This article was downloaded by: [Arizona State University]On: 10 November 2014, At: 10:24Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Information Technology for DevelopmentPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/titd20

The ETHOS project: An introductory viewTadao Takahashi a , Carlos J.P. Lucena b & Armando M. Haeberer ca Research & Development Center (Telebrás) , on leave at CNPq (Research & DevelopmentCouncil), São Paulo, Brazilb Department of Informatics , Pontifical Catholic University (PUC) , Rio de Janeiro, Brazilc Latin American Advanced School of Informatics (ESLAI) , Buenos Aires, ArgentinaPublished online: 12 Jul 2010.

To cite this article: Tadao Takahashi , Carlos J.P. Lucena & Armando M. Haeberer (1989) The ETHOS project: An introductoryview, Information Technology for Development, 4:1, 31-40, DOI: 10.1080/02681102.1989.9627139

To link to this article: http://dx.doi.org/10.1080/02681102.1989.9627139

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in thepublications on our platform. However, Taylor & Francis, our agents, and our licensors make no representationsor warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Anyopinions and views expressed in this publication are the opinions and views of the authors, and are not theviews of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should beindependently verified with primary sources of information. Taylor and Francis shall not be liable for any losses,actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoevercaused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Information Technology for Development Volume 4 Number 1.© 1989 Oxford University Press.

The ETHOS Project: An Introductory View

Tadao Takahashi*Carlos J. P. Lucena†Armando M. Haeberer‡

Abstract This paper presents an introductory (and largely non-technical) view of Project ETHOS, an Argentine-Brazilianjoint effort in research and development in InformationTechnology which aims to build a CASE (Computer AidedSoftware Engineering) workstation with several innovativefeatures.

In the first part of the paper we discuss aspects of thecontext within which the project has been launched and isexpected to bear fruits. In the second part we present therationale behind the decisions already taken in the project,as translated into research directions for Phase I (1988-1990).

1 introduction It is generally agreed that we live today amidst a process ofworldwide changes in depth and scope solely comparable tothe Industrial Revolution, and which will result in theemergence of a new social and economical order amongnations.

Information Technology has been regarded as playing acentral role in this new industrial revolution and itsresulting information-oriented society. Thus, the masteryof this technology is vital for countries which aim toparticipate in this process of changes as more than meretargets of advanced countries' policies and actions. This isthe main reason underlying the efforts of developingcountries in conceiving their own policies in Informatics; onthe other hand, this helps explain the sometimes strongreaction of some developed countries against such efforts.

Basic technology in Informatics translates into the ability

* Research & Development Center (Telebrás), on leave at CNPq (Research &Development Council), São Paulo, Brazil.

† Department of Informatics, Pontifical Catholic University (PUC), Rio deJaneiro, Brazil.

‡ Latin American Advanced School of Informatics (ESLAI), Buenos Aires,Argentina.

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

32 Tadao Takahashi et al.

to build and use artefacts which, in rough, are composed ofhardware and software. Hardware refers to the physical partof such artefacts, a complex assembly of mechanical,electrical and electronic parts following a highlysophisticated design. Software refers to instructionsequences specifically devised for an application in such away that, upon execution, the hardware will operate in themanner foreseen for that application. The processesinvolved in the design of computers and implementation ofsoftware display a level of complexity rarely (if ever) seen inmanufacturing activities. Furthermore, use has added tothe inherent complexity of applications the immensechallenge of embodying 'intelligent' behavior intocomputers, for man/machine communication, abstractreasoning, etc. Thus, the development of a computer-basedapplication all too often requires, beyond the do's anddon'ts of the application area itself, excursions throughtopics of such diverse fields as linguistics, psychology,management, etc.

1.1 The software problem Software constitutes today one of the most critical aspectsin Information Technology, due among others to thefollowing reasons:

© intrinsic flexibility: 'software' is the added-on featurewhich makes a standard equipment behave in particularways as specified for an application. This intrinsicflexibility makes it possible to use computers in anunbounded range of applications, both in nature (fromchildren's toys to nuclear plants) and in scope (frompersonal computers to transnational networks). Thesame flexibility also allows computers to be finely tunedto traits and idiosyncrasies of individuals and cultures(so that, for example, a single application can interactwith different users through commands either inPortuguese or in Japanese). However, the price of thisflexibility is the ever increasing complexitity of'software' development.

m background knowledge requisites: 'software' develop-ment for an application requires a body of knowledgedrawn from at least four areas, namely:—the application domain,—the organizational context within which the

application will operate,—hardware and software configurations adopted, and—systems design and programming techniques.

