- From: Asir S Vedamuthu <asirv@webmethods.com>
- Date: Fri, 6 Sep 2002 13:10:15 -0400
- To: "Web Service Description" <www-ws-desc@w3.org>
- Cc: "Martin Gudgin" <mgudgin@microsoft.com>, <prasad.yendluri@webmethods.com>
With permissions from Martin (thank you !!), I am posting this thread to the
WSDesc WG. Please review and address them as appropriate.
Regards,
Asir S Vedamuthu
webMethods, Inc.
703-460-2513 or asirv@webmethods.com
http://www.webmethods.com/
-----Original Message-----
From: Martin Gudgin [mailto:mgudgin@microsoft.com]
Sent: Friday, September 06, 2002 5:20 AM
To: asirv@webmethods.com
Subject: RE: Challenges raised by INLINE Schema Definitions
[inline]
> -----Original Message-----
> From: Asir S Vedamuthu [mailto:asirv@webmethods.com]
> Sent: 05 September 2002 15:35
> To: Martin Gudgin
> Subject: FW: Challenges raised by INLINE Schema Definitions
>
>
> Martin,
>
> What are your thoughts on this? Thank you.
>
> Asir
>
> -----Original Message-----
> From: w3c-xml-schema-ig-request@w3.org
> [mailto:w3c-xml-schema-ig-request@w3.org]On Behalf Of Asir S Vedamuthu
> Sent: Wednesday, August 21, 2002 11:17 AM
> To: 'w3c-xml-schema-ig'
> Subject: Challenges raised by INLINE Schema Definitions
>
>
>
> What is an INLINE Schema Definition?
>
> It is one or more schema elements; is not a root of the
> element tree; embedded within an XML document. Example, a
> schema element as a child of <types> element in a web service
> description.
>
> INLINE schema definitions are very popular and raises many
> challenges. Based on the existing usage patterns, this e-mail
> describes three challenges faced by schema processors. This
> is not an exhaustive list. There are many more challenges
> along these lines ..
>
> I request the schema WG and IG members to discuss these
> issues and issue erratta, publish a note on inline schema
> definitions, OR address these in XML Schema 1.1 version.
>
>
> [1] What is the relationship among multiple INLINE schema definitions?
>
> Current usage pattern: inline multiple schema definitions as
> children of <types> element in a web service description.
> There are references to schema components across schema
> definitions without using <xsd:include> or <xsd:import> statements.
>
> Example is,
>
> <wsd:types xmlns:A="http://www.example.com/A"
> xmlns:B="http://www.example.com/B">
>
> <xsd:schema targetNamespace="http://www.example.com/A">
> <xsd:element name="a" type="xsd:string"/>
> ..
> </xsd:schema>
>
> <xsd:schema targetNamespace="http://www.example.com/B">
> <xsd:element name="b" type="xsd:int"/>
> ..
> </xsd:schema>
>
> <xsd:schema targetNamespace="http://www.example.com/C">
> <xsd:element name="c">
> <xsd:complexType>
> <xsd:sequence>
> <xsd:element ref="A:a"/>
> <xsd:element ref="B:b"/>
> </xsd:sequence>
> </xsd:complexType>
> </xsd:element>
> </xsd:schema>
>
> </wsd:types>
>
> What is the relationship? Is this legal? If so, what are the
> processing rules?
No it's not legal, because there is no import statement in 'C'. However,
the following WOULD be legal
<wsd:types xmlns:A="http://www.example.com/A"
xmlns:B="http://www.example.com/B">
<xsd:schema targetNamespace="http://www.example.com/A">
<xsd:element name="a" type="xsd:string"/>
..
</xsd:schema>
<xsd:schema targetNamespace="http://www.example.com/B">
<xsd:element name="b" type="xsd:int"/>
..
