Bridging Between Active Objects: Multitier Programming for Distributed, Concurrent Systems

Programming distributed and concurrent systems is notoriously hard. Active objects, which encapsulate operations, state and control flow, have been investigated by researchers to alleviate this issue. In a distributed system, message exchange among active objects or actors often coincides with network boundaries, and determines a major modularization direction for the application. Yet, certain application functionalities naturally crosscut such modularization direction. For those, structuring the application architecture around network boundaries is purely accidental and does not help reasoning about programs. Recently, multitier programming has been proposed as a programming paradigm that enables code that belongs to different peers to be developed together, in the same compilation unit. The compiler then splits the code and generates the required deployment components. In this work we explore the relation between multitier programming and active objects. Multitier programming can be considered a programming paradigm based on active objects with a focus on application domains where functionalities span multiple active objects, and allows such functionalities to be encapsulated into a single object. The multitier approach keeps the asynchronous model of active objects and actors but provides a holistic view of distributed components and their interactions. Multitier programming addresses the use cases where separating components into different active objects or actors hinders encapsulation and modularization across functional boundaries. In such use cases, multitier programming can increase the level of abstraction, improve software design, simplify code maintenance, aid program comprehension and enable formal reasoning. A number of features of active objects are directly visible to programmers also in the multitier programming, resulting in an interesting combination of language abstractions available to developers.