6.1 (Soul finding 6): Commit builds a local CultMath.Random from (Zone.CombatSeed,
shot.ShotId) instead of reading and writing the shared ItemManager.Random stream.
Zone exposes CombatSeed (the same expression that already seeds its own _random),
so a shot's outcome is a pure function of zone identity and shot id -- immune to
whatever else drew from the shared stream in between.
6.2 (Soul finding 5): airburst is wired into the simulation instead of back into
Projectile.cs. PendingShot gains BurstPosition/BurstRadius, frozen at Fire time
from PredictedIntercept and the weapon's authored WeaponItemData.AirburstRange
when WeaponModifiers carries Airburst. FireControl.Step resolves such a shot via
Splash instead of Apply, never both. Projectile.cs/ProjectileManager.cs drop the
dead AirburstDistance/AirburstRange fields and write-back.
Tests: FireControlCut6Tests.cs pins both rules (208 baseline + 5 new, all green).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Cut 2 landed LoadoutGenerator's required-targeting-system rule and no targeting
system to satisfy it, so AetherDb loadout threw InvalidLoadoutException and the
player's own hull could not be generated. The turret firing-arc migration was
likewise written and never run, leaving a turret engaging only within 60 degrees
of its spawn facing.
Two targeting designs with products, four products for previously unsold designs,
and the arc migration applied: 53 designs, 43 products.
Cut 6c amends two defects in this authoring -- Resolution reads as a cost so the
premium system is worse than the cheap one, and a single seller makes ship
generation depend on whether a galaxy happens to draw that faction.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Seven code defects Soul confirmed across the campaign, every one the same shape:
a rule implemented on one path and left off the path beside it. Tracking gets a
floor instead of a cliff; a shot the shield cannot absorb now breaks it on the
discrete path as well as in Splash; a beam obeys its firing arc; one bearing
test owns the point-blank case; a destroyed subsystem stops being the aim point;
the simulation removes its own corpses instead of relying on Unity's loot-drop
subscription; and PendingShot drops two fields nothing read.
Cut 5b then repaired the mutation harness, which had been leaking un-restored
mutations into the working tree -- that leak, not a matcher bug, was also why two
anchors appeared unmatchable. Restore is now hash-verified with an explicit
tree-clean verdict, and ConstantWeapon's PowerRequest gate is pinned separately
from Execute's, which is where the one surviving mutation had been hiding.
Verified independently of the agent reports: 208 tests pass, all eleven
mutations die, the no-op control stays green, and every target file is
byte-identical to sha256 hashes taken before the run.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Precision is currently a coin flip between landing exactly on the aimed item and
landing uniformly anywhere on the hull, so aiming at a stern thruster scatters
its misses across the bow. Operator direction: misses should cluster near what
you aimed at, like throwing a dart at the grid -- and aiming should therefore be
a real tradeoff, where a thin limb risks missing the ship entirely and centre
mass is the safe play.
One Gaussian kernel over the hull schematic answers both halves: its on-hull
mass prices the stray into HitProbability, so the risk is legible on the HUD
before the trigger, and its per-cell weights place the hit. One function, so the
probability and the placement cannot drift apart -- Cut 3's named risk was
exactly two functions answering where a shot goes.
Records the rejected draft that clamped a stray dart to the nearest hull cell,
which would have quietly removed the tradeoff, and the two consequences worth
catching early: pOnHull may double-charge with the existing pSpread term, and
every AI gates on this number so they will hold fire more once it lands.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three runs of tests/mutation_tests_fire_control_cut5.sh left mutated files on
disk (ArcAllowsFire hardcoded true in Weapon.cs, InstantWeapon.cs's arc gate
deleted, FireControl.cs still carrying 5.1's own mutation) despite its
`trap restore_all EXIT INT TERM`: a trap cannot survive a SIGKILL of the
process group, and nothing before this cut ever checked that a `cp -p`
restore actually reproduced the original bytes.
Records each target file's sha256 before any mutation runs. Every restore --
per-mutation, immediately after each check_mutation, and at exit -- re-hashes
and aborts loudly on a mismatch instead of trusting `cp -p` silently. The
final summary and the exit-time trap both print an explicit tree-clean
PASS/FAIL verdict so a reader never has to infer it. The no-op control stays
first, unchanged.
Investigated the two anchors reported as "found 0" (Zone.cs's Death
subscription, Weapon.cs's InArc line) before touching anything, per the
mandate not to hand-edit them into a shape that happens to match: on this
worktree at HEAD, both anchors are single-line literals with no embedded
newline, so CRLF-vs-LF in the surrounding file cannot affect whether the
substring is found -- the CRLF theory does not hold up. Both anchors match
exactly once against the current, clean tree (verified directly with the
same perl matcher this script uses, and confirmed by a full harness run).
