Classification And Metrics For Replay Tools

Record and replay technique is widely used in practical world for various purposes. UI automation tools record user interaction and replay these actions for automated testing. Debugging tools record every step of program execution and replay these steps to reproduce bugs. These widespread usages makes replay have broad meanings.In this paper, we classify the previous replay tools (for web applications) from two aspects, their usages and their methods. We find out the causality behind these methods and explain the relationship among causality, fidelity and engineering efforts. We also give some suggestions about how to select a proper replay method to match up with the usage. Moreover, we emphasize the importance of timing precision in replay. And in the end, new comprehensive and effective metrics are proposed for verifying and evaluating replay tools. Specifically, we design novel metrics to evaluate the timing precision. The classification and metrics in this paper enrich the understanding of web replay tools and can be used to design/test future methods/tools.