Re: [stack] a concatenative language for the Semantic Web

Joe Bowbeer <[email protected]> wrote:
> William Tanksley, Jr <[email protected]> wrote:
> > > Good point about the stack.  Do your code fragments stand alone, or do
> > > they
> > > communicate via the stack(s)?  (If they stand alone, then an enclosing
> > > "let" expression could be used in Lisp to define temporary variables.)

> > "let" defines bound variables; he's working point-free, which means he
> > doesn't use them.

> Thanks.  I get that.  It's as clear to me as if someone wrote "A hammer
> drives nails, but he's using a screw driver, so he doesn't need nails."
> My question "Do the code fragments stand alone?" is more at "Why use a
> screwdriver?"

Let me quote from his paper:

"Broadly speaking, the goal of this project is to demonstrate that:
   1. linked programs have all of the advantages of generic linked data
   2. a stack language is like a path language, only better
   3. for most linked data purposes, forward traversal is all you need"

It could be stated that his goal is to prove by example that a
screwdriver is better than a hammer for driving screws :-). But enough
of that analogy. On to specifics.

His design is, from the ground up (and in the first paragraph of his
paper, which I recommend in entirety, BTW) described as taking a
"multivalued, pipeline" approach. Scheme, for all its excellent
qualities, is not a dataflow language, and is thus not suited for a
pipeline approach. Adding in the idea of multiple values (really a
clever innovation -- I don't know of anyone else who's done that)
makes bound variables even more unlikely to work (one could use
generators, like Icon, but that adds a deep nest of unusual
semantics).

So his idea is basically and fundamentally different from anything
Scheme could possibly support. Not a slam against Scheme. He just
needed something different in order to prove his point.

And his result is indeed very interesting, wouldn't you say?

I find Ripple to be in the top 3 interesting concatenative languages.
I don't know anything about RDF/Semantic Web, but you bet I'll be
learning about it now.

> --Joe

-Billy