Loading ...
Sorry, an error occurred while loading the content.

270RE: [feyerabend-project] modernism/postmodernism, was Re: CL and modernism

Expand Messages
  • Mike Beedle
    Sep 21, 2002
      Patrick Logan wrote:
      > >> I like to think of ZetaLisp, CL and
      > Smalltalk as representing a
      > bare-maximums that work to make me
      > productive in building interesting
      > or complex
      > systems. I find those bare-minimum
      > languages to be like living in a
      > tent, something that can be fun for a day
      > or two but not a serious
      > place to live. I find
      > that working with a collection of little
      > languages and tools is
      > somewhat like living in a tent city. <<

      Patrick wrote:

      Yep, sometime the tent just comes down in the middle
      of a storm.

      I recently gave up the prescriptive Java/XML
      implementation of a WFMC (workflow management coalition)
      compliant workflow management system (because of
      language and paradigm limitations), and turned the
      workflow system into a declarative CLIPS implementation
      using Jess. (Our goal is to eventually provide the
      implementation in SweetJess, Jess's XML cousin that uses
      DAML+OIL and Rule ML so that we can program in "XML" and
      satisfy what corporate America wants to hear:

      "It is in Java and in XML"

      we'll give them SweetJess :-) which is of course a
      logical-functional Trojan horse.

      Our code in Jess is about 1/5 of what it used to be and
      I can do things that are nearly impossible using
      the prescriptive and XML forms.

      Did I mention that it is also mobile, concurrent safe,
      and changeable at runtime (dynamic rules, functions, facts
      created on the fly)?

      The next release of Jess will add Lisp-like macros ....

      Btw, does anyone know of a native CLIPS implementation in
      Lisp? (Currently I added a PROLOG library to my lisp but
      it doesn't quite feel like CLIPS.)

      Patrick Logan wrote:
      > I think this is why Scheme in a JVM works
      > so well for me. The JVM enables a very
      > feature rich environment and Scheme enables
      > a very powerful way to access the
      > environment. This feels "bare maximum" in
      > the sense that a Lisp machine used to feel
      > that way, plus it grows with every useful
      > Java class I find on the net. Kind of a
      > dessert topping *and* a floor wax.

      Scheme is definitely a good place to be. In my case
      I needed features that were more akin to Curry (the
      logical Haskell cousin). In fact, in the future
      we are considering using the Xurry virtual machine (a logical-functional
      virtual machine that sits on top of Java).

      Some rough estimates indicate that our Xurry code will
      be between 1/5 and 1/10 of what our original wfmc
      prescriptive code used to look like,

      - Mike
    • Show all 5 messages in this topic