Merging of EMF models
Merging of EMF models
复制标题
EMF 模型的合并
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
B. Westfechtel
中科院分区：
文献类型：
--
作者：
B. Westfechtel
不完善的模型版本控制严重阻碍了模型驱动软件开发的应用。特别是，迫切需要对合并模型版本的复杂支持。在本文中，我们提出了一种在EMF框架中对模型进行双向和三向合并的形式化方法。该方法可以应用于任意Ecore模型的实例。我们为模型合并指定了上下文无关和上下文敏感的规则，这些规则既可以检测合并冲突，也可以解决合并冲突。基于这些规则，开发了一种合并算法，从一致的输入模型中产生一致的模型。合并算法既不假定对象标识符是唯一的，也不需要更改日志。相反，它依赖于确定公共元素的输入模型之间的匹配(基于状态的方法)。对这些匹配施加的要求被减少到最小，例如，对匹配元素的相对位置没有限制。总之，合并算法应用广泛，保持了一致性，并提供了高级功能，如在存在任意移动的情况下合并有序集合，以及处理难以检测和解决的上下文相关冲突。
Inadequate version control for models significantly impedes the application of model-driven software development. In particular, sophisticated support for merging model versions is urgently needed. In this paper, we present a formal approach to both two- and three-way merging of models in the EMF framework. The approach may be applied to instances of arbitrary Ecore models. We specify context-free as well as context-sensitive rules for model merging which both detect and resolve merge conflicts. Based on these rules, a merge algorithm is developed which produces a consistent model from consistent input models. The merge algorithm does neither assume unique object identifiers, nor does it require change logs. In contrast, it relies on matchings among the input models which identify common elements (state-based approach). The requirements imposed on these matchings are reduced to a minimum, e.g., there are no restrictions on the relative positions of matched elements. Altogether, the merge algorithm is widely applicable, preserves consistency, and offers advanced features, such as merging of ordered collections in the presence of arbitrary moves and handling of context-sensitive conflicts which are hard to detect and to resolve.