Hi, In the constraint metadata API we currently have two new methods: * getConstraintsForContainerElementType(int) (for returning a descriptor for a specific type argument) and * getConstrainedContainerElementTypes() which returns a descriptor for those type arguments which are constrained or marked for cascaded validation (themselves or nested). This is similar to BeanDescriptor#getConstraintsForProperty() and getConstrainedProperties(). Should we instead have a single method getContainerElementTypes () which returns descriptors for all the type arguments of an element, be them constrained or not? That'd be similar to ExecutableDescriptor#getParameterDescriptors() where we also return a descriptor for each parameter. It'd also allow for better evolution towards array support. That single method would make sense for array types, too, whereas getConstraintsForContainerElementType(int) is dubious. Thus I feel the latter would be a bit better. Thoughts? Thanks, --Gunnar