Benjamin S. Lerner

Searching for type-error messages

Advanced type systems often need some form of type inference to reduce the burden of explicit typing, but type inference often leads to poor error messages for ill-typed programs. This work pursues a new approach to constructing compilers and presenting type-error messages in which the type-checker itself does not produce the messages. Instead, it is an oracle for a search procedure that finds similar programs that do type-check. Our two-fold goal is to improve error messages while simplifying compiler construction. Our primary implementation and evaluation is for Caml, a language with full type inference. We also present a prototype for C++ template functions, where type instantiation is implicit. A key extension is making our approach robust even when the program has multiple independent type errors.

TeJaS: retrofitting type systems for JavaScript

JavaScript programs vary widely in functionality, complexity, and use, and analyses of these programs must accommodate such variations. Type-based analyses are typically the simplest such analyses, but due to the languages subtle idioms and many application-specific needs---such as ensuring general-purpose type correctness, security properties, or proper library usage---we have found that a single type system does not suffice for all purposes. However, these varied uses still share many reusable common elements. In this paper we present TeJaS, a framework for building type systems for JavaScript. TeJaS has been engineered modularly to encourage experimentation. Its initial type environment is reified, to admit easy modeling of the various execution contexts of JavaScript programs, and its type language and typing rules are extensible, to enable variations of the type system to be constructed easily. The paper presents the base TeJaS type system, which performs traditional type-checking for JavaScript. Because JavaScript demands complex types, we explain several design decisions to improve user ergonomics. We then describe TeJaSs modular structure, and illustrate it by reconstructing the essence of a very different type system for JavaScript. Systems built from TeJaS have been applied to several real-world, third-party JavaScript programs.

A tested semantics for getters, setters, and eval in JavaScript

We present S5, a semantics for the strict mode of the ECMAScript 5.1 (JavaScript) programming language. S5 shrinks the large source language into a manageable core through an implemented transformation. The resulting specification has been tested against real-world conformance suites for the language. This paper focuses on two aspects of S5: accessors (getters and setters) and eval. Since these features are complex and subtle in JavaScript, they warrant special study. Variations on both features are found in several other programming languages, so their study is likely to have broad applicability.

Verifying Web Browser Extensions’ Compliance with Private-Browsing Mode

Modern web browsers implement a private browsing mode that is intended to leave behind no traces of a user’s browsing activity on their computer. This feature is in direct tension with support for extensions, which can silently void this guarantee.

Seminal: searching for ML type-error messages

We present a new way to generate type-error messages in a polymorphic, implicitly, and strongly typed language (specifically Caml). Our method separates error-message generation from type-checking by taking a fundamentally new approach: we present to programmers small term-level modifications that cause an ill-typed program to become well-typed. This approach aims to improve feedback to programmers with no change to the underlying type-checker nor the compilation of well-typed programs.We have added a prototype implementation of our approach to the Objective Caml system by intercepting type-checker error messages and using the type-checker on candidate changes to see if they succeed. This novel front-end architecture naturally decomposes into (1) enumerating local changes to the abstract syntax tree that may remove type errors, (2) searching for places to try the changes, (3) using the type-checker to evaluate the changes, and (4) ranking the changes and presenting them to the user.

Supporting dynamic, third-party code customizations in JavaScript using aspects

Web sites and web browsers have recently evolved into platforms on top of which entire applications are delivered dynamically, mostly as JavaScript source code. This delivery format has sparked extremely enthusiastic efforts to customize both individual web sites and entire browsers in ways the original authors never expected or accommodated. Such customizations take the form of yet more script dynamically injected into the application, and the current idioms to do so exploit arcane JavaScript features and are extremely brittle. In this work, we accept the popularity of extensions and seek better linguistic mechanisms to support them. We suggest adding to JavaScript aspect-oriented features that allow straightforward and declarative ways for customization code to modify the targeted application. Compared to most prior aspect-related research, our work has a different motivation and a different target programming environment, both of which lead to novel design and implementation techniques. Our aspect weaving is entirely integrated into a new dynamic JIT compiler, which lets us properly handle advice to first-class functions in the presence of arbitrary aliasing, without resorting to whole-program code transformations. Our prototype demonstrates that an aspect-oriented approach to web-application customization is often more efficient than current idioms while simplifying the entire process.

Language support for extensible web browsers

Web browsers are sophisticated and crucial programs, and millions of users extend their browsers to customize their browsing experience. In this paper we argue the position that such extensions themselves constitute an important facet of web applications---one in need of serious programming-language research attention. We illustrate this position by contrasting the extension mechanisms of the two predominant extensible browsers, Mozilla Firefox and Google Chrome, and highlighting their weaknesses. We then describe very preliminary work that addresses these shortcomings.

Synthesis of Recursive ADT Transformations from Reusable Templates

Recent work has proposed a promising approach to improving scalability of program synthesis by allowing the user to supply a syntactic template that constrains the space of potential programs. Unfortunately, creating templates often requires nontrivial effort from the user, which impedes the usability of the synthesizer. We present a solution to this problem in the context of recursive transformations on algebraic data-types. Our approach relies on polymorphic synthesis constructs: a small but powerful extension to the language of syntactic templates, which makes it possible to define a program space in a concise and highly reusable manner, while at the same time retains the scalability benefits of conventional templates. This approach enables end-users to reuse predefined templates from a library for a wide variety of problems with little effort. The paper also describes a novel optimization that further improves the performance and the scalability of the system. We evaluated the approach on a set of benchmarks that most notably includes desugaring functions for lambda calculus, which force the synthesizer to discover Church encodings for pairs and boolean operations.

From Spreadsheets to Programs: Data Science and CS1 in Pyret (Abstract Only)

Data Science is at the center of many current curricular efforts. It is emerging as an integrated field that has far-reaching and important applications, from news media to policy making to business. While these applications can provide compelling uses of computer science techniques, an introduction to one is not an introduction to the other. How do topics like data structures and program design emerge from data science applications? How do we transition from data science applications to computer science topics? How can data science be integrated into other contexts with little overhead? This workshop presents assignments and curricula designed to answer these questions, and tools that support them.

Detecting conflicts among declarative UI extensions

We examine overlays, a flexible aspect-like mechanism for third-party declarative extensions of declarative UIs. Overlays can be defined for any markup language and permit extensions to define new content that is dynamically woven into a base UI document. While powerful, overlays are inherently non-modular and may conflict with each other, by defining duplicate or contradictory UI components. We construct an abstract language to capture core overlay semantics, and design an automatic analysis to detect inter-extension conflicts. We apply the analysis to a case study of Firefox extensions, finding several real-world bugs. Our analysis provides low-level feedback to extension developers and high-level reports to end users. Finally, we show how variants of overlays more expressive than those of Firefox complicate conflict detection.