- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Fri, 15 Mar 2002 10:20:55 -0000
- To: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>, "Brian McBride" <bwm@hplb.hpl.hp.com>, "RDF Core" <w3c-rdfcore-wg@w3.org>
I am sorry, I messed up. The wording I proposed does not meet up with DaveB's proposal for N-triple (i.e. that xml ltieral equality in N-triple is binary compare). Here's another attempt, with the same intent but also getting compatibility. NEW: [[[ 1. Propose that: - the exact form of the string value corresponding to any given XML Literal within RDF/XML is implementation dependent. - the string value is well-balanced XML that can be inserted as the elment content between two tags: <foo></foo> to form an XML document, satisfying both XML and XML Namespaces. - taking the exclusive canonicalization of both the original XML Literal in its containing document, and the string value within the dummy "<foo></foo>" document produce the same character string. - that the canonicalization above is without comments - that this closes the xml literal and xml literal namespaces issues 2. Propose that: The Ntriple test cases for xml literal are defined using the exclusive canonicalization without comments. The following algorithm can be used to generate n-triple output to verify test case conformance: - take the string value returned from the parser for an xml literal - create a new XML document by inserting that string value between <foo></foo> - take the exclusive canonicalization without comments of the element content of this new document and output this value as the string component of the n-triple. ]]] Moreover, we don't need to add a suite of C14N-X tests to the test suite, so how about restricting our test cases to as easy as possible ones while exercising the relevant cases, e.g.: - no predefined entity or character references in the output - keeping the number of attributes in elements down, and trying to avoid the cases where reordering of attributes happens. - no new lines at all - input should be reasonably canonical (e.g. canonical use of whitespace, use of (") not (') for attribute values) I would guess we do need to show - adding visibly used namespace declarations - PI handling? - entity reference substitution? I think we could omit: - comment handling? - removing of unused namespace declarations? The dividing line is: - the ones I think should be in show what is interoperably required of any implementation needs to do (e.g. expand entities, retain PIs, put in visisbly used namespaces) - areas which are only part of C14N and not really imported into the RDF specification except as a way of getting equality are not in the tests - areas which we are deliberately leaving vague are not in the tests. OLD: > [[[ > Propose that: > - the exact form of the string value corresponding to any given > XML Literal > within RDF/XML is implementation dependent. > - the string value is well-balanced XML that can be inserted as the elment > content between two tags: > > <foo></foo> > > to form an XML document, satisfying both XML and XML Namespaces. > > - taking the exclusive canonicalization of both the original XML > Literal in > its containing document, and the string value within the dummy > "<foo></foo>" > document produce the same character string. Equality between xml > literals is > defined on this basis, but only for the purpose of exercising the test > cases. [Such equality may be used for other purposes but there are > other notions of equality (such as via Inclusive Canonicalization) that > may also be useful.] > - that the canonicalization above is without comments > - that this closes the xml literal and xml literal namespaces issues > ]]] >
Received on Friday, 15 March 2002 05:40:02 UTC