XMLs descriptive nature is often in conflict with on-the-wire brevity.  Additionally while the descriptiveness is often useful to those new to a vocabulary, experts in a field often develop abbreviations for those things they refer to frequently, and often verbosity is unhelpful for the experts.  It would be helpful if XML schema could cater for both audiences, and this presents a dichotomy for vocabulary designers.

For example, in the H.323 voice of IP protocol there is something called a "TerminalCapabilitySet", but most experts in the field will call it a "TCS".  When defining a schema for this you have a choice of the XML instance being:

    <ns:TerminalCapabilitySet>...

or:

    <ns:tcs>...

In many respects the latter has advantages because it reduces the size of the XML, is easier to type text messages, arguably easier to read and is also in the language that the experts use.  However, when defining an element in schema such as:

    <xs:element name='tcs'...

is not particularly helpful to those learning a vocabulary, and in this case the longer name would be helpful.

Therefore it is proposed that elements, attributes and also model groups (see later) have added to them a "longName" component.  This would be an optional NCName that could be more descriptive than the "name" component.  Thus the above definition woul appear as:

    <xs:element name='tcs' longName='TerminalCapabilitySet'...

(Note that the XML instance in this case would always be <ns:tcs> and never <ns:TerminalCapabilitySet>.)

It might be suggested that such a long name could be included in the xs:documentation.  However, it helps a human schema reader to have such information in a particular place.  It can be used to suggest to developers variable/class names in software so that someone looking at the schema can have a good idea what to look for in the code to see how a particular feature is implemented.  It can also be helpful for software tools to have a specific XML schema component containing a long name that they can use it as a more descriptive name than the regular name.

I mentioned including longName in model groups.  A number of software tools, such as data binding tools turn anonymous model groups into classes; for which they need a name.  Currently this has to be derived using some heuristic rules, or the user has to define one.  This is obviously less than ideal.  If a model group had a longName, in addition to providing some description to a user, it would enable the software to avoid such heuristics.

I think this addition is very easy to add.  Just add it to the XML Representation Summary and XML Mapping Summary for elements, attributes and model groups.  It has no effect on other aspects of schema.  

In fact it's so orthogonal I've listed it as a change to both XSD 1.0 and 1.1 in the hope it can be added to XSD 1.0 in some future revision. 

Thanks.