Server Init Deadspawn And Airfield Probe

Server-Init Deadspawn Enclosure and Airfield Probe

Source-verified 2026-06-21 against master 0139a346. Paths relative to Missions/[55-2hc]warfarev2_073v48co.chernarus/ unless noted. Arma 2 OA 1.64.

Two custom claude-gaming server-only one-shots run back-to-back near the end of server init: Init_DeadspawnWall.sqf and Init_AirfieldProbe.sqf. Both are fired once from Server/Init/Init_Server.sqf immediately after serverInitFull = true (Server/Init/Init_Server.sqf:649): the deadspawn wall at Server/Init/Init_Server.sqf:657 and the airfield probe at :664. The first is a physical protection pass that rings the three per-side temporary-respawn markers with an indestructible H-barrier wall so an enemy-side AI bot cannot shoot a human parked on an adjacent side's holding marker during join. The second is a diagnostic-only probe that spawns nothing and instead logs a 5x5 grid of candidate airfield-camp positions to the RPT so blind coordinate guesses can be refuted with real surfaceIsWater/nearRoads data.

"Deadspawn" disambiguation (cmdcon30, 2026-06-30). This page owns the physical deadspawn only: the H-barrier wall ringing the per-side respawn markers so an adjacent-marker bot cannot shoot a parked, joining human. cmdcon30 introduced a second, unrelated meaning. The cmdcon30 "player stuck in deadspawn — no team / no HUD / no markers" bug is an enrollment-resolver failure, not a physical-protection issue: the PR#122 editor-slot reaper deleteGroup'd a JIP-selectable slot's group, the joiner landed in a group with no wfbe_side, and the enrollment self-heal exhausted its retries. If you arrived here chasing that report, go to JIP-Enrollment-And-Client-Data-Delivery and the group-GC reaper page — this page does not cover it.

This page is distinct from Server-Init-Bind-Cleanup, which covers duplicate compile/bind hygiene in Init_Server.sqf and touches neither of these scripts.

Boot Invocation (Init_Server.sqf)

Script	Call site	Form	Gate	Side-effect class
Server/Init/Init_DeadspawnWall.sqf	Server/Init/Init_Server.sqf:657	[] execVM "..."	if (!isServer) exitWith {} (Init_DeadspawnWall.sqf:41)	Spawns persistent global objects (H-barriers)
Server/Init/Init_AirfieldProbe.sqf	Server/Init/Init_Server.sqf:664	[] execVM "..."	if (!isServer) exitWith {} (Init_AirfieldProbe.sqf:25) plus wfbe_airfield_probe_done one-shot guard (:29-30)	None — diagnostic diag_log only

Both are placed after the world/logic init completes (serverInitFull = true at Init_Server.sqf:649) and before the town/economy server FSMs are spawned (server_town.sqf at :667).

Init_DeadspawnWall: the problem

The three per-side holding markers — WestTempRespawnMarker, EastTempRespawnMarker, GuerTempRespawnMarker (Init_DeadspawnWall.sqf:58) — sit only 64–128m apart on a bare NE-Chernarus mountaintop (all three are static in mission.sqm; a TempRespawnMarker grep returns 3 entries). AI-slot bots respawn exactly onto their own side's marker: Server/AI/AI_AdvancedRespawn.sqf:29 does _respawnedUnit setPos getMarkerPos Format["%1TempRespawnMarker",_sideText]. Because the markers are so close, an enemy-side bot has clear line-of-fire onto a HUMAN parked on an adjacent side's marker during join (the "AI killed in the deadspawn" / Smarty kill). The header records the worst sightlines as GUER↔EAST ~44.5m and GUER↔WEST ~52m (Init_DeadspawnWall.sqf:5-12).

The fix is purely additive: it rings each marker with a closed wall and touches no protected file. It deliberately does not trap players, because the joiner is teleported to base by the client-side join handshake and never walks out of the ring, and a 120s client watchdog re-enables damage even if the move stalls (Client/Init/Init_Client.sqf:21-26, the WFBE_Client_DeadspawnEscaped / 120s timeout block). It never calls setCaptive or disableAI, keeping the standing hard guardrail intact (Init_DeadspawnWall.sqf:14-23).

Init_DeadspawnWall: ring geometry

Tunable	Value	Source	Meaning
_wallCls	"Land_HBarrier_large"	Init_DeadspawnWall.sqf:47	Tall LoS/line-of-fire blocker, confirmed-present class on this CO+EP1 box (reused from Server/Init/Init_Defenses.sqf wall templates)
_radius	12 (m)	:52	Half-width of the ring → ~24m box; nearest neighbour marker is 64.5m so rings never overlap
_segLen	5 (m)	:56	Approx footprint of one H-barrier section; drives segment count
_perim	2 * 3.14159 * _radius	:78	Ring circumference
_count	ceil (_perim / _segLen), floored to min 8	:79-80	Number of barrier segments; min-8 floor guarantees even a tiny ring is sealed
_step	360 / _count	:81	Angular spacing between segments

For _radius = 12, _perim ≈ 75.4m and _count = ceil(75.4/5) = 16 segments per marker (above the 8 floor), so a full run spawns ~48 barriers across the three markers.