The mutated replacement text for each of those two anchors is byte-for-byte
identical to the exact leaked mutation the task description names as having
been left on disk by an earlier run (Weapon.cs's `return true;`, Zone.cs's
deleted Death subscription) -- i.e. the "anchor never matches" failures were
not an independent matching-mechanism bug, but a direct symptom of the same
leaked-mutation-state problem this cut's hash verification now closes. No
anchor literal was changed.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
ConstantWeaponObeysArc only asserted PowerRequest()==0 after Zone.Update had
already run Execute, which safes _firing off for its own (also-correct)
reason first. PowerRequest short-circuits on _firing regardless of what
ArcAllowsFire mutates to, so a mutation dropping ArcAllowsFire from
PowerRequest alone was an unpinned survivor even though the harness's own
Execute-line mutation was already caught. Added the same assertion before
Zone.Update, while _firing is still true and StanceAllowsFire is true, so
ArcAllowsFire is the only thing it can be testing.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The multi-tick, many-Update-call version of this test (5.1) was flaky under
repeated runs: stepping a 5s flight in 50 separate .1s ticks left the exact
commit-time reading of `now` sensitive to float-accumulation/JIT-tiering
drift, occasionally producing zero hits across all 30 shots even on
unmutated code. Rewritten with velocity 0 so each shot's flight time (and so
its commit horizon) is zero -- one Fire call, one Update call, no
accumulated ticks to drift. Confirmed clean across repeated standalone runs
and still kills the 5.1 fallback-to-0f mutation.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Cut 6b: the combat roll stops drawing from the shared ItemManager.Random stream
and becomes a pure function of (zone identity, shot id), so adding an NPC or
opening a shop can no longer reorder an existing fight's dice. Airburst resolves
in the simulation rather than going back into Projectile, which would be exactly
the presentation authority Cut 3 spent itself deleting.
Cut 6c: reviewing Cut 6a's landed catalog found the premium targeting system
strictly worse than the cheap one, because Resolution reads as a cost while its
name promises a benefit. Operator ruling: fix the formula, not the handle -- it
becomes reciprocal, higher stays better, no rename. Also: both designs ship from
one seller, and galaxy generation draws a random subset of factions, so ship
generation still throws in any galaxy that misses that draw.
Cut 7: the mutation harnesses for Cuts 1-3 that the map declared and nobody ever
wrote, plus the fifth mutation Soul left standing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Soul's pass over the whole campaign confirmed eleven findings. Seven were code
defects of one shape -- a rule implemented on one path and left off the path
beside it -- and this is all seven.
5.1 Tracking falls back to GameplaySettings.UnaidedTracking (new, 10) instead of
zero, and Commit drops the zero branch that turned a Tracking-less shooter
into a hard wall. Unaided fire was meant to be really, really bad; it was
334 shots and no hits against a target drifting at 5 u/s.
5.2 A shot the shield cannot absorb now breaks it. The decision is frozen into
ShotOutcome.ShieldBroken at commit and performed in Apply, so a shield that
recharges mid-flight cannot retroactively survive. Splash already did this;
the discrete roll never did, which made the whole shield-break mechanic
unreachable for every weapon in the catalog.
5.3 ConstantWeapon gates on ArcAllowsFire alongside StanceAllowsFire. A
side-mounted beam was firing forward.
5.4 The point-blank bearing case moves out of Weapon.ArcAllowsFire and into
InArc, which used to normalize the zero vector into NaN. The trigger said
fire and the probability said impossible.
5.5 A destroyed item drops out of IsRevealed's ranking, so Precision stops
concentrating hits onto a dead subsystem's footprint.
5.6 Zone subscribes Entity.Death and removes the corpse itself. That removal
lived only in Unity's loot-drop subscription, so headless nothing ever left
the zone, shots landed on corpses and no fight could end.
5.7 PendingShot drops PredictedIntercept and FlightTime, written and read
nowhere.
208 tests pass (197 plus eleven new).
Known defect, recorded rather than hidden: the committed mutation harness is
unreliable. Across two runs it failed to restore FireControl.cs and Weapon.cs,
leaving live mutations in the working tree -- including its own 5.1 mutation,
which is how a green suite briefly reported two failures. Its verdicts are
therefore evidence only where re-run from a verified-clean tree. Two anchors
(Zone.cs's ObserveAdd line, Weapon.cs's InArc return) fail to match at all, and
5.3's ConstantWeaponObeysArc survives its own mutation, so that rule is written
but not pinned. Cut 5b owns the repair; nothing here is trusted on the harness's
word, and the suite result above was taken by hand from a hash-verified tree.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Runs targeting-catalog apply: two new designs (Targeting Computer, Fire
Control Array) and six new products (LockOn / LockOn Pro by NiteLife Energy;
DragOnBreath by Lucent Media for plight; Leonid by Aeronautics Unlimited for
pswarm; Steadfast by Rossum & Douglas for Core Power; True North by Lightsail
Express for Large Drive).
Verified: census 53 designs/43 products (was 51/37); loadout over ten seeds
no longer throws for LonginusX and each gains exactly one interior targeting
system; dangling 0; a fresh census reopen carries 0 schema issues;
firing-arc-migrate dry run reports nothing left to migrate; dotnet test
tests/Aetheria.Shared.Tests is 197/197.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Cut 2 shipped TargetingSystemData and its FireControl/LoadoutGenerator
plumbing but never authored the catalog data it needs: no targeting-system
design exists, so tools/AetherDb loadout throws InvalidLoadoutException for
the player's own hull. This adds targeting-catalog, in the shape of the
existing *-migrate commands (dry run unless "apply"), which authors:
- Two GearData designs, "Targeting Computer" (1-cell) and "Fire Control
Array" (2-cell), each with EnergyDrawData/HeatData behaviours alongside
TargetingSystemData (the behaviour supplies no power machinery of its own),
argued from the capacitor (the catalog's only other bare Tool gadget).
