Re: [emergency] EM brouhaha (aka Issue #26)

["R. Allen Wyke" <emtc@nc.rr.com>]

On Thu, 2003-12-18 at 14:52, Rex Brooks wrote:
> So, now I have to change my mind publicly and say that I think it 
> should stay "qualified" so that locally cited elements have to 
> validate against the CAP Schema. 

I did some reading, and this is actually not the issue - whether it
validates or not that is. The validation of the document is independent
of the qualification. However, this does change "how" it is/would be
validated, so it would impact how an instance document would be written.
Here are the places I reviewed. The second one should really help if you
read through the entire thread (4 messages):

* http://www.w3.org/TR/xmlschema-0/#NS
*
http://lists.w3.org/Archives/Public/www-xml-schema-comments/2001AprJun/0169.html

Let me see if I can explain my understanding.

Couple of things first...just to make sure we are on the same page.

1. XML Schema allows you to define more than 1 root element. For
instance, we could have defined both <alert> and <status> on the same
top level. As it stands, <status> is a child of <alert>. 

2. These top level element (or attribute) definitions are called
"Global", while child elements are called "Local" - as you reference in
your email.

3. XML Schema also allows you to import definitions from other schemas.
To help bring us together on this, reference the conversations we have
had about this in regards to a longer term definition of incident types
for example, where we define the types in one schema, and import them in
to all of our efforts. Benefit is that you only define in 1 place,
rather than across multiple.

When elementFormDefault="qualified", this basically means you either
have to explicitly qualify each element using a namespace prefix for all
elements, such as first including
xmlns:cap="http://www.incident.com/cap/1.0"; where you would then see
things like <cap:status>, OR you can implicitly qualify all elements by
providing a default namespace on the uppermost element (<alert> in our
case) in the instance document - ie:
xmlns="http://www.incident.com/cap/1.0"; where you would just see
<status>.

It appears to me qualification basically controls how an author of an
instance document would expect the document to be validated when there
are either 1) multiple global elements or attributes are defined, OR 2)
there is another schema imported. So, in regards to CAP 1.0, which only
has 1 global element (<alert>) and does not import, I would propose we
address the comment by simply adding something like the following -
maybe in the Implementation Notes section (3.3):

"In order to ensure the correct determination of the appropriate
namespace in an instance document, and as defined in the schema in
Section 3.4, all CAP elements MUST be qualified."

Allen

-- 
R. Allen Wyke
Chair, OASIS Emergency Management Technical Committee
http://www.oasis-open.org/committees/emergency