Predictive program synthesize

  • Authors:
  • Raghavendra Rao Loka

  • Affiliations:
  • Department of Computer Science, Indiana University, Bloomington, Indiana

  • Venue:
  • Proceedings of the 1985 ACM SIGSMALL symposium on Small systems
  • Year:
  • 1985

Quantified Score

Hi-index 0.00

Visualization

Abstract

Program Synthesizers are becoming increasingly useful for program creation. However, a majority of these synthesizers are apparently not truly language sensitive. This is indicated by the fact that most of them use error-correcting parsers. We consider the Program Synthesizer as a formal predictive sentence generator with control given to the programmer. For the purpose of generating a (syntactically) correct sentence (or program), our method relies heavily on the underlying LL(k) parse-table for the corresponding programming language. The method demonstrates our 'Prevention is Better than Correction' notion and correspondingly the need for an error-correcting parser can be eliminated. Our method also eliminates the need for 'fetch-template' commands, thus accelerating the program entry. Finally an example session is shown with a non-trivial subset of Modula-2.