Jean-Pierre Briot, Rachid Guerraoui, and Klaus-Peter Lohr. Concurrency and Distribution in Object-Oriented Programming. ACM Computing Surveys, 30(1):291--329, September 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....and event handling functions. Hence, inter thread synchronization is based on passing on data items and control events rather than on more error prone primitive mechanisms such as locks and semaphores. The pipeline components are implemented as monitors, also known as synchronized objects [4]: each component may contain at most one active thread at any time. However, we allow threads to be preempted because running functions such as video decoders non preemptively can introduce unacceptable delay. A data processing function of one component is never called before the previous ....

....Styles in Pipeline Components In this section we discuss several styles with respect to activity that can be used in implementing pipeline components. The main distinction is between active objects that have an associated thread and passive objects that are called by external threads [4]. To make a clear distinction between the polarity of a component as introduced in Section 2.2 and its implementation style, we call implementations as active objects thread style components and implementations as passive objects function style components. We focus on neutral components with one ....

J.-P. Briot, R. Guearraoui, and K.-P. Lohr. Concurrency and distribution in objectoriented programming. ACM Computing Surveys, 30(3), September 1998.

....Beowulf ClustIS SGI Figure 5. The absolute wall clock times for the 3 D Poisson solver on the compared platforms. 6. Related Work There are two aspects of our distributed object model. One aspect is the integration of objects and processes into object oriented concurrent programming (see [5, 17] for complete reviews) The other aspect is the support for nontrivial data parallelism in the context of irregular data representations and irregular communication patterns. Concurrent object based languages have their roots in actor models [1, 2, 18] Such models have a functional view of ....

J.-P. Briot, R. Guerraoui, and K.-P. Lohr. Concurrency and distribution in object-oriented programming. ACM Computing Surveys (CSUR), 30(3):291--329, 1998.

....of programming languages in different ways and with different restrictions: Some are purely theoretical discussions of certain language constructs. The many examples range from Dijkstras famous letter Go To statement considered harmful [6] to comprehensive surveys of many languages [4, 16]. These are non quantitative and usually partly speculative. Some such works are more or less pure opinion pieces. Some are benchmarks comparing a single implementation of a certain program in either language for expressiveness or resource consumption, etc. an example is [10] Such comparisons ....

Jean-Pierre Briot, Rachid Guerraoui, and Klaus-Peter Lohr. Concurrency and distribution in objectoriented programming. ACM Computing Surveys, 30(3):291--329, September 1998.

....programming by the famous work of Pattie Maes [17] It is basically the ability of a system to watch its computation and possibly change the way it is performed. Reflection has been applied to various application domains, such as compilers [13] operating systems [34] distributed systems [19, 2], middleware [15, 1] fault tolerant systems [8] and graphic interfaces [22] An object oriented reflective system is logically structured in two (or more) levels, constituting a reflective tower. The first level is the base level and describes the computation that the system is supposed to do. ....

.... the agent receiver simply starts a new thread executing the agent: throws RemoteException new Thread(agent) start( The code of the test we have implemented is the following: public class Test public static void main(String[ args) throws Exception (1) Logger local = new Logger(args[2]) 2) ResourceManager.getResourceManager( publish( local logger , local) 3) Rebinder rebinder = new Rebinder( local logger ) 3 ) Logger logger = Logger) Reflex.createObject( sample.rebind.Logger , rebinder) 4) OperatorAgent agent = new OperatorAgent(new AnOp( logger) 5) ....

J.-P. Briot, R. Guerraoui, and K.-P. Lhr. Concurrency and Distribution in Object Oriented Programming. ACM Computer Surveys, 30(3), Sept. 1998.

....objects and, conversely, an object to several metaobjects) and is dynamic, making it possible to dynamically adapt the behavior of a base object. A second important issue is the de nition of the base events. A quick review of the literature on the applications of re ection (see, for instance, [3, 20, 17]) shows that a simple MOP providing only control of method invocation covers the needs of a large range of applications. However, application developers may need metaobjects that control other events than method invocation (e.g. object creation) or may need to handle these events in an adapted ....

