Futures and Promises

Total Page:16

File Type:pdf, Size:1020Kb

Futures and Promises Futures and promises May 7, 2014 Overview Overview ■ Futures in Clojure Clojure Javascript/JQuery Scala/Akka futures 2 / 15 Overview Clojure Future Future (example) Future (mechanics) Promises Javascript/JQuery Scala/Akka futures Clojure 3 / 15 Future Overview ■ Future: evaluates a body of code in another thread Clojure Future (def long-calculation Future (example) (future (apply + (range 1e8)))) Future (mechanics) Promises ◆ returns immediately Javascript/JQuery Scala/Akka ■ futures dereference @long-calculation ◆ blocks if value is not available yet ◆ returns saved value on subsequent calls ■ dereference with timeout (deref (future (Thread/sleep 5000) :done!) 1000 :impatient!) ;= :impatient! 4 / 15 Future (example) ( defn get−document [id] ; ... do some work to retrieve the identified ; document’s metadata ... {:url "http://www. mozilla .org/about/manifesto.en.html" : title "The␣Mozilla␣Manifesto" :mime "text/html" :content ( future (slurp "http://www. mozilla .org/about/manifesto.en.html"))}) ■ slurp fetches the document from the web ■ content is retreived once, starting immediately ◆ several threads requesting it wait for the same value ■ delay instead of future: starts only at first request 5 / 15 Future (mechanics) Overview ■ executed in the thread from the agents’ unlimited thread Clojure Future pool Future (example) Future ◆ may use IO (mechanics) Promises ■ Thread vs future Javascript/JQuery Scala/Akka ◆ futures cheaper than using thread directly ◆ more concise ◆ implements java.util.concurrent.Future 6 / 15 Promises Overview ■ delays and futures are bound to the result of a code Clojure Future block Future (example) ■ promise is not bound to any code block Future (mechanics) (def p (promise)) Promises Javascript/JQuery (realized? p) ;= false Scala/Akka futures (deliver p 42) ;= #<core$promise$reify__1707@3f0ba812: 42> (realized? p) ;= true @p ;= 42 ■ deliver assigns value to the promise ■ deref blocks until assigned, supports timeout too ■ similar to a one-time, single-value pipe ◆ sometimes called dataflow variables and are the building blocks of declarative concurrency (Oz) 7 / 15 Overview Clojure Javascript/JQuery Introduction Future event as object Making promises Deferred vs promise Pipes Pipe cascading Javascript/JQuery Scala/Akka futures 8 / 15 Introduction Overview ■ JQuery 1.4 and earlier Clojure Javascript/JQuery Introduction $.get(’/mydata’, { Future event as object success: onSuccess, Making promises failure: onFailure, Deferred vs promise always: onAlways Pipes }); Pipe cascading Scala/Akka futures ■ With JQuery 1.5 ajax functions you know ($.ajax, $.get, and $.post) now return Promises var promise = $.get(’/mydata’); promise.done(onSuccess); promise.fail(onFailure); promise.always(onAlways); ■ why this is better? 9 / 15 Future event as object Overview ■ All actions do not have to be defined on creation Clojure Javascript/JQuery ◆ Promise can be passed around Introduction Future event as ◆ effects (triggering animations, inserting HTML, object Making promises locking/unlocking user input, and so on) added Deferred vs incrementally (stacked) promise Pipes ■ the order of handlers is preserved Pipe cascading Scala/Akka futures ■ Combine promises ◆ in parallel, e.g. wait for 2 events (animation completion and ajax request result) ◆ pipeline, specify subsequent reactions to the future async results ■ resulting promise is the result of the final task in the series 10 / 15 Making promises ■ Either always+done or always+fail will be called var promptDeferred = new $.Deferred(); promptDeferred.always(function(){ console.log(’A choice was made:’); }); promptDeferred.done(function(){ console.log(’Starting game...’); }); promptDeferred.fail(function(){ console.log(’No game today.’); }); ■ triggering manually with resolve or reject $(’#playGame’).focus().on(’keypress’, function(e) { var Y = 121, N = 110; if (e.keyCode === Y) { promptDeferred.resolve(); } else