The dot is a highly valued piece of syntactic real estate. Programming languages use the dot for a range of programming constructs: In object-oriented languages, the dot is used for field and method access on classes and objects. In imperative languages, the dot is often used for field access on structs. In functional languages, the dot is often used for field access on records. And finally in some other languages, the dot is used to access the length of arrays and to index into tuples. The dot is minimal yet aesthetically pleasing. More importantly, the dot is the great enabler of auto-completion. Programmers with a specific entity in mind (e.g., an object, struct, or record) can use the dot to trigger auto-completion to be presented with a list of operations available on the entity. This simple feature improves the programming experience and makes it easier to learn new APIs. In this paper, we explore how to overload the dot to support various programming constructs within one language while supporting type inference. We present a principled approach to overloading based on type classes extended with associated types, associated effects, and compiler-generated type classes and instances. We implement the design in the Flix programming language and evaluate its impact on compiler performance and on the quality of error messages.

We propose type qualifiers based on Boolean algebras. Traditional type systems with type qualifiers have been based on lattices, but lattices lack the ability to express exclusion . We argue that Boolean algebras, which permit exclusion, are a practical and useful choice of domain for qualifiers. In this paper, we present a calculus System F <:B that extends System F <: with type qualifiers over Boolean algebras and has support for negation, qualifier polymorphism, and subqualification. We illustrate how System F <:B can be used as a design recipe for a type and effect system, System F <:BE , with effect polymorphism, subeffecting, and polymorphic effect exclusion. We use System F <:BE to establish formal foundations of the type and effect system of the Flix programming language. We also pinpoint and implement a practical form of subeffecting: abstraction-site subeffecting. Experimental results show that abstraction-site subeffecting allows us to eliminate all effect upcasts present in the current Flix Standard Library.

We present a comprehensive overview of the Datalog facilities in the Flix programming language. We show how programmers can write functions implemented as Datalog programs and we demonstrate how to build modular and reusable families of Datalog programs using first-class Datalog program values, rho abstraction, parametric polymorphism, and type classes. We describe several features that improve the ergonomics, flexibility, and expressive power of Datalog programming in Flix, including the inject and query program constructs, head and guard expressions, functional predicates, lattice semantics, and more. We illustrate Datalog programming in Flix with several applications, including implementations of Ullman's algorithm to stratify Datalog programs, the Ford-Fulkerson algorithm for maximum flow, and the IFDS and IDE algorithms for context-sensitive program analysis. The implementations of IFDS and IDE fulfill a long-term goal: to have fully modular, polymorphic, typed, and declarative formulations of these algorithms that can be instantiated with any abstract domain.

Dynamically typed array languages such as Python, APL, and Matlab lift scalar operations to arrays and replicate scalars to fit applications. We present a mechanism for automatically inferring map and replicate operations in a statically-typed language in a way that resembles the programming experience of a dynamically-typed language while preserving the static typing guarantees. Our type system—which supports parametric polymorphism, higher-order functions, and top-level let-generalization—makes use of integer linear programming in order to find the minimum number of operations needed to elaborate to a well-typed program. We argue that the inference system provides useful and unsurprising guarantees to the programmer. We demonstrate important theoretical properties of the mechanism and report on the implementation of the mechanism in the statically-typed array programming language Futhark.

We present associated effects, a programming language feature that enables type classes to abstract over the effects of their function signatures, allowing each type class instance to specify its concrete effects. Associated effects significantly increase the flexibility and expressive power of a programming language that combines a type and effect system with type classes. In particular, associated effects allow us to (i) abstract over total and partial functions, where partial functions may throw exceptions, (ii) abstract over immutable data structures and mutable data structures that have heap effects, and (iii) implement adaptors that combine type classes with algebraic effects. We implement associated effects as an extension of the Flix programming language and refactor the Flix Standard Library to use associated effects, significantly increasing its flexibility and expressive power. Specifically, we add associated effects to 11 type classes, which enables us to add 28 new type class instances. CCS Concepts: • Software and its engineering → Language features .

