B. Jonsson. A fully abstract trace model for dataflow networks. In Proceedings 16th ACM Symposium on Principles of Programming Languages, pages 155--165, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....to IO if, for any language components c 1 and c 2 and for any context C , the following holds. D(c 1 ) D(c 2 ) IO(C [c 1 ] IO(C [c 2 ] This approach has been used for many languages. For example, it has been used to define fully abstract models of nondeterministic dataflow networks in [Jon89] and [Rus89] It is not, however, applicable for object oriented languages as it is not always easy to describe a class in terms of a relation between input and output. Classes do not necessarily have any inputs or outputs, in the traditional sense, nor do they necessarily have a final state ....

....of classes if, for all structural models c 1 and c 2 such that c 1 = c 2 , the c 2 can be placed. D(c 1 ) D(c 2 ) T (C [c 1 ] T (C [c 2 ] s=0 s=2 s=1 X,Y X,Y X,Y Y Y A B Fig. 3. State transition diagrams of classes A and B . Equivalently, the following two predicates must hold (see [Jon89]) D(c 1 ) D(c 2 ) T (c 1 ) T (c 2 ) The first predicate states that the model D is compositional. The second predicate states that the model D is at least as distinguishing as the trace model T . Therefore, a model of classes which is fully abstract with respect to the trace model ....

Jonsson, B.: A fully abstract trace model for dataflow networks. In Proceedings 16th ACM Symposium on Principles of Programming Languages, pages 155--165, 1989.

....engineer and economist, a formal notation is needed to analyse the system with mathematical precision. In our approach, the model is transformed (using the method [10] into dataflow networks (DFN) that serves as the language of analysis. This non deterministic dataflow programming paradigm [6] is very suitable to describe the aspects of faults, their e#ects and error propagation at the level of functional units of the business process, i.e. activities, resources. The following tasks can be solved this way concurrently with system design: fault simulation, test generation, ....

B. Jonsson. A Fully Abstract Trace Model for Dataflow Networks. In Proceedings of the 16th ACM symposium on POPL, pages 155--165, Austin, Texas, 1989.

....in the context as illustrated above, A 1 can output two tokens, whereas A 2 can only output a single token, choosing its first behaviour. This example shows very clearly that it is necessary to look for a model that records a more detailed causality relation than the IO relations. Jonsson [17] and Kok [25] have independently given fully abstract models for nondeterminate dataflow. Jonsson s model is based on trace sets, which are sets of possible interactions between a process and its environment. Kok s model turned out to be equivalent. They showed that this model is fully abstract ....

.... due to Kahn [21] and the formal development of the subject is due to Kahn and McQueen [22] The particular axiomatization presented here builds on the ideas of Stark [38] but using the formalism of traces presented in [29] No originality is claimed for the trace model, it was Bengt Jonsson [17] who showed that traces form a fully abstract model of dataflow networks and there were several others with similar ideas at the time. We have a fixed set V of values and a fixed set P of ports. An event is a triple ha; i=o; vi where a 2 P and v 2 V . We say that ha; vi is the label of the event ....

JONSSON, B. A fully abstract trace model for dataflow networks. In POPL'89, ACM, pp. 155--165.

....above are typical examples of linear time models, where the model of synchronisation trees, event structures and Petri nets are all examples of branching time models. In some cases, a linear time model can be exactly what is needed. An example of this is given by Jonsson s full abstraction result [68] for non deterministic dataflow. As we will recall in Sec. 1.2.1 below, some branching information is necessary to reflect safety conditions such as e.g. deadlock freedom in semantics of synchronously communicating systems. 8 Interleaving versus independence. In models like labelled transition ....

....or (variations of) Buchi automata [104] Later more direct extensions of the traditional models for finite observations were proposed, such as the model of general transition systems [54] which appear in Ch. 7 and the generalised synchronisation trees [139] Also the semantics of fair dataflow in [68] uses sets of finite and infinite traces representing completed observations. Global versus local interaction This distinction is essentially the one made by Abramsky in [3] By global interaction we refer to the situation when independent agents of a concurrent system communicate via globally ....

[Article contains additional citation context not shown here]

Bengt Jonsson. A fully abstract trace model for dataflow networks. In POPL'89, pages 155--165. ACM, 1989.

....of a deterministic data flow network can be characterized either in a history based or in a state based way. In the history based approach [Kah74] each component is a function which incrementally reads from and writes onto its input and respectively output channels. In the state based approach [Kok87, Jon89, LS89], each component is a deterministic automaton whose transitions are marked with input and output messages. The extension of deterministic data flow networks to nondeterministic data flow networks proves, however, to be a nontrivial task. In the history based approach, the replacement of functions ....

