W3C EXI WG and Fujitsu

About W3C EXI

EXI (Efficient XML Interchange) WG was chartered by W3C in December 2005 with the objective of defining an efficient XML interchange standard.

W3C XBC (XML Binary Characterization) WG which harbingered EXI completed its activity early in 2005 after gathering use cases in which XML had not been well adopted and identifying document properties that could help facilitate more prevalent adoption of XML in those use cases.

XBC WG concluded in its final characterization note, while only qualitatively rather than quantitatively, that an alternative efficient format for XML Information Set could be worth being worked on by W3C to make a standard after discussing for a year on the topic not only from technical perspective but also by addressing ramifications of making such a standard in pros and cons.

EXI activity bases itself on the knowledge that XBC had accumulated. It will first attempt to make a convincing case for its next effort to define an efficient format alternative to XML 1.0/1.1 by carefully crafting a benchmark framework and running it against candidates to see if there are any EXI candidates viable to qualify as the base of the new efficient interchange format. If and only if EXI WG finds there is at least one candidate that passes the qualification threshold, it will start defining a standard format learning from those viable candidates. The initial results of the benchmark are planned for publication at the end of May 2006. It then is slated that the first public working draft of the efficient XML interchange format is to be published as early as September 2006.

According to the charter of EXI WG, the specification of the efficient XML interchange format, if there is one, will become W3C Proposed Recommendation in October 2007 soon after that EXI WG will wrap up its activity.

Our contribution to W3C EXI

Fujitsu has been conducting the study of document formats for XML Information Set in pursuit of two primary objectives.

  • To find a better way to interchange documents between information systems that are aware of XML and those existing information systems that are based on legacy document formats such as CSV (comma-separated values) and fixed-length flat files.
  • To facilitate more prevailing use of XML on new information systems that are found to have difficulties in employing XML as in the form it exists today due in part to inherent nature of the platform, examples of which include resource-constrained systems such as cell phones or PDAs. We attempted to remedy the gap by contriving an optimal encoding of XML Information Set “FXDI” (Fujitsu XML Data Interchange format) thereby reducing the burden the use of XML requires of the system.

Fujitsu is an active member of the W3C EXI WG since its start. In response to the call for EXI candidate technologies, we contributed FXDI to the EXI WG to help its effort to prove there is an existing art that meets various requirements EXI and to put forth our expertise in the process of concocting the standard EXI format.

About FXDI

Fujitsu XML Data Interchange (FXDI) Format has been designed to serve as an alternative encoding of XML infoset that allows for more efficiency both in the exchange of data between applications and in the processing of data at each end-point. The goals set for FXDI to achieve in its design included document compactness and the ability to allow implementing decoder and encoder programs that run fast, are of small footprint without involving much complexity.

Although FXDI performs much better when XML Schemas are prescribed before documents are processed, it is capable of handling schema-less documents and fragments by its support for infoset tokenization.

From its first conception until today, FXDI format has been improved continuously to find the right balance among document compactness, implementation complexity and the amount of the schema information. FXDI Schema Compiler compiles W3C XML Schemas into a FXDI Compact Schema. In general, the more information you attempt to cram into a schema, the more compactness you can get. However, since you do not want to end up with a schema that is too large, it became an aspect of balance that was taken into account in the design of FXDI. The other aspect was that the more schema information you try to explore and leverage, the more complicated the processor tends to become. The current FXDI format and its processor reflects the study that has so far been made to find a reasonable balance considering those two aspects.

Because FXDI was provided to W3C EXI WG as a candidate from Fujitsu, both its capability and performance are currently under evaluation among some others within the working group. Initial analysis with some benchmark result numbers are expected to be available as early as the end of May 2006 in the first working draft of EXI Measurement Note.