We present purity reflection, a programming language feature that enables higher-order functions to inspect the purity of their function arguments and to vary their behavior based on this information. The upshot is that operations on data structures can selectively use lazy and/or parallel evaluation while ensuring that side effects are never lost or re-ordered. The technique builds on a recent Hindley-Milner style type and effect system based on Boolean unification which supports both effect polymorphism and complete type inference. We illustrate that avoiding the so-called 'poisoning problem' is crucial to support purity reflection. We propose several new data structures that use purity reflection to switch between eager and lazy, sequential and parallel evaluation. We propose a DelayList, which is maximally lazy but switches to eager evaluation for impure operations. We also propose a DelayMap which is maximally lazy in its values, but also exploits eager and parallel evaluation. We implement purity reflection as an extension of the Flix programming language. We present a new effect-aware form of monomorphization that eliminates purity reflection at compile-time. And finally, we evaluate the cost of this new monomorphization on compilation time and on code size, and determine that it is minimal.

We propose restrictable variants as a simple and practical alternative to extensible variants. Restrictable variants combine nominal and structural typing: a restrictable variant is an algebraic data type indexed by a type-level set formula that captures its set of active labels. We introduce new pattern-matching constructs that allows programmers to write functions that only match on a subset of variants, i.e., pattern-matches may be non-exhaustive. We then present a type system for restrictable variants which ensures that such non-exhaustive matches cannot get stuck at runtime. An essential feature of restrictable variants is that the type system can capture structure-preserving transformations: specifically the introduction and elimination of variants. This property is important for writing reusable functions, yet many row-based extensible variant systems lack it. In this paper, we present a calculus with restrictable variants, two partial pattern-matching constructs, and a type system that ensures progress and preservation. The type system extends Hindley-Milner with restrictable variants and supports type inference with an extension of Algorithm W with Boolean unification. We implement restrictable variants as an extension of the Flix programming language and conduct a few case studies to illustrate their practical usefulness.

The λ_Dat calculus brings together the power of functional and declarative logic programming in one language. In λ_Dat, Datalog constraints are first-class values that can be constructed, passed around as arguments, returned, composed with other constraints, and solved. A significant part of the expressive power of Datalog comes from the use of negation. Stratified negation is a particularly simple and practical form of negation accessible to ordinary programmers. Stratification requires that Datalog programs must not use recursion through negation. For a Datalog program, this requirement is straightforward to check, but for a λ_Dat program, it is not so simple: A λ_Dat program constructs, composes, and solves Datalog programs at runtime. Hence stratification cannot readily be determined at compile-time. In this paper, we explore the design space of stratification for λ_Dat. We investigate strategies to ensure, at compile-time, that programs constructed at runtime are guaranteed to be stratified, and we argue that previous design choices in the Flix programming language have been suboptimal.

Type and effect systems have been successfully used to statically reason about effects in many different domains, including region-based memory management, exceptions, and algebraic effects and handlers. Such systems’ soundness is often stated in terms of the absence of effects. Yet, existing systems only admit indirect reasoning about the absence of effects. This is further complicated by effect polymorphism which allows function signatures to abstract over arbitrary, unknown sets of effects. We present a new type and effect system with effect polymorphism as well as union, intersection, and complement effects. The effect system allows us to express effect exclusion as a new class of effect polymorphic functions: those that permit any effects except those in a specific set. This way, we equip programmers with the means to directly reason about the absence of effects. Our type and effect system builds on the Hindley-Milner type system, supports effect polymorphism, and preserves principal types modulo Boolean equivalence. In addition, a suitable extension of Algorithm W with Boolean unification on the algebra of sets enables complete type and effect inference. We formalize these notions in the λ ∁ calculus. We prove the standard progress and preservation theorems as well as a non-standard effect safety theorem: no excluded effect is ever performed. We implement the type and effect system as an extension of the Flix programming language. We conduct a case study of open source projects identifying 59 program fragments that require effect exclusion for correctness. To demonstrate the usefulness of the proposed type and effect system, we recast these program fragments into our extension of Flix.

