A New Generation of Tools for SGML

Abstract

Exceptions are used in many standard DTDs, including HTML, because they add expressive power for DTD authors. However, there is a tradeoff: although they are useful, exceptions add significantly to the complexity of DTDs. Authoring DTDs is a difficult task, and existing tools are of limited use because of the lack of a suitable formal model for exceptions. This paper describes methods for constructing a static model that completely and precisely describes DTDs with exceptions. A software tool has been written to implement the methods and to demonstrate some practical applications. Examples are shown of how the tool is used for DTD authoring, and some useful extensions of the tool are described. For one example DTD, the output of the tool is converted into a regular expression grammar. Preliminary studies indicate that general case algorithms can be developed for this conversion. This would allow existing theory for the context free languages to be used in developing SGML applications. Statistical results are shown from running the software tool on a number of industry and government DTDs and for three successive versions of HTML. The results illustrate that the complexity of DTDs in practice is approaching, or has exceeded, manageable limits with existing tools. The formal model and its applications are needed for SGML and continued development of these methods may impact the evolution of HTML, XML, and related web publishing standards. Some specific projects are proposed, where continued development of the model can result in more powerful tools and new kinds of applications for SGML.