Jouvelot, P. and Gifford, D. K. Algebraic Reconstruction of Types and Effects. In Proceedings of the ACM Conference on Principles Of Programming Languages. ACM, New York, 1991.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....[MOTW95] we elaborate on the connections between linear logic and call by value reduction, and affine logic and call by need reduction. We conjecture that usage based program transformation in the presence of side effects could be handled by combining our usage analysis with an effect system [Luc87, LG88, JG91]. Effect systems can be used to distinguish side effecting computation from purely functional computation. 8 Related work 8.1 Linear logic The type system presented here is based on ideas taken from the linear logic of Girard [Gir87] and its successor, the Logic of Unity [Gir93] A companion ....

Pierre Jouvelot and D.K. Gifford, Algebraic reconstruction of types and effects. In ACM Symposium on Principles of Programming Languages, 1991.

....on the side effects it may cause Could we have a variant of IO that allowed exceptions but not I O Or I O but not concurrency The answer is technically, yes of course. There is a long history of research into so called effect systems, that track what kind of effects an expression can have [21]. Such effect systems can be expressed in a monadic way, or married with a monadic type system [51] However, the overhead on the programmer becomes greater, and I do not know of any language that uses such a system . An interesting challenge remains, to devise a more refined system that is ....

JOUVELOT, P., AND GIFFORD, D. Algebraic reconstruction of types and effects. In 18'th ACM Symposium on Principles of Programming Languages (POPL), Orlando. ACM, Jan. 1991. 58

....compilers and are therefore source and target language dependent. In contrast, CoSy will give a very general mechanism (programming languages) for compiler construction. There have not been many attempts to design languages neither for module re embedding nor for side effect descriptions. FX [JG90] and JADE [LR91] contain basic principles for side effect description but fSDL goes further because it allows the convenient combination of these descriptions via its domain calculus. JADE also provides a mechanism to synchronize tasks over a shared memory, however its side effect descriptions are ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In ACM Conference on Principles of Programming Languages (POPL), 1990.

....for describing precise e ects procedures with respect to the heap. The importance of procedure speci cations for pointer analysis was indicated in [14] A language for annotating software libraries is described in [6] E ects systems in general were used in functional languages with side e ects [10]. Our speci cation language bears some similarities to propositional dynamic logic [7] Similarly to [4] our e ect language speci es operations on heap. Unlike graph rewrite rules, our e ect speci cations are based on primitive e ects which correspond to statements in imperative programs. The ....

Pierre Jouvelot and David K. Giord. Algebraic reconstruction of types and effects. In ########### ## ### #### ###### ### ######### ## ### ########## ## ########### #########, 1991.

....indicating that the non key elements of the list cannot be polymorphic values. It is too bad that this purely internal data structure must be mentioned in the interface. Many researchers have recently been engaged in devising better type inference systems for polymorphic programs with references[25, 23, 17, 40, 48], which indicates that the problem of type checking references is not yet regarded as solved; some of these systems address the problem of internal, temporary references described above. The ML Grammar The designers of Standard ML worked very hard to get the semantics right, and to define the ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In Eighteenth Annual ACM Symp. on Principles of Prog. Languages, pages 303--310, New York, Jan 1991. ACM Press.

....the side effects that an expression may have during its execution. Effect system is usually coupled with a type system where the latent effect of a function is encoded in the type. Any kind of observable side effect can be analyzed by using the effect system. The original effect system [LG88, JG91, TJ92] used it to trace the memory operations, which is adopted to our work. Tofte and Talpin [TT94] used effect system for the stack allocation of all data. Xavier and Pessaux [PL99] used effect system for the analysis of uncaught exceptions. 8 3. Our approach Our langauge, etySECK, is an ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In Proceedings of The ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, pages 303--310, 1991.

....and are used to check for missing functionality. Effects were first studied in the FX system [6] a higher order functional language with reference cells. The burden of manual specification of effects is lifted through the use of effects inference as studied by Gifford, Jouvelot, and Talpin [8, 12]. The work was motivated by a desire to use stack allocation instead of heap allocation in mostly pure functional programs, and also to assist parallel code generation. The approach was demonstrated successfully for the former purpose in later work [1, 13] These researchers also make use of a ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In Conference Record of the Eighteenth Annual ACM SIGACT/SIGPLAN Symposium on Principles of Programming Languages, pages 303--310. ACM Press, New York, 1991.

