RevolutionDCM for BTS

Also you can look in the config folder, it's the revolution XML file in there. If the Revolution.ini exists in the UserSettings folder that ini file will override the xml settings though, the XML settings are more there for mod makers to set their default revolutions settings. We moved the revolutions settings over to the BUG format to make updating with BUG easier, and also BUG's system is just very logical; Emperor Fool did a great job with it, and it made sense to start using it.

I'm hoping to hear from Fuyu as well before 2.8 is released. The last remaining outstanding item on the todo list is about a bug he found where the AI was not adopting the new victory strategies in BBAI. Since Fuyu has taken over Better AI, I'm hoping he can give us the nod, and let us know if this is working correctly in the latest build before releasing (or if it's not, let us know that it's not a big deal).
 
Great thanks for the update Phungus. It's lookin like a really nice build really stable and exceptional features. We'll wait just a little bit longer it will be worth it. The last game the AI played a really aggressive cultural game which was undeniable. It felt like it knew what it was doing and won the victory very early. Not sure about other victory conditions whether the AI is winning by accident or by design.
Cheers

EDIT: One thing for sure the general unit AI is nice. Marched a stack in to the enemy interior and it was attacked, heckled, harassed and weakened to the point of pain like sticking your finger down an ant's nest.
 
I'm glad to see things progressing well. I love Civ 5, but it'll be a while until it's as polished as BTS, and this mod is what makes BTS fun for me. :)
 
Eh Jabarto
Yeah, it's a lot of raw fun this RevDCM no question! Could you tell me your thoughts on stack attack? Phungus has correctly hidden the option from main view (but not in debug view) because we don't want future generations using stack attack when it is not entirely stable. Dale put a lot of work into it originally and I put a fair bit in as well and it is bullet proof on my machine (but not others). For me, stack automation bridges the gap between civ4 and civ5 in the sense that you can lob off an entire division in animated carnage, attacking the stack as if it were a single unit in Civ5. It's sad for me not to make this option visible to players because I am very fond of it, but I understand where Phungus is coming from.
Cheers
 
I hid stack attack before the release of 2.7. The reason I did so was that we were getting multiple crash reports by users with this option on, and given the workload we have (had) I felt that the option of stack attack and fixing it should take lower precedence then other features and bugs, so I just wanted to clear it on the todo list. It might be worth it to bring it back on the radar, but I felt I made the right decision at the time. Glider if you think it is working now, and doesn't cause a critical bug, you're the producer of the mod, so feel free to bring it back, but I think I made the right decision at that time in the mod's development. RevDCM goes beyond what we are working on here, there are nearly a dozen mods that use the RevDCM gamecore, so I feel that making it a stable core to work from takes precedence over all other considerations.
 
phungus420 said:
I'm hoping to hear from Fuyu as well before 2.8 is released. The last remaining outstanding item on the todo list is about a bug he found where the AI was not adopting the new victory strategies in BBAI. Since Fuyu has taken over Better AI, I'm hoping he can give us the nod, and let us know if this is working correctly in the latest build before releasing (or if it's not, let us know that it's not a big deal).
Click to expand...

I didn't have a chance to look at it again yet. What I observed was that not a single civ ever runs any victory strategy, and that I can't see why this is happening. Whether or not that problem persists can be easily checked: hover over any player on the scroreboard while holding CTRL, and you should see all the status info on the AI, including future war plans and currently running strategies.

Why this should be a big deal: An AI can only win a cultural victory if it is running the VC stats.
Why it probably isn't: The problem with Revolution & AI is that the Revolution part massively changes game rules, and therefore should massively change AI behavior: Slower expansion ("big empire", "financial trouble"), razing cities instead of capturing them (enemy culture, initial starvation, "big empire", "financial trouble"), trying harder to get wonders (bonus for culture), less overseas settlements ("colony", "financial trouble"), higher focus on happiness and health, etcetc.. (I'm no expert, only played 2 games with Revolutions on)
I believe non of the above are happening, because if the AI knew about the effects of Revolutions and could counter them as effectively as they should, there would rarely be a Revolution happening in the game. Or so I think.
VC strats and their effects are coded for the base game, like Conquest1 which makes the AI eager to spend a bit more gold on unit upkeep - which could easily cause "financial trouble", and domination, idk for sure but probably increases the overexpansion problem even more so that's poison for the AI empire with Revolutions. Well, culture would really make sense though.
 
Thanks Fuyu
Good points on the effects of specific strategies on Revolutions. I agree but as you say it's complex. Revolutions effects also quieten down if the civ is succeeding at war but if the VC strat goes horribly wrong what happens to the AI?. We did want Jdog to take a look at the Revolutions AI but he's too busy on his Phd. The very least that should be happening is that with all these options turned off, the core DLL should behave like the Better BTS AI core. Looks like that will need to be looked at before release.
The thing is, I played a Revolutions test game by hand and the AI's culture win must have been very early premeditated because it won the victory early than I can ever recall it happening ever before. Does that confirm the Better BTS AI VC-strat mechanism or not?
Cheers
 
It does sound like it's active, fast cultural victories rarely happen by accident. Or rather, they don't happen at all. I'd still like to know if CTRL+hover will show a VC strat or not.
 
Just did a quick check Fuyu. CTRL + hover doesn't show them. From other tests I've run, the AI is achieving a whole variety of victory conditions and since that damned civ game text code file is so overly massive and bloated, it could well just be a text merging accident that is so easy to happen in that file. It could be the mechanism is working but not the text display. I'll have to check it out.
Cheers
 
Maybe it was ALT + hover? I always mix those up. But there is no text from xml involved in displaying that debug info, so that can't be the cause.
 
