Bottom-up parsing can be applied to all deterministic context-free languages. On the other hand, left parsing also has some advantages, for example, in error reporting. The left parse of a substring can be pushed on the stack together with the symbol it is reduced to. This method is reviewed in section 2 of this paper, but it does not yield the first production of the left parse until parsing is finished.

Section 3 shows how the missing prefix of the parse could be computed during left-right (LR) parsing if a certain sequence of LR(0)-items were known. The next section defines an automaton, based on the LR(k)-items, which supports the computation of the missing information. The detailed algorithm is presented in section 5. In some situations, the generation of productions may be delayed, but it is resumed as early as possible. This well-structured research paper is a valuable contribution to parsing theory.