This is to make these statistics actually match their names (and also be more accurate as number of frames can vary under heavy load)
Currently using scene frames (11.23 every second) instead of physics frames (56.18 per second)
If active, the physics module can return arbitrary stat counters that can be seen via the MonitoringModule
(http://opensimulator.org/wiki/Monitoring_Module)
This is only active in OdeScene if collect_stats = true in [ODEPhysicsSettings].
This patch allows OdeScene to collect elapsed time information for calls to the ODE native collision methods to assess what proportion of time this takes compared to total physics processing.
This data is returned as ODENativeCollisionFrameMS in the monitoring module, updated every 3 seconds.
The performance effect of collecting stats is probably extremely minor, dwarfed by the rest of the physics code.
If this is done before then on ODE agent update calls still incoming can fail as they try to use a raycastmanager that has been disposed.
Bullet plugin does nothing on Dispose()
However, I wouldn't be at all surprised if individual region restarting was buggy in lots of other areas.
If this is allowed, then the client usually gets forcibly logged out and data structures might be put into bad states.
To prevent this, the binary state machine of EMT.m_agentsInTransit is replaced with a 4 state machine (Preparing, Transferring, ReceivedAtDestination, CleaningUp).
This is necessary because the source region needs to know when the destination region has received the user but a teleport back cannot happen until the source region has cleaned up.
Tested on standalone, grid and with v1 and v3 clients.
This is to help relieve a race condition when an agent teleports then immediately attempts to teleport back before the source region has properly cleaned up/demoted the old ScenePresence.
This is rare in viewers but much more possible via scripting or region module.
However, more needs to be done since virtually all clean up happens after the transit flag is cleared .
Possibly need to add a 'cleaning up' state to in transit.
This change required making the EntityTransferModule and HGEntityTransferModule per-region rather than shared, in order to allow separate transit lists.
Changes were also required in LocalSimulationConnector.
Tested in standalone, grid and with local and remote region crossings with attachments.
This includes prim count, script count, avatar count, etc.
Information is currently the same as "show stats", though show stats can only show one scene at a time because it listens for the latest outgoing stats packet (a bad approach that needs to change).
Might be better to tie this module into the other stats module to display arbitrary stats rather than fetching directly from scene.SimStatsReporter.
Console command is "show scene" because "show region" already exists for the grid service, which is unfortunate.
Might need to make a distinction between "scene" relating to a live scene and "region" relating to more static region data (url, coords, etc.)
This is because the attachments module code was setting the 'object slam' bit by using PermissionMask.All
Solution here is to route the attachment item creation call through the existing inventory code in BasicInventoryAccessModule rather than copy/pasted code in AttachmentsModule itself.
An adaptation of part of Garmin's patch from http://opensimulator.org/mantis/view.php?id=5975, thanks!
Flag only written if the SW corner OAR is saved - this is the only one that captures object data presently (though not land or terrain data).
This adds an IRegionCombinerModule interface and the necessary methods on RegionCombinerModule
From sl docs such as http://community.secondlife.com/t5/English-Knowledge-Base/Managing-Private-Regions/ta-p/700115
agent should apply to avatars only.
This makes sense from a user perspective, and also from a code perspective since child agents with no physics or actions take up a fraction of root agent resources.
As such, the check is now only performed in Scene.QueryAccess() - cross and teleport check this before allowing an agent to translocate.
This also removes an off-by-one error that could occur in certain circumstances on teleport when a new child agent was double counted when a pre-teleport agent update was performed.
This does not affect an existing bug where limits or other QueryAccess() checks are not applied to avatars logging directly into a region.
This was because the parts in scene objects were sometimes not serialized in link order.
This is perfectly fine since the parts still have the right link numbers, but an extra fix to adjust for this
had not been done in the SerialiserModule methods that OAR loading used.
Add regression test for same.
Addresses http://opensimulator.org/mantis/view.php?id=5948, http://opensimulator.org/mantis/view.php?id=5749
Will use one of three selected methods to route avatar landing
points when using Telehubs. The setting is in [Startup] using
SpawnPointRouting = closest/random/sequence
closest: The default setting. Routes avatar to the nearest SpawnPoint
to the location.
random: Picks random SpawnPoints to land the avatar.
sequence: Follows a sequence to place the avatar on the next available
SpawnPoint location
Conflicts:
OpenSim/Region/Framework/Scenes/Scene.cs
DuplicateObject() relies on source object having correct link numbers for the duration of the dupe.
Both link and delink can change link numbers such that they are not consistent for short periods of time.
Viewers 1 and 3 are fine with doing this immediately. However, Imprudence has a small delay (<200ms, >500ms) after receiving the AgentCompleteMovement reply packet on the destination region before regarding that region as the currnet region.
If Imprudence receives a DisableSimulator in this period, it quits.
We are not restoring the full 5000ms delay since this brings back a bug where teleports permanently fail if an avatar tries to teleport back too quickly.
This commit also sends the AgentCompleteMovement packet to the client before telling the source region to release its old agent, in order to further cut down any possibility of the DisableSimulator being recieved before the AgentMovementComplete.
This adds a non-advertised wait_for_callback option in [EntityTransfer]. Default is always true.
Teleport tests disable the wait for callback from the destination region in order to run within a single thread.
This happened because the scripts were notified about control changes (e.g., the user stopped pressing the Forward key) when the animation was still WALK, so the script didn't stop the walking animation. Fixing this required: a) Update the movement animation *before* notifying the script; b) Add locking to prevent clashes with the Heartbeat thread (which also updates the animations); c) Handle the case of a user who stops walking just as the avatar is in the air: the avatar should STAND in that case, not WALK.
