This paper explores a research strategy for a uniform and rational reconstruction of AI planning techniques. The strategy relies on two assumptions: (1) classical planners like STRIPS or SNLP are restricted variants of temporal planners like DEVISER, and (2) temporal planners may be best constructed atop a time map manager (TMM). The strategy aims at a reconstruction of timeless, classical as well as temporal systems in a TMM-based architectural framework. However, this paper shows that assumed restricted variants of DEVISER cannot be adequately recast in the TMM framework: this result is shown to hold for classical nonlinear planners like SNLP, and one reasonable extension by possibly simultaneous actions. Hence, in accordance with recent complexity results, this paper calls the intutively appealing research strategy into question.