You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@cxf.apache.org by "Zarar Siddiqi (JIRA)" <ji...@apache.org> on 2007/09/13 23:36:32 UTC
[jira] Created: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
------------------------------------------------------------------------------------------------------
Key: CXF-1003
URL: https://issues.apache.org/jira/browse/CXF-1003
Project: CXF
Issue Type: Improvement
Components: JAX-WS Runtime
Affects Versions: 2.0.1
Environment: Platform independent.
Reporter: Zarar Siddiqi
This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
Thanks,
Zarar
Here's the method up for review:
private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
for (FaultInfo faultInfo : op.getFaults()) {
for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
String ns = null;
if (mpi.isElement()) {
ns = mpi.getElementQName().getNamespaceURI();
} else {
ns = mpi.getTypeQName().getNamespaceURI();
}
if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
&& qname.getNamespaceURI().equals(ns)) {
return mpi;
}
}
}
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Assigned: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
Posted by "Daniel Kulp (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CXF-1003?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Daniel Kulp reassigned CXF-1003:
--------------------------------
Assignee: Daniel Kulp
> WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
> ------------------------------------------------------------------------------------------------------
>
> Key: CXF-1003
> URL: https://issues.apache.org/jira/browse/CXF-1003
> Project: CXF
> Issue Type: Bug
> Components: JAX-WS Runtime
> Affects Versions: 2.0.1
> Environment: Platform independent.
> Reporter: Zarar Siddiqi
> Assignee: Daniel Kulp
> Attachments: BetterCheckForAFaultMessagePartInfo.patch
>
>
> This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
> I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
> Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
> I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
> Thanks,
> Zarar
> Here's the method up for review:
> private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
> for (FaultInfo faultInfo : op.getFaults()) {
> for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
> String ns = null;
> if (mpi.isElement()) {
> ns = mpi.getElementQName().getNamespaceURI();
> } else {
> ns = mpi.getTypeQName().getNamespaceURI();
> }
> if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
> && qname.getNamespaceURI().equals(ns)) {
> return mpi;
> }
> }
> }
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
Posted by "Zarar Siddiqi (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CXF-1003?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12551844 ]
Zarar Siddiqi commented on CXF-1003:
------------------------------------
I've upgraded to 2.0.3 and this issue appears to be fixed. Thanks everyone.
> WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
> ------------------------------------------------------------------------------------------------------
>
> Key: CXF-1003
> URL: https://issues.apache.org/jira/browse/CXF-1003
> Project: CXF
> Issue Type: Bug
> Components: JAX-WS Runtime
> Affects Versions: 2.0.1
> Environment: Platform independent.
> Reporter: Zarar Siddiqi
> Assignee: Daniel Kulp
> Attachments: BetterCheckForAFaultMessagePartInfo.patch
>
>
> This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
> I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
> Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
> I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
> Thanks,
> Zarar
> Here's the method up for review:
> private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
> for (FaultInfo faultInfo : op.getFaults()) {
> for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
> String ns = null;
> if (mpi.isElement()) {
> ns = mpi.getElementQName().getNamespaceURI();
> } else {
> ns = mpi.getTypeQName().getNamespaceURI();
> }
> if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
> && qname.getNamespaceURI().equals(ns)) {
> return mpi;
> }
> }
> }
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
Posted by "Daniel Kulp (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CXF-1003?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Daniel Kulp resolved CXF-1003.
------------------------------
Resolution: Fixed
Fix Version/s: 2.0.3
> WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
> ------------------------------------------------------------------------------------------------------
>
> Key: CXF-1003
> URL: https://issues.apache.org/jira/browse/CXF-1003
> Project: CXF
> Issue Type: Bug
> Components: JAX-WS Runtime
> Affects Versions: 2.0.1
> Environment: Platform independent.
