This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
refs: smlif 3.3.2 P 1,4; smlif 3.3.3; sml 3.3.1, 3.3.2 all section numbers based on submission copy There is a muddled distribution of content on the SML-defined schemes between the two specs. I think a better approach would be to: 1. Define what a scheme definition must contain (in SML 3.3 or a separate document entirely, possibly a Note... the SMLIF-imposed reqt pattern could occur again, and I'd like to not have to rev SML for it). - MUST the content comprising/marking the scheme, e.g. sml:uri - MUST where the scheme's content occurs, e.g. child of reference element; probably should ALWAYS be "where, wrt the ref element" - SHOULD whether or not the scheme MAY/MUST/etc be interpreted as an SML-IF reference (SMLIF 3.3.2 defines normative rules if they are not explicit) 2. Define the sml:uri and sml:EPR schemes, current sections, making sure above conditions are satisfied. 3. Add to SML 3.3.1 the fact that its semantic includes the implication that it can be dereferenced using the default action for the scheme (as stated in SMLIF 3.3.2 P 4) so the two sections agree. 4. SMLIF 3.3.3 P 1 S 2 end, add reference to SML 3.3. 5. SMLIF 3.3.3 P 1 S 3 end, add reference to SMLIF 3.3.2 6. SMLIF 3.3.3 change heading "that are not" to "may not be" 7. Re-write SMLIF 3.3.3 P 1 (new text) SML [1] defines two reference schemes, the URI reference scheme and the EPR reference scheme; it also permits new reference schemes to be defined without limit. Reference schemes MAY be inter-document references in the context of SML-IF [SMLIF 3.3.2]. Three consequences flow from this. 8. Re-write SMLIF 3.3.3 P 2 (new text) First, to successfully interchange models using reference schemes that are not inter-document references in the context of SML-IF, each reference element must be represented using least one reference scheme that IS an inter-document reference in the context of SML-IF. For example, [... rest of existing example]. 9. Change SMLIF 3.3.3 P 3 "of ... [end of sentence]" to "must agree on at least one reference scheme that is an inter-document reference in the context of SML-IF for each reference element in the interchange set." 10. Change SMLIF 3.3.3 P 4 "sml:ref" to "reference" 11. SMLIF 3.3.2 work in the fact that something may also be an inter-document ref (or not) if explicitly declared by its reference scheme definition. If the ref scheme definition does not make any such declaration, the ref scheme's content is treated as ordinary content under the rules above for xs:anyURI.
1. Point #2 should refer to "EPR reference schemes" rather than "sml:EPR", which is a non-existent item. 2. I'm not clear precisely how the "consequences" "flow" (are logically implied by) from the statements in P 1. I would recommend that a new paragraph be initiated, beginning: "Successful interchange of models requires the following:", followed by the three points (reworded slightly in the light of the paragraph opening). (It should be clear that the first two requirements are not normative but are physical requirements of interchange--was this a conscious change?) 3. Insert "at" between "representing using" and "least one"
I agree with section 3 "How are schemes defined?" from the ref proposal (ReferenceIssues.doc).
Resolution is to fix as per Sandy's sml reference proposal http://lists.w3.org/Archives/Public/public-sml/2007Sep/0268.html
1. Defined how a scheme must be defined. See section 4.2 Reference schemes 2. Defined the 2 schemes. See 4.2.1 & 4.2.2. Did not define how each scheme resolves, it needs discussion. Opened separate bugs for this: 5241 & 5242 3. Did not fix #3. This will be covered by the 2 new bugs. 4. Did not fix #4 ~ #11 : They will be covered when we rework interdocument references based on the proposal that Sandy is working on.