Abstract. Two general techniques for implementing a domain-specific language (DSL) with less overhead are the finally-tagless embedding of object programs and the direct-style representation of side effects. We use these techniques to build a DSL for probabilistic programming, for expressing countable probabilistic models and performing exact inference and importance sampling on them. Our language is embedded as an ordinary OCaml library and represents probability distributions as ordinary OCaml programs. We use delimited continuations to reify probabilistic programs as lazy search trees, which inference algorithms may traverse without imposing any interpretive overhead on deterministic parts of a model. We thus take advantage of the existing OCaml implementation to achieve competitive performance and ease of use. Inference algorithms can easily be embedded in probabilistic programs themselves.

High level data structures are a cornerstone of modern programming and at the same time stand in the way of compiler optimizations. In order to reason about user or library-defined data structures, compilers need to be extensible. Common mechanisms to extend compilers fall into two categories. Frontend macros, staging or partial evaluation systems can be used to programmatically remove abstraction and specialize programs before they enter the compiler. Alternatively, some compilers allow extending the internal workings by adding new transformation passes at different points in the compile chain or adding new intermediate representation (IR) types. None of these mechanisms alone is sufficient to handle the challenges posed by high level data structures. This paper shows a novel way to combine them to yield benefits that are greater than the sum of the parts.

Tony Hoare has always been a leader in writing down and proving properties of programs. To prove properties of programs automatically, the most widely used technology today is by far the ubiquitous type checker. Alas, static type systems inevitably exclude some good programs and allow some bad ones. Thus motivated, we describe some fun we have been having with Haskell, by making the type system more expressive without losing the benefits of automatic proof and compact expression. Specifically, we offer a programmer’s tour of so-called type families, a recent extension to Haskell that allows functions on types to be expressed as straightforwardly as functions on values. This facility makes it easier for programmers to effectively extend the compiler by writing functional programs that execute during type-checking. This paper gives a programmer’s tour of type families as they are supported in GHC. Source code for all the examples is available at

Abstract. Feldspar is a domain specific language, embedded in Haskell, for programming digital signal processing algorithms. The final aim of a Feldspar program is to generate low level code with good performance. Still, we chose to provide the user with a purely functional DSL. The language is implemented as a minimal, deeply embedded core language, with shallow extensions built upon it. This paper presents full details of the essential parts of the implementation. Our initial conclusion is that this approach works well in our domain, although much work remains. 1

Abstract. This paper presents a new solution to the expression problem (EP) that works in OO languages with simple generics (including Java or C#). A key novelty of this solution is that advanced typing features, including F-bounded quantification, wildcards and variance annotations, are not needed. The solution is based on object algebras, which are an abstraction closely related to algebraic datatypes and Church encodings. Object algebras also have much in common with the traditional forms of the Visitor pattern, but without many of its drawbacks: they are extensible, remove the need for accept methods, and do not compromise encapsulation. We show applications of object algebras that go beyond toy examples usually presented in solutions for the expression problem. In the paper we develop an increasingly more complex set of features for a mini-imperative language, and we discuss a real-world application of object algebras in an implementation of remote batches. We believe that object algebras bring extensibility to the masses: object algebras work in mainstream OO languages, and they significantly reduce the conceptual overhead by using only features that are used by everyday programmers. 1

Abstract. Developing rich web applications requires mastering different environments on the client and server sides. While there is considerable choice on the server-side, the client-side is tied to JavaScript, which poses substantial software engineering challenges, such as moving or sharing pieces of code between the environments. We embed JavaScript as a DSL in Scala, using Lightweight Modular Staging. DSL code can be compiled to JavaScript or executed as part of the server application. We use features of the host language to make client-side programming safer and more convenient. We use gradual typing to interface typed DSL programs with existing JavaScript APIs. We exploit a selective CPS transform already available in the host language to provide a compelling abstraction over asynchronous callback-driven programming in our DSL.

Partial evaluation aims to improve the efficiency of a program by specialising it with respect to some known inputs. In theory, it is a natural match to language implementation, in that partially evaluating an interpreter with respect to a specific source program yields an efficient translation of that program. In practice, however, there can be difficulties — we must consider e.g. binding-time improvements, function calls, recursion, code duplication, and how to deal with side-effects. These difficulties limit the practical benefits of partial evaluation and have limited its widespread adoption. In this paper, we show that partial evaluation can be an effective and, unusually, straightforward technique for the efficient implementation of domain-specific languages. We achieve this by exploiting dependent types and by following some simple rules in the definition of the interpreter. We present experimental evidence that partial evaluation of programs in domain-specific languages yields efficient residual programs whose performance is competitive with their Java and C equivalents and which are also, through the use of dependent types, verifiably resource-safe. Using our technique, it follows that a verifiably correct and resource-safe program can also be an efficient program. 1.

Haskell programmers have been experimenting with dependent types for at least a decade, using clever encodings that push the limits of the Haskell type system. However, the cleverness of these encodings is also their main drawback. Although the ideas are inspired by dependently typed programs, the code looks significantly different. As a result, GHC implementors have responded with extensions to Haskell’s type system, such as GADTs, type families, and datatype promotion. However, there remains a significant difference between programming in Haskell and in full-spectrum dependently typed languages. Haskell enforces a phase separation between runtime values and compile-time types. Therefore, singleton types are necessary to express the dependency between values and types. These singleton types introduce overhead and redundancy for the programmer. This paper presents the singletons library, which generates the boilerplate code necessary for dependently typed programming using GHC. To compare with full-spectrum languages, we present an extended example based on an Agda interface for safe database access. The paper concludes with a detailed discussion on the current capabilities of GHC for dependently typed programming and suggestions for future extensions to better support this style of programming.

Many loops in probabilistic inference map almost every individual in their domain to the same result. Running such loops symbolically takes time sublinear in the domain size. Using normalization by evaluation with first-class delimited continuations, we lift inference procedures to reap this speed-up without interpretive overhead. To express nested loops, we use multiple control delimiters for metacircular interpretation. To express loops over a powerset domain, we convert nested loops over a subset to unnested loops. 1.

One often cited benefit of pure functional programming is that pure code is easier to test and reason about, both formally and informally. However, real programs have side-effects including state management, exceptions and interactions with the outside world. Haskell solves this problem using monads to capture details of possibly side-effecting computations — it provides monads for capturing State, I/O, exceptions, non-determinism, libraries for practical purposes such as CGI and parsing, and many others, as well as monad transformers for combining multiple effects. Unfortunately, useful as monads are, they do not compose very well. Monad transformers can quickly become unwieldy when there are lots of effects to manage, leading to a temptation in larger programs to combine everything into one coarse-grained state and exception monad. In this paper I describe an alternative approach based on handling algebraic effects, implemented in the IDRIS programming language. I show how to describe side effecting computations, how to write programs which compose multiple fine-grained effects, and how, using dependent types, we can use this approach to reason about states in effectful programs. 1.