Doc No: X3H7-95-25
Doc. Date: 13 October 1995
Project:
Reply to: Jeff Sutherland VMARK Software 50 Washington Street Westboro, MA 508-366-3888 x3728 508-389-8726 FAX jsutherland@vmark.com http://www.tiac.net/users/jsuth/1.0 FIXED EVENTS - none
2.0 PRELIMINARY AND ADMINISTRATIVE MATTERS
2.1 Establish scribe - Cory Casanave (Wed), Jeff Sutherland (Thu/Fri)
2.2 Attendance and introductions
Don Belisle, IBM (Wed/Thu) belisle@austin.ibm.com
William D. Burg, Univ. Texas San Antonio (Fri) wdburg@lonestar.utsa.edu
Cory Casanave, Data Access Corporation (Scribe Wed) cory_casanave@omg.org
Tom Digre, TI digre@ti.com
Brian Henderson-Sellers, Univ. of Technology, Sydney brian@socs.uts.edu.au
Glenn Hollowell, Sematech (Chair) glenn_hollowell@sematech.org
Randy Johnson, Dept. of Defense (Fri) drj@tycho.ncsc.mil
Haim Kilov, IBM kilov@watson.ibm.com
Frank Manola, GTE fm02@gte.com
Bill Milam, Ford Motor wmilam@ford.com
Joaquin Miller, Systemhouse (Vice Chair) miller@shl.com
Jeff Sutherland, VMARK Software (Scribe) jsutherland@vmark.com
2.3 Review the proposed work items
2.4 Administrative
Date Meeting Place/Concurrent meetings
2.5 Special topics
2.6 X3H7 Annual Report
2.7 Review/discuss submitted documents
2.8 Define and schedule work to be done in plenary and breakout groups
MOTION: That the minutes of Meeting 14 be approved as X3H7-93-17. (Passed by unanimous consent)
RESPONSIBILITY PRIMARY BACKUP
Chair Glenn Hollowell
Vice-Chair Joaquin Miller
Project Editor (Object Models) Frank Manola
Project Editor (Enterprise Viewpoint) Joaquin Miller
Vocabulary Rep Joaquin Miller Haim Kilov
Secretary Jeff Sutherland
OMC/SC21 TAG Rep Elizabeth Fong
Membership Status Joaquin Miller Elizabeth Fong
Document Log/Librarian Jeff Sutherland Elizabeth Fong
Document Distribution Elizabeth Fong
X3H7/SD-6-PA 7 Mar 95 X3 Technical Committee X3H7 (OIM) Principals and Alternates Mailing List by Elizabeth Fong
X3H7-95-17 25 Jun 95 X3H7 Minutes of Meeting 14, Jun 25-28, 1995, San Jose, CA by Jeff Sutherland
X3H7-95-18 20 May 95 Accomodating SQL3 and ODMG by Jim Melton, ISO DBL:YOW-031/X3H2-95-161
X3H7-95-19 18 May 95 Object extents refined by Jim Melton and Andrew Eisenberg, ISO DBL:YOW-059/X3H2-95-162R1
X3H7-95-20 25 Jul 95 RM-ODP Enterprise Viewpoint Component Standard - New Work Area Proposal by WG7, ISO/IEC JTC 1/SC 21 N9773
X3H7-95-21 9 Oct 95 OMG Business Application Architecture White Paper, Draft 2 by Cory Casanave (Ed.), OMG/BOMSIG Business Object White Paper
X3H7-95-22 8 Aug 95 Understanding the semantics of (collective) behavior: the ISO General Relationship Model by Haim Kilov, ECOOP'95 Workshop 6: Use of Object-Oriented Technology for Network Design and Management
X3H7-95-23 1995 Seamless Object Oriented Requirements Engineering & Modelling using OMEGA by Ian Graham, Swiss Bank Corporation (slide presentation)
X3H7-95-24 22 Sep 95 Common Facilities RFP-4: Common Business Objects and Application Component Framework by OMG Technical Committee (Draft B)
Meeting schedule
Co-locate with OMG
Glenn: Expectation that X3H7 will be done at the same time/place as OMG meetings.
Concern expressed as to meeting conflict. OMG is a full week.
Frank: Significant work will be required in meetings.
Cory: Have a one day meeting and intersperse work sessions with OMG.
Glenn: Expect BOMSIG RFP to influence X3H7.
Don: X3H7 work is much more general than OMG, it will be more like a modeling language.
Glenn: OMG standard can "slide into" ANSI work.
