spaceship/enemies/EnemySpawnerComponent.gd

78 lines
3.1 KiB
GDScript3
Raw Normal View History

Turret vessel spawner enemy (#52) New EnemySpawnerComponent.gd (capability_id "spawner"): continuously instantiates another enemy type beside its host at a fixed interval, capped at max_alive concurrently, refilling as spawns die. Spawned enemies are parented under the SPAWNER'S OWN PARENT rather than the spawner itself, so they are siblings, not descendants - killing the spawner can't free them, and a surface-hosted spawner (under PlanetSetupHost) keeps its spawns under the same host instead of leaking them to a hardcoded path, the same hazard #44 already closed for pursuit/standoff/ranged-weapon components. TurretSpawnerEnemyDefinition.tres: stationary (speed 0, no pursuit or standoff component), max_hp 20 (~7x a chaser), a bulky octagon hull, and its own acid-green glow distinct from chaser/sniper/artillery. Registered in EnemyDefinitions.ORDER/DEFINITIONS - the map editor palette (#48) picks it up with no editor code change. Tests: new tests/test_turret_spawner_enemy.gd covers registration and real definition defaults, no spawns without a fleet, spawning capped at max_alive, a dead spawn's slot refilling, never moving, surviving 19 damage and dying on the 20th, and its spawns surviving its death with their pursuit component intact. Screenshot beside a chaser confirms the two read as visually distinct enemies. Full suite: 79/79. Repo notes for the next agent, confirmed empirically this session: - A node's set_process(false) call made before it has fully finished entering the tree doesn't reliably stick (already documented in tests/test_planet_atmosphere_transition.gd, and it bit this test too): BuildController._check_fleet_wipe() kept rebuilding a ship during an intended "empty fleet" test window even after disabling it in _initialize(). Do it in _process()'s first real tick instead, once _ready() cascades are confirmed complete. - Overriding an @export interval (like spawn_interval) on an already-running component does NOT reset its own in-progress countdown timer - only changes what it resets TO on its next expiry. If the countdown already started at the OLD value before your override lands, reset the timer variable itself too, or the first real interval after the override silently uses the stale value. - A capped spawner's freed-slot refill can land within a couple of frames once its interval is short - don't assert on an intermediate "just dropped by one" count that a fast interval can blow straight through before your next check runs. Assert the net effect instead (the specific killed instance is gone AND a replacement filled the slot) rather than an exact mid-transition snapshot. 🤖 Generated with Claude Code Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-08-16 09:43:44 +02:00