This reverts commit feef1dd732.
This involves a large amount of change in test scene setup code to allow test scenes to share shared modules
SetupScene is now an instance method that requires an instantiation of SceneHelpers, though other SceneHelpers methods are still static
May split these out into separate classes in the future.
folder when deleting objects from a scene. The use of the trash
folder causes assets to be created and stored everytime you delete
an object from the scene (slows down the delete and adds mostly useless
assets to your database).
Default is on (use the trash folder) which is the standard behavior.
These are identical apart from setting Velocity = zero, which has no practical effect anyway since this is zeroed when the avatar is added back to the physics scene.
This was an oversight when removing some race conditions from PhysicsActor setting recently.
Regression tests extended to probe this code path.
Extending regression tests required implementation of a BasicPhysicsPrim (there was none before). However, BasicPhysics plugin is still of no current practical use other than to fill in as a component for other parts of regression testing.
On region restart, the whole object would become physical as expected.
Observed behaviour from elsewhere is that all prims in a new linkset should take on the status of the root prim.
Add regression test for this behaviour.
storage (dictionaries and arrays of string values) for scripts and region modules.
In addition, there are operations on the storage that enable "real" distributed
computation between scripts through operations similar to those of a tuple space.
Scripts can share task queues, implement shared locks or semaphores, etc.
The structured store is limited to the current region and is not currently
persisted. However, script operations are defined to initialize a store from a notecard
and to serialize the store to a notecard.
Documentation will be posted to the opensim wiki soon.
This makes it consistent with other parts of OpenSimulator that are treating ESTATE_MANAGER and ESTATE_OWNER as different entities.
As per opensim-dev mailing list.
Another thread could come and turn off physics for a part (null PhysicsActor) at any point.
Had to turn off localCopy on warp3D CoreModules section in prebuild.xml since on current nant this copies all DLLs in bin/ which can be a very large number with compiled DLLs
No obvious reason for doing that copy - nothing else does it.
Rename IFriendsModule.GetFriendPerms() -> GetRightsGrantedByFriend() to be more self-documenting and consistent with friends module terminology.
Add some method doc.
This adds ScenePresence to IClientAPI.SceneAgent earlier on in the add client process so that its information is available to EventManager.OnNewClient() and OnClientLogin()
Also add a code comment as to why we're caching friend information for child agents.
We need to cache child agents so that friends object edit/delete permissions will work across boarders on regions hosted by different simulators.
This reverts commit d9f7b8549b.
This allows us to avoid unnecessary multiple calls to the friends service.
All friends functions originate from the root agent and only go to other root agents in existing code.
This also allows us to eliminate complex ref counting.
On osgrid and other places, I have observed that manually sending appearance updates from the console often relieves grey avatar syndrome.
Despite hunting high and low, I haven't been able to find where this packet is sometimes being lost - it might be a persistent viewer bug for all I know.
Therefore, this experimental setting resends appearance data for everybody in the scene every 60 seconds. These packets are small and the viewer only fetches texture
data if it doesn't already have it.
Default is false.
The idea is to make the critical main scene loop as skinny as possible - it doesn't need to run things that aren't time critical and don't depend on update ordering.
This will be done gradually over time to try and uncover any issues. Many non-criticial scene loop activities are being launched on separate threadpool threads anyway.
This may also allow modules to register their own maintenance jobs without having to maintain their own timers and threads.
Currently the maintenance loop runs once a second, as opposed to the 89ms scene loop.
This lets us remove the dependency of OpenSim.Framework.dll on data/avataranimations.xml, which is not necessary for ROBUST.
This commit also takes care of the odd situation where animations are stored and used internally with uppercase names (e.g. "STAND")
but scripts refer to them with lowercase names (e.g. "sit").
Update() now accepts a frames parameter which can control the number of frames updated.
-1 will update until shutdown.
The watchdog updating moves above the maintc recalculation for any required sleep since it should be accounted for within the frame.
The previous lines-per-second measurement used for top scripts report was inaccurate, since lines executed does not reflect time taken to execute.
Also, every fetch of the report would reset all the numbers limiting its usefulness and we weren't even guaranteed to see the top 100.
The actual measurement value should be script execution time per frame but XEngine does not work this way.
Therefore, we use actual script execution time scaled by the measurement period and an idealised frame time.
This is still not ideal but gives reasonable results and allows scripts to be compared.
This commit moves script execution time calculations from SceneGraph into IScriptModule implementations.
from region modules. The LSL translator is extended to generate the
modInvoke format of commands for directly inlined function calls.
A region module can register a function Test() with the name "Test".
LSL code can call that function as "Test()". The compiler will translate
that invocation into modInvoke("Test", ...)
The first llDie() could lock Scene.m_deleting_scene_object.
The second llDie() would then wait at this lock.
The first llDie() would go on to remove the second script but always abort it since the second script's WorkItem would not go away.
Easiest solution here is to remove the m_deleting_scene_object since it's no longer justified - we no longer lock m_parts but take a copy instead.
This also requires an adjustment in XEngine.OnRemoveScript not to use instance.ObjectID instead when firing the OnObjectRemoved event.
However, it looks like we should retain SP.ParentID since it's much easier to use that in places where another thread could change ParentPart to null.
Otherwise one has to clumsily put ParentPart in a reference, etc. to avoid a race.
