On Apr 23, 2006, at 12:12 PM, Bjoern Hoehrmann wrote:

* Maciej Stachowiak wrote:
Should this be taken to mean that for any other Content-Type,
implementations MUST NOT attempt to parse as XML? If so, please say
that. Optionally allowing XML parsing for types not specifically
mentioned would be bad for interoperability.

It should say that implementations use intrisinc knowledge derived from the media type registration to determine whether or not the type used to label the entity body is an XML media type, persumably one that is meant to be used for complete document entities rather than other entities as
those cannot yield in a Document.

What about media types that are not registered?

We might optionally want to require that the corresponding types in RFC 3023 and any type that follows the
+xml convention is recognized as XML document in accord with RFC 3023.

It would not make much sense for a user agent to specifically not re-
cognize specific media types as representing XML documents for the pur-
poses of XHR processing when the user agent does that everywhere else.

I think at minimum we need to require some baseline set of types, or it will be hard to write interoperable application code. I would rather list the full set of types considered to be xml. Given the widespread adoption of the +xml convention, the list of additional types would be relatively short.

However, if you propose that the implementation MAY parse any other media type as XML if it has any reason to think it is XML, then the spec should say that too.

Of these, I only know for sure that text/xsl is in common use for
sending XML content, even though it is unofficial and technically
illegal.

Not following the +xml convention is not technically illegal.

Not because it fails to follow the +xml convention but because it is unregistered and not an experimental or vendor type; because there is an alternate type registered which is the official type; and because it is under the text media type even though generally it is not useful to view the contents as plain text. "technically illegal" is probably not the best way to express that. In any case, the point is that it is in use and this won't go away.

Regards,
Maciej


Reply via email to