- One product per design from NiteLife Energy, the one manufacturer that
already sells the capacitor, so IsAvailable finds one without falling back
to "any manufacturer".
- One product each for the four designs nothing currently sells (plight,
pswarm, Core Power, Large Drive), matched to the manufacturer the brief
and the corporate roster already imply.
Not yet run against the catalog; the next commit lands its output.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
tools/AetherDb firing-arc-migrate was written in Cut 2 and never run: the
Turret hull's two Ballistic hardpoints still carried a 0-degree arc, so a
turret only ever engaged within 60 degrees of its spawn facing. Applying it
sets both to 360, matching the operator ruling (fire-control-cut.md Q1) that a
turret's arc is authored data, not a new hardpoint type. A fresh dry run now
reports both hardpoints already authored.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Soul's campaign pass confirmed eleven findings. Seven of them are code defects
of one shape -- a rule implemented on one path and left off the path beside it
-- and belong to this cut. Catalog authoring is Cut 6, alongside airburst and
the shared combat RNG; the missing mutation harnesses are Cut 7.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The last five weapon kinds route through FireControl; HullCollider's hit
and splash reporting is deleted. A beam rolls once per resolve interval.
Mines keep their proximity trigger (Q7) and resolve damage through the
same path. Also fixes a real defect: OnFire?.Invoke(FireControl.Fire(...))
never fired at all when nothing subscribed, which is the headless case.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
tests/Aetheria.Shared.Tests/FireControlCut4Tests.cs, builds its own minimal
fixtures rather than reusing FireAuthorityTests.cs's InstantWeapon-shaped
Engagement:
- BeamRollsPerInterval: over one second a beam produces exactly
1/BeamResolveInterval outcomes totalling Damage. Interval (.25s) and tick
(.125s) are both exact dyadic fractions so float accumulation cannot drift
the count, and the tick is strictly smaller than the interval so a "roll
every tick" mutant and a "roll once for the burst" mutant both fail.
- InstantWeaponFiresWithoutAnyOnFireSubscriber: pins the InstantWeapon half
of the null-conditional short-circuit fix, which no existing test could
catch (FireAuthorityTests.cs always calls FireControl.Fire directly, never
through Trigger()/Execute()).
- SplashHitsEveryEntityInRadius / SplashIsDirectional: FireControl.Splash
called directly against a zone of bare Ship entities (Entity.MapEntity,
run from the constructor, is enough to give a Ship Armor/GearOccupancy/
Hull without Activate() or a weapon catalog).
tests/mutation_tests_fire_control_cut4.py is the committed harness: a no-op
control plus one mutation per rule above, run against
tests/Aetheria.Shared.Tests via `dotnet test -p:CultLibRoot=...`. All seven
mutations behaved as declared in this pass.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Deletes the last live Physics.* hit queries and every remaining SendHit/
SendSplash call in Assets/Scripts/Gameplay/Weapons: ConstantLaser's
RaycastAll, ConstantLightning's SphereCastAll, ConstantParticleWeapon's
trigger-collider plumbing, HitscanEffect's RaycastAll, and Mine's blast
OverlapSphere. Each becomes presentation only, the same convention Cut 3
gave Projectile/GuidedProjectile/Laser/Lightning: ConstantLaser/
ConstantLightning/HitscanEffect draw to TargetTransform (or out to Range with
no target); ConstantParticleWeapon keeps only its visual emission.
ConstantWeapon.Execute (previous commit) already rolled the damage before any
of these ever draw.
Mine keeps its proximity-arming OverlapSphere (Q7: contact detection, not hit
detection) but Explode() now calls FireControl.Splash instead of walking
colliders and calling SendSplash itself.
HullCollider.Hit/Splash/SendHit/SendSplash and their event-arg classes are
deleted outright -- their last callers are gone. HullCollider is now purely a
physics surface for ship-on-ship collision impulse; no Physics.* query
decides damage anywhere in Assets/Scripts/Gameplay/Weapons except Mine's own
named exception.
Every manager (ConstantLaserManager, ConstantLightningManager,
ConstantParticleWeaponManager, HitscanManager) is updated to match: no more
Damage/Penetration/Spread/DamageType plumbing into a presentation component
that no longer decides anything with it, and HitscanManager now passes
ShotId and TargetTransform the way LaserManager already does.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
FireControl.Fire takes an optional damageOverride so a continuous weapon can
queue a shot for less than its full Damage. FireControl.Splash is the one new
rule this cut adds: an unconditional area effect (mines, airburst) applying
the shield-absorb-or-hull-takes-it branch per target, then DamageSchematic
over the directional half of each target's own hull that faces the blast --
the rule moved verbatim from the Splash subscription EntityInstance.cs
carried before Cut 3 deleted it (git show b7743789^).
ConstantWeapon.Execute rolls once per GameplaySettings.BeamResolveInterval
(new setting) for Damage*interval, through the same FireControl.Fire/Step
pair a discrete shot uses -- a beam's authored Velocity is 0, so the flight
commits and resolves in the same tick it fires (R4's short-flight
degradation). A beam is a sequence of rolls, not a continuous truth.
Fixes a real defect found while testing the above: both InstantWeapon.Execute
(Cut 3) and the new ConstantWeapon.Execute wrote `OnFire?.Invoke(FireControl.
