Particle systems cause game logic / CRC changes

[stephanmeesters]

Ideally particle systems are purely visual and disabling them should not affect game logic. However currently this does not appear to be the case.

Example:

ParticleSystem *ParticleSystemManager::createParticleSystem( const ParticleSystemTemplate *sysTemplate, Bool createSlaves )
{
	return nullptr; // <-- disable particles

	// sanity
	if (sysTemplate == nullptr)
		return nullptr;

	m_uniqueSystemID = (ParticleSystemID)((UnsignedInt)m_uniqueSystemID + 1);
	ParticleSystem *sys = newInstance(ParticleSystem)( sysTemplate, m_uniqueSystemID, createSlaves );
	return sys;
}

This causes mismatching in Golden Replay #1.

I could trace this back to TransitionDamageFX, by allowing particles to be created there but nowhere else, but this moved the mismatch just further along to another frame, so there are multiple causes.

Issue also happens in a win32 replay.

[Mauller]

There are particle systems which are physical entities within the game for certain types of weapon.

These tend to be field or spray type effects.

So there are Logical particles in the world that need to be maintained.

[stephanmeesters]

There are particle systems which are physical entities within the game for certain types of weapon.

These tend to be field or spray type effects.

So there are Logical particles in the world that need to be maintained.

That would explain it, but it still strikes me as odd as particle systems only use GameClientRandom and not GameLogicRandom suggesting it should never find its way into game logic, but I guess I should read more of the code.

[xezon]

There are particle systems which are physical entities within the game for certain types of weapon.

These tend to be field or spray type effects.

So there are Logical particles in the world that need to be maintained.

These are projectiles or missiles. Not particles.

[xezon]

It would be good to find what causes particle mismatch, because i was about to completely disable particles for headless.

[xezon]

I have now deactivated headless particles with #2709 and the replay checks passed.

[stephanmeesters]

I have now deactivated headless particles with #2709 and the replay checks passed.

But they are created, just no longer updated, is that correct? In the example above it simulates the situation that they don't exist in the data.

[xezon]

They are no longer created in #2709

[stephanmeesters]

They are no longer created in #2709

The win32 replay doesn't pass on this branch on my machine (did full rebuilds in between). It appears the CI replay test is broken

[xezon]

I suggest remove one particle after another and check which one(s) cause mismatch. I suspect some game logic is tied to the existence of a particle.

[Caball009]

Which frame does GR1 mismatch for you?

I did notice a crash in W3DTruckDraw::createWheelEmitters, because it doesn't check the ParticleSystem pointer.