Naive support for incoming mob damage - #564
Conversation
WalkthroughAdds server-side target resolution in SkillCastAttack for NPC attacks. When no targets are provided, computes targets via range prism and field queries, caps by TargetCount, records target ObjectIds, and routes through actor.TargetAttack to reuse the existing damage/effects pipeline. Adds a missing utility using directive. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor NPC as NPC Actor
participant SS as SkillState.SkillCastAttack
participant R as Attack.Range
participant F as Field
participant TA as actor.TargetAttack
participant DP as Damage/Eff. Pipeline
participant B as Broadcaster
NPC->>SS: SkillCastAttack(attack, attackTargets)
alt attackTargets provided
SS->>SS: resolvedTargets = attackTargets
else No targets provided
SS->>R: GetPrism(actor.Position, attack.Params)
R-->>SS: Area prism
SS->>F: GetTargets(prism, filters)
F-->>SS: resolvedTargets
end
SS->>SS: Cap to TargetCount (>=1)
SS->>SS: cast.Targets += target.ObjectId
SS->>TA: TargetAttack(cast)
TA->>DP: Calculate damage/effects
DP-->>TA: Results
TA->>B: Broadcast results
B-->>NPC: Notifications
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Suggested reviewers
Poem
✨ Finishing Touches
🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs (1)
79-84: Set Position/Direction when falling back to resolved targetsWhen no
attackTargetsare provided,cast.Position/cast.Directionaren’t set, which may affect effects/orientation-sensitive logic. Initialize them from the first resolved target.Apply this diff:
- int limit = attack.TargetCount > 0 ? attack.TargetCount : 1; + int limit = attack.TargetCount > 0 ? attack.TargetCount : 1; + // Initialize orientation for damage/effects when using fallback targets + cast.Position = actor.Position; + cast.Direction = Vector3.Normalize(resolvedTargets[0].Position - actor.Position); for (int i = 0; i < resolvedTargets.Count && i < limit; i++) { IActor target = resolvedTargets[i]; cast.Targets.TryAdd(target.ObjectId, target); }
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (1)
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs(2 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs (2)
Maple2.Server.Game/Model/Field/Actor/Actor.cs (3)
IActor(217-226)IActor(228-237)TargetAttack(179-211)Maple2.Server.Game/Model/Field/Actor/IActor.cs (3)
IActor(27-27)IActor(28-28)TargetAttack(30-30)
🔇 Additional comments (4)
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs (4)
5-5: Confirm necessity of new usingIf
GetPrismis an extension inMaple2.Server.Game.Util, keep this import; otherwise, drop it to avoid an unused import.
86-88: Good reuse of the existing damage/effects pipelineRouting mob damage through
actor.TargetAttack(cast)keeps behavior consistent with players and avoids duplicating logic.
69-88: Restrict SkillCastAttack to NPCs: MovementState.SkillCast invokes SkillState.SkillCastAttack for all actors—including players—so this block risks double-applying damage. Wrap the damage logic in an NPC-only guard (e.g.if (actor is MobActor)) or otherwise confirm it never runs for player actors.
71-77: Guard attack.Range before GetPrism & ensure correct target‐count handlingAdd a null check to avoid NRE when
attack.Rangeis missing, and convert a zeroattack.TargetCountinto an explicit “unlimited” limit ifGetTargetstreats 0 as “no results.” Verify your metadata andGetTargetsimplementation to confirm whether 0 means unlimited or should be handled differently.
