You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@cxf.apache.org by "Erik Godding Boye (JIRA)" <ji...@apache.org> on 2015/09/21 14:56:04 UTC

[jira] [Created] (CXF-6598) java.lang.IndexOutOfBoundsException when processing response

Erik Godding Boye created CXF-6598:
--------------------------------------

             Summary: java.lang.IndexOutOfBoundsException when processing response
                 Key: CXF-6598
                 URL: https://issues.apache.org/jira/browse/CXF-6598
             Project: CXF
          Issue Type: Bug
    Affects Versions: 3.1.2, 2.7.17, 3.0.6, 2.7.12
            Reporter: Erik Godding Boye
            Priority: Minor


We are trying to upgrade to a newer version of CXF in a web service client application. Most of our services are running fine on the new version, but one of the services are failing consequently with the following exception (line numbers depending a bit on CXF version tested):
{noformat}
Caused by: java.lang.IndexOutOfBoundsException: Index: 1, Size: 1
	at java.util.ArrayList.rangeCheck(ArrayList.java:653)
	at java.util.ArrayList.get(ArrayList.java:429)
	at org.apache.cxf.message.MessageContentsList.get(MessageContentsList.java:80)
	at org.apache.cxf.jaxws.interceptors.HolderInInterceptor.handleMessage(HolderInInterceptor.java:69)
	at org.apache.cxf.phase.PhaseInterceptorChain.doIntercept(PhaseInterceptorChain.java:307)
	at org.apache.cxf.endpoint.ClientImpl.onMessage(ClientImpl.java:802)
	at org.apache.cxf.transport.http.HTTPConduit$WrappedOutputStream.handleResponseInternal(HTTPConduit.java:1644)
	at org.apache.cxf.transport.http.HTTPConduit$WrappedOutputStream.handleResponse(HTTPConduit.java:1532)
	at org.apache.cxf.transport.http.HTTPConduit$WrappedOutputStream.close(HTTPConduit.java:1330)
	at org.apache.cxf.transport.AbstractConduit.close(AbstractConduit.java:56)
	at org.apache.cxf.transport.http.HTTPConduit.close(HTTPConduit.java:652)
	at org.apache.cxf.interceptor.MessageSenderInterceptor$MessageSenderEndingInterceptor.handleMessage(MessageSenderInterceptor.java:62)
	at org.apache.cxf.phase.PhaseInterceptorChain.doIntercept(PhaseInterceptorChain.java:307)
{noformat}

I have tried to pinpoint exactly where this error originates from, but this seems to be "buried" deep into the implementation details. I would have loved to submit a pull request, but I suspect any suggested change from my part would cause other things to fail.

However, I can try to summarize my findings, so someone with knowledge of the CXF source code hopefully can dig deeper into this:
- I seems to me that the problem is related to the fact that the cxf-codegen-plugin consider the operation to have one *in-out parameter*, and not one in and one out parameter. I do not understand why, but maybe someone with deeper understanding of WSDLs can explain?
- The problem also seems to be related to the fact that we prefer *BARE mode* over WRAPPED mode. If I configure the cxf-codegen-plugin to generate WRAPPED mode Java stubs, it all works. Changing to WRAPPED mode for this particular service seems to be the only workaround I can think of; without changing the WSDL.
- The problem also has something to do with the use of *implicit soap (response) header* in the WSDL. If I remove the implicit soap header from the WSDL, everything works fine.

I have attached a zip-file containing the WSDL in question (simplified and anonymized) and a failing test case. The test case is inspired by the system tests in cxf-systests-uncategorized, and may be added to this module when/if the bug is fixed.

Note: The WSDL may seem a bit strange (I did not make it), but it appears to be valid (at least SoapUI says it is). We are just implementing a client application to a legacy system, so changing the WSDL is not an option (unless proved invalid....)

I have tested a few different versions of CXF, and the problem seems to have appeared in version 2.7.12 (works with 2.7.11). I am not sure what changes that may have caused this to break, but I suspect changes related to CXF-5676. This particular issue is the only one in 2.7.12 (that I could find) related to implicit SOAP headers.




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)