Thanks Rickard, I am not upgrading to the latest version exactly for that reason... we already had numerous emails (on the list) regarding that issue. I am glad it is finally fixed and I will not need to maintain my patched one anymore.
Philippe Le 26 nov. 2009 à 12:09, Rickard Öberg a écrit : > Hi, > > I've found an issue with ValueComposite serialization. Basically, if you have > a base ValueComposite type Foo and then subclass it with CustomFoo, then if a > field references the base type Foo then that is what is going to be used for > serialization/deserialization, even if the actual value is a CustomFoo. > Basically the Value will be downgraded to Foo, no matter what. > > The solution I'm going with now is to check during serialization whether the > value actually is Foo, and if not, then I add a JSON field "_type" which has > the classname of the value. During deserialization I check for "_type", and > if available I use that. > > I've added checks to ValueSerializationTest for this, and it seems to work > ok. Does this sound like an ok fix to everyone? The normal case is still not > to include any type info during deserialization. It's only if the expected > type doesn't match the actual one that the _type will be included. > > /Rickard > > _______________________________________________ > qi4j-dev mailing list > [email protected] > http://lists.ops4j.org/mailman/listinfo/qi4j-dev _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
