Lexical analysis and parsing techniques for a vector machine

The use of vector oriented hardware for compilation places a special burden on the compiler writer to make use of vector operations whenever possible. This paper presents various techniques for lexical analysis and parsing using the CDC STAR-100 instruction set. Although the similarity between APL and the STAR is apparent, previous techniques developed by Lincoln and others using APL as a base cannot be used directly on the STAR. Difficulties involved in implementing such methods occur because of the non-uniform treatment of vectors of different types on the STAR as opposed to a very uniform treatment in APL. The relevant features of the STAR-100 are discussed after which techniques are described for performing lexical analysis of a source text. The goal is to reduce the text to a set of tokens of unit length. Finally some parsing methods are suggested.