class_name EnemySpawnerComponent
extends Node
# Continuously spawns another enemy type beside its host, at a fixed
# interval, capped at max_alive concurrently. The spawned enemies are
# added as children of the spawner's OWN PARENT - not the spawner itself -
# so killing the spawner never frees its already-spawned children (they
# are siblings, not descendants) and so a surface-hosted spawner keeps its
# spawns under the same host a `../Fleet`-style path assumption would miss
# (see #44's note on why that hazard exists for anything parented under a
# planet surface's PlanetSetupHost rather than directly under Main.tscn).
var capability_id: String = "spawner"
@export var spawn_interval: float = 4.0
@export var spawned_enemy_id: String = "chaser"
@export var max_alive: int = 6
@export var spawn_radius: float = 90.0
var spawn_timer: float = 0.0
var spawned: Array = []
Enemy mothership (#53) A large, slow capital enemy that periodically launches chasers and artillery on their own intervals and fires a heavily telegraphed line beam capable of killing several player vessels at once. Spawning reuses #52's EnemySpawnerComponent as-is via two preset scenes (EnemySpawnerComponentChaser.tscn: 5s/chaser, EnemySpawnerComponentArtillery.tscn: 18s/max 2/artillery) rather than adding multi-type support to the component - two instances of a single-purpose component is the composition this codebase already uses everywhere ("enemy bomber" in the original request is read here as the existing artillery enemy's slow telegraphed AOE, not a new enemy type - flagged as a judgment call in the ticket). New EnemyBigLaserComponent.gd ("big_laser") fires a standalone EnemyBigLaserBeam on a long cooldown - the beam owns its whole charge/ fire/flash sequence once launched, so the mothership dying mid-charge doesn't cancel an already-telegraphed shot, same design as EnemyArtilleryComponent/ArtilleryShell. The aim direction is locked once in launch() and never re-read - nothing in the charging phase touches the fleet's position again, which is what makes it dodgeable. Damage is a line-segment check (projection + perpendicular distance) against every player_ships member; Ship.take_damage() already handles shield absorption per-ship. MothershipEnemyDefinition.tres: max_hp 60, speed 35 (slow), a ~96-unit hull, pursuit + two spawners + big_laser, no type switch anywhere - every stat from the definition, every behaviour from a component. Tests: new tests/test_enemy_mothership.gd, split into beam mechanics (tested directly, not through the component's own cooldown timing, which would make the geometry assertions timing-fragile) and the composed definition (tested through a real Enemy.tscn instance). Screenshot of the charge telegraph mid-countdown. Full suite: 80/80. Found and fixed a real bug along the way, in EnemySpawnerComponent.gd (#52): its `enemy_scene: PackedScene = preload("res://Enemy.tscn")` member-level preload could get permanently baked in as a broken, empty PackedScene when a definition's component_scenes chain led back to Enemy.tscn while it was still mid-load - confirmed empirically via ResourceLoader.has_cached() being true (a good copy existed) while the preloaded reference's own resource_path was empty. This only surfaces when a SINGLE definition composes two instances of a script carrying such a preload (the mothership's chaser/artillery spawner presets; #52's single-spawner turret never hit it). Fixed by resolving enemy_scene lazily via load() on first use instead of at compile time, and additionally made a fresh spawner count down from a full interval before its first spawn rather than firing on tick one, which also happens to keep two sibling spawners with different intervals from ever landing their first attempt on the same frame. 🤖 Generated with Claude Code Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-08-16 10:19:02 +02:00
# Resolved lazily via load() on first use rather than preload()d as a
# compile-time constant: when a definition's component_scenes chain leads
# BACK to Enemy.tscn while it is still mid-load (e.g. two different preset
# scenes of this same script both live on one definition, as the
# mothership's chaser/artillery spawners are), the preloaded constant can
# get permanently baked in as a broken, empty PackedScene - confirmed by
# ResourceLoader.has_cached() being true (a good copy exists) while the
# preloaded reference's own resource_path was empty. load() at first real
# use instead always resolves against whatever is actually in the cache
# by then.
var enemy_scene: PackedScene = null
func _ready() -> void:
# Start counting down from a full interval rather than firing on the
# very first tick: besides reading more naturally (a freshly-placed
# spawner doesn't insta-spawn), this keeps two sibling spawners with
# different intervals (e.g. the mothership's chaser/artillery presets)
# from both landing their first attempt on the exact same physics
# frame - which hit a transient PackedScene.instantiate() failure on
# this engine build when it happened.
spawn_timer = spawn_interval
