These abstracts are for talks at this event.
NEPLS is a venue for ongoing research, so the abstract and supplemental material associated with each talk is necessarily temporal. The work presented here may be in a state of flux. In all cases, please consult the authors' Web pages for up-to-date information. Please don't refer to these pages as a definitive source.
Dissecting Transactional Implementations for Fun and Profit
There has been much recent interest in exploring the role of software transactions as an alternative concurrency control mechanism to mutual exclusion locks. While transactions can provide numerous benefits with respect to programmability, safety, and performance, their implementations are necessarily complex. This complexity arises because of the need to dynamically monitor execution within a transaction for consistency violations, and to remedy these violations once they are detected. However, by carefully selecting different features of these implementations, we can provide efficient solutions to a number of more constrained, but equally challenging, problems. In this talk, we present three such examples drawn from three very different domains: (1) safe futures for Java; (2) preemptible atomic regions to deal with priority inversion in realtime systems; and (3) a lightweight checkpointing facility for concurrent (mostly) functional programs.
jpaul - Java Program Analysis Utilities Library
jpaul is a Java implementation of several algorithms that are widely used in program analysis. Compiler writers often reimplement a (relatively small) set of algorithms: graph algorithms, constraint solvers, dataflow engines etc. The goal of the jpaul project is to have these algorithms implemented independently of any compiler. jpaul emphasizes (in order): correctness, flexibility, ease-of-use, and asymptotic complexity (we are not interested in cutting-edge low level optimizations). The intended audience consists of program analysis researchers in need of a correct, elegant, and fast way of prototyping their analysis ideas. jpaul is implemented in Generic Java (JDK 1.5) and is released under the terms of the (Modified) BSD Licence.
For downloads and more information, please visit the jpaul website on Sourceforge: http://jpaul.sourceforge.net/.
Taking Derivatives of Functional Programs: AD in a Functional Framework
Algorithm Differentiation (AD) is an established enterprise that seeks to take the derivatives of functions specified as programs through symbolic manipulation rather than finite differencing. AD has traditionally been applied to imperative programs, typically using preprocessors that apply a source-to-source transformation to derive a program for computing the derivative of a function from a program that computes the original function. We present a novel framework for performing AD within modern functional-programming languages, treating AD operators as first-class higher-order functions that map first-class function objects to first-class function objects. Our approach is more modular, allowing a library of functions based on derivatives to be built compositionally. It allows the callee to specify the necessary AD, rather than insisting that the caller provide appropriately transformed functions. Our approach is foundational: we develop a method for taking the derivative of any lambda-calculus expression over a basis of standard numeric primitives. Since the lambda calculus is both the source and target of our transformations, higher-order derivatives can be constructed naturally, without special mechanisms. In this talk, I will give a tutorial overview of AD, demonstrate the power and convenience of the functional formulation of AD by way of a series of small examples, and sketch the framework underlying our implementation. Joint work with Barak A. Pearlmutter.
