Style sheet languages for hypertext

Abstract

There are several advantages in being able to separate the structural information of a hypermedia document from the specification of the actual presentation of the document [17]. For example, one style can be applied to multiple documents, so that changing the style for all documents requires only a single change to the style information. Alternatively, a single document may have multiple styles tailored to different presentation platforms such as large screens, small screens or paper. Separation of structural and style information has long been commonplace for text, and can also be found in many hypertext models. For example, the Dexter hypertext reference model [7] separates the structural information of the storage layer from the style and layout information encoded in the presentation specifications. In most hypermedia design models, including RMM [12] and HDM [5], the two types of information are designed during different phases of the design process. Style sheets are frequently used to specify style information separate from the structural information. For example, on the Web Cascading Style Sheets (CSS [13, 2]) are used to describe the appearance of HTML documents. While CSS is designed primarily for describing how HTML documents should be formatted by Web browsers, other applications require style sheets to be applicable to more than one class of documents. The electronic publishing community, for example, uses many SGML document types other than HTML. Instead of developing a style language for every single document type, DSSSL defines a language which can be applied to any SGML-defined document type, including HTML. Given the number of different hypermedia models and applications that exist today, being able to define style sheets for different document types is an important feature of a hypermedia style sheet language. However, it is not sufficient to fully qualify as a hypermedia style sheet language. For example, a DSSSL style sheet defines a mapping from the source document to a common abstract output model. This output model is page-based, that is, it describes the output of the formatting process as a sequence