Fire(...))` inline. C#'s null-conditional short-circuits the whole
expression, argument included, whenever the event has no subscriber -- true
of any headless run with no Unity EntityInstance wired up
(docs/headless-playground-cut.md) -- so a shot silently never fired. Both
now evaluate FireControl.Fire into a local before the null-conditional
invoke.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Cut 3 landed 39 compile errors in GuidedProjectile: it uses ToCultMath
throughout and lost the using directive. Caught by batchmode on merge, not
by the pass that wrote it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The raycast hit path is gone from Projectile, GuidedProjectile, Laser and
Lightning; FireControl owns whether a shot lands. A shot freezes its gun's
stats at the trigger, commits its outcome a fixed time before impact, and
resolves on arrival against how far the target deviated. Rolls are seeded,
so a seeded engagement reproduces. Five weapon kinds still on the old path
belong to Cut 4.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Deletes the Unity physics hit path (HullCollider.Hit/Splash, and the
raycast/shield/SendHit blocks in Projectile, GuidedProjectile, Laser and
Lightning) and replaces it with a single owner in the same cut: FireControl
now decides whether a shot hits, Zone owns the pending-shot lifecycle, and
Entity.DamageSchematic/ApplyHit (moved verbatim from EntityInstance) apply the
result. Presentation subscribes to Zone.ShotCommitted/ShotResolved and applies
nothing.
FireControl.HitProbability is the pure targeting solution (Accuracy * pSensor
* pSpread, gated by detection/range/lock/arc). FireControl.Fire freezes the
weapon's payload and the shooter's resolved targeting stats at the instant the
trigger is pulled (Q6), computes the predicted intercept through the one
shared PredictedIntercept function (CombatState now calls the same function
instead of duplicating first_order_intercept), and queues a PendingShot.
FireControl.Step commits each shot's outcome at ArrivalTime - CommitHorizon --
deviation from the predicted intercept, measured once, live, at that instant
-- and applies it at ArrivalTime. A shot whose flight is shorter than the
horizon commits and resolves in the same tick it fires. Rolls draw from the
shooter's own ItemManager.Random (Q5, seeded and reproducible); an out-of-arc
or otherwise gated shot draws nothing.
Player fire is arc-gated through the same predicate AI and turret fire use
(Weapon.ArcAllowsFire, checked in InstantWeapon.Trigger -- the one point every
shooter's trigger passes through), closing Q2.
Adds GameplaySettings.CommitHorizon, SchematicCellSize and
AgentMinHitProbability (first guesses; the headless suite is the tuning
harness, operator play smoke is the arbiter).
FORK reported, not worked around silently: HullCollider.SendHit/SendSplash
cannot be deleted yet without breaking ConstantLaser.cs, ConstantLightning.cs,
ConstantParticleWeapon.cs, HitscanEffect.cs and Mine.cs, which still call them
from their own live Physics queries -- that's Cut 4's migration. They fire
into the void today: nothing subscribes to Hit/Splash any more for the four
weapon kinds this cut owns.
tests/Aetheria.Shared.Tests/FireAuthorityTests.cs pins the roll's seeding, the
gate-consumes-no-draw rule, resolve-on-arrival, the commit horizon and its
evasion window, short-flight degradation, the frozen-payload snapshot, aimed
hits, the item-then-hull damage order, the planar penetration march and the
shield-absorbs branch. FireControlTests.cs and InputCapacitorTests.cs are
updated for the new AI gate and OnFire(int) signature.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A new interior Tool behaviour carrying accuracy, resolution, precision and
tracking; unaided fire falls back to one authored setting. The two info
thresholds are finally read: a target's equipment reveals as target data
accumulates, and one predicate and one writer serve both the player and
the AI when aiming at a revealed subsystem.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
TargetingSystemTests: selection needs a real reveal (and drops it again
on decay, with no clearing loop), the reveal ranking (hardpoint before
interior, larger before smaller), the two thresholds are the ones
actually read, unaided fire falls back correctly for no/destroyed/
disabled systems, a starved system's Accuracy halves with the grant
(a real Reactor + PowerBus tick, since PowerSupply's setter is internal
to a different assembly here), and every armed generated loadout gets a
targeting system across ten seeds.
LoadoutTests' shared fixture now authors a targeting-system product too:
its "Twin" hull can randomly draw the weapon-carrying "Gun" onto its
Sensors hardpoint, and FillInterior's new required item would otherwise
throw for a fixture that predates this cut.
174 tests before this cut; 180 after (174 unchanged + 6 new), all green.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A new "Cycle Target Item" action (hand-authored into Aetheria.inputactions;
no Unity editor here to regenerate AetheriaInput.cs, so ActionGameManager
looks it up by name through Input.asset.FindAction instead of a typed
accessor) cycles the local player's aim point among the current target's
revealed items via TrySelectTargetItem. SchematicDisplay's enemy view now
filters to what FireControl.IsRevealed actually resolves for the viewer,
and marks whichever row is the current ResolvedTargetItem through a new
SchematicListElement.SelectedIndicator (unwired in any prefab yet -- a
no-op until an editor pass assigns one).
