This document discusses different types of traceability in model transformations and the challenges of traceability for imperative transformations. It defines traceability as recording the inputs, outputs, and execution of individual computations or rules. For declarative transformations, traceability is relatively straightforward, but imperative transformations pose challenges because objects are mutable and require maintaining multiple states. The document concludes that while a shared trace metamodel could support traceability between tools, imperative transformations are generally impractical for arbitrary transformations due to the state cloning required.