|
From: Ronald v. K. <rv...@ab...> - 2003-04-03 18:07:27
|
It seems that sun is not following the specs: http://nagoya.apache.org/bugzilla/show_bug.cgi?id=18050 SOAPHeader requires child elements to be of type SOAPHeaderElement. If we do not change this, AXIS will never work (I think). Ng Chi Yuen [Cyng] probeerde het volgende duidelijk te maken op 03-04-03 12:43: > Hi, > > > I understand why AXIS throws an exception, but the way they check the > > instance type isn't a bad thing to do. > Seems required even, see above > Yeah... it's not a bad thing. I just wonder we may somehow be > tempted > to create a SOAPElement instance directly using one API method and add > it to > SOAPHeader but get exception finally. > Seems a good idea. Using the SOAPEnvelope to create a SOSOAPElement, like is done done now, caused us some trouble in our application... we use an elementFactory now > > > I'm still figuring out why the Sun implementation throws an > exception when > > a SOAPHeaderElement is passed to it (via a cast). Then it would be > possible > > to have the ExtensionElement not only return a SOAPElement, but a > > SOAPHeaderElement when the ExtensionElement is an instance of > MessageHeader. > > This is also a good idea. Maybe, I should be more precise in > ExtensionElementImpl, i.e., the construction of ExtensionElementImpl > should > call addHeaderElement() and addBodyElement() instead of SOAPFactory. > createElement(). > Depending on the instance type you mean? Doesn't this require a rewrite of a large part of ExtensionElementImpl? Any tips? then I'll be willing to look into this > > If you have more information on your problems with axis, we can prevent > > doing duplicate work. I'll draw up a list with our problems tonight > > (CET+2). > > Thanks a lot for your investigation. Please go on. > Other issues we ran into when moving van sun to axis: - SOAPEnvelope.createName had some problems with namespaces (should look how we solved this) - SOAPFactory.newInstance().createElement(....) Now done with a elementFactory...(works in axis, not tested in sun) - ..... Regards, Ronald |