incremental patch discussion

[keldath]

the thing is,

the bbai code, was completely changed,

as i went through the code, 95% of the changed code was to the bbai 081, which we all agree that is needed in dune or any other mod, the rest of the changed stuff was of the revolutuion code - which is an addition, so its an easy merge, the bbai takes existing code and changes it.

even if we just merge the plane bbai, the code caphelo wrote would need to be adjusted, or even likely to be obsoleted at all.

i was able to merge every thing david koma and caphelo wrote, wasn't easy , but nothing was lost, except a koma13 part - that the bbai took a large portion of code that koma13 added inside , out = i think this made koma13'S code unneeded. and as i marked , the caphelo code.


but, on the button line, so far the code checks out, but, as you said, we need more testing,
cause i ahte to have an un stable version...


cheers and y'all have a good week.

but the fact remains, that, in order for us to keep updating the mods sdk, every now and then, each of to code writers need to adjust, bbai is very very important for every mod, im sure we all think this way.

 

[Deliverator]

Thanks for the info. I understand the problem, it is hard to merge two AI routines without really understanding what they are doing.

bbai is very very important for every mod, im sure we all think this way

I don't think it is automatic that Better AI is better for Dune Wars. Better AI is coded for the vanilla setup, but our mod is now very different from vanilla. Transport, bonuses, city placement, civ special abilities are all unique to Dune Wars. In my opinion it would be better to cherry pick the bits of Better AI that are definitely applicable to Dune Wars and leave the rest. Cephalo's code is a great example, that was written to fix a problem unique to Dune Wars and I think it is unlikely that the latest Better AI has made that code redundant.

 

[Slvynn]

I think same as Deliverator do - i had few testing runs, and still , while AI making good "vanilla" moves, it fails to do things that fit to DW and its specifics, example - as i mentioned in Feedback thread. City placement, defence and tech proritizing. Defencive /offencive tactics. All this is different in DW. At offence AI is good, but at defence he fails terrible. There is different balances between offencive/ defencive units, and defence/amount of defence units is much more important in DW.
Atm i winning Deity without any problems. Its harder to win deity in FFH even than in DW.

Because of air transportation (which is not exist in vanilla), and hight speed all terrain transportation (suspensor transports - which is not exist in vanilla too), defence tactics of AI should be reviewed. Defence should be more tough, and spread almost even. There also no roads, but other ways to use terrain. AI never uses Outposts which are nice way to catch some resource or spice. AI leaves too few units in his cities in start of game, and sending to wonder/scout too much.

 

[Deliverator]

Slvynn, have you noticed any differences between 1.6.1 and 1.6.2 with regards to the AI?

 

[Slvynn]

I played alot with 1.6.1 and started 1 game with 1.6.2 (tiny map, epic, Deity, 8 rivals)
I didnt noticed major improvement, I fought against Rabban, which attacked me after i dowed him (dogpiled) not using terrain (mesas), but homeland promo and moved through plain terrain. He had huge stack of unit (i think its because BuildUnitProb) but he had Aggressive+Protective Infantry and i had IM, which have double str of infantry. With help of Ecaz Ally Rollers (woohoo) (Ecaz and Ordos and Me = Landsrad block, and Ecaz DoWed Harks, Ordos same turn, and 2 turns later me with my Militia)) and my double str IM (6 vs 3 of infantry) and b/c anti-guardsman promotion i took for most of my stack to counter him, his stack was destoryed.
He had good defences, but Ecaz rollers were very helpfull. I was one who took all his cities.
Thats bad for him to have army of low str defencive units of one type (guardsman) and attack with it. Ai should consider Known enemy unit statistics and should try to counter, and mix his forces.
I think Monty-like AI should try to upgrade his units asap, and use his early stacks earlier, befoe they obsolete, and before Player grow good defencive army. If he will attack 20 turns earlier in this case, before i finish my bee-lined Maulas, i will start new game , because i will die veeery fast (i was 0.2 of his power ratio for long time).

If we take some basic AI's analysis algorythm:
Check Power ratio, if its < of 0.4 of own
Check what tech target can tech &#8211; if one of techs ulocks highter str unit or counter against own most used unit (of already built big stack) then DoW asap

[That means if warmonger AI already built some stack of certain units he should not simply sit and wait with them&#8230; he should check apporiate target including target's known techs into consideration, and when target gets counter/hight str unit, and if it did not and target have lower power ratio &#8211; DoW asap.]
Its not vanilla, and early wars can be very profitable.

