Kleist, J. and Sangiorgi, D., Imperative Objects and Mobile Processes, , Journal of Science of Computer Programming, Volume 44, Number 3, September 2002, 293-342.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....The first class contains works that provide semantics to objects by encoding them into process calculi. Works in the second class study calculi where primitives for objects and for concurrent processes coexist. Systematic translations of objects into calculus can be found, for instance, in [Wal95,HK96,San98,KS98]. Works that belong to the second approach are much closer to what we do here. Among these it is worth to mention the approaches of [Vas94,PT95,FMLR00] which, given a name passing calculus, build high level constructors distinctive of object oriented languages. A complementary approach is followed ....

....for the self variable is the name of the object rather then the object itself. A distinctive feature of [GH98] is the fact that the syntax of conc includes sequential composition of expressions that return results. This contrasts with what happens in most formalisms based on processes ([Vas94,PT95,Wal95,KS98]) where the operation of returning a result is translated into sending a message on a result channel. Even though we did not explicitly address the problem of returning a result, it is easy to extend our framework by endowing agent interfaces not only with methods, but also with fields whose ....

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. In PROCOMET '98 (IFIP Working Conference on Programming Concepts and Methods). North-Holland, 1998.

.... p(x) P replicated input j if [p=q] then P else Q name testing j case v of fl (x) Pg case destructor j wrong wrong Figure 9: Syntax of calculus 24 The labeled transition system is the usual one in the early style [ACS98] extended with straightforward rules for case analysis and wrong [KS98]. Transitions are of the form P , where action is: interaction) pv (free input) pv (free output) or ph( x) vi (bound output, i.e. the emission of value v containing a private name x at p) In these actions, p is the subject and v the object part. Free and bound names (fn, bn) of ....

....parameterized on two names. In an encoded jeblik term [ a] p , the parameters are used for returning its result (p) and for book keeping its current self (i c ) Objects The basic structure of the encoding is similar to the one for the imperative object calculus IOC by Kleist and Sangiorgi [KS98]: the translation of an object O, as shown in Figure 10, consists of a message that returns the object s reference s on some result channel p, a composition of replicated processes that give access to the method bodies [ b j ] r , and a new object process new.OO h s; t i that connects ....

[Article contains additional citation context not shown here]

J. Kleist and D. Sangiorgi. Imperative Objects and Mobile Processes. In D. Gries and W.-P. de Roever, eds, Proceedings of PROCOMET '98, pages 285-303. International Federation for Information Processing (IFIP), Chapman & Hall, 1998.

....such promises can be rather hard to achieve. The design and the implementation of concurrent object oriented languages, e.g. 2, 20, 1, 4] has recently prompted the investigation of the theoretical foundations of concurrent objects. Several works provide encodings of objects in process calculi [19, 18, 12, 5] or, alternatively, supplement objects with concurrent primitives [16, 3, 11] Those works promote a uni ed framework for reasoning about objects and processes, but they do not address the composition of object de nitions or its typechecking. In this work, we model concurrent objects in a simple ....

....class following the same convention as bu for initialization. 5 Related and future works There are many works on supplementing object calculi with concurrent primitives [16, 3, 11] or on supplementing process calculi with objects (usually by the mean of an encoding in the original calculus) [19, 18, 12, 5]. Our work follows the latter tradition. However, to our knowledge, it is the only one to address safe object composition in a process calculus. In [17] Odersky proposes an object oriented extension of a language based on the join calculus. His proposal amounts to adding some record structure to ....

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. June 1998.

....Riely s [11, 12] Dp, Yoshida and Hennessy s [15] Dlp, Sewell s [13] distributed p, and Vitek and Castagna s [14] Seal calculus. None of these languages are directly object oriented, although there are strong parallels between them and OO languages, using codings of OO into the p calculus such as [2, 6, 8, 9, 10]. In this paper, these two strands of research are brought together, to provide a model for distributed object based languages. In doing so, two features of distributed programming become clear: the importance of separating serializable and non serializable data using types, and the use of ....

