ls612's C2C Units

[Thunderbrd]

This is also why i wanted to have Drill-like canine promotions. Canine unit class should get the Combat promotions though. I think those combat classes should be removed and the other promotions like Combat and Drill should just include Canine unit class in who can take those promotions.

Easy to update now with promotionlines. Just give the promotionlines you want Canines to have access to the indication of UNITCOMBAT_CANINE and it will automatically allow all canine's access to the promotions on that line.

If you wanted to limit how far along the line you want that unitcombat to have access to, simply use the new NotOnUnitCombat tag on the specific promotion you wish them to NOT be able to achieve along that line. Thus if you are wanting Canines to only have access to Combat I - Combat III but no higher, then you establish Canines access according to the Combat Promotionline and cut that access off with NotOnUnitCombat at Combat IV.

 

[Hydromancerx]

Well aside from the canine specific promotions what other promotions should we have them have? I suspect it would be stuff like Combats, Drill and the Terrain promotions.

Ones like Martial Arts, Bamboo Armor, etc I don't think should be. Which since they were under the Melee unit category they were able to get, but probably should not for dogs.

 

[ls612]

Well aside from the canine specific promotions what other promotions should we have them have? I suspect it would be stuff like Combats, Drill and the Terrain promotions.

Ones like Martial Arts, Bamboo Armor, etc I don't think should be. Which since they were under the Melee unit category they were able to get, but probably should not for dogs.

OK, I'll change that along with the Gamespeeds. Which will hopefully be tomorrow.

 

[Koshling]

Wrong area but here it is: (Some memory decay also listed??)

http://www.gamefront.com/files/22424979/America1.zip

I think the fundamental problem here is that there is no UNIT_AI category that is really appropriate for them (apart from pillage).

If the AI wants to build attack units why would it ever build ambushers over light swordsmen (or even axemen, or even Atlatl)?

The reason a human would is that it's useful to have a few invisible/hidden nationality units to snipe with.

However, the AI only has broad categories like UNITAI_ATTACK, and when it decides it needs to build something for that category it has to use a consistent evaluation, which means unless they are VERY close in value (in which case you'll get an approx. 50-50 mix) it will build the dominant unit of the period. This will usually not be ambushers (for example). If we tweak the assessment parameters for UNITAI_ATTACK to make it produce ambushers the net result will be attack stacks full of (or at least half composed of) that sort of units, which is far worse than not using them at all.

I think we need either:

1) An entirely new UnitAI to handle hidden nationality units better. UNITAI_ATTACK_SNIPE say. That way we can build hidden nationality units and use them appropriately (rather than putting them in cities where they spread crime if they have defense AIs, or attack stacks where they perform poorly at attacking cities). The AI can also build an appropriate number of them without compromising other categories of unit.

or

2) Make them all default to UNITAI_PILLAGE and broaden the AI's interpretation of that UnitAI type to do more than just pillage (basically to harry and snipe as well)

(2) would be the easiest to do, and I don't really see a downside.

Anyone see a problem with taking that approach?

 

[Koshling]

Oh no not again! A few versions back the AI went crazy with sending out canine units. I guess they are doing it again!

Could someone provide BBAI logs where the AI is building lots of these please (I checked the one SO posted in regard to not building ambushers etc., but it only has occasional building of guard dogs, and always as pillage counter units to a total of just over 1 per city, which is as expected)

 

[strategyonly]

I think the fundamental problem here is that there is no UNIT_AI category that is really appropriate for them (apart from pillage).

I think we need either:

1) An entirely new UnitAI to handle hidden nationality units better. UNITAI_ATTACK_SNIPE say. That way we can build hidden nationality units and use them appropriately (rather than putting them in cities where they spread crime if they have defense AIs, or attack stacks where they perform poorly at attacking cities). The AI can also build an appropriate number of them without compromising other categories of unit.

or

2) Make them all default to UNITAI_PILLAGE and broaden the AI's interpretation of that UnitAI type to do more than just pillage (basically to harry and snipe as well)

(2) would be the easiest to do, and I don't really see a downside.

Anyone see a problem with taking that approach?

You know what, i really like your explanations of stuff, excellent work at explaining stuff I am not even going to make it harder on you, you ought to know that by now, so always take the best/easiest approach i leave it in your capable hands, as always, i trust you completely.

 

[Thunderbrd]

I think the fundamental problem here is that there is no UNIT_AI category that is really appropriate for them (apart from pillage).

If the AI wants to build attack units why would it ever build ambushers over light swordsmen (or even axemen, or even Atlatl)?

The reason a human would is that it's useful to have a few invisible/hidden nationality units to snipe with.

