[dsdl-discuss] Re: Namespaces

There is a potential problem relating to DSDL namespaces. Should there be
one that applies to all parts, or should each part have its own?

Part 2 defines a namespace which is specific for RelaxNG. It also defines a
data element with a type attribute that points to the relevant datatype
spec.

Part 5 will define DSDL specifc datatypes. It will need its own namespace as
the currently proposed RelaxNG one would not be suitable. Part 5 will also
propose some attributes that will be applicable to elements that define data
points within the model. So there will need to be something along the lines
of dsdl-5:property1 that will be added to the various elements defining data
points within the other parts.

The problem is that references to attributes defined within the Part 5
namespace may clash with those defined in other parts. For example, the type
attribute of the RelaxNG data element may need to match to something like
dsdl-5:datatype. We must avoid duplication between sections of the standard.
But at the same time we must ensure that proper management of attributes can
be achieved.

Adopting a single namespace for DSDL would simplify some issues, but would
complicate others. It would mean that each name could only be used in one
part of the standard, which will probably be a pain. But we need to have a
common front, so that people know they are dealing with a single set of
definitions and not multiple sets of disconnected definitions. We also need
to ensure that parts can interwork correctly. To do this we need some form
of shared naming scheme. ISO policy would suggest that it be something that
identifies both the standard and the relevant part number, hence the
suggestion above to use dsdl-2, dsdl-3, etc, to distinguish elements and
attributes taken from the various parts of the standard.

Martin Bryan

--
DSDL members discussion list
To unsubscribe, please send a message with the
command  "unsubscribe" to dsdl-discuss-request@dsdl.org
(mailto:dsdl-discuss-request@dsdl.org?Subject=unsubscribe)