if (e.keyCode === N) { promptDeferred.reject(); } else { return false; // our Deferred remains pending }; }); 11 / 15 Deferred vs promise Overview ■ Promise in JQuery is read-only view of a deferred Clojure Javascript/JQuery var promise1 = promptDeferred.promise(); Introduction var promise2 = promptDeferred.promise(); Future event as object console.log(promise1 === promise2); // true Making promises Deferred vs ■ In fact promise method returns self for promises promise Pipes console.log(promise1 === promise1.promise()); Pipe cascading // true Scala/Akka futures ■ convenient for protecting the deferred from external writes 12 / 15 Pipes Overview ■ you can’t attach handlers to the second task until the Clojure first one is complete Javascript/JQuery Introduction var getPromise = $.get(’/query’); Future event as object getPromise.done(function(data) { Making promises var postPromise = $.post(’/search’, data); Deferred vs promise }); Pipes // Now we’d like to attach handlers to postPromise Pipe cascading // see the problem: it is in internal scope Scala/Akka futures ■ first promise result is handled by the code, that should return second promise var getPromise = $.get(’/query’); var postPromise = getPromise.pipe(function(data) return $.post(’/search’, data); }); Effectively, pipe is a window into the future! ◆ if return simple value then it is final promise result 13 / 15 Pipe cascading Overview ■ pipe doesn’t require you to provide every possible Clojure callback: Javascript/JQuery ■ Introduction you’ll usually just want to write Future event as object var p = originalPromise.pipe(successCallback); Making promises Deferred vs promise or the following: Pipes Pipe cascading var p = originalPromise.pipe(null, failCallback); Scala/Akka futures ■ cascading: a failure is propagated to the last promise automatically var step1 = $.post(’/step1’, data1); var step2 = step1.pipe(function() { return $.post(’/step2’, data2); }); var lastStep = step2.pipe(function() { return $.post(’/step3’, data3); }); 14 / 15 Overview Clojure Javascript/JQuery Scala/Akka futures Scala/Akka futures 15 / 15.
Recommended publications
  • Composable Concurrency Models
    Composable Concurrency Models Dan Stelljes Division of Science and Mathematics University of Minnesota, Morris Morris, Minnesota, USA 56267 [email protected] ABSTRACT Given that an application is likely to make use of more The need to manage concurrent operations in applications than one concurrency model, programmers would prefer that has led to the development of a variety of concurrency mod- different types of models could safely interact. However, dif- els. Modern programming languages generally provide sev- ferent models do not necessarily work well together, nor are eral concurrency models to serve different requirements, and they designed to. Recent work has attempted to identify programmers benefit from being able to use them in tandem. common \building blocks" that could be used to compose We discuss challenges surrounding concurrent programming a variety of models, possibly eliminating subtle problems and examine situations in which conflicts between models when different models interact and allowing models to be can occur. Additionally, we describe attempts to identify represented at lower levels without resorting to rough ap- features of common concurrency models and develop lower- proximations [9, 11, 12]. level abstractions capable of supporting a variety of models. 2. BACKGROUND 1. INTRODUCTION In a concurrent program, the history of operations may Most interactive computer programs depend on concur- not be the same for every execution. An entirely sequen- rency, the ability to perform different tasks at the same tial program could be proved to be correct by showing that time. A web browser, for instance, might at any point be its history (that is, the sequence in which its operations are rendering documents in multiple tabs, transferring files, and performed) always yields a correct result.
