Practical Aspects of Declarative Languages: First International Workshop, PADL’99 San Antonio, Texas, USA, January 18–19, 1999 Proceedings

375 Pages·1998·3.44 MB·English

by  Graeme E. Moss

Lecture Notes in Computer Science 1551
Editedby G.Goos,J. Hartmanisand J.van Leeuwen
3
Berlin
Heidelberg
NewYork
Barcelona
HongKong
London
Milan
Paris
Singapore
Tokyo
Gopal Gupta (Ed.)
Practical Aspects of
Declarative Languages
First International Workshop, PADL’99
San Antonio, Texas, USA, January 18-19, 1999
Proceedings
1 3
SeriesEditors
GerhardGoos,KarlsruheUniversity,Germany
JurisHartmanis,CornellUniversity,NY,USA
JanvanLeeuwen,UtrechtUniversity,TheNetherlands
VolumeEditor
GopalGupta
NewMexicoStateUniversity,DepartmentofComputerScience
NewScienceHall(Box30001,MSCS)
StewartStreet,LasCruces,NM88003,USA
E-mail:[email protected]
Cataloging-in-Publicationdataappliedfor
DieDeutscheBibliothek-CIP-Einheitsaufnahme
Practicalaspectsofdeclarativelanguages:firstinternationalworkshop;
proceedings/PADL’99,SanAntonio,Texas,USA,January18-19,1999.Gopal
Gupta(ed.).-Berlin;Heidelberg;NewYork;Barcelona;HongKong;London
;Milan;Paris;Singapore;Tokyo:Springer,1998
(Lecturenotesincomputerscience;1551)
ISBN3-540-65527-1
CRSubjectClassification(1998):D.3,D.1,F.3,D.2,I.2.3
ISSN0302-9743
ISBN3-540-65527-1Springer-VerlagBerlinHeidelbergNewYork
Thisworkissubjecttocopyright.Allrightsarereserved,whetherthewholeorpartofthematerialis
concerned,specificallytherightsoftranslation,reprinting,re-useofillustrations,recitation,broadcasting,
reproductiononmicrofilmsorinanyotherway,andstorageindatabanks.Duplicationofthispublication
orpartsthereofispermittedonlyundertheprovisionsoftheGermanCopyrightLawofSeptember9,1965,
initscurrentversion,andpermissionforusemustalwaysbeobtainedfromSpringer-Verlag.Violationsare
liableforprosecutionundertheGermanCopyrightLaw.
(cid:1)c Springer-VerlagBerlinHeidelberg1999
PrintedinGermany
Typesetting:Camera-readybyauthor
SPIN10693172 06/3142–543210 Printedonacid-freepaper
Preface
Declarativelanguageshavetraditionallybeenregardedbythe mainstreamcom-
puting community as too impractical to be put to practical use. At the same
time,traditionalconferencesdevotedtodeclarativelanguagesdonothaveissues
related to practice as their centralfocus. Thus, there are few forums devoted to
discussion of practical aspects and implications of newly discovered results and
techniques related to declarative languages. The goal of the First International
Workshop on Practical Aspects of Declarative Languages (PADL) is to bring
togetherresearchers,practitionersandimplementorsofdeclarativelanguagesto
discuss practical issues and practical implications of their research results. The
workshop was held in San Antonio, Texas, during January 18-19, 1999. This
volume contains its proceedings.
Fifty three papers were submitted in response to the call for papers. These
papers were written by authors belonging to twenty one countries from six con-
tinents. Each paper was assigned to at least two referees for reviewing. Twenty
four papers were (cid:12)nally selected for presentation at the workshop. Many good
papers could not be included due to the limited duration of the workshop. The
workshopincludedinvitedtalksbyMarkHaydenofDEC/CompaqSystemsRe-
searchCenter, speaking on \Experiences Building Distributed Systems in ML,"
and Mark Wallace of Imperial College Center for Planning And Resource Con-
trol (IC-PARC), speaking on \ECLiPSe: Declarative Speci(cid:12)cation and Scalable
Implementation." The workshopalso included a panel discussion on\What can
declarative paradigms contribute to meet the challenges of Software Engineer-
ing?"
