W3C home > Mailing lists > Public > www-tag@w3.org > March 2004

Re: Yet another revision of the finding about qnameAsId-18

From: <noah_mendelsohn@us.ibm.com>
Date: Wed, 10 Mar 2004 20:57:00 -0500
To: Norman Walsh <Norman.Walsh@Sun.COM>
Cc: www-tag@w3.org
Message-ID: <OF89220F6F.C285657E-ON85256E54.0009F512@lotus.com>

Speaking for myself, and not necessarily for either of the WG's on which I 
sit, I like this a lot.  Congratulations on another good piece of work, 
Norm.

One quibble:   I still find a bit of wishful thinking in:

"A serializer, presented with an object model that conforms to [XML 
Namespaces] can manufacture new prefixes on the fly.  (In practice, users 
expect most prefixes to be preserved through transformations, so things 
aren't quite this simple for most developers, but this is still 
theoretically the case.)"

I'm not quite sure what object model we're talking about here.  Clearly I 
understand what you intend, but there is no such normative model set out 
anywhere.  My view is:  like it or not, prefixes count in XML, they are 
visible as characters in the serialized document and (if you care) in the 
Infoset.  The closest supporting statement I find in the Namespaces Rec 
for a prefix-oblivious model says:

"Note that the prefix functions only as a placeholder for a namespace name. 
Applications should use the namespace name, not the prefix, in 
constructing names whose scope extends beyond the containing document."

First of all, that's a lowercase "should", but even so it implies quite 
strongly that the prefix may well be significant >within< the document. 
Dsigs over prefixes, application content that refers to elements by QName, 
etc. are all examples that seem to be allowed if not encouraged by the 
above.  I guess I don't see what is gained, on balance, by promoting this 
theoretical view which is not well specificed and so rarely matched in 
practice.   I think I'd just leave out the two sentences quoted above.

I have no objection to the first part of your paragraph which says:

"If QNames are only used in element and attribute names, the processor can 
fully resolve all of the prefixes as it parses. This gives it the freedom 
to discard the prefix-to-URI mappings when they go out of scope."

Why not just stick with that and indicate:  "There may be circumstances in 
which maintaining such discipline is on balance beneficial.", or some 
such, and leaving out the theory of "object models conforming to [XML 
Namespaces]"?

For the record, I can easily live with what you have, but I think the 
paragraph would be stronger if modified.  Thanks again for this nice work.

Noah

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------
Received on Wednesday, 10 March 2004 20:58:09 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:25 GMT