P.U.R.E. 0.8 RC3

[Google_Frog]

I have completed a StarCraft / DoW build system for Resistance. It's very simple and works well, and uses standard commands. You buy a unit / building now, you pay for it now, you get it after X seconds have passed, depending on the object. Still working on a few details, but it's all basically functional, both for factories (queues, patrols, etc. all work now) and mobile stuff.

Does wait, repeat and queue inserting work?

[Argh]

Yes.

[Google_Frog]

Great. After playing THIS I realised how much wait is needed, especially for a pay when you start production system because it's impossible to save up for expensive things otherwise.

[AF]

How exactly does it work from an implementation point of view? This could mess with a system that all AIs have assumed from the very beginning and designed from, and its important that any changes are known. Unit creation and construction logic is very important.

[Pxtl]

So is it that you pay when you queue the unit, or when construction starts? I've always disliked that feature in games - I'm fine with paying at construction-start, just dislike paying at queue-time.

[Argh]

@AF: I'm not sure what you mean. This stuff is for Resistance only- and nobody's written an AI that can play it as it is in RC3, where the separation from OTA is still pretty thin, at a basic game-design level, let alone a proper AI that can make use of the specific features of the Units.

The mechanics of Hacking, for example, merely require that an AI utilize Capture commands, yet nobody's gotten anything working yet, and so the only way to get an AI to "play" Resistance, however badly, is to give the AI a ready-made base with plenty of resources.

If you're interested in that prospect, I'll be happy to discuss the mechanical details when I've settled on a final version of this, but it's not entirely satisfactory yet.

[Argh]

I have had a brilliant breakthrough, if I may be permitted to pat myself on the back. I guess insomnia is good for something, after all.

Not only do I now have a fully-functional system that can handle repeats, Wait commands, and multiple items in a queue with impunity... not only is it incredibly short code (note- I started with a Lua-monster, reinventing many wheels, but I performed a little Occam action in my head, and while I'm not going to reveal how it works right this second, lemme just say that it requires only 16 lines of Lua)...

Not only that... but the system can be very easily extended to support any number of resources, completely perfectly and without the major grunge associated with prior attempts to do this sort of thing.

So, let's see... maybe Resistance needs a third resource, eh? Shall I do StarCraft / DoW, and implement a unit-balance system, where some things eat X slot, and others eat Y? I dunno whether I want to completely rebalance everything ATM, mind ye, but... it's quite possible now.

[AF]

Argh, it is not as simple as that. Ai development needs to maximize flexibility, and thus if you can think of an alternative way of implementing something then we as AI developers may have to account for that when another content developer comes upon the same system and finds it fits their requirements.

Currently you cannot expect us to immediatley focus all our attention on support for resistance without contributing AI development time and resources yourself however AI developers can make efforts to make their AIs more flexible and resistance support may simple 'emerge' as was the case with AIs such as NTai 0.3 or KAI etc with regards to non XTA/BA mods.

You here are meddling with a system that was one of the fundamental assumptions used to design all the native AIs, and meddling with it could cause undefined behavior. It is important that unknown behavior is identified and accounted for through further refinement, so it is important we know what your change entails if we are ever to even consider supporting resistance in P.U.R.E.

[Argh]

I see what you mean... but besides the fact that this change is Resistance only, since I am keeping Overmind as close to stock OTA as I think is reasonable... I think that instead of looking at this as a difficulty, perhaps a paradigm shift is in order.

What needs to happen, imo at least, is for somebody to sit down with Chicken and flesh it out a wee bit more, to build a general structure that can be used for AI development.

Chicken already has a lot of the core functionality an AI needs- how to give orders, etc., it "merely" needs a way to handle economics that is more flexible than "go get hardcoded stuff from Spring" and a basic set of tactical rules governing unit behavior, imo.- beyond that point, it's all just refinement, and you guys have already done most of the Really Hard Stuff already, it'd just be a matter of porting some code.

I'll bet that if you tried it out, once you got past cursing at Lua you'd find it was ridiculously easy, frankly. And if it existed in Lua, well then... extending it, and connecting it with gameplay, becomes a reasonable task. Not trivial, but at least not "omg, why should I even try" territory.

