Schedulability analysis of hierarchical real-time embedded systems involves defining interfaces that represent the underlying system faithfully and then compositionally analyzing those interfaces. Whereas commonly used abstractions, such as periodic and sporadic tasks and their interfaces, are simple and well studied, results for more complex and expressive abstractions and interfaces based on task graphs and automata are limited. One contributory factor may be the hardness of compositional schedulability analysis with task graphs and automata. Recently, conditional task models, such as the recurring branching task model, have been introduced with the goal of reaching a middle ground in the tradeoff between expressivity and ease of analysis. Consequently, techniques for compositional analysis with conditional models have also been proposed, and each offer different advantages. In this work, we revisit those techniques, compare their advantages using an automotive case study, and identify limitations that would need to be addressed before adopting these techniques for use with real-world problems.