The Daffodil user community is complaining about the DFDL v1.0 restriction that nillable complex types can only have %ES; (empty data) as their nilled representation.

Multiple important data standards find this restriction quite awkward.

It has significant negative impact on usability because it eliminates polymorphism of paths.

The path needed to refer to an element value has different ending element path steps depending on whether it is nilled or not.

The value of having DFDL create an XML infoset is lost if something as basic as this cannot be done in such a way that path expressions work alike to how they would work in regular XML/XSD.

Is there a significant reason why this restriction cannot be lifted?

If not we would probably plan to define a dfdlx namespace property that can be used to enable this alternate behavior, and then we could prototype this new fuctionality in Daffodil and propose this for DFDL 2.0 revision.