1,528 Commits over 730 Days - 0.09cph!
clear sim-space frame delta properly when the sim-space changes, was leaking into the origin delta
merge from forcemode_fixes
converting all Acceleration/VelocityChange to use Force/Impulse in prep for engine change
reverted boat unbeaching change
Fixed sculpture save/load to file, added ability to send save to data to all sculptures in radius
using a proto message for serializing sculpt data rather than raw byte array storage
- should be more flexible for altering the format, let's us bake data about the space the array refers to
- still LZ4 compressing the proto bytes because we get 10:1 for data like this
replaced crc RPC-synced value with SyncVar, much cleaner
parallel censor application by-slice using concurrent NativeStream readers
initial improvement to censor chunk generation, NativeStream to parallelize shape creation
fixed sculptures requesting and loading sculpt data twice on load
replaced huge NativeList triangle alloc for a NativeStream on TempJob, 10x less memory usage with no performance loss
let generators resize internal native alloc back to 0 after 10s of not being used + codegen
handling max-triangles length properly
ripping IO out of BaseSculpture, will need to be separated later as a sub-entity for ice so it's cleaner
convert sculpting controller to us InputSystem
merge from christmas2025_DLC/ice_sculptures for code changes
merge from playerboat_beached_flipping_fix
better controlled lifting force with PlayerBoat unbeaching
adjusted torch BC controller position for better rotation adjustments
cherry pick
149555 - BC simulation frame delta fixes
BC pass on the industrial torch VM
merge from dont_bake_on_load2
just tool: `Tools/Lighting/Disable Bake On Scene Load (All Scenes)`
set serialized version as well
tool just injects the property into the .unity file instead of opening and saving it
added /ProfilerCaptures to ignore.conf
reduced physics log level from "verbose" (the default) to "default" (which isn't the default)
- also disable OnTriggerStay event generation because we don't use it and shouldn't
merge from mismatched_serialized_analyzer
analyzer optim, takes approx 40% the time it did before
last CodeAnalyzer version for these changes, catches more edge cases but didn't find any of them in our codebase
last serialized field fix
switch out batched packet send storage for HashSet to avoid multiple sends from an entity within a circuit tick
moved SendBufferedFlagChanges outside of circuit loop
using HasSubscribers method instead of manual check
skip sending batched packets to groups with no subscribers
SimpleLight sends on/off flags through batched system
added BatchedEntityFlags packet type
- allows IOCircuitSystem to buffer entity flag changes from IOEntities and send batched packets instead of per-entity packets
- test setup with 2000 lights goes from 26ms to <2ms on the server
updated CodeAnalyzer - better perf
slapdash serialized property mismatch fixes
updated CodeAnalyzer.dll - better path exclusion