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

The ETHOS Project: An Introductory View 33

It is not surprising then that software developmenttimes are unbearably long and the backlog of newlydemanded applications grows each day.state-of-the-art: the software problem is furthercompounded by the inadequacy of the developmenttools actually employed in practice. Several studiesshow that the timespan measured from the inception ofa proven tool or technique until its generalized use inproduction settings reaches on the average 15 years. Inother words, several powerful results of last decades'research activities still have to find their way intoindustrial production environments. Finally, it is fair tosay that software's inherent elusiveness has precludedthe straightforward application of managementprocedures such as interfacing (i.e., packaging)standards, version control, quality assurance methods,etc., long known in other manufacturing activities.

7.2 R & D in software Recent years have witnessed an upsurge of large projectsaddressng software issues. The Fifth Generation Project inJapan has included a major thrust in software technologyamong its several objectives. In the wake of the Japaneseproject, similar initiatives were taken in the USA andEurope. In the USA, noteworthy examples are the MCC(MicroElectronics and Computer Corporation, aconsortium of computer manufacturers) with its SoftwareTechnology Program, and the Software EngineeringInstitute, a DOD-sponsored organization which fostersresearch in Ada environments, studies in intellectualproperty of software, etc. In Europe, the best examples areprojects launched within the realm of the ESPRITProgramme, where about a third of the projects aresoftware-oriented.

Generally speaking, these research efforts aim for a reallymeaningful leap in software development productivity.Research activities may adopt an evolutionary approach(that is, aim at new results derived from the improvement ofpresent-day Software Engineering Techniques) or,alternatively, adopt a revolutionary approach (that is, try todevise techniques which clearly break away from the statusquo). A good example of a revolutionary approach isrepresented, today, by the application of techniquesoriginally developed in the Artificial Intelligence area tosoftware development problems.

In a different dimension, another distinguishing featureof current R & D efforts is that the organization itself has

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

34 Tadao Takahashi et al.

evolved into highly structured models, with an explicitlymulti-disciplinary and multi-institutional stand from theoutset. Such models have been adopted as the only way forbudget-conscious institutions to face the soaring costs ofresearch in certain areas.

Finally, the last interesting aspect to consider is themeaning of research as opposed to development in thesoftware area. As it commonly happens in other fields,many fundamental results in software theory were initiallyconceived in restricted contexts, typically in the solution ofa very particular problem, and only afterwards did theyundergo a phase of theoretical treatment andgeneralization. Examples abound, for instance, in areas likeOperating Systems and Compilers. On the other hand,research groups have frequently advanced in the oppositedirection, that is, experimenting with one or anotherapplication of a general concept and going as far asimplementing software packages with pre-competitiveindustrial quality. This advance of research groups in thedirection of experimental development, if common in theUSA and Europe, is rarely seen in less developed countries,which cannot sustain such costly activities for more than arelatively short time span. Consequently, in countries likeBrazil, where intensive communication between theresearch and industrial groups is specially critical, the rule ismore of distance and aloofness than of interaction andsynergy, with all negative consequences: research groupsneither find natural outlets for their more practical results,nor do they exert strong influence upon trends anddecisions taken in industry; in their turn, developmentgroups all too often miss a more conceptual and globalvision of the tasks they have to execute, mostly pressed bywhim of pure circumstances rather than thoughtfulplanning and analysis of experiences of the academicinstitutions.

1.3 Software research in Argentina and Brazil have several research groups inArgentina and Brazil 'software', in most cases belonging to departments of large

universities, and their production is regularly reported ininternational events and periodicals. That means thatsoftware research in both countries is lacking in quantityrather than in quality.

It is clear that, as most countries, Argentina and Brazilhave to invest much more in software technology in orderto keep abreast of future needs. Furthermore, it is clear thatnew research ventures must be carefully organized lest theirpotential results become dispersed or, even worse, fall short

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

The ETHOS Project: An Introductory View 35

of attainment. Some of the points to consider are thefollowing:

• projects should contemplate bold, long-range objectivesMulti-institutional and multi-disciplinary projects willgenerate results only after a number of years.Accordingly, the main objectives of such projects mustbe far-sighted, so as to be up-to-date when achieved andenable the bootstrapping of new research efforts.

• effective commitment of resourcesThe best way of losing money is investing less thanstrictly needed. Possibly the biggest challenge ofresearch efforts in less developed countries is ensuringthat large, strategic projects will remain so until theystart to bear fruits.