This is to deal with the hundred lines of command splurge when one previously typed "help"
Modelled somewhat on the mysql console
One can still type help <command> to get per command help at any point.
Categories capitalized to avoid conflict with the all-lowercase commands (except for commander system, as of yet).
Does not affect command parsing or any other aspects of the console apart from the help system.
Backwards compatible with existing modules.
We can now do this since the entire scene and all objects within it are now successfully gc'd at the end of these tests.
This greatly improves the time taken to run each test (by reducing teardown time, not the time to actually do the test work that we're interested in).
Slightly simplifies config read in Scene constructor to help facilitate this.
OnNewScript fires when a script is added to a scene
OnRezScript fires when the script actually runs (i.e. after permission checks, state retrieval, etc.)
This allows scripts to run in child prims that are outside region boundaries.
This is an interim patch applied from http://opensimulator.org/mantis/view.php?id=5899 though it does not resolve that bug
Thanks tglion!
On the first frame, all startup scene objects are added to the physics scene.
This can cause a considerable delay, so we don't start raising the alarm on scene loop timeouts until the second frame.
This commit also slightly changes the behaviour of timeout reporting.
Previously, a report was made for the very first timed out thread, ignoring all others until the next watchdog check.
Instead, we now report every timed out thread, though we still only do this once no matter how long the timeout.
This involves
1) On forcible teleport, call m_scene.RequestTeleportLocation() rather than ScenePresence.Teleport() - only EntityTransferModule now should call SP.Teleport()
2) When avatar is being forcibly moved due to banlines, use a 'stop movement' tolerance of 0.2 to requested position rather than 1
This prevents the avatar sometimes being stuck to banlines until they teleport somewhere else.
This aims to fix some problems in http://opensimulator.org/mantis/view.php?id=5822
potential bad update that places an object at the opposite side of the
origin sim for a moment before actually crossing it. Especially important in
grids like OSG where lag between sims is high.
This fixes the problem by fixing the permissions module to look at root part permissions rather than having to do this for every caller.
Resolves http://opensimulator.org/mantis/view.php?id=5569
At least on mono 2.6.4, running GC.Collect() is not guaranteed to force gc of all objects when run in the same method where those objects had references.
Therefore, GC.Collect() is now being done in the per-script teardown of ObjectTortureTests.
In addition, scene loop update is being run after garbage collection in order to clean out the viewer update list of scene objects in the SceneGraph.
These measures mean that scene objects/parts are now garbage collected after a test run if deleted from the scene, resulting in a much better memory usage report (though probably still not very accurate).
However, deletion takes a very long time - what's really needed is to find out now why the entire scene isn't being GC'd by this measure.
This change hasn't yet been applied to the other stress tests.
Logging level was DEBUG before 312e145 (Fri Feb 3 2012).
312e145 also accidentally removed the 'general error' log message if any shape deserialization failed.
This commit restores it, though this has no functional impact.
Setting PermissionMask.All will cause next permissions to replace current permissions when the object is rezzed, since bit 4 will be set.
This is not correct behaviour for a freshly uploaded mesh. Freshly rezzed in-world prims also do not have bit 4 set (don't yet know exactly what this is).
Should resolve http://opensimulator.org/mantis/view.php?id=5651
This illustrates that references to Scene, SOG, etc. are not currently being released when a stress test ends (or at regression test end in general).
This means even the current stress tests take much more memory than they need, a problem that will have to be addressed.
The Path.GetDirectoryName call in Compiler.CompileFromDotNetText is unnecessary since AppDomain.CurrentDomain.BaseDirectory is always a directory.
Later path concatenation is already done by Path.Combine() which handles any trailing slash.
Removing Path.GetDirectoryName() will not affect the runtime but allows NUnit to work since it doesn't add a trailing slash to AppDomain.CurrentDomain.BaseDirectory.
This resolves the recent regression from deeb728 where notecards could not be saved in prim inventories.
This looks like a better solution than deeb728 since only non-caps updates pass in a transaction ID.
Hopefully resolves http://opensimulator.org/mantis/view.php?id=5873
These can be run using the "nant torture" target. They are not part of "nant test" due to their long-run future nature.
Such tests are designed to do some testing of extreme situations and give some feedback on memory usage, etc.
However, data can be inconsistent due to different machine circumstances and virtual machine actions.
This area is under development.
Viewers since at least Linden Lab 1.23 use the script upload capability to save script changes.
It's unknown whether the commented out code was working for very old viewers or not.
Code is commented out to reduce complexity and so that useful error messages don't need to be removed.
If there is a substantial population using extremely old viewers that can't upgrade to a newer version 1 viewer (e.g. 1.23) or similar TPV then this can be revisited.
This allows NPCs to be sensed as agents by LSL sensors rather than as a specific NPC type (which is currently an OpenSimulator-only extension).
Wiki doc on this and other recent NPC functions will follow soon
llGetOwnerKey() could also be extended but this does not allow one to distinguish between an unowned NPC and some other result (e.g. 'no such object' if NULL_KEY is the return.
Also, any future extensions to LSL functions by Linden Lab are unpredictable and OpenSim-specific extensions could clash.
Doing this in every script is unnecessary since the event trigger is parameterized by the item id.
All that would happen is 2000 scripts would trigger 1999 unnecessary calls, and a large number of initialized scripts may eventually trigger a StackOverflowException.
Registration moved to UrlModule so that the handler is registered for all script engine implementations.
This required moving the OnScriptRemoved and OnObjectRemoved events (only used by UrlModule in core) from IScriptEngine to IScriptModule to avoid circular references.