Formal Semantics of Weak References
Weak references are references that do not prevent the object they point to from being garbage collected. Many realistic languages, including Java, SML/NJ, and Haskell to name a few, support weak references. However, there is no generally accepted formal semantics for weak references. Without such a formal semantics it becomes impossible to formally prove properties of such a language and the programs written in it. We give a formal semantics for a calculus called lambda-weak that includes weak references and is derived from Morrisett, Felleisen, and Harper's lambda-gc. The semantics is used to examine several issues involving weak references. We use the framework to formalize the semantics for the key/value weak references found in Haskell. Furthermore, we consider a type system for the language and show how to extend the earlier result that type inference can be used to collect reachable garbage. In addition we show how to allow collection of weakly referenced garbage without incurring the computational overhead often associated with collecting a weak reference which may be later used. Lastly, we address the non-determinism of the semantics by providing both an effectively decidable syntactic restriction and a more general semantic criterion, which guarantee a unique result of evaluation. (joint work with Kevin Donnelly and Assaf Kfoury)
DieHard: Statistical Memory Safety for Unsafe Languages
The vast majority of existing applications are written in unsafe languages like C and C++. These applications are vulnerable to memory errors such as buffer overflows, dangling pointers, and reads to uninitialized data. Such errors can lead to program crashes, security vulnerabilities, or unpredictable behavior. We present DieHard, a system that combats these errors by providing *statistical memory safety* for unsafe languages. Through randomization and replication, DieHard achieves memory safety with high probability. It thus allows programs to continue to execute correctly in the face of memory errors, *masking* errors like buffer overflows and dangling pointers while ensuring safety by detecting uninitialized reads. We show analytically that DieHard increases memory safety exponentially with the number of replicas. Because it is embarrassingly parallel by design, DieHard can leverage multiple CPUs or cores to improve program reliability. (joint work with Ben Zorn of Microsoft)
The Architecture of a World-Wide Distributed Repository to Support
Fine-Grained Sharing of Source Code
October is a new infrastructure designed to support world-wide fined-grained source-code sharing at the level of individual functions. It is organized like the Web. Just as the Web is a world-wide distributed hyperlinked indexed repository of text documents, October is a world-wide distributed hyperlinked indexed repository of function definitions. Instead of Web pages there are top-level definitions. Top-level definitions are stored in a distributed repository implemented by October servers that are analogous to Web servers. They are viewed and manipulated by October browsers that are analogous to Web browsers. October servers and browsers communicate via the October protocol which plays the role of HTTP and allows different server and browser implementations to interoperate. The October protocol is designed from the start with support for searching, caching, mirroring, and load balancing. These play the role of search engines, proxy servers, and Web caches. Top-level definitions are stored as abstract syntax trees (ASTs) which play the role of HTML. Instead of URLs there are version identifiers. Top-level definitions are hyperlinked via embedded version identifiers. When building an executable, the browser crawls the repository to fetch all top-level definitions referenced directly or indirectly by the top-level definition being built. Just as the Web supports different document styles via DTDs, October supports different programming languages and programmer preferences via programming language descriptions (PLDs). PLDs are currently written to support Scheme, C, C++, and Java. In this talk, I will describe the design goals of October, discuss how these goals motivate and influence the creation of a new infrastructure, present some of the technical problems and design tradeoffs addressed so far while creating this infrastructure, and give a live demo of the prototype implementation of October.
MetaOCaml Server Pages: Web Publishing as Staged Computation
Modern dynamic web services are really computer programs. Some parts of these programs run off-line, others run server-side on each request, and still others run within the browser. In other words, web publishing is staged computation, either for better performance, or because certain resources are available in one stage but not another. Unfortunately, the various web programming languages make it difficult to spread computation over more than one stage. This is a tremendous opportunity for multi-stage languages in general, and for MetaOCaml in particular. We present the design of MetaOCaml Server Pages. Unlike other languages in its genre, the embedded MetaOCaml code blocks may be annotated with staging information, so that the programmer may safely and precisely control which computation occurs in which stage. A prototype web server, written in OCaml, supports web sites with both static and dynamic content. We provide several sample programs and demonstrate the performance gains won using multi-stage programming.
HasSound: Generating Musical Instrument Sounds in Haskell
To complement Haskore, a Haskell library for composing music at the "note" level, we have designed a Haskell library called "HasSound" for designing instrument sounds at the level of oscillators, filters, etc. It is based on csound, a popular music software system written in C. Although csound has a rather baroque syntax, it is "mostly" functional, which of course is reflected well in Haskell. But it also has some rather ugly imperative edges that provide important functionality that can't be ignored. We discuss how we handle both the functional and imperative sides to this design, including a monad for channels, functional delay lines, and an explicit way to express recursive signals. The work is a useful example of embedded DSL design. Joint work with Matt Zamec (Yale) and Sarah Eisenstat (Brown).
|Last modified Wednesday, October 12th, 2005 10:38:14pm||Powered by|