EXPERIMENTAL: Trying the other order -- delaying MoveAgentIntoRegion until after the departing agent is released.

OpenSim/Region/Framework/Scenes/ScenePresence.cs

@@ -1329,12 +1329,6 @@ namespace OpenSim.Region.Framework.Scenes
 
             bool flying = ((m_AgentControlFlags & AgentManager.ControlFlags.AGENT_CONTROL_FLY) != 0);
             MakeRootAgent(AbsolutePosition, flying);
-            ControllingClient.MoveAgentIntoRegion(m_scene.RegionInfo, AbsolutePosition, look);
-            // Remember in HandleUseCircuitCode, we delayed this to here
-            // This will also send the initial data to clients when TP to a neighboring region. 
-            // Not ideal, but until we know we're TP-ing from a neighboring region, there's not much we can do
-            if (m_teleportFlags > 0)
-                SendInitialDataToMe();
 
 //            m_log.DebugFormat("[SCENE PRESENCE] Completed movement");
 
@@ -1361,6 +1355,13 @@ namespace OpenSim.Region.Framework.Scenes
 //                    client.Name, client.AgentId, m_scene.RegionInfo.RegionName);
 //            }
 
+            ControllingClient.MoveAgentIntoRegion(m_scene.RegionInfo, AbsolutePosition, look);
+            // Remember in HandleUseCircuitCode, we delayed this to here
+            // This will also send the initial data to clients when TP to a neighboring region. 
+            // Not ideal, but until we know we're TP-ing from a neighboring region, there's not much we can do
+            if (m_teleportFlags > 0)
+                SendInitialDataToMe();
+
             ValidateAndSendAppearanceAndAgentData();
 
             // Create child agents in neighbouring regions