I think it is important to recognise that there are three sets of namespace prefixes in play here:
- prefixes for the namespaces used in the xsds for the targetNamespace and other imported namespaces.
- prefixes for the namespaces used in the XML instance documents
- prefixes for the namespaces used in the XPath expression
Each set of prefixes is defined entirely separately, and the prefixes may be completely different.

In the case of an XPath expression, the prefixes must be defined externally and given to the XPath processor in some kind of 'context'. How this is done depends on the XPath processor.

I think the answers to the questions below follow from these facts. The XPath is querying an XML info set. Each item in that info set has a name and a namespace. It may also have a 'preferred' namespace prefix inherited from the XML document that created the info set, but that prefix is never consulted when evaluating an XPath expression. So the algorithm goes like this:
- expand every namespace prefix in the XPath to a full URL using the mappings given to the XPath processor
- match each namespace:name to the namespace:name in the XML info set.
and note that the xsd's namespace prefixes are not in play at all.

regards,

Tim Kimber,
Technical Lead for IBM Integration Bus Healthcare Pack
Hursley, UK
Internet: kimbert@uk.ibm.com
Tel. 01962-816742
Internal tel. 37246742

From: Mike Beckerle <mbeckerle.dfdl@gmail.com>
To: "dfdl-wg@ogf.org" <dfdl-wg@ogf.org>, "Costello, Roger L." <costello@mitre.org>
Date: 17/09/2014 09:37
Subject: [DFDL-WG] referencing local element declarations from path expressions between two schema documents
Sent by: dfdl-wg-bounces@ogf.org

I am looking for clarification about how path steps are to be interpreted with respect to the issue of qualified/unqualified element names, and default or no-default namespaces.

If there is a precedent we can follow set by other tools which embed paths into XSD (e.g., schematron?) or even just the selector/paths things in XSD key/uniqueness constraints then I'd like to understand it.

These are some of the conundrums:

Two Schema Files: A, and B

Schema file A has target namespace AA, and binds the AA namespace to prefix aa, and the BB namespace to prefix bb
Schema file B has target namespace BB, and the same prefix bindings.

Schema B has an element named b1 which has two local element declarations inside it named c1 and c2.

Schema A has an element named a1 which contains an element reference to bb:b1.

a1 has a DFDL annotation (e.g., dfdl:assert) which has a path expression in it. this path contains bb:b1/c1

When is the path correct?

1a) when schema B has elementFormDefault='unqualified' (that's the default), and schema A does NOT have a default namespace. Why? the default namespace - if it existed for schema A - would implicitly qualify the c1 step of the path, but since c1 is a local element in schema B with element form unqualified, it does not require (and cannot have) a namespace specified in path steps for c1.

1b) when schema B has elementFormDefault='qualified', and schema A has xmlns='BB'. Why? Because in this case the path bb:b1/c1 is equivalent to bb:b1/bb:c1, and since c1 will have a namespace-qualified name in schema B, that path will properly describe the path to the c1 element.

1c) Variation: when schema B has no target namespace (which I believe makes elementFormDefault irrelevant), and schema A has no default namespace. Because then c1 can't be qualfied as it lives in no namespace, and schema A has no implied qualifiers.

When is the path incorrect?

2a) when schema A has xmlns="BB". Why? Because the default namespace means the bb:b1/c1 is equivalent to {BB}b1/{BB}c1, but there is no {BB}/c1 since c1 is just a local element declaration, and B has elementFormDefault unqualified.

2b) when schema B has elementFormDefault='qualified', and schema A has no default namespace. In that case the step c1 needs a namespace qualifier either by default, or explicitly via a prefix, but it doesn't have one.

So I am interested in opinions on whether the above analysis is correct or not, and especially in precedent for how this stuff works in other tools.

I have made a big assumption above which is that if xmlns="BB", then any path step inside something carrying that default binding is interpreted as qualified by that namespace. I picked this because it seemed reasonable, and it seems one can always modify one's schema to properly address something in another schema without having to change that other schema. I.e., it seems to have good composition properties.

Thanks for any thoughts.

Mike Beckerle | OGF DFDL Workgroup Co-Chair | Tresys Technology | www.tresys.com
Please note: Contributions to the DFDL Workgroup's email discussions are subject to the OGF Intellectual Property Policy
-- dfdl-wg mailing list dfdl-wg@ogf.org https://www.ogf.org/mailman/listinfo/dfdl-wg

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU