Hello We have released today 0.3 draft of MXP with arkowitz. We are looking for people who can review the draft and even join the effort. Our next step is to work on proof of concept implementations on both Java and C#. Here is a short description about the protocol quoted from MXP wiki:
http://wiki.setp.info/wiki --- The goal of this wiki is to define minimal and easy to implement open protocol for distributed 3D environments (virtual environments): social 3d environments, online games and first person shooters. Distributed 3d environments are in this context systems with multiple thin clients and servers simulating single seemless virtual environment. This protocol concentrates entirely on messages required for the defined distributed 3d simulation model. This protocol does not contain application specific concepts but acts as a transport layer for custom interaction and state data which can be defined in a higher level specification. The same virtual space may be simulated by one or more servers. Distributed simulation (cloud, grid, cluster) consists of simulations nodes (bubbles, sims). Simulation node is appreviated to bubble and simulation cluster to cloud. One bubble has cubic or spherical volume but bubbles in a simulation cloud do not necessarily form a cubic lattice. Nodes may be sized and placed using entirely different strategy. Locations in messages are always presented in local simulation coordinates relative to the center of the source bubble. Each bubble is linked to coexisting and adjacent bubbles to enable injections, interactions and handovers. Client is connected to a single bubble at a time. Object is simulated by one bubble at a time. Handover is a process where object is handed over from one bubble to another. All though it is possible to support other ways of space partitioning or entirely free form distribution of bubbles, cubic lattice is still the most viable model from both implementation and network load perspectives. Cubic lattice can be applied as dual lattice to achieve local redundancy and load balancing (See: Advanced Grid Configurations). Purely peer to peer simulation model is not included to the current work due to trust issues. It is possible to allow clients to include temporary client side simulated objects to the simulation space in similar manner to avatars. In this scenario client is only responsible for internal state simulation of the object and these objects should be clearly marked as client side objects. Peer to peer protocols are considered viable for speech, video and gesture signals but not included to this protocol. --- best regards, Tommi Laukkanen --~--~---------~--~----~------------~-------~--~----~ http://groups.google.com/group/realxtend -~----------~----~----~----~------~----~------~--~---
