Restore the saturated intercept count by uncueing the embedded radar (#145)
build / mod (push) Failing after 31s

The heavy WASM scenario (wasm_flat_budget, 64 integrated complexes vs two
`airdef wave oreshnik 40 2600`) lost intercepts at the tick-2102 checkpoint:
522 (main) -> 302 (#133) -> 243 (#134). Decomposed by A/B on one fixture, one
host (2bcce64), one SDK, each slice built from its own commit:

- 522 -> 302 (#133): Track.vel became an estimate from noisy measured places
  (it was the target's true velocity); the forecast, the aim lead and the
  interceptor's launch Fix all read it. Forcing the class estimate to Unknown
  leaves the count unchanged (native 296 ~= 294), so class/roles is not the
  cause. The pre-#133 522 used the target's true velocity, which the #133
  contract "class and speed only from measurements" forbids; this portion is
  the intentional cost of honest measurement-only tracks and is kept.

- 302 -> 243 (#134): the network cue was extended from the separate engagement
  radar to every fire-quality radar, including each weapon's embedded radar.
  Cueing every relayed contact each tick flooded the embedded fire control with
  one-tick plots whose noisy velocity degraded guidance. Restricting the cue
  back to separate radars recovers the whole drop: 243 -> 325.

- own_fire_within / FIRE_QUALITY / channels are NOT causes here: relaxing the
  own-fire gate or the channel limit each changed nothing (243 -> 243). Embedded
  radars always measure Fire plots of quality >= 64 and have no channel limit.

- #140 does not affect this loadout, by fact: the fixture deploys no separate
  radar, so every profile is embedded with channels = 0; the channel-age bug
  binds only where a profile has channels, and the channels-off A/B was flat.

Fix: the cue looks only through a separate fire-quality radar
(sam.kind == SamKind::Radar); the embedded radar keeps its beam/revisit/lock
acquisition as before #134. All radar profiles, channel limits, own-fire
authority and the measurement-only class/speed contract are preserved; no
target truth is injected.

Measured after the fix (WASM, fuel budget 300M/tick): checkpoint 325 intercepts
/ 2011 hits, fuel max 118M; final (tick 3302, two late Smart raids) 385 / 2472,
fuel max 176M; zero fuel failures. The first late raid reports the #135 recon
flight ("Разведка противника: впереди налёта высланы разведчики"); the second
is silent (recon budget spent) — the fixture now expects exactly that instead
of the stale empty chat that stopped the scenario.

Tests:
- airdef-mod/tests/saturation.rs: deterministic native replica of the checkpoint
  (same sim, host block tracer) asserting the restored count and a regression
  threshold, plus the #140 loadout fact (no separate radar, channels = 0).
- Sim::launch_audit (why.rs): on-demand tolerance diagnostic counting why each
  (system, drone) pair holds fire — no track / stale / no authority / role /
  doctrine / reach / claimed / not ready / not aimed / terrain / ready.
- sensor_tests: a network plot cues a separate fire radar but not the embedded one.
- wasm_flat_budget baselines re-measured with provenance comments; the asserted
  fixture SHA in tools/airdef.py updated.

Verification: `python3 tools/airdef.py test-budgets` passes both stages and the
two late raids (283 steps, fuel failures = 0). airdef-core 292 + 8, mod lib 104,
mod scenarios 5, saturation 2 — all green. A/B record:
docs/measurements/issue-145-intercept-regression.json.

