Unmarshalling thread safe problem ?

Hello,

I am using Zeus 3.0.1. I am concerned by a non thread-safe behavior of un=
marschalling. I can't find any documentation saying if it fixed in a new =
version or it is a well known bug.

I wrote a current multi threads test where each thread does 1000 times th=
at following code :

	Unmarshaller unmarshaller =3D new Unmarshaller();
	unmarshaller.setJavaPackage("com.dexiabil.ginko.model.message.ginkoreque=
st.implementation.zeus");
	Source src =3D new StreamSource(new StringReader(XmlMessage));
	UnmarshalledObject unmarshalledObj =3D unmarshaller.unmarshal(src);
	result =3D (GinkoRequest) unmarshalledObj.getObject();

Frequently, some unmarshallings failed and i got that error "java.io.IOEx=
ception: Error in accessing mutator on new instance of class." .
As far as i understand Java, i just surround this code by a synchronized =
statement on a final Object. And it seems to work fine and better.

My only question is : am i right or am i completly mad or am i leading my=
self astray ?

S=E9bastien,


---------------------

An electronic message is not binding on its sender. =20
Any message referring to a binding engagement must be confirmed in writin=
g and duly signed.
---------------------