    [Show full text]
  • Reactive Async: Expressive Deterministic Concurrency
    Reactive Async: Expressive Deterministic Concurrency Philipp Haller Simon Geries Michael Eichberg Guido Salvaneschi KTH Royal Institute of Technology, Sweden TU Darmstadt, Germany [email protected] feichberg, [email protected] Abstract tion can generate non-deterministic results because of their Concurrent programming is infamous for its difficulty. An unpredictable scheduling. Traditionally, developers address important source of difficulty is non-determinism, stemming these problems by protecting state from concurrent access from unpredictable interleavings of concurrent activities. via synchronisation. Yet, concurrent programming remains Futures and promises are widely-used abstractions that help an art: insufficient synchronisation leads to unsound pro- designing deterministic concurrent programs, although this grams but synchronising too much does not exploit hardware property cannot be guaranteed statically in mainstream pro- capabilities effectively for parallel execution. gramming languages. Deterministic-by-construction con- Over the years researchers have proposed concurrency current programming models avoid this issue, but they typi- models that attempt to overcome these issues. For exam- cally restrict expressiveness in important ways. ple the actor model [13] encapsulates state into actors This paper introduces a concurrent programming model, which communicate via asynchronous messages–a solution Reactive Async, which decouples concurrent computations that avoids shared state and hence (low-level) race condi- using
    [Show full text]
  • A Reduction Semantics for Direct-Style Asynchronous Observables
    Journal of Logical and Algebraic Methods in Programming 105 (2019) 75–111 Contents lists available at ScienceDirect Journal of Logical and Algebraic Methods in Programming www.elsevier.com/locate/jlamp A reduction semantics for direct-style asynchronous observables ∗ Philipp Haller a, , Heather Miller b a KTH Royal Institute of Technology, Sweden b Carnegie Mellon University, USA a r t i c l e i n f o a b s t r a c t Article history: Asynchronous programming has gained in importance, not only due to hardware develop- Received 31 January 2016 ments like multi-core processors, but also due to pervasive asynchronicity in client-side Received in revised form 6 March 2019 Web programming and large-scale Web applications. However, asynchronous program- Accepted 6 March 2019 ming is challenging. For example, control-flow management and error handling are much Available online 18 March 2019 more complex in an asynchronous than a synchronous context. Programming with asyn- chronous event streams is especially difficult: expressing asynchronous stream producers and consumers requires explicit state machines in continuation-passing style when using widely-used languages like Java. In order to address this challenge, recent language designs like Google’s Dart introduce asynchronous generators which allow expressing complex asynchronous programs in a familiar blocking style while using efficient non-blocking concurrency control under the hood. However, several issues remain unresolved, including the integration of analogous constructs into statically-typed languages, and the formalization and proof of important correctness properties. This paper presents a design for asynchronous stream generators for Scala, thereby ex- tending previous facilities for asynchronous programming in Scala from tasks/futures to asynchronous streams.
    [Show full text]
  • Abstract Interface Behavior of an Object-Oriented Language with Futures and Promises
    Abstract interface behavior of an object-oriented language with futures and promises 3. September 2007 Erika Abrah´ am´ 1, and Immo Grabe2, and Andreas Gruner¨ 2, and Martin Steffen3 1 Albert-Ludwigs-University Freiburg, Germany 2 Christian-Albrechts-University Kiel, Germany 3 University of Oslo, Norway 1 Motivation How to marry concurrency and object-orientation has been a long-standing issue; see e.g., [2] for an early discussion of different design choices. Recently, the thread-based model of concurrency, prominently represented by languages like Java and C# has been criticized, especially in the context of component-based software development. As the word indicates, components are (software) artifacts intended for composition, i.e., open systems, interacting with a surrounding environment. To compare different concurrency models on a solid mathematical basis, a semantical description of the interface behavior is needed, and this is what we do in this work. We present an open semantics for a core of the Creol language [4,7], an object-oriented, concurrent language, featuring in particular asynchronous method calls and (since recently [5]) future-based concurrency. Futures and promises A future, very generally, represents a result yet to be computed. It acts as a proxy for, or reference to, the delayed result from a given sequential piece of code (e.g., a method or a function body in an object-oriented, resp. a functional setting). As the client of the delayed result can proceed its own execution until it actually needs the result, futures provide a natural, lightweight, and (in a functional setting) transparent mechanism to introduce parallelism into a language.
