Difference between cudaGraphicsRegisterFlagsWriteDiscard and cudaGraphicsMapFlagsWriteDiscard flags?

Can an NVIDIA OpenGL/CUDA interop wizard help me understand if a write discard flag passed to cudaGraphicsResourceSetMapFlags() is redundant if cudaGraphicsRegisterImage() has also been passed a write discard flag?

cudaGraphicsGLRegisterImage    (..., ... | <b>cudaGraphicsRegisterFlagsWriteDiscard</b>);
cudaGraphicsResourceSetMapFlags(...,       <b>cudaGraphicsMapFlagsWriteDiscard</b>);

The documentation is pretty thin on this subject.