Apply this diff:- if (resolvedTargets.Count == 0) { - // Fallback: query targets from attack range - Maple2.Tools.Collision.Prism prism = attack.Range.GetPrism(actor.Position, actor.Rotation.Z); - foreach (IActor target in actor.Field.GetTargets(actor, new[] { prism }, attack.Range.ApplyTarget, attack.TargetCount)) { - resolvedTargets.Add(target); - } - } + if (resolvedTargets.Count == 0 && attack.Range != null) { + // Fallback: query targets from attack range + Maple2.Tools.Collision.Prism prism = attack.Range.GetPrism(actor.Position, actor.Rotation.Z); + int effectiveLimit = attack.TargetCount > 0 ? attack.TargetCount : int.MaxValue; + foreach (IActor target in actor.Field.GetTargets(actor, new[] { prism }, attack.Range.ApplyTarget, effectiveLimit)) { + resolvedTargets.Add(target); + } + }
| if (resolvedTargets.Count > 0) { | ||
| int limit = attack.TargetCount > 0 ? attack.TargetCount : 1; | ||
| for (int i = 0; i < resolvedTargets.Count && i < limit; i++) { | ||
| IActor target = resolvedTargets[i]; | ||
| cast.Targets.TryAdd(target.ObjectId, target); | ||
| } | ||
|
|
||
| // Reuse existing pipeline to calculate and broadcast damage/effects | ||
| actor.TargetAttack(cast); | ||
| } |
There was a problem hiding this comment.
Prevent cross-attack-point target accumulation (duplicate damage across APs)
cast.Targets is mutated and never cleared, so subsequent attack points may re-hit prior targets unintentionally. Clear the targets after applying damage, or add/remove only the newly added ids within this AP.
Apply this diff (simple/safe clear after damage):
- actor.TargetAttack(cast);
+ // Ensure per-AP targeting to avoid double-hits on later APs
+ actor.TargetAttack(cast);
+ cast.Targets.Clear();If other systems rely on cumulative targets, instead track only newly added ids and remove them post-attack.
📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| if (resolvedTargets.Count > 0) { | |
| int limit = attack.TargetCount > 0 ? attack.TargetCount : 1; | |
| for (int i = 0; i < resolvedTargets.Count && i < limit; i++) { | |
| IActor target = resolvedTargets[i]; | |
| cast.Targets.TryAdd(target.ObjectId, target); | |
| } | |
| // Reuse existing pipeline to calculate and broadcast damage/effects | |
| actor.TargetAttack(cast); | |
| } | |
| if (resolvedTargets.Count > 0) { | |
| int limit = attack.TargetCount > 0 ? attack.TargetCount : 1; | |
| for (int i = 0; i < resolvedTargets.Count && i < limit; i++) { | |
| IActor target = resolvedTargets[i]; | |
| cast.Targets.TryAdd(target.ObjectId, target); | |
| } | |
| // Ensure per-AP targeting to avoid double-hits on later APs | |
| actor.TargetAttack(cast); | |
| cast.Targets.Clear(); | |
| } |
🤖 Prompt for AI Agents
In Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs around
lines 79 to 88, cast.Targets is being appended to each attack point and never
cleared, causing targets to accumulate across attack points and receive
duplicate damage; after calling actor.TargetAttack(cast) clear only the
per-attack targets (e.g., cast.Targets.Clear() or remove the IDs added for this
AP) so subsequent APs start with an empty target set; if cumulative behavior is
required elsewhere, instead track the newly added IDs in a temporary list and
remove them from cast.Targets after actor.TargetAttack(cast).
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (13)
Maple2.Tools/Dotenv.cs (2)
27-29: Handle escapes and inline comments robustly.
Current quote stripping doesn’t unescape sequences and won’t strip inline comments when unquoted (e.g., FOO=bar # note).Apply:
- if ((value.StartsWith("\"") && value.EndsWith("\"")) || (value.StartsWith("'") && value.EndsWith("'"))) { - value = value.Substring(1, value.Length - 2); - } + bool isQuoted = value.Length >= 2 && + ((value[0] == '"' && value[^1] == '"') || (value[0] == '\'' && value[^1] == '\'')); + if (isQuoted) { + value = value[1..^1] + .Replace("\\n", "\n") + .Replace("\\r", "\r") + .Replace("\\t", "\t") + .Replace("\\\"", "\"") + .Replace("\\'", "'"); + } else { + int hash = value.IndexOf('#'); + if (hash >= 0) value = value[..hash].TrimEnd(); + }
31-35: Clarify empty-var override semantics.
