aboutsummaryrefslogtreecommitdiffstatshomepage
path: root/OpenSim/Region/Physics/OdePlugin (follow)
Commit message (Collapse)AuthorAgeFilesLines
...
* move geom/actor map maintenance into ↵Justin Clark-Casey (justincc)2011-11-212-10/+9
| | | | | | DestroyODEStructures()/AvatarGeomAndBodyCreation(). This saves us having to do it separately when a character capsule size is changed
* When changing avatar size in ODE, remove the old actor from the name and ↵Justin Clark-Casey (justincc)2011-11-212-8/+14
| | | | actor maps
* When an ODECharacter is removed (e.g. when an avatar leaves a scene), remove ↵Justin Clark-Casey (justincc)2011-11-212-10/+23
| | | | | | the actor reference in OdeScene.actor_name_map rather than leaving it dangling. This also largely centralizes adds/removes in OdeScene.AddCharacter()/RemoveCharacter()
* Have ODECharacter and ODEPrim both use PhysicsActor.Name instead of ↵Justin Clark-Casey (justincc)2011-11-212-9/+5
| | | | maintaining their own properties
* Instead of generating a new list for bad characters on every physics pass, ↵Justin Clark-Casey (justincc)2011-11-211-3/+12
| | | | keep reusing the same list.
* Stop OdePrim and OdeCharacter insanely overriding set LocalID to set their ↵Justin Clark-Casey (justincc)2011-11-165-35/+19
| | | | | | | own private m_localID property but leaving get to return the then unset PhysicsActor.LocalId! Instead, just have both subclasses use the PhysicsActor.LocalID property. This restores collision functionality that fell away in 45c7789 yesterday
* Instead of having scene add/remove collision events directly to the OdeScene ↵Justin Clark-Casey (justincc)2011-11-151-20/+38
| | | | | | | | | collision event dictionary, marshall them via a change dictionary first. 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.
* use a more efficient dictionary in OdeScene._collisionEventPrim rather than ↵Justin Clark-Casey (justincc)2011-11-151-7/+4
| | | | a list
* Add comment for experimental effect of removing the Thread.Sleep(20) in ↵Justin Clark-Casey (justincc)2011-11-042-1/+4
| | | | ODEPrim.changevelocity()
* Stop llPushObject() from causing problems by adding force via a taint rather ↵Justin Clark-Casey (justincc)2011-11-041-2/+25
| | | | | | | than directly. 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.
* Add taint target velocity for ODECharacters as is already done for ↵Justin Clark-Casey (justincc)2011-10-291-7/+12
| | | | | | | ODECharacter position and position and velocity for ODEPrims. 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
* tidy up OdeCharacter so that we just use OpenMetaverse.Vector3 assignment ↵Justin Clark-Casey (justincc)2011-10-291-51/+39
| | | | | | directly where possible, instead of transferring X, Y and Z components separately some of this is probably a hold over from using ODE.Vector3, which is still necessary in some places.
* Move position set from taint to logically better position at top of ↵Justin Clark-Casey (justincc)2011-10-291-12/+12
| | | | ODECharacter.ProcessTaints() though this makes no practical difference
* Stop setting _position as well as m_taint_position in ODECharacter.PositionJustin Clark-Casey (justincc)2011-10-291-4/+1
| | | | | setting position at the same time as taint appears to undermine the whole purpose of taint testing doesn't reveal any obvious regressions in doing this
* Fix a bug I introduced yesterday in ODE physics where prim scripts would ↵Justin Clark-Casey (justincc)2011-10-271-1/+22
| | | | only receive the very first collision.
* Clear OdeCharacter CollisionEventUpdate when we subscribe or unsubscribe ↵Justin Clark-Casey (justincc)2011-10-251-0/+2
| | | | from collision events
* Remove unused fields from CollisionEventUpdateJustin Clark-Casey (justincc)2011-10-252-2/+2
|
* When sending object collision updates, don't null out and recreate the ↵Justin Clark-Casey (justincc)2011-10-251-13/+3
| | | | CollisionEventUpdate() if the number of collisions falls to zero. Reuse the existing one instead.
* For ScenePresence collision events, instead of creating a new ↵Justin Clark-Casey (justincc)2011-10-251-5/+3
| | | | | | CollisionEventsThisFrame every time we need to send some new ones, reuse the existing one instead. This assumes that the listener is using the data synchronously, which is currently the case.
* Fix bug where collision event listeners were not removed once the listener ↵Justin Clark-Casey (justincc)2011-10-251-4/+5
| | | | | | | had gone away. This was causing continuous use of temporary memory even when all avatars had left the scene. Memory does leak but it does cause more calls to the garbage collector, which would pause the scene thread for a very short while during collection.
