From furm@npac.syr.edu Tue Sep 22 17:00:52 1998 Date: Tue, 22 Sep 1998 16:07:21 -0400 From: Wojtek Furmanski To: shrideep@npac.syr.edu, timucin@npac.syr.edu, gcf@npac.syr.edu Cc: furm@npac.syr.edu Subject: sep 22 book meeting minutes Minutes from the book meeting Sept 22 1998 Lists/databases to collect/build: -------------------------------- List of Books we have to read/page (Wojtek) List of Articles, URLs (Hasan) List of References (Shrideep) List of Issues (Wojtek - starts with this email) Table of Contents (Fine Grain) ------------------------------ Assume 1/2 page for an atomic homogeneous snippet of text, and 2 pages for a self-contained 'story' (with about 4 snippets for say intro, body, summary, conclusions). 500 pages means 250 stories or 1000 snippets. Current plan/table of contents has ~30 items. To get down to the snippet level, we need to split each current item into some ~30 new subitems. First, however, we need to modify the current table of contents / coarse grain as below. First draft for coarse grain tomorrow, for fine grain Thursday. Fine grain TOC will take ~25 pages by itself; or 6 pages if we stop at the story level (reasonable for the printed TOC in the actual book). RCI 100 pages is 150 Wiley pages. Shrideep to combine subsets of RCI and previous book drafts towards some 180 Wiley pages by the end of Sept. Table of Contents (Coarse Grain) -------------------------------- Items 3,4,5 to be merged/repackaged with 4 Stakeholders at the top level, i.e. the rough plan is: Part I: Introduction (or Distributed Systems; or Web is Computer) Part II: Stakeholders 1. Web 2. Java 3. CORBA 4. COM Part III: Pragmatic Object Web Part IV: Future Each stakeholder item to be split as follows: Overview 101 Course Advanced Fetures Latest Developments Future Trends Total Solution Perspective POW Stakeholder Role Core Total Latest Our take ........................................................ COM Millenium COM+ Front-End, Desktop, DirectX (but also clustering etc.) CORBA OMA CORBA3, BEA Open impl in JWORB Java Framework Soup JINI JWORB base, JINI vs HLA WOM WebFlow (?) XML RPC, SOAP FlowML etc. Other edits/updates for the current Table of Contents: a) include WebHLA as item in 7. Proposed Integration Technologies b) rename 8. as More/All about POW (not JWORB) c) Include OMBuilder, Jager Donuts, CareWeb etc. in 9. Title ----- Building Distributed Computing on Pragmatic Object Web - The Best of Web, Java, CORBA and COM. Synergies/Integration Components: --------------------------------- Java <--> COM: COM+, Beans/ActiveX bridge COM <--> HLA: DirectPlay HLA <--> XML: attribute/message/event codec format XML <--> CORBA: DOM/IDL CORBA <--> Java: Object Web, JWORB Java <--> XML: DOM/Java, parsers, 'Give Java something to do' Java <--> HLA: Object Web RTI, Javelin CORBA <--> HLA: DMSO SIG in OMG, CorbaCap, Object Web RTI CORBA <--> COM: OMG Bridge XML <--> COM: XML file format (I/O) for all MS Office tools XML for WebFlow --------------- VML uses JPEG only as external entity, no mixing of Unicode and binary streams Suggestion: adopt/extend AVS file format with XML Unicode preamble, separated by new line from the binary body. WebFlow module is a general purpose XML->XML transform (or XML->something viewer/sink of something->XML source); includes the following computational paradigms: a) trivial computation (e.g. E-commerce, can be done by scripting) b) single node and/or medium perf computing (Java) c) LAN distributed and/or higher perf batch mode computing (CORBA) d) desktop viewers (COM) e) interactive distributed computing (HLA) Need to collect all existing XML applications / WebFlow data formats -------------------------------------------------------------------- OSD CDF SMIL ChemML MathML VML (HealthML?) what else? is there a site with XML apps? design/include/find good names for our stuff: FlowML? Field formats (check AVS, POOMA, what else) Learn more about and decided where to put Metadata -------------------------------------------------- MCF (Netscape) XML-Data (Microsoft) RDF (W3C) how about UML metamodel and meta-metamodel? Previous/old material attached below for completeness ----------------------------------------------------- Abstract: ---------------------------------------------------------------------------- We are experiencing an unprecedented revolution in computer hardware and software. This is driven by relatively low-end commodity applications and features powerful multimedia PC and workstation nodes, wide area networks and complex distributed systems of systems. The median configuration is only typical of the very heterogeneous reality with POTS (classic telephones) to gigabit networks and nodes that span entry level network computers to backend supercomputers. The underlying architecture has a compelling universality as it encompasses for instance business enterprise systems, home entertainment and information delivery and large scale distributed simulations. The hardware revolution has been accompanied by remarkable new software technologies whose capabilities and pace of innovation bring opportunities and challenges. Most remarkably, the World-Wide Web took the computer world by storm itself, but it also unleashed and exposed other, more sophisticated object/component technologies such as COM, CORBA or Java. In turn, the somewhat simplistic Web of mid '90s is now consolidating its latest technologies such as XML, RDF and DOMJ towards an agile and dynamic native framework that might be called Web Object Model (WOM). Other communities will likely join as players of this new global game of 'Object Web' - for example Defense Modeling and Simulation Office (DMSO) which develops object based High Level Architecture (HLA) with the mission of world-wide wargaming interoperability and with the natural commodity potential for the multiplayer online gaming markets. Object