Init_DeadspawnWall: per-marker build loop

The script forEach-iterates the three markers (:62-116):

1. Resolve centre. _c = getMarkerPos _mk (:64). getMarkerPos returns [0,0,0] for an unknown name, so if both X and Y are 0 (:70) it logs a WARNING via WFBE_CO_FNC_LogContent and skips only this marker using a guarded else block rather than exitWith — exitWith would abort the whole forEach and skip the remaining sides (:66-72).
2. Walk the circle. For each segment _i from 0 to _count-1 (:84): bearing _ang = _i * _step (:85), point on the ring at _px = (_c select 0) + _radius * sin _ang, _py = (_c select 1) + _radius * cos _ang (:88-89).
3. Spawn the barrier. _prop = createVehicle [_wallCls, _pos, [], 0, "NONE"] (:93) — createVehicle (GLOBAL) is used deliberately so the wall is server-authoritative and AI LoS/collision sees it on a dedicated server, per the standing rule against createVehicleLocal here (:28-30).
4. Graceful class-miss handling. If isNull _prop (:94), log one WARNING and keep going — modelled on Server/Functions/Server_SpawnStructureDressing.sqf:48-51 (:36-38, verified: same isNull/WARNING pattern).
5. Orient and pin. setDir (_ang + 90) orients the long face TANGENT to the ring so consecutive barriers overlap into a continuous wall instead of radial spokes (:98-100); setPosATL [_px,_py,0] pins each barrier to the marker's own terrain elevation (the three markers differ by up to ~12m, so no height is hardcoded — :34-35, :101-102); setVectorUp [0,0,1] (:103).
6. Make it cheap and permanent. enableSimulation false (no physics cost) and allowDamage false (bots can't shoot the wall down) — :104-106.

Each surviving prop is appended to _allProps (:107), and a per-marker INITIALIZATION line reports the marker, position, segment count, and radius (:112).

Init_DeadspawnWall: handle storage

After the loop, the prop list is stored non-broadcast for debugging / potential teardown: missionNamespace setVariable ["WFBE_DEADSPAWN_WALL_PROPS", _allProps] (:119) — the 2-argument setVariable form (no true publish flag), so it stays server-local. A closing INITIALIZATION line reports the total barrier count across the side markers (:121). The variable is set in exactly one place and read nowhere else in the mission (verified).

Init_AirfieldProbe: diagnostic-only grid scan

Init_AirfieldProbe.sqf changes NO coordinates and spawns NO objects (:1-4). Its purpose is to settle where the airfield capture camps should be placed: the airfield camps (mission.sqm LocationLogicCamp id=308 NWAF / id=310 Balota) sit ~300m south of the real airfields, and blind coordinate guesses were refuted, so instead of guessing it probes (:6-8).

Aspect	Detail	Source
Side gate	if (!isServer) exitWith {}	Init_AirfieldProbe.sqf:25
Async body	[] spawn { ... }	:27
One-shot guard	if (!isNil "wfbe_airfield_probe_done") exitWith {} then wfbe_airfield_probe_done = true	:29-30
Settle delay	uiSleep 20 (so world/roads are loaded before sampling)	:32
Offsets per axis	_offsets = [-120, -40, 0, 40, 120] (5 samples → 5x5 = 25 candidates/field)	:34-36
Airport anchors	["Balota", 4550, 2280], ["NWAF", 4479.3252, 10618.404] (from mission.sqm LocationLogicAirport, Balota id=7 / NWAF id=8)	:38-42

For every candidate it builds _pos = [_ax + _px, _ay + _py, 0] (:57) and samples two on-land validity signals:

• _water = surfaceIsWater _pos (:59) — true means in the sea, reject.
• _roads = count (_pos nearRoads 8) (:61) — a non-empty list means on/at a road, reject for an apron.

It then emits one greppable pipe line per candidate via diag_log (:63-70):

AIRFIELD_PROBE|v1|field:<NWAF|Balota>|x:..|y:..|water:<bool>|roads:<count>

The run is bracketed by AIRFIELD_PROBE|v1|begin|grid:5x5|offsets:-120,-40,0,40,120 (:44) and AIRFIELD_PROBE|v1|end (:75), so an operator can isolate the whole block in the RPT and pick a verified on-land, off-road apron. The nested forEach/forEach/forEach structure (fields → X offsets → Y offsets) reuses the _x iterator name at each level (:48, :53, :55), which is valid A2-OA scoping.

A2-OA command discipline

Both scripts are written to A2-OA-only command rules. The probe header explicitly notes it uses surfaceIsWater, nearRoads, getPosASL/getPosATL and NO isOnRoad / A3 commands (Init_AirfieldProbe.sqf:22). The wall uses createVehicle (not createVehicleLocal), setDir, setPosATL, setVectorUp, enableSimulation, allowDamage, capitalized Private/Format, and Call WFBE_CO_FNC_LogContent — all valid A2-OA forms. Neither script uses remoteExec, BIS_fnc_MP, isEqualTo, or any A3-only construct.

Continue Reading

• Server-Init-Bind-Cleanup
• Mission-Entrypoints-And-Lifecycle
• Lifecycle-Wait-Chain
• Respawn-And-Death-Lifecycle-Atlas
• Server-Gameplay-Runtime-Atlas