DITA 1.0 changes

[Christopher Wong <cwong@idiominc.com>]

The following is my attempt to document the changes going from pre-1.0 DITA (public toolkit 1.3.2) and the official OASIS DITA 1.0 standard. Hopefully, this will make it easier for DITA adopters to migrate to the official standard. I'm posting this on the TC list first in case any of you can catch any mistakes I made.

DITA 1.0 changes

• Topic level elements' bodies are now optional. This affects topic and its specializations. So topic/body, task/taskbody etc are now optional.
• The boolean element is eliminated. This element is redundant, as the state element provides a superset of its functionality.
• The xref element can now contain the desc element to be consistent with the topicref element..
• The choice element's content model and attributes has been expanded to be consistent with the li element. This includes the id attribute so this element can now be a link or conref target.
• The query attribute was added to the link element to be consistent with the topicref element.
• The xml:lang and translate attributes were added to the topicref element to be consistent with the link element.
• The table element is now essentially the OASIS Exchange model. It was previously CALS-based.

  • The spanspec and tfoot child elements have been eliminated.
  • The expanse attribute has been renamed to pgwide.
  • char and charoff attributes have been added to the entry and colspec elements.
• Namespace support:

  • The DTDVersion attribute in topic level elements has been changed to a namespace-prefixed DITAArchVersion attribute.
  • All other elements have a null namespace declaration (xmlns="").
• The public IDs have been changed. In keeping with the move to a vendor-independent standard, the DITA public IDs have changed the owner component from "IBM" to "OASIS". They are otherwise unchanged. For example, topic.dtd's public ID changed from "-//IBM//DTD DITA Topic//EN" to "-//OASIS//DTD DITA Topic//EN".
• The use of the conditional processing attribute otherprops has been clarified. As I documented before, it can contain expressive multi-axes spaced-delimited conditional properties like "language(C C++ Java) skill(programming) coffee(black)".
• The use of topicref has been clarified. In place of the now-rejected mapref proposal to include other DITA maps in map, the committee simply decided to make clear that topicref can serve this purpose. The format attribute of topicref is now documented to explicitly allow the "mapref" value (i.e., the resource is a DITA map).
• keyref attribute is CDATA instead of NMTOKEN. This relaxes the attribute's model.
  
• The keyword element's definition now parametrizes the tm element to be consistent with other base phrase types.