fundamental to the learning aspect and the semantic/rule duality. And learning is fundamental to crystal. Based on my lisp background I see the great advantage that I can create s-expressions "on the fly", store them in files, transmit them, modify them, and evaluate them. Doing this in any other language requires a pile of parsers everywhere. In another project I'm doing we have several parsers (XML, and several "intermediate languages"). They aren't hard to write but each one uses a different surface syntax with slightly different semantics. Each one is more unnecessary machinery. Each one is sliding down the slope of "optimizing", trying to generate "better" code, adding "types", working on bringing in more compiler technology. Each little language has to duplicate the function of the other languages (e.g. storing symbols). Each adds maintenance overhead. Using lisp allows intermediate languages but they are just "more lisp" so it is trivial to manipulate, store, transmit, modify, and evaluate. But this all quickly devolves into language wars and religious debate. > And replacing LaTeX with some lisp typesetter is, in my opinion, the most > stupid thing that could happen to axiom. Eh? I never suggested this. I would prefer that the browser "spoke" latex so that I could use it in conjunction with html. But that's a project for someone else. Perhaps some bright spot will make a Tex4ht "plugin". On the other hand I see the browser as an "input" program as well as a display program so I'd like to see it generate information in a format that Axiom can use. That is, I'd like to see equations (mathml?), s-expressions (e.g. graph structures) and latex pamphlets coming back to axiom from the browser front end. This requires a decent programming language in the browser. t _______________________________________________ Axiom-developer mailing list Axiom-developer@nongnu.org http://lists.nongnu.org/mailman/listinfo/axiom-developer