Co-Authored-By: Kimi Code <noreply@moonshot.ai>

Integration note (shipped engine 8e711f969): the tick-2102 checkpoint baseline is
unchanged and passes on the shipped engine (325 intercepts / 2011 hits). The final
stats baseline was re-measured on the shipped engine with #146 integrated (one
knowledge snapshot per Smart order changes the two late raids): 414 / 2463, world
8885 sections / 420671 rays / 3442845 cells. The asserted fixture SHA in
tools/airdef.py was updated to match. python3 tools/airdef.py test-budgets passes
all 283 steps with zero fuel failures (tick max 184608537 / 300000000).
This commit is contained in:
2026-11-21 10:24:56 +03:00
parent c76f620642
commit b796a439df
9 changed files with 606 additions and 7 deletions
@@ -0,0 +1,58 @@
{
"schema": 1,
"issue": 145,
"title": "Intercept regression under WASM saturation: 522 -> 302 (#133) -> 243 (#134)",
"scenario": "airdef-mod/tests/budgets/wasm_flat_budget.toml",
"checkpoint_tick": 2102,
"engine": "2bcce642362ab5d3f04de193e390d8293cd1bf30",
"fuel_budget_per_tick": 300000000,
"method": "One unchanged fixture, one WASM host (cargo-horizon-mod test --fuel-profile budget), one SDK; each slice built from its own commit against the pinned engine. The checkpoint stats line is read by a deliberate expect.feedback mismatch that prints the real line; fuel counters come from the host. Native replica: airdef-mod/tests/saturation.rs (same sim, host block tracer, deterministic).",
"ab_checkpoint": [
{"slice": "main release v0.4.0-rc1", "commit": "c525feb", "kills": 522, "hits": 1789, "missiles_airborne": 5},
{"slice": "parent of #133", "commit": "95739c95b", "kills": 522, "hits": 1789, "missiles_airborne": 5, "fuel_max_tick": 113866321},
{"slice": "#133 class/speed from measurements", "commit": "ac7521936", "kills": 302, "hits": 2024, "missiles_airborne": 8, "fuel_max_tick": 114068675},
{"slice": "#134 radar profiles", "commit": "858735a4c", "kills": 243, "hits": 2083, "missiles_airborne": 24, "fuel_max_tick": 135803282},
{"slice": "kimi-work HEAD", "commit": "a983177d3", "kills": 243, "hits": 2083, "missiles_airborne": 24, "fuel_max_tick": 136067073}
],
"decomposition_variants_wasm": [
{"variant": "#133 with class estimate forced Unknown", "base": "ac7521936", "kills": 249, "hits": 2092, "note": "confounded: Unknown also disables last-chance and halves the ballistic salvo; class/roles is not the #133 cause (native class-off 296 ~= baseline 294)"},
{"variant": "#134 cue restricted to separate radars (cue-off)", "base": "858735a4c", "kills": 325, "hits": 2011, "fuel_max_tick": 124349576, "note": "recovers the whole #134 drop: the embedded cue flood is the cause"},
{"variant": "#134 own_fire_within relaxed to own_within (no FIRE_QUALITY/product gate)", "base": "858735a4c", "kills": 243, "hits": 2083, "fuel_max_tick": 135800131, "note": "zero effect: embedded radars always produce Fire plots of quality >= 64"},
{"variant": "#134 channel limit never binds", "base": "858735a4c", "kills": 243, "hits": 2083, "fuel_max_tick": 135438135, "note": "zero effect: embedded profiles have channels = 0"},
{"variant": "cue precise + carry network velocity (Fix D)", "base": "a983177d3", "kills": 260, "hits": 2051, "note": "partial: early cue authority fires with a stale carried velocity"},
{"variant": "cue restricted to separate fire-quality radars (Fix F, chosen)", "base": "a983177d3", "kills": 325, "hits": 2011, "fuel_max_tick": 118325396, "note": "the fix; restores embedded acquisition to pre-#134 beam/revisit/lock"}
],
"native_replica_checkpoint": {
"note": "Deterministic native replica (tests/saturation.rs); chaotic combat answers render rays a tick later than the WASM host, so counters sit a few apart from WASM. Same drops, same fix.",
"pre133": {"kills": 605, "hits": 1706},
"s133": {"kills": 294, "hits": 2037},
"s134_head": {"kills": 247, "hits": 2079},
"class_off": {"kills": 296, "hits": 2045},
"cue_off": {"kills": 314, "hits": 2022},
"authority_relaxed": {"kills": 247, "hits": 2079},
"channels_off": {"kills": 247, "hits": 2079},
"fix_f": {"kills": 314, "hits": 2022}
},
"causes": {
"drop_133_522_to_302": "Track.vel became an estimate from noisy measured places (was the target's true velocity); the forecast, the aim lead and the interceptor's launch Fix all read it. Class-off ~= baseline, so class/roles is not the cause. The pre-#133 522 used the target's true velocity, which the #133 contract ('class and speed only from measurements') forbids; this portion is the intentional cost of honest measurement-only tracks.",
"drop_134_302_to_243": "#134 extended the network cue from the separate engagement radar to every fire-quality radar, including each weapon's embedded radar. Cueing every relayed contact each tick flooded the embedded fire control with one-tick plots whose noisy velocity degraded guidance. Restricting the cue back to separate radars recovers the drop (243 -> 325).",
"own_fire_FIRE_QUALITY_channels": "Not causes here: relaxing the own-fire gate or the channel limit each changed nothing (243 -> 243). Embedded radars always measure Fire plots of quality >= 64 and have no channel limit.",
"issue_140": "Does not affect this loadout, by fact: the fixture deploys 64 integrated complexes and no separate radar, so every profile is embedded with channels = 0; the channel-age bug binds only where a profile has channels. The channels-off A/B changed nothing."
},
"fix": {
"file": "airdef-mod/src/sim/sight.rs",
"change": "The network cue looks only through a separate fire-quality radar (sam.kind == SamKind::Radar), never the weapon's embedded radar; the embedded radar keeps its beam/revisit/lock acquisition as before #134.",
"preserved": "All radar profiles, channel limits, own-fire authority (own_fire_within / FIRE_QUALITY), and the measurement-only class/speed contract (#133). No target truth is injected.",
"tests": [
"airdef-mod/tests/saturation.rs (deterministic checkpoint replica + audit + #140 loadout fact)",
"airdef-mod/src/sim/sensor_tests.rs::a_network_plot_cues_a_separate_fire_radar_but_not_the_embedded_one",
"airdef-mod/src/sim/why.rs::Sim::launch_audit (on-demand tolerance diagnostic)"
]
},
"after_fix_wasm": {
"checkpoint_tick_2102": {"kills": 325, "hits": 2011, "airborne": 184, "missiles_airborne": 4, "fuel_max_tick": 118325396, "fuel_failures": 0},
"final_tick_3302": {"kills": 385, "hits": 2472, "airborne": 9, "missiles_airborne": 2, "fuel_max_tick": 176588234, "fuel_failures": 0},
"late_raid_chat": ["Разведка противника: впереди налёта высланы разведчики"]
},
"target_range": "Post-fix checkpoint ~325 intercepts (WASM) / 314 (native replica) at tick 2102, fuel max ~118M/300M. The residual gap to the pre-#133 522 is the measurement-only velocity contract cost (#133), kept deliberately: class and speed come from measurements, never the target's own figures."
}