Using icon-derived technologies to drive model transformations.

Model transformations are currently the object of much interest and research.
Current proposals for model transformation languages can be divided into two
main camps: those taking a ‘declarative’ approach, and those opting for an ‘imperative’
approach. The Icon programming language is a SNOBOL derivative which
contains several unique constructs which make it particularly well suited to the
job of analyzing and transforming strings. In this paper we discuss model transformations,
analyze the relevant parts of Icon that lend themselves to transforming
strings, and then propose how some of Icon’s unique features could be incorporated
into a model transformation approach that partially blurs the distinction between
‘declarative’ and ‘imperative’ approaches.