Repairing structurally complex data

  • Authors:

  • Sarfraz Khurshid;Iván García;Yuk Lai Suen

  • Affiliations:

  • Department of Electrical and Computer Engineering, The University of Texas at Austin, Austin, TX;Department of Electrical and Computer Engineering, The University of Texas at Austin, Austin, TX;Department of Electrical and Computer Engineering, The University of Texas at Austin, Austin, TX

  • Venue:
  • SPIN'05 Proceedings of the 12th international conference on Model Checking Software
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present a novel algorithm for repairing structurally complex data. Given an assertion that represents desired structural integrity constraints and a structure that violates them, the algorithm performs repair actions that mutate the given structure to generate a new structure that satisfies the constraints. Assertions are written as imperative predicates, which can express rich structural properties. Since these properties can be arbitrarily complex, our algorithm is sound but not complete, and it may not terminate in certain cases. Experimental results with our prototype implementation, Juzi, show that it is feasible to efficiently repair a variety of complex data structures that are routinely used in library code. Juzi can often repair structures comprising of over a hundred objects (even when majority of the objects have some corrupted field) in less than one second. Our algorithm is based on systematic backtracking but does not require storing states and can easily be implemented in a variety of software model checkers, such as the Java PathFinder, SPIN, and VeriSoft.