Add configuration option - DEBUG to enable debugging methods. This is temporary for helping users testing teleport routing be able to report back the data with the test cases. We can remove when finished with this, or leave it if it proves to be useful.
Users: set DEBUG = true in OpenSim.ini to get more information from teleport routing. The default is false. It presently prints the TeleportFlags value.
Switch to our TeleportFlags enum instead of LibOMV because we need to define a type for HG Logins. Also moved some debugging in ScenePresence into a function to make it simpler to enable/disable.
Incoming HG owner/estate manager, etc. is routed according to the rules defined for teleports within the local grid. Left some commented debugging code inside so we can test other cases. Will remove when tings are settled in.
Update basic profile to use the replaceable interface, making configuration less error-prone. Add support to query avatar's home user account and profile service for regions usng the updated OpenProfileModule with Hypergrid.
This is a nasty situation. The map tile UUID is, in principle, stored authoritatively in RegionSettings. However, it also needs to be stored in the Grid Service because that's how other sims can retrieve it to send it in Map Blocks to non-V3 viewers. So every time the tile image changes, that change needs to propagate to the Grid Service, and this is done via RegisterRegion (ugh!). Interestingly, this problem didn't affect grids because by default AllowRemoteDelete is false, so the prior images aren't being deleted from the asset servers -- but they were not being correctly updated in the map either, the map was stuck with old images.
Naturally, default is true.
When set to false, "phantom" flags on prims can be set as usual but all prims remain phantom.
This setting is for test purposes.
This switch does not affect the collision of avatars with the terrain.
The viewer warns in the log if it receives this.
Stopping this doesn't appear to have adverse effects on viewer 1 or viewer 3 - the viewer gets its own appearance from body parts/clothes and self-baked textures.
Further filters "debug packet <level>" to exclused [Request]ObjectPropertiesFamily if level is below 25.
Adjust some method doc
Minor changes to some logging messages.
There were two problems here:
1) On object group update, we looked for the group is the IClientAPI group cache rather than in the groups service. This fails to groups created newly in that session
2) On object group update, we weren't setting the HasGroupChanged flag. This meant that the change was not persisted unless some other action set this flag.
This commit fixes these issues and hopefully addresses http://opensimulator.org/mantis/view.php?id=5588
This commit also moves HandleObjectGroupUpdate() to the GroupsModule from the Scene.PacketHandlers.cs file
Stop hiding RemoveAvatar failure, add log messages when characters are removed through defects or re-added unexpectedly.
Add commented out log lines for future use.
Use automatic property for PhysicsActor for better code readability and simplicity
This prints out both exception message and stacktrace (Exception.ToString()) isn't enough on Windows.
This also uses m_log.*Format() which is more efficient than string concat.
The only caller is the LLUDP stack and this has to validate the UDP circuit itself, so we know that it exists.
This allows us to eliminate another null check elsewhere and simplifies the method contract
This means that avatar/appearance data of other avatars and scene objects for a client will be sent after the ack rather than possibly before.
This may stop some avatars appearing grey on login.
This introduces a new OpenSim.Framework.ISceneAgent to accompany the existing OpenSim.Framework.ISceneObject and ISceneEntity
This allows IClientAPI to handle this as it can't reference OpenSim.Region.Framework.Interfaces
If sp becomes null right after we've checked or created it, then behaviour down the line is going to be wrong anyway.
So instead retain the check/create ScenePresence reference and use this.
It makes far more sense anyway to use TryGetRootScenePresence().Scene, in common with the rest of the code
This method could also return any scene for child or root agents, depending in which order the scenes happened to lie in the list
This is required for the substitution of different HTTP servers or the newer HttpServer.dll without having to commit to a particular implementation.
This is also required to write regression tests that involve the HTTP layer.
If you need to recompile, all you need to do is replace OSHttpRequest/OSHttpResponse references with IOSHttpRequest/IOSHttpResponse.
This required an option to be added to NullRegionData via ConnectionString for it to act as a non-static instance, so that regression tests (which only load this class once) don't get hopeless confused and complex to compensate.
Normal standalone operation unaffected.
This reverts to situation where animation updates are made each frame on SP.PhysicsCollisionUpdate (though a packet is only sent if the anim actually changes).
m_updateCount was not being update on various avatar state changes, causing the correct animations to never be sent.
Always setting in HandleAgentUpdate() is not enough since the avatar is continually sending AgentUpdate packets.
One would need to identify all the conditions under which animations need to play out and set m_updateCount appropriately in SP.HandleAgentUpdate()
Neither of these can have any effect on child agents
Now leaving warning about trying to set animation on a child agent active. Might temporarily pop up now and again.
This involves getting IScene.RequestModuleInterfaces() to return an empty array (as was stated in the method doc) rather than an array containing one null entry.
Callers adjusted to stop checking for the list reference being null (which never happened anyway)
These are just the result of an attempt to canonicalize received messages - it's not important that we constantly log them.
Also finally get the deregister grid service message working properly
Executing this asynchronously allows a race condition where subsequent friends fetches hit a cache that FetchFriendsList() had not yet populated.
Changing this to synchronous may improve issues where a user does not see friends as online even though they are.
I don't believe synchronous is a problem here, but if it is, then a more complicated signalling mechanism is required. Locking the cache isn't sufficient.
prim update to only triple queuing. Existing method was:
1. Schedule prim for update, adding to scene update list
2. Update on SOGs during heartbeat queues update onto each SceneViewer
3. Update on SPs during heartbeat queues update onto each IClientAPI
4. ProcessEntityUpdates queues updates into UDP send stack
Now the SceneViewer has been eliminated so updates are scheduled at any
time and then put onto the IClientAPI priority queues immediately during
SceneGraph.UpdateObjectGroups.
