Re: [mmox] 3-world OGP interop scenario

[Jon Watte <[e-mail]>]

Rob Lanphier wrote:
> Yes, one can assume client A can talk to servers B and C if we actually
> go about the work of defining a standard client-server protocol spelling
> out how the client is expected to behave in that transaction.
>   

Do you seriously think it's a good idea to specify all the minutiae of 
how a server and client interact, as a standard protocol? Including how 
GUI is tied to local and remote capabilities, how a scene graph is 
constructed and animated, how version patching is done, how user control 
is forwarded, etc?
And all that work, just to be able to show up in world B instead of 
world A? What about objects you want to use cross worlds; are you saying 
you also want to specify the exact runtime requirements for objects 
(scripting language, physics interface, etc)?

Personally, I believe that trying to do that means that there will be no 
cross-world interoperability except for the worlds that share the same 
DNA (meaning SecondLife <-> OpenSim, and perhaps ActiveWorlds <-> 
Worlds.com, and perhaps OLIVE <-> There.com). I think that would be a 
huge loss. Additionally, I think that all of that effort will actually 
not solve any problem that users care about, because all it really 
solves is saving users from having to install an extra client. But even 
in the ActiveWorlds <-> Worlds.com scenario, or the OLIVE <-> There.com 
scenario, because the platforms share technology DNA, using OGP is 
probably more work than it's worth, because there is already an existing 
infrastructure that is less resistance, and achieves the same amount of 
interop. OGP doesn't help interop, in this case, except for the Second 
Life <-> Open Sim case.

Meanwhile, I believe that the real enabler of interop is to be able to 
merge the simulations of different worlds, something like LESS or MXP. 
If you think about it, that model actually provides the same benefit you 
seek: the user being able to visit other places. It even provides the 
benefit of the user not having to install more than one client, ever, as 
long as the user's "home" world (or "virtual world service provider" if 
you will) is interoperable. However, it also provides the benefit of 
being able to use a There.com car, while letting ActiveWorlds users see 
you do that (and perhaps even ride the car). Also, the LESS/MXP approach 
requires much less engineering to implement for the majority of world 
vendors. I know this from experience, having done it for different 
protocols in the past.

Thus, if the goal is general virtual world interoperability, I believe 
the OGP model does not achieve it, whereas the LESS and MXP model does.

If the goal is to make systems with Linden Labs DNA interoperate, such 
as Open Sim or RealXTend, then I suggest that you state that as an 
explicit goal. I believe the feedback you're getting right now is that 
OGP, as proposed, is not suitable for interop outside that technology 
sphere.

Sincerely,

jw