proposal for issue 462

Hi, 

I'd like to propose the following comment as a resolution to issue 462:

The commentator is concerned with base64 data in the infoset that
contains whitespace; and the handling of such data in XOP.

The primary use-case for XOP is that an application wants to include a
chunk of binary data (that the application has as an octet stream, not
encoded for example as base64) in an XML document in some optimized
form. Where XOP requires canonical base64 data, that is only a
conceptual requirement, it is expected that any implementation will skip
encoding the original data into the infoset (or data model) but will
pass the data directly to the XOP layer which will put it directly into
the MIME envelope. Conceptually therefore, the data may be viewed as
present in the infoset in base64 canonical, even if the implementation
doesn't contain any base64-encoding code.

For the other use cases where the base64 form is actually used, we must
choose one unambiguous form (because we are required to reconstruct the
precise infoset at the receiving end), and we chose to reuse XML
Schema's canonical form (yet to be published?).



To avoid such comments in the future, such text as the above should be
present in XOP, perhaps in the introduction, where the limitation to
canonical base64 is introduced, with no readily-apparent reason.

                   Jacek Kopecky

                   Systinet Corporation
                   http://www.systinet.com/

Received on Wednesday, 25 February 2004 08:09:10 UTC