You currently override when existing is empty. If “empty means intentionally set by host,” switch to null-only.Option A (do not override when empty):
- string? existing = Environment.GetEnvironmentVariable(key); - if (string.IsNullOrEmpty(existing)) { + string? existing = Environment.GetEnvironmentVariable(key); + if (existing is null) { Environment.SetEnvironmentVariable(key, value); }If the current behavior is intended, add a brief comment stating that empty is treated as “not set”.
Maple2.Server.Game/Program.cs (1)
57-82: Solid exponential backoff with jitter; add a per-RPC deadline and de-dupe delay calc.
Prevent a single RPC from hanging too long and avoid duplicating delay math.Apply:
-while (true) { +while (true) { try { - response = worldClient.AddChannel(new AddChannelRequest { + response = worldClient.AddChannel(new AddChannelRequest { GameIp = Target.GameIp.ToString(), GrpcGameIp = Target.GrpcGameIp, InstancedContent = overrideInstanced || Target.InstancedContent, - }); + }, deadline: DateTime.UtcNow.AddSeconds(2)); if (response != null && response.GamePort != 0 && response.GrpcPort != 0) { break; // Success with valid ports } // Received an invalid allocation (likely due to a race). Retry. - attempt++; - int baseDelayMs = (int)Math.Min(30000, 1000 * Math.Pow(2, Math.Min(attempt, 6))); - int jitterMs = Random.Shared.Next(250, 1000); - int delayMs = baseDelayMs + jitterMs; + int delayMs = ComputeBackoff(++attempt); Log.Warning("World returned invalid ports. Retry {Attempt} in {DelayMs}ms", attempt, delayMs); await Task.Delay(delayMs); - } catch (RpcException) { - attempt++; - int baseDelayMs = (int)Math.Min(30000, 1000 * Math.Pow(2, Math.Min(attempt, 6))); - int jitterMs = Random.Shared.Next(250, 1000); - int delayMs = baseDelayMs + jitterMs; + } catch (RpcException) { + int delayMs = ComputeBackoff(++attempt); Log.Warning("World not ready yet. Retry {Attempt} in {DelayMs}ms", attempt, delayMs); await Task.Delay(delayMs); } }Add (outside this hunk):
static int ComputeBackoff(int attempt) { int baseDelayMs = (int)Math.Min(30_000, 1000 * Math.Pow(2, Math.Min(attempt, 6))); int jitterMs = Random.Shared.Next(250, 1000); return baseDelayMs + jitterMs; }Optional: also honor cancellation (ApplicationStopping) to exit the loop on shutdown.
scripts/stop_servers.ps1 (1)
15-21: Expose a stop timeout parameter.
Allows graceful vs. fast stops, works with v1/v2.Apply:
-param([string[]]$Service) +param([string[]]$Service, [int]$Timeout = 10) @@ - Compose stop @Service + Compose stop -t $Timeout @Service @@ - Compose stop + Compose stop -t $TimeoutMaple2.Server.World/Service/WorldService.Migrate.cs (2)
36-51: Simplify channel selection and avoid double endpoint checks.
You validate active endpoint twice (Lines 38–40 and again at 53–55). Keep a single check after selection.Apply:
- if (request.InstancedContent && channelClients.TryGetInstancedChannelId(out int channel)) { - if (!channelClients.TryGetActiveEndpoint(channel, out _)) { - throw new RpcException(new Status(StatusCode.Unavailable, "No available instanced game channel")); - } - } else if (request.HasChannel && channelClients.TryGetActiveEndpoint(request.Channel, out _)) { + if (request.InstancedContent && channelClients.TryGetInstancedChannelId(out int channel)) { + // channel selected; active check occurs below + } else if (request.HasChannel && channelClients.TryGetActiveEndpoint(request.Channel, out _)) { channel = request.Channel; } else { // Fall back to first available non-instanced channel, then instanced if none channel = channelClients.FirstChannel(); if (channel == -1) { if (!channelClients.TryGetInstancedChannelId(out channel)) { throw new RpcException(new Status(StatusCode.Unavailable, "No available game channels")); } } }Optional: replace
channelClients.Count == 0withchannelClients.FirstChannel() == -1to reflect the actual non‑instanced availability check used later.