J.-P. Briot, R. Guerraoui, and K.-P. Lhr. Concurrency and distribution in object oriented programming. ACM Computer Surveys, 30(3), September 1998.

....to object oriented based concurrency. The main benefit of an object oriented based concurrent programming model is a higherlevel of abstraction in which some of the lower level concurrency aspects are implicit. However, this is not a complete review of the existing languages and systems (see [26, 112] for complete surveys) The chapter describes three levels of integration between objects and processes. Each of the approaches is discussed in turn by using an example of language implementation. We describe how concurrency is achieved in each of the approaches and what kind of applications can ....

....is difficult to implement, and it requires special mechanisms to ensure consistency. For efficiency reasons, fully concurrent objects are usually implemented as passive objects (standard objects without any activity) without any synchronization, and they are replicated on every processor [26]. Other models distinguish between data and behavioral objects [84] Thus, one type of objects is used to structure shared memory as a collection of passive objects , while a process is considered a special kind of active process object . The actions on passive shared objects are performed ....

[Article contains additional citation context not shown here]

Jean-Pierre Briot, Rachid Guerraoui, and Klaus-Peter Lohr. Concurrency and distribution in object-oriented programming. ACM Computing Surveys, 30(3):291--329, 1998.

....a complex system must implement. In particular, synchronization problems always arise while trying to give a system the capability of executing concurrent threads. To solve these problems, one must be able to exploit additional programming constructs, able to ensure synchronization. According to [6], synchronization may be associated with objects and with their communication means (i.e. message passing) through various sub levels of identification. Message passing, in a concurrent environment, requires synchronization between sender and receiver, as the classical blocking policy cannot ....

J-P. Briot, R. Guerraoui and K-P. Lohr, "Concurrency and Distribution in Object-Oriented Programming", ACM Computing Surveys , Vol. 30(3) pp. 291-329, 1998.

....not truly integrate the SR concurrency model (e.g. no VMs) and Java. 15 Java SR does not address the extension of the SR constructs in terms of the Java programming language. More generally, numerous concurrent object oriented languages have been proposed, e.g. as discussed in a recent survey [13]. These languages have various concurrency models, ways of expressing synchronization, etc. Some concurrent object oriented languages take an object oriented approach to synchronization (as does JR) including Simtalk [9] Actalk [14] GARF [25, 18] and BAST [19, 20] SimTalk, for example, ....

....JR di ers in its overall approach of building synchronization via the operation abstraction, its overall concurrency model, and its de nition as an extension of Java. Furthermore, the survey classi es the di erent approaches in object oriented concurrent programming into three coarse categories [13]. The three categories are: the library approach, the integrative approach, and the re ective approach. The library approach provides class libraries that encapsulate concurrency components (e.g. Java threads are represented as objects) The integrative approach uni es concurrency concepts with ....

J.-P. Briot, R. Guerraoui, and K.-P. Lohr. Concurrency and distribution in object-oriented programming. ACM Computing Surveys, 30(3):291-329, September 1998.

....Keywords Distributed objects, middleware, abstractions, remote invocations. 1. INTRODUCTION In the last decade, many object oriented languages have been extended towards distribution, and new object oriented languages have been designed from scratch with distribution in mind ( 1] 2] 3] [5], 10] 16] 18] The objective is to provide the developers of distributed applications with higher level abstractions than what operating systems or network protocols usually offer. For instance the CORBA approach to distributed programming goes along these lines [14] For several ....

....philosophers uses Java RMI and consists of the remote interface, the server and the client implementation. Interface interface Philosopher extends Remote void think ( void eat ( Server class PhilImpl extends UnicastRemoteObject implements Philosopher static Fork forks[ new Fork[5]; int left, right; static for(int i=0; i 5; i ) forks[i] new Fork ( PhilImpl ( throws RemoteException void think ( void eat ( synchronized (forks[left] synchronized (forks[right] public static void main(String args[ throws Exception for(int ....

