If by "this" you mean to deal with a user defined exception, I attached the AxisFaultAdaptor for you to look at.
The AxisFaultAdaptor expects the Exception to have been registered with type mapping; which we do in the WSM and it uses that and the "correct" (based on the WSDL definition) element to serialize the content under. The question I guess is do you see the user defined exception without serialization as a use case for Axis? Or does axis depends on the types and faults to have their own serialization (or in helper classes) in which case this run time feature won't be of use to an Axis user. Daryoush > -----Original Message----- > From: Davanum Srinivas [mailto:[EMAIL PROTECTED] > Sent: Tuesday, June 07, 2005 6:49 PM > To: Beehive Developers > Subject: Re: beehive-717 and AxisFaults > > Daryoush, > > do you have a suggestion/diff on how to fix this in axis? > > thanks, > dims > > On 6/7/05, Daryoush Mehrtash <[EMAIL PROTECTED]> wrote: > > The bug in the JIRA is for Axis to properly serialized AxisFaults, that > > is the AxisFualts that Axis would generate if you do wsdl2Java or if you > > write your own AxisFaults. > > > > Until AxisFault serialization is fixed, you would not be able to throw > > an AxisFault subclass from your JWS and except a correct serialization. > > There is noting in our code that we do to work around this problem, we > > are expecting Axis to fix its problem. > > > > The AxisFaultAdaptor does something else. It is an Adaptor for the user > > defined exception to make it look like a fault that is generated from > > Axis source code generation tools. Because it is in the control of the > > serialization it is not making the same errors as the generated > > AxisFault does today. > > > > FYI, if you look at the AxisFault class the > > AxisFault.makeFault(Exception) ends up in the following method. > > Essentially the only thing they do on a user defined exception is > > "toString()" and not a real serialization as AxisFaultAdaptor does. > > > > /** > > * fill in soap fault details from the exception, unless > > * this object already has a stack trace in its details. Which, > > given > > * the way this private method is invoked, is a pretty hard > > situation to ever achieve. > > * This method adds classname of the exception and the stack trace. > > * @param target what went wrong > > */ > > private void initFromException(Exception target) > > > > Hope that helps. > > > > Daryoush > > > > > > > -----Original Message----- > > > From: Eddie O'Neil > > > Sent: Tuesday, June 07, 2005 4:29 PM > > > To: Beehive Developers > > > Subject: Re: beehive-717 and AxisFaults > > > > > > > > > Right, so, just to be clear then, AxisFaultAdapter is intended to > > > work around the serialization problem with faults in Axis. > > > > > > Correct? > > > > > > > > > > > > Daryoush Mehrtash wrote: > > > > No. If the default AxisFault would have worked we would not have > > had > > > > the serialization issue. Please read my comments in JIRA/code. > > > > > > > > Daryoush > > > > > > > > > > > >>-----Original Message----- > > > >>From: Eddie ONeil [mailto:[EMAIL PROTECTED] > > > >>Sent: Tuesday, June 07, 2005 3:39 PM > > > >>To: Beehive Developers > > > >>Subject: beehive-717 and AxisFaults > > > >> > > > >>Daryoush-- > > > >> > > > >> Another thought occurred to me regarding the AxisFault fix in JIRA > > > >>717 -- for the beginning of the discussion, see here: > > > >> > > > >> http://issues.apache.org/jira/browse/BEEHIVE-717 > > > >> > > > >>Instead of creating a new AxisFaultAdapter, could we just call the > > > >>AxisFault.makeFault(Exception) method which according to the Javadoc > > > >>here: > > > >> > > > >> > > > > > > > > > > http://ws.apache.org/axis/java/apiDocs/org/apache/axis/AxisFault.html#ma > > > > ke > > > > > > > >>Fault(java.lang.Exception) > > > >> > > > >>would just return the AxisFault or wrap if needed? > > > >> > > > >>Eddie > > > > > > > > > > > > > > > > > > > > > > > > > > > > -- > Davanum Srinivas - http://webservices.apache.org/~dims/
