FLOC 2018: FEDERATED LOGIC CONFERENCE 2018
Towards Incremental and Modular Context-sensitive Analysis

Authors: Isabel Garcia-Contreras, Jose F. Morales and Manuel V. Hermenegildo

Paper Information

Title:Towards Incremental and Modular Context-sensitive Analysis
Authors:Isabel Garcia-Contreras, Jose F. Morales and Manuel V. Hermenegildo
Proceedings:ICLP Proceedings of ICLP 2018
Editors: Paul Tarau and Alessandro Dal Palu'
Keywords:Program Analysis, (Constraint) Logic Programming, Abstract Interpretation, Fixpoint Algorithms, Incremental Analysis, Modular Analysis
Abstract:

ABSTRACT. Context-sensitive global analysis of large code bases can be expensive, which can be specially problematic in interactive uses of analyzers. However, in practice each development iteration implies small modifications which are often isolated within a few modules, and analysis cost can be reduced by reusing the results of previous analyses.

This has been achieved to date on one hand through modular analysis, which can reduce the memory consumption and often localize the computation during reanalysis mainly to the modules affected by changes. In parallel, context-sensitive incremental fixpoints have been proposed that achieve cost reductions at finer levels of granularity, such as changes in program lines. However, these fine-grained techniques are not directly applicable to modular programs.

This paper describes, implements, and evaluates a context sensitive, fixpoint analysis algorithm for (Constraint) Logic Programs aimed at achieving both inter-modular (coarse-grain) and intra-modular (fine-grain) incrementality, solving the problems related to propagation of the fine-grain change information and effects across module boundaries, for additions and deletions in multiple modules.

The implementation and evaluation of our algorithm shows encouraging results: the expected advantages of fine-grain incremental analysis carry over to the modular analysis context. Furthermore, the fine-grained propagation of analysis information of our algorithm improves performance with respect to traditional modular analysis even when analyzing from scratch.

Pages:2
Talk:Jul 17 16:45 (Session 123: Technical Communications II)
Paper: