Tampa, Florida
January 13-17, 1997
Note: This draft is Craig's notes, plusFrank's notes in blue.
Internet SIG continues to influence OMG's platform efforts providinga much needed longer term roadmap than OMG has had lately. Craig formeda Composition Architectures Working Group, which will meet in Austin, toassess why OMG is generally not building common facilities using objectservices, e.g., its architecture is less compositional than expected. TheC4I working group is making a little progress. We also made some progresson getting the Event-Condition-Action Rules RFP progressed. The first meetingof the Distributed Simulation working group was very interesting. Qualityof Service did not meet due to an ISO meeting on that subject at the sametime. Visigenics, ORB licensor to Netscape, approached us to do some workfor them -- we'll be told more in a week or so. Coordinatedpresentations were made of submissions to the Meta Object Facility RFP,Business Object Facility RFP, and Object Analysis and Design RFP. Thenext OMG meeting will be in Austin the week of March 10.
Craig Thompson led this one-day meeting of the Internet SIG, which attracted40 participants, and also a joint meeting with Common Facilities Task Forcewith around 50 participants. See Minutes.Recommendations from ISIG (mainly based on OBJS and MITRE RFI responses)were adopted by ORBOS and Common Facilities Tasks Forces and a WorkingGroup on Composable Architectures was formed, with Thompson todraft a white paper and others to review it on composition issues in OMGOMA.
This group met Sunday and Monday evening and is still getting startedthough there is a good core of interest. About 10-20 people attended. Responsesto the C4I Survey were reviewed the first evening. These responsesare on the OMG server.
Based on a vote, existing OMG specs this community feels it wouldneed for C4I are: DCE/CORBA, Secure IIOP, Messaging, Analysis and Design,DCOM/CORBA, IDL/Java, Internet, Meta-object, data interchange, electroniccommerce, interoperability, events, time.
Based on some discussion, future specifications this communitywill need are: OODB, shared data, C4I QoS Profile, Replication, CommonMessaging Facilities, Object Definition (Rules, Roles, relate), Simulation,Security facilities and admin, MLS, Service bridging/binding, geospatial,mobile computing/MIOP. We then defined the latter items. QoS and CommonMessaging ranked first, Simulation, MLS, IDL/Java, Geospatial, and OODBranked next.
This group did not meet at this OMG meeting since there is an importantISO meeting in Australia on QoS and principals attended that meeting. Thereseems to be growing interest in OMG in QoS. C4I, Telecom, Real-time andInternet groups all are interested.
Thompson, Stanley Su and Hermann Lam made copies of the draft Event-Condition-ActionRFP and ECA RulesWhite Paper available and lobbied for RFP acceptance at the upcomingAustin OMG meeting.
I only attended a little of ORBOS. Here is a summary. ORBOS had initialpresentations on these RFP response submissions: IDL to Java Mapping, Messaging,and SSL. They examined but did not issue new RFPs for Reverse Java Mapping[96-12-12], DCE/CORBA Interworking [96-12-01], and Persistent Object Service2.0 [96-12-07] (people are unhappy with the current Persistent Object Serviceand so there is a move to go to a round two). They had a presentation onenhancements to the OMG Security Concepts and Services. They extended dealdlinesfor Multiple Interfaces (to Feb 17), ORB Portability revised submissions(to Feb 17), Objects by Value (Apr 14), and Real-time RFI (Feb 14).
The meeting covered several topics:
Mobile Agents. After discussion with Sankar, I forwarded a motionto defer the deadline for the Mobile Agents RFP responses to the next OMGmeeting. The IBM submission is more polished but not something the ArchitectureBoard will approve; Sankar's submission with General Magic is less wellformed, since they attempted to work with IBM, unsuccessfully, and so donot have a completed specification they are ready to submit.
Workflow. There were two presentations: one from the WorkflowManagement Coalition (WfMC) describing their architecture and a claim thatthey may provide an IDL interface. Then there was an interesting presentationfrom Wolfgang Schulze from U Dresden that points out that WfMC architectureis not based on OMG services; their group actually has developeda componentized workflow system.
Name change. We considered adding "Internet" into thename or mission statement of Common Facilities Task Force, finally rejectinga name change, though addition to the mission statement may happen at theOMG meeting in Austin.
Relationship to Internet SIG and ORBOS. Suggestion by Bill Coxis to continue to meet jointly and use Internet SIG as a coordination bodythe looks out for long-term OMG directions.
Rules. I alerted Common Facilities that information on the RulesRFP and White Paper would be available at the next OMG meeting.
(TRC was just purchased by Perot Systems.) OT has not delivered on reuse,does not handle industry's rate of change, which is accelerating. So weneed to make sure our reuse starts at the model of the business throughreengineering down to code. We can build systems fast but change is a constantand most OT is driven from static requirements and a static view of theenvironment. Many errors are introduced in the front end. Further on lifecycle,the more expensive to repair. OT is not an issue of software construction,but effectiveness, do the right thing. KIF is an ISO standard now.
Cory Casanave (chair of the Business Object DTF)gave a presentation on the potential impact of the Business Object Facility(and the related MOF and OA&D work) on the OMG object model. Cory notedthat these areas of work involved the use of high level abstractions thatwere really object model concepts, and that some integration of these conceptsinto the object model would constitute an attempt to recapture the originalOO goal of turning design abstractions directly into implementations. Hethen presented an overview of their (Data Access') BOF submission as anexample. The submission proposes adding new concepts to the OMG Core ObjectModel to form a Business Object (BO) profile (much as concepts were addedto the Core to form CORBA and ODMG profiles). Their BO profile includesall of the CORBA profile, and some of the ODMG profile (specifically, relationships,keys, extents, and parameterized collections). Additional concepts includeassociations (a generalization of relationships based to some extent onthe ISO General Relationship Model), rules (including constraints and ECArules), roles (dynamic subtypes), states (nested and concurrent, takenfrom the Sematech CIM model), and events.
Shirdhar Iyengar (heading work on the Meta ObjectFacility) gave a presentation on the potential impact of the Meta ObjectFacility on the OMG object model. He described the role of the MOF withrespect to the OA&D and BOF work, and noted that the MOF was supposedto be used to create and hold metamodels in various domains (e.g., OA&Dschemas). He noted that the primary focus of the MOF right now was on theOA&D and BOF (due to the extent of the modeling issues raised, andthe timing), together with the OMG Object. In particular, they are usingthese as proofs of the adequacy of their ideas, by showing that the metalevelconcepts required to represent them can be captured in the MOF. Shirdharindicated potential impacts on the OMG Object Model in two areas: reflectionand new features. Reflection is an issue because the submissions generallyinvolve the ability to explicitly define object model features (metaconceptssuch as types, methods, attributes, inheritance) and (to varying degrees)their semantics. New features are an issue because (as Cory noted in hispresentation), the submissions involve new features such as roles, relationships,and constraints, that might need to become explicit parts of the OMG ObjectModel. Shirdhar said that the MOF activity is working on a comparison ofthe various metameta models involved in the various submissions, and thatthey wanted to also determine the implied metameta models of the CORBAand DCOM object models.
Frank spoke to Shirdhar after the meeting regardingthe ability of the MOF to deal with other types of metadata, with particularemphasis on the types of metadata we are investigating within the SSISSS.He expressed a great deal of interest in seeing the MOF tested against,e.g., Internet-related metadata, and Frank intends to follow this up inthe context of the Metadata and MDR tasks.
In addition, to these presentations, there wereto have been reports from separate working groups on Object Identity andthe Type System; these were cancelled due to the presenters needing tobe in other meetings. There was a report from a newly-formed Semanticsworking group, headed by Haim Kilov. These working group activities arebecoming so significant that future OMSC meetings will be structured asparallel working group meetings, followed by a plenary (there were over50 people in the room for this meeting).
Submissions to the Meta Object Facility (MOF)RFP, Business Object Facility (BOF) RFP, and Object Analysis and Design(OA&D) RFP were all due January 17 (the last day of this OMG meeting).Due to the potential interdependence between these RFPs, the chairs ofthe respective task forces decided to schedule pre-presentations of thesubmissions to these RFPs as a combined agenda, followed by separate meetingsof the respective task forces, and a joint evaluation and boundary group.(They are "pre-presentations" because the "official"presentations will be made in Austin, as the official deadline for submissionshad not yet expired). The interdependence between these RFPs arises becausethe MOF is intended to support interoperability across the entire applicationdevelopment life cycle, which includes OA&D as one phase (and the MOFmust specify the metametamodel that defines how OA&D metamodels willbe specified). The MOF is also intended to address standardizing othertypes of metamodels (e.g., relational DBMS schemas). At the same time,the BOF recognizes that OMG IDL may not be adequate to represent the semanticsof business objects, and that this additional semantic information maybe represented as additional metadata. The need to coordinate these activitieswas explicitly recognized in the RFPs, which requested information fromsubmitters about the appropriate relationship between these facilities.The evaluation and boundary group is intended to try to help address thecoordination problem, in conjunction with the submitters (the lines betweenthe facilities could not be too tightly drawn in advance, since, due tothe OMG technology adoption process, much of this depends on the individualsubmissions, and the integration of them that typically takes place followingthe initial submissions, and prior to any technology adoption).
The following is just a brief summary of the variouspresentations and sessions. All this MOF, BOF, and OA&D activity ispotentially relevant to the Metadata and MDR tasks of the SSISSS; hencewe will be looking further at these submissions.
Craig: Most provide more semantics than IDL; the specs are portmanteauand monolithic. It is not clear whether these specs depend on object services.Frank: There is explicit discussion of dependencyon object services (this is required to satisfy the RFPs). However, thisis a mixed bag. In many cases, dependencies are explicitly identified,but these are only to the simpler services (e.g., LifeCycle). In a numberof other cases, submissions use their own variants of an OMG service. Ingeneral, additional architecting of the submissions would be required todefine them as a composition of current OMG services.
Responses to the BOF RFP could propose technologyfor either (or both) the Business Object Facility (BOF) or one or moreCommon Business Objects (CBOs). CBOs are "objects representing thosebusiness semantics that can be shown to be common across most businesses",while the BOF is "the infrastructure (application architecture, services,etc.) required to support business objects operating as cooperative applicationcomponents in a distributed object environment" (quoting in both casesfrom the RFP). The presentations were separated depending on whether theywere addressing the BOF or CBOs (only IBM proposed both).
The SSA BOF submission is based on a fullyintegrated business object architecture, using the concept of an "ExecutableObject" (XO; basically a large-grained component built from PL-levelobjects). An XO has a single entry point, and is invoked by the BOF whenthere is a message for it. XOs exchange messages and information in theform of "Semantic Data Objects" (SDOs), which are basically setsof attribute/value pairs. The work is based on an in-progress SSA productcalled BPCS, and ESPRIT's OBOE project (work done with Prism).
The Data Access/SEMATECH/Prism/Iona BOFsubmission is based on a fully integrated business object architecture,and is based on Data Access' "DataFlex" products and the SEMATECHCIM Framework. The proposal includes a separate Component Definition Language(based on OMG IDL, and ODMG ODL and OQL, plus extensions) for definingBOs, and a BOF object model with its own type hierarchy. The submissionincludes a reflective metaobject system that directly represents both BOFand CDL concepts. The specification describes how CDL maps to CORBA IDL,and uses the transaction, query, event, and collections services.
The IBM BOF submission appeared to be moreof a framework in the conventional OOP sense (a set of object classes,as opposed to being based on an integrated product). The proposal did notattempt to provide the type of fully-integrated architecture adopted bysome of the other submissions, arguing that otherwise it would be difficultto integrate legacy software (which they claimed could not realisticallybe pried away from legacy data, since the legacy software generally incorporatedtoo much of the data semantics). The framework also included "location","home", "server", "container", and "handle"concepts for (abstractly) representing the physical characteristics ofthe system and deploying objects within it.
The EDS BOF submission also appears tobe a framework in the conventional OOP sense; the technology base is aset of C++ classes. The submission uses the OMG transaction, security,naming, query, and lifecycle services (in some cases with extensions);it doesn't use the relationship, collection, event, persistence, or concurrencyservices. It also proposes an ORB extension.
The Technical Resource Connection BOF submissionbasically proposed the use of ontologies and logic-based representations(e.g., KIF) as tools to define common semantics for business concepts.Their argument was that trying to get a 1-1 mapping between common businessobjects and software components was expecting too much. A demonstrationof Stanford's Ontolingua was used as an example. They have a tool thatconverts Ontolingua definitions to CDIF via Prolog.
The IBM CBOs submisson proposed a numberof objects from its "San Francisco" project. The specific objectsproposed were Currency, Involved Party, Address, Datetime, Description,and Decimal Structure.
The NIIIP CBOs submission proposed whatit calls "Task and Session" objects that represent organizations,people, places, resources, and processes. Specifically, it proposed thefollowing object types: Group, User, Role, Workspace, Task, and ResourceProxyobjects.
Overall, the submissions exhibited a wide rangeof approaches to (supposedly) deal with the problems addressed. The fullyintegrated approaches seemed generally to be higher-level abstractionsof standard OOP stuff, with some simplification, and a lot of things pre-defined.While it is plausible that such approaches might provide betterplug and play, it is not clear that it necessarily will. In thecase of CBOs of the type proposed by IBM, it's not clear how much has reallybeen gained by adopting this primitive level of CBO. [It also appearedthat OMG must have been running a "most illegible viewgraph"contest in conjunction with these presentations].
The Data Access MOF submission describesa fully reflective MOF that is completely integrated with their BOF proposal.Metaobjects in the MOF represent aspects of their BOF components, and arethemselves components. An attempt was made to address issues of how tointegrate their MOF with other OMG metadata-related components, such asthe Interface Repository, Name Service, and Trader. (It appeared to methat some of the scaling issues in world-wide distributed object systems(such as the Internet) might not have been totally thought out.)
The DSTC MOF submission described the needto support multiple metaobject type systems and their definitions, as wellas relationships between them (e.g., relationships between types in differentschemas that used different type systems). Their type concept (as wellas other concepts in their proposal) were based on RM-ODP ideas.
The Gemstone/Objectivity MOF submissionwas based on ODMG technology, and proposed to build the OMG MOF on newODMG metaobject facilities being defined as part of the ODMG 2.0 specifications.
The Unisys/Oracle/IBM/ICL/SSA/MCI SystemhouseMOF (JMOF) submission provides what appears to be a fairly reasonableand straightforward meta-metamodel with IDL definitions for each (meta)type(around 25 of them), and corresponding factory interfaces and a few utilitytypes. Their meta-metamodel is a superset of UML, and the submitters havebeen collaborating with both the Rational, et.al. and the IBM/ObjectTimeOA&D submissions to align them (the submission is also aligned withCDIF).
The IBM/ObjecTime OA&D submission proposes:(a) a Core Meta-Model (CMM) which can be used to build both general purposeand domain specific modeling languages; (b) an Object Constraint Language(OCL) usedto specify formal semantics for the meta-model. This ensuresthat the modeling concepts are formal enough to be interoperable and automatedby software development tools; (c) a Model Schemes mechanism to supportdifferent modeling languages as formal extensions to the Core Meta-Model;(d) a foundation for patterns, frameworks and component-based developmentbased on composite model elements. The submission emphases its formal foundations,which enable it to represent the semantics of the various models and meta-modelsit captures. Examples of how to represent Smalltalk, C++, Java, UML, ROOM,and Catalysis concepts are provided.
The Rational/Microsoft/HP/Oracle/TI/MCI Systemhouse/Unisys/ICON/Intellicorp OA&D submission was presented in two parts, a formalpre-presentation, and a longer evening session hosted by Rational and theother submitters of this proposal. Booch,Jacobsen, and Rumbaugh gave a description of their Unified Modeling Language.There is both a linear language and a mapping to a 2D icon language. Theynow have a collection of 83 classes that define their meta model and makeA LOT of distinctions. They claim it is stable. It is definitely complex.If one uses all this, it is not clear to me one could program in an OOPLwhere this is mapped to. So I suspect they will need an OOPL to go alongside.Also, I suspect one can come up with another 100 sorts of semantics thatcan make it more complex. That is, there is no notion of completeness.Still, I suspect this will be a highly respected spec and we will haveto pay attention to it if we call ourselves OO guys. EG, when people ask,where is your OA&D design of your system, oh DARPA researcher, we willbe embarrassed if we say we have no rational answer (since the masses havebought into OA&D).
The Platinum OA&D was presented ata separate breakfast meeting sponsored by Platinum.) Platinum is proposinga core meta-model and extensibility features. Their meta-metamodel is identicalwith the CDIF meta-metamodel (except that CDIF's MetaEntity is changedto MetaClass). Their metamodel consists of8 independent metamodels (calledsubject areas) that are integrated by sharing common metaobjects and inheritingcommon supertypes. the subject areas are: foundation, common, object model(describes the static class structure of an OO system), component model(describes a component-oriented view of a system), dynamic model (describesthe dynamic properties of classes using state-transition diagrams), usagemodel (describes use cases), architecture model (describes implementationconcepts such as processor, process, thread, and network), physical databasemodel (describes mappings of concepts from other areas to database storageconcepts such as tables).
Detailed notes weren't taken on either the SofteamOA&D submission or the Taskon/Reich/Humans and Technology/DataAccess OA&D submission.