Presentation only; decides nothing. Unverified here -- Unity batchmode
is unavailable in this environment.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
LoadoutGenerator.FillInterior equips a targeting-system product,
required: true, on any entity EquipHardpoints already gave a weapon --
Q4's "not an upgrade" ruling made structural. CombatState now selects
the target's most dangerous revealed weapon item each tick through the
same IsRevealed predicate and TrySelectTargetItem writer the player path
uses, instead of never touching TargetItem at all.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
TargetingSystem is required interior Tool gear: Accuracy/Resolution/
Precision/Tracking are PerformanceStats reached through the usual
Evaluate path, so lot quality, wear, heat and PowerSupply all apply
(a starved system rolls worse, never switches off). FireControl.Accuracy/
Resolution/Precision fall back to GameplaySettings.UnaidedAccuracy (new,
authored low per Q4) whenever an entity has no targeting system or its
own is destroyed/disabled.
FireControl.IsRevealed derives R5's reveal rule from EntityInfoGathered:
non-hull equipment ranks hardpoint-mounted first, then by size, and
reveals between TargetArmorInfoThreshold and TargetGearInfoThreshold --
the first code to read either setting. Entity.TargetItem/
TrySelectTargetItem/ResolvedTargetItem give both the player and AI one
shared, revoke-on-decay aim-point slot with a single writer.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
HardpointTransforms and its renderer readback are gone; mount direction
comes from the item's rotation and arcs are simulation data, default 120
with a per-hardpoint override. Combat no longer throws headless: a real
engagement steps two ticks and decides to fire through the same arc gate.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A Hands scratch console upsert, same shape as ShieldMigrate/BrownoutMigrate
(dry run unless "apply" is passed) -- follows the Cut C precedent named in
docs/fire-control-cut.md's Cut 1. Scoped to the Turret hull's own Ballistic
hardpoints, the only ones the cut names (Q1: turrets author FiringArc: 360).
Not applied here: GameData/Aetheria.cc is left untouched. Dry run against
the live catalog reports two hardpoints to migrate:
hull hardpoint position was now
Turret Ballistic (1,2) 0 360
Turret Ballistic (5,2) 0 360
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
One check per rule Cut 1 lands, each with a stated mutation that would
defeat it: mount direction follows the item's authored rotation and not
the hull's facing (ArcFollowsMountRotation), FiringArc is a full width so
the boundary sits at half of it (ArcBoundaryIsHalfWidth), the arc test is
planar per R7 (ArcIsPlanar), and a per-hardpoint override wins over the
default while an authored 0 still means "use the default" rather than a
zero-width arc (HardpointOverrideBeatsDefault).
CombatStateStepsHeadless is the cut's real deliverable: a Minion with a
target in range steps through two Zone.Update ticks without Combat.cs
throwing (the KeyNotFoundException this cut exists to unbreak), and the
fight actually decides to fire, not just avoids a crash.
Builds its own hull/catalog rather than reusing IffAndCombatTests'
private BuildWorld/NewShip (fixed to one hardpoint) -- and along the way,
surfaces three catalog-authoring gotchas any headless-combat fixture
needs: Mass defaults to 0 (NaNs UpdateTemperature via ThermalMass),
default temperature bounds [0,100] don't cover wherever the heat model
settles, and Entity._orderedEquipment is only ever populated by TryEquip
so an entity that never equips anything throws on its first Update.
docs/fire-control-cut.md, Cut 1.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Entity.HardpointTransforms was written only by Unity (EntityInstance.cs),
so the simulation's notion of where a gun points was a readback from a
renderer that never runs headless -- Combat.cs threw KeyNotFoundException
on the first combat tick outside Unity. Deletes the field, its writer, and
the readback branch in Behavior.Direction, leaving the item-rotation path
(the same convention reaction thrusters use, R6) as the only source of
mount direction.
Adds FireControl (ServerShared), a static class owning whether a weapon
bears on a point: MountDirection (planar, from ItemRotation), ArcFor (the
per-hardpoint HardpointData.FiringArc override, falling back to the new
GameplaySettings.FiringArc default of 120), and InArc (the planar dot-cone
test, R7 -- target height never enters it, and an arc of 360+ passes
unconditionally so a turret hardpoint needs no new type, just a number).
Combat.cs and TurretController.cs now gate AI fire through InArc instead
of the old hardcoded 0.99 dot-product test against a live Unity transform.
Settings.cs drops AgentFiringMinDot, which was declared and read by
nothing.
docs/fire-control-cut.md, Cut 1.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A power request is what the item wants at full supply, not what it is
currently managing, so the circularity dies at the root and radiators and
drives can carry their brownout curve again. The static refusal goes; the
modifier-chain check stays. Six skipped tests pass on their merits and the
shipped catalog is legal again.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A power request may not depend on power supply (Cut 6/F6), and continuous
consumers brown out through a power term on their own performance stats
(Cut 7). For Radiator (PumpedHeat) and AetherDrive (Torque) those were the
same stat, making six shipped catalog records illegal and leaving six
BrownoutTests cases skipped.
Resolved per the operator's ruling: a power request is evaluated
nominally -- what the item wants at full supply, not what it is currently
managing. EquippedItem.EvaluateNominalPower (mirroring ConditionRatio's
own NominalContext) pins only PowerSupplyFactor to 1; Heat/Durability
stay live, since they are not the term a request would be circular
through. Every PowerRequest/RefreshReserve/RefreshInputCapacitor
implementation now reads a StatValidation.PowerRequestFields stat through
it, so the same stat can carry a brownout curve and be a request field
without the request depending on its own answer.
