<html>
<head>
<style>
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
FONT-SIZE: 10pt;
FONT-FAMILY:Tahoma
}
</style>
</head>
<body class='hmmessage'>Of course that sounds reasonable, I was just reflecting over the fact that PhysicalScene.AddAvatar( pos ) creates a _new_ physActor everytime; but PhysicalScene.RemoveAvatar( physActor ) just detaches it.<BR>
<BR>
On one hand, I don't think you have to 'know the inner works' but on the other hand, you shouldn't be afraid to 'follow the code' and muck about in the core;<BR>
<BR>
we will have very slap-on physics if you don't, which would be bad. The Physics interaction and wireup is one of the big white spots on the chart, and a very important such.<BR>
<BR>
But in the end, you do what you think is fun/worthwile. :-D<BR>
<BR>
/Stefan<BR><BR><BR><BR><BR>
<HR id=stopSpelling>
<BR>
> Date: Wed, 19 Sep 2007 15:17:38 -0700<BR>> From: danbmil99@yahoo.com<BR>> To: opensim-dev@lists.berlios.de<BR>> Subject: Re: [Opensim-dev] clients, agents, avatars, ODE<BR>> <BR>> from my narrow POV as physics guy, what should happen is this:<BR>> <BR>> when a client *really* enters a region, thru logon or across a border,<BR>> AddAvatar() should be called, so the avatar can interact with the<BR>> environment (walk on terrain, bump into prims/other avs...)<BR>> <BR>> when the client leaves the region, RemoveAvatar() is called to remove it<BR>> from the physics for that region.<BR>> <BR>> How the code got into the state it's in, I have no idea. Some of this was<BR>> inherited from pre-sugilite days, when things were mighty different.<BR>> <BR>> I'm willing to get into the logistics of the rest of the codebase to some<BR>> degree, but frankly I think my energies are better spent staying focused on<BR>> physics issues. While not ideal, the physics engine plugin API is a<BR>> reasonable first pass, and is a good point of interaction between physics<BR>> and the rest of the project.<BR>> <BR>> What I think might be helpful would be for me to add debug messages along<BR>> the lines of:<BR>> <BR>> [PHYSICS] AddAvatar called; avatar count for this scene is now X<BR>> [PHYSICS] RemoveAvatar called; avatar count for this scene is now X<BR>> <BR>> or something along those lines. Then you geniuses can mull over the code<BR>> and mangle it into something that makes sense from the physics angle.<BR>> <BR>> Does this sound reasonable?<BR>> <BR>> -dan<BR>> <BR>> <BR>> --- Stefan Andersson <stefan@tribalmedia.se> wrote:<BR>> <BR>> > I guess you noticed by now, but I have done extensive restructuring in<BR>> > these areas yesterday and today; still, I suspect there is something odd<BR>> > (from looking at the code) with the chain of events. Also, we're<BR>> > re-creating physActors everytime we move into a region, but we just detach<BR>> > it from the physScene, we don't explicitly destroy (close) the old one<BR>> > when we leave; which I would suspect a native-wrapper would very much like<BR>> > us to.<BR>> > <BR>> > /Stefan<BR>> > <BR>> > <BR>> > <BR>> > > Date: Tue, 18 Sep 2007 23:00:53 -0700> From: danbmil99@yahoo.com> To:<BR>> > opensim-dev@lists.berlios.de> Subject: Re: [Opensim-dev] clients, agents,<BR>> > avatars, ODE> > followup to chat with lbsa71:> > in ClientView.cs, we<BR>> > have:> > this.m_scene.AddNewClient(this, false);> > so it's always passing<BR>> > child=false. But InitNewClient is called by> AuthUser, which is spawned as<BR>> > a thread by ClientView, and... climbing up the> stack of calls, we get<BR>> > OnReceivedData creating a new client thread when it> gets a<BR>> > UseCircuitCode. This seems to happen once for each region, and I> have no<BR>> > idea how to tell what is a child and what is the 'real' client, to><BR>> > hopefully pass it down the call stack or something of that nature.> ><BR>> > -dan> > > _______________________________________________> Opensim-dev<BR>> > mailing list> Opensim-dev@lists.berlios.de><BR>> https://lists.berlios.de/mailman/listinfo/opensim-dev<BR>> <BR>> _______________________________________________<BR>> Opensim-dev mailing list<BR>> Opensim-dev@lists.berlios.de<BR>> https://lists.berlios.de/mailman/listinfo/opensim-dev<BR><BR></body>
</html>