Re: [mmox] MMOX Progress tracking (Jon Watte)

[Charles Krinke <[e-mail]>]

I believe the place we are at is one of partial interop between some similar
virtual worlds. And I further believe that one of the goals is to extend
this interop either in its current form or some negotiated form to other
virtual worlds.

With respect to OGP, I see that as a partial solution currently in that an
although an avatar can teleport from a Linden grid to an OpenSim grid, there
is not yet a full handoff of the client's server connections. When we can
perceive of how an avatar from a Linden grid can perform a "full handoff"
from a Linden simulator to an OpenSim simulator *and* conversely, the same
or different avatar can perform a "full handoff" from an OpenSim simulator
to a Linden simlator, then we get to the next stage.

With respect to MXP, it is currently possible, to log onto an OpenSim
simulator with either a SecondLife compatible client or an MXP compatible
client as OpenSim has implemented the MXP logic from the MXP folks. This
needs further testing.

With respect to HyperGrid, an avatar can perform a full handoff between
grids using OpenSim and work is ongoing as folks creare more logic.

So, my vision of interop is that of an avatar from one virtual world
starting on his own, home, world or region and then
teleporting/transporting/moving/<whateverWeWishToCallI> to another virtual
world where the avatar can perform chat, examine objects, interact with
other avatars and clients in a different virtual world.

And we are part way there.

The trick is to define the nuances of the logins and interconnections
currently existing for additional virtual worlds so that we can all add
support in server and client software to allow as much interoperability as
possible.

To date, the work has concentrated on virtual worlds with very similar
architecture. SecondLife and OpenSim are similar virtual worlds, but let us
not make the mistake of trying to say they are the *same* virtual world.
They are not.

How we can perform a login from an OpenSim world to a Croquet or Wonderland
or other virtual world is one of the things I believe it is to our advantage
to understand as we try to move forward in increasing interoperability.

Charles Krinke
OpenSim Core Developer
OSGrid Director

On Thu, Mar 12, 2009 at 1:14 AM, zedmaster <[e-mail]> wrote:

> Jon makes an excellent point here; true interop implies that there is no
> single world owner/vendor, but suggests that any world is made up from
> contributions from multiple sources.  Everything in the mail trail so far
> seems to be heading towards isolated worlds (islands) served by a common
> server-client protocol, between which a user might be able to teleport, but
> this isnt really going to help build the sort of detailed worlds that i
> think people are wanting.  All you will be able to do is hop beteween
> experiences - a bit like the "old" web model of hoppping between sites.
> Mashups have changed all that - people are now building sites which deliver
> an aggregated experience i.e. using webservices etc. they can pull together
> diverse sets of data and present something new, something the original sites
> couldnt do alone.
>
> Imaging that I have access to some VR technology that can put immersive
> users (full motion tracking) into complex mechanical assemblies and let the
> user operate and interact with the items e.g. operate a back hoe on a JCB,
> or study the detailed assmembly instructions for an entire airliner.   I
> would like to deploy that technology to a shared world where others could
> come view my models; I wouldnt expect everyone to have access to the full
> body tracking, but i would like them to be able to walk around and touch and
> press buttons etc..
>
> True interop to me implies that I could run the simulator which "owns" my
> mechanical entity, and I could do this wouthout needing to know or care how
> others get to view this; it doesnt matter to my simulation, as long as I get
> 'standardised' inputs (telling me when someone touches something) and I get
> to send 'standard' outputs saying how the results of my simulation have
> changed.  I dont want to have to code my simulation in N different scripting
> languages for N different worlds - i just want to be able to register that
> my aeroplane is 'somewhere' in 3D space and let the worlds work out who sees
> it.
>
> This model isnt new.  A number of the early networked VR systems (dvs,
> dive, simnet/dis)supported this federated model.  Some existing/upcoming
> systems (Sirikata is a good example) are proposing that this sort of
> federated architecture is going to be key in building really complex worlds.
>  It would be fantastic to see them contribute to this discussion.
>
> Dare I ask the simple question here which is why not build on top of
> existing examples e.g. DIS/HLA?
> If not directly extending them, learn from them and build something better.
> -dirk
>
>
>
>  ----- Original Message -----
>>
>>> Date: Wed, 11 Mar 2009 17:46:17 -0700
>>> From: Jon Watte <[e-mail]>
>>> Subject: Re: [mmox] MMOX Progress tracking
>>>
>>
>> .> I am claiming that a client/server protocol does not achieve virtual
>>
>>> world interop, because there is only one world involved in the
>>> connection. Thus, there is no interop between virtual worlds (plural). Thus,
>>> it's not applicable to the "vendor neutral virtual world interoperability"
>>> that is in the charter for this group, because, by necessity, only one
>>> vendor and one world is involved at a time. Let's call it scope control.
>>>
>>> Separately, as I keep having to repeat every week, because new people
>>> come to the list: I think that a standard client/server protocol solves a
>>> very small, uninteresting problem. It solves the problem of having to
>>> install two, three or four clients, which is a small problem. In fact, I
>>> doubt you could even get very many people to pay a single dollar for the
>>> convenience of not having to install two clients to be able to visit two
>>> different worlds (assuming versioning could be solved at all). It doesn't
>>> solve anything else -- specifically, if I want to be with person A, who is
>>> in virtual world A, and person B, who is in virtual world B, at the same
>>> time, a standardized server/client protocol makes no difference; I still
>>> have to choose one or the other. That is a problem we have now, have no good
>>> vendor neutral solution for, and need a standard for.
>>>
>>> If you want a standard for client/server connections that lets you stay
>>> within the same client while switching connection between different,
>>> independent worlds, just put a plugin in a browser, and write it in Flash or
>>> Java. The language and the browser is the standard; it can be done today
>>> (and has been done). I can, from my browser, easily surf on over to another
>>> site that uses a different flash/java plug-in, and connect to that other
>>> world, without losing my browser context. For all intents and purposes, this
>>> looks to the user as if "the same client" connects to those different
>>> worlds. The fact that very few implementors do that means that that kind of
>>> interop simply isn't that interesting.
>>>
>>> If you think it's valuable to have a client/server protocol standardized
>>> through the IETF, then you should do so within a group that is appropriately
>>> chartered. In my opinion, the MMOX charter is not appropriate for that at
>>> the current time (and, in my opinion, that's a good thing).
>>>
>>> Sincerely,
>>>
>>> jw
>>>
>> .
>>
>
>
> _______________________________________________
> mmox mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/mmox
>



-- 
Charles Krinke
OpenSim Core Developer
OSGrid Director