StatValidation.ValidateNoPowerSupplyOnRequest (F6's static half of the
rule, forbidding the term outright) is deleted -- nominal evaluation
removes the circularity at the root instead. Only the dynamic
modifier-chain half (StatModifier.ValidateNoPowerSupplyChain) still
refuses anything, since EvaluateNominalPower forwards ScaleModifier/
ConstantModifier unchanged (real, not nominal).
Un-skips BrownoutTests' six Radiator/AetherDrive cases and adds two new
regression tests (a direct EvaluateNominalPower probe, and a two-tick
Radiator demand check) backed by a new mutation harness. Updates the four
PowerCurveTests/AetheriaStoresTests cases that pinned the now-superseded
static refusal to instead assert acceptance. GameData/Aetheria.cc is
unchanged and still validates.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
F1 power supply multiplies the resolved value (operator ruling), so zero
supply means zero and a flat stat still responds. F2 the bus agrees with
Active in both directions. F3 an empty tier reports no demand instead of
full supply, and sub-epsilon demand is rationed and subtracted. F4/F5 the
shield restores while inactive and CanTakeHit no longer mutates. F6/F7 the
validator names every stat a request reads and the tools validate before
writing. F8 re-anchors two dead mutation harnesses.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Both scripts aborted with "anchor must occur exactly once, found 0" -- later
cuts moved the code their anchors pinned:
- cut3's fourth mutation anchored PowerBus.Step's old flat "write GrantRatio
to every consumer" loop, which Cut 5 replaced with AllocateTiers' own
tiered closing loop (draw.Item.PowerSupply = ratios[draw.Tier]). Re-anchored
to that loop; the rule (every draw gets written, not just the first) and
its test (PowerBusTests.EveryConsumingItemReadsTheSameGrantRatio, a
single-tier fixture) are unchanged.
- cut4's shield mutation anchored CanTakeHit's old single-expression body
(`return Item == null || _reserve.CanSpend(...)`), which the 2026-09-19
shield reserve ruling split into an Item-null check, a Broken check and
this CanSpend call as its own line, and F5 (this same pass) then made the
whole method a pure query. Re-anchored to the CanSpend line; the rule (a
partial draw is allowed, not gated on covering the whole reserve) and its
test are unchanged.
Both scripts now run clean end to end against the pinned CultLib worktree:
every mutation kills its named test, the no-op control does not.
Scope note: mutation_tests_stats_power_cut0/1/2 also abort on stale anchors
(spot-checked while re-anchoring cut3/cut4, then the check was cancelled
partway through cut1 once the pattern was clear) -- not touched here, as F8
named only cut3 and cut4.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
CultRecordRefs.Upsert is the one function every catalog write is supposed to
go through, but tools/AetherDb/Program.cs's Dangling, ShieldMigrate and
BrownoutMigrate commands each repair an existing record in place and must
preserve its key -- Upsert's own cache.UpsertAsync(document) cannot take a
caller-supplied key, so all three wrote raw and unvalidated through
`db.Cache.Commit(batch => batch.Upsert(document.GetType(), document, key))`.
`brownout-migrate apply` is how the shipped catalog's PowerSupply terms were
actually authored, unvalidated, the whole time.
Extracted Upsert's validation body into CultRecordRefs.Validate(object), and
call it directly from all three tool commands immediately before their own
identity-preserving commit -- the same rules, invoked where Upsert itself
cannot reach. The comment above StatValidation naming the write-bypass holes
(landed with F6) now names all four instead of two: CultCache Studio's
generic editor (still open -- not this repository's to patch) plus the three
tool commands above (closed).
Pinned in tests/Aetheria.Shared.Tests/AetheriaStoresTests.cs:
ValidateRefusesAnEquippableItemWithAZeroSpanHeatRangeTheSameWayUpsertWould and
ValidateRefusesAPowerRequestCarryingAPowerSupplyTermTheSameWayUpsertWould pin
that the extracted method actually refuses what Upsert would (fail under a
reverted Validate that is a no-op); ValidateIsANoOpForADocumentTypeItDoesNotGovern
pins that a document type it has no opinion about doesn't crash it.
Verified tools/AetherDb/AetherDb.csproj still builds clean against the pinned
CultLib worktree.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
StatValidation.PowerRequestFields registered only each consumer's own
top-level request field (Thruster.EnergyUsage, Shield.Capacity, and so on),
on the premise that PowerRequest reads exactly one PerformanceStat. That was
false for five of the eight consumers: Radiator.PowerRequest also reads
PumpedHeat and WasteHeat (its own early-out gate), AetherDrive.PowerRequest
also reads Torque, LambdaMultiplier, MaximumRpm and PassiveCoupling (its
whole spin-up arithmetic), Shield.PowerRequest (via RefreshReserve) also
reads RefillDuration/RestoreDuration, InstantWeapon.PowerRequest (via
RefreshInputCapacitor) also reads Cooldown/Count, and Sensor.PowerRequest
also reads PingCooldown. Any of those could carry a PowerSupply term and sail
straight past ValidateNoPowerSupplyOnRequest, corrupting the very request it
feeds -- exactly Soul's reproduced "OK Disperser" oscillation
(SOUL_RadiatorRequestDependsOnItsOwnGrantAndOscillates).
Audited directly against each PowerRequest method's own body (and the
RefreshReserve/RefreshInputCapacitor helpers it calls into), not guessed from
the behaviour's public surface.
