Orlando, Florida
June 7-12, 1998
OMG made further progress in its work on a number of key technologies.XML and related Web technologies (e.g., RDF) are gradually making theirway into a number of OMG activities, making the integration of these technologiesincreasingly important. I co-chaired the Internet SIG meeting, and helpedestablish a Web/OMG Integration Architectures Working Group. I also attendeda number of meetings of OMG groups who are considering the use of XML,and noted the existence of our new Working Group to these groups as a sourceof possible collaboration. I also contributed to the Object and ReferenceModel Subcommittee's Extension Green Paper (on extending the Object ManagementArchitecture).
Outline:
Shel Sutton and I co-chaired this one-day meeting -- see separatemeeting minutes. We heard presentations on Web Compatible CORBA Metadata(using XML) from Ron Zahavi and Robbie Tischer (Concept 5 Technologies),an update on the HTTP-NG effort (Mike Spreitzer, Xerox PARC), and ComponentTechnology for Dynamic CSCW Environments (Eric Hughes, MITRE). We alsoheard brief presentations from John Weiler (OMG's Government Liaison) onthe Interoperability ClearingHouse, and from Juggy Jagannathan (CareFlowNet) on their XML|IT toolkit(which automatically generates XML-tagged CORBA service calls, and providesutilities to convert Java structures to and from XML).
The ISIG then formed three Working Groups:
The first two of these working groups had breakout meetings during theafternoon. I chaired the Web/OMA Integration WG breakout. During this breakout,the WG defined as its general goals:
These are becoming very important, given the increasing number of systemsbeing built which incorporate both CORBA-based distributed object technologyand Web technology, and the increasing frequency with which Web technologiessuch as XML are being referenced in OMG Task Force activities (some ofthe other activities reported below illustrate this trend). I will be producinga draft initial charter, and distributing it via email to get discussionstarted prior to the Helsinki OMG meeting in July (I also intend to pointpeople to our Web Object Modelreport, if they haven't seen it, as an introduction to some of therelevant technologies).
Following the breakouts, the first two WG's exchanged cross-briefings.Henry Rothkopf indicated that the goal of his WG was to examine and reporton opportunities for fulfilling emerging requirements for CSCW within distributed,component-based enterprises. The group intends to work on a Green Paper,and cooperate in depth with the Web/OMA Integration WG.
Mike Bigrigg then led a working session on OTAM. This is aimed at definingan OMG facility, based loosely on the ISO FTAM specifications, for a formof virtual file system that allows access to, and maintenance of, virtuallyany type of data. Mike circulated both a draftOTAM Green Paper and a draftRFI. There was considerable discussion of the need for the proposedfacility, how it related to existing OMG specifications, scenarios in whichit would be used, etc. I had written up a list of questions/comments onthe Green Paper, which I agreed to send to the entire group via email followingthe meeting.
The ORMSC (a subcommittee commissioned by the OMG Architecture Board)held a working meeting Sunday and Monday, and a plenary on Thursday. Duringthe working meeting, the group produced a new revision of the ReferenceModel Extension Green Paper (ormsc/98-06-01). This paper motivates workon a revision of the OMG's basic Object Management Architecture, and describessome of the changes needed. The paper contains a significant amount ofmaterial on architectural requirements from the paper "The Next GenerationObject Management Architecture", by Craig Thompson, Ted Linden, andBob Filman (ormsc/97-09-01). During the meeting, I contributed some additionaltext on the role of platform technology requirements in justifying theneed for OMA revisions. The group also developed a draft outline (to whichI contributed as well) for the OMA Revision document (the outline is documentormsc/98-06-02).
During the plenary session, the group heard a presentation from SteveMellor, Van-si Nguyen, and Dirk Epperson on "the need for precisespecification of processing in the context of UML" This presentation,which was also given to the Object Analysis and Design Task Force, wasan attempt to justify the need for an additional language (an action language)to specify behavioral semantics at the design (UML) level. A goal wouldbe the ability to automatically generate complete code from design-levelspecifications. The presenters said they weren't entirely sure what kindof language would be appropriate for this. While there is clearly a needfor some way to specify behavioral semantics in a high-level way, thereis a considerable amount of concern about, for example, whether choosingone language for this purpose might adversely affect the language-independenceof OMG-based systems. The plenary also heard a presentation by ToshiakiKurokawa on the conceptual object model developed by the Japan StandardsAssociation (JIS TR X0007, "Conceptual Model of Objects"). Thismodel is related to the set of object model features developed by the NCITSH7 committee in developing its Object Model Features Matrix (the two groupsinteracted during the early phases of that activity).
