In the TxText model, a text consists of a double-linked list of spans, containing the text content and its attributes. I wonder how to efficiently map a hierarchical text model (like that of Pillar) to that. Stephan