As type and effect systems become more expressive there is an increasing need for efficient type inference. We consider a polymorphic effect system based on Boolean formulas where inference requires Boolean unification. Since Boolean unification involves semantic equivalence, conventional syntax-driven unification is insufficient. At the same time, existing Boolean unification techniques are ill-suited for type inference. We propose a hybrid algorithm for solving Boolean unification queries based on Boole’s Successive Variable Elimination (SVE) algorithm. The proposed approach builds on several key observations regarding the Boolean unification queries encountered in practice, including: (i) most queries are simple, (ii) most queries involve a few flexible variables, (iii) queries are likely to repeat due similar programming patterns, and (iv) there is a long tail of complex queries. We exploit these observations to implement several strategies for formula minimization, including ones based on tabling and binary decision diagrams. We implement the new hybrid approach in the Flix programming language. Experimental results show that by reducing the overhead of Boolean unification, the compilation throughput increases from 8,580 lines/sec to 15,917 lines/sec corresponding to a 1.8x speed-up. Further, the overhead on type and effect inference time is only 16% which corresponds to an overhead of less than 7% on total compilation time. We study the hybrid approach and demonstrate that each design choice improves performance.

We present the design values and design principles of the Flix programming language, a functional-first, imperative, and logic programming language. We explain how these values and principles came into being and how they have influenced the design of Flix over the last several years.

We present a simple, practical, and expressive relational nullable type system. A relational nullable type system captures whether an expression may evaluate to null based on its type, but also based on the type of other related expressions. The type system extends the Hindley-Milner type system with Boolean constraints, supports parametric polymorphism, and preserves principal types modulo Boolean equivalence. We show how to support full Hindley-Milner style type inference with an extension of Algorithm W. We conduct a preliminary study of open source projects showing that there is a need for relational nullable type systems across a wide range of programming languages. The most important findings from the study are: (i) programmers use programming patterns where the nullability of one expression depends on the nullability of other related expressions, (ii) such invariants are commonly enforced with run-time exceptions, and (iii) reasoning about these programming patterns requires not only knowledge of when an expression may evaluate to null, but also when it may evaluate to a non-null value. We incorporate these observations in the design of the proposed relational nullable type system.

Traditionally, web applications have been written as HTML pages with embedded JavaScript code that implements dynamic and interactive features by manipulating the Document Object Model (DOM) through a low-level browser API. However, this unprincipled approach leads to code that is brittle, difficult to understand, non-modular, and does not facilitate incremental update of user-interfaces in response to state changes. React is a popular framework for constructing web applications that aims to overcome these problems. React applications are written in a declarative and object-oriented style, and consist of components that are organized in a tree structure. Each component has a set of properties representing input parameters, a state consisting of values that may vary over time, and a render method that declaratively specifies the subcomponents of the component. React’s concept of reconciliation determines the impact of state changes and updates the user-interface incrementally by selective mounting and unmounting of subcomponents. At designated points, the React framework invokes lifecycle hooks that enable programmers to perform actions outside the framework such as acquiring and releasing resources needed by a component. These mechanisms exhibit considerable complexity, but, to our knowledge, no formal specification of React’s semantics exists. This paper presents a small-step operational semantics that captures the essence of React, as a first step towards a long-term goal of developing automatic tools for program understanding, automatic testing, and bug finding for React web applications. To demonstrate that key operations such as mounting, unmounting, and reconciliation terminate, we define the notion of a well-behaved component and prove that well-behavedness is preserved by these operations.

Datalog is a declarative logic programming language that has been used in a variety of applications, including big-data analytics, language processing, networking and distributed systems, and program analysis. In this paper, we propose first-class Datalog constraints as a mechanism to construct, compose, and solve Datalog programs at run time. The benefits are twofold: We gain the full power of a functional programming language to operate on Datalog constraints-as-values, while simultaneously we can use Datalog where it really shines: to declaratively express and solve fixpoint problems. We present an extension of the lambda calculus with first-class Datalog constraints, including its semantics and a type system with row polymorphism based on Hindley-Milner. We prove soundness of the type system and implement it as an extension of the Flix programming language.

