Re: Market Penetration of FP and Haskell

[S. Alexander Jacobson]

On Tue, 2 Mar 1999, Philip Wadler wrote:
> Bad news: The proposal to split XSL refers to the current working
> draft, which already contains a transformation language (Section 2)
> and a style language (Section 3).  Preliminary versions of the
> transformation language are already in wide use (including Internet
> Explorer 5 by Microsoft; LotusXSL by IBM Alphaworks; and xt, by James
> Clark, editor of the XSL document).  So it's way too late, there is no
> chance of getting the W3C to switch to Haskell.
> 
> Good news: The charter requires that XSL be `a declarative language',
> and in fact the transformation part of XSL is a functional language.
> It is largely based on DSSSL, the stylesheet language for SGML, an ISO
> standard based on Scheme.  (I was gobsmacked when I first discovered
> Scheme was part of an ISO standard for document production!)
> 
> Bad news: XSL is not nearly as nice as Haskell.  Sigh.

So, if they are making sure that all platforms have a scheme interpreter,
why not make the spec language agnostic.  i.e. allow the user to use any
language that can be compiled/translated into scheme (by code written in
scheme?).
So the user would be able to say:
<transformation language=http://www.haskell.org/haskell2scheme>
</transformation>

The question for the compiler/interpreter writers here is
1. how hard would it be to write code that translates haskell into scheme?
2. would this translator be excessively slow?
3. would the resulting scheme be excesively slow/large?

If the answer to these questions is (easy, no, no), then this minor change
would dramatically increase the language options for the whole web
community.

If the answer to these questions are hard,slow,slow and/or large, then is
there a well defined VM which declarative languages like scheme,
haskell, prolog &tc. can target?  For example, why not use a JVM that
allows tail-call recursion, as the target machine? or the Ocaml VM?  

Such a generic language system would benefit more than just the haskell
(or declarative programming community).  I would guess that translator
writers could define declarative subsets of C, Java, Perl, TCL, etc.

It also seems like XSL sort of resembles a production system so languages
like Clips/Jess might be better suited to process an XML stream (for
some applications) than any of the functional programming langues.

In any case, wedding everybody to one language seems like a mistake.
(e.g. Java)

> PS.  If anyone has any suggestions about *small and reasonable* changes
> to the XSL working draft that would be an improvement, please let me know.


I guess that is the question.  I don't know enough about language design
to answer.  Are the above suggestions *small and reasonable*?  

-Alex-
 
> XSL working draft: http://www.w3.org/TR/WD-xsl
> XSL W3C home page: http://www.w3.org/Style/XSL/
> 
> -----------------------------------------------------------------------
> Philip Wadler                             [EMAIL PROTECTED]
> Bell Labs, Lucent Technologies      http://www.cs.bell-labs.com/~wadler
> 600 Mountain Ave, room 2T-402                   office: +1 908 582 4004
> Murray Hill, NJ 07974-0636                         fax: +1 908 582 5857
> USA                                               home: +1 908 626 9252
> -----------------------------------------------------------------------
> 

___________________________________________________________________
S. Alexander Jacobson                   Shop.Com
1-212-697-0184 voice                    The Easiest Way To Shop