The workshop is sponsored and organized by COMPULOG AMERICAS
(http://www.cs.nmsu.edu/~complog),a networkof researchgroupsdedicated to
promoting research in logic programming and related areas, by the Association
for Logic Programming (http://www.cwi.nl/projects/alp),and the Department
of Computer Science, New Mexico State University (http://www.cs.nmsu.edu).
TheworkshopisbeingheldincooperationwiththeACMSpecialInterestGroup
onProgrammingLanguages(ACMSIGPLAN).Thesupportofmanyindividual
was crucial to the success of this workshop. My thanks to Enrico Pontelli for
his tremendous help in managing and organizing every aspect of the workshop.
He alsotookuponhimself the time-consumingtaskofdevelopingandmaintain-
ing the PADL’99 web-pages. Thanks are also due to Krzysztof Apt, President
of ALP, and David S. Warren, past President of ALP, for their support, help,
andencouragement,to AndrewAppel,POPLGeneralChair,andDonnaBaglio
and Maria Triver of ACM, for answering various organizational questions, and
to the referees for their prompt reviewing of papers. My thanks to the program
committee members for all their help in reviewing and their advice. Finally, my
thanks to all the authors who took interest in PADL’99 and submitted papers.
November 1998 Gopal Gupta
Program Committee
M. Carlsson SICS, Sweden
T. Chikayama University of Tokyo, Japan
B. Demoen KU Leuven, Belgium
H. Gill DARPA, USA
G. Gupta New Mexico State University, USA (chair)
M. Hermenegildo Technical University of Madrid, Spain
B. Jayaraman SUNY Bu(cid:11)alo, USA
N. Jones University of Copenhagen, Denmark
R. Kieburtz OregonGraduate Institute, USA
M. Martelli University of Genoa, Italy
S. Peyton Jones Microsoft Research, UK
E. Pontelli New Mexico State University, USA
I.V. Ramakrishnan SUNY Stonybrook, USA
D. Schmidt Kansas State University, USA
V. Santos Costa University of Porto, Portugal
P. van Hentenryck Brown University, USA
D.S. Warren SUNY Stonybrook, USA
P. Wadler Lucent Technologies Bell Labs, USA
Organizing Committee
E. Pontelli New Mexico State University, USA (chair)
G. Gupta New Mexico State University, USA
List of Referees
A. K.Bansal,C. Baoqiu,M. Bruynooghe,F. Bueno,M. Carlsson,M.Carro,M.
M. T. Chakravarty, H. Davulcu, B. Demoen, I. Dutra, M. Florido, G. Gupta,
M.Huth, J.Hatcli(cid:11),N.Jacobs,B.Jayaraman,N.D.Jones,S.PeytonJones,R.
B. Kieburtz, K. Narayan Kumar, J. P. Leal, S. W. Loke, R. Lopes, W. Luk , J.
Marino,R. F. P.Marques,M. Martelli,S.Mckeever,T.Mogensen,A. H. Nieva,
O. Olsson, E. Pontelli, G. Puebla, C. R. Ramakrishnan, P. Rao, R. Rocha, A.
Roychoudhury,V.SantosCosta,D.Schmidt, M.H.B.Sorensen,K.Sagonas,S.
