5463 2008-02-07 20:58:01 +0000 Must conforming producers be able to spin straw into gold? 2008-02-20 15:20:17 +0000 1 1 1 Unclassified SML Interchange Format unspecified Macintosh All RESOLVED FIXED P2 normal LC 1 cmsmcq kumarp public-sml oldest_to_newest 18841 0 cmsmcq 2008-02-07 20:58:01 +0000 The text of section 5.1 as approved in bug 4675 reads in part A conforming SML-IF Producer MUST be able to generate a referentially conforming SML-IF Document from an SML model. If the SML is non-conforming, or invalid in certain ways (e.g. because its SML references are messed up), is it clear that the generation of a referentially conforming SML-IF document is possible in principle? Full validity is clearly not a logical prerequisite, but is conformance? The current unqualified wording seems to require a lot of a conforming producer. 19001 1 kumarp 2008-02-13 08:20:07 +0000 Proposal: Change the quoted sentence, from: A conforming SML-IF Producer MUST be able to generate a referentially conforming SML-IF Document from an SML model. to: A conforming SML-IF Producer MUST be able to generate a referentially conforming SML-IF Document from a conforming SML model. 19038 2 pratul.dublish 2008-02-14 05:01:49 +0000 Agree with Comment #1 19064 3 virginia.smith 2008-02-14 21:08:13 +0000 Fix per proposal in comment #1 and add non-normative note stating that the expectation is that the input and output model are 'equivalent' but we are not defining this equivalence. (editors to wordsmith) 19097 4 kumarp 2008-02-16 09:35:31 +0000 Updated per proposal in comment# 1 and added the following non-normative note: When a producer generates a referentially conforming SML-IF document from a conforming source model, it is expected that the source model and the generated model are equivalent. That is, the source model and the destination model both have the same validity, same number of documents with similar structure and content differing only in places where references are updated to have equivalent SML URI scheme representation. However, this specification does not normatively define the notion of model equivalence. 19146 5 johnarwe 2008-02-20 15:20:17 +0000 The disposition in comment #3 was a wg decision from the 2/14 telecon.