Program comprehension through dynamic analysis

拟议研究的目标是探索新的方法，以帮助开发人员通过探索执行痕迹来理解软件系统的行为。从历史上看，痕迹很难处理，因为它们可能非常大。因此，我们需要研究如何让软件工程师理解和分析他们的内容。在这个方案中，我们打算研究以下技术：首先，我们将研究如何将轨迹的不同部分视为相同的模式。我们的目标是开发先进的轨迹简化算法。其次，我们将研究如何从大量轨迹中自动提取高级行为设计模型。这些模型将记录痕迹的本质，以便工程师更容易理解信息。第三，我们将开发一种标准化的表示痕迹的交换格式。这项工作将基于一种称为轨迹压缩格式(CTF)的现有格式。我们将研究如何最好地扩展CTF的表达能力，以支持多线程系统生成的各种类型的轨迹。最后，我们将研究如何在可视化环境中最好地表示轨迹。由此产生的痕迹分析工具将纳入上述研究的结果。

The objective of the proposed research is the exploration of new approaches to aid developers in comprehending the behaviour of a software system through the exploration of execution traces. Traces have historically been difficult to work with since they can be extremely large. Therefore, we need to investigate ways to allow software engineers to understand and analyse their content. In this proposal, we intend to investigate the following techniques:Firstly, we will study how various portions of a trace can be treated as the same pattern. The objective is to develop advanced trace simplification algorithms.Secondly, we will study how high-level behavioural design models can be automatically extracted from large traces. These models will record the essence of traces making it easier for an engineer to comprehend the information.Thirdly, we will develop an exchange format for representing traces in a standardised way. This work will be based on an existing format called the Trace Compact Format (CTF). We will investigate how to best extend the power of expression of CTF in order to support various types of traces generated from mutli-threaded systems.Finally, we will study how to best represent traces in a visualization environment. The resulting trace analysis tool will incorporate the outcome of the above-mentioned studies.