-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]
Sent: Thursday, October 07, 1999 10:46 AM
To: [EMAIL PROTECTED]
Subject: BOUNCE [EMAIL PROTECTED]: Non-member submission from
["Mike Fletcher" <[EMAIL PROTECTED]>]


>From [EMAIL PROTECTED]  Thu Oct  7 10:46:03 1999
Received: from vrtmail.robocity (vrtmail.vrtelecom.com [209.195.170.44])
        by ariadne.iz.net (8.8.7/8.8.7) with ESMTP id KAA25611
        for <[EMAIL PROTECTED]>; Thu, 7 Oct 1999 10:46:02 -0700 (PDT)
        (envelope-from [EMAIL PROTECTED])
Received: from cr706570a (cr706570-a.yec1.on.wave.home.com [24.112.128.169]) by 
vrtmail.robocity with SMTP (Microsoft Exchange
Internet Mail Service Version 5.5.2448.0)
        id 4DFA2B7W; Thu, 7 Oct 1999 13:36:50 -0400
From: "Mike Fletcher" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]>
Subject: Re: Persistence and server record models
Date: Thu, 7 Oct 1999 13:47:43 -0400
Message-ID: <[EMAIL PROTECTED]>
MIME-Version: 1.0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0)
X-Mimeole: Produced By Microsoft MimeOLE V5.00.2314.1300
In-Reply-To: <[EMAIL PROTECTED]>
Importance: Normal

>> From the (Fruits of/Core) Living Worlds camps...

Most systems use either a scanning or registration approach:

Registration:
        Living Worlds, Core Living Worlds and Fruits of Living Worlds use a root
node from which the software builds a "multi-user graph", which contains all
of the multi-user (and therefore all of the persistent and especially
serviced) properties of the vrml sceneGraph.  This multi-user graph
structure is communicated to the server (along with where to get the static
source) by the author's system at creation/adding time.
        In the abstract, any given node in the multi-user graph may have associated
with it a "description", which in most cases is the URI of the source code,
as well as some number of "data points" (states)[1]. To ensure that the
state of the sceneGraph is consistent with the stored state, it is necessary
only to send the "description" and values of all data points (states).
Because there is a single registrar (Living Worlds provide the concept of
multiple zones, but this has never been implemented to my knowledge), or a
set of known registrars, the structure is very easily resolved into record
IDs.[2]
        That is, the creator enters a set of records in the database which contains
the "description" and the list of "data points".  All others connecting to
the database retrieve both the description and the data points from the
database, applying the latter to the former. The content creator is
responsible for ensuring that the content will be able to handle non-ordered
update among different data points (i.e. State1 and State2 might come in in
any order, so you cannot use the one to initialize the other).  In practice,
this has not proved to be a significant limitation.

Scanning:
        The Fruits of Living Worlds allows for (but does not require) a more
opportunistic model, where the entire sceneGraph may be scanned for
potential "multi-user graphs", and any of the these may be registered with
the database.  The key is abstracting somewhat from the original Living
Worlds proposal to create a world-level record. All multi-user graphs which
are discovered within a loaded set of nodes are conceptually parented to
this world-level record (i.e. their root namespace is one (compound) record,
and they are stored as such (by implication, they must have unique names
within that namespace (not generally a problem, as they're normally authored
by a single entity))).

Enjoy yourselves,
Mike

[1] In Living Worlds and Core Living Worlds, this description is only
mentioned in relation to the SharedObject.
[2] My understanding is that this is Blaxxun's strategy as well, with their
BlaxxunZone nodes.

-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of
[EMAIL PROTECTED]
Sent: October 7, 1999 11:49 AM
To: Jeff Sonstein
Cc: steve guynup; [EMAIL PROTECTED]
Subject: RE: soon to be developing again


I would like to say yes, they are completely solved -- =)

More to reality - I worked through some of them when I built my test
system, which I am happy to share, there tend to still be some unresolved
issues, primarily in that the more that is done to solve persistency the
more integration need between the VRML scene file and the datbase records
need to exist.  The problem will be (and maybe it is not a problem) but I
see a need to build a **something** that can make sure that the VRML file
and the database record match -  maybe this is the user that is building
the world, or maybe it is in the software, I dunno...   THoughts ?

-A


Andrew Phelps
Department of Information Technology
Rochester Institute of Technology
http://www..it.rit.edu/
http://andysgi.rit.edu/



On Thu, 7 Oct 1999, Jeff Sonstein wrote:

> > If these issues of persistence are worked through [...]
>
> and I think they shall be...
>
> jeffs
>
> --
> Jeff Sonstein, M.A.     http://ariadne.iz.net/
>         http://ariadne.iz.net/~jeffs/jeffs.asc
> ==============================================
> there are no bugs
> there are just undocumented features
>
>

Reply via email to