Re: Deployment and media types

On 2/29/2012 1:56 PM, David Carlisle wrote:
> No, if you send non well formed XML to an XML processor, it has to flag an
> error, but the application is free to do anything it likes after the error
> is flagged. If parsing with xml-er and parsing with xml produce the same
> result on well formed xml, then it seems perfectly OK to me to
> parse with xml-er instead asn as I aid initially externally it is the same
> as parsing with xml and just parsing with xml-er on failure.

Right.

I think Mark B. would be correct in his concern if we start advertising any 
non-XML content as "conforming", e.g. if we start encouraging the use of 
unquoted attribute values. If we do that, then I think a new media type is 
indeed necessary.

Otherwise, I think we can leave RFC 3023 unchanged, and make clear that any 
content other than what's allowed by the RFC is in error and 
non-conforming, regardless of whether processed by a traditional XML 
processor or an XML-ER processor. As far as I know, that's parallel to the 
way that HTML5 treats (will treat once the media type is registered) things 
like improperly nested tags; I.e. they are errors, but the browser does its 
best with them anyway. An HTML5 validator will show them as being in error.

Noah

Received on Wednesday, 29 February 2012 20:13:21 UTC