[relax-ng-comment] Proposal for new Examplotron based on RNG

[John Cowan <jcowan@reutershealth.com>]

Here's a proposal for a new version of Examplotron based on translation to
RNG rather than translation to XSLT.  The eg:assert attribute is not
supported.  I don't really understand the eg:placeHolder attribute,
so I am not mentioning it here.

The eg:occurs attribute has the values *, ?, +, and . as in existing
Examplotron: it causes the element to be wrapped in a rng:zeroOrMore,
rng:optional, rng:oneOrMore, or nothing at all.

Elements are replaced with rng:element elements.  Attributes are made
optional (change from existing Examplotron) by wrapping the rng:attribute
element in an rng:optional element.

If an element has character content that is not whitespace, its content
model is wrapped in an rng:mixed element.

However, if an eg:type attribute is present, then the content model of the
element will be rng:data if there is no character content, or rng:value if
there is.  An inheritable eg:lib attribute specifies the database library.

The attribute eg:cons can be "|", "&", or "." and causes the content
model of the element to be wrapped in rng:choice, rng:interleave, or
rng:group respectively.

eg:import is mapped to rng:externalRef.

Every rng:element is given an rng:attribute child which allows any eg:
attribute, so that the Examplotron schema will validate against itself.

There is no support for non-Russian-doll schemas or attribute types.
All attributes are optional (there could be an operating mode or eg:att
flag forcing all attributes to be required).

-- 
With techies, I've generally found              John Cowan
If your arguments lose the first round          http://www.reutershealth.com
    Make it rhyme, make it scan                 http://www.ccil.org/~cowan
    Then you generally can                      jcowan@reutershealth.com
Make the same stupid point seem profound!           --Jonathan Robie