The plenary then took up the new revision of the Reference Model ExtensionGreen Paper. The original plan was for the plenary to approve the textof the paper, and send it to the Architecture Board for approval as a WhitePaper. Andrew Watson, the OMG Chief Architect, said that he had numerousdetailed changes which he would like made, and others also wanted somerevisions in the way the RM-ODP specifications were referred to in thepaper. As a result, it was decided to table consideration of the paperuntil the Helsinki meeting, with a 10-day deadline for submitting additionalcomments.
I only attended a small part of this meeting, due to conflicts withother TF meetings. During the part of the meeting I attended, ORBOS hearda presentation of the revised Firewall submission (orbos/98-05-04). Thesubmitters described products by Orbix (Wonderwall) and Visibroker (Gatekeeper)as proofs of concept of the ideas contained in the submission. A vote onthis submission is scheduled for the Helsinki meeting. The general issuesdealt with in this submission (providing an "object model" offirewalls, including proxies, operations/interfaces that a CORBA firewallshould support, etc.) could provide useful insight into describing otheraspects of the Internet/Web infrastructure in OMG terms.
ORBOS also heard presentations on:
and discussed:
The Messaging submission and Benchmark RFI were approved. Scripting,Persistent State Service, Tagged Data Facility, DCE/CORBA, Security RFPs,and Interoperable Naming are on the agenda for the Helsinki meeting.
I attended part of this meeting, which discussed the BOCA InteroperabilitySpecification (bom/98-05-03) and the CBO Revised Final Submission (bom/98-05-01).
The BOCA (Business Objects Component Architecture) InteroperabilitySpecification describes the interoperability requirements of design conceptsspecified by the BOCA metamodel described in other specifications. Thishad previously been approved by the BODTF, and was returned by the ArchitectureBoard for more work. The main technical issue appears to be the need toalign these specifications with whatever OMG decides to do about componentsin general (there have been several submissions to an OMG Component ModelRFP; revisions are scheduled for discussion at the Helsinki meeting, butit is not clear this schedule will be met). Since CORBA Components do notcurrently exist, it is not clear what changes need to be made to the BOCAspecification. A motion to recommend the BOCA specification to the DomainTask Force failed. A motion to extend the submission deadline to July 6passed, but it is not clear that any revisions will actually be submitted.Fred Cummins (EDS), one of the submitters, indicated that the submitterswould be interacting with the submitters to the Components RFP to try toaddress the alignment issues.
The CBO (Common Business Objects) Revised Final Submission containsthe specifications for a number of proposed common business objects. Ratherthan voting on the complete submission, independent votes were taken onseveral related groups of CBOs. The only set approved was the set of objectsdescribing Task and Session. It was unclear to me what this approval meansin practice, since the document now contains material on quite a numberof non-approved concepts; this may have been resolved later in the meeting.
Dave Zenie (NIIIP) also described the work of the Organizational Modelworking group. They are working toward an RFP to describe interfaces foraccessing organizational charts and related information (the current draftis bom/98-03-06). This would contain types such as Person, Business Unit,Group, etc., and their interrelationships. The RFP would also address requirementsfor different views of this information, and support for reorganizations.A vote on this RFP is scheduled for the Helsinki meeting.
During the meeting, I had a short talk with Liz Fong (NIST), who isnow working in the areas of metadata and Electronic Commerce. I mentionedour Web Object Model report as possibly being of interest to her.
I attended the OMG Chair's Dinner (in my role as acting ISIG co-chair).I spoke to Sridhar Iyengar, who leads MOF-related activities. He said Unisyshad mapped MOF to XML (as a wire format). I suggested he look at RDF asa specific metadata model (with a richer semantics), and told him a littlebit about it. In the MOF context, it would be necessary to look at a UML/RDFmapping, since the MOF metamodel is basically a UML subset. I mentionedthe new ISIG Web/OMA Integration Working Group, and suggested we coordinateon this and other subjects (this is one of many XML applications floatingaround at this OMG meeting). I also listened to a few people involved inBODTF activities express their frustration with the Architecture Board'sholding up approval of business-object-related specifications (my own view,unexpressed during these comments, is that the AB is simply doing its jobin attempting to ensure that OMG approve relatively coordinated specs whichalso reflect real products that have a reasonable chance in the marketplace,as opposed to letting OMG serve primarily as a marketing tool for specificvendors).
