High-level query constructs help greatly improve the clarity of programs and the productivity of programmers, and are being introduced to increasingly more languages. However, the use of highlevel queries in programming languages can come at a cost to program efficiency, because these queries are expensive and may be computed repeatedly on slightly changed inputs. For efficient computation in practical applications, a powerful method is needed to incrementally maintain query results with respect to updates to query parameters. This paper describes a general and powerful method for automatically generating incremental implementations of high-level queries over objects and sets in object-oriented programs, where a query may contain arbitrary set enumerators, field selectors, and additional conditions. The method can handle any update to object fields and addition and removal of set elements, and generate coordinated maintenance code and invocation mechanisms to ensure that query results are computed correctly and efficiently. Our implementation and experimental results for example queries and updates confirm the effectiveness of the method. Categories and Subject Descriptors D.3.3 [Programming Languages]: Language Constructs and Features—classes and objects;

This paper describes the use of a powerful graph query language for querying programs, and a novel combination of transformations for generating efficient implementations of the queries. The language supports graph path expressions that allow convenient use of both vertices and edges of arbitrary kinds as well as additional global and local parameters in graph paths. Our implementation method combines transformation to Datalog, recursion conversion, demand transformation, and specialization, and finally generates efficient analysis programs with precise complexity guarantees. This combination improves an O(V E) time complexity factor using previous methods to O(E), where V and E are the numbers of graph vertices and edges, respectively. We also describe implementations and experiments that confirm the analyzed complexities. Categories and Subject Descriptors D.3.2 [Programming Languages]: Language Classifications—Constraint and logic languages, Very high level languages; D.3.3 [Programming Languages]: Language Constructs and Features—Patterns; D.3.4

Re-distributed by Stanford University under license with the author.

Abstract not found