A temporal model for multi-level undo and redo

Abstract

divergent and convergent timelines as a way to support collaboration, and provides facilities for conflict resolution. Even in systems without such rich models of history, time is often an explicit-and directly manipulable-part of the user interface, and user experience. Systems such as TimeMachine Computing [15] and Lifestreams [6] are exemplars of this trend.All of these systems rely on an explicit model of history, which can be scanned to support search or "navigation" over a timeline, and all allow their timelines to be "traversed" to move the application's state to other points in its history. However, as powerful as these applications are, their timeline representations are for the most part exceedingly simple. They typically support only linear, not branching timelines (GINA and Timewarp are exceptions, however); the "nodes" in a timeline must represent atomic operations with side effects that are well understood at the time the application is created; and, typically, the timeline of the entire application must be navigated or traversed as a whole-it is impossible to have a portion of the timeline exist in a "bubble" that can be manipulated separately.While we don't commonly encounter such rich models of time in our day to day experience, they can be extremely useful nonetheless. Divergent timelines, for example, can be employed to allow users to interact with different but related versions of an artifact, and then reconcile those differences later. The ability to expand the representation of time in ways that better accommodate side effects can make applications easier to write. And being able to separate the history of one nested artifact from the history of the application as a whole can allow users to work locally on a document, project source code, et cetera, and still integrate their changes globally. This paper presents an expansion of the most traditional representation of timelines, which is based on the command object idiom. The research here makes two contributions. First it extends this traditional model of history to better support the causality effects often found in "real" applications. This first extension works for both linear and divergent timelines. Second, this new causal model is then extended to support multi-level or "interleaved" timelines, in which the various components of an artifact can exist at ABSTRACT A number of recent systems have provided rich facilities for manipulating the timelines of applications. Such timelines represent the history of an application's use in some session, and captures the effects of the user's interactions with that application. Applications can use timeline manipulation techniques prosaically as a way to provide undo and redo within an application context; more interestingly, they can use these same techniques to make an application's history directly manipulable in richer ways by users. This paper presents a number of extensions to current techniques for representing and managing application timelines. The first extension captures causal relations...

Graphical Histories for Visualization: Supporting Analysis, Communication, and Evaluation

“…To design our history tools, we first conducted a design space analysis to enumerate these decisions. We surveyed prior work spanning general history mechanisms [2,8,9,23,29] and interface designs in the areas of graphical design tools [8,16,17,22,26], web browsing [1,5,12,14,15,30], and visualization and simulation [3,4,6,10,11,13,18,19,20,24,25]. In this section, we outline the design space of history tools using examples from this body of work.…”

“…For example, a spreadsheet may maintain separate histories per worksheet, while a graphics editor could maintain local histories for objects in the scene. Edwards et al [8] propose a transactional model to support local histories in which actions may have global side-effects. In all cases, applications must support the ability to merge local histories into a global timeline.…”

Temporal trajectories in shared interactive narratives

“…For example, Timewarp enabled participants to create and navigate divergent edit histories, including moving backwards and forwards in a document's history (story time in our terms) before creating a final definitive version (an example of synthesizing a historic trajectory) [9]. A later extension to this work involved multiple local edit histories that could be edited independently and/or related to a common overall document history [8], a more complex example of synchronizing multiple trajectories. The MASSIVE-3 collaborative virtual environment platform included a nested recording facility where scenes in one virtual world could be enacted, recorded and then played back within another live world.…”

“…The relationship between time and interaction has been a longstanding concern within HCI, spanning wide ranging discussions of responsiveness, pace and interaction [5], revealing and communicating delays [18], synchronizing multiple users' interactions [14,19], and visualizing, browsing and synchronizing convergent and divergent histories of interaction [8,9,17]. This paper extends HCI's concern with time to address the challenges raised by a new generation of narrative-driven experiences such as computer games, artistic performances and hypermedia stories.…”

A Recovery-Oriented Approach to Dependable Services: Repairing Past Errors with System-Wide Undo

“…Edwards et al also propose a hybrid approach in a somewhat different context-executing pluggable user interface code in an extensible digital whiteboard system-as a way to address the problem of commands with unknown side effects or side effects that might change between the original generation of that command and the time it is undone/redone [33]. However, they quickly reject the hybrid approach on the grounds that it would make redo too expensive, and instead favor a pure commandbased approach that requires all side effects to be encoded as new commands in the history.…”