* minor: rename a parameter in OdeScene.Simulate() from actor -> prim since ↵Justin Clark-Casey (justincc)2011-10-251-4/+4
| | | | it's an OdePrim
* very minor removal of old commented out line of code in OdeSceneJustin Clark-Casey (justincc)2011-10-211-5/+3
|
* remove unnecessary null check on _collisionEventPrimJustin Clark-Casey (justincc)2011-10-201-3/+0
|
* Get OdeScene to use passed in time step rather than hard-coded 0.089Justin Clark-Casey (justincc)2011-10-201-22/+23
| | | | However, I still don't recommend changing MinFrameTime from 0.089, high values do not work well and lower values don't seem to make much difference
* For now, stop passing timeStep into methods where it's not actually used.Justin Clark-Casey (justincc)2011-10-202-13/+5
|
* Temporarily put in log lines to record time taken to set terrain in OdeScene.Justin Clark-Casey (justincc)2011-10-181-0/+6
|
* Store scene identifier passed in to OdeScene for later debug messagesJustin Clark-Casey (justincc)2011-10-181-2/+5
|
* Don't bother taking OdeLock during OdeScene construction, since there can be ↵Justin Clark-Casey (justincc)2011-10-181-16/+13
| | | | no contention until the object is constructed.
* Remove unnecessary lock of OdeLock in OdePrim.changeadd()Justin Clark-Casey (justincc)2011-10-181-18/+13
| | | | This taint can only ever be processed from the OdeScene.Simulate() loop, which already locks OdeLock.
* minor: method doc to explain a lock of OdeLockJustin Clark-Casey (justincc)2011-10-182-1/+3
|
* Remove the unused CollisionLocker from ODEJustin Clark-Casey (justincc)2011-10-184-97/+81
| | | | Despite its name, this wasn't actually being used in any collision checking
* Implement osNpcSit(). This is still in development so don't trust itJustin Clark-Casey (justincc)2011-10-171-0/+1
| | | | | | | | | 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 :)
* reduce access to ODECharacter methods to make code analysis easier. ↵Justin Clark-Casey (justincc)2011-10-152-54/+36
| | | | Eliminate redundant argument on ProcessTaints()
* restrict unnecessary access levels on ODEPrim fields/propertiesJustin Clark-Casey (justincc)2011-10-151-19/+17
|
* refactor: make methods that do not need to be public in ODE private or ↵Justin Clark-Casey (justincc)2011-10-152-128/+195
| | | | internal to aid code reading/analysis. Remove some unused method arguments
* Change hardcoded ODE total frame time to match the default total frame time ↵Justin Clark-Casey (justincc)2011-10-141-1/+1
| | | | | | | (0.09375 -> 0.089). No apparant ill effects - because the default stepsize is 0.2, there are still 5 physics steps per physics frame. This is a precursor to using the elapsed value passed in (and now changeable in config).
* More method doc and formatting changes. Makes DestroyOdeStructures() privateJustin Clark-Casey (justincc)2011-10-131-6/+10
|
* refactor: move 3x copy/pasted ode structure removal code in ODECharacter ↵Justin Clark-Casey (justincc)2011-10-131-90/+57
| | | | | | into a DestroyOdeStructures() method also adds some method doc
* Remove unused local variable i from OdeScene.Simulate()Justin Clark-Casey (justincc)2011-10-051-3/+0
|
* adjust some whitespace to trigger another build, to check the last failure ↵Justin Clark-Casey (justincc)2011-09-171-2/+2
| | | | was just a glitch
* Stop the avatar stalling on its first boarder cross when using the ODE pluginJustin Clark-Casey (justincc)2011-09-161-6/+13
| | | | | | 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
* If SP.MoveToTarget has been called with a force walk, begin by landing the ↵Justin Clark-Casey (justincc)2011-08-111-1/+5
| | | | | | avatar. There is a bug here - once an avatar has landed it glides to its new position instead of performing a walk animation
* Add a OS_NPC_LAND_AT_TARGET option to osMoveToTarget()Justin Clark-Casey (justincc)2011-08-101-0/+2
| | | | | | | 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
* Move common gemo/agent map name code into CreateGeom()Justin Clark-Casey (justincc)2011-08-011-11/+5
| | | | Fix build break.
* Instead of preserving old name in geom_name_map in change size or shape, use ↵Justin Clark-Casey (justincc)2011-08-011-7/+3
| | | | | | the Name property instead. This is equivalent since the prim 'name' is never changed. In fact, this propery is never used for prims
* comment out unused code in OdeScene.TriCallback()Justin Clark-Casey (justincc)2011-08-011-12/+12
|
* On geom removal, remove the name from the OdeScene.geom_name_map tooJustin Clark-Casey (justincc)2011-08-011-0/+1
|
* remove the unused ODEPrim.prev_geom fieldJustin Clark-Casey (justincc)2011-08-011-3/+0
|
* When an ODE geom is removed (as when a non-phantom prim is deleted, resized ↵Justin Clark-Casey (justincc)2011-08-011-0/+2
| | | | | | or shape changed, also remove the OdeScene.actor_name_map entry pointing to the phys actor This is to stop a small memory leak over time when prims are deleted or phantom-toggled
* refactor: centralize prim geom removal code from four places to oneJustin Clark-Casey (justincc)2011-08-012-43/+44
|
generated by cgit v1.1 at 2024-11-01 08:07:46 +0000