A Layered Approach to XML Canonicalization

XML Canonicalization enables reliable textual and binary comparison of XML documents through the removal of irrelevant differences in structure and content. Though XML Canonicalization is critical for XML Signatures, it also has value in other XML applications such as version control. Currently, the approach to XML Canonicalization is to write a single specification that details how all parts of XML instances are to be canonicalized. This position paper suggests an alternative approach in which canonicalization is specified separately for the different layers in XML data stack (core, schema, and namespace). In this way, an application can select different canonicalization methods for the different layers; this allows the customization of canonicalization to the application's needs while simplifying the task of writing canonicalization specifications because the specifications need not be all encompassing. As well, it may also be feasible to canonicalize data belonging to different namespaces differently; for example, numbers in an accounting spreadsheet could be canonicalized differently than numbers appearing in descriptive text.