Developer's Musings. By Hasse Haitto, Synex Information AB, Stora Nygatan 20, S-111 27 Stockholm, Sweden, Tel: +46 8 791 88 81; Fax: +46 8 791 88 89; Email: haitto@synex.se; WWW: http://www.synex.se.
SGML and its related standards are still very much a niche market. XML, being developed within the W3C with endorsement from companies such as Microsoft, Netscape, Sun etc., has the potential to broaden SGML usage. XML also presents a danger in that it may dilute the interest in SGML and standards-based document processing, much like the proverbial cuckoo's egg.
In the wake of the harmonization between SGML, HyTime, and DSSSL, there is clearly a number of new features product vendors will be required to support. XML is also gaining a lot of support among companies. The response to these requirements is initially likely to boil down to a question of market demand. The requirements do not stem from standards development driving software development, but from a combination of user awareness and customer demand: it is the interplay of software that supports new features, and users who have found a need for these features that ultimately brings about implementations and new markets.
Looking at the initial interest in XML, it seems that concern of an adverse effect on SGML's future may be unfounded, as emerging XML uses are heavily slanted towards meta-data, in contrast to typical SGML usage. Microsoft's Channel Definition Format (CDF) is a specification for turning a web site into a push channel using XML. The Open Software Description (OSD), also an XML application, recently jointly announced by Microsoft and Marimba (and endorsed by several companies including Netscape Communications and Lotus Development) provides means to describe "software components, their versions, their underlying structure, and their relationships to other components." Conceptually, this use of XML is closer to HyTime than current SGML practices.
Though any assessment of XML's ultimate use at this point is bound to be guesswork at best, it would seem that whereas the standardization process has addressed complex requirements carrying the standards into the next century, the simplified SGML of XML is being rolled out to meet complementary needs essentially as a data description language for the immediate and near future, with XML web delivery as another obvious use.
The XML effort is being moved forward with urgency and quite some determination, in a course typical of industrial de facto standards. XML is easy enough to grasp, and its applications readily apparent. The market is there.
Currently, the apparent complexity and lack of applications for advanced HyTime and DSSSL stand in the way of bringing the potential of these standards to a larger audience, resulting in a chicken and egg situation: the market has yet to be created for vendors and users to cross-pollinate applications development. HyTime, an application of SGML, has especially in its second edition (ISO/IEC 10744:1997) addressed a number of key requirements not covered by SGML and which are the key to very interesting uses. These additions, new facilities, and the processing model of groves - a formalism introduced with DSSSL - combine to make the three standards a powerful conceptual apparatus indeed.
Groves are an application-independent abstraction of the result of parsing, and which therefore can be unambiguously understood between applications. Effectively, there is now a model for different standards-based tools to share any piece of SGML information. The groves formalism also forms the backbone in linking and addressing, for hyperlinking SGML documents and other data, regardless of their notation. Supporting hyperlinking based on these novel concepts will be an important feature in the years to come.
Property sets define object classes and their properties; the SGML property set is used by DSSSL and HyTime, and will become part of the revised SGML standard. The output of an SGML parser can thus be described in these implementation-independent terms, as a low-level abstract API to the document.
The new HyTime standard has two parts: one covering the domain of HyTime per se (for describing time-based, hypertextual multimedia, with modules for linking and addressing, event schedules, and rendition) and another covering the SGML Extended facilities, essentially a number of separate, but important, specifications, one of which is the Architectural Form Definition Requirements. Architectural forms are building blocks in deriving documents or specializing generic architectures (similar to a superclass in object-oriented terms).
Whereas XML is designed for simplicity, with design goals rather easily met even when starting from scratch, the standards' new facilities require product vendors to meet specific requirements at a rather profound level of complexity within the standards. Grove-based document processing defines steps for grove construction, interpretation, and providing results of such processing. Addressing development of this functionality typically requires person years. Luckily, much of this new SGML/DSSSL functionality is already implemented in the freely available sp parser and Jade DSSSL engine, respectively, both from James Clark (http://www.jclark.com/).
As far as tool vendors are concerned, there are several avenues to explore. XML tools may well explode to meet demands of DTD-less authoring (improvising markup as you go along) and applications to support authoring DTDs from existing XML document instances can also to be expected. Transformation tools between full-fledged SGML and its simpler offspring XML format is likely to also be in demand. Another class of tool may be for tracking the evolution of DTDs; versioning is probably more important in XML-based document management due to the greater freedom of expression and rather loose validation requirements. Mid-range HyTime application can solve specific uses like Topic Map addressing. At the other end of the application spectrum one can see the advanced HyTime/SGML/DSSSL-based tools powering a new breed of programs, for applications that will bring about the unification of data formats, hyperlinking, and the way we deal with information.
As we undertake this journey, no doubt the applications will influence, and be influenced by, the ways we would like to use our data. The goal is distant still, but we may one day remember with fondness this brief moment in time, of pioneering efforts and hard work ahead.
Contact Robin Cover with corrections and updates, or to submit contributions to the ISUG online document database.
