userDaniel Pcancel

3,070 Commits over 730 Days - 0.18cph!

1 Hour Ago
Merge: from mt_snapshot_saving - Bugfix for ExactArrayPool not being thread-safe Tests: TestMTSave
1 Hour Ago
Bugfix: rewrite ExactArrayPool to be a ConcurrentDictionary Was not MT safe for Jobs 4 parallel serialization as it could lazy initialize Tests: TestMTSave
2 Hours Ago
Update(tests): make SendEntityUpdates perf test spawn as many recievers as threads available Allows to fully saturate the CPU in a test, giving a bit fairer comparison between Jobs 3 and 4 Tests: ran new test
3 Hours Ago
Merge: from mt_snapshot_saving - Optim: in UsePlayerUpdateJobs 4 entitiy saving for network is done in parallel. Use server.ParallelNetworkQueueBatchSize to tweak number of tasks Tests: 2k procgen, built a base and a boat, rode vehicle, heli, zipline, boat, horse, fly across the world. ran all TestMTSave unit tests
3 Hours Ago
Clean: fixed whitespacing Tests: none, trivial change
3 Hours Ago
Merge: from main
4 Hours Ago
Buildfix: add missing server guard Tests: compile sim
4 Hours Ago
Update: add an default TransformHandle check to TransformEx.Unsafe.ExtractTranformAccess Better we lose a smidge of perf than have a native crash Tests: ran all TestMTSave
4 Hours Ago
Update: add GetPosMT and GetRotMT to TransformEx - bugfix Elevator Save accessing transform chcked internals, should be safe Tests: fly across all of 2.5k procgen, no errors triggered
4 Hours Ago
Bugfix: avoid resolving transformHandle on an optional BatteringRam.doorPhysicsHinge Checked other cases I've modified, rest are required Tests: none, trivial change
4 Hours Ago
Bugfix: resolve game controlelr at init of BaseCardGameEntity - use cached controller instead of trying to lazy create it in Save Avoids invoking non MT-safe APIs in save Tests: visited bandit camp on procgen, no errors. Ran TestMTSave tests
5 Hours Ago
Bugfix: avoid NRE if bike doesn't have a sidecar Tests: spawned a motorbike
5 Hours Ago
Update(tests): TestServer now also creates a dummy BiomeMap Not sure why pre-merge TestMTSave(GrowableEntity) didn't trip over, but sorted now Tests: all TestMTSave
5 Hours Ago
Merge: from main
6 Hours Ago
Clean: add comments for new perf test Tests: none, trivial change
6 Hours Ago
New(tests): SendEntityUpdates perf test, creates a bunch(100, 1000, 10000) of players that get sent to 10 players Runs Jobs 3(parallelize dependency chains) vs Jobs 4(parallelize queues) serialization modes and whether to use network cache. Doesn't create any entity hierarchy (best case for Jobs 3). With network cache, Jobs 4 beats by ~2x (10k: 16ms vs 9ms), without cache Jobs 4 scales much better, up to 3x faster (10k: 62ms vs 14ms). Tests: ran SendEntityUpdates
Today
Update(tests): assign prefabIDs to entities created in TestServer Avoids tests derping out trying to save global managers/players Tests: SendEntityUpdates perf test passes
Today
Bugfix: fix double-free on a pooled buffer This was causing a buffer to be shared between 2 tasks when it was fetched from the pool, leading to chaos Tests: ran SendEntityUpdates perf test (gonna submit in a bit)
Today
Update: Jobs 4 - intercept exceptions from SendAsSnapshot/ShouldNetworkTo and log once per task - Also if ToStream threw, don't save the resulting buffer as a network cache, as that would poison everything Instead of bubling them all up to the top like usual, handle locally by just skipping sending the data, and report error once to avoid filling console Tests: booted into craggy with CCTV fix reverted, no more spam
Yesterday
Optim: Implement parallel serialization in Jobs 4 mode - fix CCTV_RC doing GetComponentInParent on an MT path, cache on init - add server.ParallelNetworkQueueBatchSize, controls how many players end up being batched in a task. Higher number - less tasks Offloading ShouldNetworkTo, serialization and sending all to worker threads. Main thread steals last task to run on main thread before spinning for other task completion. Tests: booted craggy, gathered resources, build a base, rode ahorse, shot a gun, drove a car, heli, zipline, built a boat, used codelock
Yesterday
Update: rewrite ToStreamForNetwork to be thread safe - EntityMemoryStreamPool is now a concurrent queue Now, theoretically, we just throw it at threads and drink champagne Tests: craggy, built a base and checked decay timer
Yesterday
Update: initialize every SaveInfo with ThreadSafeTime.TakeSnapshot - also removed some jobs 2 code I missed previously Tests: on craggy rode a horse, built a base, checked decay timer was updating as I added/removed resources
Yesterday
Bugfix: avoid multiple BuildingPrivilege.CalculateProtectedMinutes calls in a frame for 1 entity Atomic latch wasn't correctly checking if it was updating from an old value only, so we could CompareExchange(1,1,1) and pass Tests: none
2 Days Ago
Update: avoid Time calls in BasePlayer.GetUnderwearSkin - UnderweatManifest now initialized in ServerMgr.StartServer Tests: switched to jobs 4 and toggled underwear
2 Days Ago
