Home // International Journal On Advances in Intelligent Systems, volume 10, numbers 3 and 4, 2017 // View article

Why We Need Static Analyses of Service Compositions --- Fault vs. Error Analysis of Soundness

Authors:
Thomas M. Prinz
Wolfram Amme

Keywords: Service Composition; Analysis; Case Study; Mojo; Soundness

Abstract:
The programming of classic software systems is well-supported by integrated development environments. They are able to give immediate information about syntax and some logic failures. Although service compositions are widely used within modern systems, such a support for building service compositions is expandable. In this paper, we plead for the creation of an integrated development environment for service compositions, which enables immediate failure feedback during the development. To this end, there is a need for new research activities on occurring failures and how they can be found. Since most current failure finding techniques are based on dynamic approaches, e.g., state space exploration, we show in a case study on soundness that the application of dynamic techniques is not a suitable solution for integrated development environments. In most cases, they are either too time consuming or their output does not lead easily to the root of a failure. As a result, we suggest new advanced (static) analyses of service compositions. To accentuate that pleading, the paper demonstrates a static analysis tool, Mojo, which can be used to check soundness and to get detailed fault diagnostics. With the help of this tool, it was possible to compare the behaviour of dynamic and static analysis techniques in a practical context. For this, a benchmark of real world service compositions was checked regarding soundness with a state space-based (dynamic) and a compiler-based (static) tool. Altogether, the case study and the comparison in a practical context show that dynamic analyses are not suitable for development support. Static analyses should be used instead.

Pages: 458 to 473

Copyright: Copyright (c) to authors, 2017. Used with permission.

Publication date: December 31, 2017

Published in: journal

ISSN: 1942-2679