Hi Darin, I'd say this is definitely a bug. I've fixed a couple of problems in this area, but the whole area of handling pre-get/post-get/post-set methods is not nearly as clean as I'd like. You might try a recent version of the 1.2.2 code from http://www.jibx.org/jibx_1_2_2-SNAPSHOT.zip to see if the particular bug you're encountering has been fixed.
Part of the problem is that object creation and reuse is just not well-structured in the current code, and when there are multiple layers of class hierarchies involved the code loses track of the methods that should be applied to the "current" type of the object. I'm once again trying to push forward with the JiBX 2.0 code (as soon as 1.2.2 is out!). As part of that I'm revising the binding structure and clarifying exactly how these options will work, including for cases of multiple levels of mapping. - Dennis Dennis M. Sosnoski XML and Web Services in Java Training and Consulting http://www.sosnoski.com - http://www.sosnoski.co.nz Seattle, WA +1-425-939-0576 - Wellington, NZ +64-4-298-6117 Darin Clark wrote: > I'm seeing some rather strange behavior in JiBX with regards to abstract > mappings, post-set and its ilk, and the '<structure map-as="foo"/>' > construct. The oddness isn't described in the docs, so I figure I'll check > on the list to see if it's a bug or just a surprise. In short, post-set is > sometimes called, and sometimes not.. An example: > > Class: > > public class TestClass > { > public String one; > public String two; > public String three; > > public void postSet() > { > three="three"; > } > > public static void main(String args[]) throws Exception > { > String testString = "<TestClass></TestClass>"; > IBindingFactory factory = > BindingDirectory.getFactory(TestClass.class); > IUnmarshallingContext c = factory.createUnmarshallingContext(); > TestClass tc = (TestClass) c.unmarshalDocument(new > StringReader(testString)); > System.out.println(tc.three); > } > } > > Binding: > > <binding > > <mapping class="TestClass" abstract="true" type-name="level-one" > > <value name="one" field="one" usage="optional"/> > </mapping> > > <mapping class="TestClass" abstract="false" post-set="postSet" > name="TestClass"> > <structure map-as="level-one"/> > <value name="two" field="two" usage="optional"/> > </mapping> > </binding> > > In this case, the post-set method is called - the output will be "three". > > That's good, I suppose... but... let's mix it up a little. > > Remove the 'two' value element from the second mapping and run the example > again. Result? The postSet method is _not_ called, and 'three' remains > null. > > Of course, we can always just move the post-set attribute up to the abstract > mapping, and it gets called, once again. Not that this helps, if my entire > reason for creating a concrete base mapping is to use a post-set that wouldn't > be used against the abstract class. Joy? > > So.. it appears that the concrete mapping only calls its own post-set if > there's something other than a 'map-as' present for mapping? > > I admit I'm not entirely sure just what's going on here. I can recognize > it, and I can see a(n ugly) way to work around it, but...Is this a bug, or > just something I should know to watch for? > > ------------------------------------------------------------------------------ The Planet: dedicated and managed hosting, cloud storage, colocation Stay online with enterprise data centers and the best network in the business Choose flexible plans and management services without long-term contracts Personal 24x7 support from experience hosting pros just a phone call away. http://p.sf.net/sfu/theplanet-com _______________________________________________ jibx-users mailing list jibx-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/jibx-users
