Spring RTS Engine

Speaking of buildtrees, AF, can you move the buildtree info in AI to either an NTAI map or better in aidll/globalai/ntai? The data in AI is not general AI data, it's data specific to your NTAI.

No you need to use a different name, look at the tag in the mod_info.tdf inside the AA mod.

And jelmer, tobi has no objections and even created the folder in the svn for me.

Also OTAI 1.14 loads from the same folder so it is not NTai specific. Also OTAI mex caches and DTBuild caches are identical to NTai caches.

And finally it can be changed using AI.tdf, there is a tag datapath that defaults to the AI folder that tells NTai where to load and save.

As for AA buildtree, there is someone doing stuff with an AA buildtree but its not for sure atm.

It should default to aidll/GlobalAI/NTAI/AI. Unexperienced users will assume that AI is the general path for all AIs (and would you like it if another AI used the same folder and overwrote your files?).

Data written to that folder is versioned, if another AI wrote over the top of my data with a different format then it would break their AI's when they loaded NTai caches that where of a different format.

Instead they'd have to sue a new version, for example having a different version tag in the learning files, or using mv5 instead of mv4 for mex cache file extensions.

It should default to aidll/GlobalAI/NTAI/AI. Unexperienced users will assume that AI is the general path for all AIs (and would you like it if another AI used the same folder and overwrote your files?).

Exactly, for this reason it should be /NTAI IMO. The fact is that it is not shared data, except for OTAI. Even though you would like to see data like this unified, you have to accept that multiple AIs will use multiple methods. If you insist on an AI path instead of aidll/globalai, let's make it ai/ntai. OTAI can load its metal cache files from there too.

Instead they'd have to sue a new version, for example having a different version tag in the learning files, or using mv5 instead of mv4 for mex cache file extensions.

Again ai/ntai would be the better solution, since you then keep complete freedom as an AI developer to put your data, and are not forced to support a specific format.

I still would like to know what exactly is wrong with aidll/globalai/ntai ...

Make a new thread about it, I've heard this allbefore and unless you can give em an easy to use install maker that can rename and move everything from AI -> /aidll/globalai/data then go away, i have enough to get on with as it is, and right now your just being annoying.

How come people start new threads for other AI's and my thread gets bloated with masses of banter.

Normally I would appreciate what your saying but right now I dont want to say what I think because I know there are young people who surf these forums.

It's my 18th birthday today and rather than feeling like celbrating I'm tired I have a headache and I'm very annoyed.

congrats !

Make a new thread about it, I've heard this allbefore and unless you can give em an easy to use install maker that can rename and move everything from AI -> /aidll/globalai/data then go away, i have enough to get on with as it is, and right now your just being annoying.

Ok here's your thread. It has indeed been said before but you haven't given any good arguments for moving from the standard location aidll/globalai.
Furthermore I'm just trying to discuss AI data paths here, trying to compromise. It would be nice if you did the same.
Also I'm making the installer, so in the end it's my call. That means in the end we need to agree with eachother somehow.

Normally I would appreciate what your saying but right now I dont want to say what I think because I know there are young people who surf these forums.

"Normally I would flame you but I won't", please what kind of behavior is this?

Normally I would flame you? huh?

Normally I would appreciate what your saying

Now I know your the one making the spring Installer but I'm not talking about the spring installer, I'm talking about the NTai installer. I dont want people complaining they have a deadweight AI folder sitting there for nor eason once I move the data folder, or to have to explain to people to delete the folder, its enough that I have to make them upgrade from the ancient XE7.5 or face them dissing NTai XE9 in #main because they think their XE7.5 is the latest version.

Well, the way it is I had a deadweight aidll/globalai/NTAI folder that I had to manually delete. When did that change to /AI?

Congratulations!
And about the AI dir structure, why use "aidll" as an AI directory?
dll's are a windows thing and buildtree data are plain text files.
If things get moved around I would move it all inside the AI dir and be done with it.

Then maybe an internal re-arrangement of the AI folder is in order, perhaps a dll folder, an NTai fodler than a cache folder for AI nonspecific data

Whats there to share? a couple 1kb metal files that take milisseconds to generate?
I dont really care but it would be pretty nice if the AI stuff stayed in ONE AI folder, maybe aidll/globalai/shareddata?

Then maybe an internal re-arrangement of the AI folder is in order, perhaps a dll folder, an NTai fodler than a cache folder for AI nonspecific data

That I can agree with, the point was that NTAI was holding its data in a location that should be usable by all AIs (spring/AI/* instead of spring/NTAI/*)

Maybe a structure like:

/ai
/ai/win/ all windows dlls
/ai/linux/ linux so
/ai/mac/ mac so
/ai/data/
/ai/data/ntai
/ai/data/kai

If otai still wants to use ntai metal data, it could read from /ai/data/ntai, but that would be a discussion between af and veylon...

No because that would mean either OTAI is needed as a prerequisite for NTai or NTai is needed for OTAI to work, neither of which is accepable.

And mex data isnt the only thing generated by both NTai and OTAI thats sharable.

otherwise I agree with jelmer, but an AI/data/generic/ would be more useful.

It would make sense to rearrange some dirs as AI development is something new in svn. We still need to find a way to fit AI development in with the rest.

And about the shared data, you would need multiple AI's that use the same data format before it would be useful.
Managing a data format for beta software like Spring would be quite hard, but if we/you get other AI developers that like to do this I would be fine with it.
But I would first get that commitment from other AI developers before you spend to much energy on it.

No energy needed, nearly all the AI's in existence for sprign sue krogothes metal class, and OTAI and NTai use near identical classes with identical data output for several thigns including mex data.

It's a matter of working togetehr on the classes and loading from the same directory.

AF wrote:No because that would mean either OTAI is needed as a prerequisite for NTai or NTai is needed for OTAI to work, neither of which is accepable.

...

I don't understand?

Ow wait, OTAI is your "old" AI right?

Edit: No it's not..?
Anyway, I'm done, I'm going to call it a day over here.
I need a beer and a good seat.

no otai was made by veylon IIRC