506 Commits over 123 Days - 0.17cph!
Update: ShopFront pools ItemContainer
- also deleted empty virtual override
Test: 2 player session with a shop front, one was interacting with the front while I destroyed it - items dropped, next front clear
Clean: Remove redundant StorageContainer.ResetState
All the cleanup is already done in DoServerDestroy. Might be worth consolidating these callbacks in the future.
Tests: none, trivial change
Update: StorageContainer now pools ItemContainer
- had to introduce a member variable to avoid funky pool API usage because oif property
- also had to relax empty container assert, as there already are cases that don't clean up their inventory on kill
Tests: on Craggy killed all entities via console, saw the pool container stats go down to 0
Update: ItemContainer supports IPooled
- Minor fix - avoid double returning container UID
- replace a static-readonly with a const
Right now there's only 1 place that pools ItemContainers(reclaim manager), so impact of this is low.
Tests: in softcore killed self and looted reclaim backpack.
Tests:
Optim: avoid copying items on container clear
- Also consolidated it into one function.
Did an overview of existing code, didn't spot cases where we can try to modify the container as we're clearing it.
Tests: killed a bunch of animals, loaded up a wooden storage box then killed it
Update: Adding a couple tail recursion perf tests
- also comes with a vailidity unit test
Going to be used to do a write up for the weekly perf blog post
Tests: Ran these perf tests in editor and player, got kind-of expected results.
Merge: from main
Tests: none
Merge: from fix_hitinfo_pooling
Tests: all modes build in editor, got attacked by animals then injure->kill - correct killer (and no leaks of HitInfo in Pool)
Bugfix: show valid last attacker when suiciding
We used to cache pooled HitInfo which could cause it to be randomly modified by other hit processing. Now we cache a value struct isntead of hitinfo.
Tests: solo session, get attacked by an animal, then run injure -> kill. Used to see self as killer, now see animal
Bugfix: Fix server-side leak of HitInfo for player melee attacks
Still on the lookout for the sourc of bad death screen info
Test: local session, couple hits by p2 and the pool numbers for HitInfo stayed stable
Merge: from main
Tests: none, trivial changes
Merge: from geiger_counter_leak
Brings across the pool.print_memory filter feature.
Tests: none, trivial change
Merge: from main
Tests: built all modes in editor
Clean: remove obsolete logs from GeigerCounter
Tests: none, trivial change
Update: Take marker telemetry every measurement
- Also avoid activating markers during Setup/Shutdown
- Also avoid trying to convert Undefined sampler type to "seconds"
Without these changes perf telemetry was too coarse(1 total value for all measurements * iters)
Tests: Ran it for existing bench tests
Bugfix: Don't record GC data from Warmup runs
Fallout from my first patch
Tests: ran new perf tests with warmup, observed no allocations being recorded(as expected)
Clean: fix 2 Pool obsolete warnings
Tests: none, trivial changes
Merge: from main
- There's 2 new warnings for obsolete pool usage, will fix in next
Tests: Built all modes in editor
Merge: from main
Tests: built all modes in editor
Update: Switch to Dynamic measurements for Pool perf tests
- Also moved the tests into it's own nested class since there'll be more in the future.
Surprised this is not default behavior, but ah well. It does run less measurements, so it avoids random spikes.
Tests: ran the pool perf tests, results line up with previous
Bugfix: Avoid perf degradation with many small Perf Test iterations
Was caused by aggressive GC invocation, which I just ripped out - there's a better alternatives if it's truly needed.
Tests: Ran the Pool bench tests, execution time went down from 3m to 3-5s
Bugfix: Avoid tracking allocations from Setup/Cleanup
This fixes the bug that was invalidating measurement tracking. That said, this causes a perf regression because ExecuteSingleIteration is slow - will fix in next CL.
Tests: Observed a reduction in allocation calls for Pool perf test (but not to 0, as it's picking up allocations from GC.Collect)
New: Adding PoolTests allocation perf test
Estimates the allocation speed compared to basic heap alloc when we have full capacity available. This is just for perf testing package validation(revealed 2 bugs).