> Reporter: Zarar Siddiqi
> Assignee: Daniel Kulp
> Fix For: 2.0.3
>
> Attachments: BetterCheckForAFaultMessagePartInfo.patch
>
>
> This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
> I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
> Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
> I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
> Thanks,
> Zarar
> Here's the method up for review:
> private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
> for (FaultInfo faultInfo : op.getFaults()) {
> for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
> String ns = null;
> if (mpi.isElement()) {
> ns = mpi.getElementQName().getNamespaceURI();
> } else {
> ns = mpi.getTypeQName().getNamespaceURI();
> }
> if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
> && qname.getNamespaceURI().equals(ns)) {
> return mpi;
> }
> }
> }
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
Posted by "Zarar Siddiqi (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CXF-1003?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Zarar Siddiqi updated CXF-1003:
-------------------------------
Attachment: BetterCheckForAFaultMessagePartInfo.patch
I've submitted a patch which scans for a matching MessagePartInfo for a fault if the type's namespace didn't match.
To reproduce this issue all you have to do is the following:
package com.mycompany.services;
@WebService(targetNamespace="http://mycompany.com/MyService", ...)
public class MyService ... {
public String doSomething() throws WebServiceException {...}
}
package com.mycompany;
@WebFault
public class WebServiceException extends Exception {
...
}
Then try calling it....
Thanks,
Zarar
> WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
> ------------------------------------------------------------------------------------------------------
>
> Key: CXF-1003
> URL: https://issues.apache.org/jira/browse/CXF-1003
> Project: CXF
> Issue Type: Improvement
> Components: JAX-WS Runtime
> Affects Versions: 2.0.1
> Environment: Platform independent.
> Reporter: Zarar Siddiqi
> Attachments: BetterCheckForAFaultMessagePartInfo.patch
>
>
> This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
> I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
> Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
> I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
> Thanks,
> Zarar
> Here's the method up for review:
> private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
> for (FaultInfo faultInfo : op.getFaults()) {
> for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
> String ns = null;
> if (mpi.isElement()) {
> ns = mpi.getElementQName().getNamespaceURI();
> } else {
> ns = mpi.getTypeQName().getNamespaceURI();
> }
> if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
> && qname.getNamespaceURI().equals(ns)) {
> return mpi;
> }
> }
> }
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
Posted by "Daniel Kulp (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CXF-1003?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12548730 ]
Daniel Kulp commented on CXF-1003:
----------------------------------
Much of this SHOULD be fixed for 2.0.3 and I think a little more for 2.0.4 SNAPSHOT.
> WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
> ------------------------------------------------------------------------------------------------------
>
> Key: CXF-1003
> URL: https://issues.apache.org/jira/browse/CXF-1003
> Project: CXF
> Issue Type: Bug
> Components: JAX-WS Runtime
> Affects Versions: 2.0.1
> Environment: Platform independent.
> Reporter: Zarar Siddiqi
> Assignee: Daniel Kulp
> Attachments: BetterCheckForAFaultMessagePartInfo.patch
>
>
> This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
> I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
> Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
> I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
> Thanks,
> Zarar
> Here's the method up for review:
> private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
> for (FaultInfo faultInfo : op.getFaults()) {
> for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
> String ns = null;
> if (mpi.isElement()) {
> ns = mpi.getElementQName().getNamespaceURI();
> } else {
> ns = mpi.getTypeQName().getNamespaceURI();
> }
> if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
> && qname.getNamespaceURI().equals(ns)) {
> return mpi;
> }
> }
> }
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
Posted by "Zarar Siddiqi (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CXF-1003?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12548725 ]
Zarar Siddiqi commented on CXF-1003:
------------------------------------
Any updates regarding this? I'm surprised more people haven't run into this because it seems like you can't use @WebFault without encountering this problem. Am I missing something?
> WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
> ------------------------------------------------------------------------------------------------------
>
> Key: CXF-1003
> URL: https://issues.apache.org/jira/browse/CXF-1003
> Project: CXF
> Issue Type: Bug
> Components: JAX-WS Runtime
> Affects Versions: 2.0.1
> Environment: Platform independent.
