[
Lists Home |
Date Index |
Thread Index
]
- From: Rick JELLIFFE <ricko@geotempo.com>
- To: xml-dev@lists.xml.org
- Date: Tue, 14 Nov 2000 23:50:14 +0800
"Sean B. Palmer" wrote:
> My question again: "Now, I want to 'attach' a scheme [sic] to the namespace,
> that tells both humans and parsers not only what the content and structure
> of the document is, but what it means as well. Question: How do I best go
> about doing this?"
There are several approaches, each of which have different tradeoffs:
1) If you control the domain of the namespace name, then put just plonk
some semantic schema there. Pro: easy. Con: lucky dip.
2) If you control the domain of the namespace, and it already has an XML
Schema there, then stick your semantic schema inside the XML Schema
using appinfo. Pro: have cake and eat it. Con: still a lucky dip.
3) Use the namespace URL as the base: get a directory listing back (in
what format?) and look for any file with an extension that matches the
well-known extension for your file type. Pro: extensible. Con:
to-and-fro.
4) Use content negotiation with the request on the namespace name, to
request that the resource be sent back as the MIME type you want. Pro:
extensible. Con: content negotiation is meant to retrieve different
renderings of the same resource, but this uses it to get data that has
different meaning.
5) Put a catalog in some format at the URL (SOCAT, DrLove, UDDI??) and
then use that to locate the appropriate resource. Pro: see 3. Con: see
3.
6) Establish some resource discover service, perhaps using RDF, and then
have that tell you the schema for the namespace. Pro: distributed. Con:
publicity.
7) Add some strong hinting mechanism to the instance to tell it where
the schema is, and then use the appropriate declarations in that schema
for the namespace. (i.e. xml-schemas solution) Pro: attachment is by
markup not by convention. Con: people will think that this is somehow
not simple.
Cheers
Rick Jelliffe
|