Tests: ran it and validated results in the Test Report window
Add: Adding Performance Test Framework(v3.0.3) as a source package
While evaluating I discovered a couple issues, so will be patching it over time
Tests: wrote a pool perf test, it ran
Merge: from main
Tests: built all modes in editor
Update: document and test StringView.IndexOf
Tests: ran the new unit tests
Merge: from main
Tests: built all modes in editor
Update: RCon test page now accepts custom endpoint
Think at this point it's good enough, so I'll stop here
Tests: tested with connecting to local host
Update: RCon test page now follows log coloring output
- Parses color tags that we receive in the stream
- Recolored background to follow unity's style (since we sometimes send white colored logs)
Tests: connected to session with high debug level with colored output
Update: RCon page displays Logs
- Logs the pages connection flow, rcon commands and the feed from the server
- Log scroll sticks to bottom
Tests: ran with local session
▅▌▆▆▄▋: ▄▌▋▋ ▆▄▄█ ▉█▋▋ ▇▇▊▆▆▍▌▊▌▊▍▊
- ▍█▌ ▉ ▋▊▍▌▄▅ ▊▊▌▇ ▄▄▊▋▋▍▇ ▋▍▋▍▌ ▊▆█ █▌ ▇▍▄▌▋▆ ▉▋▍▉ ▌▋▆▋▍█▌▌
- ▋▄▅▌▄▆ ▅▉ ▅▊▌▄▊█▊▆▆ ▋▋█▇ ▌▅▍ ▌█▉▄ ▊▍▍ ▇▋█▊▇▋█▌ ▌▊▅▄▌▌▍ ▍█ ▅▄▉▆▋▇▉▍▍▅ ▄▊▋▅
- ▍▆▋▉▆▅▌▆▉ ▌▌▌▋ ▅▋▆ ▅▍█▄▌▊█▊ ▊▇▌ ▊█▉ █▆▅▉▆▋▄▇ ▊▋ ▉ ▌▉▊ ▇▌▊█▅▊▄ (▄▅▊▆▇▌▄▌ ▍ ▆▅▋▍█▊, ▋▆▌▄ ▆▌▇▋█▄▅▋▇ ▋▋ ██▌▉▇)
▄█▇▍▍: ▍█▋██▄▉▆▊ ▅▇ ▉▅█ ▅▋█▅▆ ▆▅▋▅▅▌ ▋▍▇▄▉▇▍, ▍▅▆▇▉▆▄ ▍▉▉▊▊▇▌▇▇▇▉, ▊▆▄▋▌▆▇ ▉▊▊▅ ▅▅▄▌█▊ ▄▅▌ ▅▅▅▆
Update: Add entity counter to Save Viewer
This should be enough for now, I'll add more functionality as needed.
Tests: loaded a large save, validated counter updates with filtering applied
Update: SaveViewer triggers loading on changed event
- Also disabe the export buttons if no valid save is loaded
Allowed to get rid of an extra variable and a first-run exception
Tests: loaded large save game multiple times, all func still worked
Bugfix: avoid parsing invalid paths in SaveViewer
Tests: tried to use a malformed path
Bugfix: Avoid leaking protobuff entities in SaveViewer
Tests: loaded a save multiple times
Clean: remove redundant code in SaveViewer
Tests: filtered a small save
Bugfix:SaveViewer no longer constantly filters
- Also clear up cached results when user clears save file path
Tests: stepped through the code, confirmed execution happened only once
Update: Migrating ConvertSaveFile func to SaveViewer
- Removed old CovertSaveFile
Tests: checked sorting works on a small save, export generates valid csvs
Merge: from main
I also renamed this branch(used to be itemcontainer_ipooled)
Tests: spawned on cragy, killed a scientist, confirmed m92 was marked to not save before kill, then marked for save after kill
Bugfix: Fix equipped item entity leak into savegames
Fixed by tracking whether owher player-entity will be saved or not, and banning equipped items from saving.
Tests:
- Ran in local editor with known offender 'scientistnpc_junkpile_pistol.prefab' - it no longer orphaned it's items on save-load
- Built a local server executable, let it simulate for 30 mins, saved-loaded - there were no instances of 'geiger_counter' and only 1 instance of m92 (out of 253 on the server), it's child `lasersight` also was 0 (out of 143 original)
Buildfix: invalid macro scope
Tests: built in none mode - passed
Update: Allow control for individual log type level
- dev level works as before by overwriting all of log type levels
This will help avoid flooding logs in specific scenarios. To be fully complete it needs to be extracted and consolidated in it's own utility.
Tests: changed logs on a procedural map - saw log types cut off/appear in the stream
Update: Temp logging of geiger counter equip events
Tests: local session
Optim: defer boxing when calling LogEntry
Reduces the overhead of logging a smidge when we won't be emitting any logs due to debug level.
Tests: ran local session, logs still there
Update: Converting custom combat logs to use LogEntry
Tests: local session, checked that collisions generated hurt logs
Update: minor log format changes to help with searchability
By wrapping the category in it'll be easier to do simple searches without hitting popular words (like Spawn)
Tests: ran locally
Clean: Making frequent logs less spammy in normal detail mode
- Bradley velocity report is removed (as it didn't add much value)
- Network update logs moved to highest detail level, 3
Otherwise, it floods the output too much, making it hard to parse logs
Tests: ran locally, checked logs are less spammy
Bugfix: Fix logging name frequently missing a NetID
It was possible to generate and cache name too early
Tests: local editor session, observed that same game object's identity updated properly
Update: When logging emit an instance ID
Without it can be hard to aggregate logs per game object
Tests: ran in editor
Bugfix: RCon test page now works
It's barebones, but enough.
Tests: tested in local session, could see commands take effect