....representation of the program code. 4. 1 Semantic Objects We assume monomorphic types ( which classify values, communication effects ( which classify communication behaviours we are concerned with, and abstract function closures (v) Drawing inspiration from previous work by several authors [19, 20, 21] to expose memory allocation or communication behaviours of programs, we annotate types with a combination of effects and abstract function closures. v : j hae; li j v [ v : Unit j Int j Bool j chan (j j) j v; v : j out(L) j in ( j [ The set v conservatively estimates the ....

P. Jouvelot and D.K. Gifford. Algebraic reconstruction of types and effects. In Proceedings of POPL'90, pages 303--310. ACM Press, 1990.

....towards formalizing the encapsulation of state in Haskell, the encapsulation of state in ML is of independent interest as witnessed by the large amount of previous, current, and planned work on the topic. Much of this work has been carried out in the context of effect and region inference [LG88, JG91, TJ92, TT97] In contrast to the monadic ap proach, effect systems do not restrict the interactions between the functional and imperative sublanguages at all. Expressions with global effects are allowed but are given different type descriptions than expressions with no effects. The type system ....

....The type system composes effects generated by imperative operations to annotate enclosing expressions with their effects. A special rule, called effect masking, may be used at any time to restrict the set of visible effects by eliminating local (i.e. encapsulated) ones. In some systems [LG88, JG91, TJ92] effect masking is purely an artifact of the type system and has no operational significance. For example, the absence of visible effects provides a simple criterion for generalizing type variables that avoids the subtle problems that occur when mixing polymorphism and assignments [Tof90, ....

Pierre Jouvelot and David Gifford. Algebraic reconstruction of types and effects. In Conference Record of the Eighteenth Annual ACM Symposium on Principles of Programming Languages, Orlando, Florida, pages 303--310. ACM Press, January 1991.

....on optimisations made on the basis of compile time analysis of programs. The main approaches fall into three broad classes: those based on abstract interpretation (e.g. 21] 1] those based on projections (e.g. 24] 20] 9] 8] and those based on type systems (e.g. 19] 27] 23] [18]) The interrelations of these methods has attracted some study, with Burn [7] comparing abstract interpretation with projection analysis, and Jensen [17] and Benton [5] investigating links between abstract interpretation and the typing approach of Kuo and Mishra. The main contribution of this ....

P. Jouvelot and D. K. Gifford. Algebraic reconstruction of types and effects. In 18th Annual ACM Symposium on Principles of Programming Languages, 1991.

....there is a possibility of allocating a new mutable object or updating an existing mutable object. This has to be achieved in a flexible but sound manner within and across function and local block boundaries. Many type systems in the literature follow this general framework [Dam85, Tof90, LG88, JG91, Wri92, TJ92] The various systems differ in their notion of a store abstraction and the amount of information propagated across function boundaries. An illustrative comparison of some of these systems is presented in [OJ91] First, we will briefly describe two such systems that are simple ....

....effects for expressions and that they are computable using an efficient inference algorithm. At least two effect based systems [TJ92, Wri92] propose such inference mechanisms based on structural unification [Rob65] The effect system of FX 91 [GJSO91] uses the more complex algebraic unification [JG91] which permits unification modulo algebraic identities such as associativity and commutativity. This provides more expressive power to the inference system, albeit at the cost of simplicity and efficiency. Here, we will only discuss the inference system based on standard structural unification. ....

Pierre Jouvelot and David K. Gifford. Algebraic Reconstruction of Types and Effects. In Proceedings of the 1991 ACM Conference on Principles of Programming Languages, pages 303--310. ACM, 1991.

....to restrict communication with other threads. The same technique would remove timing channels, and could be applied to our scheme. It is not clear how well this scheme works in practice; it seems likely to restrict timing and communication quite severely. Static side effect and region analysis [JG91] which aims to infer all possible sideeffects caused by a piece of code, may be able to capture effects like timing channels. 8 Conclusions This work was motivated by a desire to provide better security when using downloaded applications, by making finegrained information flow control ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In ACM Symposium on Principles of Programming Languages, pages 303--310, January 1991.

