2. Document Object Model (Core) Level 1

2.3 Design goals for the DOM Level One specification

It is important to note that the DOM is an "abstraction", or an idealized model of how documents are represented and manipulated in the products that support the DOM interfaces. Or, even more ideally, one might think of the DOM as the "abstract base classes" that products supporting the DOM actually implement. Thus, in general, HTML and XML products that support the DOM merely "expose" the DOM as a means of accessing and manipulating their (possibly proprietary) internal data structures and operations; there is no assumption that they are internally built on the DOM APIs or that they do not expose proprietary APIs that have no relationship to the W3C DOM.

The DOM objects and interfaces are designed to be:

high-fidelity: sufficient for representing the content of parsed HTML and XML documents without loss of [significant] information. The supported HTML version is 4.0; the supported XML version is 1.0.
isomorphic with XML: The DOM should be sufficient to construct an entirely new document instance programmatically that is identical to the parsed form of a given HTML or XML document. This means that it has sufficient constructive power to build any useful document object hierarchy, and that an implementation could be written such that the external document parser merely calls the methods specified in the level one specification to build the object hierarchy.
extensible: The DOM core is the foundation for the rest of the document object model levels, which means it must be simple, flexible, and extensible.
thread-safe: The operations supported by the DOM will not corrupt the document object or return corrupted state (as far as this API is concerned). Higher level consistency support mechanisms such as explicit locks or transactions are outside of the scope of the level one specification. For level one of the DOM, the assumption is that only one thread operates on the document at a time.

Note: In the current specification, some operations can modify the document tree, but there is no model for handling concurrent access. The WG also recognises that in some situations, a document, or some of its components, will not be modifiable, and a method for dealing with such situations needs to be defined in a subsequent revision level of the DOM.

2.3.1 Entities and the DOM Core

In the DOM core, there are no objects representing entities. Numeric character references and references to the pre-defined entities in HTML and XML, are replaced by the single character that makes up the entitity's replacement. For example, in:

   <p>This is a dog &amp; a cat</p>

the "&" will be replaced by the character "&", and the text in the <p> element will form a single continuous sequence of characters. The representation of general entities, both internal and external, are defined within the XML-specific portion of the level one specification. Note: When a DOM representation of a document is serialized as XML or HTML text, applications will need to check each character in text data to see if it needs to be escaped using a numeric or pre-defined entity. Failing to do so could result in invalid HTML or XML.