[legalxml-comment] Draft UBL Naming Standards

[John McClure <hypergrove@olympus.net>]

Hello,
As mentioned in a previous posting, I believe LegalXML needs to revisit -- from an organizational view -- the naming standards we
REQUIRE going forward. I stated before it's a good idea to follow UBL, since it's not only the XML implementation of ebXML (now
UN/CEFACT) but it's an OASIS working group... In this regard, I've attached an early version of the UBL naming standards (the Arofan
paper) for XML elements and attributes. There, the policy is that all UBL element names for object classes are to be in
UpperCamelCase with "leaf elements" in lowerCamelCase. This stands in contrast to LegalXML's current naming convention, which I
understand is that _all_ elements are lowerCamelCase.

The Arofan paper explicitly calls for a "Dictionary" while it raises issues beyond naming standards worth considering, in particular
XML element/attribute names v. object/property names. As you know, this is the arena that I have been looking at closely for awhile
now. However, I have concluded that RDF Schema is a better solution for representing a dictionary because RDF supports multiple
inheritance while XML Schema does not. XML Schema is somewhat comparable to going back to a world organized around relational
tables, albeit their source data is arranged hierarchically. In effect, the UBL grafts OO onto XML Schema, while RDF starts with OO
classes and properties, and provides the tools to accommodate the requirements met by XML Schema. [You may have noticed that XML
Schema is being bashed universally as "too complex" -- prompting the desire within many quarters to go to the "stripped-down"
language called RELAX-NG.] Because RDF is not trying to define an on-the-wire protocol of elements and attributes, RDF is able to
standardize that attributes are to be viewed the same as those XML elements that represent properties of an object, meaning that
<Elem foo='x'> is the same as <Elem><foo>x</foo> -- this simplifies standards efforts such as ours because it means we only define
data models, not XML protocols, as our primary item of work.

So, while I disagree with UBL's proposal to represent their Dictionary in XML Schema, I do suggest that we look hard at adopting
their naming standards.
Thanks,
John McClure


PS I've posted an RDF dictionary (with an XSL stylesheet for viewing) for 8 document-types at www.dataconsortium.org/documents.html.
The package there includes a nifty (alpha-version) Javascript "DCN Editor" that manages the input of data to any HTML <span>, <div>,
<td>, etc. element -- using the contents of an RDF dictionary to get the information it needs to perform field-level formatting
before and validation after data entry. I am planning this kind of presentation of LegalXML material, albeit using the
UpperCamelCase convention. Also, I've begun the XSL stylesheets that convert an RDF dictionary into either a DTD or an XML Schema.
Comments off-line please.

Attached:
Position Paper: Code Lists (Eve Maler)
Position Paper: Elements versus Attributes (Gunther Stuhec)
Position Paper: Modularity, Namespaces and Versioning (Bill Burcham)
Position Paper: Definition of Elements, Attributes and Types (Mark Crawford, Arofan Gregory, Eve Maler)

Attachment: draft-arofan-tagspec-03_SWIFTcomments.doc
Description: MS-Word document

Attachment: draft-stuhec-elemvsattrib-03SWIFTcomments.doc
Description: MS-Word document

Attachment: PP Namespaces Review FormSWIFT.xls
Description: MS-Excel spreadsheet

Attachment: draft-maler-codelists-04_SWIFTcomments.doc
Description: MS-Word document