Spring RTS Engine

Thank you for adjusting auto-completion. Also, the error I was getting seems to be a Lua problem. After disabling widgets, everything works fine again.

New problem. I was testing my air code after the previously mentioned update, but spring kept crashing without any relevant information. So I ran it under gdb, expecting to find something wrong with my air code, but the only output I got from GDB was Starting program: E:\eclipse\workspace\trunk\game...

Wow, how silly of me. Thank you.

edit: Then what need to be done to get auto-complete working for /give? It still works for .give but not for /give.

I just synchronized with SVN, and now cheating is not working. I am assuming that this has something to do with the new character set handling being done. Specifically, I'm launching a modified XTA v9 sdd on Mars. I type .cheat, with no response. I type .give arm_peeper (auto-complete is working for...

The changes I have made to AirMoveType so far should not have changed any functionality. It was just refactoring so I could wrap my head around what it was doing, in preparation to write an entirely new air movement class (which is what I am working on now). If it appears I broke anything, please le...

I agree with Kloot. IMHO, it would be far to difficult and even dangerous to multi-thread the synced code. Threading the unsynced portions, however (as previously mentioned, AIs and display) would be almost pure win.

Two more things that would be cool: 1. Use subfolders for the different music types, and search for the music files (aka, don't hard-code the names) so that people can put whatever music they want in there. 2. Search the mod archive for a music folder and subfolder before defaulting to the spring mu...

I'm currently refactoring some of the CAirMoveType, CTAAirMoveType and CAirCAI to better define the boundry between the Command AI and the Move Type, in preperation for creating a CNewAirMoveType, which I intend to make usable as a mod and/or unit option. I'll probably be committing my current chang...

After much though and some experimentation, I don't think there is anything inherently unbalanced about the network. I need more more games to be sure, but I think my concern currently lies in system/network games, where the flow, vulnerable to both pointers and bytes, is not a highly used unit. Thi...

It does? For some reason I thought that a random seed was sent out to all of the clients before the game started, or does Lua use a different synced random number generator?

One thing you need to remember is that sin/cos on a computer are (nearly) always approximations due to the maximum size of a float. If we can increase speed significantly and keep the error below 1%, I think we should go for it.

CMobility does not seem to be used for anything, but it's created for every unit. Can anyone think of any reason I shouldn't just delete and remove references to it?

Sounds great. I know we could use it in some of the float3 methods at least.

(beaten by Zpock) IANAL. Trademark != Copyright Trademarks can be lost by inaction and failure to defend them. However, even if Atari has/does lose the Total Annihilation trademark, and someone else does pick it up, they could create a game Total Annihilation 2, but could not sell the original TA. I...

It reminds me of a ghost-busters trap for some reason. Not saying that is good or bad.

The latest bug comment

In CNetProtocol::InitClient and CNetProtocol::InitLocalClient,
CNet::InitClient is called with wantedNumber, and stores what the number ended up being in myNum. It then calls SendAttemptConnect with wantedNumber. Should that be myNum instead?

I have a thought that might fix bug 759, but I have no idea how to test it. Any ideas? Also could I get a sanity check from someone more familiar with that section of the code (CNet & CNetProtocol)?

I had sync problems once when I had an .sdd and an .sd7 of the same mod in my spring folder. Also, I believe .sdd to .sdd syncing does not work.

While I agree that keeping current functionality (crashing etc.) on by default is a good thing, anyone who doesn't think the the air move type code needs a major rewrite is either foolish or hasn't seen the code. It uses a goto for peet's sake.

Deciding whether/which units to attack, and therefore when the aircraft goes into an attacking state is the job of the CommandAI, not the move type.

Edit: The attacking function should run while in full flight and owner->userTarget.