Update: in UseParallelSaves mode, Ragdol component samples transform handles Tests: on playground enabled Jobs 4 and ragdolled a bunch of times
2 Days Ago
Update: use handles to get pitch and yaw rotations for AttackHelicopterTurret in UseParallelSaves mode - amended test to cover turret saving path Tests: TestMTSave pass
2 Days Ago
Update: move OrientableLight interpolation invoke from Save to rpc + ondeply Scheduling invokes won't be safe in MT Tests: none
2 Days Ago
Update: avoid transform call in TravellingVendor.Save when it's following a spline Tests: none
2 Days Ago
Update: move Privelege caching from DisplayingBoxStorage.Save to OnPrivelegeUpdated We can't have it in Save as it is self-modifying and can call Vis.Entities - lets see how this holds Tests: none
2 Days Ago
Update: force TimeZone caching in WipeTimer.ServerInit While smashing TestMTSave(WipeTimer) I saw it NRE while resolving timezone. Couldn't repro again, but this should make it safe Tests: none, trivial change
2 Days Ago
Update: get rid of WipeTimer rounding to second, just pass DateTime from external source Old version would always have a sporadic fails when serialization test would run across second boundary, this way we can control it in tests Tests: a bunch of times for hacked version of TestMTSave(WipeTimer) that has 1k serialization tasks
2 Days Ago
Update(tests): add various missing TestServer setup and unit testing bypass for entity logic Tests: ran all TestMTSave, no sporadic failures, except WipeTimer - guess my prev fix didn't work >.>
2 Days Ago
Update: made MarketTerminal.Save MT friendly - get rid of TimeUntil aliases from Entity.proto - ran protobuf gen Doesn't need protocol update, as the underlying type/serialization doesn't change Tests: TestMTSave - all 636 tests pass, though gotta work on cleanup consistency
2 Days Ago
Update: WipeTimer now returns time rounded to a second No change for gameplay, but makes serialization tests consistent Tests: 20 runs of TestMTSave(WipeTimer), all pass
2 Days Ago
Update(tests): TestMTSave - move protobuf serialization to worker threads to catch more invalid API usages Revealed 2 failures Tests: ran all TestMTSave - MarketTerminal (TimeUnitl) and WipeTimer(missmatching stream) failed
2 Days Ago
Bugfix: avoid Time call in ChickenCoop.Save to make it MT friendly - added chicken as part of test setup to trigger this path Tests: TestMTSave pass
2 Days Ago
Update: BasePlayer - don't mutate own model state on save We modify the copy as we send it, and the original value should be modified on every tick, so should be safe Tests: none, trivial change
2 Days Ago
Update: make ItemBlueprint.FindBlueprint MT friendly GetComponent is not thread safe, so caching item->blueprint pairs at init time Tests: setup TC, made sure it has right upkeep calculated
3 Days Ago
Bugfix: BuildingPrivlidge(sic) - add GetOrUpdateProtectedMinutes that's MT safe and does a bit of a cheeky If we end up sending this entity to many players in 1 frame, it will update the cached protected minutes only for 1 player, and the rest are in limbo if they got old or new value. This gives us same perf as original lazy-eval solution. Tests: TestMTSave - got 0 fails!
3 Days Ago
Update(tests): patch MapMarkerMissionProvider Tests: TestMTSave passes (down to 2)
3 Days Ago
Update: TestServer now initializes PaintballColorLookup.instance Tests: TestMTSave - 5 -> 3 fails
3 Days Ago
Bugfix: make TravelingVendor.Save MT friendly - get rid of wheel flags processing, they were updated on server and sent over, but not used - cache linear velocity every fixed step (we're missing on interpolation, but we only send the snapshot every fixed update anyway) - use manually interpolated steeringAngle instead of querying the wheel collider Tests: TestMTSave(down to 5 remaining), ran TravelingVendorTest scene and looked at the vendor as it drove around
3 Days Ago
Bugfix: move tin can updates out of Save, as it's not MT safe Tests: TestMTSave and placed it on the ground, walked over and opened inventory
3 Days Ago
Update(tests): TestServer now also initialized TrainWagonLootData and Translate classes Tests: TestMTSave - 13 -> 7 fails
3 Days Ago
Bugfix: batch of bugfixes to Save being not MT safe - removed more Time dependency - replaced Transform access with TransformHandle bypass Tests: TestMTSave - 20 -> 13 fails
3 Days Ago
Bugfix: move Signage.EnsureInitialized from Save to ServerInit, as it's not MT safe Tests: TestMTSave - 25 -> 20 fails
3 Days Ago
Bugfix(tests): batch of patches for unit test spawned entities to avoid NREs in Save Tests: TestMTSave - 28 -> 25 fails
3 Days Ago
Bugfix: move CarvablePumpkin.EnsureInitialized out of Save, since it's not thread safe Tests: TestMTSave(CarvablePumpkin) and spawned and carved one on playground
3 Days Ago
Bugfix: batch of Transform replacements with TransformHandle in save code when using Jobs 4 Tests: TestMTSave - 32 -> 29 fails
3 Days Ago
Bugfix: batch of replacement for Time usage with ThreadSafeTime Only 1 skipped out for now, BuildingPriviledge - that one does lazy-eval, which is problematic Tests: TestMTSave - 42 -> 32