How tools in IDEs shape developers' navigation behavior

Abstract

Understanding source code is crucial for successful software maintenance, and navigating the call graph is especially helpful to understand source code [12]. We compared maintenance performance across four different development environments: an IDE without any call graph exploration tool, a Call Hierarchy tool as found in Eclipse, and the tools Stacksplorer [7] and Blaze [11]. Using any of the call graph exploration tools more developers could solve certain maintenance tasks correctly. Only Stacksplorer and Blaze, however, were also able to decrease task completion times, although the Call Hierarchy offers access to a larger part of the call graph. To investigate if this result was caused by a change in navigation behavior between the tools, we used a set of predictive models to create formally comparable descriptions of programmer navigation. The results suggest that the decrease in task completion times has been caused by Stacksplorer and Blaze promoting call graph navigation more than the Call Hierarchy tool.

Foraging and navigations, fundamentally: developers' predictions of value and cost

“…These and other findings about developers' foraging and navigations (e.g., [26,31,37,44]) have led software engineering researchers to produce tools that help to reduce developers' navigation costs as they look for the information they need [8,17,18,19,22,30,33,43]. Despite these gains, however, little is known at the foundational level of developer navigation-how well developers go about choosing where to navigate.…”

Toward Principles for the Design of Navigation Affordances in Code Editors

“…One key pattern is that most navigations tend to revisit code the programmer recently visited. For example, studies of predictive models of programmer navigation have consistently found that one of the strongest predictors of which code method (i.e., subroutine) a programmer would click in next was how recently he/she had visited the method (i.e., more recently implies more likely) [10,26,29,33,38,45]. Studies have also found that between 82% [16,45] and 95% [33] of programmer navigations were to previously visited methods.…”

Poster: Discovering Code Dependencies by Harnessing Developer's Activity

“…The most frequent developer's action is navigation in source code space [4]. Developer may navigate using references inferred by a syntactic analysis, but may also pick components from codebase with her own intent, even at random.…”

“…From a software developer's view, dependencies between various source code entities are of the most interest during development and maintenance [2,4]. Syntactic analysis, which is prominently used for discovering source code dependencies, is however insufficient for identifying hidden task-related connections [2,3], for dynamically typed languages, or when combining programming languages [5].…”