53-55: Unify error message for inactive/unknown channel.
Consider specifying instanced/non-instanced in the message for clarity.Example:
- throw new RpcException(new Status(StatusCode.Unavailable, $"Channel {channel} not found")); + throw new RpcException(new Status(StatusCode.Unavailable, $"Channel {channel} not available (inactive or missing)"));Maple2.Server.World/Containers/ChannelClientLookup.cs (3)
123-125: Be robust to hostnames in gameIp (avoid Parse-only path).
IPAddress.Parse(gameIp)will throw for hostnames (e.g., "localhost" or DNS names). Prefer TryParse + DNS fallback.- IPAddress ipAddress = IPAddress.Parse(gameIp); + IPAddress ipAddress = IPAddress.TryParse(gameIp, out var parsed) + ? parsed + : Dns.GetHostAddresses(gameIp).First(a => a.AddressFamily == System.Net.Sockets.AddressFamily.InterNetwork); IPEndPoint gameEndpoint = new IPEndPoint(ipAddress, newGamePort);
143-147: Use Task.Run for async delegate; StartNew + async loses exceptions and the CTS isn’t tracked.
Task.Factory.StartNewwith an async-returning delegate fire-and-forgets a nested task and can drop exceptions. Also, theCancellationTokenSourceis not retained, so you can’t cancel monitors.- var cancel = new CancellationTokenSource(); - Task.Factory.StartNew(() => MonitorChannel(activeChannel, cancel), cancellationToken: cancel.Token); + var cancel = new CancellationTokenSource(); + _ = Task.Run(() => MonitorChannel(activeChannel, cancel), cancel.Token);If you want controlled shutdown, store
cancelon the Channel and cancel on disposal. For example (outside this hunk):// In Channel class public CancellationTokenSource? MonitorCts { get; init; }And set it when creating
activeChannel, then cancel it when removing the channel.
157-159: Attempt cap off-by-one.
attempts++precedes the check, so this allows 101 tries. Use>= 100.- if (attempts > 100) { + if (attempts >= 100) { logger.Error("Failed to allocate a channel after {Attempts} attempts", attempts); return (0, 0, -1); }scripts/start_servers.ps1 (3)
37-46: Treat 'unhealthy' explicitly to aid diagnostics.Right now only 'exited' emits logs. Add handling for 'unhealthy' to surface recent logs.
- } elseif ($status -match '^(running|starting|created)$') { + } elseif ($status -match '^(running|starting|created)$') { # still waiting - } elseif ($status -match 'exited') { + } elseif ($status -eq 'unhealthy') { + Write-Warning "$Service is unhealthy. Showing last logs:" + try { Compose logs --no-color --tail=200 $Service } catch { } + # keep waiting; timeout logic below will handle exit + } elseif ($status -match 'exited') { Write-Warning "$Service exited unexpectedly. Showing last logs:" try { Compose logs --no-color --tail=200 $Service } catch { } if ($Soft) { return $false } else { throw "$Service exited" } }
119-121: Include web in the suggested tail command.Useful for first-run troubleshooting.
