It is possible to achieve the implementation of generative grammars and parsers of Sanskrit using various methodologies which have varying degrees of affinity to those of Paninian grammar. The current paper compares obvious methods to implement a few aspects of Sanskrit grammar computationally, comments upon the degree to which they approach or depart from Paninian methodology and exemplifies methods that would achieve a closer model. Two questions essential to determining a basic framework in which to implement Paninian grammar computationally are dealt with in some detail: the question of levels and the role of semantics.