L'OBJET 1.3: Le texte anglais du document ANIMO
AGREEMENT ON CORE OBJECT METHODOLOGY PRINCIPLES (ACOMP)
1. CHARTER
The goal is to quickly achieve a written concensus on the fundamentals
of object-oriented development methodsto be endorsed by as many
published methodologists as is practicable. Initial focus will be on
those things that we either already informally agree upon or those
which we can agree to easily and rapidly. Official endorsement need
not mean 100% agreement in all items; rather it may represent a
acceptable compromise for the sake of standardization. The
signatories agree to migrate their approaches (publications,
presentations) towards consistency with the standard within a
reasonable and agreed timeframe.
The scope includes terminology (syntax and semantics), potentially
notation and documentation (number and type of diagrams vs icons),
modelling techniques, high-level process, life cycle underpinning.
We do not want to promote any single methodology; neither do we wish
to preclude any existing methodologies or any future advancements in
the field. We promote extensibility in the framework. Minimum but not
necessarily sufficient conditions.
The motivation is to promote the acceptance of object technology,
avoid the current ad hoc "tower of Babel", support for CASE tool
vendors; allow the methodologists to concentrate on their important
value-added characteristics which differentiates different OOA&D
approaches; elimination of unnecessary differences.
2. PLAN
-
1. Produce charter
2. Announce formation of working group 2.1 Email to methodologists
-
-
2.2 Announce on Comp.object
-
2.3 Announcements in SIGS publications 2.4 Announcement at conferences
-
2.5 Setup liaisons with other organizations and working groups (OMG,
X3H7, ISO, CDIF)
-
2.6 CACM letter
-
3. Obtain membership
-
4. Construct initial list of potential agreements 5.
Voting/acceptance/making consensus official
process - how to do. How to deal with exceptions and caveats.
-
6. Allocate workitems to subgroups
-
7. Construct process by which results and comments are iterated around
membership
-
8. Iterate within subgroups. Combine material from subgroups into full
group document and iterate this full group document.
-
9. (Vote) on group document and eliminate violent disagreements
-
10. Publish OOAD manifesto. Wide publication. Short paper ACM and if
possible full paper CACM; conference presentations; SIGS publications.
-
11. Determine if end of Phase I should lead to commencement of Phase
3. INITIAL SUBGROUPS AND VOLUNTEERS
Concepts: Terminology and definitions (Firesmith)
Diagrams: terminology and definitions (Firesmith)
Overall models: e.g. static model contains diagram + .... (Odell and
Firesmith)
Techniques: e.g. CRC, role playing scenarios
Notations: (Henderson-Sellers and Constantine)
Lifecycle: iterative, incremental, parallel/concurrent development
[Candidates: spiral, fountain, baseball, pinball, vortex]
Henderson-Sellers, Walden, Nerson, Firesmith
Deliverables (maybe)
3. POTENTIAL MEMBERSHIP
Kent Beck
Ed Berard
Michael Blaha
Grady Booch
Andy Carmichael
Peter Coad
Ed Colbert
Derek Coleman
Steve Cook
Ward Cunningham
John Daniels
Desmond D'Souza
Fred Eddy
David Embley
Don Firesmith
Martin Fowler
Adele Goldberg
Ian Graham
Brian Henderson-Sellers
Ralph Hodgson
Ivar Jacobson (ask if any of co-authors)
Gerald Kristensen
Barry Kurtz
William Lorensen
James Martin
Steve Mellor
Bertrand Meyer
Jean-Marc Nerson
Peter Nevermann
James Odell
Karin Oppel
Meilir Pages-Jones
Bill Premerlani
Kenny Rubin
James Rumbaugh
Sally Shlaer
Jim Thomann
Kim Walden
Katherine Whitehead
Brian Wilkerson
Scott Woodfield
Rebecca Wirfs-Brock
Ed Yourdon
4. COMMA, ACOMP and other acronyms
Convergence is in the air. Methodologists are talking to each other
and agreeing to combine their analysis and design techniques in such a
way as to offer to industry a more rational choice. Booch and OMT,
arguably the most popular approaches, are converging on, at least, a
single notation. A group of other methodologists also see the
contribution that combining their strengths might have on the
introduction of object technology into that vast, as yet untapped,
market --- especially the areas of traditional MIS/COBOL. Established,
yet perhaps less widely known, quality OO methodologies to participate
include MOSES, SOMA and Martin/Odell. The new methodology, code-named
OMEGA, had its first public appearance at OOPSLA in October and will
be consolidated over the next few months.
At the same time, many methodologists have agreed that metamodelling
may provide a way forward to both underpinning the methodologies
themselves by something a little more rigorous than the current
informality and also to creating a potentially agreeable core
metaobject model for all methodologies to adhere to. This project,
known as COMMA (Common Object Methodology Metamodel Architecture),
was proposed in late 1994 (Henderson-Sellers, 1994) and was funded for
the first six months of 1995. COMMA aims to take the published
descriptions of the leading OO methodologies and derive their
metamodels using a metamodelling notation. These descriptions are then
to be circulated to the method developers for their ratification
and/or correction. Based on these agreed individual metamodels, a
proposed core metamodel will then be circulated for agreement.
To date, fourteen of the leading OO methodologies have been
metamodelled. Comments received back from the developers have led to
minor modifications. A full report, containing details of all these
fourteen metamodels will then be created and placed in the public
domain.
In parallel, an initiative known as ACOMP (Agreement on Core Object
Methodology Principles), initiated at the TOOLS conference in August
1995, aims to build on the COMMA metalevel ideas and the collection of
data embodied in the Object Dictionary of Firesmith and Eykholt
(1995). This dictionary has collected together ALL the definitions
within the current object-oriented analysis and design techniques and
provides a rich source of data to complement the metalevel data of
COMMA. ACOMP's goal is "to quickly achieve a written consensus on the
fundamentals of object-oriented development methods to be endorsed by
as many published methodologists as is practicable". ACOMP will not
promote any single methodology, nor will it preclude any future
advancements in the field; rather it provides minimum but not
necessarily sufficient conditions for the development of coherent and
flexible object development frameworks and aims to eliminate any
unnecessary differences. Small panels will be formed to address
specific methodological issues and the results circulated widely in
order to foster agreement.
In parallel to COMMA, the OMG is considering the established of a Task
Force, under the guidance of Mary Loomis, to evaluate standardization
of OO methodologies. COMMA is one of the inputs to that OMG process.
The results of ACOMP will similarly be placed at the disposal of the
OMG in an attempt to create, as rapidly as possible, a tenable
standard, yet flexible, approach to object modelling and systems
development.
As Richard Soley foresaw in the July issue of Object Magazine, this
(northern hemisphere) summer has been an exciting time for OO
methodologies and the establishment of new standards within the
object-oriented community.
5. REFERENCES
-
Firesmith, D.G. and Eykholt, E.M., 1995, The Dictionary of Object
Technology, SIGS Books, New York
-
Henderson-Sellers, B., 1994, COMMA: an architecture for method
interoperability, Report on Object Analysis and Design, 1(3), 25-28
|
