Dear group,
I would like to invite you to the Master thesis defense of Judi Abdullah. Judi works on a merging topic, that is currently flying under the radar of research, and explores whether we should change this.
The meeting will be completely remote. I am sorry for the short notice and hope you can still attend.
Cheers
Alex
Title:
An Empirical Study of Context Modifications in Git Merge Conflict Resolutions
Abstract:
Merge conflicts are a common challenge in collaborative software development. Resolving them often requires edits that go beyond the conflicting lines, affecting surrounding code and even non-conflicting files. While previous work has studied conflict resolution, the role of context modifications remains less explored. To better understand the impact of these changes in this, so called context, we empirically examine the nature, frequency, and distribution of context changes during Git merge conflict resolution. We focus on changes within conflicting files as well as modifications made to non-conflicting files. We first classify the types of changes developers perform—such as substantive additions, deletions, and modifications, as well as comment edits, whitespace adjustments, and con- tent movements—and measure their frequency in Python and Java projects. We find that substantive changes dominate, but comment edits and structural changes are also common, particularly in Java. Next, we analyze how changes are distributed across merge commits, conflict files, and context chunks. Most context changes are small and local, but some involve extensive edits and restructuring, showing a long-tail distribution. We also study modifications in non-conflicting files, where developers add or remove substantive content. A partial correla- tion exists between the stability of conflict and non-conflict files, suggesting that while many conflict resolutions are local, some require broader adjustments across the codebase. Finally, we examine the impact of resolution strategies (e.g., “ours,” “theirs,” “manual”). Manual strategies are linked to more extensive modifications, while canonical strategies tend to yield more stable results. These findings show that conflict context and non-conflicting file changes occur frequently and are important for understanding merge conflict resolution. The results highlight patterns that can inform the design of better automated merge tools, especially those that account for unstable regions and broader edits beyond conflicts. This work contributes both practical guidelines and a dataset for future research. Our study provides a labeled dataset of merge conflicts with detailed information on context stability and resolution strategies. Overall, it includes more than 28,000 conflicting files containing over 37,000 conflict chunks and 72,000 context chunks, as well as more than 13,000 non-conflicting files.
________________________________________________________________________________
Microsoft Teams Need help?<https://aka.ms/JoinTeamsMeeting?omkt=en-US>
Join the meeting now<https://teams.microsoft.com/l/meetup-join/19%3ameeting_OGQzNDU3M2UtZjNhZS00…>
Meeting ID: 337 079 862 127 9
Passcode: 68E3LM79
________________________________
For organizers: Meeting options<https://teams.microsoft.com/meetingOptions/?organizerId=05aca28f-23b8-4ea6-…>
[https://unibeidbistoragedeskmgmt.blob.core.windows.net/intune-blob-storage/…]
https://serviceportal.unibe.ch/sp
________________________________________________________________________________