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. In Proc. IFIP Working Conference on Programming Concepts and Methods, 1998. 8

....The present paper gives an in depth account of the former aspect, and only hints at the latter (see Section 6) leaving a detailed treatment for future work. Several proposals of formalisms for foundational study of concurrent object oriented programming can be found in the literature (e.g. [9, 12, 15, 16, 18, 7, 14]) however, to our knowledge, no previous work directly embeds methods into calculi of mobile agents. 2 Outline of the model The definition of the model is given parametrically on the underlying calculus of processes and agents so that it can be adapted to formalisms such as Mobile Ambients [5] ....

J. Kleist and Sangiorgi D. Imperative objects and mobile processes. Unpublished manuscript.

....to reusing the advances in type system of object oriented programming and static analysis. This will be one of the natural continuations of this work. Several proposals of formalisms for foundational study of concurrent object oriented programming can be found in the literature (see for example [11, 13, 16, 17, 20]) However, to our knowledge, this is the rst attempt to directly embed methods in calculi of mobile agents. 2 Mobile Objects Our model of mobile objects results from extending calculi of mobile agents over distributed systems, endowing agents with methods. The de nition of the model is largely ....

J. Kleist and Sangiorgi D. Imperative objects and mobile processes. Unpublished manuscript.

....presented a first attempt to formalize the semantics of jeblik in terms of a process calculus. The reason for that choice was the immediate availability of notions of program equivalence and, in particular, proof techniques. Furthermore, previous work on the semantics of objects as processes (cf. [Wal95,HK96,San98,KS98]) clearly showed the value of such approaches. However, our first attempt failed in the sense that we did not capture the intuition of all examples that we could come up with ourselves. One problem of the approach apparently was the missing higher level semantics with which to compare our ....

J. Kleist and D. Sangiorgi. Imperative Objects and Mobile Processes. In D. Gries and W.-P. de Roever, eds, Proceedings of PROCOMET '98, pages 285--303. International Federation for Information Processing (IFIP), Chapman & Hall, 1998.

....of our operational semantics. Various formalisms in the calculus family have been used to model imperative or concurrent objects, for instance, in the work of Honda and Tokoro (1991) Jones (1993) Vasconcelos (1994) Pierce and Turner (1995) Walker (1995) Fournet and Gonthier (1996) Kleist and Sangiorgi (1997), and Dal Zilio (1998) All these models use formalisms based on processes, computations with no concept of returning a result, instead of expressions. The operation of returning a result is translated using continuations into sending a message on a result channel. Our concV calculus is based on ....

Kleist, J. and D. Sangiorgi (1997). Imperative objects and mobile processes. Draft.

....such promises can be rather hard to achieve. The design and the implementation of concurrent object oriented languages, e.g. 2, 23, 1, 4] has recently prompted the investigation of the theoretical foundations of concurrent objects. Several works provide encodings of objects in process calculi [22, 20, 12, 6] or, alternatively, supplement objects with concurrent primitives [16, 3, 11] Those works promote a uni ed framework for reasoning about objects and processes, but they do not address the composition of object de nitions or its typechecking. Microsoft Research, 1 Guildhall Street, Cambridge, ....

.... chemical typings modulo process reduction (i.e. taking 7 for ) 6 Related and future works There are many works on supplementing object calculi with concurrent primitives [16, 3, 11] or on supplementing process calculi with objects (usually by the mean of an encoding in the original calculus) [22, 20, 12, 6]. Our work follows the latter tradition. However, to our knowledge, it is the only one to address safe object composition in a process calculus. In [17] Odersky proposes an object oriented extension of a language based on the join calculus. His proposal amounts to adding some record 29 ....

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. To appear, June 1998.

