This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
As part of the SML reference proposal, the references semantic had covered these situations: V2.2. If at least one of the recognized schemes resolves to more than one target element, then the model is invalid. V2.3. If one recognized scheme resolves to a target that's different from the target resolved by another recognized scheme, then the model is invalid. V2.4. If one recognized scheme resolves and another doesn't, then the model is invalid. http://lists.w3.org/Archives/Public/public-sml/2007Sep/0268.html The SML spec covers v2.2 in the 4.2 Reference semantics section, under 4.2.1 At Most One Target We need a normative section to cover v2.3 and v2.4 under the same 4.2 Reference semantics section The proposal is to add this section to 4.2 Reference semantics 4.2.2 Consistent Reference Schemes The model is declared invalid when a recognized scheme resolves to a target that's different from the target resolved by another recognized scheme or when a recognized scheme resolves and another doesn't.
fixed as per description
I do not see the need to add sub-sections that duplicate information already present in sibling sub-sections. In this case, section "4.2.5 Resolving an SML reference" clearly defines the sematics associated with reference resolution/model validity. Repeating a part of that information in sibling section 4.2.2 should be avoided.
Re comment #2 The actual problem is that section 4.2.5 Resolving an SML reference duplicates other section in the spec, such as : - section 4.1.3 Unresolved SML Reference Bullet 2.a in 4.2.5 is covered under 4.1.3 >>If none of the recognized schemes resolves, then R is unresolved. - section 4.2.1 At Most One Target Bullet 2.b in 4.2.5 is covered under 4.2.1 >>> When one or multiple recognized schemes in a reference resolve to more than one target then the model is declared invalid. The new section, 4.2.2 Consistent Reference Schemes, introduced under this defect is adding the two cases ( v2.3, v2.4 ) covered under the reference proposal. These two cases are also covered under the 4.2.5 Resolving an SML reference which makes 4.2.5 duplicate of this new section as well. There are two options to solve this duplicate information: Option 1: Remove section 4.2.5 Resolving an SML reference a.Keep 4.1.3 Unresolved SML Reference, 4.2.1 At Most One Target and 4.2.2 Consistent Reference Schemes the way they are now b. The only thing covered by 4.2.5 which is not described anywhere else is bullet 2. If it recognizes R as using N > 0 schemes it supports, then it MUST attempt to resolve R using all N schemes Add this sentence at the beginning of 4.2 : A model validator MUST attempt to resolve an SML reference using all reference schemes of which the reference is recognized as an instance. Option 2: Remove sections duplicated by 4.2.5; which means remove 4.1.3 Unresolved SML Reference, 4.2.1 At Most One Target and 4.2.2 Consistent Reference Schemes Proposal : Apply Option 1 above ( remove section 4.2.5 and add a new sentence to 4.2 ) If this change is applied the spec will look like this ( see also attached doc): 4.2 Reference Semantics A model validator MUST attempt to resolve an SML reference using all reference schemes of which the reference is recognized as an instance. 4.2.1 At Most One Target Every non-null reference MUST target at most one element in a model. When one or multiple recognized schemes in a reference resolve to more than one target then the model is declared invalid. 4.2.2 Consistent Reference Schemes The model is declared invalid when a recognized scheme resolves to a target that's different from the target resolved by another recognized scheme or when a recognized scheme resolves and another doesn't. [Section 4.2.5 Resolving an SML reference - removed]
Created attachment 497 [details] SML spec SML spec where option 1 from comment #3 is applied
Resolved as won't fix since Bug 4992 has superceded this change.
As with any needsReview defects, the fix has been applied to CVS already The content may be updated based on 4992, but this doesn't change the fact that this defect has been applied to the spec. By marking wontfix the requirement is to back out this change ? If not the defect should be moved to fixed and not wontfix ; please clarify
Reopen it as per comment #6, so that we don't lose track of it. Even though text for bug 4992 may cover the same set of changes, this bug is not part of or a duplicate of 4992, and warrants a separate disposition.
This thread is related to this bug report: http://lists.w3.org/Archives/Public/public-sml/2007Dec/0025.html http://lists.w3.org/Archives/Public/public-sml/2007Dec/0035.html
Recommend changing the sentence "References MUST be supported by model validators that conform to this specification" in the proposal reference in Sandy's previous comments (comment #8) as per Ginny's rephrasing in: http://lists.w3.org/Archives/Public/public-sml/2007Dec/0042.html.
To expand on comment #9, the sentence in section 4 Reference MUST be supported by model validators that conform to this specificatio should be revised to: A conforming model validator MUST support SML references as defined in this specification. (SHOULD this now say a "fully conforming" model validator ??)
Fix per comment #10. No need to mark 'needsReview' after editorial change.
fixed as per comment #10 The text reads : Conforming model validators MUST support SML references as defined by this specification.