....our automata have parallel interfaces and work in a timed environment. However composition of timed automata is not studied in [Kok87] because the main emphasis of this work is to give a fully abstract denotational semantics for untimed data flow networks. Similar timed automata were also used in [Jon89]. These are however composed by taking all possible synchronized interleavings. The relational theory of nondeterministic data flow networks is also the subject of intensive research. The model presented in this paper is a slight variation of the one given [GS95] In that model we used ....

B. Jonsson. A fully abstract trace model for dataflow networks. In Proc. 16th Annual ACM Symposium on Principles of Programming Languages, pages 155--165, 1989.

....Our main contribution is that we have formalized mobility in the context of streams and functions on streams. In this respect our work is in the tradition of Kahn [Kah74, KM77] The close relationship between stream based models and models based on traces as for example advocated by Jonsson [Jon89] is well known. Hence, it makes sense to ask why we use a stream based model and not one based on traces Well, we think that each model has its own merits, and it is hard to classify one as better than the other. Each model was pursued in different schools, and each school prefers its own ....

....Well, we think that each model has its own merits, and it is hard to classify one as better than the other. Each model was pursued in different schools, and each school prefers its own model. This might be also a matter of tradition. We are not aware of work on mobility within the framework of [Jon89] How our approach to mobility should be reformulated in a trace based setting, and whether this reformulated trace based version would be simpler or more elegant than the one presented in this paper, is a matter of further research. We think that the trace based, automata like models are closer ....

B. Jonsson. A fully abstract trace model for dataflow networks. In Proc. 16th Annual ACM Symposium on Principles of Programming Languages, pages = 155--165, 1989.

.... to detect the information from the system and how to 2 express it in the model [God96, WN94] The literature provides examples where independence is detected by a notion of events happening at different localities [MN92] belonging to different processes [God96] or happening at different ports [Jon89, PS88]. When defining an independence model, one typically chooses to provide a structure that can model one or more of these features. A natural goal is that this should be as abstract as possible, such that the semantics does not make any distinctions not imposed by the independence. This is one of ....

....in the mid sixties [Kah74, Den74, Den84] The basic idea is that data flows between nodes, that are interconnected by channels, acting as FIFO queues, into a dataflow network. There exist several formal definitions of the basic units (nodes) and how to combine them into network (see discussion in [Jon89]) The fundamental unit we have in mind, is the monotone input output port automata [Jon89, PS88] We assume that V is a fixed set of values and P a fixed set of port names. For two sets A; B P of port names (not necessarily disjoint) we define a port alphabet Sigma A;B = fig Theta A Theta ....

[Article contains additional citation context not shown here]

Bengt Jonsson. A fully abstract trace model for dataflow networks. In PoPL'89, pages 155--165. ACM, 1989.

....formulated compositional semantics of this kind also for the more general classes of P T nets, but without comparing it with the already existing token game semantics. These seminal works in dataflow nets and in Petri nets inspired and strongly influenced the research of modularity for net models. [1, 4, 5, 6, 12, 15, 16, 18, 19, 20, 22]. Note that in these works modularity and compositionality are not clearly distinguished. Modularity issues for models based on the net concept constitute also one of the major goals in our previous papers [5, 18, 19, 20] In [5] we used the Mazurkiewicz algebraical approach to formulate an ....

B. Jonsson. A fully abstract trace model for dataflow networks. In Proceedings of the 16-th ACM Symposium on Principles of Programming Languages, 1989.

....a poor way of thinking about it that we are better off learning to deal with concurrency in its own right than persisting with reductionism. Reductionists counter that primitivism is not fully abstract, that is, draws unobservable distinctions. In the context of dataflow nets Kok [Kok87] Jonsson [Jon89], Rabinovich and Trakhtenbrot [RT88] Russell [Rus89] and others have shown that reductionism, unlike primitivism, is fully abstract. This work was supported by the National Science Foundation under grant number CCR8814921. The primitivist answer is that these results presume the observer ....

B. Jonsson. A fully abstract trace model for dataflow networks. In Proc. 16th Annual ACM Symposium on Principles of Programming Languages, pages 155--165, January 1989.

....on set theoretic input output relations can give results consistent with token pushing semantics. Subsequently, a rather large literature has developed on this subject. A variety of sophisticated approaches, such as powerdomains [1, 8] categories [3] scenarios [6, 7] sets of linear traces [4, 11, 12, 17, 23], pomsets [10, 26, 27] multilinear algebra [22] and other ideas [18, 19, 25, 28, 29, 30, 34, 37, 35] have been tried, but up until now none of these approaches has resulted in a truly simple and natural generalization of Kahn s principle that also maintains a clear connection with operational ....

B. Jonsson. A fully abstract trace model for dataflow networks. In Sixteenth Annual ACM Symposium on Principles of Programming Languages, pages 155--165, January 1989.

