At 10:16 AM 5/7/99 -0500, Shane P. McCarron wrote:
Within an agent, once the basic container has been
determined, the agent will examine the contents further to figure out
how to process the content. Part of this is parsing the XML data
stream. While an agent might have a single XML parser, it might not.
And it doesn't matter. Agents need to know a lot more than we could or
should put in a mime type in order to faithfully reflect the intent of
the content author.
But why is it such a problem to provide the agent with as much information
as possible? If the agent knows that the basic content type is XML, it
knows that it's safe to poke at it with an XML processor. If the agent
knows what kind of XML it's looking at, it can pick and choose which files
it wants on that basis if that behavior is appropriate. At least we'd be
giving the agent as much information as possible within the constraints of
MIME rather than punting and requiring it to search through _everything_ to
find what it wants.
I don't see this as a case where 'less is more'. I still like Rick's
proposal.
Simon St.Laurent
XML: A Primer / Building XML Applications (June)
Sharing Bandwidth / Cookies
http://www.simonstl.com