Performant String-Wise Peer-to-Peer Group Editing With Selective Undo in Real-Time Environments

Collaboration exists in various forms but it can be categorized in two major groups: asynchronous and synchronous. Asynchronous collaboration is the most widespread form of collaboration, with the first example of it perhaps being e-mail, which is adopted nowadays by more than 3,500 million users. Synchronous collaboration, on the other hand, allows people to interact with one another in real-time. This has been shown to be an extremely valuable productivity tool, especially in the context of writing software. The situation for editing the same file simultaneously, however, is quite fragmented and suboptimal tools, such as screen-sharing, are often employed. Screen-sharing is the most basic form of synchronous collaboration, and it consists in continuously sending snapshots of one of the collaborators’ screen, sometimes also allowing other participants to control the screen by taking turns. Even with nowadays high-speed networks, the latency of this approach is inadequate; for instance, it takes 60ms to respond to a keystroke sent from Western Europe to Eastern America, assuming the optimal case of data traveling at the speed of light without network congestion or bottlenecks. Other solutions, such as cloud-based IDEs, exist. Such products solve the latency issue, at the cost of forcing users to leave their usual text editor and use tools they are not comfortable with. For all of the above reasons, it’s crucial to develop new techniques and improve the performance of existing ones to add support for low-latency real-time collaboration into existing editors. This thesis provides a background on some basic aspects of convergence and intention-preservation in a distributed text editing setting. Then, a string-wise algorithm based on existing research will be presented. It will be shown that such algorithm advances the state of the art in terms of Big-O complexity while greatly reducing the amount of transmitted data.