-$joined = ($started + @('world','login')) -join ' ' +$joined = ($started + @('world','login','web')) -join ' '
12-20: Optional: support a project name to avoid compose namespace collisions.Add
-p $ProjectNamepass-through when provided.-param( +param( [int[]]$NonInstancedChannels, [switch]$IncludeInstanced, - [switch]$NoBuild + [switch]$NoBuild, + [string]$ProjectName ) @@ function Compose { param([Parameter(ValueFromRemainingArguments=$true)][string[]]$Args) - if ($UseV2) { & docker compose @Args } else { & docker-compose @Args } + if ($UseV2) { + if ($PSBoundParameters.ContainsKey('ProjectName') -and $ProjectName) { & docker compose -p $ProjectName @Args } else { & docker compose @Args } + } else { + if ($PSBoundParameters.ContainsKey('ProjectName') -and $ProjectName) { & docker-compose -p $ProjectName @Args } else { & docker-compose @Args } + } }compose.yml (1)
21-39: MySQL port is published to host; ensure this is intended.Exposing 3306 is fine for local dev but risky otherwise. Consider removing the port mapping or limiting to 127.0.0.1 in non-dev.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (7)
Maple2.Server.Game/Program.cs(1 hunks)Maple2.Server.World/Containers/ChannelClientLookup.cs(1 hunks)Maple2.Server.World/Service/WorldService.Migrate.cs(2 hunks)Maple2.Tools/Dotenv.cs(2 hunks)compose.yml(2 hunks)scripts/start_servers.ps1(1 hunks)scripts/stop_servers.ps1(1 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
Maple2.Server.Game/Program.cs (2)
Maple2.Server.Core/Constants/Target.cs (2)
Target(5-77)Target(29-76)Maple2.Server.World/Service/WorldService.GamePorts.cs (1)
Task(7-15)
🔇 Additional comments (5)
Maple2.Tools/Dotenv.cs (1)
15-17: Good: comment handling now ignores leading whitespace.
This fixes common ".env" patterns like " # comment".scripts/stop_servers.ps1 (1)
9-14: Nice compose v1/v2 shim.
Good portability and strict mode.Maple2.Server.World/Service/WorldService.Migrate.cs (1)
31-35: Reasonable fallback when no non-instanced channels are active.
This avoids hard failures when an instanced channel exists.compose.yml (2)
78-84: Ensure healthcheck uses grpc_health_probe or installs bash
Ripgrep shows nobashinstallation orgrpc_health_probein any Dockerfile; the currentbash -c 'exec 3<>/dev/tcp/…'will fail if bash isn’t present. Either add bash to each image or replace the healthcheck with:healthcheck: test: ["CMD","grpc_health_probe","-addr=127.0.0.1:21001"] interval: 5s timeout: 3s retries: 30 start_period: 5s
99-114: Healthcheck port is correct
Target.GrpcLoginPort is set to 21000 inCore/Constants/Target.csand the Login service binds gRPC on port 21000 internally; no changes needed.
| bool isDocker = Environment.GetEnvironmentVariable("DOTNET_RUNNING_IN_CONTAINER") == "true"; | ||
| if (isDocker) { | ||
| // When running in Docker, use the provided grpcGameIp as hostnames | ||
| grpcUri = new Uri($"http://{grpcGameIp}:{newGrpcChannelPort}"); | ||
| } else { | ||
| // Outside of Docker, parse the IP addresses normally | ||
| IPAddress grpcIpAddress = IPAddress.Parse(grpcGameIp); | ||
| grpcUri = new Uri($"http://{grpcIpAddress}:{newGrpcChannelPort}"); | ||
| } |
There was a problem hiding this comment.
Don't assume grpcGameIp is an IP outside Docker; parsing will throw on hostnames.
Unify URI construction and allow DNS names in all environments.
