The Eagle has Landed!

[MMC]

SDK + Patch 1.61 now available...

http://www.2kgames.com/civ4/downloads.htm

[MMC]

Hummm... On second thoughts, the SDK is kind of disappointing. I was expecting some tools of some kind, plus the ability to open the .vcproj file in Visual C++ 2005.

Neither are available. It is only the source code for the game DLLs.

[Solver]

"Only" the source code ? That's what you can change the entire game with...

[MMC]

EDIT: My Bad... I didn't notice that all the files are set to read-only, so the conversion to VC++2005 didn't work... switching that off fixes the problem.

But I still say that extra tools should have been included, such as some form of .nif editor. Also a GUI editor for the .thm interface files would be nice.

[Unspeakable Horror]

There are more files at Firaxis site.

[Snoddasmannen]

Very cool indeed! 8007 lines of AI code to read
CvPlayerAI.cpp

EDIT: Interesting tidbit number one, do you know that the AI will never raze a city during the first 20 turns? Even more interesting, it will never raze a holy city or one with an active wonder in it.

void CvPlayerAI::AI_conquerCity(CvCity* pCity)
{
	CvCity* pNearestCity;
	bool bRaze;

	if (canRaze(pCity))
	{
		if (GC.getGameINLINE().getElapsedGameTurns() > 20)
		{
				if (!(pCity->isHolyCity()) && !(pCity->hasActiveWorldWonder()))
				{

For a programmer with an interest in AI this is a lot of fun

[Solver]

There's a lot more interesting stuff there .

Not razing holy cities is relatively new, though, I think...

[rogerbacon]

Originally posted by Snoddasmannen
Very cool indeed! 8007 lines of AI code to read
CvPlayerAI.cpp

EDIT: Interesting tidbit number one, do you know that the AI will never raze a city during the first 20 turns? Even more interesting, it will never raze a holy city or one with an active wonder in it.

Yeah, I noticed that. Lots of godo reading in there.

One of the first things I wantto do is change the AI routine so that whenever you get a negative relation for some action (You didn't help civ X in their war against civ Y) that you also get a positive relation in the other direction (in this case, a bonus for civ Y for not helping civ X fight against them). I haven't seen that part of the AI code yet, but I haven't looked through all 800 lines.

Roger Bacon

[Snoddasmannen]

I'm more interested in improving the tactical prowess of the AI. It actually does a great job of getting troops to roughly the right place, but then it often seems to get a little off track and really tilt the odds in favour of the human.

I have a background in computer chess programming, so I was hoping to jump right in and fix up a little 1 ply alpha-beta search and make it behave optimally, at least when attacking. Of course that was desperately over-optimistic

One of the bigger problems seems to be that the order in which units move is decided in a different location than where the units actually move. Having them seperate is probably a nice thing to have architecturally, but it may make implementing a highly effecient AI in tight tactical situations more tricky. Basically you seem to be forced to rely on assumptions such as catapults always move first, and a unit will always want to use all its MPs at once, and stuff like that.

Correct me if I'm wrong I'm probably rambling here, it's almost midnight and I already spend most of my days looking at code I think the overall AI design seems to be very nice and flexible, not to mention extremely interesting. The whole missions concept is a cool way to solve things. Many kudos to Firaxis for releasing this, awesome.

[Dale]

Anyone need help in deciphering the combat code? Let me know.

Dale

[Solver]

Don't you have something to post ?

[Dale]

Not for the latest patch, gotta convert it first.

Dale