    [Show full text]
  • N4244: Resumable Lambdas
    Document Number: N4244 Date: 2014-10-13 Reply To Christopher Kohlhoff <[email protected]> Resumable Lambdas: A language extension for generators and coroutines 1 Introduction N3977 Resumable Functions and successor describe a language extension for resumable functions, or “stackless” coroutines. The earlier revision, N3858, states that the motivation is: [...] the increasing importance of efficiently handling I/O in its various forms and the complexities programmers are faced with using existing language features and existing libraries. In contrast to “stackful” coroutines, a prime use case for stackless coroutines is in server applications that handle many thousands or millions of clients. This is due to stackless coroutines having lower memory requirements than stackful coroutines, as the latter, like true threads, must have a reserved stack space. Memory cost per byte is dropping over time, but with server applications at this scale, per-machine costs, such as hardware, rack space, network connectivity and administration, are significant. For a given request volume, it is always cheaper if the application can be run on fewer machines. Existing best practice for stackless coroutines in C and C++ uses macros and a switch-based technique similar to Duff’s Device. Examples include the coroutine class included in Boost.Asio1, and Protothreads2. Simon Tatham’s web page3, Coroutines in C, inspired both of these implementations. A typical coroutine using this model looks like this: void operator()(error_code ec, size_t n) { if (!ec) reenter (this) { for (;;) { yield socket_->async_read_some(buffer(*data_), *this); yield async_write(*socket_, buffer(*data_, n), *this); } } } The memory overhead of a stackless coroutine can be as low as a single int.
    [Show full text]
  • Concepts and Paradigms of Object-Oriented Programming Expansion of Oct 400PSLA-89 Keynote Talk Peter Wegner, Brown University
    Concepts and Paradigms of Object-Oriented Programming Expansion of Oct 400PSLA-89 Keynote Talk Peter Wegner, Brown University 1. What Is It? 8 1.1. Objects 8 1.2. Classes 10 1.3. Inheritance 11 1.4. Object-Oriented Systems 12 2. What Are Its Goals? 13 2.1. Software Components 13 2.2. Object-Oriented Libraries 14 2.3. Reusability and Capital-intensive Software Technology 16 2.4. Object-Oriented Programming in the Very Large 18 3. What Are Its Origins? 19 3.1. Programming Language Evolution 19 3.2. Programming Language Paradigms 21 4. What Are Its Paradigms? 22 4.1. The State Partitioning Paradigm 23 4.2. State Transition, Communication, and Classification Paradigms 24 4.3. Subparadigms of Object-Based Programming 26 5. What Are Its Design Alternatives? 28 5.1. Objects 28 5.2. Types 33 5.3. Inheritance 36 5.4. Strongly Typed Object-Oriented Languages 43 5.5. Interesting Language Classes 45 5.6. Object-Oriented Concept Hierarchies 46 6. what Are Its Models of Concurrency? 49 6.1. Process Structure 50 6.2. Internal Process Concurrency 55 6.3. Design Alternatives for Synchronization 56 6.4. Asynchronous Messages, Futures, and Promises 57 6.5. Inter-Process Communication 58 6.6. Abstraction, Distribution, and Synchronization Boundaries 59 6.7. Persistence and Transactions 60 7. What Are Its Formal Computational Models? 62 7.1. Automata as Models of Object Behavior 62 7.2. Mathematical Models of Types and Classes 65 7.3. The Essence of Inheritance 74 7.4. Reflection in Object-Oriented Systems 78 8.
    [Show full text]
  • From Events to Futures and Promises and Back
    From Events to Futures and Promises and back Martin Sulzmann Faculty of Computer Science and Business Information Systems Karlsruhe University of Applied Sciences Moltkestrasse 30, 76133 Karlsruhe, Germany [email protected] Abstract. Events based on channel communications and futures/promises are powerful but seemingly different concepts for concurrent program- ming. We show that one concept can be expressed in terms of the other with surprisingly little effort. Our results offer light-weight library based approaches to implement events and futures/promises. Empirical results show that our approach works well in practice. 1 Introduction There exists a wealth of programming models to make concurrent programming easier. A popular form is Communicating Sequential Processes (CSP) [7] which has found its way into several programming languages in one form or another. For example, Haskell [9] provides native support for a primitive channel type called MVar which is effectively a buffered channel of size one. The Go pro- gramming language [5] follows more closely CSP and has native support for synchronous channel-based message passing and selective communication. Con- current ML (CML) [11] provides for even richer abstractions in the form of events. Another popular concurrency approach is the concept of futures/promises [1, 4]. For example, Alice [12] provides native support for futures/promises whereas Scala [6] follows a library-based approach and provides a powerful set of combi- nators to compose futures/promises. Our interest lies in establishing some connections among channel-based events and futures/promises by expressing one concept in terms of the other. This is interesting and helpful for several reasons.
