The schema spec requires that all parts of a schema from the target
namespace of a given schema be present in that schema document or
<include>d.  A location is required in the <include> element because
there may be more than one.  The normal externalSchemaLocation
mechanism would not allow multiple items for the same namespace, but
the honour-all-schemaLocations feature when set to true allows
multiple settings for a given namespace all to be processed.  It is
not clear from the docs whether the externalSchemaLocations are
consulted for <include> processing.

The spec also requires that if a different namespace is referenced
from the given schema, that that namespace be explicitly <import>ed.
In this case a location is optional; the application is allowed to use
an entity resolver or some other mechanism to associate a namespace
with its component declarations.

If a location is provided for an imported namespace, the normal
behavior is for the location to be treated as a hint that the
application is free to ignore if it has another source (such as a
local copy) for the declarations.  This is also important for the sake
of security and verifiability of contracts, etc.  It is not clear to
me how the Xerces honour-all-schemaLocations interacts with this
aspect of the spec.

Jeff

On 12/7/06, Caleandro <[EMAIL PROTECTED]> wrote:
Hi Michael,

Yes, I have only the Xerces 2.8.1 of xercesimpl.jar on my machine.

Actually the matter is: can I use "external-schemaLocation" to locate all
the schemas needed for validation INSTEAD of including some of them in a
whole main schema by means of schema- include end/or import? How can I reuse
in that way some common portions of schemas of a unique namespace?
To do this must I specify  honour-all-schemaLocations feature in Xerces, if
yes I not yet got it.

I would not have to specify include and/or import for common components in
all the schemas, but if possible reach the same target from the outside of
the schema and message instance (by means of Xerces features and
properties).

The above questions seems to be linked to another problem; the soap-envelope
schema says that some of his elements like body, fault and headers have the
processContents attribute set to "lax"; and this is right but, I need to
validate a soap message whose parts (particularly the body) are defined in
many schemas or namespaces, but Xerces seems to take in count only the first
schema one declare. I had to modify the "lax" in "strict" to validate
against compound schemas.
So I remain with some uncertainty.

Thank for your help.
Vito


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to