Based on discussion with some of the participants,there seems to be considerable potential for integrating aspects of severalof the submissions, for example, integrating the formal base that is partof the IBM/ObjecTime submission with the UML metaconcepts (or is it metametaconcepts?). A number of the submitters use UML notation and concepts anyway,as it is a widely-used and respected approach which has already gone tosome lengths to integrate what are considered best practices in the variousOA&D methodologies. (The same comment applies to integration of generalaspects from the various OA&D, MOF, and BOF submissions. Some of theBOF and OA&D submissions indicated a willingness to address this, oncethe submitters had seen the other submissions).
This group met to address issues involving therelationship/overlap of the MOF, BOF, and OA&D submissions. A draftBoundary Guidelines document had been prepared (OMG Document BOM/96-12-02),documenting discussion at previous meetings (Hyannis and Nice), the currentdocument being draft .3. The document contains an identification of somepotential discrepancies between the terms used in the three RFPs togetherwith the start of a common glossary, and a set of boundary questions andissues that should be dealt with. During the meeting, some other "boundaries"were noted, e.g., with vertical domain task forces, such as Finance, andwith the new Persistent Object Service (which has a need to store metadata).This group will not be dealing with the boundaries between CBOs and verticaldomains, just the MOF, BOF, and OA&D boundaries. The group decidedthat the first priority would be to determine what (of the boundary issues)was to be within the purview of the MOF first, as this would simplify resolutionof the remaining issues. It was also decided that there would be two kindsof presentations (by submitters) at the next meeting: one to the individualRFP group, and another specifically on boundary issues (the latter willbe a plenary meeting of all groups, the individual RFP group meetings willbe allowed to overlap). There was a short discussion of whether the fourlayer metamodel architecture (user objects, model, metamodel, and meta-metamodel)was adequate, or whether there was even a need for the meta-metamodel layer(Jim Rumbaugh argued that the meta-metamodel layer was unnecessary, asa good model could describe itself, and some method of bootstrapping themeta-level was needed instead). Cris Kobryn (MCI Systemhouse, one of theparticipants in the Rational/UML submission) noted that an additional sourcefor common glossary material was the Rational UML glossary (which alsoincludes terms from the OMA and the OA&D RFP) at http://www.rational.com/uml/1.0/glossary.html.Cris noted that the MOF glossary is a superset of a subset of this glossary(a subset of this glossary is also used in the Unisys MOF submission).It was also noted that both the Rational OA&D submission and the UnisysMOF submission contain a matrix mapping the meta-metaobjects of UML, CDIF,and the MOF to each other (this matrix is also available at the Rationaland Unisys Web sites). The chair of this group is open; this will be discussedfurther at the Austin meeting.
This group, which consisted almost entirely ofsubmitters, met to discuss how evaluation of the BOF submissions wouldproceed (Frank narrowly escaped, by firmly declining, being named chairmanof the evaluation activity, as being only one of two non-submitters present).A draft BOF Evaluation Guidelines document had been prepared (OMG DocumentBOM/96-12-01) and revised over several OMG meetings (the current documentis the fourth revision). The document summarizes the requirements fromthe RFP, and records discussions from previous meetings on evaluation criteria,and contains an example model/scenario, consisting of a simple invoicingapplication, together with additional discussion on what features shouldbe demonstrated by scenarios. During the meeting, it was decided that submitterscould provide several variants on the general scenario, to highlight particularcharacteristics of their submission. It was also decided that submittersshould write a comparison of the strengths and weaknesses of their submissionversus the others.
Fred Koul (MITRE) led this meeting-around 15 people attended. The firstdiscussion covered background of the participants and their range of simulationinterests: C4I, mission rehearsal, training, business war gaming, decisionsupport, computing and distributed systems simulation, manufacturing simulation,engineering simulation, human factors simulation, virtual space, virtualoffice, MUD/MOO collaboration, planning, distributed data gathering, others.Two other perspectives were: cycle sucking, scale-changes from massivesimulations with billions of active elements.
Dimensions of the problem area are: models and simulation, variationsin model of time (real-time simulation, synchronization with real-worldevents, faster than real-time), whether there is an integration of realinteractions and virtual systems as in C4I systems (interactive), whetherthe simulation is recreating an environment (virtual reality, virtual co-location,virtual space) or for analysis purposes to answer questions, extensiblesimulations where sub-simulations can join or leave, task mixes can change,behavioral aspects, active autonomous agents in the simulation.
There is a relationship to the Internet SIG, Real-time SIG, QoS WG,as well as ORBOS and Common Facilities, and also, the area of planning,not covered by OMG currently.
