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]


"Thomas B. Passin" wrote:

> Marcus Carr wrote -
> >How do you maintain the currency of the declarations? What do you do if the
> >schema changes? Reprocess the instances either to include the modified
> >attribute value or by changing the declaration?
>  These are good questions but  they apply even if the schema is included and
> processed.  If I change a schema so as to change a defaulted attribute value,
> what should you do about those documents I've already processed?  The question
> is no different.

True - I guess it comes down to the nature of the change. If the defaulted
attribute was to be applied to all documents, I'd favour leaving the information
in the schema - if it was to be applied to documents from this point on, it would
make more sense for the documents to carry the information. Really, neither way is
going to be very satisfactory in every case.

> >I don't think it's appropriate to let the instance dictate terms to the
> >system.
> What, you mean it's better to leave information out of a document and require
> a processor to refer to another document to recover it?

In some cases, yes. I'd really prefer to have another layer controlling which
schema gets applied to which instance and when.

> >For one thing, it may mean that an instance could be required to be processed
> >against a schema, even though in the context of the current process, type did
> >not matter and no schema had been provided.
> I thought I was favoring exactly this possibility.  If you didn't think so,
> then I wasn't clear.  Sorry.

It's probably me who wasn't clear - if you think of an instance as containing
information relevant to several processes, it may be unduly restrictive to
reference any particular schema. It would be preferable to call a schema expressly
based on the purpose of the process. My understanding is a bit sketchy, but
perhaps RDDL would be useful for doing the dereferencing?


Marcus Carr                      email:  mrc@allette.com.au
Allette Systems (Australia)      www:    http://www.allette.com.au
"Everything should be made as simple as possible, but not simpler."
       - Einstein