Spring RTS Engine

Yes, it learns what units/defences are effective vs classes of targets - gorund sea air hover etc.

Kewl, now all I need is the compatible version.

submarine can you give us an ETA when you'll have a compatible version up?
This waiting is driving me insane.

i mailed a version to the testers yesterday...

however, both university and soccer slow down development heavily :=

more hot updates please :)

The testers (I am at least) are reporting bugs/improvements to Submarine as we find them, it's just a case of him having enough time to implement them... watch this space!

well update:

i extended the unit learning a bit, aai now learns about the usefulness of different categories (air, ground, sea, hover) not only based on a specific map but also based on the mod itself

the total usefulness will be a mixture of both values. i hope aai now doesnt have to learn that hovercraft in aa are not useful on land maps on each land map again....

AAI works nicely in my experience, 'cept for a bug that I e-mailed Submarine about a couple days ago... And haven't heard back or gotten an acknowledgement about.

@fizz: sorry i read your mails but i havent found the time to respond...

but you got a new version some minutes ago :)


especially the "sufficient ressoruces but idle factories" obeservation is really anoying, i'm thinking about a good way to make sure aai does not waste any metal

*dances like a child on christmas morning*

update: some minor things fixed, some primitive reclaiming added but development now stalled until i can use the debugger again...

argh it seems as if the interface has been broken once again -> "my" aai version wont work with curent spring release

when you get it working again i've got a good suggestion/addition:
"ai has turned off cheating "

the problem is:

current spring release -> doesnt support debugging

current svn version -> perhaps it may be compiled to support debugging but is not compatible with the 0.72 spring release

The current svn version can run 0.72 aidll's and vice versa, onyl you have the otpion to use extra events using the abstract interface additions cain and zaphod added, such as WeaponFired events.

You also get a stability boost as they recently fixed an issue with not having virtual destructors

so an ai dll without this destructor will not work anymore...

AF wrote:The current svn version can run 0.72 aidll's and vice versa, onyl you have the otpion to use extra events using the abstract interface additions cain and zaphod added, such as WeaponFired events.

You also get a stability boost as they recently fixed an issue with not having virtual destructors

no it doesnt,

tobi said "Anyway, for now it means that AIs compiled against current SVN don't work with
either current release or spring compiled from current SVN. As a workaround
(for AI devs) go some revisions back and compile your AI against that
checkout."

besides those virtual destructors completly broke compilation for me, it's not possible to even compile the ai atm...

Maybe you should setup your own destructors as virtual, after all it's good practice, and if you derive from those classes it can mean bugs.

Destructors should always have the virtual keyword.

of course i set up my own destructors, thats not the problem, have a look at the dev mailing list, tobi described the problem more detailed

submarine wrote:besides those virtual destructors completly broke compilation for me, it's not possible to even compile the ai atm...

Did you try removing this IGlobalAI::~IGlobalAI() {} thingie in AAI.h?

Because it did compile here when I did that, I only couldn't get spring to load it, no matter how hard I tried.

yep i did, but then the ^linker error appears again, i'll try to "downgrade" to an older svn version until you fix the problem with visibility