Hiam: This work is well focused and based on the paper, let BOMSIG do the more concrete work.
Frank: It is unclear if this group or BOMSIG is doing the more concrete work.
Conclusion, next meeting: Sunday Jan 7 - Monday 8th 1995 in San Diego.
Glenn: Presentation to ANSI: May 13th, Kansas city. One meeting between January and this will be required.
OMG Sept 16-19th, 1996 - Hyannis, Ma
Tentative meeting: Sunday Sept 15-16th, Hyannis, MA
Approval of Minutes
Minutes of June 25th, 1995
Call for correction: None
Glenn: Motion to accept by white ballet
Motion to approve June 25th minutes passes by unanimous consent
Membership status
No action
Annual report
Glenn: Did report last year. Official schedule is November, but this can be changed. Will be an opportunity to "get blessing" with what we are doing. It will be scheduled with November or January. Draft will be sent prior to presentation.
Action item delegated to the Chair.
Liaison report
BOMSIG liaison will be discussed later.
X3T3 (Glenn), work item submitted to X3T3. They helped with draft. They expect this group to do the technical work.
X3J21 (Haim): "Z" is a commited standard. Report on use of formal method available. International relations model "General Relationship Model" 10265-7.
Break for lunch
Glossary
Haim: Work done is 2 years old, should not be used without review.
Glenn: Glossary is a collection of definitions for each term.
Josquin: Volunteers to bring up to date and circulate the results for January.
But, glossary is not necessarily intended to go in the report.
semantic enterprise mode
Ottawa meeting results
Started with report. U.K. was concerned that this work conflicted with the Conceptual schema modeling facility of ISO under the NWI proposal WG3CSMF.
Glenn asserts that this relates only to the system model that may be used to implement.
The relationship between the two efforts is sumarised in paragraph 2.3. It was agreed that the groups would stay in touch with each others work.
Document requirements
Frank: What is the connection between the computational entity that implements a business object and the ODP concept of enterprise viewpoint?
Josquin: The essence of a purchase order is the commitment that the business enters into to execute a trade. It is this concept that the business object is representing.
Secondary agenda:
-Purpose
-How to narrow work item, make work item more precise
-Our relation to OMG
9:00 RM-ODP Enterprise Viewpoint
Assumption: there is a basic infrastructure (CORBA) that allows message passing between objects
Glenn would be happy with a roadmap t interoperability. We don't need to get all the way to interoperability.
Application Architecture Semantics - Don
Business Application Architecture? Tends to rule out some main line business? No, we are trying to rule out basic operating system domain.
Behavioral semantics of business - Cory
Needs to be generally applicable - Don
Level of abstraction between model and thing you are modeling is important. Talking about business means talking about higher level of abstraction.
OMG has defined several technical object frameworks.
Application components sounds better than business objects - Cory, Don
If I want to go to the object store I need to be able to say what I want and know whether the product I am looking at does it - Haim
How can we specify a system object model.
SQL specification describes behavior. The fact that it is a language is not the point. But for SQL, we had initially Codd's 10 page paper that laid out all the concepts. First SQL standard was not very good. There has been a lot of evolution.
We do not have to define an enterprise language perfectly on the first attempt.
Could we write the paper that is analogous to the original Codd paper? RM-ODP references an Enterprise Language. I resist language. It is an arcane form of expression - Cory. RM-ODP language is a general term for a wide variety of notations. It does not describe a particular language, but states that the language must be able to express these concepts. Different viewpoints may use different languages - Haim.
Before SQL there were multiple languages, QUEL, etc. But Codd's paper could help distinguish whether on not the language supports relational concepts - Haim. This was because the notion of relational completeness was defined - Frank.
Technique <OR> Specs
if do X, y happens, with side effects z
language a la C++
language a la (subset of English, set of concepts)
"Enterprise completeness"
Working rule - assume whole world has one vendor/language/os underneath - Joaquim. Better yet, assume OMG has taken care of the plumbing - Glenn. I can use any implementation I want.
BOMSIG RFP picture above is very useful for defining layers. Application Component Framework is not the right words, behavior semantics of business components (Haim).
Goal: Vendors can specific complete behavior of their product by using this language that specifies components - Don.
Specify sets of concepts.
In picture, what are we specifying. We are looking at common supertypes of objects in the picture.