This is to avoid http://opensimulator.org/mantis/view.php?id=5783 when a collision with a ground sitting avatar causes that avatar to automatically stand and sometimes not be able to move
A better solution may be to keep gound sitting avatars solid but remove their collision status. However, this requires some physics code work.
This means that if the avatar is within 10 meters of the selected target, it sits on it immediately without walking.
Existing autopilot outside this range will be disabled in a later commit
This is to partially address http://opensimulator.org/mantis/view.php?id=5769
We don't need to call SP.HandleAgentSit() again if we are within 10m since the autopilot won't trigger.
By calling it twice, the position of the sitting NPC was wrongly adjusted, ending up near <0,0,0>.
However, this change does mean that NPCs further than 10m away will not attempt to autopilot to the prim, though this code was broken anyway (is actually a different mechanism to normal NPC movmeent).
Hopefully this can be addressed soon.
This was meant to help with the script in http://opensimulator.org/mantis/view.php?id=5772 but it doesn't work.
Probably the event is fired before the physics actor has been set up again for the stood avatar.
Fixing that would be much more complicated, but processing the event last of all seems like a good idea in any case.
from previous commit which sort out which iterator is used are left
intact. A discussion is needed as to what constitutes an avatar vs a
ScenePresence.
the 3 iteration functions so more of them are using the correct
iteration for the action they are performing. The 3 iterators that seem
to fit all actions within OpenSim at this time are:
ForEachAvatar: Perform an action on all avatars (root presences)
ForEachClient: Perform an action on all clients (root or child clients)
ForEachRootClient: Perform an action on all clients that have an avatar
There are still a dozen places or so calling the old
ForEachScenePresence that will take a little more refactoring to
eliminate.
UpdateFlag is now referenced/used only within SOP and SOG. Outsiders are
using ScheduleFullUpdate, ScheduleTerseUpdate or ClearUpdateSchedule on
SOP consistently now. Also started working toward eliminating those
calls to ScheduleFullUpdate, ScheduleTerseUpdate or ClearUpdateSchedule
from outside SOP in favor of just setting properties on SOP and let SOP
decide if an update should be scheduled. This consolidates the update
policy within SOP and the client rather than everywhere that makes
changes to SOP. Some places forget to call update while others call it
multiple times, "just to be sure".
UpdateFlag and Schedule*Update will both be made private shortly.
UpdateFlag is intended to be transient and internal to SOP so it has
been removed from XML serializer for SOPs.
In AvatarFactoryModule.HandleAppearanceUpdateTimer(), we loop through appearance save and send requests and dispatch via a FireAndForget thread.
If there was more than one request in the save or send queue, then this led to a subtle race condition where the foreach loop would load in the next KeyValuePair before the thread was dispatched.
This gave the thread the wrong avatar ID, leaving some avatar appearance cloudy since appearance data was never sent.
This change loads the fields into local references so that this doesn't happen.
This is necessary so that code in HttpServer can use framework facilities such as the thread watchdog for monitoring purposes.
Doing this shuffle meant that MainServer was moved into OpenSim/Framework/Servers
Also had to make OpenSim.Framework.Console rely on OpenSim.Framework rather than the other way around since it in turn relies on HttpServer
MainConsole and some new interfaces had to be moved into OpenSim/Framework to allow this. This can be reverted if parts of OpenSim.Framework stop relying on console presence (cheifly RegionInfo)
SinceLastFrame was calculating the interval between any sleep that had occurred to pad out the frame time and the start of the next frame.
This would usually be below MinFrameTime but occasionally if the sleep was long it would be above, often due to the time required to update the watchdog.
This doesn't appear to play much practical role right now.
ODE was actually ignoring it entirely. Bullet might be helped slightly by receiving a non-varying value.
Format is osNpcSit(<npc-uuid>, <target-uuid>, OS_NPC_SIT_IMMEDIATE)
e.g. osNpcSit(npc, llGetKey(), OS_NPC_SIT_IMMEDIATE);
At the moment, sit only succeeds if the part has a sit target set.
NPC immediately sits on the target even if miles away - they do not walk up to it.
This method is in development - it may change so please don't trust it yet.
Standing will follow shortly since that's kind of important once you're sitting :)
This had stopped working. However, at the moment it still allows the physics flag to be set even though this has no effect. This needs to be fixed.
Default for this flag is true as previously.
Settings are at bottom of [Startup] in OpenSimDefaults.ini, override in OpenSim.ini to change
Defaults are the same as previously.
More information to come on opensim-dev shortly.
Feel free to tweak but if you do please don't expect any support unless feedback on certain tweaks is explicitly requested.
Unlike the other 3 stats mechanisms, monitor data can be queried per individual region, which makes this useful.
This doesn't affect an of the existing monitored stats.
ways to access the list/dictionary of child regions and locking was
inconsistent. There are now public properties which enforce locks.
Callers are no longer required to create new copies of lists.
Original request URLs that end with / will still work, but this will allow one to type /simstatus as well as /simstatus/
Can't do this with webstats yet since it does insane things to the path.
Some of the places where agentMS was added were in separate threads launched by the update loop. I don't believe this is correct, since such threads are no longer contributing to frame time.
Some of the places were also driven by client input rather than the scene loop. I don't believe it's appropriate to add this kind of stuff to scene loop stats.
