OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

The Division of Schema - A Practical Perspective (Re: Are we losingout because of grammars?)

Eric van der Vlist wrote:

> Looks like we are going towards 4 different types of XML "schema"
> languages to define:
> - The structure
> - The datatypes
> - The rules
> - The semantic.
You made an excellent point.

Let me use a project I am working on to illustrate this division.
Although SWIFT has a plan to adopt XML, the legacy SWIFT format is expected
to coexist with whatever new standard under construction. We are undertaking
a project that maps traditional SWIFT standard directly to some "interim"
XML format.
DTD is sufficient to express the structure:
<!ELEMENT message (header1, header2, header3?, body, trailer?)>
<!ELEMENT body (field*)>
<!ELEMENT field (fieldline*|subfield*)>
XSL transformation rules are based on processing on these structural elements
The Datatypes
When it gets to datatypes, then schema definitions have to be used for the
In plain English, a SWIFT message rule is expressed as something like:
"If the value of field 17U is 'Y', then field 15D must exist."
I write it in XML Schema form:
<complextype name='F17URule'>
<any maxOccurs='*'/>
  <any maxOccurs='*'/>
  <element name='FIELD' tag='17U'>
      <element name='FIELDLINE' fixed='Y'/>
  <any maxOccurs='*'/>
  <element name='BLOCK'>
      <element name='FIELD' tag='15D'/>
      <any maxOccurs='*'/>
  <any maxOccurs='*'/>
We relax the schema to allow multiple types to apply the rules. This way, we
can limit ourselves to only XML Schema standard. A preprocessor can process this
into an enormous standard schema by using union. I am a compiler guy, so I
prefer writing my rules in grammar approach and apply preprocessing for
standard compliance.
Can anyone show this rule in other forms? Please criticize my approach.
The definition of semantics can start a philosophical discussion. However,
we are computer people, so we only need semantics useful for computer
Definition: Semantics. All the basic relationships involving an object, that
are sufficient to deduct all the relationships required for computing tasks
about this object.
In our specific case, a SWIFT message is used to relate to other SWIFT
messages of the same or different types. A collection of the match, amendment,
cancellation, and field correlation rules among SWIFT messages make up
the semantics of the SWIFT messages.
Jay Zhang
IntermicsTech, Inc.