bugfix(particlesys): Delay particle system destruction to prevent orphaning slaved finite-lifetime particle systems

[Mauller]

Alternative to: #3071

Closes: #3071
Closed: #2645

This PR alters the lifetime handling of linked particle systems. Preventing a master particle system destroying itself before any slaved particle systems have been destroyed.

The result of a slave system losing its master are the following:

• Positional data is lost as this comes from the master system, so slave system particles render at the origin.
• The slave systems m_isDestroyed variable gets set which cuts the systems lifetime short.

The above occurs due to the random nature of particle lifetimes. And when a master particle system destroys itself,
the destructor clears the master of the slave system and calls destroy() which sets the slave systems m_isDestroyed.

With this fix a master particle system checks if it has a finite-lifetime slaved particle system and will not destroy itself till the slaved particle system has been destroyed.
This allows the slaved particle system to fully run the course of its lifetime and render properly.

EDIT - This only applies to finite lifetime particle systems now, it has been tweaked to destroy infinite lifetime systems like it did prior to this change.

[qodo-free-for-open-source-projects]

PR Summary by Qodo

Prevent master particle systems outliving linked slave cleanup

🐞 Bug fix 🕐 10-20 Minutes

AI Description

• Keeps destroyed master systems alive until linked slave systems finish destruction.
• Prevents orphaned slave particles from rendering at the world origin.
• Routes lifetime expiry through normal deferred destruction handling.

Diagram

stateDiagram-v2
  direction TD
  [*] --> Active
  Active --> DestroyPending: expiry or destroy
  DestroyPending --> Draining: particles remain
  Draining --> SlaveCheck: particles cleared
  DestroyPending --> SlaveCheck: no particles
  SlaveCheck --> Waiting: slave linked
  Waiting --> SlaveCheck: slave clears link
  SlaveCheck --> Removed: no slave
  Removed --> [*]

Loading

High-Level Assessment

The targeted lifecycle guard is the appropriate approach because it preserves the existing bidirectional master-slave ownership contract and deferred particle cleanup. A broader ownership redesign would add disproportionate risk for this isolated lifetime bug.

Files changed (1) +10 / -3

Bug fix (1) +10 / -3

ParticleSys.cppDefer master removal until slave destruction completes +10/-3

Defer master removal until slave destruction completes

• The update lifecycle now keeps a destroyed master alive while a valid slave remains linked, allowing the slave destructor to clear the relationship first. Natural lifetime expiry marks the system destroyed instead of immediately returning false, routing cleanup through the same deferred path.

Core/GameEngine/Source/GameClient/System/ParticleSys.cpp

[qodo-free-for-open-source-projects]

Code Review by Qodo

🐞 Bugs (1) 📘 Rule violations (0) 📎 Requirement gaps (0) 📜 Skill insights (0)

1. Forever slave retains master 🐞 Bug ☼ Reliability

Description

Natural lifetime expiry now sets only the master's m_isDestroyed, even though destroy() is the
path that propagates destruction to its slave. A linked slave with the supported default
SystemLifetime of zero is forever-lived, so it never clears the link and the master's new wait
guard returns true indefinitely, leaking and updating both systems for the rest of the session.

Code

Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[2152]

+			m_isDestroyed = true;

Evidence

The template default is a zero system lifetime, and construction explicitly interprets zero as
forever. Slave systems are instantiated from independently configured templates, while only
destroy() propagates destruction to the slave; the changed natural-expiry branch bypasses that
propagation, and the added wait then retains the master as long as the forever slave remains linked.

Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[752-765]
Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[1139-1144]
Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[1195-1210]
Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[1320-1328]
Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[2128-2153]
Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[2666-2692]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
Natural lifetime expiry marks only the master particle system as destroyed. If its slave has the supported default zero system lifetime, the slave never destroys itself, so the master waits on the slave forever and both systems remain registered and updated.

## Issue Context
`ParticleSystem::destroy()` already marks the current system destroyed and propagates destruction to its slave. A zero `SystemLifetime` is initialized by default and is interpreted as `m_isForever = true`, while the newly added master guard refuses deletion until the slave clears the relationship.

## Fix Focus Areas
- Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[1320-1328]
- Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[2139-2153]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

Context sources

✅ Tickets: 🎫 Explosions in map corner around coordinates 0, 0, 0

Tip of the day

💡 Did you know, you can enable the Remediation agent and Qodo fixes findings in a dedicated fix PR

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

[qodo-free-for-open-source-projects]

1. Forever slave retains master 🐞 Bug ☼ Reliability

Natural lifetime expiry now sets only the master's m_isDestroyed, even though destroy() is the
path that propagates destruction to its slave. A linked slave with the supported default
SystemLifetime of zero is forever-lived, so it never clears the link and the master's new wait
guard returns true indefinitely, leaking and updating both systems for the rest of the session.

Agent Prompt

## Issue description
Natural lifetime expiry marks only the master particle system as destroyed. If its slave has the supported default zero system lifetime, the slave never destroys itself, so the master waits on the slave forever and both systems remain registered and updated.

## Issue Context
`ParticleSystem::destroy()` already marks the current system destroyed and propagates destruction to its slave. A zero `SystemLifetime` is initialized by default and is interpreted as `m_isForever = true`, while the newly added master guard refuses deletion until the slave clears the relationship.

## Fix Focus Areas
- Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[1320-1328]
- Core/GameEngine/Source/GameClient/System/ParticleSys.cpp[2139-2153]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

[Mauller]

fixed by checking for forever-lived particles and acting like before the change.

[greptile-apps]

Greptile Summary

The PR changes linked particle-system lifetime handling so a destroyed or naturally expired master remains alive while a finite-lifetime slave finishes.

• Retains masters that still own finite-lifetime slaves.
• Preserves immediate master removal for forever-lived slaves.
• Adds comments documenting the intended lifecycle behavior.

Confidence Score: 4/5

The PR does not appear safe to merge because a forever-lived slave remains registered after its finite master expires.

The prior correction claimed that the master destructor calls destroy() on its slave, but the current destructor only clears their links; the forever-lived slave therefore survives the master and has no natural expiry path.

Files Needing Attention: Core/GameEngine/Source/GameClient/System/ParticleSys.cpp

Important Files Changed

Filename	Overview
Core/GameEngine/Source/GameClient/System/ParticleSys.cpp	Adds finite-slave retention, but the forever-slave exemption still deletes the master while merely detaching and retaining the slave.

_{Reviews (5): Last reviewed commit: "fix(particlesys): Prevent master particl..." | Re-trigger Greptile}

[Mauller]

Fixed the initial lifetime issue of forever particle systems brought up by the bots.

[xezon]

This does not look bullet proof yet.

[Mauller]

Tweaked from recent feedback, should be good now.