Keith-
Okay, the situation is a little clearer now (I do vaguely remember
reading about attributes and namespace inheritance)... This brings me
to the question of how to handle namespaces for attributes in a schema?
This problem comes up when I try to round trip Jitesh's example
(unmarshal, marshal, unmarshal) with validation turned on -- the first
time works fine because in his original xml file the attributes are not
prefixed, but when Castor marshals the objects back out it adds
prefixes, and then on unmarshalling Xerces complains when it tries to
validate against the schema (Attribute 'ns1:name' is not allowed to
appear in element 'logger'). I believe Castor is doing the right thing
with the namespaces, and the original xml file is probably flawed, but I
haven't been able to find the correct combination of xml/schema/mapping
file to keep everyone happy...
At this point I think we're into the "just for future reference"
category, but any advice you can give would be appreciated :)
Thanks!
Stephen
Keith Visco wrote:
I'm skipping most of the conversation as I don't have too much to add,
but here I can explain a bit more:
Stephen Bash wrote:
OH! And looking back through comments I now remember, the reason I had
to be screwy with the mapping file namespaces is to get the mapped
attributes into the right namespace. Otherwise when marshalling,
Castor would put the elements in the correct namespace, but put the
attributes back in the castor namespace... It was weird... There's
probably a better way to do that.
The reason Castor defaults to the "castor" namespace is because Jitesh's
mapping file declares this as the default namespace as such:
<mapping xmlns="http://castor.exolab.org/";
So Castor is using that namespace when determing the proper namespace
for a field when not prefixed.
Namespaces in the mapping file are a bit tricky to work with. We should
probably update the mapping file to use a more explicit approach where
ns-prefix and ns-uri are spelled out everywhere including the bind-xml
elements...it would make it clearer, but it would also make it more
verbose and tedious.
Hope that helps to make it a bit clearer as to how Castor determines the
namespace of a field when not prefixed.
--Keith
-------------------------------------------------
If you wish to unsubscribe from this list, please send an empty message
to the following address:
[EMAIL PROTECTED]
-------------------------------------------------
-------------------------------------------------
If you wish to unsubscribe from this list, please
send an empty message to the following address:
[EMAIL PROTECTED]
-------------------------------------------------