[Article contains additional citation context not shown here]

J-P Briot, R. Guerraoui and K-P Lohr. Concurrency and Distribution in Object-Oriented Programming. ACM Computing Surveys, September 1998.

....a given object may even vary according to the client performing the invocation. 1. Introduction In the last decade, many object oriented languages have been extended towards distribution, and new object oriented languages have been designed from scratch with distribution in mind ( 1] 2] 3] [5], 9] 15] 16] The objective is to provide the developers of distributed applications with higher level abstractions than what operating systems or network protocols usually offer. For instance the CORBA approach to distributed programming goes along these lines [13] For several ....

J-P Briot, R. Guerraoui and K-P Lohr. Concurrency and Distribution in Object-Oriented Programming. ACM Computing Surveys, September 1998.

....are considered best practice , so the aspects of synchronisation provide a way of thinking and communicating about synchronisation. Some researchers are already coming to the same conclusions as we have in their survey of concurrency and distribution in object oriented systems, Briot et al. [Bri98] recognise the three primary aspects that we identify, but they restrict their discussion to being only one concern within one form of concurrent object oriented programming. They do not recognise that the aspects of synchronisation exist in all forms of concurrent object oriented programming; but ....

Briot J., Guerraoui R. and Lhr K., "Concurrency and Distribution in Object-Oriented Programming", ACM Computing Surveys, Vol. 30 No. 3, pp291-329, September 1998

....called object, may have a look at paragraph 8.7. Our reactive object shows that an active object can almost always be ready to accept incoming calls, and thus block these calls as shortly as possible. It is close to Actors or to POOL, described in the following, and to the reflexive approach [13], or at least it can easily be adapted to mimic these concepts. 4.4 Inheritance The inheritance of active objects is often linked to the socalled inheritance anomaly [11] namely, the synchronization incoherence that could be introduced by a possible redefinition of the body of a base class by a ....

J.P. Briot, R. Guerraoui and K.P. Lohr, Concurrency and Distribution in Object-Oriented Programming, in ACM Computing Surveys, September 1998, Vol. 30, No. 3, pp. 291-329.

....technology. This calls for a combination of object and process; this combination is captured in the notion of an active object. In research communities it is now well accepted that the object oriented paradigm provides a good foundation for the challenges of concurrent and distributed computing [10, 16, 19, 20, 21, 23, 25, 29], and a number of widespread object oriented programming languages are supporting concurrency [4, 7, 28] As such it is at least in theory fairly straightforward to combine the two programming paradigms. 1.1. Transformational Approaches to Combining OO and Concurrency For practitioners, ....

....instead, but this was merely a matter of taste [9] we apply the await statement for simplicity and because this is the approach we have adopted in the C framework. 3. A Framework for C There are numerous ways to combine object oriented programming languages with concurrency constructs: [10] identifies a coarse classification in three approaches: the library approach, the integrative approach, and the reflective approach; and [14] discusses specific issues related to making concurrent extensions of C . Our approach to extending C with concurrency constructs is to integrate object ....

[Article contains additional citation context not shown here]

Briot J.-P., Guerraoui, R. and Lhr, K.-P. "Concurrency and Distribution in Object-Oriented Programming", ACM Computing Surveys, Vol. 30, No. 3, September 1998, pp. 291-329.

....describe flexible abstractions: ffl Expressiveness. Several abstraction policies are supported by specializing the design pattern associated with the abstraction. For instance: an object concurrency pattern has, among others, specializations for sequential and concurrent objects. 1 Briot et al.[4] considers reflection as classifying a set of approaches. This classification is centered on the properties of integration mechanisms. ffl Modularity. Each concern s design pattern must separate the abstraction from the object s functionality. Modularity facilitates incremental development, for it ....