Thanks Fuyu yep ALT+hover is the one. Double checked with Better BUG AI and RevolutionDCM. Text displays in both and considering that there is some evidence that the AI is making progress on it's victory strats, it looks like all is well in both. It'll be interesting to hear what Chris has to say from his play testing of RevDCM.
Cheers
 
Eh Chris. No problems here either. There is a question about AI late game victory strategies but there is no evidence to say it is broken and there is some evidence to say that it is working. I guess you will not have enough time to play through an entire game to see the AI's late game behaviour true? Suspect that I'll just go into release phase in the next 24 hours.
Cheers
 
If the line that starts with "Vic Strats: " doesn't end right there but actually lists at least one or 2 strategies, and if the AI players actually win early victories too sometimes then there is no doubt left that it's working now.
glider1 said:
I have added some modding level commenting to Fuyu's code where I think it wasn't commented previously and so RevDCM's Better BUG AI code is probably more up to date than Fuyu's code for a short instant in time.
Click to expand...
I don't comment changes to my own code more than necessary, so most of those "//RevDCM - uncommented Fuyu changes" is PlayerAI aren't really necessary. If you compare CvPlayerAI::AI_civicValue() of Better BTS AI with the one from Better BUG AI you should see it very clearly, I think that function is 5 times longer now.
What I did in TeamAI was a reaction to what jdog did to AI_isLandTarget().. I still don't know for sure if it's ok as it is now, it was done to reduce early warring , which was happening because AI_isLandTarget() returned true a lot more often than before jdog's changes. (Afforess used a compltely different approach to counter that)

There's still a lot to be done which can be summed up as "Revolution-specific AI" but that can always be added later., in case you ever find someone willing and able to work on that.
For now I'm committing CvPlayerAI with city razing levels back to what jdog thought was right. Only when NO_REVOLUTIONS is on, my reduced version will come into play.

I think you're ready for release now.
 
Thanks Fuyu!
Ok with the definite early culture win and the correct text display VC strat is working. I'll make a note that next release of RevDCM to take another look at those sections. The early war AI must be a lot of fun to tune! It will be interesting to study your approach, verses Jdog's approach, verses Afforess approach. Considering that Jdog is temporarily unavailable and that Afforess has probably moved onto civ5, their code will possibly remain static. It will be interesting whether you do any more work on it. I've traditionally felt that the ancient era's are too stable (for my taste) and thus have pretty much always played with the start-as-minors option which has evolved over time to be quite nice. However if the early war code shows itself to be more flexible and dynamic, start-as-minors is not technically necessary for players like myself.

I'll make a note about it and examine it again next release. Thanks!
Cheers
 
Early wars are nice for human players, who can use small numbers of units to great effectiveness, but hurt the AI, which simply can't. The AI performs better in peace, as long as it can expand - which is the reason Afforess wrote something to avoid war-declaring when there's enough land to settle left. Starting as minors means early war for everyone, so that probably doesn't favor the AI either.
If you want war in the Ancient era you need to increase number of players. If there's no place left to settle, the AI will come after you soon enough.
 
glider1 said:
Thanks Fuyu!
Ok with the definite early culture win and the correct text display VC strat is working. I'll make a note that next release of RevDCM to take another look at those sections. The early war AI must be a lot of fun to tune! It will be interesting to study your approach, verses Jdog's approach, verses Afforess approach. Considering that Jdog is temporarily unavailable and that Afforess has probably moved onto civ5, their code will possibly remain static. It will be interesting whether you do any more work on it. I've traditionally felt that the ancient era's are too stable (for my taste) and thus have pretty much always played with the start-as-minors option which has evolved over time to be quite nice. However if the early war code shows itself to be more flexible and dynamic, start-as-minors is not technically necessary for players like myself.

I'll make a note about it and examine it again next release. Thanks!
Cheers
Click to expand...

Fuyu said:
Early wars are nice for human players, who can use small numbers of units to great effectiveness, but hurt the AI, which simply can't. The AI performs better in peace, as long as it can expand - which is the reason Afforess wrote something to avoid war-declaring when there's enough land to settle left. Starting as minors means early war for everyone, so that probably doesn't favor the AI either.
If you want war in the Ancient era you need to increase number of players. If there's no place left to settle, the AI will come after you soon enough.
Click to expand...

Eventually, once I merged Fuyu's corrected early war starting AI, I removed my changes, simply because I didn't notice a huge difference anymore.

On the subject of AI though, I found a bug in Phungus's Inquisition AI that will prevent inquisitors from ever moving to a needed city, and more importantly, am trying to write some AI code to make the AI create some specialized military production cities, like humans will often do.
 
Hi ya Afforess
Yes I just have noticed your contribution to the CvGameCoreDLL for RevDCM as well as Fuyu's as well. Thanks! Ok so there is still consistency in the early war approaches between us. The inquisitor AI is nice to tune as well. I'll now wait another 24 hours just to see if the dust settles ok before releasing. It doesn't matter anyway because the weekend is almost over and the October CIV4 mod announcement is already out. However I will release once it does completely settle down.
Thanks again.
Cheers
 
You must log in or register to reply here.

Similar threads

H
Custom music folder troubleshooting
Replies
1
Views
563
<Nexus>
<Nexus>
MantaRevan
Can't seem to find CIV4TraitInfos.xml for trait editing
Replies
7
Views
628
<Nexus>
<Nexus>
D
How to run a mod running under ProtonGE
Replies
1
Views
323
a pen-dragon
A
gavenkoa
BULL hover popup text "Speeds Up ..." on F6 is not shown
Replies
5
Views
403
f1rpo
f1rpo
S
All game crashing for a unknown reason
Replies
15
Views
768
burticus
B
Back
Top Bottom