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.
Programming with Rectangles, Triangles, and Circles
We will argue that by properly generalizing the type system and expression syntax, it is possible for any modern object oriented language to provide first class support for manipulating both relational and hierarchical data in a sound and statically typed manner. The type system extensions, however, are not based on XML Schemas. We show that XSDs and the XML data model do not fit well with the class-based nominal type system and object graph representation of our target languages. Instead we propose to extend object-oriented type system with new structural types that model XSD sequences, choices, and all-groups. We demonstrate our language and type system by translating a selection of the XQuery use cases.
Controlling Gemini capsules: classical logic, intersection types
and strong normalization
Curien and Herbelin have defined a typed term calculus supporting a Curry-Howard interpretation for sequent proofs in classical logic. The untyped calculus underlying that of Curien and Herbelin can naturally be considered as the foundation of a functional programming language with explicit representation of control. We investigate some fundamental properties of the reduction relation for this calculus (which we call Gemini). In particular we characterize the strongly normalizing computations in Gemini by defining a system of intersection types, itself presented in sequent-style. As a corollary we deduce strong normalization for the Curien-Herbelin calculus. The proof that all strongly normalizing terms are typable relies on a definition we give of a perpetual strategy for reduction. The proof that all typable terms are strongly normalizing uses a variation on the traditional reducibility method which seems particularly suited to terms associated with sequent calculi, and which may be a useful tool in related settings. Joint work with: Silvia Ghilezan, University of Novi Sad Pierre Lescanne, ENS Lyon
A Typed Approach to Multiple Inheritance
Multiple Inheritance (MI) in Object-Oriented Programming (OOP) can be of great use in software practice.B However, current models for statically typed OOP often either completely avoid addressing MI or address it through the use of some greatly involved mechanisms that seem neither natural nor amendable to further extensions.B In this talk, we present a typed approach to modeling OOP that takes a view of ``objects as functions'', in contrast with the popular view of ``objects as records''. We show that this approach can not only address various common OOP features but also provide a simple and general account for MI.B The object encoding in our OOP model is primarily rooted in Applied Type System (ATS), a recently proposed framework for facilitating the design and formalization of advanced type systems. Joint work with Hongwei Xi.
Engineering Calling Conventions in the Quick C-- Compiler Infrastructure
Many compilers have been written that translate high-level languages to C. This technique eases portability, but C makes it difficult to provide efficient implementations of some features of modern languages, e.g., proper tail calls, accurate garbage collection, and efficient exception dispatch. C-- is designed as a better alternative, one that should be as easy to use as C while providing performance approaching that of custom code generators. C-- requires a balancing act: To be easy to use, it must abstract away from machine details, but to make high performance possible, it must expose such details. Nowhere is this balancing act more delicate than in the treatment of procedure calls. This talk provides a brief overview of the C-- design, emphasizing the mechanisms C-- uses to support procedure calls. It then explains how procedure calls are implemented in the Quick C-- compiler and shows how this implementation can be used to achieve two goals: Tuning the performance of calls and interoperating with calls in other languages. (Joint work with Simon Peyton Jones, Christian Lindig, Joao Dias, and Reuben Olinsky)
Once a program run exposes a bug, the programmers' first task is fault localization: identifying the portions of code they need to change to correct the bug. In this talk, I will present a pattern common among state-of-the-art fault-localization tools. Tools based on this pattern represent program runs as sets of events, and contrast the representations of successful and failing program runs to provide a hint as to the location of the bug. Isolated events rarely suffice, though, because their impact on the outcome of a run depends heavily on the context in which they execute. Existing research addresses this context problem by employing program-specific knowledge. I will present a novel enhancement of the pattern that addresses the context problem in a program-independent way. The technique employs a distance metric between program runs and, given a failing run, selects its most similar successful run to contrast with it. To quantify the success of the technique, I developed a metric for the quality of fault-localization tools. This is the first metric I know of, and it allows direct comparison between different techniques. I will use this metric to evaluate two implementations of my technique.
Mark-Copy: Fast copying GC with less space overhead
Copying garbage collectors have a number of advantages over non-copying collectors, including cheap allocation and avoiding fragmentation. However, in order to provide completeness (the guarantee to reclaim each garbage object eventually), standard copying collectors require space equal to twice the size of the maximum live data for a program. We present a mark-copy collection algorithm (MC) that extends generational copying collection and significantly reduces the heap space required to run a program. MC reduces space overhead by 75-85% compared with standard copying garbage collectors, increasing the range of applications that can use copying garbage collection. We show that when MC is given the same amount of space as a generational copying collector, it improves total execution time of Java benchmarks significantly in tight heaps, and by 5-10% in moderate size heaps. We also compare the performance of MC with a (non-generational) mark-sweep collector and a hybrid copying/mark-sweep generational collector. We find that MC can run in heaps comparable in size to the minimum heap space required by mark-sweep. We also find that for most benchmarks MC is significantly faster than mark-sweep in small and moderate size heaps. When compared with the hybrid collector, MC improves total execution time by about 5% for some benchmarks, partly by increasing the speed of execution of the application code. Joint work with J. Eliot B. Moss.
Object Equality Profiling
This talk presents Object Equality Profiling (OEP), a new technique for helping programmers discover optimization opportunities in programs. OEP discovers opportunities for replacing a set of equivalent object instances with a single representative object. Such a set represents an opportunity for automatically or manually applying optimizations such as hash consing, heap compression, lazy allocation, object caching, invariant hoisting, and more. To evaluate OEP, we implemented a tool to help programmers reduce the memory usage of Java programs. Our tool performs a dynamic analysis that records all the objects created during a particular program run. The tool partitions the objects into equivalence classes, and uses collected timing information to determine when elements of an equivalence class could have been safely collapsed into a single representative object without affecting the behavior of that program run. We report the results of applying this tool to benchmarks, including two widely used Web application servers. Many benchmarks exhibit significant amounts of object equivalence, and in most benchmarks our profiler identifies optimization opportunities clustered around a small number of allocation sites. We present a case study of using our profiler to find simple manual optimizations that reduce the average space used by live objects in two SpecJVM benchmarks by 47% and 38% respectively. This is a joint work with Robert O'Callahan from the IBM T. J. Watson Research Center.
The HOW of Programming Language Research -- Starting Point for a Discussion
At many past NEPLS meetings we have heard and talked about programming language research RESULTS. I maintain that an equally valid subject for these meetings is HOW we do our research -- the tools and techniques that we use to get our work done. I propose to initiate a discussion of this topic by describing some tools that I have developed, and am developing, for our work on language inter-operation. I hope that this will stimulate other people to talk about tools that they use, as such a discussion should be mutually beneficial. At the very least, if there are wonderful tools out there that I have missed I would like to know about them. My presentation will describe two sets of tools that I am developing. One set takes an S-expr-based representation of a set of syntax, type, and semantic rules as input, and produces LaTeX representations of the rules. It also produces a corresponding set of Prolog inference rules for type-checking and another set of Prolog inference rules for execution of a program written in the language. The other set of tools help to verify proofs about the system of rules by taking a similar representation of the proof, verifying (most of) the steps of the proof and generating a LaTeX version of it.
|Last modified Wednesday, October 15th, 2003 12:28:52pm||Powered by|