ietf-xml-mime
[Top] [All Lists]

Re: Informal last call: Announcement of draft-murata-xml-06.txt

2000-06-22 01:32:59

Interesting proposal.

But let me first introduce myself: i am a chemist involved in XML (CML to
be precise, with some SVG and MathML) and am working on a web server that
will be able to the following:

on the server side there are database with data with a limited set of 
MIME types (mostly XML: CML, SVG, MathML etc). But since not all web clients
are able to work with XML, the server will convert the data into a MIME type
which the client *can* handle...

The link with XML Mime types is ofcourse that NS6 and IE5 can handle XML and
appropriate MIMEs are needed.

Oke, now my comment on the draft.

First of all, i like to say its an important draft (but you all know that).
But what i miss in the draft is mixed XML content (not referring to PCDATA),
but XML documents like this:

<xhtml:xhtml xmlns:xhtml="blabla">
  ... some XHTML code ...
  
  ... some article's text: ... take for example the molecular structure of 
      acetic acid ...

  <cml:molecule xmlns:cml="http://www.xml-cml.org/";>
     ... 2D molecular structure ...
  </cml:molecule>

  ... some more text: ... and if we apply the following equation to this
      structure we get ...

  <mathml:mrow xmlns:mathml="foofoo">
  </mathml:mrow>

  ... and possibly more text and more foreign XML fragments ...

</xhtml:xhtml>

It is easy to give this document a text/xml MIME type, but i would say that
the mechanism which is proposed: application/mathml+xml is very interesting.

But how would that mechanism apply to the document? 
application/xhtml,mathml,cml+xml
Obviously not, because you cannot register all combinations...

The reason why i think this is important is the content negotiation process, 
where the client has to say something like: application/mathml+xml,
application/cml+xml etc... And in this case the client could say:
Accept: application/mathml+xml, application/cml+xml;

But this is not complete. Since namespaces are not part of XML1.0 but an 
extension
text/xml does not include namespaces. So the proposal also lacks a mechanism
to say that a document uses namespaces.

These are some thoughts i had when reading draft... when considering my project
i have the two problems above:
1. there is not MIME type for XML Namespaces
2. what MIME type must the server use when sending a document like the one 
above...

I hope that several can comment on this...

Egon