These changes hopefully have the nice affect of making the broken out frame stats actually add up to the total frame time
SP still has an implementation but this is now just a public method on SP rather than an abstract one in EntityBase.
No point making the code more complex until it actually needs to be,
Doing this to see if addresses inventory object deserialization problems in http://opensimulator.org/mantis/view.php?id=5708, though if it does I'm really surprised not to have seen it before now.
Really need to go through and systematically set the culture for every timer and change all BeginInvoke calls to FireAndForget instead.
But don't want to do something like that this close to a release.
When a slider parameter is changed, the viewer uploads a new shape (or other asset) and the item is updated to point to it.
Viewer 1 uploaded the data in the initial request itself, so the asset references was almost always correctly updated.
However, viewer 3/2 always uploads data in a subsequent xfer, which exposed a race condition where the viewer would make the item update before the asset had uploaded.
This commit shuffles the order of operations to avoid this race, the item is updated with the new asset id instead of the old one while the upload was still taking place.
A second race had to be fixed where avatar appearance would also be updated with the old asset id rather than the new one.
This was fixed by updating the avatar appearance ids when the appearance was actually saved, rather than when the wearables update was made.
I thought that I had implemented this but must have accidentally removed it.
Adds a regression test to detect if this happens again.
Temporarily disables automatic landing of NPC at a target. Will be fixed presently.
This stops the npc walking backwards if the target is directly behind.
This means that the npc no longer returns to its original rotation once movement has finished.
If you want this behaviour, please store and reset the original rotation after movement.
This is somewhat to address http://opensimulator.org/mantis/view.php?id=5678
When upgrading the previously child agent to a root, the code was setting the Size parameter on the ODECharacter PhysicsActor.
This in turn reset Velocity, which cause the border stall.
I'm fixing this by commenting out the Velocity = Vector3.Zero lines since they don't appear to play a useful purpose
Using the RemoteAdmin API to close then recreate a region would fail if the estate name had not changed. If the estate name /was/ changed then the existing estate would be renamed rather than a new one being created. The problem really arose from a lack of distinction in the data storage layer between creating new estates and loading existing ones.
Viewer 2 no longer contains the default avatar assets (i.e. "Ruth") that would appear if the user had insufficient body part/clothing entries.
Instead, avatars always appear as a cloud, which is a very bad experience for out-of-the-box OpenSim.
Default is currently off. My intention is to switch it on for standalone shortly.
This is not particularly flexible as "Ruth" is hardcoded, but this can change in the future, in co-ordination with the existing RemoteAdmin capabilities.
Need to fix creation of suitable entries for users created as estate owners on standalone.
Avatars still appear with spooky empty eyes, need to see if we can address this.
This commit adds a "Default Iris" to the library (thanks to Eirynne Sieyes from http://opensimulator.org/mantis/view.php?id=1461) which can be used.
If we do this, then viewer 3 crashes when we try and rez a script directly in an attachment's prim inventory.
Sending an empty file name was already being done if the prim's inventory had never been touched.
Now we always do that if there are no items in that inventory.
Hopefully addresses the remaining point in http://opensimulator.org/mantis/view.php?id=5644
Viewer 2/3 will sometimes attempt to rewear attachments, even though they have already been attached during the main login process.
This change ignores those attempts.
This stops script failures during login, as the rewearing was racing with the script startup code.
It might also help with attachments being abnormally put into deleted state.
Hopefully resolves some more of http://opensimulator.org/mantis/view.php?id=5644
As far as I can see, this is only invoked by a PUT request to ObjectHandlers, which is not being used anyway.
Invoking attachments code at this point is probably inappropriate since it would still be invoked when the client entered the scene.
Being commented to simplify analysis of attachments issues. Can be uncommented when in use.
Also, small tweak to lock and log removal of a SOG from the SceneObjectGroupsByLocalPartID collection in SceneGraph.GetGroupByPrim() if an inconsistency is found.
This is the message sent to the client when the object is returned.
We were sending byte[0] in the binary bucket. This didn't kill viewer 1 but did terminate viewer 3 (don't know about viewer 2).
So sending "\0" instead.
This is to address http://opensimulator.org/mantis/view.php?id=5683
Attach and detach packets are processed asynchronously when received from a viewer.
Bugs like http://opensimulator.org/mantis/view.php?id=5644 indicate that in some situations (such as attaching/detaching entire folders of objects at once), there are race conditions between these threads.
Since multiple data structures need to be updated on attach/detach, it's not enough to lock the individual collections.
Therefore, this commit introduces a new IScenePresence.AttachmentsSyncLock which add/remove operations lock on.
Leaving them at UUID.Zero meant that when a viewer 2 logged into a region that had been freshly created, it received UUID.Zero for these textures, and hence display the land as plain white.
On a simulator restart, the problem would go away since when the database adapators loaded the new region settings, RegionSettings itself has code to use default textures instead of UUID.Zero.
This commit resolves the problem by saving the default texture UUIDs instead of Zero.
However, we currently have to do this in a roundabout way by resaving once the RegionSettings have been created by the database for the first time. This needless complexity should be addressed.
This change will also have the effect of replacing any existing UUID.Zero terrain textures with the default ones.
However, this shouldn't have any effect since the UUID.Zeros were already being replaced in memory with those same UUIDs.
This is only called by a region console command.
We should also be locking m_partsUpdateQueue when dequeueing the next part, or locking m_pendingObjects in QueuePartForUpdate().
However, I won't do this now since I don't have time to analyze how this would affect liveness.
