HeapVision:Debugging by Interactive Heap Navigation

With the proliferation of online source code repositories such as GitHub and BitBucket, programmers have instant access to countless code examples. With the availability of these resources the focus in software development shifts away from writing code towards understanding source code. Many common algorithms, that see a lot of reuse across different problem domains, involve complex manipulations of data structures and hence the program’s heap (e.g., sorting algorithms). Understanding these manipulations and the effect onto the heap’s structure is a cognitively demanding and complex task. However, despite the importance of such algorithms for many applications, there is little tool support build into current IDEs (Integrated development environment) to help programmers in understanding and debugging such algorithms and data structures. In particular, there currently does not exist any sophisticated tool to visually explore the heap structure and to interactively experience the impact of the algorithms instructions on the heap throughout the execution of the program. In the course of this thesis we have developed a tool called HeapVision, which allows for inspection of and interaction with dynamic visual representations of the heap structure. The interface allows programmers to more easily understand how a data structure and it’s content are modified by a particular sequence of instructions. Furthermore, we have developed a novel pen and touch based interface that allows developers to more directly interact with the running process and to concentrate on the essence of the program. Pen and touch gestures can be used to dynamically change the visual representation of the heap, allowing the user to concentrate on important parts of the heap. We leverage formal program analysis tools in order to automatically abstract away unimportant aspects of the heap, while keeping the parts of a data structure that are being manipulated concrete. In other words we aim to capture the essence of an algorithm’s manipulations so that the user may understand the underlying working principle. Our application can improve the comprehension of algorithms, especially for algorithms that manipulate data structures recursively – such as but not limited to reversing linked lists.