Lambda the Ultimate

Contestants have 72 hours to implement the challenge task, and the deadline ends Sunday, July 29th. There's also a special lightening division for submissions within 24 hours. The organizers have already received 39 entries in the lightening division!

Besides the prize money, most contestants are,of course, competing for the unlimited bragging rights that are up for grabs.

The challenge task this year is to design and implement an optimizer for SML/NG which is a simple XML-based markup language.

While the objective is to show off how your favorite functional programming language is better, a whole host of languages (or even a combination of several languages) are allowed.

Keep checking the latest news updates.

Cheers!

Kaushik

I tend to put a lot of white space in my documents because I think one should be able to discern the structure by just glancing at the indentation - XML/XHTML documents are in many ways like reading code. Unfortunately, such whitespace tends to eat up bandwidth.

What would be nice is if the compression/optimization could be build into the internet server engine, so you could author your pages without consideration of bandwidth issues and let the server take care of that chore.

as far as i've tested Bourin is better than Python Team, but a python solution can't be real opponent for this kind of problem ;p

• static typing with type inference vs dynamic typing (esp. for higher order programming)
• python data structures don't handle variants nicely vs ocaml which has pattern matching on variants
• python data structures are memory costly
• efficiency difference

let tags2env tags = fold_left add_tag_to_env default_env (rev tags)

This function transforms a list of tag (eg: B, U, I, U) into the context (eg: { default_env with b = true ; u = 2 ; i = true })

It can just as easily be written in scheme *if* you manage to write it the good way. In OCaml, you don't have to care, if it compiles it will run, and that's quite true :)

Indeed. Here's an snippet from the SSAX XML parser written entirely in Scheme. The snippet resolves the names of attributes with respect to the current Namespace context, and verifies that after the resolution, the attribute names remain unique.

 (fold-right	; fold-right, fold-left, etc. are in SRFI-1
     (lambda (name-value attlist)
       (or
	(attlist-add attlist
	   (cons (SSAX:resolve-name port (car name-value) namespaces #f)
		 (cdr name-value)))
	(parser-error port "[uniqattspec] after NS expansion broken for " 
	       name-value)))
     (make-empty-attlist)
     proper-attrs)

	(attlist-fold add-default-decl result decl-attrs)

The SSAX parser itself is a fold -- an enhanced tree fold over an XML document.

Folds and similar generic control structures are very powerful and useful indeed. Church numerals are folds!

A new icfp2001.tar.bz2 has just been uploaded (beware it's *big*: 19MB compressed, 121MB expanded)

See icfp news for some official news: <quote> We have 172 entries, more than the total of all previous years !

The first 7 files (000 to 007) have eliminated more than 100 entries </quote>