....such promises can be rather hard to achieve. The design and the implementation of concurrent object oriented languages, e.g. 2, 22, 1, 4] has recently prompted the investigation of the theoretical foundations of concurrent objects. Several works provide encodings of objects in process calculi [21, 20, 13, 6] or, alternatively, supplement objects with concurrent primitives [17, 3, 12] Those works promote a uni ed framework for reasoning about objects and processes, but they do not address the composition of object de nitions or its typechecking. In this work, we model concurrent objects in a simple ....

....of access privileges for every message along the reduction. 11 5 Related and future works There are many works on supplementing object calculi with concurrent primitives [17, 3, 12] or on supplementing process calculi with objects (usually by the mean of an encoding in the original calculus) [21, 20, 13, 6]. Our work follows the latter tradition. However, to our knowledge, it is the only one to address safe object composition in a process calculus. In [18] Martin Odersky proposes an object oriented extension of a language based on the join calculus. His proposal amounts to adding some record ....

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. To appear, June 1998.

....to reusing the advances in type system of object oriented programming and static analysis. This will be one of the natural continuations of this work. Several proposals of formalisms for foundational study of concurrent object oriented programming can be found in the literature (see for example [11, 13, 16, 17, 20]) However, to our knowledge, this is the first attempt to directly embed methods in calculi of mobile agents. 2 Mobile Objects Our model of mobile objects results from extending calculi of mobile agents over distributed systems, endowing agents with methods. The definition of the model is ....

J. Kleist and Sangiorgi D. Imperative objects and mobile processes. Unpublished manuscript.

....more complex cases. Many authors observe that inheritance and synchronization then often conflict [14] In contrast, process calculi provide a simpler, finer grain parallelism, and this formal simplicity is appealing for a core calculus of objects. Several works encode objects in process calculi [20, 19, 12, 7] or, vice versa, supplement objects Microsoft Research y Dipartimento di Scienze dell Informazione, Universit a di Bologna, Mura Anteo Zamboni 7, 40127 Bologna, Italy z INRIA Rocquencourt, BP 105, 78153 Le Chesnay Cedex France. with concurrent primitives [15, 4, 11] In these works the ....

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. To appear, June 1998. 17

....for translating parallel object oriented languages and for proving the validity of certain program transformations. But the source languages studied are untyped and rather simple. In [24] Sangiorgi gives the first interpretation of the typed Abadi Cardelli s calculus with subtyping in , and in [19], the authors give an interpretation of the imperative object calculus. There are also some works on the definition of concurrent calculus of objects obtained by extending sequential languages with concurrent operators: 12, 14] 2 The Calculus The blue calculus is a variant of the mini ....

Josva Kleist and Davide Sangiorgi. Imperative objects and mobile processes, 1998. (to appear in Proceedings of PROCOMET'98).

....of certain program transformations. But the source 33 languages are untyped and rather simple. In [36] Sangiorgi gives the rst interpretation of Abadi Cardelli typed functional calculus with subtyping in (a related work is [27] This interpretation is extended to the imperative case in [29]. These interpretations, and the type system used, are very dioeerent from ours. For example, in the coding of method update, we do not use irelay constructsj. Intuitively, in our encoding, the number of reductions when invoking a method does not depend on the number of method updates applied on ....

Josva Kleist and Davide Sangiorgi. Imperative objects and mobile processes. In Proc. of PROCOMET '98 Working Conference on Programming Concepts and Methods. North-Holland, 1998.

No context found.

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. In Proc. PROCOMET'98. North-Holland, 1998.

....In previous work on the Imperative Object Calculus (IOC) AC96] equivalence between IOC terms was defined in a contextual way [GHL97] similar to Definition 4.3.1. In many cases, it seems simpler to use a semantics by translation into # calculus [MPW92] to establish the equivalence between terms [KS98] The main advantage is the large number of equivalences and algebraic laws that can be used to reason about expressions. Since IOC is (almost) a concurrency free subset of jeblik, we chose a similar path for establishing the safety of external surrogation. An early sketch of a proof for a ....

