MIME/RFC822 <-> XML interop (was RE: Binary Data in XML)

Jonathan A. Borden (jborden@mediaone.net)
Wed, 30 Sep 1998 15:22:34 -0400

So what you are saying here is that the use of a attribute is totally
compatible with the use of NOTATIONs and when DTDs are used both are useful.
I suggest that xml:content (or xml:content-type) be used instead of xml:mime
because MIME has lots of potential headers. In the past few minutes I've
hacked my MIME class library to add XML serialization support. What I've
found is that while the vast majority of RFC 821/822 headers are single
values, several are frequently multivalues e.g. the SMTL Received: header
that everyone has seen a gazzilion times :-) for this reason, I have
represented generic RFC 821/822 messages (which incorporate MIME) as: (the
point being that elements may be needed to represent generic RFC822 headers)

<MIME>
<Received> ... </Received>
<Received> ... </Received>
<From>jborden@mediaone.net</From>
<To>xml-dev@ic.ac.ul</To>
<Reply-To>jborden@mediaone.net</Reply-To>
<Content-Type>multipart/mixed</Content-Type>
<Body>
<MIME>
<Content-Type>text/plain</Content-Type>
<Body>This is an example of an e-mail message
however the text here will need to be encoded
</Body>
</MIME>
<MIME>
<Content-Type>image/jpeg</Content-Type>
<Content-transfer-encoding>base64</Content-transfer-encoding>
<Body> ... base64 encoded data here </Body>
</MIME>
</Body>
</MIME>

Is this reasonable?

John Cowan wrote:

>
> Here's how. Insert the following declarations into the DTD
> (internal or external as desired):
>
> <!NOTATION chemical-x-chemdraw PUBLIC "-//whoever/whatever">
> <!ATTLIST CML:molecule
> xml:mime NOTATION(chemical-x-chemdraw | another | another)
> #REQUIRED>
>
> and then just use it as above. A validating parser will guarantee
> that the value of the xml:mime attribute is one of the specified
> notations, and it will be possible to retrieve the external
> identifier for each notation so that you know what it means.
>
> The nice thing about this is that it just works if you are DTD-blind,
> but provides the information that a generalized processor for
> notation-governed elements needs to figure out what to do
> (e.g. render using a chemical/x-chemdraw renderer).
>
Jonathan Borden