GameMessage::getCommandTypeAsAsciiString looks inefficient

[xezon]

1. Allocates a new string on every call.
2. Uses if condition instead of switch case. Unclear if the compiler will make it nice.

AsciiString GameMessage::getCommandTypeAsAsciiString(GameMessage::Type t)
{
#define CHECK_IF(x) if (t == x) { return #x; }
	AsciiString commandName = "UnknownMessage";
	if (t >= GameMessage::MSG_COUNT)
	{
		commandName = "Invalid command";
	}
	CHECK_IF(MSG_INVALID)
	CHECK_IF(MSG_FRAME_TICK)
	CHECK_IF(MSG_RAW_MOUSE_BEGIN)
	CHECK_IF(MSG_RAW_MOUSE_POSITION)
	CHECK_IF(MSG_RAW_MOUSE_LEFT_BUTTON_DOWN)
	CHECK_IF(MSG_RAW_MOUSE_LEFT_DOUBLE_CLICK)
	CHECK_IF(MSG_RAW_MOUSE_LEFT_BUTTON_UP)
	CHECK_IF(MSG_RAW_MOUSE_LEFT_CLICK)
	CHECK_IF(MSG_RAW_MOUSE_LEFT_DRAG)
...
// hundreds more

[roossienb]

I investigated this issue using the following code:

	auto start = std::chrono::high_resolution_clock::now();
	for (int i = 0; i < 1000000; i++)
	{
		// Third message
		// auto ret = GameMessage::getCommandTypeAsAsciiString(GameMessage::Type::MSG_RAW_MOUSE_BEGIN);

		// Last message
		auto ret = GameMessage::getCommandTypeAsAsciiString(GameMessage::Type::MSG_ENABLE_RETALIATION_MODE);
	}

	auto end = std::chrono::high_resolution_clock::now();
	std::chrono::duration<double> duration = end - start;
	std::cout << "Time taken for 1000 iterations: " << duration.count() << " seconds" << std::endl;

I called the method 1 million times for both the third and last messages in the list, expecting a performance difference due to the varying number of if-statements each must traverse—the third message requiring only three checks, whereas the last message processes all of them.

To improve efficiency, I refactored getCommandTypeAsAsciiString() into a switch statement. The recorded execution times are averaged over multiple runs.

MESSAGE                         CHECK_IF()          Switch
----------------------------------------------------------
MSG_RAW_MOUSE_BEGIN                0.331 s         0.168 s
MSG_ENABLE_RETALIATION_MODE        0.454 s         0.167 s

I'll make a PR accordingly

[xezon]

This is not meant to be changed yet. We focus on stability things.

[Caball009]

I called the method 1 million times for both the third and last messages in the list, expecting a performance difference due to the varying number of if-statements each must traverse—the third message requiring only three checks, whereas the last message processes all of them.

Always using the same enum value makes it way too easy for the branch predictor, so the results don't provide much insight imo.

[Caball009]

This is not meant to be changed yet. We focus on stability things.

I happened on this function for the umpteenth time and decided that it was time to do something about it.