....instance, we are about to show that join(fork(a) # =a holds, although this may only be true under certain self infliction conditions, and we also expect some laws of Moggi s computational # calculus [Mog89] to hold naturally. Similar properties have already been proved by Kleist and Sangiorgi [KS98] using a # calculus semantics for the IOC, of which our translation in [MHKN99] is a rather direct generalization. As noted in the Introduction, our conjectures on safe surrogation will only hold unless sites may fail, which is obvious by considering that the proxy site (where the object ....

J. Kleist and D. Sangiorgi. Imperative Objects and Mobile Processes. In D. Gries and W.-P. de Roever, eds, Proceedings of PROCOMET '98, pages 285--303. International Federation for Information Processing (IFIP), Chapman & Hall, 1998.

....Objects, Migration 1. INTRODUCTION The work presented in this paper is in line with the research activity to use the # calculus as a tool box for reasoning about object based programming languages. Former works on the semantics of objects as processes showed the value of this approach: while [35, 12, 27, 15] focused on providing formal semantics to object oriented languages and language features, the work of others [24, 29] has been driven by a specific programming problem. Our work tackles a problem in Cardelli s lexically scoped distributed programming language Obliq [4] Cardelli proposed to ....

....after having itself in the local scope of x assigned 0 to x. In this respect, our proof concerns the class of all programs in which clients may freely perform surrogation on object di#erent from itself. In Section 6.2 we explain this in more detail. 1.3. Related work The work closest to ours is [15] where an interpretation of Abadi and Cardelli s object calculus [1] into typed # calculus is presented. Unlike [15] we focus on a concurrent object calculus. Gordon and Hankin [7] and Di Blasio and Fisher [5] describe two concurrent object calculi, but no account of object migration is given ....

[Article contains additional citation context not shown here]

Josva Kleist and Davide Sangiorgi. Imperative objects and mobile processes. In David Gries and Willem-Paul de Roever, editors, Proceedings of PROCOMET '98, pages 285--303. International Federation for Information Processing (IFIP), Chapman & Hall, 1998.

....file 33 In previous work on the Imperative Object Calculus (IOC) 1] equivalence between IOC terms was defined in a contextual way [8] similar to Definition 4.1. In many cases, it seems simpler to use a semantics by translation into calculus [15] to establish the equivalence between terms [12]. The main advantage is the large number of equivalences and algebraic laws that can be used to reason about expressions. Since IOC is (almost) a concurrency free subset of jeblik, we chose a similar path for establishing the safety of external surrogation. An early sketch of a proof for a ....

....For instance, we are about to show that join(fork(a) a holds, although this may only be true under certain self infliction conditions, and we also expect some laws of Moggi s computational calculus [16] to hold naturally. Similar properties have already been proved by Kleist and Sangiorgi [12] using a calculus semantics for the IOC, of which our translation in [14] is a rather direct generalization. As noted in the Introduction, our conjectures on safe surrogation will only hold unless sites may fail, which is obvious by considering that the proxy site (where the object returned ....

Josva Kleist and Davide Sangiorgi. Imperative objects and mobile processes. In Gries and de Roever [9], pages 285--303.

....1 Introduction The work presented in this paper is in line with the research activity to use the calculus as a toolbox for reasoning about object based programming languages. Former works on the semantics of objects as processes showed the value of this approach: while [Wal95, HK96, San98, KS98] focused on providing formal semantics to object oriented languages and language features, the work of others [PW98, San99b] has been driven by a speci c programming problem. Our work tackles a problem in Cardelli s lexically scoped distributed programming language Obliq [Car95] Cardelli ....

....by the translation. Section 6 lays the ground work for the proof of safe surrogation, and in Section 7 we prove the main result of the paper. Section 8 contains conclusions. Finally, Appendix A contains proofs omitted from the main part of the paper. 1. 3 Related work The work closest to ours is [KS98] where an interpretation of Abadi and Cardelli s object calculus [AC96] into typed calculus is presented. Unlike [KS98] we focus on a concurrent object calculus. Gordon and Hankin [GH98] and Di Blasio and Fisher [DF96] describe two concurrent object calculi, but no account of object migration ....