....the semantic soundness of any algorithm that decides a conservative approximation to these rules. More expressive type systems have been developed by Leroy and Weis [LW91] and, independently, by Talpin, Jouvelot and others [TJ92, Wri92] partly based on the framework of Gifford s FX system [JG91]. In contrast to the SML NJ algorithm and preceding ML compilers, which keep track of the types of all values in the store, these approaches approximately identify the values in the store that are accessible in a particular context. This additional accesibility information leads to type systems ....

P. Jouvelot and D. Gifford. Algebraic reconstruction of types and effects. In 18th ACM Symposium on Principles of Programming Languages, pages 303--310, 1991.

....in cubic time. We do not know of any other automatic transformation on inference rules that provides a terminating evaluation strategy for this rule set. Tractable rule sets also play an important role in type inference systems for computer programming languages [Talpin and Jouvelot, 1992] [Jouvelot and Gifford, 1991]. Although we have not yet investigated connections between the notion of locality used here and known results on tractability for type inference systems, this seems like a fruitful area for future research. From a practical perspective it seems possible that general purpose bottom up evaluation ....

Pierre Jouvelot and David Gifford. Algebraic Reconstruction of Types and Effects. In Proceedings of the 1991 ACM Conference on Principles of Programming Languages, 1991.

....to restrict communication with other threads. The same technique would remove timing channels, and could be applied to our scheme. It is not clear how well this scheme works in practice; it seems likely to restrict timing and communication quite severely. Static side effect and region analysis [JG91] which aims to infer all possible sideeffects caused by a piece of code, may be able to capture effects like timing channels. 8 Conclusions This work was motivated by a desire to provide better security when using downloaded applications, by making finegrained information flow control ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In ACM Symposium on Principles of Programming Languages, pages 303--310, January 1991.

....Unitary, and Idempotent algebras. Since effects define a set al..gebra, general set constraint solvers are another solution. As an aside, the type inference problem for a language with records does also requires a nonstructural unification algorithm. Effect inference rules can be found in [TJ92] and [JG91]. Slide 14 Type Effect inference (reconstruction) ffl Analogue to algorithm W (structural unification) ffl Effects have no structure, simple unification does not work. ffl Type inference = Constraint satisfaction problem. ffl Standard type inference rules define a set of equality ....

....research. Polymorphic effect systems are first described in Lucassen s PhD thesis [Luc87] That work got published in [LG88] where the authors describe an explicitly typed system for store effects that infers effect masking. The inference problem for effect systems is described and solved in [JG91]. In [TJ92] the authors add sub effecting to the inference algorithm. More recently, Tofte and Talpin [TT93] show how effect inference can be used to completely stack allocate the typed call by value lambda calculus. 11 Slide 16 Background ffl Originated in work on FX for parallel ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In Proceedings of the 18th Annual ACM SIGACT-SIGPLAN Symposium on Principles of Programming Languages, pages 303--310, January 1991.

....(as with ILC and Idealized Algol) and that the syntax of assignments, pure, and return are somewhat restrictive. Far less attention has been paid to controlling side effects in the call by value realm. Some of the most welldeveloped approaches are the type and effect systems for assignments [17, 19, 20, 41, 42], continuations [16] and exceptions [10] In each of these languages, a static system assigns both types and possible effects to expressions; the type and effect can then be used in the generation of code for parallel architectures and other applications. For example, in the type and effect ....

P. Jouvelot and D. K. Gifford. Algebraic reconstruction of types and effects. In Conference Record of the Eighteenth Annual ACM Symposium on Principles of Programming Languages, pages 303--310. ACM, 1991.

....input syntax in the form of labels. 4. 1 Semantic Objects We assume monomorphic types ( which classify values, communication effects ( which classify communication behaviours we are concerned with, and abstract function closures (v) Drawing inspiration from previous work by several authors [19, 20, 21] to expose memory allocation or communication behaviours of programs, we annotate types with a combination of effects and abstract function closures. v : j hae; li j v [ v : Unit j Int j Bool j chan(j j) j v; v : j out(L) j in( j [ The set v conservatively estimates ....

