[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: XSD substitution groups, subtypes, nillable .. thewhole sm

• To: xml dev <xml-dev@l...>
• Subject: Re: XSD substitution groups, subtypes, nillable .. thewhole smack
• From: Soren Kuula <dongfang@d...>
• Date: Fri, 24 Feb 2006 17:51:56 +0000
• User-agent: Mozilla Thunderbird (X11/20050322)

Hi,

Ok, a little more clarification:

My ad hoc notation

 >There is a declared element with name x, with type b, and a required 
attribute xsi:type with values in {b} and >an optional attribute xsi:nil 
with values in{false}

cannot be expressed in XML Schema, because the declarations resulting 
from my translating away type inheritance would result in all sorts of 
violations:

- Translation could have more than one top level element of the same name
- Translation of local declarations could violate EDC
- Translations declare XML Schema Instance attributes.
- And maybe more.

Im just trying to express the simple, single-type** language struggling 
to get out from within XML Schema. My idea was, again, that:

-An element declared with a type extended from another type really is 
two or more declarations, with the same name, and with different types. 
They are made distinguishable, keeping the schema single-type, by 
requiring the xsi:type attribute

-An element declared nillable really is two declarations with the same 
name, one with the declared content, and one with empty content. They 
are made distinguishable, keeping the schema single-type, by requiring 
the xsi:nil attribute

-The above two combine, although I'm not quite sure how.

My ad hoc'er

 >There is a declared element with name x, with type b, and a required 
attribute xsi:type with values in {b} and >an optional attribute xsi:nil 
with values in{false}

is, in pseudo XSD: XSD without type extension, without nillable and with 
explicit declaration of xsi attributes:

<element name="x">
 <complexType>
   <complexContent>
     <sequence>
       <!-- content of type a in original schema-->
       <!-- content of type b in original schema-->
     </sequence>
   </complexContent>
   <attribute name="xsi:type" use="required">
     <simpleType>
       <restriction base="string">
         <enumeration value="b"/>
       </restriction>
     <simpleType>
   </attribute>
   <attribute name="xsi:nil" use="optional">
     <simpleType>
       <restriction base="string">
         <enumeration value="false"/>
       </restriction>
     <simpleType>
   </attribute>
 </complexType>
</element>

(I might have used fixed instead of the sinlgton type for the xsi:type 
attribute, but to stay consistent, avoid more confusion and trouble with 
declaration of a fixed, optional attribute -- that is impossble, is it 
not? -- I will just stick with singleton simple types).

Hope that clarified,

Thank you for trying to help ;)

Soren


** Single type, but not ancestor typeable. If xsi:type and xsi:nil 
attributes are considered part of the name of the element that holds 
them, then XML schema is ancestor typeable.

• Prev by Date: XSD substitution groups, subtypes, nillable .. the wholesmack
• Next by Date: Re: Mailmen, POST, Intent, and Duck Typing
• Previous by thread: XSD substitution groups, subtypes, nillable .. the wholesmack
• Next by thread: XSD processContents attribute usage question
• Index(es):
  • Date
  • Thread