Re: [stack] Cat article submission to Doctor Dobbs Journal

On Feb 10, 2008 12:25 PM, William Tanksley, Jr <[email protected]> wrote:
> Christopher Diggins <[email protected]> wrote:
> > > Now I find a real disagreement. First-class functions aren't at all
> > > part of the definition of "concatenative"
>
> > My bad, I didn't mean to imply that. They are key to why Joy is so
> > interesting (to me at least) though.
>
> Definitely. You're right that they belong in the intro.
>
>
> > > nor are branch statements
> > > antithetical to concatenativity.
>
> > I really didn't mean to imply that one neither. My point was that the
> > fact that Manfred leaving branch statements (by which I mean goto
> > statements) out was significant (is that contentious?). I wasnt trying
> > to say anything there about concatenative languages in general. I've
> > tried to clarify.
>
> What previous concatenative language had GOTO statements?

Forth has BRANCH and BRANCH? Do you consider Forth a concatenative
language? Either way I am talking explicitly about stack-based
languages so Forth definitely has to be considered.

>  I don't
> think Manfred "left them out"... They're a little hard to put in, so
> they were rare at best in other concatenative languages. I assumed you
> were talking about special syntax for conditionals, which Manfred
> removed the need for when he supplied function literals.

I consider conditionals (and other control flow constructs) as simply
syntactic sugar for branch statements. However, I will try to make it
less contentious, and my intention more clear.

> > > A possible replacement:
>
> > > "Manfred's primary innovation with Joy was to introduce first class
> > > functions and function literals (called quotations). The resulting
> > > language shares many of the advantages of functional languages (e.g.
> > > it is easy to reason about) and is particularly expressive."
>
> > I like that, but the goto removal is significant, and should be left
> > because with gotos you wouldn't have the nice rewriting properties
> > that you have now.
>
> Good point... But that's because of the removal of syntax (or the
> regularization of syntax), not so much because of the removal of GOTOs
> (which are merely a particular type of syntax, and one which most
> concatenative languages don't have anyhow).

You keep saying concatenative languages, and I am talking about
stack-based languages. I am still not sure which languages are
technically concatenative (is Forth? is JVML? is PostScript?). I am
referring to things like MSIL, JVML, and Forth all of which have
explicit gotos (i.e. branches / jumps / conditionals / etc.).
Historically branching is part of stack
languages, because they lacked higher-order functions. However
PostScript breaks this rule, and I am concerned that I may be beging
inaccurate when we consider it. It seems that PostScript does have a
notion of function literals before Joy, so I have to be careful how I
refer to the contributions of Joy.

> > > Unfortunately, this removes all mention of "concatenative" from your
> > > article, but since you didn't really discuss it I don't think it's a
> > > loss. I do think you could be more precise than "is particularly
> > > expressive", though... I'm not sure what you mean by it, although it
> > > looks like a compliment :-).
>
> > By expressive I meant that we can say a lot with a little. (e.g. a few
> > instructions can convey a sophisticated concept). I've tried to
> > rephrase.
>
> Cool.
>
>
> > I look forward to more of your comments, thanks!
>
> Having skimmed the rest, I like what you have from there on... I could
> hunt for more nits to pick, but don't have the time and might not find
> any anyhow.

Thank you very much!!

- Christopher