Turret vessel spawner enemy (#52) New EnemySpawnerComponent.gd (capability_id "spawner"): continuously instantiates another enemy type beside its host at a fixed interval, capped at max_alive concurrently, refilling as spawns die. Spawned enemies are parented under the SPAWNER'S OWN PARENT rather than the spawner itself, so they are siblings, not descendants - killing the spawner can't free them, and a surface-hosted spawner (under PlanetSetupHost) keeps its spawns under the same host instead of leaking them to a hardcoded path, the same hazard #44 already closed for pursuit/standoff/ranged-weapon components. TurretSpawnerEnemyDefinition.tres: stationary (speed 0, no pursuit or standoff component), max_hp 20 (~7x a chaser), a bulky octagon hull, and its own acid-green glow distinct from chaser/sniper/artillery. Registered in EnemyDefinitions.ORDER/DEFINITIONS - the map editor palette (#48) picks it up with no editor code change. Tests: new tests/test_turret_spawner_enemy.gd covers registration and real definition defaults, no spawns without a fleet, spawning capped at max_alive, a dead spawn's slot refilling, never moving, surviving 19 damage and dying on the 20th, and its spawns surviving its death with their pursuit component intact. Screenshot beside a chaser confirms the two read as visually distinct enemies. Full suite: 79/79. Repo notes for the next agent, confirmed empirically this session: - A node's set_process(false) call made before it has fully finished entering the tree doesn't reliably stick (already documented in tests/test_planet_atmosphere_transition.gd, and it bit this test too): BuildController._check_fleet_wipe() kept rebuilding a ship during an intended "empty fleet" test window even after disabling it in _initialize(). Do it in _process()'s first real tick instead, once _ready() cascades are confirmed complete. - Overriding an @export interval (like spawn_interval) on an already-running component does NOT reset its own in-progress countdown timer - only changes what it resets TO on its next expiry. If the countdown already started at the OLD value before your override lands, reset the timer variable itself too, or the first real interval after the override silently uses the stale value. - A capped spawner's freed-slot refill can land within a couple of frames once its interval is short - don't assert on an intermediate "just dropped by one" count that a fast interval can blow straight through before your next check runs. Assert the net effect instead (the specific killed instance is gone AND a replacement filled the slot) rather than an exact mid-transition snapshot. 🤖 Generated with Claude Code Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-08-16 09:43:44 +02:00
func tick(delta: float) -> void:
spawned = spawned.filter(func(node): return is_instance_valid(node))
if spawned.size() >= max_alive:
return
spawn_timer -= delta
if spawn_timer > 0.0:
return
spawn_timer = spawn_interval
var fleet: Node2D = get_tree().get_first_node_in_group("player_fleet")
if fleet == null or fleet.ships.is_empty():
return
var spawner: Node2D = get_parent()
var host: Node = spawner.get_parent()
if host == null:
return
var angle: float = randf() * TAU
var offset: Vector2 = Vector2(cos(angle), sin(angle)) * spawn_radius
Enemy mothership (#53) A large, slow capital enemy that periodically launches chasers and artillery on their own intervals and fires a heavily telegraphed line beam capable of killing several player vessels at once. Spawning reuses #52's EnemySpawnerComponent as-is via two preset scenes (EnemySpawnerComponentChaser.tscn: 5s/chaser, EnemySpawnerComponentArtillery.tscn: 18s/max 2/artillery) rather than adding multi-type support to the component - two instances of a single-purpose component is the composition this codebase already uses everywhere ("enemy bomber" in the original request is read here as the existing artillery enemy's slow telegraphed AOE, not a new enemy type - flagged as a judgment call in the ticket). New EnemyBigLaserComponent.gd ("big_laser") fires a standalone EnemyBigLaserBeam on a long cooldown - the beam owns its whole charge/ fire/flash sequence once launched, so the mothership dying mid-charge doesn't cancel an already-telegraphed shot, same design as EnemyArtilleryComponent/ArtilleryShell. The aim direction is locked once in launch() and never re-read - nothing in the charging phase touches the fleet's position again, which is what makes it dodgeable. Damage is a line-segment check (projection + perpendicular distance) against every player_ships member; Ship.take_damage() already handles shield absorption per-ship. MothershipEnemyDefinition.tres: max_hp 60, speed 35 (slow), a ~96-unit hull, pursuit + two spawners + big_laser, no type switch anywhere - every stat from the definition, every behaviour from a component. Tests: new tests/test_enemy_mothership.gd, split into beam mechanics (tested directly, not through the component's own cooldown timing, which would make the geometry assertions timing-fragile) and the composed definition (tested through a real Enemy.tscn instance). Screenshot of the charge telegraph mid-countdown. Full suite: 80/80. Found and fixed a real bug along the way, in EnemySpawnerComponent.gd (#52): its `enemy_scene: PackedScene = preload("res://Enemy.tscn")` member-level preload could get permanently baked in as a broken, empty PackedScene when a definition's component_scenes chain led back to Enemy.tscn while it was still mid-load - confirmed empirically via ResourceLoader.has_cached() being true (a good copy existed) while the preloaded reference's own resource_path was empty. This only surfaces when a SINGLE definition composes two instances of a script carrying such a preload (the mothership's chaser/artillery spawner presets; #52's single-spawner turret never hit it). Fixed by resolving enemy_scene lazily via load() on first use instead of at compile time, and additionally made a fresh spawner count down from a full interval before its first spawn rather than firing on tick one, which also happens to keep two sibling spawners with different intervals from ever landing their first attempt on the same frame. 🤖 Generated with Claude Code Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-08-16 10:19:02 +02:00
