Java and Scala's Type Systems are Unsound: The Existential Crisis of Null Pointers

We present short programs that demonstrate the unsoundness of Java and Scala’s current type systems. In particular, these programs provide parametrically polymorphic functions that can turn any type into any type without (down)casting. Fortunately, parametric polymorphism was not integrated into the Java Virtual Machine (JVM), so these examples do not demonstrate any unsoundness of the JVM. Nonetheless, we discuss broader implications of these findings on the field of programming languages.

Java and Scala's Type Systems are Unsound The Existential Crisis of Null Pointers

Nada Amin

We present short programs that demonstrate the unsoundness of Java and Scala's current type systems. In particular, these programs provide parametrically polymorphic functions that can turn any type into any type without (down) casting. Fortunately, parametric polymorphism was not integrated into the Java Virtual Machine (JVM), so these examples do not demonstrate any unsoundness of the JVM. Nonetheless, we discuss broader implications of these findings on the field of programming languages.

Assoc Computing Machinery2016

Miniboxing: Improving the Speed to Code Size Tradeoff in Parametric Polymorphism Translations

Martin Odersky, Cristian Talau, Vlad Ureche

Parametric polymorphism enables code reuse and type safety. Underneath the uniform interface exposed to programmers, however, its low level implementation has to cope with inherently non-uniform data: value types of different sizes and semantics (bytes, integers, floating point numbers) and reference types (pointers to heap objects). On the Java Virtual Machine, parametric polymorphism is currently translated to bytecode using two competing approaches: homogeneous and heterogeneous. Homogeneous translation requires boxing, and thus introduces indirect access delays. Heterogeneous translation duplicates and adapts code for each value type individually, producing more bytecode. Therefore bytecode speed and size are at odds with each other. This paper proposes a novel translation that significantly reduces the bytecode size without affecting the execution speed. The key insight is that larger value types (such as integers) can hold smaller ones (such as bytes) thus reducing the duplication necessary in heterogeneous translations. In our implementation, on the Scala compiler, we encode all primitive value types in long integers. The resulting bytecode approaches the performance of monomorphic code, matches the performance of the heterogeneous translation and obtains speedups of up to 22x over the homogeneous translation, all with modest increases in size.

2013

Higher-Order Subtyping with Type Intervals

Sandro Stucki

Modern, statically typed programming languages provide various abstraction facilities at both the term- and type-level. Common abstraction mechanisms for types include parametric polymorphism -- a hallmark of functional languages -- and subtyping -- which is pervasive in object-oriented languages. Additionally, both kinds of languages may allow parametrized (or generic) datatype definitions in modules or classes. When several of these features are present in the same language, new and more expressive combinations arise, such as (1) bounded quantification, (2) bounded operator abstractions and (3) translucent type definitions. An example of such a language is Scala, which features all three of the aforementioned type-level constructs. This increases the expressivity of the language, but also the complexity of its type system. From a theoretical point of view, the various abstraction mechanisms have been studied through different extensions of Girard's higher-order polymorphic lambda-calculus F-omega. Higher-order subtyping and bounded polymorphism (1 and 2) have been formalized in F-omega-sub and its many variants; type definitions of various degrees of opacity (3) have been formalized through extensions of F-omega with singleton types. In this dissertation, I propose type intervals as a unifying concept for expressing (1--3) and other related constructs. In particular, I develop an extension of F-omega with interval kinds as a formal theory of higher-order subtyping with type intervals, and show how the familiar concepts of higher-order bounded quantification, bounded operator abstraction and singleton kinds can all be encoded in a semantics-preserving way using interval kinds. Going beyond the status quo, the theory is expressive enough to also cover less familiar constructs, such as lower-bounded operator abstractions and first-class, higher-order inequality constraints. I establish basic metatheoretic properties of the theory: I prove that subject reduction holds for well-kinded types w.r.t. full beta-reduction, that types and kinds are weakly normalizing, and that the theory is type safe w.r.t. its call-by-value operational reduction semantics. Key to this metatheoretic development is the use of hereditary substitution and the definition of an equivalent, canonical presentation of subtyping, which involves only normal types and kinds. The resulting metatheory is entirely syntactic, i.e. does not involve any model constructions, and has been fully mechanized in Agda. The extension of F-omega with interval kinds constitutes a stepping stone to the development of a higher-order version of the calculus of Dependent Object Types (DOT) -- the theoretical foundation of Scala's type system. In the last part of this dissertation, I briefly sketch a possible extension of the theory toward this goal and discuss some of the challenges involved in adapting the existing metatheory to that extension.

EPFL2017

Higher-Order Subtyping with Type Intervals

Sandro Stucki

EPFL2017