Web offers an explosive combination of object/component reusability with the global access - the result could be viewed as an emergent global componentware marketplace but there are still several obstacles on the way. As of mid '98, the field is still fragmented by several competing and rapidly evolving object/component technologies listed above and so keeping abrest of all these new opportunities is a non-trivial challenge for policy makers, users, designers and developers. Here is where this book might help. It is not yet another text on COM, CORBA, Java or the Web - there are many excellent books covering various aspects of these technologies. Instead, we bring here these four leading Object Web candidates together and we seek for their common and complementary aspects that would allow us to glimpse into the near term future and to predict some possible convergence patterns towards what we termed Pragmatic Object Web (POW). Today, we view POW as a practical recipe to proceed and succeed with our software plans, design and development in complex times of the ongoing global software wars. Hece, POW tries to combine the elegance of Java, completeness of CORBA, marketing muscles of COM and the universal deployment of the Web. Today we can only see a fragmented distorted view of this express train to software heaven - mererly a mosaic of emerging software technologies which we feel are capable to build global distributed systems more poweful than the sum of their components. Tomorrow, POW can indeed start converging towards a global system along some of the possible trajectories discussed in the book. We review here various plausible convergence scenarios and we propose our own - based on Java Web Object Request Broker (JWORB). JWORB is a multi-protocol Java network server, currently supporting HTTP, IIOP and DCE RPC (not yet by July 98..) protocols and hence capable to adapt and act as a Web, CORBA or COM server. We also illustrate how one could natually construct high level distributed object services such as DMSO HLA, metacomputing, commodity cluster or heterogeneous database management on top of self-managing adaptive multi-server multi-user meshes of JWORB nodes. We have designed this book to be read standalone by those interested in distributed systems. It can also be used in training or University curricula where the book provides a framework or architecture for a set of courses covering the individual technologies of Java, CORBA, COM or the evolving Web in great detail. >From furm@npac.syr.edu Tue Sep 22 15:28:01 1998 Date: Sat, 23 May 1998 11:27:20 -0400 From: Wojtek Furmanski To: timucin@npac.syr.edu, shrideep@npac.syr.edu, gcf@npac.syr.edu Cc: furm@npac.syr.edu Subject: book bullets->chapters: one more chapter Sorry - forgot 'Point Client-Server Solutions' which are included below: 1. Introduction 2. Broad Technology Approaches 3. Low Level Technologies (or should we replace Low Level -> Core?) 3.1 COM 3.2 CORBA/ORB part 3.3 Java 3.4 WOM 4. High Level Technologies 4.1 Microsoft Transaction Server (+Wolfpack+Falcon+..) 4.2 CORBA Services/Facilities i.e. OMA 4.3 Java RMI, IDL, Spaces, JSDT etc. 4.4 WebBroker (as per DataChannel Proposal) 5. Point Client-Server Solutions 5.1 Active Server Pages + ActiveX with IE4 front-end (FMS Training?) 5.2 CORBA enterprise + collaboratory examples (Shrideep) 5.3 TVR over JSDA plus something from HPJava/Grande domain? 5.4 Not clear (Check CommerceNet: they gave up on CORBA and do XML) 6. Integration of Approaches - General Discussion of Synergies etc. 7. Our Proposed Integration Technologies: 7.1 WebFlow - visual (language-independent) programming front-end 7.2 JWORB - multi-protocol (protocol-independent) middleware 7.3 HyperWorld - multi-vendor/storage.. data backend (OLEDB,PSS,RDF..) 8. More/All about JWORB 9. POW Integration Examples: 9.1 WebFlow via JWORB over Globus 9.2 WebFlow linked with UML and VBA/COM front-end (VisSimTools) 9.3 Object Web RTI 10. Future 10.1 Distributed Databases 10.2 Agents 10.3 What else >From gcf@npac.syr.edu Tue Sep 22 15:28:08 1998 Date: Sun, 24 May 1998 17:28:00 -0400 From: Geoffrey Fox To: Wojtek Furmanski Cc: timucin@npac.syr.edu, shrideep@npac.syr.edu, gcf@npac.syr.edu Subject: Re: book bullets->chapters > Some proposed title changes (the idea is to both define POW > and include some familiar keywords that might attract more/other readers): > > 1) Building Distributed Systems on Pragmatic Object Web that > integrates COM, CORBA, Java and WOM > > 2) Building Distributed Systems on Pragmatic Object Web: > Towards the Integration of COM, CORBA, Java and WOM. How about 3) Building Distributed Systems on Pragmatic Object Web: Using the best of the Web, Java, CORBA and COM (I don't think many will know WOM) > > Book parts as in my meeting notes: > > Intro > Broad Technology Approaches > Low Level Technologies > High Level Technologies > Point Client-Server Solutions > Integration of Approaches > JWORB > 3 Examples > The Future (includes Distributed Databases, Agents etc.) > > Possible Book Chapters (or just expanding bullets listed above): We should decide on order which should poehaps optimize "pedagogy" I suggest: 1. Web (Natural precursor of Java except WOM not as well known as Java) 2. Java (or interchange 1 and 2 as per above remark) 3. CORBA (as easier to understand than COM but not so real) 4. COM (the rubber hits the road ...) as standard for each of following chapters that discusses bits of each > > 1. Introduction > 2. Broad Technology Approaches > 3. Low Level Technologies (or should we replace Low Level -> Core?) Here we should review "other technologies" such as VRML (not used) and Web-linked databases (extensively needed) > 3.1 COM > 3.2 CORBA/ORB part > 3.3 Java > 3.4 WOM Reply by Geoffrey Fox gcf@npac.syr.edu, http://www.npac.syr.edu, Phone 3154432163(3154431723 npac central) Fax:3154434741