[Article contains additional citation context not shown here]

J. Kleist and D. Sangiorgi. Imperative Objects and Mobile Processes. In D. Gries and W.-P. de Roever, eds, Proceedings of PROCOMET '98, pages 285-303. International Federation for Information Processing (IFIP), Chapman & Hall, 1998.

....1 Introduction The work presented in this paper is in line with the research activity to use the calculus as a toolbox for reasoning about object based programming languages. Former works on the semantics of objects as processes showed the value of this approach: while [Wal95, HK96, San98, KS98] focused on just providing formal semantics to object oriented languages and language features, the work of others [PW98, San99b] has been driven by a speci c programming problem. Our work tackles a problem in Cardelli s lexically scoped distributed programming language Obliq [Car95] Cardelli ....

....most counterexamples presented in [NHKM00, Mer00] exhibiting the problems of Obliq s original semantics) were actually discovered while using some calculus semantics to understand Obliq programs and trying to prove the correctness of surrogation. 1. 3 Related work The work closest to ours is [KS98] where an interpretation of Abadi and Cardelli s object calculus [AC96] into typed calculus is presented. Unlike [KS98] we focus on a concurrent object calculus. Gordon and Hankin [GH98] and Di Blasio and Fisher [DF96] describe two concurrent object calculi, but no account of object migration ....

[Article contains additional citation context not shown here]

J. Kleist and D. Sangiorgi. Imperative Objects and Mobile Processes. In D. Gries and W.-P. de Roever, eds, Proceedings of PROCOMET '98, pages 285-303. International Federation for Information Processing (IFIP), Chapman & Hall, 1998.

....model. In previous work on the Imperative Object Calculus (IOC) AC96a] equivalence between IOC terms was defined in a contextual way [GHL97] similar to Definition 1. In many cases, it is simpler to use a semantics by translation into cal culus to establish the equivalence between terms [KS98]. The main advantage is the large number of equivalences and algebraic laws that can be used to reason about expressions. Since IOC is (almost) a concurrency free subset of jeblik, we chose a similar path for establishing the safety of external surrogation. Proof (Overview) A proof of a ....

J. Kleist and D. Sangiorgi. Imperative Objects and Mobile Processes. In D. Gries and W.-P. de Roever, eds, Proceedings of PROCOMET '98, pages 285--303. International Federation for Information Processing (IFIP), Chapman & Hall, 1998.

....1 Introduction The work presented in this paper is in line with the research activity to use the calculus as a toolbox for reasoning about distributed object oriented programming languages. Former works on the semantics of objects as processes have shown the value of this approach: while [22, 9, 19, 10] have focused on just providing formal semantics to object oriented languages and language features, the work of others [18, 20] has been driven by a speci c programming problem. Our work tackles a problem in Obliq, Cardelli s lexically scoped distributed programming language [4] In this setting, ....

....the overall result channel p and the latest current self i n . Thus, the responsibility to signal a result on p is passed on to the respective object manager waiting at y. Objects. The semantics [ O ] k p of an object O : l j = s j ; x j )b j ] j2J , as shown in Table 4 (again similar to [10]) consists of a message that returns the object s reference s together with the current self k on channel p, a composition of replicated processes that give access to the method bodies [ b j ] k 0 r , and a new object process newOO h s; e t i that connects invocations on s from the outside ....

J. Kleist and D. Sangiorgi. Imperative Objects and Mobile Processes. In Proceedings of PROCOMET '98. Chapman & Hall, 1998.

No context found.

Kleist, J. and Sangiorgi, D., Imperative Objects and Mobile Processes, , Journal of Science of Computer Programming, Volume 44, Number 3, September 2002, 293-342.

No context found.

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. In Proc. IFIP Working Conference on Programming Concepts and Methods (PROCOMET'98). North-Holland, 1998.

No context found.

J. Kleist and D. Sangiorgi. Imperative objects and mobile processes. In Proc. of PROCOMET '98. North-Holland, 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