That's pretty much all I have to say about this at this time.

I've been a wee bit frustrated about this issue. I kept my door open, invited all of the AI developers here to participate in the early stages of P.U.R.E.'s development... other than Kloot, nothing happened, and KAIK is the only AI that can play even Overmind worth a goddamn.

Something needs to change, and I really wish you AI people would get your heads out of your arses, quit futzing around with exotic languages and DLLs and reinvention of the wheel, start using Lua, and do things in a way that made sense. I feel that the entire purpose of the AI interface was made completely irrelevant with the arrival of Lua, in terms of practical stuff (fancy uni projects notwithstanding) and it's not my fault you guys haven't noticed.

[Hoi]

Please keep the resource stuff as in ota (pay while building) It's really more strategic than pay and then make the unit, changing it would fuck stuff up imo.

[Argh]

It's really more strategic than pay and then make the unit, changing it would fuck stuff up imo.

I dunno, tbh there's not a lot of difference in practical terms, except that the way that you manage the economy as a player is a little different. You don't pay the cost until the unit starts building, after all. It's not like you queue up 20 tanks, and have to pay for all of them immediately. It's like Dawn of War, where you pay when the unit reaches the head of the queue.

So there's not much difference either way, during the crucial part of early game- most decent players are building a set of units for rush or defense.

Most commercial games use that system, and I'm kinda excited about implementing it for one side, so that people who are more used to that model can play that way, now that I have it working correctly. Plus it really opens up some doors in terms of game design possibilities down the road.

[Hoi]

I'm warning you, it'll give balance issues, but on the bright side, it'll help out noobs not allowing them to build stuff there economy won't support, if you are going to do it like this, please make it a mod option, and don't do it for energy, that'd make no sense.

[Argh]

it'll give balance issues

Well, yes- any change to a fundamental area of the game design invariably does. That said, it seems like a good time to do it- from here on out, other than balance fixes, I want to concentrate on missions, custom maps, and shader stuff, so this is pretty much the last time I have to concentrate on the core game design, other than throwing in the occasional new unit.

don't do it for energy, that'd make no sense

Er, why not? It's a resource, just like Materials... I suppose I could make things just cost Materials, but that would truly screw up the game balance, such as it is. If I'm going to screw it up that much, I might as well go ahead and introduce a third resource, frankly, and just get all of the pain over with in one go.

[Hoi]

What I mean is that energy should be used while something is being build, or atleast I would like it to, this is one of the things that made ta win, I know this is not ta, but in general I like it better the ta way than the red alert way.

About the balance, if you don't spend on building start you can build something you don't have the resources for yet, so you will complete it faster than the other faction which can only start to build if they have the resources, that way one of the factions will always be ahead of the other when not just spamming units.

Anyway, make it as you want it to be, if people say it's a problem (people who played) it's always possible to change some things.

[Google_Frog]

If you've got 1 lab spamming units that cost 300 and the other spamming units that cost 600 with the default system for every 600 unit that gets built you'll have 2 300 units. With pay when you start the 300 units will keep on taking the resources when the start completely preventing any 600 units from being built. That's my only problem with that system.

[Hoi]

To be more clear, lets say core can build when he wants to, and arm has to wait till it has resources, core will, if both are good players, have t2 (example) way before arm, and will pwn them

[Gota]

Argh.
Do you have a final plan on how pure will play?like gameplay wise.
Is there a point at which it will enter the unit balance/tweaking mod?
or will it always be under constant development that alters the gameplay every x months?

[clericvash]

Gota bear in mind it is still testing versions at the moment so changes are of course going to happen from feedback and such.

[smoth]

yan and hoi, it is his project, just let him do it, he could add an unplayable fluffy bunny faction if he wants, It is a beta meaning it is not the final product.

[Gota]

I'm not ordering anyone around,just asking.
If its gonna be like CA i just wont bother checking up on it.
thats why I want to know.
Unit balancing is a part of the beta phase isn't it?