Singh, A. Stoughton, P. Szeredi, A. Tiwari, P. van Hentenryck, P. Wadler, and
D. S. Warren.
Table of Contents
Software Engineering
Automated Benchmarking of Functional Data Structures.....................1
G.E. Moss and C. Runciman
np-spec: An Executable Specification Language for Solving
All Problems in NP ........................................................16
M. Cadoli, L. Palopoli, A. Schaerf, and D. Vasile
Prototyping a Requirements Specification through an Automatically
Generated Concurrent Logic Program......................................31
P. Letelier, P. Sa´nchez, and I. Ramos
Multi-agent Systems Development as a Software Engineering Enterprise.....46
M. Bozzano, G. Delzanno, M. Martelli, V. Mascardi, and F. Zini
Innovative Applications I
From Functional Animation to Sprite-Based Display........................61
C. Elliott
Beyond Pretty-Printing:Galley Concepts in Document
Formatting Combinators....................................................76
W. Kahl
Lambda in Motion: Controlling Robots with Haskell........................91
J. Peterson, P. Hudak, and C. Elliott
Implementation
CHAT: The Copy-Hybrid Approach to Tabling............................106
B. Demoen and K. Sagonas
The Influence of Parallel Computer Architectural Parameters on the
Performance of ParallelLogic Programming Systems......................122
M.G. Silva, I.C. Dutra, R. Bianchini, and V. Santos Costa
Or-Parallelismwithin Tabling.............................................137
R. Rocha, F. Silva, and V. Santos Costa
Declarative Query Processing
Mnesia – A Distributed Robust DBMS for Telecommunications
Applications...............................................................152
H. Mattsson, H. Nilsson, and C. Wikstr¨om
An AQUA-Based Intermediate Language for Evaluating an
Active Deductive Object-Oriented Language...............................164
B. Siddabathuni, S.W. Dietrich, and S.D. Urban
Implementing a Declarative String Query Language
with String Restructuring.................................................179
R. Hakli, M. Nyka¨nen, H. Tamm, and E. Ukkonen
VIII Table of Contents
Systems Applications
Client-Side Web Scripting with HaskellScript..............................196
E. Meijer, D. Leijen, and J. Hook
MCORBA: A CORBA Binding for Mercury...............................211
D. Jeffery, T. Dowd, and Z. Somogyi
Analysis
Dead Code Elimination through Dependent Types.........................228
H. Xi
Multiple Specialization of WAM Code.....................................243
M. Ferriera and L. Damas
A Flexible Framework for Dynamic and Static Slicing of
Logic Programs...........................................................259
W.W. Vasconcelos
Innovative Applications II
Applying Logic Programming to Derive Novel Functional
Information of Genomes...................................................275
A.K. Bansal and P. Bork
An Application of Action Theory to the Space Shuttle.....................290
R. Watson
Developing a Declarative Rule Language for Applications in
Product Configuration.....................................................305
T. Soininen and I. Niemela¨
Constraint Programming
University Timetabling Using Constraint Logic Programming..............320
H-J. Goltz and D. Matzke
Constraint-BasedResource Allocation and Scheduling in
Steel Manufacturing.......................................................335
M. Carlsson, P. Kreuger, and E. ˚Astro¨m
Using Constraints in Local Proofs for CLP Debugging.....................350
C. La¨ı
Declarative Languages and Software Engineering (Invited)
A Return to Elegance: The Reapplication of
Declarative Notation to Software Design...................................360
D.A. Schmidt
ECLiPSe: Declarative Specification and Scaleable Implementation ........365
M. Wallace and J. Schimpf
Author Index...........................................................367
Automated Benchmarking of Functional Data
Structures
Graeme E. Moss and Colin Runciman
Department of Computer Science, University of York,UK
fgem,[email protected]
Abstract. Despite a lot of recent interest in purely functional data
structures, for example [Ada93, Oka95, BO96, Oka96, OB97, Erw97],
fewhavebeenbenchmarked.Ofthese,evenfewerhavetheirperformance
quali(cid:12)ed by how they are used. But how a data structure is used can
signi(cid:12)cantly a(cid:11)ect performance. This paper makes three original con-
tributions. (1) We present an algorithm for generating a benchmark
according to a given use of data structure. (2) We compare use of an
automated tool based on this algorithm, with the traditional technique
of hand-picked benchmarks, by benchmarking six implementations of
random-access list using both methods. (3) We use the results of this
benchmarkingtopresent adecision treefor thechoice of random-access
list implementation, according to how thelist will beused.