Dynamic supertyping? Dynamic multiple inheritance as roles. Component needs to specific behaviors as events which other components can register for. This is a type of relationship - Haim. If we clarify these types of concepts and clarify them (there are probably not more than 10 of them) we can solve the problem. Vendors can integrate without being dependent on the internal behaviors of their systems.
We are talking about a connectivity framework - Frank. Need metadata - Cory.
Agenda Part A
What is our purpose?
How do we narrow the work item?
How do we make the work item more precise?
What are out relations with OMG?
We have come to an important consensus - Haim/agreed by group
It was an eye opener for a customer that use cases were helpful, irrespective of whether these were used to implement a computer system - Haim. BPR is independent of computing systems, even though IT shops are driving a lot of BPR efforts - Frank.
11:20 Agenda - Part B
What is "Enterprise Viewpoint" vs. "Enterprise Modelling"?
What is a "Business Object" and/or is this concept off our agenda?
What needs to be added to this list:
What needs to be added to this list:
What needs to be added to this list:
6. Is role equivalent to "dynamic supertype" or "dynamic class" or "dynamic multiple inheritance"? Answer: Yes. But there are a number of definitions for role - Haim.
X3H7 will remain X3H7 but will have standing. We will be the advisory group to the U.S. national body that votes in the ISO process. We will be a part of WG7 through the auspices of X3T3 who is the official U.S. TAG for RM-ODP and may have an independent role.
We need to progress our draft of the RM-ODP Enterprise Viewpoint Component Standard. There are five or six other standards groups that will comment and add to the draft. We will be the editor. If we are the Secretariat, we determine the rate of progress of the draft.
Membership in X3H7 is a company membership and company vote. The cost is $300/ year. With an international piece of work there is an additional $300 charge per company. Any number of people from a company may attend the meetings, but only one vote per company.
ACTION: Glenn will present our annual report in November to X3 and clarify the relationship of X3H7 to ISO and make a recommendation back to the Committee whether we should be the Secretariat for our item of work.
1:25 Reconvene after lunch
We buy with fuzzy specification. We know what we buy needs to work with what we've got (PC, 486, ISA bus, etc.)
Consider DBMS procurement. Some people write books for specifications. Some people write a few pages. Neither are very precise if you look carefully at them. They are not necessarily complete. You may be look at supplier to help educate you on what the complete specification should be.
If TI were to say to build the users a system with off the shelf components wherever you can find them, building only what you have to, then I would go to the users, interview them, and flesh out a specification. Then I would go to the "object store" to find what I could buy as component pieces. These would not be applications, but pieces that could be used by applications. They make be in multiple languages. They may even talk to each other. Tom is in the business of doing just that for TI - Glenn.
There are a number of application development groups, a number of legacy systems, a number of purchased systems. All need to interoperate. We want to by off-the-shelf components. The SEMATEC framework specifies some interoperable components. We need to extend this across the enterprise with an enterprise model. No ever wafer fab will be the same so we need to be able to tailor the components and frameworks - Tom.
The SEMATEC framework has generic concepts populated with domain specific components - Glenn/Tom.
I want to stay with the object store idea. Is the catalogue in the object store similar to a Motorola or TI chip catalogue? Or is it more like I am shopping for a new video board for my PC. Describing the video board is more complicated in detail, but more simple in that I don't need to know all the details - Frank. What TI is trying to put together deals with both levels - Tom.
Most people will be buying the PC board, rather than the chip. The video board supports two standards, ISA bus and VGA output. Both are complex standards but the buyer doesn't need to know the details. But some parts of large organizations may need to be able to build special boards that plug into an ISA bus and need to understand ISA in detail - Glenn.
I think we are uses examples that are too simple to be relevant to software. Software is more like a very complicated stereo component with many controls on it - Joaquin.
If we are really going to talk about this problem, we need to talk about multiple specifications defined by many industry groups (VGA, for example), that are transparent at a higher level. At the higher level, we only need to know it is compliant at a lower level.
We are working on how to specify an entire system based on a set of user business requirements. OMG has specified CORBA and object services. We don't need to do that. When our user wants something that only works with a SCSI connection, that we can specify a request for a component that with plug into a SCSI connection with certain behavior characteristics. We want to be able to define the specification so that we can go to the object store (like Circuit city for PC boards) and buy software components - Glenn.
SAP comes with 41000 switches which are equivalent to rules. How long does it take to set up? It can take 6 months to two years just to migrate from one release of SAP to another in a large production environment.
For a video board, it is understood that when a signal is sent through the board when it is plugged into the TV monitor, a picture will appear. The HR analogy is the employee object is the signal. When you plug in an employee, how does jury duty get handled?
