Search for: All records

Program comprehension is a vital skill in software development. This work investigates program comprehension by examining the eye movement of novice programmers as they gain programming experience over the duration of a Java course. Their eye movement behavior is compared to the eye movement of expert programmers. Eye movement studies of natural text show that word frequency and length influence eye movement duration and act as indicators of reading skill. The study uses an existing longitudinal eye tracking dataset with 20 novice and experienced readers of source code. The work investigates the acquisition of the effects of token frequency and token length in source code reading as an indication of program reading skill. The results show evidence of the frequency and length effects in reading source code and the acquisition of these effects by novices. These results are then leveraged in a machine learning model demonstrating how eye movement can be used to estimate programming proficiency and classify novices from experts with 72% accuracy. 

Eye tracking tools are used in software engineering research to study various software development activities. However, a major limitation of these tools is their inability to track gaze data for activities that involve source code editing. We present a novel solution to support eye tracking experiments for tasks involving source code edits as an extension of the iTrace community infrastructure. We introduce the iTrace-Atom plugin and gazel—a Python data processing pipeline that maps gaze information to changing source code elements and provides researchers with a way to query this dynamic data. iTrace-Atom is evaluated via a series of simulations and is over 99% accurate at high eye-tracking speeds of over 1,000Hz. iTrace and gazel completely revolutionize the way eye tracking studies are conducted in realistic settings with the presence of scrolling, context switching, and now editing. This opens the doors to support many day-to-day software engineering tasks such as bug fixing, adding new features, and refactoring. 

An efficient and scalable rule‐based syntactic differencing approach is presented. The tool srcDiff is built upon the srcML infrastructure. srcML adds abstract syntactic information into the code via an XML format. A syntactic difference of srcML documents is then taken. During this process, the differences are further refined using a set of rules that model typical editing patterns of source code by developers. Thus, the resulting deltas model edits that are programmer centric versus a purely syntactic tree edit view. Other syntactic differencing approaches focus on obtaining an optimal tree edit distance with the assumption that this will produce an accurate difference. While this may work well for small or simple changes, the differences quickly become unreadable for more complex changes. By contrast, the approach presented here purposely deviates from an optimal tree edit difference in order to create a delta that is both easier to understand and better models changes between the original and modified. To evaluate the approach, a comparison user study against a state‐of‐the‐art syntactic differencing approach and two line‐based differencing tools is conducted as an online within‐participant study with about 70 subjects on 14 sample changes. The results provide support that the rule‐based syntactic differencing produces more accurate and understandable deltas.