The document compares different approaches to implementing feature model composition through merge operators, including: 1. Separate feature models with intersection - Provides basic merging but lacks optimality in some cases. 2. AGG - Limited to union mode and struggles with non-trivial examples in intersection mode due to expressiveness limitations. 3. Kompose - Too restrictive due to two-stage local reasoning approach, making constraints and post-conditions difficult. 4. Kermeta - Provides better semantics than Kompose but still has issues with different hierarchies and constraints. 5. Boolean logic - Provides full expressiveness but lacks hierarchy and variability information on its own. An algorithm is needed to reconstruct feature models from