 

[keldath]

well,

in my opinion the bbai is a must, but, i agree that picking the needed things from it is better.
jdog5000m improves the code that dune uses, some of the stuff koma wrote for instance - was already made in the new bbai by jdog.

anyway,

i think we should get stuck on one merged version of revdcm or bbai, we must develop with those mod comps, but, i agree that we may not need all of this.

the thing is, that we lack a fully sdk coder here, that can really put him self and make dw a fully unique code.

im open to suggestions, i brought 1.6.2 but, i need the co-operation of koma13, caphelo and david in order to set the dune codes with bbai.




edit: ok guys, version 1.6.2 is compatible with deliverator's patch 1.6.1b
you can use it both on 1.6.1 and 1.6.2.

enjoy.

 

[Deliverator]

Slvynn pointed out a nasty bug with the Infantry unit breaking the interface. I have updated this post to 1.6.1c with the fix.

 

[Ahriman]

Playing 1.6.1b, just discovered something weird.... build infantry unit (in a town with barracks), and when it is selected much of the UI issues disappear (promotion access, other units in the same tile, etc.)

Check the attached screenshots:
infantry1 if I select a worker, then an infantry (displayed UI: can build mine.
Infantry2 if I select a soldier, then an infantry (displayed UI; can upgrade to infantry)
Infantry3 if I alt-tab and just select infantry. (displayed UIl; none).

Weirdness......

 

[Deliverator]

Yeah, the infantry unit breaks the interface in 1.6.1b. I've just posted an amended 1.6.1c.

 

[keldath]

nobody plays 1.6.2??

me very sad

 

[Ahriman]

Ok thanks, will install.

Another thought I had for tweaking religion; what if we made the missionaries for Shai-Hulad, Imperium and Landsraad (but not Quizarate or Technocracy) require that the religion was your State religion? So only members of the Imperium religion could build imperial missionaries, etc.

Can you tell me what XML in which files I would need to make myself in order to try this?

 

[Ahriman]

What exactly does 1.6.2 *do*? It was hard for me to tell.

 

[Deliverator]

Can you tell me what XML in which files I would need to make myself in order to try this?

In the file Civ4UnitInfos.xml, locate the units called UNIT_IMPERIAL_ADVOCATE, etc, then find a tag called StateReligion and change it e.g. RELIGION_IMPERIAL. It is easy as the tag PrereqReligion is right below it with the tagname you need.

 

[keldath]

Ahriman,

it improves the bbai from 0.78 to 0.81
see revdcm forum for more details.
also it makes the game faster by using elements of car mod.

also, it removed many useless code that lied on the files,
it has buf 4 that has better interface options.

i dont understand why you all reluctant to work with it.

 

[Ahriman]

It is easy as the tag PrereqReligion is right below it with the tagname you need.

Thanks.

i dont understand why you all reluctant to work with it.

Because that still doesn't tell me what it does. I'm not going off to some other forum to figure out what your patch does, and other people who did know what it did seemed skeptical (better vanilla AI != better Dunewars AI).
Maybe I will install it later.

To clarify: if I don't know what it does, I don't know how to test it.

 

[Slvynn]

I like a visual odds features and other thing that come with 1.6.2
cool stuff, really

but, as Deliverator and Ahriman telling - and i agree

(better vanilla AI != better Dunewars AI).

I playing 1.6.2 (1 game started so far) and will, and hope David and koma will touch it soon in future to ensure it is better DW AI.

also afaik there is a bug with HW screen in 1.6.1 where unit list is not updating
it being fixed in 1.6.2

Queston: is this patch 1.6.1c compatible with 1.6.2?

 

[keldath]

i dont like this,

i dont like this at all.

 

[Deliverator]

the thing is, that we lack a fully sdk coder here, that can really put him self and make dw a fully unique code.

Yep, having a full time SDK egghead working on Dune Wars would be fantastic. Calling all SDK geniuses...

i dont understand why you all reluctant to work with it.

Don't be discouraged keldath. We'll work out the best solution.

Now that this is an established mod with a reasonable following, we need to discuss the best course of action more. If you can't convince the forum that something is a good idea, then perhaps it isn't. Generally it is good to test your plans/ideas here and see if people like them. In this case, I feel you went ahead and did something without really listening to the concerns. Just saying "trust me on this one" isn't enough - you need to convince people. If your idea is a good one it will stand up to scrutiny.

 

[Slvynn]

keldath bro - i like 1.6.2 , really and all those features, moe than that you made a work to make it working, and thats great, lets hope codemasters will touch it soon enough and it will be already in 1.6.3 fixed.

Atm i playing it and enjoying all cool stuff. AI anyways require more deep insight and fixing.

Others cool boys can be also more cool if they at least will try it and then will say no.
And atm i am keeping that cool and start another game with cool 1.6.2 thingy game.

 

[keldath]

ok you guys,

despite what i think,

i have a saying - "why replacing a good horse".

if its good, why bother to get another.


ill let everyone decide, im with the majority.