This correctly refuses the real catalog now: `GameData/Aetheria.cc` carries
exactly six violations --
cold like my heart RadiatorData.PumpedHeat
Arctica RadiatorData.PumpedHeat
Skiron RadiatorData.PumpedHeat
OK Disperser RadiatorData.PumpedHeat
Iapyx RadiatorData.PumpedHeat
Traction AetherDriveData.Torque
-- re-authoring them (moving the PowerSupply term off the now-covered field,
or redesigning Radiator/AetherDrive's own PowerRequest to read a
nominal/undegraded value instead) is catalog and design work for Self/roles
authoring, not done here.
BrownoutTests.cs's own Radiator/AetherDrive brownout fixtures hit the same
wall -- they authored a PowerSupply term on PumpedHeat/Torque to test Cut 7's
"curve on the performance stat," which is now correctly refused since both
behaviours' PowerRequest reads that same stat. Neither behaviour has an
untouched field left to curve, so there is currently no valid catalog shape
that gives them a working brownout curve without also feeding it back into
their own request -- a real, scoped conflict between Cut 6 and Cut 7 for
these two consumers. The six affected tests are marked Skip with a header
comment explaining the conflict rather than silently reworked or deleted;
Thruster and ConstantWeapon are unaffected (their own PowerRequest never
reads the separate field their tests curve).
Pinned in tests/Aetheria.Shared.Tests/PowerCurveTests.cs:
UpsertRefusesARadiatorWhosePumpedHeatCarriesAPowerSupplyTerm and
UpsertRefusesAShieldWhoseRefillDurationCarriesAPowerSupplyTerm -- both fail
under the reverted (top-level-field-only) registry.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
F4: a hit that breaks the shield usually also cooks it (TakeHit's own
AddHeat), which can knock the item thermally offline in the same tick. The
reserve's recharge and its Broken-clearing check lived inside Execute, which
Entity.Update only runs while Item.Active is true -- so a shield broken by a
hit that also shuts it down could never come back, no matter how long
RestoreDuration allows for. That logic now lives in Shield.Update
(IAlwaysUpdatedBehavior), which Entity.Update calls every tick unconditionally
(Cooldown.cs already relies on the same interface for the same reason).
F5: CanTakeHit broke the shield and drained the reserve as a side effect of
the query itself, on the premise that every caller queries it exactly once
per hit. That premise was false: RaycastAll/SphereCastAll/OverlapSphere hand
every caller under Assets/Scripts/Gameplay/Weapons both the shield collider
and the hull collider for one shot, so the same shield's CanTakeHit was
queried twice per hit. CanTakeHit is now a pure read (Broken + the reserve's
own CanSpend); the mutation moved to a new Shield.Break(), idempotent, called
by each weapon file exactly where it already decided to route a hit past the
shield instead of absorbing it.
Pinned in tests/Aetheria.Shared.Tests/ShieldReserveTests.cs:
RestorationProceedsWhileTheItemIsInactive (F4),
CanTakeHitNeverMutatesRegardlessOfHowManyTimesItIsCalled and
BreakIsIdempotentAcrossRepeatedCalls (F5). Updated every existing test that
relied on CanTakeHit's old mutating side effect to call Break() explicitly,
matching the new caller pattern.
Cannot compile-verify Assets/Scripts/Gameplay/Weapons/*.cs here (UnityEngine,
Unity batchmode unavailable in this environment) -- reviewed by hand for
brace/control-flow correctness; Self or a Unity batchmode pass should confirm.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
AllocateTiers treated any tier whose summed demand was at or below 1e-4 as
"fully met" (ratio 1) without ever subtracting it from `remaining`. Two bugs:
a tier with no consumers equipped in it at all read as 1f, the exact "fully
supplied" signal a UI is meant to trust, even on a fully blacked-out ship
(measured: Critical 0 demand reads "fine" right next to a genuinely starved
High); and a tier with a real but tiny demand (down to 5e-5) was granted out
of nothing and never charged against the ship's actual generation, letting a
lower tier get overfed by the sliver the higher one was never billed for.
AllocateTiers now tracks which tiers actually have a consumer separately from
their summed demand. A tier with no consumer reports the new PowerBus.NoDemand
sentinel (NaN), never 1f. A tier with a consumer whose own request truly is
zero this tick is trivially satisfied (ratio 1, nothing to be short of). Any
other demand, however small, goes through the same grant/remaining arithmetic
as every other tier.
Pinned in tests/Aetheria.Shared.Tests/PowerTierTests.cs:
SubEpsilonDemandGetsNoGrantAtTotalBlackout,
HigherTierSubEpsilonDemandIsStillSubtractedFromRemaining, and
EmptyTierAtBlackoutReportsNoDemandNotFullySupplied -- all fail under the
reverted (<=1e-4 free-pass) mutation. Updated
ATierWithNoDemandPassesItsShareDown's own empty-tier assertion, which had
pinned the bug itself ("empty tiers read as fully met").
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
PowerBus.Step billed every equipped IPowerConsumer's request unconditionally
and admitted a reactor on Item.Online alone, but Entity.Update only ever runs
a behaviour's Execute when the item is Active (Enabled && Online). A consumer
the player switched off never executes and never spends anything, yet was
still billed, starving live consumers and draining capacitors for nobody. A
reactor switched off stayed Online and kept generating even though
Reactor.Execute (which produces the heat receipt) never ran for it: free,
heat-free power.
Both directions now gate on the same predicate the rest of Entity uses:
item.Active.Value.