® careful planningResearch objectives and Development objectives mustbe clearly identified, and dependencies must bedecoupled and continuously monitored. A contractualapproach to task assignment must be adopted. Issuesand mechanisms for technology transfer andintellectual property of results must be adopted at thebeginning of a project and there must exist a legalmeans of ensuring such.

2 The ETHOS project

2.1 Background

In February 1986, a workshop took place in Campinas(Brazil), which brought together researchers fromArgentina and Brazil with the objective of identifyingprospective areas for cooperative research in Informatics.The discussions were organized into seven committees(Signal Processing, Computer Networks, SoftwareEngineering, Microelectronics, Industrial Automation,Programming Theory, Computer Architecture, andArtificial Intelligence) which, at the end of the meeting,presented a plan of activities and projects in each area forfurther consideration by the then debut-making Program ofCooperation in Informatics between Argentina and Brazil.

By mid-1986, the Coordination of the Program decidedto propose a new project, covering the areas of SoftwareEngineering, Programming Theory, ComputerArchitecture and Artificial Intelligence. The idea was tojoin the identified interests in all four areas into a singleencompassing project, so that the generated whole wasmore meaningful than the mere juxtaposition of its(formerly) self-contained parts.

Thus began the Project ETHOS.

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

36 Tadao Takahashi et al.

Several meetings, discussions and preparatory activitiesfollowed, with the participation of researchers from nearly20 institutions in Brazil and Argentina. As a first attempt,the project was divided into three phases: a PreliminaryPhase (1987) during which the detailed specification of theproject should be generated; a Research Phase (1988-1990),during which research-oriented activities would prevailover experimental development; and a Development Phase(1900-1992), where quality-level development wouldbecome an important issue. In February 1988, the PreliminaryPhase was successfully concluded, and the Final Reportshows that ETHOS has evolved into one of the largestresearch projects in Informatics ever launched in Brazil orArgentina.

2.2 Focus and scope of ETHOS aims to promote, carry out, and/or integrate theresearch necessary R & D activities for the construction of a

prototype workstation for methodology-centred softwaredevelopment. Research activities will be conducted in areassuch as Formal Methods, Software Tools, Data Bases,Knowledge Representation, Graphical Interfaces, etc.Development activities though experimental as a rule, willabide by a number of recommendations especiallyenvisioned to ease the transfer of results to industrialpartners. The scope of research activities is daringly broad,exploring four inter-related dimensions which are seldomconsidered together in a single project: programming-in-the-large, environment generation, artificial intelligence, andformal support.

m Programming-in-the-large: there is a generalizedconsensus about essential differences betweenprogramming-in-the-large and programming-in-the-small. Roughly, the former refers to the construction ofsoftware artefacts whose own size and complexitydemands a collective approach; this distribution ofdevelopment tasks imparts the need for a strictlycontrolled integration and system configuration phase.On the other hand, the latter does not raise suchmanagement needs and concentrates itself more on thebasic task of software construction.

ETHOS intends to support real-life developmentmethods. This implies that the workstation will have toallow software engineers to:

—model software life cycles,—generate and manage configurations of systems

described as composites of smaller units,

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

The ETHOS Project: An Introductory View 37

—associate and control cost/time estimates ofindividual tasks,

—develop and integrate method-dependent tools andtechniques, etc.

Therefore, ETHOS will support programming-in-the-large.Environment Generation: ETHOS as a Meta-environment: admitting as realistic that in theforeseeable future several development methods will co-exist, it is evident that one of the basic goals of a projectlike ETHOS must be the construction of general toolswhich are adaptable and configurable to supportdifferent methods. The generalization of this objectiveleads to the idea that ETHOS must primarily aim at theconstruction of meta-environments which can be (semi-)automatically customized in order to produce specificenvironments. Therefore, ETHOS will implement ameta-environment.Artificial Intelligence: software development processesare far from well-structured, insofar as we understandthem today. Furthermore, no matter what phaseprecedes the materialization of a system, thespecification seems to remain an essentially cognitivetask. From specification to the full detailing of aprogram, development steps are chosen based on alarge amount of heuristic criteria. Therefore, it is notsurprising that an important line of current research inSoftware Engineering starts from the premise thatsoftware development is essentially a knowledge-basedactivity and, consequently, is best modelled by ArtificialIntelligence techniques. This knowledge can be par-titioned in two levels, namely:

—general knowledge upon which the structure of adevelopment environment can be built, by:*representing project activities in hierarchicalnetworks with attributes of time, action, authority,etc., and

*representing software structures in hierarchicalnetworks with pertinent attributes attached to eachcomponent.

