OMG Internet Platform Special Interest Group
Minutes of Meeting #19
September 14-15, 1998
co-chairs: Craig Thompson
(OBJS) and Shel Sutton (MITRE)
OMG document internet/98-09-01
OMG Internet Platform SIG homepage: http://www.objs.com/isig/home.htm
Pranab Baruah - Boeing - firstname.lastname@example.org
Greg Berghorn - Raytheon - email@example.com
Michael Bigrigg - Carnegie Mellon University - firstname.lastname@example.org
Larry Bugbee - Boeing - email@example.com
Alan Burger - Siemens Nixdorf Informationssysteme - firstname.lastname@example.org
Roger Burkhart - Deere & Company - email@example.com
Dan Chang - International Business Machines - firstname.lastname@example.org
Martin Chapman - IONA - email@example.com
Thomas Culpepper - 3M - firstname.lastname@example.org
Andrew Eisenberg - Sybase - email@example.com
Nils Fischbeck - Humboldt-Universitaet - firstname.lastname@example.org
Joseph Fontaine - Software Productivity Consortium - email@example.com
David Gamble - Micro Focus - firstname.lastname@example.org
Ronald Graves - Statistics Canada - email@example.com
Lothar Grimm - Deutsche Telekom AG - firstname.lastname@example.org
Bob Hodges - International SEMATECH - email@example.com
Ed Hong - VISA International - firstname.lastname@example.org
Hide Horiuchi - AT&T Wireless Services - email@example.com
PolarHumenn - Syracuse University - firstname.lastname@example.org
Yuri Ismailov - Ericsson - email@example.com
Juggy Jagannathan - CareFlow/Net - firstname.lastname@example.org
Bill Janssen - Xerox - email@example.com
Gene Jarboe - Promia - firstname.lastname@example.org gjarbo
George Karabatis - Bellcore - email@example.com
Marc Laukien - Object-Oriented Concepts - firstname.lastname@example.org
Guangtian Liu - Southwestern Bell - email@example.com
Henry Lowe - Object Management Group - firstname.lastname@example.org
Frank Manola - Object Services and Consulting - email@example.com
Kathleen Maryman - NationsBank - firstname.lastname@example.org
David Mattox - MITRE - email@example.com
Henrik Nielsen - W3 Consortium - firstname.lastname@example.org
Karl Popp - SAP AG - email@example.com
Simon Raik-Allen - Viant - firstname.lastname@example.org
Bapa Rao - TIS Labs@Network Associates - email@example.com
Daniel Riscalla - Information Management Associates - firstname.lastname@example.org
William Robinson - GTE Laboratories - email@example.com
Henry Rothkopf - MITRE/Open Systems Center - firstname.lastname@example.org
William Ruh - Concept 5 Technologies - email@example.com
Harold Solbrig - 3M - firstname.lastname@example.org
Larry Streepy - HIE - email@example.com
Yasuaki Sugino - Fujitsu - firstname.lastname@example.org
Shel Sutton - MITRE/Open Systems Center - email@example.com
Craig Thompson - Object Services and Consulting (OBJS)
- firstname.lastname@example.org - co-chair
Jim Trezzo - Oracle - email@example.com
Peter Trimmel - PSE/Siemens AG - firstname.lastname@example.org
Amjad Umar - Bellcore - email@example.com
Leo Uzcategui - International Business Machines - firstname.lastname@example.org
John Weiler - The OTG - email@example.com
Ron Zahavi - Concept 5 Technologies - firstname.lastname@example.org
Jan van der Bijl - KPN Telecom - email@example.com
set up email list for firstname.lastname@example.org, email@example.com, and firstname.lastname@example.org. (FYI
There is already an email list email@example.com -- get on it by sending email
The presentations are a more reliable source of information than the
notes below. The notes are included for convenience.
The HTTP-NG Next Generation Architecture, Bill
Janssen, Xerox PARC
HTTP-NG is a W3C experimental project. Members
are Xerox, Microsoft, IBM, Netscape, others.
The motivation of the project was how do you deploy
apps (not just documents) to the WWW via a standard means. No guarantee
different ways will not conflict. Other problems: how to use HTTP as a
reliable datagram protocol due to tunneling, how to allow programs to talk
to each other, when there are multiple OOPLs, and congestion.
The approach is to redevelop HTTP on top of an
ORB-like object underpinning (the Xerox PARC ILU system).
The approach is two part: a web characterization
effort just broke out to become a separate W3C activity. And a protocol
design group, working on the HTTP-NG architecture, specs, interchange formats,
and prototype. The goal is a reference implementation and real evidence
of performance improvements, then possible standards.
Architecture shows Apps, App I/F Stubs, Messaging,
and Transports. All layers are isolated from each other.
Much more on architecture not covered in notes. Modular
architecture with means of adding in new modules. Predictable architecture.
App Layer: The Classic Web App (TCWA) - apps cannot
interfere with each other. Notion of operation inheritance so you can version
methods. The Web as an app: interface Webserver with methods get, head,
Interface between application layer and messaging
Type System vs. CORBA - varies from CORBA in many
ways (needed based on experience) though mostly simular. Fixed and floating
point cover CORBA fixed and floating but are more general. No char distinct
from string. ...
