xml-dev - Re: Syntax and Semantics

[Peter Murray-Rust <peter@ursus.demon.co.uk>]

At 03:01 PM 5/18/00 -0700, Marc.McDonald@Design-Intelligence.com wrote:
>I think the cases of actually processing the infornation in a document and
>documents from different sources are often overlooked.
>
>
>For example, a document with some MathML, ChemML, and HTML mixed together. A
>DTD or schema may allow the syntax of the sections to be validated, but it
>won't render the formula, molecule or text. The rendering is done by an
>application which understands the semantics of the language involved.
>
>There is no current specification for the cooperation of rendering engines
>for MathML, ChemML and HTML - except perhaps through browser plugins and a
>delegating XML browser. 
>
>All namespaces provide is avoidance of ambiguity. All tying a namespace to a
>schema adds is validity checking.

This states my problem precisely. I need a means of attaching semantics to
a document so that people receiving that document know how to process it.
The success o browser plugins (despite their drawbacks) mean that the
community at large (the 99.99% of web users who haven't even heard of XML)
will expect us to solve it transparently, *in a short period of time*.

It *is* solved - in large part - for documents carrying single MIME types.
I and Henry Rzepa are in the process of mounting a number of *.SVG
documents on our web site. They can be used by someone who has never heard
of SVG. A user clicks on the link and is sent a document with MIME type
image/svg. The browser hasn't heard of this, but asks if I want a plug-in
and then goes to try to find it. I don't know where it gets it from, but it
manages it. Therefore "*.svg" and "image/svg" both carry effective
semantics today. In contrast "http://www.w3.org/TR/someVersion/SVG/SVG.dtd"
does not carry any semantics and many XML-DEVers are arguing that it
shouldn't.


The MIME strategy works pretty well, even outside the formal mechanisms of
the IETF. ca. 4 years ago Henry and I wished to use MIME for stamping files
with molecules in. We proposed a top-level type of "chemical/" to the IETF
- lots of fun discussion for 6 months - and it didn't make it. BUT there is
a de facto convention for "chemical/x-*)(enshrined in the formal
publication of the International Union of Pure and Applied Chemistry"). It
became adopted very quickly. All the chemical sites and most software
suppliers use it. If anyone receives a file with "chemical/x-pdb" MIME
type, the browser will go out and retrieve a plugin. All this without any
official registration of the type. For many pragmatists, this is how the
Web actually works. Theoretically broken but extremely useful. It's a 99%
solution.

There are lots of things "wrong" with this approach, of course:
	- chemical/x-pdb is not guaranteed to be unique (its not registered). It
could be spoofed. 
	- it does not label a resource. If the "PDB" specification changes (which
it does, fairly frequently !) the software can break on new files. But it
breaks anyway.
	- horizontal vendors (e.g. browser m'facturers) tend not to support
discipline-dependent MIME types.

BUT it works for us. We accept all these failings. What pragmatic XML users
will not accept is being told that there is no way of carrying semantics
for XML files. Or wait 3 years while the name/use/mention/id/resource/label
problems are sorted out. 

The immediate problem is the multiDTD file. We cannot use a single scalar
wrapper to describe the semantics. CML documents will contain CML
intimately mixed with SVG, MathML, XHTML, CALS/OASIS and 1-2 other common
technical DTDs. There either has to be a new wrapper specifying this or
there has to be semantics in the file. Either of these require the
browser-writers to agree on a common convention.

I suggest we decide on such a convention as quickly as possible and promote
it. A possible solution could involve PIs. [I choose these because most
people agree you can do anything you like with them and not upset anyone
else - at least that is what the spec says, I think]. PIs are "ignorable
semantics"! So the document contains:

<?xdev mime="image/svg"?>
<?xdev mime="chemical/x-pdb"?>
<?xdev mime="text/xml"?>

[xdev represents XML-DEV, but I don't think we can use "xml-dev", can we?]

Then purists can ignore these. People who want to use this can do so. The
enthusiasts will see how far it goes before it breaks. Meanwhile we shall
have developed a more robust semantic/ontological/semiotic approach which
can be implemented as the next phase.

	P.



***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************