I attended part of this TF meeting primarily to hear some of the WorkingGroup status reports. The UML Revision Task Force (RTF) reported that theyhad a database containing 362 issues as of June 1, with about 130 not yetaddressed (primarily because they had been submitted very recently). Theyexpect to address these within the next few weeks. The MOF RTF reportedthat they had resolved 26 issues, 6 were still on the table (they didn'tidentify them), and they had recommendations for 3 of the 6. (The issueslists for these activities are on the OMG server).
The RTAD (Real Time Analysis and Design) Working Group is working onthe use and usability of UML for real-time, embedded, and fault-tolerantsystems. They indicated that these systems required numerous timing- andresource-related concepts beyond UML, but there was no consensus on thedetails of the concepts required. They particularly noted that schedulingwas a big issue in many of these systems, that there was a massive literatureon scheduling, but that UML was fuzzy on specifying enough informationto feed known scheduling techniques.
The (A&D) Process WG is looking at a possible RFP on a softwareprocess definition language and model(s). Jim Rumbaugh asked if OMG shouldreally be doing some of the things the TF is doing, noting that OMG isn'tsupposed to be doing things just because it can do them, and that OMG isnot the appropriate place to do research. (This same comment was also directedtoward the Mellor, Nguyen, Epperson presentation on behavioral modeling,which came later in the OA&D TF meeting).
Wednesday morning I attended part of the ADSS (Autonomous DecentralizedService System) SIG meeting. The group is working on environments in whichagents and brokers negotiate to provide services to customers (many ofthe ideas come out of "service management" concepts currentlybeing developed in the context of telecommunications networks). They arecurrently working on an RFP on "Inter-domain Service Brokerage"to be issued by the Telecomm DTF. This RFP defines (service) "requestor","provider", and "mediator" concepts, and the interfacesbetween them. I was actually there just as an observer, to get some feelfor the area in preparation for the joint SIMSIG/ADSS SIG meeting thatafternoon (see below) which was mainly to talk about metadata. However,I was able to make some comments which the group found helpful (based onmy RM-ODP experience and my telecom experience at GTE) regarding exactlyhow the mediator might have to differ from the Trading Service alreadydefined by OMG, and material the RFP should explictly include about whetherit required support for dynamic renegotiation of service agreements (based,e.g., on a provider not meeting quality of service guarantees).
That afternoon I attended a joint meeting of the Simulation and ADSSSIGS on "Managing Policy and Context Metadata in Simulations and AutonomousSystems", the basic idea being that metadata on policies, contextinformation, etc. is important in governing both interactions between federatesin a (federated) simulation, and negotiations between autonomous agentsin transactions of various types. There were three presentations:
Dr. Funabashi described a variety of target applications for ADSS, includinghome intermedia services, electronic commerce, and community informationsystems. He noted the need to have agent behavior in such systems accuratelyreflect the policies of their owners, and described a system to do policymanagement in these systems. Particular issues include:
This sounded like a good application for ECA rules, although the talkdid not describe particular representation techniques for the rules.
Dr. Kuhl works with the Defense Modeling Simulation Office (DMSO) ontheir HLA (High Level Architecture) for distributed component-based simulation.The HLA specifies a way for independently-developed simulations to worktogether. Part of the HLA is a metamodel (the Information Model) that allowsthese separate simulations ("federates") to interoperate. Hedescribed the objects used in their model, and a publish/subscribe mechanismthat helps integrate the relatively-independent federates. More informationis available at http://hla.dmso.mil,and in document mfg/98-05-02 on the OMG server (I have a copy of his overheads,which will presumably also be put on the OMG server).
Dr. Mack described the need to share compound documents (possibly derivedfrom multiple sources) between different organizations in a collaborativecrisis understanding application, and the use of XML and RDF metadata inassisting with this process. The idea is that each component of a documentwould have its own (embedded RDF) metdata. The explicit markup providedby XML and RDF would help in rationalizing the different vocabularies usedamong different organizations, and also allow different tools to operateon the documents. He also described the use of metadata in capturing interationsabout the documents during collaborations, and helping to maintain changesover time. He noted that embedding RDF in ACL or KQML-like messages allowedinformation in XML/RDF form to be directly moved into messages and instructionsexchanged between agents.
Following the presentations, some general discussion ensued. The generalquestion arose as to whether IDL interfaces that essentially define "datarepresentations" in various OMG specifications preclude the use ofXML in those applications. Andrew Watson was at the meeting. He indicatedthat he had been attending the W3C DOM working group meetings as part ofthe OMG/W3C liaison activity, noted the work on mapping UML/MOF to XML,and had some comments on the relationship between XML and a number of OMGactivities. I noted the potential of DOM (as well as more specialized interfacesgenerated for the purpose) to provide the basis of general interfaces betweenOMG activities and XML representations, said that, if necessary, XML representationscould easily be defined to correspond to the interfaces defined withinOMG specifications, and mentioned the existence of ISIG's Web/OMA IntegrationWG to help address these issues.