We present a simple, practical, and expressive type and effect system based on Boolean constraints. The effect system extends the Hindley-Milner type system, supports parametric polymorphism, and preserves principal types modulo Boolean equivalence. We show how to support type inference by extending Algorithm W with Boolean unification based on the successive variable elimination algorithm. We implement the type and effect system in the Flix programming language. We perform an in-depth evaluation on the impact of Boolean unification on type inference time and end-to-end compilation time. While the computational complexity of Boolean unification is NP-hard, the experimental results demonstrate that it works well in practice. We find that the impact on type inference time is on average a 1.4x slowdown and the overall impact on end-to-end compilation time is a 1.1x slowdown.

Modern programming languages support concurrent programming based on channels and processes. Channels enable synchronous and asynchronous message-passing between independent light-weight processes making it easy to express common concurrency patterns. The implementation of channels and processes in compilers and language runtimes is a difficult task that relies heavily on traditional and error-prone low-level concurrency primitives, raising concerns about correctness and reliability. In this paper, we present an automatic program generation technique to test such programming language implementations. We define a type and effect system for programs that communicate over channels and where every execution is guaranteed to eventually terminate. We can generate and run such programs, and if a program fails to terminate, we have found a bug in the programming language implementation. We implement such an automatic program generator and apply it to Go, Kotlin, Crystal, and Flix. We find two new bugs in Flix, and reproduce two bugs; one in Crystal and one in Kotlin.

The Java Virtual Machine (JVM) offers an attractive runtime environment for programming language implementers. The JVM has a simple bytecode format, excellent performance, multiple state-of-the art garbage collectors, robust backwards compatibility, and it runs on almost all platforms. Further, the Java ecosystem grants access to a plethora of libraries and tooling, including debuggers and profilers.

Program development tools such as bug finders, build automation tools, compilers, debuggers, integrated development environments, and refactoring tools increasingly rely on static analysis techniques to reason about program behavior. Implementing such static analysis tools is a complex and difficult task with concerns about safety and soundness. Safety guarantees that the fixed point computation -- inherent in most static analyses -- converges and ultimately terminates with a deterministic result. Soundness guarantees that the computed result over-approximates the concrete behavior of the program under analysis. But how do we know if we can trust the result of the static analysis itself? Who will guard the guards?

Implicit parameters allow programmers to omit certain arguments from function calls and have them automatically inferred by the compiler based on their types. At every call site, the compiler determines the values of the implicit parameters based on their declared types and the bindings currently in implicit scope. The programmer controls this mechanism in two ways: by adding bindings to the implicit scope, or by explicitly providing the implicit parameters for the function call.

Recently, promises were added to ECMAScript 6, the JavaScript standard, in order to provide better support for the asynchrony that arises in user interfaces, network communication, and non-blocking I/O. Using promises, programmers can avoid common pitfalls of event-driven programming such as event races and the deeply nested counterintuitive control ow referred to as “callback hell”. Unfortunately, promises have complex semantics and the intricate control– and data- ow present in promise-based code hinders program comprehension and can easily lead to bugs. The promise graph was proposed as a graphical aid for understanding and debugging promise-based code. However, it did not cover all promise-related features in ECMAScript 6, and did not present or evaluate any technique for constructing the promise graphs. In this paper, we extend the notion of promise graphs to include all promise-related features in ECMAScript 6, including default reactions, exceptions, and the synchronization operations race and all. Furthermore, we report on the construction and evaluation of PromiseKeeper, which performs a dynamic analysis to create promise graphs and infer common promise anti-patterns. We evaluate PromiseKeeper by applying it to 12 open source promise-based Node.js applications. Our results suggest that the promise graphs constructed by PromiseKeeper can provide developers with valuable information about occurrences of common anti-patterns in their promise-based code, and that promise graphs can be constructed with acceptable run-time overhead.