    [Show full text]
  • Integrating Rx and Async for Direct-Style Reactive Streams
    RAY: Integrating Rx and Async for Direct-Style Reactive Streams Philipp Haller Heather Miller Typesafe, Inc. EPFL [email protected] heather.miller@epfl.ch ABSTRACT (it's even possible for a stream to never complete.) Given Languages like F#, C#, and recently also Scala, provide the richer substrate of reactive streams, the Async model \async" extensions which aim to make asynchronous pro- has to be generalized in several dimensions. gramming easier by avoiding an inversion of control that is inherent in traditional callback-based programming models We call our model RAY, inspired by its main constructs, (for the purpose of this paper called the \Async" model). rasync, await, yield, introduced later in the paper. This paper outlines a novel approach to integrate the Async model with observable streams of the Reactive Extensions This paper makes the following contributions: model which is best-known from the .NET platform, and of which popular implementations exist for Java, Ruby, and • A design of a new programming model, RAY, which in- other widespread languages. We outline the translation of tegrates the Async model and the Reactive Extensions \Reactive Async" programs to efficient state machines, in a model in the context of the Scala Async project [5] (pro- way that generalizes the state machine translation of regular posed for adoption in mainline Scala [6]); Async programs. Finally, we sketch a formalization of the Reactive Async model in terms of a small-step operational • An operational semantics of the proposed program- semantics. ming model. Our operational semantics extends the formal model presented in [1] for C#'s async/await to 1.
    [Show full text]
  • LEAGER PROGRAMMING Abstract
    LEAGER PROGRAMMING Anthony Chyr UUCS-19-002 School of Computing University of Utah Salt Lake City, UT 84112 USA February 26, 2019 Abstract Leager programming, a portmanteau of “lazy” and “eager” or “limit” and “eager,” is an evaluation strategy that mixes lazy evaluation and eager evaluation. This evaluation strategy allows iterators to precompute the next value in a separate thread, storing the result in a cache until it is needed by the caller. Leager programming often takes the form of an iterator, which alone allows data to be prefetched, and when chained together can be used to form concurrent pipelines. We found a dramatic reduction in latency on par with code written with asynchronous callbacks, while making minimal modifications to the initial sequential code. LEAGER PROGRAMMING by Anthony Chyr A senior thesis submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Bachelor of Computer Science School of Computing The University of Utah May 2019 Approved: / / Matthew Flatt H. James de St. Germain Supervisor Director of Undergraduate Studies School of Computing / Ross Whitaker Diretor School of Computing Copyright c Anthony Chyr 2019 All Rights Reserved ABSTRACT Leager programming, a portmanteau of “lazy” and “eager” or “limit” and “eager,” is an evaluation strategy that mixes lazy evaluation and eager evaluation. This evaluation strategy allows iterators to precompute the next value in a separate thread, storing the result in a cache until it is needed by the caller. Leager programming often takes the form of an iterator, which alone allows data to be prefetched, and when chained together can be used to form concurrent pipelines.
    [Show full text]
  • A Programming Model and Language for Concurrent and Distributed Object-Oriented Systems
    A Programming Model and Language for Concurrent and Distributed Object-Oriented Systems Jan Schäfer Vom Fachbereich Informatik der Technischen Universität Kaiserslautern zur Verleihung des akademischen Grades Doktor-Ingenieur (Dr.-Ing.) genehmigte Dissertation Datum der wissenschaftlichen Aussprache: 5. November 2010 Dekan: Prof. Dr. Karsten Berns Vorsitzender der Promotionskommission: Prof. Dr. Reinhard Gotzhein Erster Berichterstatter: Prof. Dr. Arnd Poetzsch-Heffter Zweiter Berichterstatter: Prof. Dr. Einar Broch Johnsen D 386 Cover image: http://www.flickr.com/photos/handersongomes/2471526525/ (c) Handerson Gomes under the Creative Commons Licence (CC BY 2.0) http://creativecommons.org/licenses/by/2.0/deed.en To my family Abstract The wide availability of multi-core processors and the ubiquitous presence of the Internet lead to new challenges in software design and implementation. Software has to be written in a parallelizable way to profit from multiple cores. Interaction with distributed Internet services requires coping with message delays and network failures. These challenges reach application domains, like desktop applications, which have been mainly written in a sequential way in the past. The concurrency model of mainstream object-oriented programming languages is based preemptively scheduled threads, which concurrently work on a shared object-heap. This programming model is highly prone to race conditions, i.e., hard to find concurrency-related errors that are not easily reproducible. To synchronize threads and prevent data races, operating system mechanisms like locks have to be used. Experience shows that this programming model is too difficult for most programmers, is not very modular, and is not well suited for the behavioral description of software components. Furthermore, the thread-based model is not appropriate for realizing distributed systems, due to its inherent synchronous communication model.
