In large-scale software development ecosystems, developers work across multiple projects and often come across peers they have worked with before. When developers with different levels of experience interact with each other, the varying experiences add a distinct dimension to the interaction. Our objective is to quantify the work experience of developers into different levels and capture the different types of interaction among these developers. We investigate the impact of these interaction types on an outcome of interest, by examining data from three large software development ecosystems. We considered the activity around each bug as a unit of analysis and segregated the developers into four experience levels. Based on these levels, we identified 11 different categories of developer interactions. We then used established metrics and statistical techniques to investigate the impact of developer familiarity on bug resolution time. We found that developers are more inclined to interact with peers with whom they have already worked before. We also found statistically significant evidence that this repeated interaction relates to more time taken to resolve bugs. Of the factors that have the potential to impact the bug resolution time, interaction among the participating developers is a preeminent one. In an essentially people-intensive enterprise such as software development, categorizing and observing the interactions among participants offers interesting insights for various stakeholders.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Developer Interaction Across Experience Levels: A Study of Three Systems

  • Reshma Roychoudhuri,
  • Sayantak Karar,
  • Subhajit Datta,
  • Subhashis Majumder

摘要

In large-scale software development ecosystems, developers work across multiple projects and often come across peers they have worked with before. When developers with different levels of experience interact with each other, the varying experiences add a distinct dimension to the interaction. Our objective is to quantify the work experience of developers into different levels and capture the different types of interaction among these developers. We investigate the impact of these interaction types on an outcome of interest, by examining data from three large software development ecosystems. We considered the activity around each bug as a unit of analysis and segregated the developers into four experience levels. Based on these levels, we identified 11 different categories of developer interactions. We then used established metrics and statistical techniques to investigate the impact of developer familiarity on bug resolution time. We found that developers are more inclined to interact with peers with whom they have already worked before. We also found statistically significant evidence that this repeated interaction relates to more time taken to resolve bugs. Of the factors that have the potential to impact the bug resolution time, interaction among the participating developers is a preeminent one. In an essentially people-intensive enterprise such as software development, categorizing and observing the interactions among participants offers interesting insights for various stakeholders.