In JavaScript programs, asynchrony arises in situations such as web-based user-interfaces, communicating with servers through HTTP requests, and non-blocking I/O. Event-based programming is the most popular approach for managing asynchrony, but suffers from problems such as lost events and event races, and results in code that is hard to understand and debug. Recently, ECMAScript 6 has added support for promises, an alternative mechanism for managing asynchrony that enables programmers to chain asynchronous computations while supporting proper error handling. However, promises are complex and error-prone in their own right, so programmers would benefit from techniques that can reason about the correctness of promise-based code. Since the ECMAScript 6 specification is informal and intended for implementers of JavaScript engines, it does not provide a suitable basis for formal reasoning. This paper presents λ p , a core calculus that captures the essence of ECMAScript 6 promises. Based on λ p , we introduce the promise graph, a program representation that can assist programmers with debugging of promise-based code. We then report on a case study in which we investigate how the promise graph can be helpful for debugging errors related to promises in code fragments posted to the StackOverflow website.

Many bugs in JavaScript applications manifest themselves as objects that have incorrect property values when a failure occurs. For this type of error, stack traces and log files are often insufficient for diagnosing problems. In such cases, it is helpful for developers to know the control flow path from the creation of an object to a crashing statement. Such crash paths are useful for understanding where the object originated and whether any properties of the object were corrupted since its creation.

We present Flix, a declarative programming language for specifying and solving least fixed point problems, particularly static program analyses. Flix is inspired by Datalog and extends it with lattices and monotone functions. Using Flix, implementors of static analyses can express a broader range of analyses than is currently possible in pure Datalog, while retaining its familiar rule-based syntax. We define a model-theoretic semantics of Flix as a natural extension of the Datalog semantics. This semantics captures the declarative meaning of Flix programs without imposing any specific evaluation strategy. An efficient strategy is semi-naive evaluation which we adapt for Flix. We have implemented a compiler and runtime for Flix, and used it to express several well-known static analyses, including the IFDS and IDE algorithms. The declarative nature of Flix clearly exposes the similarity between these two algorithms.

Many JavaScript programs are written in an event-driven style. In particular, in server-side Node.js applications, operations involving sockets, streams, and files are typically performed in an asynchronous manner, where the execution of listeners is triggered by events. Several types of programming errors are specific to such event-based programs (e.g., unhandled events, and listeners that are registered too late). We present the event-based call graph, a program representation that can be used to detect bugs related to event handling. We have designed and implemented three analyses for constructing event-based call graphs. Our results show that these analyses are capable of detecting problems reported on StackOverflow. Moreover, we show that the number of false positives reported by the analysis on a suite of small Node.js applications is manageable.

In JavaScript, and scripting languages in general, dynamic field access is a commonly used feature. Unfortunately, current static analysis tools either completely ignore dynamic field access or use overly conservative approximations that lead to poor precision and scalability. We present new string domains to reason about dynamic field access in a static analysis tool. A key feature of the domains is that the equal, concatenate and join operations take $\mathcal{O}$ (1) time. Experimental evaluation on four common JavaScript libraries, including jQuery and Prototype, shows that traditional string domains are insufficient. For instance, the commonly used constant string domain can only ensure that at most 21% dynamic field accesses are without false positives. In contrast, our string domain $\mathcal{H}$ ensures no false positives for up to 90% of all dynamic field accesses. We demonstrate that a dataflow analysis equipped with the $\mathcal{H}$ domain gains significant precision resulting in an analysis speedup of more than 1.5x for 7 out of 10 benchmark programs.

JavaScript is a language that is widely-used for both web- based and standalone applications such as those in the upcoming Windows 8 operating system. Analysis of JavaScript has long been known to be challenging due to its dynamic nature. On top of that, most JavaScript applications rely on large and complex libraries and frameworks, often written in a combination of JavaScript and native code such as C and C++. Stubs have been commonly employed as a partial specification mechanism to address the library problem; however, they are tedious to write, incomplete, and occasionally incorrect.

Developers of JavaScript web applications have little tool support for catching errors early in development. In comparison, an abundance of tools exist for statically typed languages, including sophisticated integrated development environments and specialized static analyses. Transferring such technologies to the domain of JavaScript web applications is challenging. In this paper, we discuss the challenges, which include the dynamic aspects of JavaScript and the complex interactions between JavaScript, HTML, and the browser. From this, we present the first static analysis that is capable of reasoning about the flow of control and data in modern JavaScript applications that interact with the HTML DOM and browser API.