Program
9:15-9:30 Opening remarks
9:30-10:30 Invited Talk
Phil Bernstein, Microsoft. Schema and Mapping Evolution in an Object-Relational Mapper Talk Slides
ABSTRACT: Schema evolution is an unavoidable consequence of the application development lifecycle. In a database application, the conceptual model, the persistent database model, and the mapping specification between them must co-evolve so they are always mutually consistent and can compile into executable code. We study scenarios where the conceptual model changes and the database and mapping must evolve in kind, in the context of Microsoft's ADO.NET Entity Framework. We present two new techniques that, in most cases, allow those evolutions to progress automatically. The first technique treats the mapping as data, mines it for mapping patterns such as table-per-hierarchy or table-per-type, and automatically derives proper store and mapping changes that are consistent with the pattern. The second technique incrementally compiles the mapping specification into executable views, thereby avoiding the expense of a recompiling the entire mapping specification. Together, the techniques enable a developer to modify the conceptual model and let the system do the rest.
This is joint work with James Terwilliger, Jorge Perez, and Adi Unnithan.
ABSTRACT: Schema evolution is an unavoidable consequence of the application development lifecycle. In a database application, the conceptual model, the persistent database model, and the mapping specification between them must co-evolve so they are always mutually consistent and can compile into executable code. We study scenarios where the conceptual model changes and the database and mapping must evolve in kind, in the context of Microsoft's ADO.NET Entity Framework. We present two new techniques that, in most cases, allow those evolutions to progress automatically. The first technique treats the mapping as data, mines it for mapping patterns such as table-per-hierarchy or table-per-type, and automatically derives proper store and mapping changes that are consistent with the pattern. The second technique incrementally compiles the mapping specification into executable views, thereby avoiding the expense of a recompiling the entire mapping specification. Together, the techniques enable a developer to modify the conceptual model and let the system do the rest.
This is joint work with James Terwilliger, Jorge Perez, and Adi Unnithan.
11:00-12:30 Session 1: Update Semantics and Analysis
- Formal Reasoning about Runtime Code Update. Nathaniel Charlton, Ben Horsfall and Bernhard Reus Talk Slides
- Towards a categorical framework to ensure correct software evolutions. Sylvain Bouveret, Julien Brunel, David Chemouil and Fabien Dagnat
- Predicting Upgrade Failures Using Dependency Analysis. Roberto Di Cosmo and Pietro Abate
12:30-2:00 Lunch Break
2:0-3:30 Session 2: Database Upgrades
- Schema Evolution Analysis for Embedded Databases. Shengfeng Wu and Iulian Neamtiu Talk Slides
- Causes for Dynamic Inconsistency-tolerant Schema Update Management. Hendrik Decker
- Propagating Evolution Events in Data-Centric Software Artifacts. George Papastefanatos, Panos Vassiliadis and Alkis Simitsis
3:30-4:00 Break
4:00-5:30 Session 3: Approaches and Systems
- Agnes Cristèle Noubissi, Julien Iguchi-Cartigny and Jean-Louis Lanet . Hot updates for Java-based Smart Cards
- Non-disruptive Large-scale Component Updates for Real-Time ControllersUpgrade. Michael Wahler, Stefan Richter, Sumit Kumar and Manuel Oriol.
- State Transfer for Clear and
Efficient Runtime Upgrades. Christopher Hayden, Edward
Smith, Michael
Hicks and Jeffrey
Foster. Talk Slides
5:30-6:00 Closing Remarks and Business Meeting