branchrust_reboot/main/projectile_jobscancel
155 Commits over 30 Days - 0.22cph!
Update: BaseRaidBench - wip weapon aiming
- Weapon debug graphics confirm aiming direction
- Player model doesn't face the right way (need to run client updates, that's next)
- added benchmark-related safety checks in local client logic
This also hijacks main camera, I'll have to fix that later
Tests: ran the scene
Bugfix: BaseRaidBench - make sure weapon flags are synced with held state
Tests: ran the scene
added Projectile.UseBatched ClientVar to switch between serial and batched tick for projectiles
Update: BaseRaidBench - give players projectile weapons on spawn
For what-ever reason, doesn't appear in hands(instead on back), despite being reported as "held". Still investigating
Tests: ran the bench scene
Jobified trace-testing in HitboxSystem, also replacing local NativeArray helper function with existing Expand static extension
corrected InitializeVelocity on projectile to keep Projectile performance test consistent between runs (increased run-count)
Optim: reduced unnecessary Transform reads in batched position/rotation update, also added AddRange(BufferList<>) function to BufferList to change to that for marginally faster loop speed
Test: consistency tests passing with serial using old method still
Update: BaseRaidBenchmark - adjust the camera positioning to get the full circle in view
Tests: ran the scene
Update: BaseRaidBenchmark - players are visible and dressed
Tests: ran the scene
New: BaseRaidBenchmark - stress test client via a raid scenario
- Doesn't do much, just ground framework
Tests: ran the scene - saw players spawn (but not their visuals)
Update: World Setup Benchmark -> BiomeBenchmark World Setup rename
- replaced default params on BiomeBEnchmark with placeholder to indicate they don't do anything
Tests: none, trivial changes
Clean: Refactor free-slot searching logic
Hope is to make it harder to mess up index calculation when writing more code of this style
Tests: GamePhys unit tests + projectile consistency unit test
Added extra sanity checks to projectile consistency testing
Tests: GamePhysics.TraceRrays and TraceSpheres coverage
Wasn't covered previously which led a bug slip in. But has already been fixed.
Tests: ran it with bugfix revwerted - failed(as expected). Reapplied fix - passed
added missing effects update on batched projectiles and retiring into the prefab pool
clear static test field in pet-test teardown
Clean: simplify previous bugfix
Tests: ran the unit test
Update: Projectile.TestConsistency now checks for number of phys commands issued
Tests: unit test passes
Bugfix: Projectile.TestConsistency - when substepping movement, use the right trace result
Tests: consistency unit test now passes
Bugfix: TestProjectileEntity now works in C+S
Tests: ran the previously failing unit test
Bugfix: Projectile.Batch - don't over-simulate prjectile when reacting to a hit/ricochet
Also noticed serial test vs entity is failing - will fix next
Tests: single-projectile-batch test passes, but the whole batch fails
Update: ProjectileTests.TestConsistency - debug code to isolate projectile
Still fails on 520, but now with different results(proj moves, but different height).
Tests: ran unit test - fails (expected)
Update: ProjectileTests.TestConsistency - rewrote logic to run both serial and batched in one test
- Removed old serial validation logic
- Renamed PerfSerial test to Performance (since it also runs batched variation)
Detects an issue currently, investigating.
Tests: ran the new test - fails (kind of expected, but still suspicious)
Update: TestLineOfSightBatched
A lil cheeky 44k raycasts test. Missing the actual batched query impl.
Tests: ran the test - failing as expected
cached native allocations, batched projectiles now save between 30% - 50% of the runtime compared to serialized in performance tests (gains increase with projectile count)
Update: GamePhysics tests - move box grid creation to individual tests
Avoids polluting the world for other tests(problem for LineOfSightBatched)
Tests: ran the updated unit tests
Update: GamePhysics tests - spawn ignore volumes per tests
Tests: ran the 3 affected tests - they pass
multiple significant fixes to GameTrace, new Projectile and GameTrace tests are passing but seeing a slight differences in projectile counts with performance test that's indicating some untested edgecases
Update: TestTraceAll - don't spawn Terrain boxes
We already have terrain at the bottom of the ocean
Tests: ran TestTraceAll
Tests: new GamePhysics tests - TestLineOfSight
- Also making craggy loading in unit test stable by applying same fixes from Test.Projectile
No batched variant as we haven't implemented it yet
Tests: ran the new unit tests
Tests: new GameTrace tests - TestsTraceAll and TestsTraceAllBatched
TestTraceAllBatched is currently failing, as it's wip
Tests: ran new unit tests
initial pass on batched projectiles, functional except for LOS checks but failing consistency tests which needs more diagnosing
- added batch support to multiple tests for comparison
Update: projectile perf test
Tests: ran the perf test
Buildfix: rename UnitTests.WaterLevel to WaterLevelTests
It was colliding with a non-namespace WaterLevel and causing a headache
Tests: editor build in C+S
Update: refactor TestConsistency and make it cover a larger area
- increased proj velocity, as it was possible for proj to start inside the shield, breaking validation
Tests: ran unit test
Update: TestConsistency now validates repl projectiles as well
Test: ran the unit test
Update: TestConsistency now passes
- validates auth proj hit their players/shields
- added utility that helps revalidate projectile hit and get more info (auth only for now)
Need to expand to re-validate non-auth hits
Tests: ran projectile consistency test
Update: wip fixing the consistency test
- fix invalid tracking of on which frame a hit is expected
- extra validation code to confirm hits to expected players - revealed missmatches
Need to redo how I setup the players to avoid random overlapping
Tests: ran the test
Update(wip): new comprehensive projectile test on craggy
- spawns a bunch of players on craggy and drops projectiles on them
Currently fails as majority of hits miss (but expect 100% hit rate) - likely bad test setup. Will investigate tomorrow
Tests: ran unit tests
Update: projectile tests against players with shields
Test: ran new unit tests
Update: rewrite TestProjectileWater to attempt to hit a player in water
Tests: ran new TestProjectileWater - pass all
Update: extract all miss cases from projectile tests into separate case
It takes the same path internally, so no point in rerunning the same test case
Tests: ran all proj tests - pass
Update: projectile-water test via Craggy
Pretty close to covering all cases
Tests: ran new tests
Update: Projectiles that are managed by unit tests only run 1 substep
Tests: projectile unit tests now pass
Update: Projectile Tests now run inside Craggy
Most tests are failing because now their substepping eventually hits water/terrain - will fix next.
Tests: ran unit tests
Update: TerrainTexturing - defer init in editor, relly on manual init by user
This prevents it from screwing unit tests when loading scenes
Tests: ran part of updates projectile unit tests (to be committed)
Update: adding mounted-player related tests
- Fixed names of tests as turns out if they share a name, then all of them get executed when trying to run only one
- Fixed not cleaning up palyer's physics objects at the end of tests (caused tests to fail unexpectedly when ran all together)
- test projectile now has thickness (relevant to test with players in same vehicle)
Tests: ran all unit tests
Clean: minor comment fix
Tests: none, trivial change
Update: projectile-entity test now covers more internal code
- cleaned dead code - not sure I'll need it in the future, so to the grave it goes
Tests: ran the unit tests