一句话：以 CRDT 支撑去中心化近同步编程协作，并研究复制粒度与模仿学习式冲突合并。
屏幕共享与视频会议仅提供基础协作能力，而 IDE 内嵌协作工具能精确追踪源码变更并保留开发者控制权。CRDT 可实现去中心化协作、免除中心服务器依赖，但现有 CRDT 在编程文件中难以保留用户意图、难以处理复杂格式，导致效率低下。本文采用设计科学研究方法，围绕四个研究问题展开：(1) 识别现有 CRDT 算法在保留编程文件用户意图方面的局限；(2) 针对编程协作需求优化 CRDT 实现；(3) 确定编程环境中复制数据类型的合适粒度；(4) 探索监督式模仿学习以增强系统合并冲突的能力。工作目标是改进协作编程工具，提升一致性、效率与用户满意度。
This dissertation investigates near-synchronous collaboration in programming environments by means of Conflict-free Replicated Data Types (CRDTs). While screen sharing and video conferencing offer basic features for collaborative programming, IDE-embedded collaboration tools provide precise tracking of source code changes and developer control. CRDTs enable decentralized collaboration, eliminating central server dependencies. However, existing CRDTs fail to preserve user intent and handle complex formats, leading to inefficiencies. Using a Design Science Research approach, this dissertation addresses four key research areas: (1) Identifying the limitations of existing CRDT algorithms in preserving user intent for programming files, (2) Optimizing CRDT implementations for programming collaboration requirements, (3) Determining the appropriate granularity for replicated data types in programming environments, (4) Exploring supervised imitation learning to enhance system capabilities in merging conflicts. This work aims to improve collaborative programming tools, ensuring higher consistency, efficiency, and user satisfaction.
DOI 原文 ·
@inproceedings{paperbot8489,
title = {Decentralized Near-Synchronous Local-First Programming Collaboration},
author = {Leon Freudenthaler},
booktitle = {ISSTA},
year = {2024},
doi = {10.1145/3650212.3685555}
}