- bool isDocker = Environment.GetEnvironmentVariable("DOTNET_RUNNING_IN_CONTAINER") == "true";
- if (isDocker) {
- // When running in Docker, use the provided grpcGameIp as hostnames
- grpcUri = new Uri($"http://{grpcGameIp}:{newGrpcChannelPort}");
- } else {
- // Outside of Docker, parse the IP addresses normally
- IPAddress grpcIpAddress = IPAddress.Parse(grpcGameIp);
- grpcUri = new Uri($"http://{grpcIpAddress}:{newGrpcChannelPort}");
- }
+ // Allow IPs or hostnames uniformly
+ grpcUri = new Uri($"http://{grpcGameIp}:{newGrpcChannelPort}");📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| bool isDocker = Environment.GetEnvironmentVariable("DOTNET_RUNNING_IN_CONTAINER") == "true"; | |
| if (isDocker) { | |
| // When running in Docker, use the provided grpcGameIp as hostnames | |
| grpcUri = new Uri($"http://{grpcGameIp}:{newGrpcChannelPort}"); | |
| } else { | |
| // Outside of Docker, parse the IP addresses normally | |
| IPAddress grpcIpAddress = IPAddress.Parse(grpcGameIp); | |
| grpcUri = new Uri($"http://{grpcIpAddress}:{newGrpcChannelPort}"); | |
| } | |
| // Allow IPs or hostnames uniformly | |
| grpcUri = new Uri($"http://{grpcGameIp}:{newGrpcChannelPort}"); |
🤖 Prompt for AI Agents
In Maple2.Server.World/Containers/ChannelClientLookup.cs around lines 128 to
136, the code currently parses grpcGameIp as an IP address outside Docker which
will throw for hostnames; instead unify URI construction so grpcGameIp is used
verbatim as the host in both cases (do not call IPAddress.Parse), ensure
grpcGameIp is non-empty and the port/newGrpcChannelPort is validated, and create
the Uri with string interpolation like
$"http://{grpcGameIp}:{newGrpcChannelPort}" for all environments so DNS names
are supported.
|
@gugarosa can you make a new PR with only the docker stuff? |
Of course, my bad. I ended up doing some additional work on top of the same branch. It is reverted now, only the initial change for damage is on this PR. |
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (1)
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs (1)
86-88: Prevent cross-attack-point target accumulation (duplicate damage across APs).cast.Targets is appended but never cleared; subsequent APs will re-hit prior targets. Clear after applying damage.
- // Reuse existing pipeline to calculate and broadcast damage/effects - actor.TargetAttack(cast); + // Reuse existing pipeline to calculate and broadcast damage/effects + actor.TargetAttack(cast); + // Ensure per-AP targeting to avoid double-hits on later APs + cast.Targets.Clear();
🧹 Nitpick comments (1)
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs (1)
81-84: Optionally guard against self-targeting.If ApplyTarget or upstream lists ever include the attacker, skip it to avoid self-damage.
- IActor target = resolvedTargets[i]; - cast.Targets.TryAdd(target.ObjectId, target); + IActor target = resolvedTargets[i]; + if (target.ObjectId != actor.ObjectId) { + cast.Targets.TryAdd(target.ObjectId, target); + }
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs(2 hunks)
🔇 Additional comments (3)
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.cs (3)
5-5: LGTM: required using for range utilities.The added using appears necessary for Range.GetPrism/ApplyTarget extensions.
73-75: Verify collision angle units. actor.Rotation.Z is in degrees; confirm thatSkillMetadataRange.GetPrism(and underlyingRectangle/Trapezoid) interprets the angle parameter as degrees—not radians—to avoid skewed target prisms.
80-84: Honor TargetCount==0 as unlimited instead of forcing single‐target.- int limit = attack.TargetCount > 0 ? attack.TargetCount : 1; - for (int i = 0; i < resolvedTargets.Count && i < limit; i++) { + int limit = attack.TargetCount > 0 + ? Math.Min(attack.TargetCount, resolvedTargets.Count) + : resolvedTargets.Count; // 0 ⇒ unlimited + for (int i = 0; i < limit; i++) { IActor target = resolvedTargets[i]; cast.Targets.TryAdd(target.ObjectId, target); }Verify that Field.GetTargets fallback correctly treats TargetCount==0 as unlimited.
2 participants
What changed
Maple2.Server.Game/Model/Field/Actor/ActorStateComponent/SkillState.csattackTargetswhen present.attack.Range.GetPrism(actor.Position, actor.Rotation.Z)andField.GetTargets(...)when empty.cast.Targets.actor.TargetAttack(cast)to reuse the established damage flow (DamageCalculator, HP updates via StatsPacket.Update, SkillDamagePacket.Damage, and effect triggers).Why this fixes it
Summary by CodeRabbit
Bug Fixes
Improvements