One performance test of page accesses with 42
embedded links, integers and larger objects, fetched page many times, data
should swamp headers. Using W3ng/webmux, much faster and smaller than IIOP
1.0, Pipelines 1.1, and HTTP 1.0.
HTTP-NG in the IETF since HTTP is in IETF domain.
Gave BOF at IETF in Chicago in August. Proposing to Transport Area. Not
currently headed for W3C standard. What about HTTP-NG and CORBA?
They currently have functional compatibility with OMG, Java, and the web,
with some omissions. They hope to reuse CORBA services on top of
http-ng substrate. They are just now thinking about a detailed OMG rendezvous
strategy and this was the topic of hallway discussions. Could mean
that either IDL might change by deltas or a mapping from IDL to NTTP-NG
- both upstream. Would want equal time for Java and DCOM. Want XML interface
definition to httpng. Another puzzle is services under the httpng hood
versus similar services in systems riding on top - or whether this is an
artifact of the current differences between the OMG and ILU IDL.
Q: QoS? Many IETF discussions on service
discovery. The HTTP-NG architecture could work with many QoS parameters.
HTML and XML Directions, Shel
Presentation is internet-98-09-03.
This presentation briefly described and made OMGers more aware of W3C standards
but we did not quite cover the "so what, how does
this affect OMG?" aspect.
HTML - HTML 4.0 is finished. Will be an XML app,
allows HTML in XML.
RDF - RDF adds some semantics to XML. Can use RDF
for security labels, ip rights, … its just a graph data model. It is encoded
VL and PGML are proposals for graph languages.
XSL - Define content once and display into many HTMLs.
One xml for each of cellular phone, TV, computer, … write once display
everywhere. (Correction to presentation is XSL not based on ECMAsript.)
Xlink - Xlink is how to point outside the resource
Xpointer - Xpointer is within the page
SMIL is synchronized multimedia language.
CDF is Microsoft's channel format.
OSD open sw description provides a way to distribute
software over the net. Signed document markup language.
IMS metadata spec.
XML Metadata Interchange,
Namespaces in XML.
Math Markup Language.
Open Financial Exchange.
XML EDI. Chemical Markup Language.
DNA sequencing markup language.
Mentioned Alan Doyle at GTE/BBN is looking at XML
Some Web Object Model Construction Technologies, Frank
Presentation is internet-98-09-04.
See related papers by Frank:
What web technologies might, when taken together, provide something akin
to an object model? Object models involve state, interface,
inheritance, …. OOPLs closely couple these but web object models might
not do so. The talk shows how various web technologies can meld into
a sort of web object model. These include: XML, Xlink, Xpointer,
XML Namespace, XML schema and data typing facilities, scriptlets in IE5
Beans. SGML community keeps semantics separated from structure. Style
sheets are one possible place to put the semantics. Talks about how to
build web on objects as in httpng and vice versa as in webBroker, XML-RPC,
Q: XML and HTTP data typing work. Will they be type consistent. Hope
so within W3C. No telling from W3C to OMG. Will need more coordination.
Interoperability Clearing House, John
Wieler, The OTG
John made a quick informal presentation on the need for a mapping from
business terms to technology. A sub problem is to describe component software
by its metadata descriptions. Another is to insure the components would
fit with each other and interoperate. A third is how the business problem
can be mapped to the technology standards. The need is for an open logical
repository of specs. The users have a way to vote. There is a facility
for validation. The OBJECTive Technology Group is working toward
this. They held a separate two day meeting at OMG Seattle.
internet-98-09-01.html Minutes of
Internet PSIG Meeting #19
Craig Thompson, OBJS
The HTTP Next Generation Architecture
Bill Janssen, Xerox PARC
HTML and XML Directions, Shel Sutton, MITRE
Some Web Object Model Construction Technologies, Frank Manola, OBJS
internet-98-09-05.html Minutes of Internet
PSIG WG Meeting on Computer Supported Cooperative Work
Craig Thompson, OBJS
Henry Rothkopf, MITRE
Henry Rothkopf, MITRE
internet-98-09-08.html Minutes of Internet
PSIG WG Meeting on Web/OMA Integration Architectures
Frank Manola, OBJS
internet-98-09-09.html Minutes of Internet
PSIG WG Meeting on Object Transfer and
Manipulation Facility (OTAM), Mike Bigrigg, CMU
internet-98-09-10.html Minutes of ECDTF/Internet
PSIG WG Meeting #1 on Agents
Craig Thompson, OBJS, and Steve McConnell, TBD
Foundation for Intelligent Physical Agents
Francis G. McCabe
OMG MASIF Specification
Stefan Covaci, GMD Fokus
Strawman Agent Reference Architecture
DARPA Control of Agent Based Systems Program, Craig Thompson, OBJS
Agent Research versus Agent Standardization
James Odell, TBD; Roger Burkhart, Deere and
Co and Santa Fe Institute
MIAMI, European Agent Project
Stefan Covaci, GMD Fokus
Jeffrey Bradshaw, Boeing