Strictly speaking, we could also stop bothering to clear the m_updateTimes and m_partsUpdateQueue if we are sure that the whole SceneViewer is shortly to be garbage collected anyway, but we'll leave them around for now.
The code in question is over three years old and just be catching an inconsistency rather than being wholly necessary.
This commit still carries out the check and prints all the previous log warnings but a 'failure' no longer prevents avatar region crossing or teleport, and it doesn't give the client the error message.
This will have some kind of impact on http://opensimulator.org/mantis/view.php?id=5672
This was happening because we were using the source avatar's item IDs in the clone appearance.
Switch to using the asset IDs of attachments instead for NPCs.
The InventoryAccessModule and AttachmentModule had to be changed to allow rezzing of an object without an associated inventory item.
Hopefully goes some way towards resolving http://opensimulator.org/mantis/view.php?id=5653
This method wasn't actually doing anything since dropped attachments retain a PCode of 9.
Also, behaviour of dropped attachments in other places appears to be that they persist after avatar logout rather than get deleted.
Even though we don't use these on rez they are still present after an unlink, after which selecting them causes various viewers to crash
Hopefully really does address http://opensimulator.org/mantis/view.php?id=5664
It's never possible for SOG to have no RootPart, except in the first few picosends of the big bang when it's pulled from region persistence or deserialized
The only times when ParentGroup might be null is during regression tests (which might not be a valid thing) and when scene objects are being constructed from the database.
At all other times it's not possible for a SOP not to have a SOG parent.
The approach here, as in other parts of OpenSim, is to return a copy of the list rather than the attachments list itself
This prevents callers from forgetting to lock the list when they read it, as was happening in various parts of the codebase.
It also improves liveness.
This might improve attachment anomolies when performing region crossings.
On making a root agent, we need to reset the ScenePresence.m_movement_flag so that it doesn't remember the
movement registered to the client when it exited the initial region.
If this is remember, then the client avatar movement isn't updated and it appears to stall in mid-air, though this is resolved with a prod/release of any other direction key.
This bug was probably introduced a few weeks ago. Surprised that nobody brought it up.
Apart from one obvious bug, this was failing because attempting to serialize the script from inside the script (as part of saving the attachment as an inventory asset) was triggering an extremely long delay.
So we now don't do this. The state will be serialized anyway when the avatar normally logs out.
The worst that can happen is that if the client/server crashes, the attachment scripts start without previous state.
llRegionSay will now message avatars on chan 0
and will message attachments on the avatar that
listen on channels other than 0.
This behavior is consistant with the LL
implementation as tested on regions in Agni
with one exception: this implementation does
not include issue:
https://jira.secondlife.com/browse/SCR-66?
This meant punching in another AddUser() method in IUserManagement to do a direct name to UUID associated without the account check (since NPCs don't have accounts).
May address http://opensimulator.org/mantis/view.php?id=5645
This is done by introducing a PresenceType enum into ScenePresence which currently has two values, User and Npc.
This seems better than a SaveAttachments flag in terms of code comprehension, though I'm still slightly uneasy about introducing these semantics to core objects
Default for this function is now not to automatically land.
This allows better control by scripts when an avatar is going to be landing on a prim rather than the ground.
Stopping the avatar involves faking a collision, to avoid the pid controller making it overshoot.
A better approach would be to gradually slow the avatar as we near the target
this is to allow walking on prims. it will be up to the script writer to be sure that there is a continuous path.
currently implemented in osNpcMoveToTarget(), but none of this is final.
This works by serializing and deserializing NPC AvatarAppearance to a notecard in the prim inventory and making the required baked textures permanent.
By using notecards, we avoid lots of awkward, technical and user-unfriendly issues concerning retaining asset references and creating a new asset type.
Notecards also allow different appearances to be swapped and manipulated easily.
This also allows stored NPC appearances to work transparently with OARs/IARs since the UUID scan will pick up and store the necessary references from the notecard text.
This works in my basic test but is not at all ready for user use or bug reporting yet.
This doesn't help where the target is a prim surface. In these situations, it might be better to provide manual overrides so the script can control whethre an avatar flys there/lands, etc.
This makes the movement exact. Regression test changed to check avatar reaches exact target.
Also has the nice side effect of making NPC animations continue to work after the first movement (which wasn't working). However, avatar still pauses in mid-stride
Avatar moves and stops. However, will stop in mid stride.
And if the move to position is in the air, avatar will continue to make vain and quite hilarious attempts to take off (but never doing so).
Clearly more work is needed.
This now works again except that it requires a click or avatar mvmt to get going
This is because the ScenePresence.HandleAgentUpdate() method doesn't trigger until the client does something significant, at which point autopilot takes over.
Even clicking is enough to trigger.
This will be improved presently.
This is not used for anything - appearances are always properties of objects with ids (ScenePresence, AgentCircuitData) and just has the potential to get out of sync when the appearance is cloned.
Had to stop using AvatarService for now since it doesn't store baked texture IDs (which is why this was failing).
Also failing because cloning appearance was also cloning the AvatarApperance.Owner field, which we weren't then changing.
Extended TestCreate() to check this.
This is to prevent situations where the first name returned by GridService.GetRegionsByName is not one that exactly matches the given region name, even when there is an exact match later on in the list.
Only the above two functions call this teleport method (the map uses a different routine) so this seems safe to change.
Addresses http://opensimulator.org/mantis/view.php?id=5606
This is to accomodate situations where the authorization service is being used by the hypergrid, where visitors have no user account.