if enemy_scene == null:
#84: move 215 root files into a folder hierarchy Everything lived in one flat directory, so a diff gave no clue what area it touched and finding the file you wanted meant knowing its exact name. Twelve folders now, grouped by what a thing IS rather than what type it is - a vessel's definition, its component scenes and its upgrade resources sit together, because that is the set you read when you change one of them. buildings core controllers enemies logistics planets resources shaders surface ui vessels vfx world 332 files moved and every res:// reference rewritten - 306 files touched, including the autoload and main-scene paths in project.godot. Nothing renamed: this is purely where files sit, so the diff is a pile of pure moves and no behaviour is buried in it. ui_theme.tres, icon.svg and web_shell.html stay at the root, where Godot and the export preset expect them. Verified three ways, because a suite alone would not have caught all of it: the full 117 pass, the web export builds clean, and the game renders. The export is why that matters. The FIRST export after the move failed with "Attempt to open script 'res://GameState.gd'" - the old path - while project.godot on disk plainly said res://core/GameState.gd. That is this project's documented Android FUSE mtime trap: Godot served a cached copy of a file whose content had changed. A `touch` fixed it. Nothing in the test suite would have found that, because the suite runs from a copy on a native filesystem - the one place the bug cannot happen. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-21 19:44:14 +02:00
enemy_scene = load("res://enemies/Enemy.tscn")
Turret vessel spawner enemy (#52) New EnemySpawnerComponent.gd (capability_id "spawner"): continuously instantiates another enemy type beside its host at a fixed interval, capped at max_alive concurrently, refilling as spawns die. Spawned enemies are parented under the SPAWNER'S OWN PARENT rather than the spawner itself, so they are siblings, not descendants - killing the spawner can't free them, and a surface-hosted spawner (under PlanetSetupHost) keeps its spawns under the same host instead of leaking them to a hardcoded path, the same hazard #44 already closed for pursuit/standoff/ranged-weapon components. TurretSpawnerEnemyDefinition.tres: stationary (speed 0, no pursuit or standoff component), max_hp 20 (~7x a chaser), a bulky octagon hull, and its own acid-green glow distinct from chaser/sniper/artillery. Registered in EnemyDefinitions.ORDER/DEFINITIONS - the map editor palette (#48) picks it up with no editor code change. Tests: new tests/test_turret_spawner_enemy.gd covers registration and real definition defaults, no spawns without a fleet, spawning capped at max_alive, a dead spawn's slot refilling, never moving, surviving 19 damage and dying on the 20th, and its spawns surviving its death with their pursuit component intact. Screenshot beside a chaser confirms the two read as visually distinct enemies. Full suite: 79/79. Repo notes for the next agent, confirmed empirically this session: - A node's set_process(false) call made before it has fully finished entering the tree doesn't reliably stick (already documented in tests/test_planet_atmosphere_transition.gd, and it bit this test too): BuildController._check_fleet_wipe() kept rebuilding a ship during an intended "empty fleet" test window even after disabling it in _initialize(). Do it in _process()'s first real tick instead, once _ready() cascades are confirmed complete. - Overriding an @export interval (like spawn_interval) on an already-running component does NOT reset its own in-progress countdown timer - only changes what it resets TO on its next expiry. If the countdown already started at the OLD value before your override lands, reset the timer variable itself too, or the first real interval after the override silently uses the stale value. - A capped spawner's freed-slot refill can land within a couple of frames once its interval is short - don't assert on an intermediate "just dropped by one" count that a fast interval can blow straight through before your next check runs. Assert the net effect instead (the specific killed instance is gone AND a replacement filled the slot) rather than an exact mid-transition snapshot. 🤖 Generated with Claude Code Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-08-16 09:43:44 +02:00
var new_enemy: Node2D = enemy_scene.instantiate()
new_enemy.enemy_id = spawned_enemy_id
# spawner.position is already expressed in `host`'s local space, and
# new_enemy is about to be added to that same `host` - unlike
# EnemyRangedWeaponComponent's projectile, which reparents to root and
# needs global_position, there is no space conversion needed here.
new_enemy.position = spawner.position + offset
host.add_child(new_enemy)
spawned.append(new_enemy)