Re: [unrev-II] update on jpOHS

From: Jack Park (jackpark@verticalnet.com)
Date: Mon Mar 26 2001 - 15:00:40 PST

  • Next message: Eric Armstrong: "Re: [unrev-II] update on jpOHS"

    Eric,
    Like the 1950's GE ads, "progress is our most important product!"

    There has been a bunch of learning experience that I have been hinting at
    lately. Firstly, I am no longer passing everything through jLinda4 -- my
    simple java implementation of a Linda server -- Linda being the name of an
    invention of David Gelernter and his students at Yale in 1989, a kind of
    blackboard that allows asynchronous communication between heterogeneous
    processes. He called it Tuple Spaces, and wrote a book _Mirror Worlds_. His
    arguments are compelling, so much so that I implemented a version just to
    satisfy my need to play with it!

    In jpOHS, which, by the way, is content generation/management centric, as
    opposed to the navigation centric aspect of HyperScope, there exist two
    classes of processes: those that are intended to fill some display
    function -- which are typically blocking, synchronous, and those that are
    really background activities, like, say, storing a new occurrence. Because
    of all this, it is clear to me that some type of DDOM as suggested by Lee
    Iverson is going to be very important. So will a good grove engine.

    I should point out here that jpOHS is building on the following thesis: XTM,
    the XML Topic Maps markup language should be adequate for serving all
    messaging and persistence activities.

    To this end, each "tab" on the GUI, in one way or another, builds elements
    of topic maps. No more, no less. In IBIS, you don't use that vernacular.
    Instead, you talk in terms of issues, claims, and arguments. But, those map
    directly to topic maps, one topic map for each issue. I suppose the thesis
    will break eventually, but I haven't found its limit yet.

    Because of the need to handle both synchronous and asynchronous
    communication, I have begun to implement a SOAP-like message
    passing/interpreting system. SOAP calls for remote method invocation --
    RMI. I'm not presently using that. Rather, each XML message is parsed (at
    each end) and the appropriate method is selected by an interpreter.
    Exploration here involves the difference between interpreting and using rmi.
    Hope to come up with some metrics by which the tradeoffs (scalability,
    speed) can be assessed. I discuss this a bit more in the latest version of
    http://www.thinkalong.com/ohs/jpOHS.pdf

    And, I think it appropriate to stress the idea that we need lots more
    experimental implementations, some of which should be instances of
    HyperScopes.

    Jack

    From: Eric Armstrong <eric.armstrong@eng.sun.com>

    > Jack Park wrote:
    > >
    > > http://www.thinkalong.com/ohs/jpOHS.pdf
    > > <http://www.thinkalong.com/ohs/jpOHS.pdf>
    > >
    > > The project continues. The spirit of the project rests on the notion
    > > that
    > > it is time for many to build variants of an OHS so that we will have
    > > lots to
    > > compare and discuss once a "real" OHS gets off the ground.
    > >
    > Wow... Impressive.
    >
    > > I have combined Content and Knowledge into what I call the DKR:
    > > dynamic because everything (in my version) is represented
    > > internally as topic maps...
    > >
    > Sounds good.
    >
    > > InContext, there exists a Topic Maps tab,
    > > an IBIS
    > > tab, and an SDS tab. Later, there will be TSC, Augment, etc, tabs.
    > >
    > This is really intriguing.
    >
    > > The
    > > common interlingua between Context and DKR is XML messages, with the
    > > "cargo"
    > > written in XTM format. Context, called client, talks to a DKR server,
    > > which
    > > couples client to jLinda4, a Linda-like coordination system. From
    > > there,
    > > DKR, which is also a jLinda4 client, process persistence and knowledge
    > > issues. By coordinating with jLinda4, the system can accept many
    > > clients,
    > > and many DKR servers. In some sense, jLinda4 becomes a kind of Von
    > > Neumann
    > > bottleneck. Film at 11...on that one:-)
    > >
    > Looking forward to that film.
    >
    > > Progress to date: got client and server to talk to jLinda4. Client
    > > and
    > > server (prior to addition of jLinda4) already could manipulate topic
    > > maps:
    > > display them, create and edit them, and delete them from the
    > > database. Now,
    > > to get the tuples working!
    > >
    > What is JLinda4, again?
    >
    > and congrats on the progress!

    ============================================================================
    This message is intended only for the use of the Addressee(s) and may
    contain information that is PRIVILEGED and CONFIDENTIAL. If you are not
    the intended recipient, dissemination of this communication is prohibited.
    If you have received this communication in error, please erase all copies
    of the message and its attachments and notify postmaster@verticalnet.com
    immediately.
    ============================================================================

    ------------------------ Yahoo! Groups Sponsor ---------------------~-~>
    Do you have 128-bit SSL encryption server security?
    Get VeriSign's FREE Guide, "Securing Your
    Web Site for Business." Get it now!
    http://us.click.yahoo.com/2cW4jC/c.WCAA/bT0EAA/IaAVlB/TM
    ---------------------------------------------------------------------_->

    Community email addresses:
      Post message: unrev-II@onelist.com
      Subscribe: unrev-II-subscribe@onelist.com
      Unsubscribe: unrev-II-unsubscribe@onelist.com
      List owner: unrev-II-owner@onelist.com

    Shortcut URL to this page:
      http://www.onelist.com/community/unrev-II

    Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/



    This archive was generated by hypermail 2b29 : Mon Mar 26 2001 - 15:27:07 PST