First, when logging a JIRA for any of the OASIS tests can you PLEASE describe the problem. Just posting a list of fixes tells very little - the failure MAY be a result of Tuscany doing something wrong, not the testcase.
Second, these changes are unacceptable. TestComposite5.composite in Contribution1 is design to be "language independent" - that is, it must not contain references to any artifacts that are specific to an implementation language. So any implementation must be a composite and any interface must be a WSDL. Replacing interface.wsdl with interface.java violates this principle. So does replacing implementation.composite with implementation.java.
The final proposed change of the value for the property makes no sense to me - the fixed value originally present was intended to uniquely identify this composite from other versions of the same composite that exist in the test suite and which might get used if the runtime has contribution resolution mechanisms that violate the spec.
When I run this test against the latest code, the test fails. The failure reported by Tuscany is:
"Component service interface incompatible with implementation service interface: Component = TestComponent1 Service = Service1" in TestComposite5.composite
Let's pick the bones out of this error.
ASM_12005_TestCase specifies that the following Contributions are used in conjunction with the root composite Test_ASM_12005.composite:
"ASM_12005", "Contribution1", "General", "General" + _Lang
which means when running with Java as the language:
"ASM_12005", "Contribution1", "General", "General_Java
Looking at contribution ASM_12005
Test_ASM_12005.composite has a component TestComponent1 that has as implementation "test2:TestComposite5" where the test2 namespace is the "2nd" composites namespace for the test suite, ie.
The sca-contribution.xml file for contribution ASM_12005 has:
<import namespace="http://docs.oasis-open.org/ns/opencsa/scatests/2009032" location="Contribution1"/>
<import namespace="http://test.sca.oasisopen.org/"/> <!-- WSDL namespace -->
...note that there is a specific import for the "test2" namespace explicitly from the "Contribution1" contribution, which is the nub of the test.
OK, so let's turn our attention to contribution Contribution1. It contains TestComposite5.composite, which is declared in the "test2" namespace, but which in turn uses a) the main "test" namespace for an implementation.composite and b) WSDLs for the interfaces it declares (Service1 in both cases).
The sca-contribution.xml in Contribution1 is like this:
<import namespace="http://docs.oasis-open.org/ns/opencsa/scatests/200903"/> <!-- Composites -->
<import namespace="http://test.sca.oasisopen.org/"/> <!-- WSDL files -->
<export namespace="http://docs.oasis-open.org/ns/opencsa/scatests/2009032"/> <!-- Composites 2nd namespace -->
...so here, "test2" is being exported, but "test" and the WSDL namespace are being imported - the design is that these namespaces will resolve to General and General_Java
So far, so good. Where does the reference to "test:TestComposite5" resolve?
Looking in contribution "General", there is no TestComposite5.composite in the "test" namespace.
Looking in contribution "General_Java", there is a TestComposite5.composite in the "test" namespace.
Looking at TestComposite5.composite in "General_Java", it offers a service called "Service1" but which offers the service interface:
Note that this is Service2 NOT Service1
So the real error is that composite TestComposite5.composite in Contribution1 wants an interface Service1 while the TestComposite5.composite it uses from General_Java has an interface Service2. These two interfaces are by design incompatible and as a result, we correctly get the error that Tuscany gives for this testcase as currently laid out....
I'll post separately for a fix....