Run, XTATIC, run: Efficient implementation of an object -oriented language with regular pattern matching

Michael Y Levin, University of Pennsylvania


Schema languages such as DTD, XML Schema, and Relax NG have been steadily growing in importance in the XML community. A schema language provides a mechanism for defining the type of XML documents; i.e., the set of constraints that specify the structure of XML documents that are acceptable as data for a certain programming task. A number of recent language designs---many of them descended from the XDUCE language of Hosoya, Pierce, and Vouillon---have showed how such schemas can be used statically for type-checking XML processing code and dynamically for evaluation of XML structures. The technical foundation of such languages is the notion of regular types, a mild generalization of nondeterministic top-down tree automata, which correspond to a core of most popular schema notations, and the notion of regular patterns---regular types decorated with variable binders---a powerful and convenient primitive for dynamic inspection of XML values. This dissertation is concerned with one of XDUCE's descendants, XTATIC. The goal of the XTATIC project is to bring the regular type and regular pattern technologies to a wide audience by integrating them with a mainstream object-oriented language. My research focuses on an efficient implementation of XTATIC including a compiler that generates fast and compact target programs and a run-time system that is designed to support efficient manipulation of XML fragments. Many techniques described here are applicable not only to XTATIC but also to other XDUCE derivatives such as CDUCE and Cω.

Subject Area

Computer science

Recommended Citation

Levin, Michael Y, "Run, XTATIC, run: Efficient implementation of an object -oriented language with regular pattern matching" (2005). Dissertations available from ProQuest. AAI3211170.