P. Jouvelot and D.K. Gifford. Algebraic reconstruction of types and effects. In Proceedings of POPL'90, pages 303--310. ACM Press, 1990.

....as its sole form of memory management. 1.2 Region Inference Rules Tofte and Talpin [16] introduced region inference not as an algorithm, but as a set of inference rules in the style of type inference rules. Region inference is based on earlier work on effects and (a different notion of) regions [13,3,6]. In particular, we use effects for determining the lifetime of values. As a first approximation, let us take an effect of an expression to be a finite set of region variables. The set is an upper bound on the set of regions that the evaluation of the expressions uses for storing and retrieving ....

P. Jouvelot and D.K. Gifford. Algebraic reconstruction of types and effects. In Proceedings of the 18th ACM Symposium on Principles of Programming Languages (POPL), 1991.

....operations that communicate with other threads. Stoughton s model [Sto81] also uses this local global approach. The same technique can be used to control timing channels. This approach could be applied to JFlow and even checked statically, similarly to static side effect and region analysis [JG91] which aims to infer all possible side effects caused by a piece of code. However, it is not clear how well this scheme works in practice; it seems likely to restrict timing and communication quite severely, particularly if applied directly to a programming model in which objects are shared ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In ACM Symposium on Principles of Programming Languages, pages 303--310, January 1991.

....system divides the store into disjoint regions and assigns each reference to one of them. An effect system approximates the effect of each expression, i.e. the set of regions that may be accessed by evaluation of e. The simplification with respect to other published region inference systems [45, 50, 70, 76] is the omission of polymorphism, leading to an analysis which is monovariant with respect to types, effects, and regions. This restriction leads to a polynomial time algorithm, as opposed to the exponential algorithm for polymorphic region and effect inference (see Sec. 4.4) In the example ....

Pierre Jouvelot and David K. Gifford. Algebraic reconstruction of types and effects. In Proc. 18th Annual ACM Symposium on Principles of Programming Languages, pages 303--310, Orlando, Florida, January 1991. ACM Press.

....heavily on optimisations made on the basis of compile time analysis of programs. The main approaches fall into three broad classes: those based on abstract interpretation (e.g. Myc81, AH87] those based on projections (e.g. WH87, Lau91, Dav93, Bur90b] and those based on type systems (e.g. [KM89, WBF93, Wad91, JG91]) The interrelations of these methods has attracted some study, with Burn [Bur90a] comparing abstract interpretation with projection analysis, and Jensen [Jen91] and Benton [Ben93] investigating links between abstract interpretation and the typing approach of Kuo and Mishra. The main contribution ....

P. Jouvelot and D. K. Gifford. Algebraic reconstruction of types and effects. In 18th Annual ACM Symposium on Principles of Programming Languages, 1991.

....expressiveness. val one o = 1 p and fun a fac b (n : int c ) int d = if e fint f g (n g 2 h ) then one else (n j (fac k (n Gamma l 1m ) n Figure 21: The Factorial with a Global Constant In polymorphic effect system that neither have notions of subtyping nor polymorphic recursion, such as [Jouvelot al. 1991], Talpin al. 1994] Thomsen 1993] or [Wadler 1991] an alias analysis will produce a type similar to 8a; b: int a b;a;c int c ) d (c being the region of one) and isolate regions as shown. The integers (n : are in the same region x as one. The integers (n Gamma 1) are in the same ....

Jouvelot, P., and Gifford, D. Algebraic Reconstruction of Type and Effect. In Proceedings of the 1991's ACM Symposium on Principles of Programming Languages. ACM Press.

No context found.

Jouvelot, P. and Gifford, D. K. Algebraic Reconstruction of Types and Effects. In Proceedings of the ACM Conference on Principles Of Programming Languages. ACM, New York, 1991.

No context found.

Jouvelot, P., and Gifford, D.K. Algebraic Reconstruction of Types and Effects. POPL 1991, 303310.

First 50 documents  Next 50

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