More specific names and types, use for_each in CComponentDataHolder::ResetState.

Changed the data type of SComponentDataGenerator::m_LastIndex from bool to uint_32,

Previous upload was a mistake, this one is better hopefully.

Allocator is now able to expand to arbitrary capacity.

Formated some comments.

reworked CGame::IsGameFinished, mostly changed it back to it was before. Seems like a dependency injection issue.

Simplified CGame::IsGameFinished

exclude graph2.*

included tests.

Added a test file.
Standardized instance of SComponentDataGenerator names to 'gen'.
Renamed some instances of the ordered list given by GetComponentsByInterface from ents to comps.
Changed ComponentType.dealloc to ComponentType.destruct.
Changed type name of ComponentType.dealloc from DeallocFunc to DestructFunc.

Renamed GetEntitiesWithInterface to GetComponentsByInterface and GetEntitiesWithInterfaceUnordered to GetComponentsByInterfaceUnordered.

Fixed broken upload from yesterday. Sorry.

std::byte* instead of void*. Explicit keyword on SComponentDataGenerator constructor.

I haven't tested multiplayer or rejoin, Ill have to dig up my old laptop.

On Combat Demo Huge I see a 4% improvement vs. main. This is kinda a worst case scenario for this code since we have a single peak that trails off rather then the multiple peaks and valleys pattern it was written for.

Added spaces after where and for.

Ah, ok! Ill implement the average and total code inside prolfile2 instead then.

Profile2 as i understand it is for seeing the performance of a running game in real time. This is an alternative to graph.html and graph.js as used in simulation replay mode.

Moved CComponentDataHolder and SComponentDataGenerator from heap to stack. Added alignment.

Indentation, voidptr

formatting whitespace, change some let to const.

Default arguments for EjectOrKill, removed VIM from .gitignore

generated with git diff -U9999 instead of git diff --no-prefix -U9999

I don't know what Vulcan is trying to tell me here.

Diff has context now.

renamed attacker and attackerOwner parameters to changeSource and changeSourceOwner in Health.RegisterHealthChanged and throughout GarisonHolder.

Potential optimization to the event listener?

git diff --no-prefix -U9999

@wraitii Ok, thanks for the help :)

style conformity changes, no semantic change.

Remove DamageTaken, add attacker and attackerOwner to HealthChanged

HealthChanged is fired in RegisterHealthChanged
RegisterHealthChanged is called in Reduce
Reduce is called in TakeDamage
TakeDamage has attacker

@wraitii Another issue with storing which entity they died trying to get out of is that the entity might not be destroyed at that moment.

@wraitii Storing in the stats tracker and integrating at the end sounds ok. It won't assign loot.

oh wait, the last line of
void CComponentManager::PostMessage(entity_id_t ent, const CMessage& msg)
is
SendGlobalMessage(ent, msg);

Looking at source/simulation2/system/ComponentManager.cpp line 974 it seems to be saying that when you provide an entity to PostMessage only the components of that entity will be notified.

@Stan would something like

@Stan Is that true when the event is targeting a single entity as well?

I put some time into thinking about how to do this without a message but didn't come up with anything better. JS event emitters are generally kinda cheap I thought? Just a hash table lookup with a JIT symbol and a run through an array of callbacks. Should cost almost nothing if there are no listeners. O(1)?

included previously unincluded files

fixing which files are included hopefully