>From the responses I have gotten from some of the members of the XML-WG, it
is clear that we can't add BASE64 section to the spec. As you pointed out,
BASE64 section is not helpful enough for XML applications. Tim suggested
that we write up a proposal for the use of a reserved attribute or namespace
to signal, as a convention in XML 1.0, that the contents of an element are
base64 encoded. Such a proposal would serve the need right now and could be
adopted by XML 1.1 in the future.
I would like to form a small team to write the proposal. Since we are
dealing with a focused subject, I would like to fasttrack this proposal.
Let me get the ball rolling with following brief summary of the proposal:
1. Name
Names are important since they serve as mental hooks to hang knowledge. The
choices I can think of are:
a) XML-Binary
b) XML-Blob
c) Encoded XML Content
I would like to use a short easily understandable name like XML-Binary so
that vendors can say their product supports XML-Binary.
2. Mechanism
I tend to prefer the use of reserved attribute(s) than namespace. I would
very much like to see something like xml:space attribute used.
For the kind of applications I am familiar with, adding following two
special attributes would be enough:
xml:encoding="base64"
xml:mimetype="image/gif"
Should we limit it to base64 and just have xml:encoded attribute with true
and default as possible values?
Should we be using some standard encoding standard names? Frankly I am not
aware of any such standard (duh!).
Do we need xml:mimetype? My application sure could use it since I can
fireup a content handler based on the mimetype and pass it the decoded data.
The content handler returns a component which is inserted into the tree to
display the content.
This should be enough get the discussion going.
Sincerely,
Don Park