I attended the afternoon session of the one-day meeting of this SIG.Shel Sutton (MITRE), the chair, reported on several SIG activities. TheOGC (Open GIS Consortium) SimpleFeatures Specification represents an attempt to document "simplefeatures" the group can agree on. The group is still discussing thegeometries of more complex features. The intention is for the CORBAgisSIG to look at this document, discuss it via email, and be prepared toseriously work on it in Helsinki. (As reported in previous minues, OGChas "task force" status with OMG, since OMG has agreed not toset up its own GIS TF. The role of CORBAgis is basically to "profile",or "CORBAize", OGC specifications.)
Shel also discussed a Green Paper "Geospatial Information Transfer:An XML and RDF Based Approach" (no copies were distributed, but hedisplayed some text on overheads). He noted that the paper was based ona talk given to CORBAgis in Salt Lake City, and to the Internet SIG inManchester, by Henry Rothkopf, but that the original idea was providedby the Web Object Model talk I had given to the SIG at the New Brunswickmeeting. The goal is to get agreement by the SIG on the use of XML forexchanging GIS metadata (and presumably data as well). Shel described anumber of GIS data transfer formats to which this approach would apply,including:
Shel mentioned the goal of defining a single XML DTD (Document TypeDefinition) that would accomodate all these data formats. I suggested thata better starting point would be to define a DTD for each format individually,and then work the problem of mapping into a single DTD separately. Forexample, having a separate DTD for each format would provide a basis fortaking XML data (which might be generated in the XML format by new tools),validating it, and translating it back into the original format for useby legacy tools. There seemed to be a lot of support for this idea, butit remains to be seen how the actual work on the Green Paper will progress.Shel said that the paper will be distributed on the SIG's email list, andthat volunteers to work on it were welcome.
Shel also discussed a Green Paper on the relationship of CORBAgis toother GIS-related standards organizations. It was noted that there hadbeen some discussion in the Analysis and Design SIG group working on theMeta Object Facility about generating XML from MOF data and UML (e.g.,as an exchange format between CASE tools). It seemed to me that work, saywithin the Manufacturing SIG, on exchanging mechanical CAD designs andother product data might also be relevant here.
The NCITS (National Committee for Information Technology Standards)H7 committee met briefly on Tuesday afternoon. As noted in earlier reports,H7 is the U.S. group working on the Enterprise Language for specifyingthe Enterprise Viewpoint (EV) within the ISO Reference Model for Open DistributedProcessing (RM-ODP). RM-ODP is a general reference model to which OMG specificationsare supposed to conform (OMG RFPs generally contain an explicit requirementfor RM-ODP compliance; and CORBA IDL has been made an ISO standard forobject interface specification within RM-ODP).
At the meeting, H7 discussed several concepts (e.g., "delegation"and "artefact") which had been included in the draft EV specificationby other national bodies, and which some members felt were unnecessary.H7 also discussed the adequacy of the current set of concepts for describing"process-like" information (it was unclear that any new conceptswere needed). H7 wants to progress the current specification to CommitteeDraft status, and there will be an email vote on a U.S. contribution tothe next ISO meeting on the EV specifications.
The Domain Task Force plenary heard presentations from the various DomainTask Forces. It also approved the issuing of a number of documents including:
During the plenary I looked over a copy of a draft initial submission(by six companies, including 3M) to CORBAmed's Clinical Observations AccessService (COAS) RFP. COAS defines a mechanism for passing clinical observationsdata between information repositories and applications (the CIAS mentionedabove is intended to provide image-related access services to COAS). Thesubmission includes a section on "XML Usage" which, unfortunately,seems to contain a number of wrong ideas (e.g., it seems to suggest thatXML and IDL are alternative data interchange formats). I intend to emailthe submitters and suggest some corrections.
The Platform Task Force plenary heard presentations from the variousPlatform Task Forces. It was reported that the Java/IDL mapping revisionwas near approval. The OA&D TF reported that they might be consideringan RFP for a process (behavioral) specification language (mentioned above)at the Helsinki meeting, and that they had heard pre-presentations of submissionsfor a Stream Model Interchange Format (SMIF), one being based on CDIF andanother on XML. The Real Time SIG reported that they were working on anRFP for open nested transactions.
The complete list of OMG Technical Committee work in progress is athttp://www.omg.org/library/schedule.htm.