.... u [P ] if n : N u [P ] I;O (N u [P ] D[n Omega m] D[n] Omega D[m] We define the observation semantics of a network as the input output behavior of its semantic denotation as follows: O[n] f(x; y) j 9f 2 D[n] y = f(x)g Now, we define full abstraction as in [Jon89] Definition 13 Full abstraction The model D is said to be fully abstract with respect to O if for all networks n; m : N u [P ] I;O (N u [P ] 1) D[n] D[m] O[n] O[n] fD is more distinguishingg (2) 8C( D[n] D[m] D[C(n) D[C(m) fD is compositionalg (3) ....

B. Jonsson. A fully abstract trace model for dataflow networks. In Proc. 16th Annual ACM Symposium on Principles of Programming Languages, pages 155--165, 1989.

....a language designed to model distributed process interaction using ideas from [121] 2. 3 The 1980s Dataflow continued to be an area of widespread research during the 1980s and several additional semantic models for dataflow were introduced, for example, 76] 21] 197] 198] 199] 131] and [120]. Logic programming languages also began to be used to model SPSs. A modification of PROLOG used to model what have become termed perpetual processes (see [147] within logic programming was introduced in [19] Functional programming languages were also used widely to model SPSs. Notable in this ....

B Jonsson. A fully abstract trace model for dataflow networks. Technical Report 88016, Swedish Institute of Computer Science, 1988.

....a language designed to model distributed process interaction using ideas from [85] 3. 3 The 1980s Dataflow continued to be an area of widespread research during the 1980s and several additional semantic models for dataflow were introduced, for example, 61] 18] 142] 143] 144] 94] and [84]. Logic programming languages also began to be used to model SPSs. A modification of PROLOG used to model what have become termed perpetual processes (see [107] within logic programming was introduced in [16] Functional programming languages were also used widely to model SPSs. Notable in this ....

B Jonsson. A fully abstract trace model for dataflow networks. Technical Report 88016, Swedish Institute of Computer Science, 1988.

....no longer sufficient to compute the behaviour of a network. This situation, known as merge or Brock Ackermann anomaly, was solved by adding information to the input output behaviour by using scenarious, traces, or oracles, etc. Extensions to the nondeterministic case were suggested for instance in [SN85, Bro87, Kok87, Jon89, Bro93]. In this paper we take the viewpoint of [Park83, Bro87] and model nondeterministic dataflow networks with the help of oracles. An oracle provides a priori global information on the choices in all the nondeterministic points and it allows to give the semantic of a nondeterministic network by a set ....

B. Jonsson. A fully abstract trace model for dataflow networks. In: Prooc. 16th ACM Symp. POPL'89, 155--165, 1989.

....semantics generalizing Kahn s model to incorporate non determinism. A variety of models has been proposed, including hiatons [Fau82] scenarios [BA81] I O automata [Sta89, RT89] and sets of continuous functions [Abr90] Various trace theoretic models have also been proposed, including [KP84, Kok87, Jon89, Rus90]. Although each of these models attempted to stay faithful to Kahn s spirit, typically retaining some form of continuity assumption, none is as simple and elegant as Kahn s original model. Moreover these approaches have achieved only limited success, usually being incapable of properly modelling ....

....Instead, a deadlocked process manifests itself in our model as infinite stuttering, which after all is how it will appear to any process attempting to interact with it: a deadlocked process will never change the state, and never terminates. In contrast to several earlier trace theoretic models [Jon89, Kok87, Rus90, KP84] we take a different view of state, and of the nature of a step in a trace, and we build in a different combination of closure conditions on trace sets. Typically these earlier models were concerned with the search for fully abstract models of Kahn networks, with respect to a kind of observable ....

B. Jonsson. A fully abstract trace model for dataflow networks. In Sixteenth Annual ACM Symposium on Principles of Programming Languages, pages 143--154. ACM Press, 1989.

....properties of presheaves automatically imply the usually postulated axioms for asynchronous, monotone automata [27, 32] 2. we get a notion of bisimulation, which is crucial when one includes both synchronous and asynchronous primitives together, 3. it is closely connected to the extant models [15] expressed in terms of trace sets, but also provides a relational viewpoint which allows one to think of composing network components as a (kind of) relational composition, 4. gives a semantics of higher order networks almost for free by using the passage from traced monoidal categories to ....

....computing agents, that are interconnected by communication channels acting as unbounded buffers. Traditionally, the observable behaviour is taken to be the inputoutput relation between sequences of values on respectively input and output channels, sometimes referred to as the history model [15]. For dataflow networks built from only deterministic nodes, Kahn [19] has observed that their behaviour could be captured denotationally in the history model, defining network composition by the least fixed point of a set of equations describing the components, which was later shown formally by ....