J.-P. Briot, R. Guerraoui, and K.-P. Lohr. Concurrency and distribution in object-oriented programming. ACM Computing Surveys, 30(3):291--329, September 1998. 5

....capability of the bus makes possible the evolution of execution policies (e.g. object migration) We propose an overall solution that allows the object bus to be adaptive, making the broker plug and play . The classification of concurrent and distributed programming proposed by Briot et al. BRI 98] constitutes an interesting framework for tackling the adaptive issue. The authors distinguish three approaches: library approach applies object oriented concepts in order to structure the concurrent and distributed systems through class libraries; integrative approach consists in unifying ....

BRIOT J.P., GUERRAOUI R., LÖHR K.P. --- Concurrency and Distribution in Object-oriented Programming. In ACM Computer Surveys, vol.30, n°3, p.291329, September 1998.

No context found.

J-P Briot, R. Guerraoui and K-P Lohr. Concurrency and Distribution in Object oriented Programming. ACM Computing Surveys, September 1998.

.... 1 Introduction Context For a couple of decades now, many projects have been devoted to the design and implementation of object based concurrent and distributed systems, mainly focusing on building support for object based concurrency control, remote object invocation and object migration [5]. Very few however have been devoted to building support for replication and object groups. This is actually not surprising as most object based distributed systems have been experimented on academic applications for which fault tolerance is rarely a mandatory feature. This situation has recently ....

J.P. Briot, R. Guerraoui and K.P Lohr. Concurrency and Distribution in Object-Oriented Programming. ACM Computing Surveys, September 1998.

....this paper. Note that this paper is a revised version of a technical report, published as: No 96 01, Dept. of Information Science, the University of Tokyo, January 1996, and also as: No 96 190, D epartement d Informatique, Ecole Polytechnique F ed erale de Lausanne, May 1996. Another paper [19], to appear in late 1998, has been based on this analysis, but includes many more examples, notably with numerous addings from the parallel computing field. ....

J.-P. Briot, R. Guerraoui, and K.-P. L ohr, Concurrency and Distribution in Object-Oriented Programming, ACM Computing Surveys, to appear in late 1998.

....identifies the notion of object with the notion of activity, 1 The objective of this integrative approach is to provide a simple and unified conceptual model to the programmer. Note that this is not the only way to relate object concepts with concurrency, parallelism, and distribution concerns. BGL 98] considers three main approaches (applicative, integrative, reflective) and studies their specificities, differences as well as their complementary aspects and levels. leading to the notion of an active object. It also identifies message passing between objects with the synchronization of the ....

J.-P. Briot, R. Guerraoui, and K.-P. L ohr, "Concurrency and Distribution in Object-Oriented Programming," ACM Computing Surveys, Vol. 30, n o 3, September 1998, pages 291--329.

....order multicast. 7 Summary For a couple of decades now, many projects have been devoted to the design and implementation of object based concurrent and distributed systems, mainly focusing on building support for object based concurrency control, remote object invocation and object migration [4]. Very few however have been devoted to building support for replication and object groups. This is actually not surprising as most object based distributed systems have been experimented on academic applications for which fault tolerance is rarely a mandatory feature. This situation has recently ....

J.P. Briot, R. Guerraoui and K.P Lohr. Concurrency and Distribution in Object-Oriented Programming. ACM Computing Surveys, September 1998.

No context found.

Jean-Pierre Briot, Rachid Guerraoui, and Klaus-Peter Lohr. Concurrency and Distribution in Object-Oriented Programming. ACM Computing Surveys, 30(1):291--329, September 1998.

No context found.

J-P. Briot, R. Guerraoui, K-P. Lohr, Concurrency and Distribution in Object-Oriented Programming in ACM Computing Surveys, Vol. 30, No. 3, September 1998.

No context found.

J.-P. Briot, R. Guearraoui, and K.-P. Lohr. Concurrency and distribution in objectoriented programming. ACM Computing Surveys, 30(3), September 1998.

No context found.

Briot JP, Guearraoui R, Lohr KP. Concurrency and distribution in object-oriented programming. ACM Computing Surveys 1998; 30(3).

No context found.

Briot J., Guerraoui R. and Lhr K., "Concurrency and Distribution in Object-Oriented Programming", ACM Computing Surveys, Vol. 30 No. 3, pp291-329, September 1998

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC