* Nasty hack to reduce the incidence of spurious exceptions where a user deletes a newly rezzed object before the persistence thread gets to it from its queue.

* This should greatly reduce but not eliminate the problem - elimination probably requires a redesign of the prim persistence processes

1 files changed, 9 insertions, 1 deletions

diff --git a/OpenSim/Region/Environment/Scenes/InnerScene.cs b/OpenSim/Region/Environment/Scenes/InnerScene.cs
index 9a720d9..ef98599 100644
--- a/OpenSim/Region/Environment/Scenes/InnerScene.cs
+++ b/OpenSim/Region/Environment/Scenes/InnerScene.cs
@@ -236,7 +236,15 @@ namespace OpenSim.Region.Environment.Scenes
                     // Don't abort the whole update if one entity happens to give us an exception.
                     try
                     {
-                        m_updateList[i].Update();
+                        // A null name signals that this group was deleted before the scheduled update
+                        // FIXME: This is merely a temporary measure to reduce the incidence of failure, when
+                        // an object has been deleted from a scene before update was processed.
+                        // A more fundamental overhaul of the update mechanism is required to eliminate all
+                        // the race conditions.
+                        if (entity.Name != null)
+                        {
+                            m_updateList[i].Update();
+                        }
                     }
                     catch (Exception e)
                     {