Proceedings

Abstract:

The W3C's XML Binary Characterization Working Group was chartered to produce a recommendation as to whether or not a binary XML format should be specified by the W3C. Talk of binary XML makes lots of people nervous - the results on this investigation could radically change how XML is used and how XML-based applications are developed. This talk paper exposes the rationale for binary XML and report on the published findings of the working group to date. It may or may not allay your fears but at least it will help to understand why this old issue is being taken so seriously today.

Why might XML need a binary format?

Bandwidth: How many bits it takes to send an XML message across a wireless link. Wireless devices have limited bandwidth and greater use of bandwidth increases transmission time and the error rate. Binary XML can reduce the bandwidth needed to send an XML message. Compression techniques include redundancy elimination (e.g., putting common strings in a string table) or domain-based (using knowledge of the structure derived from the schema to encode content more efficiently).

Processing speed/parsing: Generally, XML is too slow! Binary XML addresses this problem by making the data format closer to datatypes used in programming languages and by using compression techniques to reduce the number of bits processed.

Progressive download/streaming: Eliminates the necessity of reading the entire XML document before it can be processed, supporting applications such as partial rendering, packetization, and interleaving.

Random access: Various techniques for including an index of the XML document, enabling direct access to particular sections of interest without sequential processing of the document.

Dynamic update: Various techniques, such as delta records, for updating an XML document without modifying the original, eliminating stream rewriting or tree node manipulations.

New data types: Direct support for various data types useful in non-text based applications of XML, including native numeric data types, arrays, and graphs.

Link support: Fundamental support for XLINK or other hyperlink primitives in core XML.

Compactness: Similar to limited-bandwidth approaches but the applications usually involve archiving of very large amounts of data.

The Working Group has developed application Use Cases around these technology roadblocks and Properties of a possible alternative XML format that would solve these problems and address the Use Cases. These Use Cases and Properties will be discussed and possible outcomes of the W3C activity will be examined.