[Article contains additional citation context not shown here]

B. Jonsson. A fully abstract trace model for dataflow networks. In POPL'89, pages 155-- 165. ACM, 1989.

....1 Introduction The general theory of asynchronous systems has reached a state of maturity comparable to the much more extensively studied theory of process calculi. The fundamental studies of Stark [27, 28] Arnold and Nivat [2] Mazurkiewicz [15] Pratt [22] Kok [12] Kwiatkowska [13] Jonsson [7] and Rabinovich and Trakhtenbrot [23] just to mention an arbitrary subset, have placed the theory on firm foundations. Among the possible models of asynchronous computation a particularly pleasing one is the dataflow model. We have a clear conceptual model, originally due to Kahn and McQueen [9, ....

....in operational terms using port automata [14] and show that the Kahn principle is correct with respect to this model. We have generalizations to indeterminate networks [10, 1] and we have expressiveness results that classify the power of different primitives [19, 20, 17] The result of Jonsson [7, 24] shows that traces form a fully abstract semantics for such networks. The most important expressiveness result shows that there is a class of automata, called monotone automata that enjoy nice properties but which exclude simple primitives such as fair merge [19] One of the main results of the ....

[Article contains additional citation context not shown here]

Bengt Jonsson. A fully abstract trace model for dataflow networks. In Conference Record of the Sixteenth Annual ACM Symposium on Principles of Programming Languages, pages 155--165, Austin, Texas, January 11--13, 1989. ACM SIGACT and SIGPLAN.

....very pleasant application of Scott s semantic ideas. In almost any elaboration of Kahn s model, the situation becomes much more difficult. In the context of indeterminate dataflow, recent work by Kok, Jonsson and others has shown that one gets fully abstract models from the traces of computations [9, 11, 17]. Traces do not, however, give one the same level of abstraction that is provided by being able to think of processes as functions. Similarly, though process algebra has now reached a high degree of mathematical maturity and elegance, see, for example, the recent books by Milner and by Hennessy ....

B. Jonsson. A fully abstract trace model for dataflow networks. In Proceedings of the Sixteenth Annual ACM Symposium On Principles Of Programming Languages, 1989.

....subset of E denoted by the regular expression (coin choc) coin e) But, this semantics provides no information about the process commitments, namely, that a chocolate must always be delivered by the machine once a coin is inserted. The idea of quiescence, developed in [Jonsson 85, Jonsson 89] within the context of a logic to reason about specifications, and developed in [Costa 89, Costa and SernadasA 90b] within an algebraic framework, provides a solution for the problem of representing state dependent commitments in a process. We say that a process is in a quiescent state iff it is ....

B.Jonsson, "A fully abstract trace model for dataflow networks", in Proc. of the 16th Annual ACM Symposium on Principles of Programming Languages, 1989, 155-165

....to IO if, for any language components c 1 and c 2 and for any context C , the following holds. D(c 1 ) D(c 2 ) IO(C [c 1 ] IO(C [c 2 ] This approach has been used for many languages. For example, it has been used to define fully abstract models of nondeterministic dataflow networks in [Jon89] and [Rus89] It is not, however, applicable for object oriented languages as it is not always easy to describe a class in terms of a relation between input and output. Classes do not necessarily have any inputs or outputs, in the traditional sense, nor do they necessarily have a final state which ....

....corresponding to c 1 and c 2 can be placed. D(c 1 ) D(c 2 ) T (C [c 1 ] T (C [c 2 ] A Fully Abstract Semantics of Classes for Object Z 43 s=0 s=2 s=1 X,Y X,Y X,Y Y Y A B Fig. 3. State transition diagrams of classes A and B . Equivalently, the following two predicates must hold (see [Jon89]) D(c 1 ) D(c 2 ) D(C [c 1 ] D(C [c 2 ] D(c 1 ) D(c 2 ) T (c 1 ) T (c 2 ) The first predicate states that the model D is compositional. The second predicate states that the model D is at least as distinguishing as the trace model T . Therefore, a model of classes which is fully ....

Jonsson, B.: A fully abstract trace model for dataflow networks. In Proceedings 16th ACM Symposium on Principles of Programming Languages, pages 155--165, 1989.

No context found.

B. Jonsson. A fully abstract trace model for dataflow networks. In Proceedings 16th ACM Symposium on Principles of Programming Languages, pages 155--165, 1989.

No context found.

B. Jonsson. A fully abstract trace model for dataflow networks. In Proc. 16th Annual ACM Symposium on Principles of Programming Languages, pages 155-- 165, 1989.

No context found.

B. Jonsson. A fully abstract trace model for dataflow networks. In Proc. 16th Annual ACM Symposium on Principles of Programming Languages, pages 155-- 165, 1989.

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