Martin Odersky. A functional theory of local names. In Proceedings of 21st Annual Symposium on Principles of Programming Languages, pages 48--59, Portland, Oregon, January 1994. ACM Press.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....in first order logic. An equation a = b between names is unsatisfiable if and only if a and b are distinct. The ae calculus ensures that all abstractions are equipped with a unique name. On creation of a new abstraction a new name is created, relying on a mechanism independently proposed in [14, 10, 9]. Without the above invariant, confluence of the ae calculus would fail. For instance, consider the following expression E being a composition of two abstractions with the same name a: a: x = a: y = E reduces to non joinable expressions when composed with an application a z : E E a z ....

Martin Odersky. A Functional Theory of Local Names. In POPL, pages 48--59, January 1994.

....of abstract types. Roughly, a system declaration is very similar to a packed abstract type, while a system use corresponds to a combination of unpacking and repacking. The idea of embedding names into a lambda calculus for controlling side effects can also be found in Odersky s work [22]. While Odersky proposes a calculus of local names, our module calculus which controlIs the side effect caused by structure and type generation could be called a calculus of global names. While we favour a strict separation of implementation and module language, other authors try to integrate ....

Martin Odersky. A functional theory of local names. In POPL-21 [24], pages 48-59.

....of fresh names that can be tested for inequality. Use of this seemingly small computational effect (which in fact has rather subtle interactions with higher order functions: see [26] of course has a long history in functional programming, from Lisp s gensym to more recent advocates, such as [21, 7]. However, combining it with name swapping and controlling the two through the use of the abstraction type #bty#ty is unique to FreshML. Since the original design was published in [25] the use of swapping and freshness to deal with # equivalence and name restriction has been taken up by others, ....

M. Odersky. A functional theory of local names. In POPL 1994.

....location types [20] and provides a static type system which prevents access control to resources for which the required capability is not held. But, because there is no solid notion of an object, or object reference, it is unclear how to apply their system to object systems. Odersky s calculus [36] borrows names from the calculus to model local names in functional programming languages. But here only scoping of names, not of values or references, is modelled, and there is no type system which localises reference sharing. Leroy [27] formalises security properties which protect a web ....

Martin Odersky. A functional theory of local names. In 21th ACM Symposium on Principles of Programming Languages, pages 48--59, January 1994.

....well typed, however, every context visible in the object must be outside or equal to the representation context. In particular, to guarantee that 0 , we must have 0 , since we have the assumption . The new operation was inspired by Odersky s functional theory of local names [36]. The new expression evaluates as follows: new in [ 0 ; hide = in where and are fresh, and 7 [ 0 : is a newly created object with object identi er . The elds of the new object are also initialised in the evaluation. This is how objects are ....

Martin Odersky. A functional theory of local names. In 21th ACM conference on Principles of Programming Languages, January 1994.

....p 0 in v:A) as : p 00 ; x:A 0 in b:B) 1 ; 1 ; u) Figure 14: Dynamic Semantics Unfortunately, the type system cannot express or enforce such distinctions. 11 Related Work Our new term which generates new names has been studied by Pitts and Stark [29] and independently by Odersky [26] in a di erent context. It also resembles operator in the calculus and related calculi [23] These do not include names in types. Recently Cardelli, Ghelli and Gordon use names at the level of types to prove security properties about the calculus [7] and the Ambient calculus [6] and to ....

Martin Odersky. A functional theory of local names. In 21th ACM conference on Principles of Programming Languages, January 1994.

....Case 1 of our example proof. 6 4. TYPE SYSTEM M 2 In the second case, however, D 0 = pD 0 q has functional type, and therefore ndseq must traverse the binder before it recurses. It does so by extending the regular world by a new parameter block (indicated by the new notation [Ode94] u : nd G 1 ; h : hyp G 1 . The result of the recursive call on (D 0 u) yields a sequent derivation (E 0 h) which is parametric in h and to which impR is applied before it is returned as result. The third case however simply returns the appropriate LF object after recursing on D 1 and D 2 ....

Martin Odersky. A functional theory of local names. In Proceedings of 21st Annual Symposium on Principles of Programming Languages, pages 48-59, Portland, Oregon, January 1994. ACM Press.

