On failure of the pruning technique in "error repair in shift-reduce parsers"

A previous article presented a technique to compute the least-cost error re pair by incrementally generating configurations that result from inserting and deleting tokens in a syntactically incorrect input. An additional mecha nism to improve the run-time efficiency of this algorithm by pruning some o f the configurations was discussed as well. In this communication we show t hat the pruning mechanism may lead to suboptimal repairs or may block all r epairs. Certain grammatical errors in a common construct of the Java progra mming language also lead to the above kind of failure.