There exist a large amount of research efforts that claim cross-layer design in multi-hop wireless networks. However, almost all of such works use the idea of cross-layering in some ad hoc manner, and different cross-layer designs generally do not work together. To this end, we argue that cross-layer design in such networks should follow some systematic approach, in order to provide a common ground for alternative comparison and integration. In this position paper, we identify two key components toward achieving this ultimate goal of systematic cross-layer design. First, a rigorous mathematical framework is needed to provide the unifying approach for cross-layering, and in this work we choose the well-recognized Generalized Network Utility Maximization (GNUM) approach. Second, with the components availed by the mathematical framework, a composable software "workbench" is needed to "glue" the various components into some prototype alternatives for inspection. We will discuss how such composability may be achieved via our unique ontology-based modeling and simulation approach.