1 Motivation
Recent years have seen renewed interest in purely functional data structures:
sets[Ada93],random-accesslists[Oka95],priorityqueues[BO96],arrays[OB97],
graphs [Erw97], and so on. But, empirical performance receives little attention,
and is usually based on a few hand-picked benchmarks. Furthermore, the per-
formance of a data structure usually varies according to how it is used, yet this
is mostly overlooked.
For example, Okasaki [Oka95] uses (cid:12)ve simple benchmarks to measure the
performance of di(cid:11)erent implementations of a list providing random access. He
points out that three of the benchmarks use random access, and two do not.
However, all the benchmarks are single-threaded. How do the data structures
perform under non-single-threadeduse? We simply do not know.
Okasakipresentsmanynewdatastructuresinhisthesis[Oka96],butwithout
measurements of practical performance. He writes in a section on future work:
\The theoryandpractice ofbenchmarking[functional]data structuresis still in
its infancy."
Howcanwe make benchmarking easier andmore reliable? A major problem
is(cid:12)ndingarangeofbenchmarksthatweknowusethedatastructureindi(cid:11)erent
ways. If we could generate a benchmark according to a well-de(cid:12)ned use of the
data structure,we couldeasily makeatable listing performanceagainstarange
of uses.
G.Gupta(Ed.):PADL’99,LNCS1551,pp.1{15,1998.
(cid:13)c Springer-VerlagBerlinHeidelberg1998
2 Graeme E. Moss and Colin Runciman
To make precise \the use of a data structure" we need a model. Section 2
de(cid:12)nes such a model: a datatype usage graph, or dug. Section 2 also de(cid:12)nes
a pro(cid:12)le, summarising the important characteristics of a dug. Section 3 gives
an algorithm for generating a benchmark from a pro(cid:12)le. Section 4 introduces a
benchmarking kit, called Auburn,that automates benchmarking using the algo-
rithmofSections3.Section4thencomparesbenchmarkingsiximplementations
ofrandom-accesslistsmanuallyagainstusingAuburn.Section5discussesrelated
work. Section 6 concludes and discusses future work.
Some of the details of this paper are only relevant to the language we use:
Haskell,apurefunctionallanguageusinglazyevaluation.Suchdetailsareclearly
indicated.
2 Modelling Datatype Usage
Howcanwecapturethewayanapplicationusesadatastructure?TaketheSum
benchmark of [Oka95] as an example of an application. Sum uses an implemen-
tation of random-access lists (see Fig.1) to build a list of n integers using cons,
and then sum this list using head and tail. Code for Sum is given in Fig.2(a).
Let us use a graph to capture how Sum uses the list operations. Let a node
represent the result of applying an operation, and let the incoming arcs indi-
cate the arguments taken from the results of other operations. Let any other
arguments be included in the label of the node. Figure 2(b) shows this graph.
Node 1 represents the result of empty, which is an empty list. Node 2 repre-
sents the result of applying cons to 1 and empty, which is a list containing just
theinteger1.Andsoon,tillnoden+1representsalistofncopiesoftheinteger
1. This is how Sum builds a list of n integers.
Noden+2representstheresultofapplyingheadtothislist,whichisthe(cid:12)rst
element in the list. Node n+3 represents the result of applying tail to this list,
which is all but the (cid:12)rst element. Node n+4 represents the result of applying
head to the list of node n+3, giving the second element. Every other element
of the list is removed in the same way, till node 3n represents the last element,
and node 3n+1 represents the empty list. This is how Sum sums the list of n
integers.
The authors introduced such a graph in [MR97], given the name datatype
usage graph, or dug. The de(cid:12)nition was informal in [MR97] but we shall now
give a brief formal de(cid:12)nition. To abstract over many competing data structures
providing similar operations, we insist on a dug describing the use of an adt.
The same dug can then describe the use of any implementation of that adt.
We restrict an adt to being simple.
De(cid:12)nition 1 (Simple adt)
A simple adt provides a type constructor T of arity one, and only operations
over types shown in Table 1.