It's actually the case that the first draft that Jerry wrote of the PREMIS
schemas had all the documentation from the data dictionary in it (i.e.,
definition, examples, notes, etc.). After a discussion on the then
PREMIS-schemas review group, we had him take it out because many preferred
leaner schemas (and also it would have taken him much longer to finish the
work on it). But the point that it would be nice to be able to generate
the data dictionary from the XML is an important one; that was mentioned
at the time, but we decided to wait until after the implementation period
to consider whether to go that route. Having one XML document that is used
to generate various views (and I would go for having specific tags for
portions of the data dictionary, e.g. usage notes, definition, etc.) is an
attractive approach. Something to be considered in the future; I wouldn't
do that right now since we don't plan to make changes for the
implementation period (unless there are clear errors).
Any further discussion what are now 3 ways of writing the schema in terms
of mandatory/repeatable tags:
1. don't specify the defaults when they're used, i.e. minOccurs=1
2. do specify the defaults regardless
3. annotate the change section saying what the defaults are when something
else isn't specified.
On Mon, 12 Sep 2005, Richard Davis wrote:
> It's hard to imagine this particular issue arising at all for schemas
> expressed in Relax NG; and I feel it takes me a fraction of the time to
> get to grips with an RNG schema (XML or compact). I often doubt XML
> Schema was designed for or by humans.
> People who don't have a need to work with raw XML Schema every day are
> much better served by versions formatted specifically for human readers
> - like, for example, the OxygenXML schema documenter generates. It's
> only an XSLT transformation away. Ideally, the human-readable version of
> the Data Dictionary would be generated like that, directly from the
> Schema Definition.
> For those who do prefer schemas à la Tartare, don't some XML editors
> have a facility for inserting optional default content automagically?
> I'm sure I've done that in the past in Oxygen and Emacs, though I can't
> now remember how (can anyone remind me?)
> \ Richard M Davis
> / Digital Archives Specialist
> \ University of London Computer Centre
> / Tel: +44 (0) 20 7692 1350
> \ mailto: [log in to unmask]