However, the AI only has broad categories like UNITAI_ATTACK, and when it decides it needs to build something for that category it has to use a consistent evaluation, which means unless they are VERY close in value (in which case you'll get an approx. 50-50 mix) it will build the dominant unit of the period. This will usually not be ambushers (for example). If we tweak the assessment parameters for UNITAI_ATTACK to make it produce ambushers the net result will be attack stacks full of (or at least half composed of) that sort of units, which is far worse than not using them at all.

I think we need either:

1) An entirely new UnitAI to handle hidden nationality units better. UNITAI_ATTACK_SNIPE say. That way we can build hidden nationality units and use them appropriately (rather than putting them in cities where they spread crime if they have defense AIs, or attack stacks where they perform poorly at attacking cities). The AI can also build an appropriate number of them without compromising other categories of unit.

or

2) Make them all default to UNITAI_PILLAGE and broaden the AI's interpretation of that UnitAI type to do more than just pillage (basically to harry and snipe as well)

(2) would be the easiest to do, and I don't really see a downside.

Anyone see a problem with taking that approach?

It would seem to me to be a more complex route to differentiate between pillage ais that should arrange themselves in cross-counter stacks to go in and destroy infrastructures during war from hidden nationality units, but otherwise I can't see a downside per se. But the end result would seem to me as if it would end up cleaner and easier to perceive within the code design if we designed a new ai entirely. Additionally, we may wish to begin adding some unique missions that only those sorts of units might be capable of accessing and it would make it easier for us to include mission ai in those cases with a more defined role.

But I'd be happy with an improvement there either way.

I've begun to consider that we might wish to take a more 'generic' approach to ai that allows us to have a lot more xml control over their behaviors but I'll discuss that concept in better depth later when I have a better idea of how it could be set up.

 

[ls612]

2) Make them all default to UNITAI_PILLAGE and broaden the AI's interpretation of that UnitAI type to do more than just pillage (basically to harry and snipe as well)

(2) would be the easiest to do, and I don't really see a downside.

Anyone see a problem with taking that approach?

2 sounds fine to me if it's easiest for you. I always interpreted PILLAGE in my design as both actual pillage and killing of random nearby units, so making the AI actually do that can't hurt.

 

[strategyonly]

See question in pic 1, thx.

 

[ls612]

See question in pic 1, thx.

Smithing is three rows behind and one era before Civil Service, so the Flailman would be superior to the Swordsman. The Heavy Swordsman comes at about the same time as the flailman, and is more equivalent strength-wise.

 

[Hydromancerx]

Yeah there are main 3 melee lines ...

Very Good vs Melee
Stone Axeman -> Axeman -> Heavy Axeman

Very Good vs City Attack
Light Swordsman -> Swordsman -> Heavy Swordsman

Sort of Good at Melee and City Attack
Stone Manceman -> Maceman -> Flailman

 

[strategyonly]

Smithing is three rows behind and one era before Civil Service, so the Flailman would be superior to the Swordsman. The Heavy Swordsman comes at about the same time as the flailman, and is more equivalent strength-wise.

I understand that but whats the use of buy/building a Swordsman, when ONLY three techs later you have the same STR but way better % of other fighting?? I might be writing this wrong, i hope you understand now??

 

[Vokarya]

I understand that but whats the use of buy/building a Swordsman, when ONLY three techs later you have the same STR but way better % of other fighting?? I might be writing this wrong, i hope you understand now??

I do it the opposite; I never build Flailmen, because I'd rather build Swordsmen for my city attack stats (and then if I get to Armor Crafting, upgrade to Heavy Swordsman) with Axe/Spear units as stack guards.

 

[Vokarya]

Is anyone planning a separate combat class for Corporate Executives? I'd like to be able to use the new <UnitCombatProductionCostModifier> tag to speed up Executive production, but all Executive units have a <Combat> of NONE.

 

[Thunderbrd]

Not a bad idea to give them their own then. COMBATCLASS_EXECUTIVE... sure... why not? All units should have a combat class anyhow.

There are some interesting combat executive types in the units section here on the CivFanatics... perhaps we could have a few exec types that act like Criminals as well. (eventually anyhow)

 

[Vokarya]

Not a bad idea to give them their own then. COMBATCLASS_EXECUTIVE... sure... why not? All units should have a combat class anyhow.

There are some interesting combat executive types in the units section here on the CivFanatics... perhaps we could have a few exec types that act like Criminals as well. (eventually anyhow)

I think that would fit into Cyberpunk very well. Allow training Megacorporation Execs that can fight.

 

[Thunderbrd]

So I'll put adding that Combat Class on my todo list.

 

[ls612]

So I'll put adding that Combat Class on my todo list.

I can make it tomorrow.

 

[Thunderbrd]

Sounds good to me then. It's a fairly minor task either way and I think you've got a bit more time for this sort of thing than I do. I take it you'll add the CC to all the Execs as well then?

 

[Hydromancerx]

@ls612

Here is an icon/button for the Bear Rider. The old icon can still be used for the Bear Knight. This is so we don't have 2 units with the same icon/button.