RE: Atom Syndication Format and the Infoset

["Bob Wyman" <bob@xxxxxxxx>]

Bill de hÓra wrote:
> if someone wanted to call their non-XML encoding 
> of Atom "my non-XML encoding for Atom", I think 
> that would be fine, 

	The question is: "How do I *say* 'my non-XML encoding for
Atom'?"

	My assumption is that MIME-type for Atom would be "atom+xml."
Given this, and given that I might want to use a "fastinfoset encoding
for Atom", then I should be able to register
"application/atom+fastinfoset" as a MIME type. Or, if I was using an
encoding which, unlike xml or fastinfoset, was not one that was useful
in a schema-less environment, I should be able to create and use:

	
application/atom;er="uper";schema="http://example.com/atom.asn1";

	(where :
	"er" indicates the encoding rules used,
	"uper" indicates ASN.1 unaligned PER,
	"schema" is the URI of the Atom schema)

	If I can't do this, then what I'm forced to do is copy the
Atom spec into a new document, give it a new name and then continue as
before. This is simply silliness. I do not believe it is useful to
insist that I copy-paste a specification and give it a new name simply
to be able to use it. If nothing else, once a "copy" of Atom exists
with a new name, there exists the risk of diverging specifications --
THAT would not be good for interop.
	If you want to say that the string "Atom", when used on it's
own and without further qualification, *always* refers to "atom+xml"
then that is fine. But you are doing more damage to interop than good
if you insist that I maintain a parallel definition of Atom simply so
that I can get a MIME type to identify it.

		bob wyman