Alex Milowski wrote: > Although I believe the XML Schema specification is notably silent on the > following > issue, I believe that the XMLEditor product should handle this differently. > > Suppose we have a substitution group amongst three schema target namespaces > A, B, C with the following properties: > > 1. Schema 'A' is the top schema for the document. > 2. Schema 'A' uses schema 'B' and 'C' directly via import statements. > 3. Schema 'B' uses 'C' directly via an import statement. > 4. Schema 'C' knows nothing of 'A' or 'B'. > 5. Schema 'C' defines some kind of fundamental element as abstract and > intends others to use substitution to associate actual elements. > 6. Schema defines a container type that uses this abstract element. > It also > defines an element 'C-container' associated with that type. > 7. Schema 'B' participates in the same substitution structure by > adding its > element 'custom' to the mix. > 8. Schema 'A' defines an element 'A-container' associated with C's > container > type in (6). > > Now, in the current version (2.4), if you go to insert an element in > 'A-container' > you will see 'custom' listed. If you go to insert an element in > 'C-container', you > will not see 'custom' listed. > > I can fix this by importing schema B in schema C--which breaks my model > for who knows > whom. It seems to me that the concept of substitution group--which > transcends individual > schemas--should be consistent within the editor. > > While the current behavior is understandable, I don't think it is > necessarily correct and > it certain stands in the way of developing "mix-in" vocabularies.
It is very hard for me to see where is the problem without some sample schemas and XML instances. May be you have simply found a bug, but in such case, I need to be able to reproduce the bug in order to fix it.