> Reporter: Zarar Siddiqi
> Assignee: Daniel Kulp
> Attachments: BetterCheckForAFaultMessagePartInfo.patch
>
>
> This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
> I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
> Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
> I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
> Thanks,
> Zarar
> Here's the method up for review:
> private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
> for (FaultInfo faultInfo : op.getFaults()) {
> for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
> String ns = null;
> if (mpi.isElement()) {
> ns = mpi.getElementQName().getNamespaceURI();
> } else {
> ns = mpi.getTypeQName().getNamespaceURI();
> }
> if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
> && qname.getNamespaceURI().equals(ns)) {
> return mpi;
> }
> }
> }
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
Posted by "Zarar Siddiqi (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CXF-1003?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Zarar Siddiqi updated CXF-1003:
-------------------------------
Issue Type: Bug (was: Improvement)
I think this might even be considered a bug...
> WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
> ------------------------------------------------------------------------------------------------------
>
> Key: CXF-1003
> URL: https://issues.apache.org/jira/browse/CXF-1003
> Project: CXF
> Issue Type: Bug
> Components: JAX-WS Runtime
> Affects Versions: 2.0.1
> Environment: Platform independent.
> Reporter: Zarar Siddiqi
> Attachments: BetterCheckForAFaultMessagePartInfo.patch
>
>
> This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
> I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
> Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
> I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
> Thanks,
> Zarar
> Here's the method up for review:
> private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
> for (FaultInfo faultInfo : op.getFaults()) {
> for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
> String ns = null;
> if (mpi.isElement()) {
> ns = mpi.getElementQName().getNamespaceURI();
> } else {
> ns = mpi.getTypeQName().getNamespaceURI();
> }
> if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
> && qname.getNamespaceURI().equals(ns)) {
> return mpi;
> }
> }
> }
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (CXF-1003)
WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it
also has private visibility
Posted by "Daniel Kulp (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/CXF-1003?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12528496 ]
Daniel Kulp commented on CXF-1003:
----------------------------------
More thoughts at:
http://www.nabble.com/Re:-Thoughts-on-CXF-1003-tf4469823.html
> WebFaultOutInterceptor.getFaultMessagePart() might be too draconian and it also has private visibility
> ------------------------------------------------------------------------------------------------------
>
> Key: CXF-1003
> URL: https://issues.apache.org/jira/browse/CXF-1003
> Project: CXF
> Issue Type: Bug
> Components: JAX-WS Runtime
> Affects Versions: 2.0.1
> Environment: Platform independent.
> Reporter: Zarar Siddiqi
> Attachments: BetterCheckForAFaultMessagePartInfo.patch
>
>
> This implementation of this method is a little draconian (at least to me). It forces a match on the localPart and the namespace of MessagePartInfo and the fault QName before returning the MessagePartInfo. In the case of where the exception class annotated with @WebFault (default namespace so it "inherits" the services) is also the fault bean (i.e.: no getFaultInfo method specified) the name of the namespace URI will never match because only mpi's that are NOT elements are going to be considered. And since the fault class is a java class, it's corresponding MPI's getTypeQName() will always return the namespace corresponding to the package name and NOT the real namespace of the exception (which really lies behind getElementQName()).
> I propose changing this method so that if no MessagePartInfo is found using the current method, we loosen the criteria and return a match based on the elementQName. This will also require modification of the MessagePartInfo class.
> Now the alternate is to make this method public or protected so it can be overridden. But that's not really a solution.
> I'm working on a patch but before I upload it I was wondering what the CXF team thought of this
> Thanks,
> Zarar
> Here's the method up for review:
> private MessagePartInfo getFaultMessagePart(QName qname, OperationInfo op) {
> for (FaultInfo faultInfo : op.getFaults()) {
> for (MessagePartInfo mpi : faultInfo.getMessageParts()) {
> String ns = null;
> if (mpi.isElement()) {
> ns = mpi.getElementQName().getNamespaceURI();
> } else {
> ns = mpi.getTypeQName().getNamespaceURI();
> }
> if (qname.getLocalPart().equals(mpi.getConcreteName().getLocalPart())
> && qname.getNamespaceURI().equals(ns)) {
> return mpi;
> }
> }
> }
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.