</xsd:schema>
<xsd:schema targetNamespace="http://www.example.com/C">
<xsd:import namespace='http://www.example.com/A' />
<xsd:import namespace='http://www.example.com/B' />
<xsd:element name="c">
<xsd:complexType>
<xsd:sequence>
<xsd:element ref="A:a"/>
<xsd:element ref="B:b"/>
</xsd:sequence>
</xsd:complexType>
</xsd:element>
</xsd:schema>
</wsd:types>
>
>
> [2] AD HOC Built-in Types and Declarations
>
> Per XML Schema 1.0, there are a few built-in types: 'anyType'
> and 'anySimpleType' from Part 1 and built-in simple types from Part 2.
>
> Current usage pattern: certain type definitions and element /
> attribute declarations that are germane to web service domain
> are implicitly treated as built-in types and decls. Here is
> an example,
>
> Example 3, WSDL Version 1.2: Bindings
> http://www.w3.org/TR/2002/WD-wsdl12-bindings-20020709/#_soap-e
>
> <types>
> <schema targetNamespace="http://example.com/stockquote/schema"
> xmlns="http://www.w3.org/2000/10/XMLSchema">
> ..
> <complexType name="ArrayOfFloat">
> <complexContent>
> <restriction base="soapenc:Array">
> <attribute ref="soapenc:arrayType"
> wsdl:arrayType="xsd:float[]"/>
> </restriction>
> </complexContent>
> </complexType>
> </schema>
> </types>
This is illegal too. There should be a
<xsd:import namespace='http://schemas.xmlsoap.org/soap/encoding/' />
>
> In this example, complex type soapenc:Array and attribute
> decl soapenc:arrayType are considered to be built-ins. I
> believe that this sample schema definition is invalid per XML
> Schema 1.0 spec. Because, there aren't any import statements
> for the soapenc namespace,
> http://schemas.xmlsoap.org/soap/encoding/. However, this
> usage pattern demonstrates a need for extensible built-ins.
I disagree, I think people should put the import statements into the
schemas
>
> Today, this usage pattern is widely used and enforced by ad
> hoc groups.
WSDL 1.2 could ( and should ) fix this.
>
>
> [3] Mandatory targetNamespace
>
> Per XML Schema 1.0, targetNamespace is optional. If absent,
> then this schema definition defines and declares components
> that are not considered to be in any namespace.
>
> Current usage pattern: targetNamespace is mandatory for
> inline schema definitions within a web service description.
What makes you think this? Given that all that appears in a 'wsdl:types'
element are <xsd:schema> elements, I don't see how anyone can mandate
targetNamespace without changing the schema spec. Is this a tools issue?
>
> This takes away a huge functionality. I do know that
> implementers have gone to lengths to implement this
> restriction. Let me describe our experience. There are
> several webMethods users that depend on ABSENT
> targetNamespace. To overcome this restriction and support
> these users, we have introduced a special namespace URI in
> markup that represents absent [namespace name] in XML
> Information Sets - http://www.webMethods.com/noNamespace/ and
> this is the targetNamespace :-( Here is a sample,
>
> <xsd:schema targetNamespace="http://www.webMethods.com/noNamespace/"
> xmlns="http://www.webMethods.com/noNamespace/">
>
> <xsd:complexType name="purchaseOrder">
> <xsd:sequence>
> <xsd:element name="shipTo" ..
> ..
> </xsd:sequence>
> </xsd:complexType>
>
> <xsd:simpleType name="SKU">
> <xsd:restriction base="xsd:string">
> <xsd:pattern value="\d{3}-[A-Z]{2}"/>
> </xsd:restriction>
> </xsd:simpleType>
> </xsd:schema>
>
> I believe that making targetNamespace mandatory is a huge
> loss of functionality.
I'm not sure it's huge, but it is certainly a loss!
It is not clear to me why these are issues for the schema WG/IG. Surely
these are issues for the WSDesc WG, no?
Gudge
Received on Friday, 6 September 2002 13:09:56 UTC