....[MT91] In the full paper we present some key technical properties and a proof that the compiler optimisation performed after so called strictness analysis is still sound in the presence of observable sharing. 4. 4 Relation to Other Calculi Similar languages have been considered by Odersky [Ode94] call by name semantics) and Pitts and Stark [PS93] call by value semantics) A reductioncalculus approach to call by need was introduced in [AFM 95] and extended to deal with mutable state in recent work of Ariola and Sabry [AS98] The reduction calculi approach in general has been ....

Martin Odersky. A functional theory of local names. In POPL'94, pages 48--59, ACM Press, 1994.

....u : nd G 1 ; h : hyp G 1 val (E h) ndseq (D 0 u) in (impR (h: E h) end The argument to ndseq is the representation of D from the proof of Lemma 1. Since D 0 = pD 0 q is hypothetical, D 0 has functional type, and we must introduce two new parameters (indicated by the new notation [Ode94]) u : nd G 1 ; h : hyp G 1 . The result of the recursive call on (D 0 u) yields a sequent derivation (E 0 h) which is parametric in h and to which impR is applied before it is returned as result. 2 6 4. TYPE SYSTEM M 2 Let Phi : Delta j Phi; u : nd pGq; h : hyp pGq fun ndseq u = ....

Martin Odersky. A functional theory of local names. In Proceedings of 21st Annual Symposium on Principles of Programming Languages, pages 48--59, Portland, Oregon, January 1994. ACM Press.

....1.1: Contributions of the present dissertation to the design of functional programming languages with assignment. flexible relationship between locality of names and locality of store computations. We also have a correction to make to previous type systems proposed for both ILC and l var [Chen and Odersky, 1994] these published type systems are now known to violate the essential property that reductions preserve typings. The error is classical, having been noticed by Reynolds almost two decades ago in [Reynolds, 1978] In this dissertation, we adapt a type system proposed in [Launchbury and Peyton ....

....stuck without yielding one of the set of sanctioned answer terms) With a notion of runtime error in the calculus it becomes possible to study the issue of type systems for guaranteeing safe purifiability as separate entities from the underlying computational formalism. Chen and Odersky [Chen and Odersky, 1994] give a type system for l var based on the general techniques of ILC, but 3 We give a more detailed exposition of this problem in Chapter 6. 6 this type system suffers from the same subtle flaw as ILC; the type system we present in Chapter 6 remedies this flaw. The results of our previous work ....

[Article contains additional citation context not shown here]

Martin Odersky. A functional theory of local names. In Conference Record of POPL '94: 21st ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, Portland, Oregon, pages 48-- 59, New York, January 1994. ACM Press.

....from the limitations suggested by Backus and sets up a truly object based paradigm for thinking about imperative programs. Reynolds s program for the semantics of imperative languages was further developed by Oles and Tennent [22,23,38 40] and continued and expanded in a number of works [13,16,24,18,20,36,17,35]. In a separate line of development, a model based more explicitly on a notion of objects has been formulated in [28,29] Reynolds s conception of imperative programming expressed above formed an important pre theoretic motivation for this work, though its theoretical development also draws ....

M. Odersky. A functional theory of local names. In Conf. Record 20th ACM Symp. on Principles of Programming Languages, Charleston, South Carolina, 1993. ACM, New York.

....the ffi module calculus is to integrate the side effect of generating module instances into a calculus with a simple rewriting semantics. The idea to use names for representing instances is inspired by calculi which integrate side effects into functional languages, especially Odersky s calculus [19]. While is essentially a calculus of local names, ffi may be characterized as a calculus of global names. The ffi module calculus is a predicative calculus, where types depend on values. A non standard variant of weak sums is used to explain module generativity. Similar in spirit, Mitchell and ....

Martin Odersky. A functional theory of local names. In POPL-21 [20], pages 48--59.

....executed in a continuation that tests the store in ways that are not expressible in Scheme itself. It is possible to construct denotational semantics in which some equivalences like (1) are true, but they are quite complex, and none extend far enough to encompass real languages [MS88, Ole85, OT95, Ode94] We seek to prove the correctness of such source level transformations and translations in compilers for languages like Scheme and ML. This setting is challenging in a number of respects: ffl We seek methods that are applicable to both untyped and typed languages, including Scheme, ML, and ....

