Views over distributed information sources, such as data warehouses, rely on the stability of the schemas of underlying databases. In the event of meta data changes in the sources, such as the deletion of a table or column, such views may become undefined. Using meta data about information redundancy, views can be evolved as necessary to remain well defined after source meta data changes. Previous work in view synchronization focused only on deletions of schema elements. We now offer an approach that makes use of additions also. Our algorithm returns view definitions to previous versions by using knowledge about the history of views and meta data. This technology enables us to adapt views to temporary meta data changes by canceling out opposite changes. It also allows undo/redo operations on meta data. Last, in many cases, the resulting evolved views even have an improved information quality. In this paper, we give a formal taxonomy of schema and constraint changes and a full description of the proposed history-driven view-synchronization algorithm for this taxonomy. We also prove the history-driven view-synchronization algorithm to be correct. Our approach falls in the global-as-view category of data integration solutions but, unlike prior solutions in this category, it now also deals with changes in the information space rather than requiring source schemas to remain constant over time. PUBLICATION ABSTRACT