4,907 Commits over 762 Days - 0.27cph!
Ensure input is displayed brighter than logs
Can now submit console commands to the new console ui
- Split off the display from the processing part of the console UI
Disable per item heights when a grid is used
Fixed visibility calculations still expecting fixed height - use accumuative per row
Use row math for height and spacing
- Fix flex virtual scroll not calculating the correct top and bottom starting rows after dynamic cell changes
- Increased the threshold for console messages to be put into the same bucket (1.5 seconds now)
- Console messages will now group according to log type, multiple warnings will be grouped together etc
- VirtualScroll and FlexVirtualScroll now support elements with different heights - dynamic sizing
- Console is now displaying properly
Fix console not working on first open (eg youd open it to a blank screen)
Group buckets via log type (log types that are different from the 'running log' will create a new bucket
Setup correct warning colour
- Setup virtual flex
- Got something printing, buffer is working
Initial console page backend work:
- Timestamp bucketing system
- Fixed size queue system
- Simple rewrite of console
- Added RPC name under the info section of RPC packets
- Added global filters for server/client realms
- Added packet type breakdown of the top packets
- Clean restyle of the net graph
First draft of a network profiler. Featuring:
- Simple packet stats
- Advanced stats (entity name, packet size, timestamp etc)
- Filtering via timestamp, name, type, direction, realms
- Per entity bandwidth usage, showing how much % of the bandwidth is taken up by what packets
Store:
- RPC inbound/outbound
- SyncVar inbound/outbound
- PackedSyncVar inbound/outbound
- Effect inbound/outbound
Missed some metrics here and there, but critically I've managed to grab the id its going to for each one
Setup data structures to store more detailed packet information, this will be stored and handled locally rather than being sent off for analytics.
Data will be stored in a big ass circular buffer that we can then sort through to do the analytics.
These include:
- Basic data we stored before (byte size, type, etc)
- Associated entity with the packet
- Optional full packet data
- Timestamp
More NRE fixes - lods were swapping before seat ref was established
- Setup system to enable/disable leg animator on LOD
- Setup on PT Boat Rear turrets so that the expensive leg animators will only be enabled when up close
Legs animator:
- Re-eneable leg animator on pt boat walk around turrets
- Allow foot gluing to work lower down under the foot position, this combats the higher waves
Tested the static and regular ballista on a few terrain types. Responded fine to the changes.
turret_head_animations -> naval_update
turret_head_animations -> naval_update
Stop the reload animations of a turret if you hop off
Fix some issues with my one shot cancellation
Stop current player one shot if mounting or dismounting
Made front turret way smoother to look around with
naval_update -> turret_head_animations
multi_drop_nre_fix -> main
- Setup a method to emulate the twitch drop crash from before
- Codegen
Wrap the whole menu init sequence in a try/catch
Stop a null drop NRE bricking the whole Bootstrap sequence
- Apply root motion (not needed but helps rule things out)
- Setup correct reload time on the front turrets
Reorient the front point pivot
Fixed position offset on the front turret
- Fixed scuffed clamping on the front turret seat
- Use tickrate for lerping (cant use deltatime since im predicting)
turret_head_animations -> naval_update
- Fixed muzzle flash playing at the wrong location (still bugged on front turret?) - Add model scripts to the top most parent of each of the 50cal entities
- Remove any clientside angle smoothing - bad idea (causing minor stutters)
- Deleted unused head anims folder
Dont bother with the snapshot update - we dont need it. Handle it in SV_ReloadStart instead
Yaw needs to be set as well due to where our animation starts
Actually set the pitch values to prevent them trying to continue where they left off
- Save client predicted weapon rotation adjustments into the state circular buffer
- Ensures this forced position update is classified as valid on the client, preventing reconcilliation
Without this the turret was reconcilling at the end of a reload as the predicted rotation change wasnt saved as a valid state
Fix some causes of reconcilliation after a reload
- Dont generate predicted states when reloading
- Force update visual weapon rotation when reconcilling, dont just update internal values
- Increase degree offset required for reoncilliation
naval_update -> turret_head_animations
scientist_boats_ai_improvements_3 -> naval_update
Some more tuning:
- Reduce group cohesion a little (they were a bit too stuck into groups)
- Can jump to a new macro target earlier (makes them less stubborn)
- Increase how far out they need to be before they will begin pulling towards the centre
- Greatly increase the applied centre force (they cant get out of moving to the centre as easily)
Reynolds wander isn't working very well for us - boats seem to bunch up on the same map edges
Rewrote all of the boat wander state:
- Now pick macro and micro directions
- Macro is a fixed target point (within deepsea bounds or in random direction)
- Micro is the same Reynolds stuff but done simpler
- Combine them together for a bit of a better approach
Boats now traverse around the whole of the deepsea bounds rather than sticking to an edge. Much cleaner and more reliable
