Added callbacks for shape and body changes in GetBodyAndShape() so the linkset
constraints can be picked up and restored. A better design might be to have
a "prim shape changed" event. Think about that.
Added constraint types to general constraint class.
pointers at runtime and at taint-time. Now passes the body into the
taint.
Vehicles zero inertia when active to eliminate Bullet's contribution
to vehicle motion.
Static objects are set to ISLAND_SLEEPING rather than DISABLE_SIMULATION.
Might reconsider this and, alternatively, have dynamic objects force activation.
Clean up use of DetailLog().
Rearrangement and cleanup of shape collection code. Much more readable.
Enabling and use of collision filters and masks.
Addition of ID to body creation BulletSimAPI calls so always set in
shape for collision reporting.
Change default of ShouldSplitSimulationIslands and ShouldRandomizeSolverOrder
from 'false' to 'true'. When 'false', this suppresses NO_CONTACT_RESPONSE
which makes volume detect fail.
so the members case is consistant. Caused modifications everywhere.
New logic in BSShapeCollection to track use and sharing of shapes.
I just reslized, though, that shapes cannot be shared because the
shape's UserPointer is the localID of the prim and is required
for tracking collisions. More changes coming.
Added DuplicateCollisionShape2() to API and changed BuildNativeShape2
to take a ShapeData structure so don't have to pass so many parameters.
This matches the latest version of BulletSim.dll.
Additions and removal of DetailLog() statements for debugging.
Fix avatar height calculation to properly account for the capsule ends.
Rearrange some locking in TerrainManager to eliminate possible race conditions.
Move DetailLog() definition into common BSPhysObject class.
Some variable renaming to make usage clearer (refactor.rename makes this so easy).
and into the parent class BSPhysObject.
Rework collision logic to enable extra collision after done colliding.
Rename 'Scene' to 'PhysicsScene' to differentiate it from the simulator 'Scene'.
Fix collision code to properly sense mega-region children regions as terrain.
When setting an object physical, reset all the physical properties (friction, ...).
Many changes to BSDynamic for readability and commentary.
Linkset hacking for vehicles: don't over mass the root prim.
Add parameter for link constraint solver iterations.
Correct uses of timestep in timescale calculations for vehicles.
Reorganize code/logic for making objects static and dynamic for readability
and use of API2.
Changed most calls in BSPrim to use API2 calls (the new way).
Avatars do not generate default Bullet collision events but do call up
to the simulator for every avatar. Reduces overhead.
Objects added to collision list only if they are processing collisions.
Reduces overhead especially for large numbers of avatars.
Generalize call for water height to GetWaterHeightAtXYZ().
Catch and correct exception getting terrain height when out of bounds.
Correct race condition in Terrain Manager where creation wasn't at taint-time.
Add API calls for constructing compound shapes.
Move NeedsMeshing() logic into object class.
Reorganize logic for object meshing to reduce rebuilding of meshs/hulls.
'Body' to 'BSBody' for disambiguation when reading code.
Complete the API2 interface so nearly all methods on bullet
classes are available to the managed code. The efficient
single call simulation step is kept in place while all
other creation/destruction/parameterization can be done
in the managed code.
Major rework of terrain management which finally makes mega-regions work.
Update heightmap of terrain by rebuilding the terrain's body and shape.
There is a problem with just replacing the shape so this workaround
will do for the moment but it will need to be resolved for
mesh and hull switching.
Update DLL API for new terrain and shape/body pattern methods.
Terrain creation and modification uses new shape/body pattern.
Move debug logging callback set to initialization call so logging
is per physics engine.
Moved all terrain code out of BSScene and into new BSTerrainManager.
Added logic to manage multiple terrains for mega-regions.
Added new functions to BulletSimAPI to match the library.
Moved all of the terrain creation and setup logic from C++ code to C# code.
The unused code has not yet been removed from either place. Soon.
Moved checks for avatar above ground and in bounds into BSCharacter.
no effect on non-logging running.
Capture region name that is passed to the physics engine and use
it for detail logging file name prefix.
Fix problem with avatars dropping when flying across region boundries.
Update BulletSimAPI with new constraint related function calls.
Reorganize locking in BS6DofConstraint.
Update BS6DofConstraint to do constraint reset correctly.
Add new 'midpoint' construction of 6Dof constraint.
debugging.
Linkset bugs fixed where accounting of children would get lost.
Moved scene based vehicle tracking logic from prim to the scene.
Added GetCollisionFlags2 method to BulletSimAPI.
Updated DLLs and SOs.
6dof constraint but eventually others will be exposed so future
features can use all the Bullet capabilities.