See http://opensimulator.org/mantis/view.php?id=5517, this code is somewhat adapted/cleaned up from Michelle's patch
I'm a little ambivalent about this since visitors could put anything in firstname/lastname so it's not much of an auth measure.
It's up to the auth service to decide which data it actually uses.
Possibly we should be passing through other info such as agent circuit ip
Some items had completely wrong permissions - this is easier than correcting them all.
The ability to set permissions in xml is retained since there are use cases for this (e.g. to create no-mod library scripts)
Library items always need the same permissions, so it doesn't make sense to load them from the xml files. This just opens the door to permissions mistakes.
Fixed this by inspecting Shape.SculptEntry at various places instead of Shape.SculptType. Sculpties actually have a SculptType of Cylinder - only true mesh is SculptType.Mesh
This addresses http://opensimulator.org/mantis/view.php?id=5595
This stops problems when we undo a few steps and start off down another path.
Surprisingly, apart from this now fixed problem, redo appears to be working too.
I think (ha ha) this largely fixes undo, except for the fact that rotation a set of prims with 'edit linked parts' selected doesn't quite work properly (though this works fine if the checkbox isn't selected).
Also, the double undo bug for resize is still present.
Redo might be incredibly buggy, haven't even looked at that yet.
The only obviously broken things right now are the undo of the position of just a root prim (stays in place) and the fact that resizes need two undoes.
However, what happens now is that undo just doesn't do anything when the root prim is selected on its own. This requires more code than just fiddling with undo states.
This involves implementing a boolean in UndoState to signal whether the undo needs to be done for an entire group/linkset or just a single prim
Resizing individual components of linksets is still dodgy.
Resizing still has to be down twice, since for some reason the client is sending two multiobjectupdate packets on every resize except the very first. This applies to single prims and linksets. Need to look into this.
Undo rotation and position appear to be working.
Resizing a single prim appears to be working, though the undo has to be done twice.
Resizing a group of prims still does not work properly - possibly because in the UndoState we don't store a knowledge of when we're resizing a whole group rather than individual prims.
This needs to be addressed.
Also fiddle a bit with undo. This is not currently working properly, though to be fair it also didn't appear to work in 0.7.1.1 either (at least for resize).
Will get some more attention soon.
Unable to get to the bottom of why resizing a mesh fails to properly reset the physics proxy, when toggling phantom does
After a mesh is generated, the existing sculptdata is set to zero in PrimitiveBaseShape to save memory
When phantom is toggled, the sculptdata is regenerated before remeshing.
But on resize, the sculptdata is not regenerated.
So clearly, resetting sculptdata is possible, but haven't quite been able to pin down how this is being done when phantom is toggled.
If a user with a very large inventory right-clicks on their "My Inventory" folder, viewer 1 code will send a massive number of Fetchinventory requests.
Even though each is handled asynchronously via a pool thread, the sheer frequency of requests overwhelms the pool and freezes inbound packet handling.
This change makes the first Fetchinventory thread also handle subsequent requests, freeing up the other threads.
Further efficiencies could be made by handling all the items in a particular FetchInventory request together, rather than separately.
This is to avoid problems with corrupt inventories where an inventory link target points back at the source's folder
No viewer has been observed to set these up as of yet. If this ever happens, we will need a more sophisticated solution to track sent folders within the recursion
This now creates an avatar but appearance is always cloudy.
Move doesn't work.
Really, creating an NPC should only involve a ScenePresence rather than doing anything with IClientAPI, since an NPC has no viewer to communicate with!
Many thanks to the aurora project for pioneering this.
This code is almost certainly not bug free, but it does at least appear to handle simple meshes (except when the viewer crashes - but it is beta!).
This switch stops any assets being saved in the oar.
This can be useful if you're using OAR to backup regions and you know you'll always have the original asset database available.
To get this to work, I had to disable the dupe link check I put in a couple of commits ago.
When the viewer adds wearables to an existing outfit, it first requests deletes of all the existing links before creating a new set.
Since these messages are async, the creates were being received before the deletes had a chance to complete, resulting in missing current outfit links.
However, the dupe check shouldn't be as important now that broken links have been fixed - it was the broken links that were causing the client to create dupes.
Tested on kokua 0.1.0 WIP and SL 2.6.3. I now have no problems managing outfits on my standalone.
It appears that if the viewer requests a folder containing links, we must also send the folders that contain the link targets first.
This was tested with Kokua 0.1.0 WIP though I predict it will also work with other viewer 2s
I believe this is safe since there is a 1-1 correspondence between link item and worn item (i.e. you can't be wearing the same item at two spots simultaneously in one outfit).
This should stop lots of duplicate links being created when viewer 2 is used.
However, this doesn't prevent broken inventory links, which I believe is timing related since the effect is not consistent (e.g. keep relogging and the viewer should end up seeing them correctly) . I think we actually see this problem on viewer 1 as well.
It might be easier just to implement the Fetch*2 inventory caps which are documented at http://wiki.secondlife.com/wiki/Inventory_API. WebFetch* has been deprecated by Linden Lab since viewer 2.5.1 and according to the sl wiki, "has numerous bugs".
Addresses http://opensimulator.org/mantis/view.php?id=5444
Fix is to stop the asset transaction calling UpdateInventoryItem() since the caller is doing it anyway, which is more correct.
This did not effect scripts.
and currently used for all of an avatars attachments by the other
policies. Also changed the way items are pulled from the update queues
to bias close objects even more.
Justin Clark-Casey (justincc)