Chandi Perera, CSIRO, Australia
Erland Oreby, University of Oslo, Norway
Eric van Herweijen, CERN, Switzerland
Stephen Buswell, STILO, United Kingdom
Per-Ake Ling, Eurostep AB, Sweden
Dianne Kennedy, GCA, USA
The meeting began with introductions and roll call. Minutes from the previous meeting were read by Dianne Kennedy. She then provided an agenda for the meeting that would result in a series of work items in support of a new version of ISO 12083.
The first discussion centered around the content of the Revision. The options include:
The group determined that we should do only one XML-compliant DTD. This would be an XML DTD, not an SGML DTD. An SGML DTD could be easily created by eliminating the <?xml processing instruction. This would allow the DTD to work with XML and SGML tools.
The new DTD would refer to ISO standard WebSGML to as the basis of our work, but also reference W3C XML Version 1.0. In addition, the group recommended that a style sheet be provided to support Web delivery. A simple CSS-2 style sheet will be prepared as an appendix. If XSL is a recommendation by the time a revision is provided, an XSL style sheet will be provided as well.
There was a great discussion about what 12083 should be for the future. The way the DTD is now, many organizations have developed technically "illegal" variants. That is not a good thing. It was suggested that we define parameter entities for a general body to be used for article, book, serial, as well as options for tables, math, and metadata (instead of front and rear matter). Many of the problems people have with the current DTD are because the current modularization does not allow for the many varieties of front and back matter that exist in published materials. In today's world, with DSSSL and XSL transformation tools, we can and should define an ISO 12083 interchange and archive format that simply reflects the data content but does not necessarily have to be a one-for-one match with the element ordering in the printed manuscript. Hence, the suggestion for substituting a metadata parameter entity for "front" and "back".
Eric pointed out that that is just how the current DTD functions and that we could change our units of modularization without changing the actual mechanisms we use to modularize. He also pointed out that there is a large audience that already understands the current ISO DTD organization and modification rules. To change this would involve unnecessary work and confusion.
We decided to study how public identifiers could be added into the model so we can clearly see what we are using for each part, such as standard body.
Per-Ake Ling will prepare a draft of how this can be done.
We must have a provision for including TeX . We must use #PCDATA with comments that you must put a Marked Section "IGNORE" since we cannot use CDATA any longer.
IBM Web site (Bob Sutter) use that approach.
We need to include one or more table models. We decided to include the current model, even though there is little authoring software support, because some in the community may be using this model. In addition an XML version of the CALS model (the new OASIS XML table model) as well as an XML version of HTML tables.
Several math fragments will be referenced or included. The first is the AAP model, since a number of leading publishers still use that model and since there is authoring support for that model. Second is the current math model, even though use is highly limited. We also will reference MathML as well as OpenMath.
Graphics notations needs updating to include new graphic formats as well as Web-compatible graphics.
Eric raised the issue of chemical equations. It was always the goal of ISO 12083 to include a DTD for chemical equations, but one was never developed. The group decided to provide a reference to the new XML-based Chemical Markup Language (CML) developed by Peter Murray Rust.
Most bugs reported to the current DTDs are elements that are present in one DTD (like book) but not in article. Dividing the DTD into metadata and one generalized body for all types (book, article, serial) should correct these issues.
There is a known bug in index parameter entity.
<!ENTITY % m.idx "(%m.sec | ((indxname | indxsubj)*, pages*)*)*">
<LIT tag should not be CDATA -- Paul Grosso (this should be fixed)
ISO 12083 was one of the first "standard" DTDs to contain support for accessiblitiy. We need to check ICAD elements with Murray Mallony or Harvey Bingham to be sure these elements are still valid and conform with new Web accessibility guidelines.
There have been some requests to add new elements special purposes. Those will be added.
Some sort of digital identifier is needed in references. The suggestion is to add this element to bibliography, let the user specify the namespace. There is some debate about the format of idenifier that should be used (PII Vs. DOI). We will specify the ID type in an attribute for the DigitalId with CDATA value to allow for extensions in the valid type of digital ids.
Cross references must be reviewed. The group decided to allow linking to rely on Xpointer/ XLink in the future. Comments to this effect must be included in the specification.
The meeting was adjorned.