XML as a Tool for Domain-Specific Languages

Abstract

Abstract: Computers are general-purpose machines for manipulation of symbols, which means they can be applied in almost any field whose problems can be expressed in terms of symbols. But the creators of computer systems and the potential users of those systems do not always think the same way and do not always find communication easy. Much of the history of information technology can be glossed as a series of attempts to bridge this communication gap. One current approach to this problem is to design ‘domain-specific languages’ (DSLs): formal languages suitable for computer processing, with vocabulary and semantics drawn from the intended application domain. In retrospect, the design of the Extensible Markup Language (XML) can be viewed as an attempt to encourage domain-specific languages and make them easier to specify. Like DSLs as conventionally conceived of, XML vocabularies allow concise descriptions of interesting states of affairs in a particular application area and tend to be more accessible to domain experts than conventional programming languages. Unlike conventional DSLs, most XML vocabularies are specified as having declarative not imperative semantics; this is both a blessing (declarative information is almost always easier to verify and easier to apply in new and unexpected ways) and a curse (many conventional programmers find declarative semantics hard to come to terms with). Examples will be drawn largely from XML vocabularies for the encoding of culturally significant textual materials.