Architectural Unit Testing

被引:3
作者
Scollo, Giuseppe [1 ]
Zecchini, Silvia [1 ]
机构
[1] Univ Verona, Dept Comp Sci, Verona, Italy
关键词
formal testing methods; white-box testing; test purpose; test selection; automated test case generation;
D O I
10.1016/j.entcs.2004.12.006
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
A formal testing methodology is outlined in this paper, that proves applicable to validation of architectural units in object-oriented models, and its use is illustrated in the context of the design of a robot teleoperation architecture. Automated generation of test cases to validate the functionality of the robot trajectory generation unit showcases the key features of this methodology. A disciplined use of UML state diagrams, to model the unit's dynamics consistently with its static properties as modeled by class diagrams, enables one to provide such models with Input/Output Labelled Transition Systems (IOLTS) semantics, whence a rich machinery of testing theories and tools based on those theories become readily available. Our case study tells that, besides black-box testing of final implementation units, white-box analysis of architectural units may greatly benefit from the flexibility of parameterized I/O-conformance relations. Test purposes turn out to be a useful methodological link between functional requirements, which they are drawn from, and conformance relations, which they help one to instantiate, thereby delimiting test selection to purposeful tests. Contingent aspects of our methodology include: a mechanical translation of state diagrams in Basic LOTOS, a non-mechanical, use-case driven synthesis of test purposes, expressed in the same language, and the use of the TGV tool for automated test case generation. Other choices in these respects are well possible, without affecting the characteristic traits of the proposed methodology, that are rather to be found in: 1) the combination of object-oriented architectural modeling with IOLTS semantics; 2) the aim at maximizing the potential for test generation from UML models, in a broad view of testing which applies throughout the development process; 3) the specific proposal to consider internal actions as testable actions, in view of a better coordination between testing (discovery of faults) and debugging (discovery of internal sources of faults).
引用
收藏
页码:27 / 52
页数:26
相关论文
共 28 条
[1]  
APFELBAUM L, 1997, 10 INT SOFTW QUAL WE
[2]   INTRODUCTION TO THE ISO SPECIFICATION LANGUAGE LOTOS [J].
BOLOGNESI, T ;
BRINKSMA, E .
COMPUTER NETWORKS AND ISDN SYSTEMS, 1987, 14 (01) :25-59
[3]  
Booch G, 1997, UNIFIED MODELING LAN
[4]  
Brinksma E., 1988, INFORM TECHNICAL REP
[5]  
Dalal S. R., 1999, Proceedings of the 1999 International Conference on Software Engineering (IEEE Cat. No.99CB37002), P285, DOI 10.1109/ICSE.1999.841019
[6]  
de Vries Rene G., 2000, INT J SOFTW TOOLS TE, V2, P382, DOI [10.1007/s100090050044, DOI 10.1007/S100090050044]
[7]  
El-Far I.K., 2001, ENCY SOFTWARE ENG
[8]   An experiment in automatic generation of test suites for protocols with verification technology [J].
Fernandez, JC ;
Jard, C ;
Jeron, T ;
Viho, C .
SCIENCE OF COMPUTER PROGRAMMING, 1997, 29 (1-2) :123-146
[9]  
FERNANDEZ JC, 1996, LECT NOTES COMPUTER, V1102, P437
[10]  
FERNANDEZ JC, 1996, LECT NOTES COMPUTER, V1102, P348