[
Lists Home |
Date Index |
Thread Index
]
Henry S. Thompson wrote:
>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>Thanks for clarifying -- before going further in constructing a reply,
>we need to fix something:
>
>Given
>
>
>
>><complexType name="a">
>> <!--- some model --->
>></complexType>
>>
>><complexType name="b">
>> <extension base="a">
>> <!--- some model -->
>> </extension>
>></complexType>
>>
>>
>
>then
>
>
>
>><element name="x" type="a" substitutionGroup="y"/>
>>
>><element name="y" type="b"/>
>>
>>
>
>is backwards, i.e. not allowed as written. For x to be in the
>substitution group of y, x's type must be or be derived from y's type,
>but in this case it's the other way around.
>
>Does this invalidate the rest of your questions about this example, or
>should we just reverse 'a' and 'b' in this example, or . . .?
>
>
The other examples do not have this feature. At least, they were not
intended to, and they are not supposed to depend on anything defined in
that example (scenario 2)
The question in scenario 2 can be reformulated to:
Given the declarations, what is the case?
- The schema is plain "statically" invalid, and should be rejected as such
- The schema is a valid XML Schema, but no instance document that
substitutes an y-element by an a-typed x-element will validate
- The schema is a valid XML Schema, but no instance document that
substitutes an y-element by any x element will validate
- The schema will validate a document where an a-typed x element appears
in place of an y element
- Something else?
I have a hard time reading the intended meaning out of the spec, except
that I have a fairly good idea that the 4th option is not the right one :)
If this particular example is too much trouble, we can leave it for
later, and take the others first.
Meanwhile, I begin to realise that I have probably misunderstood
something about the final / disallowed substitutions is related to
- type substitutions, not only substitution groups
- substitution by substitution group members can be switched on or off
entirely with final / disallowed substitutions
Substitution group exclusions (aka block) limits the types of
substutiteable elements (but it does not restrict anything that does not
have to do with substitution groups)
--- anyway I think I know of a way to squeeze that semantics into my
"many constructs, single declarations --> few constructs, many
declarations" simplification.
Thanks again
Soren
|
