SGML Extended Facilities

Charles F. Goldfarb

HyTime 2 (ISO/IEC 10744:1997)

• Two standards in one document
  • SGML Extended Facilities
  • Enhanced HyTime architecture

SGML Extended Facilities Overview

• For all SGML, not just HyTime
• Definition requirements for
  • Architectural Forms (AFDR)
  • Groves and Property Sets (PSDR)
  • Formal System Identifiers (FSIDR)
  • Lexical Types (LTDR)
• "General Architecture"
• Modular conformance: use any or all

Architectural Form Definition Requirements (AFDR)

• Define "enabling architectures" for SGML documents
• Element type, attribute, and notation forms
• Permits common processing tools across multiple document types
• Documents are "derived" from architectures
• Multiple derivation levels and/or bases

Key AFDR Concepts

• Client document is "derived" from one or more "base" architectures (meta-DTDs)
• Client identifies its base architectures via "architecture support declarations"
• Any DTD can be a meta-DTD (even a derived DTD)
• An SGML parser can recognize "architectural instances" (as well as normal document instance)

Architecture Benefits

• Proven technique since 1990
• Provides OOPS-like inheritance of attributes, content models
• Manages multiple name spaces without conflict
• No new syntax: use with any DTD (e.g., HTML)
• Transparent to normal SGML parsing
• Free implementation in James Clark's SP

Property Set Definition Requirements (PSDR)

• Classes and properties recognized by a parsing process
  • SGML is one such process
  • HyTime datatok is another
• SGML classes include elements, attributes, data characters
• Properties of SGML elements include element type, markup, etc.
• Property set definition document for any notation

Graph Representation Of property ValuEs (GROVE)

• Structure created by a parsing process
  • Graph nodes are instances of classes
  • Graph arcs are properties of nodes
• "Grove plan" determines which classes and properties are included
• Grove contains multiple connected trees
• Groves linked to form hypergroves
• Common object model for any notation

General Architecture

• Extensions to SGML semantics
• Immediate reference typing
• Attribute inheritance with default value lists for implied attributes
  • DVL: Inheritance from ancestors
  • AFDR: Inheritance from base classes
• Modular conformance: use any or all

Formal System Identifier Definition Requirements (FSIDR)

• Coexists with informal system IDs
• FSI definition document describes all system storage facilities
• Multiple storage managers permitted
  • Local: osfile, osfd
  • Portable: url, neutral, mimetype
  • Global: literal, ThisOne
  • Container: sbento, tar, mime

Lexical Type Definition Requirements (LTDR)

• Lexical types for attribute values, data content, pattern matching, addressing
• Single definition mechanism for all uses
• Use any modeling notation: Posix regular expressions, HyLex, et. al.
• LTDR + AFDR + SGML provide generalized extensible schema facility
  • Transparent to parsers
  • Fully standardized; no ad hoc code
  • Doesn't confound lexical and structural

HyTime Architecture Enhancements

• Facilities aligned with DSSSL
  • SDQL with HyTime addressing
  • Core expression language
  • Common SGML grove
• Location addressing: grove-based (transparently), extensible
• Hyperlinking: more link forms
• Scheduling: fcsloc, calibrated schedules

Foundation for Future of SGML and XML

• Data modeling and precise document type definition (LTDR, AFDR)
• Network-awareness with no change to document instances (FSIDR)
• Common object model for all notations (PSDR, groves)
• Name spaces (AFDR)
• Inherited attribute values and content models (AFDR, GA)
• Rich linktyping and locating (HyTime)

For Further Information ...

Visit the HyTime Users' Group at www.hytime.org

Copyright ©2008 Charles F. Goldfarb. All rights reserved.