How can business rules be expressed? How do they relate to the rest of the system.
If I go to the object store, I can specify the recruiting function to the supplier. I want to be able to do various things with resumes using pre- and post-condition. Then when I go to the interview function, I talk about persons and resumes are no longer relevant.
I want to be able to buy the component, throw out one piece, and rewrite it on my own.
There are other ways to build a framework.
I don't want to standardize how to express business rules. This area needs to evolve. I want to allow for using various ways of specifying business rules - Cory.
We can allow business rules to be specify rules in many ways - language, english, etc. But we should be able to ask a vendor to specify rules in a way they will interoperate in a framework - Haim.
There is going to be at least one way to specify any concepts of interest. The problem in providing a generic way to specify.
What we are doing is not stating how to specify the catalogue. It is more like an ASICs catalogue of ready made chips that we will select and use in multiple combinations and permutations - Glenn.
3:00 PM Draft Document to present to ISO in April
We need to write a draft of a document with section headers, some of which may be populated. Enough should be done so that it can be reviewed and effectively decided whether further work should proceed - Glenn.
It should be multiple parts. Part I should be design considerations of what an enterprise viewpoint standard is about - Haim.
We need to approve a draft in January that will be sent out six weeks before a May 1996 meeting. This is Attachment 3 with is Draft 0.1 (a skeleton) of a proposed standard. We could take the outline of RM-ODP Part I and II and rework it.
What we have talked about is general justification and approach which should be Clause 0 of Part I. Examples should be in Part I. The several bullet consensus developed above should be the beginning of Part II.
ACTION: Joaquin volunteered to be document editor through 1 April 1996.
ACTION: Frank volunteered the GTE FTP server for storing current document versions.
ACTION: Jeff volunteered VMARK exploder for mailing list for work on ISO proposal.
3:40 PM Begin Outline of Proposal
RM-ODP Part II Proposal
0 Introduction
1 Scope
2 References
3 Definitions
4 Abbreviations
5 The Descriptive Model
6 Basic Interpretation Concepts
7 Basic Linguistic Concepts
8 Basic Modelling Concepts
9 Specification
10 Organization Concepts
11 Properties of Systems and Objects
12 Naming concepts
13 Concepts for behavior
14 Management concepts
ODP Role: Identifier for a behavior … which is associated with one of the component objects of a composite object.
In Part III there is a statement about the Enterprise Viewpoint that we need to take a look at.
9:10 Reconvene
Next meeting is second week of January. We will target the next meeting in March before our 1 April deadline for submission of our ISO proposal.
Enterprise Language - RM-ODP Part III
X3H7 Additions
Where in this standard will we specify things in enough detail so that code can be written to it? RM-ODP is a meta-standard today. It is not possible to determine conformance today to RM-ODP. For example, is the OMG work conformant. What we are doing here can either be an RM-ODP abstraction with definitions or we can specify where OMG work is an instance of an Enterprise component, for example. RM-ODP has a conformance testing section.
The difference between the CORBA model and RM-ODP need to be worked out. This is being discussed in the OMG Object Model group. Objects with multiple interfaces, for example. What RM-ODP calls an object is an interface. If may be possible to use RM-ODP concepts in a way the relates them to CORBA. Glenn did an RM-ODP/CORBA analysis two years ago there was nothing that conflicted except the multiple interfaces.
If OMG IDL becomes an RM-ODP standard (as we think it will), we will have to describe interfaces for computational objects in IDL. These would be examples for our work at a higher level of abstraction.
10:30 Assignments for January Meeting
11:35 X3H7 Web Page Discussion
Frank Manola call X3 and tell them to put pointers from X3H7 Home Page to pages maintained by Frank Manola and Jeff Sutherland.
11:45 OOPSLA Workshops
We should talk about our new work item at workshops at OOPSLA - Haim. We will verbally announce it and have copies of the work item available.
The work item needs a cover sheet that says who we are and who to contact to participate.
BOMSIG Meeting tomorrow starts at 2PM at Sematech. Visitors lobby is not open. Come in the east entrance. Guard station in lobby. Conference room FC on first floor. X3J21 will be all day in the fifth floor board room.
12:00 Adjourn
TOP of Page
Jeff Sutherland's Object Technology Home Page
This WWW page was built by jsutherland@vmark.com.
Your feedback is appreciated.
| Page Hits | |
|---|---|
| This Page | |
| Since 10/13/95 | |