[Opensim-dev] defintion of terms

[Stefan Andersson]

I'm cool with MW's updates; can't think of anything to add/change.
 
/Stefan


Date: Thu, 18 Oct 2007 21:48:14 +0100From: michaelwri22 at yahoo.co.ukTo: opensim-dev at lists.berlios.deSubject: [Opensim-dev] defintion of termsJust wanted to make a few remarks about the definitions on wiki, thought it was best to make them here rather than change things on wiki, so that we can all try to get agreement on the terms.

Definition of some common terms: 
SceneA scene is part of a region, currently there is a one to one relationship, i.e. one region has one scene. A scene contains all the 3d objects, called Entities, entities can either be ScenePresences or SceneObjectGroups. 
A ScenePresence can be represented by an avatar, if it is a root scene presence. SceneObjectGroups consists of a number of SceneObjectParts.****Although it might not really be implemented fully at the moment, ScenePresence is meant to be able to "HAVE" a avatar. If its a what I'm taking that we are calling a  "root Scene Presence" , then it should have a avatar, if not then it should be thought of as a ghost scene presence...it can see the region but others in the region can't see it.*****
Avatar/ScenePresenceEach viewer is represented by one or more avatars. The Root avatar or ScenePresense functions as a kind of cursor, into the grid. Where the region will keep the viewer updated on the state of the simulation, taking place in the scene. Besides the root presence, a number of child presences will exist in the neighboring regions, enabling the viewer to know what is going on in the neighboring regions.
****** 
A viewer should only ever have one Avatar, here is where due to some of the names that we have in use in opensim that things get confusing I think. 
I believe the best way to think of things, is that a view has multiple agents. A "agent" is really a connection to a region. so it has a main agent to the region that its avatar will be in, then child agents to the neighbouring regions. 
A agent (connection) has a scene presence inside the scene of that region. With the main agent, it's scene presence has a avatar , that the other users see. But with child agents, there scene presence have no avatar so are in effect ghosts in that region. 
As said, we kind of in opensim use the same terms for multiple things, like agents is used for a few different things, where I personally think the correct usage of it would be for a connection to a region. So the flow from viewer to region is something like:  viewer <-circuit -> ClientStack (although we must likely need to look at changes in the stack as thats mostly old code ) <--->Region Agent <---> ScenePresence (which may or maybe not have a Avatar)At the moment we don't really have any thing like this Agent, but I do think as part of the refactoring of the OpenSim.Region.Enviroment , that a lot of the things currently in Scene should be moved out to a outter layer (most likely called RegionService/manager), and most likely have a Agent class in this layer for each connected client, so that not everything goes into the scene layer, and some client requests are dealt with in this region layer. *****
ViewerAn application which is used to view the state of the simulation. Currently only three different viewers are known: the LL viewer, the OpenSL test client and AjaxLife.
****
Isn't OpenSL dead? or do we mean a libsl test client?
*****
SessionEach connection between a viewer and an region, regardless if the avatar is the root or a child, is a session.****I think a session is a login session, so a single session covers all agents, scenepresences etc. A viewer should only have one session at a time.*****



For ideas on reducing your carbon footprint visit Yahoo! For Good this month. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20071019/2122f750/attachment-0001.html>