Force children to generate a position update when unlinked.
This also fixes computation of avatar mass.
Added parameter MaxPersistantManifoldPoolSize.
Fixed a parameter setting bug which caused crashes of there were
more than 400 or so physical objects. I tested up to 5000.
Updated BulletSim DLLs and SOs.
themselves. For that, the physics module simply calls RequestAssetMethod, which
in turn points to Scene.PhysicsRequestAsset. This gives physics access to
the asset system without introducing unwanted knowledge of the scene class.
Possibly this was for a feature that was never implemented or was otherwise removed.
Thanks to SignpostMarv for the spot of the warning that shows this parameter was never changed.
and into its own class. The BulletSim data structures track
individual prims as linksets of 1 so most of the prim code is not
different between a linked and unlinked object.
Physical linksets are fully functional.
Tweeking of the vehicle code to make it semi-work.
Utilize the new API2 for some setting operations.
Add GetOrientation() API call for proper reporting of children of linksets.
Changes the interface between C# and C++ code so old DLLs won't work!
with an ini configuration parameter.
Correct computation of relative offsets of children in a linkset.
Remove a prim from any link relationship before deleting it.
Minor code flow cleanups.
This occurred in b18c8c8 (Thu May 17 2012).
This was a cause of very occasional race conditions and likely memory leakage as clients came and went from the region.
These will act as a sanity check with the main scene stats, to show that physics scene entities are being managed properly.
Total prims will not match scene total prims since physics total does not include phantom prims
Also zeros collisions scores on all prims after report collection, not just the top 25.
As before, this collision scores are only reset after a report is requested, which may give unrealistic numbers on the first request.
So to see more realistic scores, ignore the first report and then refresh the request after a couple of seconds or so.
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.
Since this is done directly from ScenePresence, it can lead to a race condition with the simulator loop.
There's no real point doing it anyway since the clear will be done very shortly afterwards by the simulate loop and either there are no events (for a new avatar) or events don't matter (for a departing avatar).
This matches existing behaviour in OdePrim
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.
prim_geom == IntPtr.Zero only before a new add prim taint is processed (which is the first taint) or in operations such as scale change which are done in taint or under lock.
Therefore, we can remove these checks which were not consistently applied anyway.
If there is a genuine problem, better to see it quickly in a NullReferenceException than hide the bug.
It seems that ODE calls the avatar collision handling routine even
if there are no collisions. This causes the animation to be updated.
So, for instance, going from HOVER to FLY is caused by the physics engine
calling the collision routine each frame with 0 collisions.
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.
In over 4 years this never progressed beyond an unimplemented stub.
This doesn't mean that it can't come back if someone is interested in implementing PhysX support.
In theory, this means that a 64-bit Windows OS user can now run OpenSim.exe with ODE and use more than 2 (or 3) GB of memory.
However, this is completely untested since I don't currently own a 64-bit Windows box. Feedback appreciated.
Using OpenSim.32BitLaunch.exe should continue to work. Other platforms are unaffected.
This will currently not work with sqlite - I will add that too if this works.
If velocity reaches 256 in any vector then bad things happen with ODE, so we now clamp this value.
In addition, a falling avatar is clamped by default at 54 m/s, which is the same as a falling skydiver.
This also appears to be the value used on the linden lab grid.
This should resolve http://opensimulator.org/mantis/view.php?id=5882
It turns out that calls to Collide() are not thread-safe even for objects in different ODE physics worlds due to ODE static caches.
For simulators running multiple regions, not serializing calls from different scene loops will sooner or later cause OpenSim to crash with a native stack trace referencing OBBCollider.
This affects the default OPCODE collider but not GIMPACT. However, GIMPACT fails for other reasons under some current simulator loads.
ODE provides a thread local storage option, but as of ODE r1755 (and r1840) DLLs compiled with this crash OpenSim immediately.
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 is to avoid a complicated tri-thread deadlock on region crossing for avatars with attachments, where
1) XEngine starting up scripts can lock XEngine.m_Scripts and then try to lock OdeScene._collisionEventPrim while starting up a script due to avatar border crossing
2) An existing collision event will lock OdeScene._collisionEventPrim and then try to lock SP.m_attachments while trying to send the collision event to attachments
3) The avatar still entering the region will lock SP.m_attachments and then try to lock m_Scripts to start more attachment scripts.
This isn't a perfect solution since there can be a race between the taint processing and taint setting, as force needs to be reset after processing.
Needs careful locking in the future.
This is to help stop surprises if the velocity is set in the middle of physics calculations, though this probably isn't a huge problem.
It's more for consistency and for the next step of removing some scene locks
Robert Adams