    [Show full text]
  • The Essence of Event-Driven Programming
    The Essence of Event-Driven Programming Jennifer Paykin1, Neelakantan R. Krishnaswami2, and Steve Zdancewic3 1 University of Pennsylvania, Philadelphia, USA [email protected] 2 University of Birmingham, Birmingham, United Kingdom [email protected] 3 University of Pennsylvania, Philadelphia, USA [email protected] Abstract Event-driven programming is based on a natural abstraction: an event is a computation that can eventually return a value. This paper exploits the intuition relating events and time by drawing a Curry-Howard correspondence between a functional event-driven programming lan- guage and a linear-time temporal logic. In this logic, the eventually proposition ♦A describes the type of events, and Girard’s linear logic describes the effectful and concurrent nature of the programs. The correspondence reveals many interesting insights into the nature of event- driven programming, including a generalization of selective choice for synchronizing events, and an implementation in terms of callbacks where ♦A is just ¬ ¬A. Digital Object Identifier 10.4230/LIPIcs... 1 Introduction Event-driven programming is a popular approach to functional concurrency in which events, also known as “futures,” “deferred values,” or “lightweight threads,” execute concurrently with each other and eventually produce a result. The abstraction of the event has been extremely successful in producing lightweight, extensible, and efficient concurrent programs. As a result, a wide range of programming languages and libraries use the event-driven paradigm to describe everything from message-passing concurrency [24] and lightweight threads [26] to graphical user interfaces [11] and I/O [13, 21, 25]. Although these systems vary considerably in the details of their implementations and APIs, they share a common, basic structure.
    [Show full text]
  • Software Transactional Memory
    Software Transactional Memory Jaakko Järvi University of Bergen <2018-10-02 Tue> Jaakko Järvi (University of Bergen) Software Transactional Memory <2018-10-02 Tue> 1 / 11 Outline 1 Examples Jaakko Järvi (University of Bergen) Software Transactional Memory <2018-10-02 Tue> 2 / 11 Both transactions maintain invariant, yet without early fail can “crash” Examples Early Rollback is Important Assume invariant x == 0 && y == 0, and two concurrently executing threads: atomic { if (x != y) crash(); } atomic { ++x; ++y; } Jaakko Järvi (University of Bergen) Software Transactional Memory <2018-10-02 Tue> 3 / 11 Examples Early Rollback is Important Assume invariant x == 0 && y == 0, and two concurrently executing threads: atomic { if (x != y) crash(); } atomic { ++x; ++y; } Both transactions maintain invariant, yet without early fail can “crash” Jaakko Järvi (University of Bergen) Software Transactional Memory <2018-10-02 Tue> 3 / 11 Examples Deposit with lock synchronization (Java) public void deposit(double amount){ System.out.println("Depositing "+ amount); double nb=0; balanceChangeLock.lock(); try{ nb= balance+ amount; balance= nb; } finally{ balanceChangeLock.unlock(); } System.out.println("New balance is "+ nb); } Jaakko Järvi (University of Bergen) Software Transactional Memory <2018-10-02 Tue> 4 / 11 Examples Deposit with STM (hypothetical Java) public void deposit(double amount){ System.out.println("Depositing "+ amount); double nb=0; atomic{ nb= balance+ amount; balance= nb; } System.out.println("New balance is "+ nb); } Jaakko Järvi (University of Bergen) Software Transactional Memory <2018-10-02 Tue> 5 / 11 Lock all accounts Manually decide what needs to be undone after each kind of exception Side-effects might be visible in other threads before undone Examples Composing critical sections (lock syncrhonization) class Bank{ Accounts accounts; ..
    [Show full text]