* Deleting a map is now routed through MapComponent deletion.
* Remove map and grid deletions from map networking, just delete the entity if you want to remove the map.
* Moved the chunkSize property of created grids from the networked MapData to the MapGridComponent state.
* Remove unused IMapManager.DefaultMap property.
* Removed MapCreationTick field from MapManager.MapCollection.
* Removed _maps hashset field from MapManager.
* Removed CreatedMaps array from network MapData.
* MapGrid.ParentMapId is now derived from the bound TransformComponent, and isn't required in the MapGrid ctor.
Removed MapGrid.CreatedTick, it can be found on MapGridComponent.CreationTick.
* Remove a bunch of ApplyGameStatePre code duplication.
* Completely refactored CreateGrid.
* Adds AddComponentUninitialized to the ECS system. This allows you to access a component before it is initialized in a using block.
* Use AddComponentUninitialized to allocate a grid after the component is allocated.
* MapLoader now creates the grids after creating the map entities.
* Chunksize and TileSize properties are now actually read out of the map yaml.
TileSize has a public Setter.
* Minor cleanup.
* Moved grid allocation onto the MapGridComponent.
* Final Cleanup.
* Merge Fail.
* Fixed test, grid was getting deleted because it was empty.
* Remove DeletedChunkDatum from grid networking.
* ApplyMapGridState moved from map manager to the MapGridComponent.
Simulation input and Update() does not happen when prediction is disabled. Both of these can be re-opted in on a per-handler/system basis with a bool flag. Stuff like physics opts out of this now.
* ComponentNames are not sent over the network when components are created.
* Removed ComponentStates array from EntityState, now the state is stored directly inside the CompChange struct.
* Remove the unnecessary NetID property from ComponentState.
* Remove Component.NetworkSynchronizeExistence.
* Change GetNetComponents to return both the component and the component NetId.
* Remove public usages of the Component.NetID property.
* Adds the NetIDAttribute that can be applied to components.
* Removed Component.NetID.
* Revert changes to GetComponentState and how prediction works.
* Adds component netID automatic generation.
* Modifies ClientConsoleHost so that commands can be called before Initialize().
* Completely remove static NetIds.
* Renamed NetIDAttribute to NetworkedComponentAttribute.
* Fixing unit tests.
FullState was getting updated before ResetPredictedEntities ran instead of after. So ResetPredictedEntities was applying data to stuff like containers that depends on entities that hadn't been made yet.
* Refactors EntityManager to not do any networking.
ServerEntityManager and ClientEntityManager now do the networking instead.
* Rename property for "backwards compat."
* Remove comented out code in robust server simulation
* Refactors MapManager to not do any networking.
Now, ServerMapManager and ClientMapManager handle any networking.
* it's christmas in april!
* Remove comented line
* Remove useless seal
* Fix incorrect semicolon
* Event is no longer overriden, has a protected Invoke method instead
* Split entity lookups from entitymanager
* Helps if you subscribe dingus
* Handle map changes
* Stacks instead
* Make mapchanges use a queue because it's probably better
Moves likely only care about the latest position
* IoC what you did there
* IoC refactor
* Minor optimisations
* Apply feedback
* My IQ dropped 3 sizes that day
* Rest of acruid's feedback
* final_no_actual commit
* enlightenment?
* Liftoff
* final_commit_v2_actual
Co-authored-by: Metal Gear Sloth <metalgearsloth@gmail.com>
* Added new ComponentEventBus, combined it with IEventBus.
* Removed all traces of IEntity from ComponentDependencies.
Removed IEntityManager dependency from ComponentManager.
* Added entity create/delete events to IEntityManager.
* ComponentEvents now use EntitySystemMessages instead of their custom ComponentEvent class.
* Component events are now just overloads of entity events.
* Removed obsolete EntitySystemMessage, now everything uses the base EntityEventArgs.
* Add a bool argument for if the message should be broadcast as well as directed.
Fix ordering and init issues of events in EntityManager.
* Changed names from Component/Entity events to Directed/Broadcast.
* Fix bugs and unit tests.
* Removed the Interfaces folder.
* All objects inside the GameObjects subfolders are now in the GameObjects namespace.
* Added a Resharper DotSettings file to mark the GameObjects subfolders as not providing namespaces.
* Simplified Robust.client.Graphics namespace.
* Automated remove redundant using statements.
1. Set timeBeginPeriod(3) on the server to reduce scheduler latency in the lidgren thread.
2. Add 16ms of guaranteed lag bias to client prediction calculations to account for scheduler latency.
Both of these changes are to account for how the windows scheduler seems to handle time periods in related to socket polls. See this Discord conversation for why, details down below as well: https://discord.com/channels/310555209753690112/770682801607278632/798309250291204107
Basically Windows has this thing called time periods which determines the precision of sleep operations and such. By default it's like 16ms so a sleep will only be accurate to within 16ms.
Problem: Lidgren polls the socket with a timeout of 1ms.
The way Windows seems to handle this is that:
1. if a message comes into the socket, the poll immediately ends and Lidgren can handle it.
2. If nothing comes in, it takes the whole 16ms time period to actually process stuff.
Oh yeah, and Lidgren's thread needs to keep pumping at a steady rate or else it *won't flush its send queue*. On Windows it seems to normally pump at 65/125 Hz. On Linux it goes like 950 Hz as intended.
Now, the worst part is that (1) causes Lidgren's latency calculation to always read 0 (over localhost) instead of the 30~ms it SHOULD BE (assuming client and server localhost).
That 30ms of unaccounted delay worst caseis enough to cause prediction undershoot and have messages arrive too late. Yikes.
So, to fix this...
On the server we just decrease the tick period and call it a day. Screw your battery life players don't have local servers running anyways.
On the client we bias the prediction calculations to account for this "unmeasurable" lag.
Of course, all this can be configured via CVars.
* Fix client-side input system not sending sequence numbers for input messages.
This caused message ordering issues and keys getting stuck down.
* Make sure EyeUpdateSystem runs after physics.
* IGameTiming.TickFraction helper.
* Subtick data for input commands.
* Literally a unit test to verify that I wasn't going insane while debugging input message ordering issues.
* More prediction logs behind net.predict.
* Move physics to shared and run it on the client.
* Synchronize grid gravity.
* Fix ResetPredictedEntities crash when entities get deleted server-side.
* Fix CollidableComponent setters not calling Dirty()
* Fix going into space while predicting.
* Watch window uses history line edit.
* Fix unpredicted objects stuck-jittering.
* VV tags for interp vars on transform.
* Fix 0 mass objects on client.
* Fix friction calculations being TPS dependent.
* Reset predict flag on handelComponentState
update a bunch of packages and use JetBrains.Annotations as private asset as needed
mark some hot math methods agg inline to benefit from loop opts
use FMA for interp
use canny min/max/clamp
make Quaternion NormalizeAngle fixed time and faster
clean up YamlDotNet references