We reviewed the mission statement. I made a number of recommendationsand it will be revised overnight preparing for a vote in the Domain TCto form a SIG. One issue arose: should this be a domain or platform SIG.
HLA. Fred gave a briefing on HLA (High Level Architecture) andRTI (Real-time Infrastructure). HLA is a DoD mandated federation architecturefor distributed simulation (as of Aug96). See http://www.dmso.mil/projects/hla/,Dr. Weatherly, weather@mitre.org, Tom Stark at DMSO tstark@msis.dmso.mil.The Familiarization Version will be available in Java by summer 1997. HLAsubsumes ALSP (Aggregate Land Simulation Protocol) and DIS (DistriibutedInteractive Simulations), the latter providing standards only for the bitlevel with no place to add some higher level simulation objects. They areworking with CMU's David Garlan, software architecture guru. They viewan architecture as providing building codes and ordinances and street planning.HLA has a structural object model, it has four plug in (parameterized timemodels), its thesis is, there is no single monolithic simulation but rathera federation of simulations, rules are meta amounting to requirements HLAmust meet.
Capt.Hollenbach, Director of DMSO, presentation.jwh@dmso.mil. Motivations: information dominance, virtual training,with less money, do more, operations other than war. Modeling and simulationis a way for DoD to do business. History: limited scope simulations andlittle interoperability in 1988; SIMNET in 1988; DIS standards in 1990;ALSP for federation in 1992; HLA in 1996. DoD hierarchy: William Perry(Secretary of Defense), J White, Paul Kaminsky (Under Secretary of Defense),Anita Jones (Defense Research and Engineering), Hollenbach (DMSO). DARPAunder Anita Jones too.
HLA architecture "facilitates" but does not guarantee interoperability.$20-30M/year effort. Support Utilities/Simulations/Interface to live playersall talk to Run-time infrastructure (RFI). All the modeling happens abovethe RTI. Object model template. Only one federate can own/change an attributeof an object at a time. Kinds of object models: simulation object model(SOM) and federation object model (contract between n simulations), bothdocumented via object model template. Mapping from HLA to DMSO terms: systemmanagement, events, lifecycle, security, time, and load balancing. NowDoD standard since no pay after FY99 for non-HLA, and retire non-HLA inFY01. Stick and carrot. Want this to be an industry standard (for Hollywoodtoo). Corporate decisions on standards. Prototype one: MITRE and MIT LincolnLabs using Orbix; not mandating specific RTI implementation; but developingshareware version. Familiarization F0 is give-away to DoD contractors foruse, not make money.F1.0 in Java this summer, International approval there,give away approval under way. Competitive procurement also underway. ArchitectureLeadership Group is a guidance committee.
Q from me: HLA federates simulation; how does it accommodatethe many object models in manufacturing to C4I; does it provide interfacesto GIS, C4I; how is it related to planning? How do you share world modelsacross these boundaries? Ans: There is much more work to be donein this area. The HLA interfaces partition off the rest of applicationslike C4I from the simulation service. The line between simulation and planningis fuzzy. Most GIS data is pre-loaded and incremental change made at simulationnodes.
www.itsi.disa.mil / standards activities / ismc / papers and presentations/ - contains Shel's current overall CIIF reference architecture. Kurt Buehler(Open GIS) attended. Shel Sutton showed an impressive video, "Imagingfor the Next Millennium," of command and control handling two crises,one on mythical Islandia. Then Shel talked about the range of earth imagingstandards at various levels. Thompson suggested an Architecture/InteroperabilityWorking Group.
As tight set of five designers are working on new protocols to possiblyreplace the HTTP protocol. Bill Janssen (Xerox PARC) gave a presentation.The team also includes: Bill Gettys (W3C), Paul Leech (Microsoft), McScenter?(chair of HTTP-WG at IETF), Spritzer (?). They are working on two things:
These summarized activities and actions of all committees at the meeting.Half the architecture board slots had come open and new folks were votedin by ballot.
Among the many OMG activities of potential interest,a couple to note:
CORBAmed is issuing an RFP on Lexicon Query Services(supporting access to heterogeneous controlled vocabulary services). Thisis of potential relevance to our SSISSS query and metadata tasks. The RFP(OMG Document corbamed/97-01-04) identifies potential overlaps with theMOF, BOF, and OA&D facilities.
The Document Management PSIG is currently workingprimarily on Revision Control facilities for compound documents. It isalso working on a view facility for such documents (different users seedifferent views of the same document and/or its components).