[bt-spec] XML Schema and default values

[Alex Ceponkus <alex@ceponkus.org>]

> > 50: yes .. I agree that the XML spec section needs to be
> updated, but not
> > the XML schema.  I have been trying to only have the basic xml structure
> > reflected in the schema.  For example, in Fault, the
> > superior-identifier is
> > only present if this fault is being sent to a superior .. there
> is no way
> > for that kind of information to be reflected in the schema, so the
> > superior-identifier is just optional.  I find the default
> values fall into
> > this category (however, there is a way to specify default values in the
> > schema, I don't think there is much benefit.)  Perhaps we
> should call the
> > schema 'non-normative' to indicate that the spec body text takes
> > precedence.
>
> Do you mean the default should be at the semantic level but not at syntax
> level. I guess this makes a difference to whether it is in the
> post-validation infoset (at least if it is taken through a validator that
> knows the schema).

(moving discussion that started from a vote comment to the list)

Although I cannot think of a concrete case in the current spec, the schema
will get confusing if we have a situation where the default value might
change depending on what state the transaction might be in or the role that
the sender/receiver has taken.  (I believe this situation existed before the
renaming of the message set work took place .. but I would have to go look
in an old spec to confirm.)  So to have values show up in the
post-schema-validation infoset might cause confusion to somebody
implementing BTP.  Aside .. IMO the schema will be used as a tool for
developers during development, but not at runtime for the actual validation
of the BTP messages that are being received (the same way that SOAP
implementations don't use the SOAP Envelope schema at runtime.)

Alex