Re: Deployment and media types

Hey Anne,

On Wed, Feb 29, 2012 at 1:07 PM, Anne van Kesteren <annevk@opera.com> wrote:
> On Wed, 29 Feb 2012 19:00:01 +0100, David Carlisle <davidc@nag.co.uk> wrote:
>>
>> Others can speak up if they disagree, but I think there has been an
>> implicit assumption so far that xml-er would be used as an alternative
>> parser for things served with an xml media type.
>
>
> Indeed. We do not want to introduce a new format, just a new way to process
> an existing format.

If I send some HTML as text/plain, I'm using the plain text format.
It's true that it's *also* HTML (as well as SGML and some forms of
rich text), but the media type serves as a key for the processing
model with which it's supposed to be interpreted.

Likewise, if I send "<hello>there</goodbye>" as application/xml, I'm
evoking the XML 1.0 specification via RFC 3023 to indicate how my
message should be interpreted. As we all know, that's not well-formed
XML and so it's interpretation is "null".

If you're trying to write a new spec to modify that chain so that my
message can be interpreted and processed not as "null", but as well
formed XML, then you've got to at least update one of the
specifications in that chain, either RFC 3023 or XML 1.0. Which would
you prefer? 8-)

Mark.

Received on Wednesday, 29 February 2012 18:36:17 UTC