Ch 10. MultiplayerGameState's Sorting After Receiving SpawnEnemy

[kyuhyunp]

Since the program uses TCP, the packets are guaranteed to be received in order.

In GameServer's tick, the value of mBattleFieldRect.top decreases for every function call. Thus, the value of mWorldHeight - mBattleFieldRect.top + 500 increases for every function call.

if (now() >= mTimeForNextSpawn + mLastSpawnTime)
	{	
		// No more enemies are spawned near the end
		if (mBattleFieldRect.top > 600.f)
		{
			std::size_t enemyCount = 1u + randomInt(2);
			float spawnCenter = static_cast<float>(randomInt(500) - 250);

			// In case only one enemy is being spawned, it appears directly at the spawnCenter
			float planeDistance = 0.f;
			float nextSpawnPosition = spawnCenter;
			
			// In case there are two enemies being spawned together, each is spawned at each side of the spawnCenter, with a minimum distance
			if (enemyCount == 2)
			{
				planeDistance = static_cast<float>(150 + randomInt(250));
				nextSpawnPosition = spawnCenter - planeDistance / 2.f;
			}

			// Send the spawn orders to all clients
			for (std::size_t i = 0; i < enemyCount; ++i)
			{
				sf::Packet packet;
				packet << static_cast<sf::Int32>(Server::SpawnEnemy);
				packet << static_cast<sf::Int32>(1 + randomInt(Aircraft::TypeCount-1));
				packet << mWorldHeight - mBattleFieldRect.top + 500;
				packet << nextSpawnPosition;

				nextSpawnPosition += planeDistance / 2.f;

				sendToAll(packet);
			}

			mLastSpawnTime = now();
			mTimeForNextSpawn = sf::milliseconds(2000 + randomInt(6000));
		}
	}

Therefore, the height is already sorted in increasing order and sorting the enemy after every SpawnEnemy packet is wasteful. Using a deque to sort the height for networked state would be more efficient.

[Bromeon]

Hello! Thanks a lot for the report, and for the interest in the book! This marks your 10th issue 💪

Some thoughts on this matter:

1. As of today, the book is over 12 years old and can't be updated anymore. New techniques have emerged in the meantime, different people would implement certain things differently, but there were definitely also objective mistakes.

2. The book's primary focus lies on teaching SFML. Networking is such a huge topic that it's impossible to cover all best practices and optimizations in just 1 out of 10 chapters, while also teaching SFML usage at the same time. There are entire libraries dedicated to networking that do a better job at conveying subtle details and involved trade-offs.

3. We do not want to make fundamental changes to the codebase at this point, as it will no longer match the text in the book, and we cannot expect readers to go through GitHub issues to piece together what changed. There is also some history linked to it by now (e.g. links from other websites to concrete lines in this repo, etc.)

As such, the issues you opened represent a useful collection of things that could be improved. You've spent a lot of time on detailed reports. However, I'm not sure if we can still reach the intended audience. From the issues, it looks like you have a lot of knowledge in the area of networking and synchronization. Have you considered writing about some of the things you've learned along the way, e.g. in a blog? I feel that a prose format of optimizations and best practices could deliver a coherent experience and potentially interest readers on its own. But it's just an idea, there are many options 🙂

[kyuhyunp]

I'm sorry if I bothered you or anyone else by raising these issues. I also don't expect any changes to the existing codebase, but I thought it would be helpful for myself and other serious learners. If you don't want me to post any more issues, please let me know.

I don't have much knowledge in networking and synchronization, and I am still learning about the topic. Can you let me know which libraries are dedicated to networking?

Happy Thanksgiving!

[eXpl0it3r]

No bother at all. We just want to make sure you're not having false expectations, i.e. that we would fix/change the code base in anyway.

I too think, your issues might help others who run into the same problems, so I'm happy to keep them open and you're free to report more things that you notice. 🙂