Re: FW: VoiceXML 2.1 and PI issue

My hope is to review the discussions and options with the VoiceXML 2.1 group next Thursday.  MattO is responsible for that agenda; he'll send out appropriate invites. 

The idea is that these PI semantics could be applicable to any XML document.  There's was also a goal of allowing the addition of the processing instruction to existing XML documents (like existing RSS feeds) without having to modify the structure of the document.  So if people use single-namespace processors and editors for there content, the addition of this access-control enhancement shouldn't interfere.

I'm personally still leaning toward the PI myself, but I think I can present balanced trade-offs.  The XML encryption standard looks like it would work, but is perhaps overkill for our needs (key management challenges are the main concern).  It may be something that we should try to support in addition.

Brad

Max Froumentin wrote:

I agree with DanC that the main inconvenient about PIs is that they
aren't bound to namespaces. And it would be a problem if VoiceXML was
mixed with other markup in a document. Unless the PI's semantics
applied to any XML document, which I don't expect VoiceXML 2.1 to do.

  

Dave gave another: the DOM doesn't give access to them.
    

It does and XSLT too.
See http://www.w3.org/TR/2004/REC-DOM-Level-3-Core-20040407/core.html#ID-1004215813

On the other hand, SOAP 1.2 doesn't allow them. I know the reasons for
that have been discussed many times, but I'm still hold the grudge that
SOAP has unilaterally killed PIs in any XML document that might be wrapped
in it. Just like doctypes. But I digress.

So my point is: the only reason why PIs should be avoided is
namespaces.  The other ones mentioned in this thread aren't valid, in
my opinion.

Max.