[jira] Updated: (CXF-1198) Element with a refName pointing to an unqualified element that isn't in the same schema.

["Benson Margulies (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/CXF-1198?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Benson Margulies updated CXF-1198:
----------------------------------

    Attachment: diffs

I've written a validator for all the schema references in the service model. If you apply this patch and run the SerializationTest, it will present you with the first of what may be a sequence of issues here.

> Element with a refName pointing to an unqualified element that isn't in the same schema.
> ----------------------------------------------------------------------------------------
>
>                 Key: CXF-1198
>                 URL: https://issues.apache.org/jira/browse/CXF-1198
>             Project: CXF
>          Issue Type: Bug
>          Components: JAXB Databinding
>    Affects Versions: 2.1
>            Reporter: Benson Margulies
>            Assignee: Daniel Kulp
>         Attachments: diffs
>
>
> I made a change to the classes for the Javascript test harness service, and I hit an apparently unrelated problem.
> If you remove the @Ignore from DocLitWrappedTest, and run with -ea, you'll hit an assertion. The assertion is because there is a type with an element with no schema type, no schema type name, and a refName of {}testBean1. Now, there's no such beast. Really. testBean1 derives from the class TestBean1, which looks like ...
> @XmlRootElement
> @XmlType(namespace = "uri:org.apache.cxf.javascript.testns")
> public class TestBean1
> In the past, I hit some cases like this where the XML Schema objects were more or less acting like a faithful recording of what the XML might look like: ref='bloop' for bloop in the TNS. Whether or not that's a good idea, it isn't what's happening this time, in that the type that contains the element that causes this havoc lives in a different namespace than the TestBean1, so in XML the ref would have an actual prefix.
> Dan may already be 'in process' on this issue, I don't know why adding a wrapper type to enable something else entirely uncovered this particular manifestation.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.