Saturday, May 30, 2009

Webizing Common Lisp

Lisp is a very old language, but it is by no means old in the derogatory sense. Being the oldest language (except Fortran) has given it time to mature and grow into a language capable of amazing things. It has been said many places that Lisp and XML are similar in that they are both data (and if you include XAML and XSLT, code too). I would argue, though, that Common Lisp has vastly more features. However, to make a true comparison, we have to webize Lisp. At the very least, this requires a URI for every symbol. We can start by giving a namespace to all symbols in the COMMON-LISP package.

For convinience, we will use the cl prefix for this namespace. We can now start relating Web standards and Lisp, or even writing ontologies using Lisp functions.

Lets compare lists, which are a fundamental part of Lisp. In Lisp, a cons (x . y) would correspond to the RDF graph given by [ a rdf:List ; rdf:first _:x ; rdf:rest _:y ] and cl:nil would correspond to rdf:nil. We can do this with almost all Lisp syntax, except perhaps for the syntax based on the hash (#) symbol. For example, #c(3 4) would correspond to the MathML <cn type="complex-cartesian">3<sep/>4</cn>, so the mapping can get complicated.

Considering such a mapping between Lisp and XML has many advantages. One thing they both have in common is the idea of homogenous data, however, when viewed as two separate syntaxes, they are incompatible, because XML would be a string in Lisp, and Lisp would be a string in XML, losing all structure. When considering such a mapping, it may seem equally advantageous to express XML in Lisp, and conversely, Lisp in XML. If only it were a bijection...

No comments:

Post a Comment