—specific knowledge incorporated into special-purposetools (e.g. module interconnection, symbolicdebugging, etc.) to actively assist users in their properuse or, even, take over clerical tasks.

ETHOS will include intensive research in ArtificialIntelligence techniques applied to Software Engineer-ing.

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

38 Tadao Takahashi ef al.

Formal Support: several researchers argue that only asolid formal approach can impart rigor, conceptualintegrity and global coherence to the softwaredevelopment process and, by extension, to the qualityof the resulting products. A brief survey of the areashows that proposed formalisms are mainly devoted tomodelling software structures (i.e., products) and, evenin this case, are lacking in general applicability: forinstance, some approaches are ill-suited to modellingparallelism; others are not easily extensible toprogramming-in-the-3arge, etc. A different (and lessfamiliar) approach is the formal description of softwareprocesses (rather than structures or products) as thewell-controlled application of transformations chosenat each step from a base of canonical developmentoperation. It is agreed today that a most generalapproach will have to start with an overallencompassing model for software processes (accountingfor heuristic transformations as well) under whichseveral complementary formalisms will model softwarestructures and techniques. At any rate, the importanceof forma! approaches cannot be overestimated. ETHOSwill include intensive research in Formal Methodsapplied to Software Engineering.

2.3 Phase II(1988-1390): researchplus miniETHOS

As mentioned before, the ETHOS project now enters theResearch Phase, during which research-oriented activitieswill prevail over development-oriented ones. Given thedifferent nature of these activities, a clear decoupling ofinterdependencies was planned. In order to ensure that therelative autonomy of the research and development frontswith respect to each other would not harm the conceptualintegrity of the whole, a number of preliminary decisionswere taken which cannot be altered without a newconsensus.

2.3.1 Theoretical Foundations and General Paradigms

® An extension of Lehrnan-Turski proposals, the so-called Inverted-U Meta-model, will be the basicformalism for the description of software developmentprocesses, complemented by other formal methods(algebraic specification, CCS, etc.) as it may seemnecessary.

© Frames and semantic nets will be the basic paradigm forknowledge representation, complemented by rules,axioms, etc., as it may seem necessary.

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

The ETHOS Project: An Introductory View 39

Equational logic will be the main implementationformalism.Object orientation will be the basic paradigm ofsoftware design and implementation for theworkstation itself.

2.3.2 Minimum Common Environment: MiniETHOS

A minimum common environment for the initialdevelopment of ETHOS, the integration of separatelydeveloped results, and the preparation of the Phase II infra-structure will be developed during this phase, usingstandard workstations (SUN, Apollo, etc.) and softwaretools. Activities leading to this objective will be organizedinto a sub-project called MiniETHOS.

• The Minimum Common Environment will include:

—a generalized user interface management system(UIMS),

—a syntax-oriented environment generator,—a knowledge-base and manipulation tools,—a production-oriented development chain for an

extension of the C language, and—a software database.

« The development of MiniETHOS will be supported bythe following tools:

—UNIX,—Gandalf (CMU)—IDL (UNC),-X-Windows (MIT), and-GKS-2b (LOWELL).

2.3.3 Testbeds

Experimental environments will be generated to exploreand test ETHOS for at least the following two applicationdomains:

® information systems, particularly office automationsystems, adopting one or more of commonly usedmethods such as Jackson's, Warnier's, etc.

® real-time control, exploring facets such as concurrency,time constraints, etc., not existing (or not so critical) inthe information system area.

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014

40 Tadao Takahashi et al.

3 Final remarks 'ETHOS' is the acronym of 'Software Oriented HeuristicWorkstation' in Portuguese as well as in Spanish. Each ofthe composing terms contributes with one aspect to thecomplete meaning of the project name.

On the first acquaintance with the objectives of theETHOS project, one cannot help wonder whether they arenot too ambitious for the Argentinean and Braziliancontexts, both from the technical and organizational pointsof view.

After two years of work, pooling technical expertise inArgentina and Brazil and painstakingly planning minute-detailed activities of the project, we can confidently refutedoubts on the feasibility of the ETHOS project per se.

Giving it a second thought, ETHOS in fact poses thesober question of whether, and to which extent, it ispossible to foster, in less developed countries, a researchparadigm and a scientific culture with a distinct, indigenouspersonality. That is, ethos.

Dow

nloa

ded

by [

Ari

zona

Sta

te U

nive

rsity

] at

10:

24 1

0 N

ovem

ber

2014