....Their translation uses a monadic style in place of our continuation passing style. Their treatment does not include a store, and it is not clear whether their system can be extended to deal with latently recursively typed languages like Scheme. Most work on store semantics [MS88, Ole85, OT95, Ode94] deals with stores in stack discipline languages, and are complicated by the need to model that restriction. Most of this work depends crucially on the type structure of the language; our work is applicable to untyped or latently typed languages as well. PS93] treats heap allocated names without ....

Martin Odersky. A functional theory of local names. In Conference Record of POPL '94: 21st ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, pages 48--59, 1994.

....of abstract types. Roughly, a system declaration is very similar to a packed abstract type, while a system use corresponds to a combination of unpacking and repacking. The idea of embedding names into a lambda calculus for controlling side effects can also be found in Odersky s work [22]. While Odersky proposes a calculus of local names, our module calculus which controlls the side effect caused by structure and type generation could be called a calculus of global names. While we favour a strict separation of implementation and module language, other authors try to integrate ....

Martin Odersky. A functional theory of local names. In POPL-21 [24], pages 48--59.

....in first order logic. An equation a = b between names is unsatisfiable if and only if a and b are distinct. The ae calculus ensures that all abstractions are equipped with a unique name. On creation of a new abstraction a new name is created, relying on a mechanism independently proposed in [14, 10, 9]. Without the above invariant, confluence of the ae calculus would fail. For instance, consider the following expression E being a composition of two abstractions with the same name a: a: x = a: y = E reduces to non joinable expressions when composed with an application a z : E E a z ....

Martin Odersky. A Functional Theory of Local Names. In POPL, pages 48--59, January 1994.

.... and imperative programming, e.g. Swarup et al. 1991; Wadler, 1990b; Wadler, 1990a; Peyton Jones and Wadler, 1993; Guzm an and Hudak, 1990; Launchbury and Peyton Jones, 1995) Conservative extension results of the kind considered here have been a specific concern in (Odersky et al. 1993; Odersky, 1994; Riecke, 1993; Riecke and Viswanathan, 1995) 2 Idealized Algol Idealized Algol extends simply typed functional programming with primitive types for imperative features. We take the language PCF, a typed calculus with recursion and basic arithmetic constructs, as our representative pure ....

Odersky, M. 1994. A functional theory of local names. In: Conf. Record 21st ACM Symp. on Principles of Programming Languages. ACM, New York.

....prefixed by a ( replicator. This is similar to the role of abstractions in linear calculus [1] Second, an abstraction and its argument are matched by name rather than by position. Fresh local names are introduced by a restriction prefix x (this has also been studied in the context of calculus [22, 18]) Notational Conventions. fn(M) the set of free names of a term M , is given by fn(x) fxg fn(x:M) fn(M)nfxg fn(x y:M) fxg [ fn(M)nfyg) fn(xM) fxg [ fn(M) fn(M j N) fn(M) fn(N) fn( M) fn(M) fn(0) fg: x=y]M denotes substitution of x for all free occurrences of y in M . To ....

Martin Odersky. A functional theory of local names. In Proc. 21st ACM Symposium on Principles of Programming Languages, pages 48--59, January 1994.

No context found.

Martin Odersky. A functional theory of local names. In Proceedings of 21st Annual Symposium on Principles of Programming Languages, pages 48--59, Portland, Oregon, January 1994. ACM Press.

No context found.

Odersky, Martin. (1994). A functional theory of local names. Pages 48--59 of: Symposium on Principles of Programming Languages, POPL'94.

No context found.

Martin Odersky. A functional theory of local names. In Proc. 21st ACM Symposium on Principles of Programming Languages, pages 48--59, January 1994.

No context found.

Martin Odersky. A functional theory of local names. In Proceedings of 21st Annual Symposium on Principles of Programming Languages, pages 48--59, Portland, Oregon, January 1994. ACM Press.

No context found.

M. Odersky. A functional theory of local names. In Conference Record of the 21st Annual ACM Symposium on Principles of Programming Languages, pages 48--59. ACM Press, 1994.

No context found.

M. Odersky. A functional theory of local names. In Conference Record of POPL '94: 21st ACM Symposium on Principles of Programming Languages, pages 48--59. ACM Press, 1994. (p. 124)

No context found.

Martin Odersky. A functional theory of local names. In ACM, editor, Proceedings of 21st Annual ACM SIGACT-SIGPLAN Symposium on Principles of Programming Languages (POPL), pages 48--59, New York, NY, USA, January 1994. ACM Press.

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