Triggers: A Dynamic Approach to Testing Multiuser Software Robert F. Dugan, Jr. Ephraim P. Glinert Multiuser Software Computer Supported Cooperative Work Testing In this paper we describe triggers, a novel technique for testing multiuser applications. Triggers provide a framework in which selected application events induce the execution of virtual user test scripts. Unlike conventional execution based verification systems which require complete prescription of multiuser tests, our framework allows the tester to interact with virtual user scripts in situ. Additionally, the extensible framework supports any application event or sequence of events and provides an editor to configure the specifics of a triggering event at runtime. Triggers can be used to investigate a variety of problems common to multiuser applications including: synchronization, performance, human-computer interaction, and human-human interaction. We evaluated our framework on a conventionally tested, mature multiuser application. Our evaluation found triggers particularly effective at uncovering race conditions and problems with response time under load. Department of Computer Science, Rensselaer Polytechnic Institute, Troy, NY cs-01-05
Triggers: A Dynamic Approach to Testing Multiuser Software
Robert F. Dugan, Jr.
Ephraim P. Glinert
Multiuser Software
Computer Supported Cooperative Work
Testing
In this paper we describe triggers, a novel technique for testing multiuser applications. Triggers provide a framework in which selected application events induce the execution of virtual user test scripts. Unlike conventional execution based verification systems which require complete prescription of multiuser tests, our framework allows the tester to interact with virtual user scripts in situ. Additionally, the extensible framework supports any application event or sequence of events and provides an editor to configure the specifics of a triggering event at runtime. Triggers can be used to investigate a variety of problems common to multiuser applications including: synchronization, performance, human-computer interaction, and human-human interaction. We evaluated our framework on a conventionally tested, mature multiuser application. Our evaluation found triggers particularly effective at uncovering race conditions and problems with response time under load.
Department of Computer Science, Rensselaer Polytechnic Institute, Troy, NY
cs-01-05