Pinned in tests/Aetheria.Shared.Tests/PowerBusTests.cs:
DisabledConsumerContributesNoDemand and DisabledReactorGeneratesNoPower --
both fail under the reverted (Online-only / unconditional) mutation.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
32 of 51 designs declare roles, 183 stats point at one, 28 of 37 products
author matching role quality. Adds the validation that refuses a stat
naming a role its design lacks, and roles-migrate alongside the other
catalog tools. Product means are placeholders pending operator review.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
PerformanceStat.Evaluate folded PowerSupply into the same Min/Max lerp factor
as Heat/Durability/Quality, so zero supply bottomed out at Min instead of 0
(measured: a radiator at zero power still pumped 19-25% of full; three of Cut
7's eight authored terms had Min == Max and were completely inert, since
lerp(x, x, anything) == x).
PowerSupply terms are now accumulated into their own multiplier and applied to
the fully-interpolated, fully-modified value instead of to the interpolation
factor. The term still lives in PerformanceStat.Terms -- it is still authored
per stat and still censused/validated the same way -- it is just no longer
blended into `factor`. Corrected the comments that stated the old premise
(Radiator.cs, Entity.cs, ItemData.cs's IStatContext.PowerSupplyFactor doc).
Pinned in tests/Aetheria.Shared.Tests/PowerCurveTests.cs:
APowerSupplyTermedStatWithNonzeroMinResolvesToExactlyZeroAtZeroSupply and
AMinEqualsMaxStatStillRespondsToItsPowerSupplyTerm -- both fail under the
reverted (lerp-based) mutation.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
roles-migrate apply, run once: 32 of 51 designs now declare roles (183
stats pointed at the role that governs them; the other 19 designs' kinds
either carry no quality-bearing behaviour at all, or every candidate stat
is authored flat with no observable role-vs-workmanship difference). 28 of
37 products authored matching ProductRole quality; the rest sell an
unrolled design or a design no product sells (Core Power, Industrial
Thermostatic Heater, pswarm, plight -- all already flagged unsold in the
Cut 1 census).
docs/stats-power-cut7-roles.md is the legible review table the map's O7
calls for: every design's roles and which stat took each, grouped by item
kind, plus every product's authored mean/deviation and its basis (one real
brand identity -- Arctica/Alakrita on heat efficiency, from
item-provenance-target.md -- the rest generic placeholders flagged for
operator review).
Verified before/after: AetherDb dangling/station-fit/hardpoint-fit output
is byte-identical; `loadout 1` output is byte-identical (no generation
behaviour moved); census reports 0 designs with roles and no matching
product spread; 144 Aetheria.Shared.Tests pass (was 141); the six-mutation
harness in mutation_tests_stats_power_cut8.py all behave as declared.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
roles-migrate declares roles on designs, points each design's own
StatTerm.Role at the role that governs it, and authors matching ProductRole
quality on every product selling a now-role-bearing design -- all through
AetherDb's own CultCache-backed types, one commit, nothing written if any
record fails to derive (same all-or-nothing contract as shield-migrate and
brownout-migrate). Field->role maps are per item kind (Weapon/Ballistic,
Weapon/Energy, Weapon/Launcher, Radiator, Reactor, Sensors, Thruster,
AetherDrive, Hull/Ship); Tool is keyed by design name since it is not a
shared kind. A stat only gets a role when it carries an unrouted
StatSource.Quality term AND still varies (Min != Max) -- Cut 1's flat
placeholder stats carry no observable role-vs-workmanship difference, so
routing one would be authoring fiction.
Product role means are deterministic generic placeholders (stable FNV hash
of product+role, not System.String.GetHashCode, which is randomized per
process and would make a dry run and the later apply disagree) flagged
"needs operator review" in the printed table, except Arctica/Alakrita's
`fins`, which item-provenance-target.md already gives a real brand identity
(heat efficiency).
census gains a role-coverage check: every sold, role-bearing design's
product must author a matching ProductRole spread, reported as "(product,
role) pairs where ... authors no matching spread" -- the gap the ruling
names directly (a newly minted lot from a product with no spread rolls the
default mean .5, silently pulling every maker toward average).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
StatValidation.ValidateRoleUsage walks every PerformanceStat field on every
behaviour and refuses a StatSource.Quality term whose Role is not in the
design's own declared Roles -- wired into AetheriaStores.Open and
CultRecordRefs.Upsert alongside the existing heat/modifier/power-supply
checks. Lot.QualityForRole already falls back to workmanship for a role an
old lot never rolled (forward-compat for a role added after minting); this
is the other direction: a term naming a role the design never declared at
all is an authoring mistake, not a lot predating the role, and must fail
loudly.
RoleAuthoringTests.cs pins three rules with both a positive and negative
control: Upsert refuses an absent-role term but accepts a declared one, and
two products of one design with different ProductRole means mint lots that
resolve a role-termed stat to different, deterministic values.
tests/mutation_tests_stats_power_cut8.py (this repo's own numbering follows
cut7.py's brownout work) mutates ValidateRoleUsage in both directions
(never-refuse, always-refuse) and QualityForRole's fallback, plus
ItemManager.CreateLot's mean read -- all six behave as declared.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
One ratio per item, resolved performance against the same item in perfect
conditions (quality and lot stay part of perfect; only durability, heat and
power count as degraded). Thruster and drive particles multiply intent by
it, and presentation stops computing durability itself.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
