Re: [mmox] One more time: The LESS model vs the Generic Client model

I've been avoiding this list lately since I've not had the time to read the (very) significant number of posts made in recent history, however I am going to chime in here again and say we should be utilizing URI-style constructs as much as possible.

I believe we [may] have a rough consensus that there are elements we want to standardise, this being authentication and identity circuits. However there is no reason why the client<->region connection needs to be fixed as a specific protocol, simply the result from the authentication mechanism can return a URI.

For instance,

MMOX defines a auth protocol, client makes attempt to connect to a environment via this auth protocol, is returned a URI with "mxp://ip/regionID/", or a "olive://ip/region", or a "secondlife://region/x/y/z". (or perhaps a combination of the above for servers capable of handling more than one [eg opensim].)

This then allows others to pitch client<->region standards and let the market weed out what is deemed to be effective / noneffective. It would still then be entirely in the remit of this group to propose a standard for the above too - however by utilizing the URI schema, it does not force implementers into using it (only recommends it as a 'common reference point').

Adam

> -----Original Message-----
> From: mmox-bounces@ietf.org [mailto:mmox-bounces@ietf.org] On Behalf Of
> Jon Watte
> Sent: Monday, 16 March 2009 8:10 AM
> To: Morgaine
> Cc: MMOX-IETF
> Subject: Re: [mmox] One more time: The LESS model vs the Generic Client
> model
>
> Morgaine wrote:
> > The goal is simply to get us beyond the current pro-client/anti-
> client
> > discussion, which like almost all polarized pro/anti discussions, is
> > rarely constructive.
>
> The reason we ended up here was because I said OGP will not be widely
> adopted outside the Second Life / OpenSim sphere, because it requires
> significant engineering in the client/server connection to achieve any
> kind of interop through OGP, for any kind of cross-technology interop.
> In the OGP model, currently, if Open Sim wants to connect to OLIVE,
> they
> need to implement the OLIVE protocol. If OLIVE wants to connect to
> Metaverse, they need to implement the Metaverse protocol. I consider
> that broken.
> You suggest that this problem will be solved by engineering a standard
> client/server protocol. I point at the history of synchronous
> interactive online systems (note: web is not synchronous, and hardly
> interactive), and say that I don't believe such a standard is feasible
> (politically or technically).
> Thus, I believe that an "interop" protocol that requires all technology
> clients to be able to talk to all other technology clients to not
> actually deliver feasible interop (nevermind that it doesn't solve the
> world mash-up problem).
> While there may be, in your words, "generic" clients within a specific
> technology domain, that doesn't mean much for cross-technology interop.
>
> Note: because words seem to be mis-understood, in the above comments,
> when I say "technology," I mean "virtual world platform DNA" --
> Multiverse is a technology; OLIVE is a technology, Second Life is a
> technology, Unreal 3 is a technology. When it comes to client/server,
> interop between different worlds using the same technology is easy (at
> least if the technology makes it so); it's interop between different
> technologies that needs standardization to happen.
>
> Sincerely,
>
> jw
>
> _______________________________________________
> mmox mailing list
> mmox@ietf.org
> https://www.ietf.org/mailman/listinfo/mmox