• We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
I think there is some sort of misunderstanding. Let me quote \ai\ai_file_doc.txt:



I would like to have a fourth paramenter production = 0.5 that make the ai never allocate more than 50% of the effective ic for production. In the later game ai events could decrease the value to 30% or 10% if short on rares etc..
About adding new serial to production. Would it be sufficient to simply allow new projects only if the ic allocated for production is higher than the ic required for the current production?

Yes, I was talking about a different parameter, that will prevent AI adding new units into production queue, for example production_limit = 2 , if units in queue more than 2 * base IC, nothing can be added into production any more.

As for your suggestion to have the production parameter behave as upgrading and reinforcement, this is not possible with the current system, because production is allocated last with everything remaining from the other five sliders.
 
Last edited:
as for adding leaders and ministers, please refrain from doing so. AoD gfx and DB is already a mess. better hunt down people without a pic (mostly late war ministers/leaders for eg USA and UK). completely new pics can always come with mods.
same for province pics and stuff. please keep the base game as slim as possible, modders can do the rest.

Well, I suggest skip my Romania suggestions since they are based very much on bestmajor pack but keep the minor tech teams and Brazil need new materials.
Some of my province pictures should be accepted not all but some of them.

province pictures
ill_prov_1422 new picture fixed just place it in folder and it replace the old. ill_prov_501 and ill_prov_900 are already added by pang bingxun I think.

ill_prov_31 picture of Malta is nice maybe that can be added.
ill_prov_1457 is a picture of Calcutta maybe useful to have some variation in British India.

ill_prov_1121 picture of Pretoria in South Africa not the best picture but it is ok even so not a must either.
ill_prov_804 picture of an oil field in south America cant remember its name. its a nice picture but I has blunt cut in it. The reason is to make room for the province name.

Straits
Ill_prov_397 sea of Mamara ok, but nos so good
Ill_prov_2430 Gibraltar strait nice picture
Ill_prov_2453 Suez channel not the best
Ill_prov_2575 Panama channel nice picture but maybe to small.
 
Last edited:
Yes, I was talking about a different parameter, that will prevent AI adding new units into production queue, for example production_limit = 2 , if units in queue more than 2 * base IC, nothing can be added into production any more.

Imo this limit must be as small as possible. If the total ICD in the queue is bigger than ic allocated x max batch time, then something is wrong.

As for your suggestion to have the production parameter behave as upgrading and reinforcement, this is not possible with the current system, because production is allocated last with everything remaining from the other five sliders.

Unfortunatly this makes the AI allocate much too less into reinforcements and upgrades when needed.
 
A question regarding auto-merge of trades or trade convoys: does it also affect trades set up with the parameter cancel = no (AI can't cancel them unless at war with the other nation) ?

EDIT: using relative build sheme is good as AI would then prioritize units it considers as having insufficient numbers, otherwise it would just be a dumb IC allocation to unit types according to build prefs. But it would be a lot of work to change all ai files to relative, even most (all?) german switch files don't use relative.

Pang: in ger_1936.ai garrison attachment prefs are set to none (to save manpower ?!?), please change this to 100% Police. That's what they used these kind of units for. DD are set up with 100% ASW, in reality german DD had a focus on surface combat.
 
Last edited:
There would be a shortage on manpower for most countries. This could be solved by increasing the starting manpower and the daily mnpower gain by 50% aswell. The advantage would be more accurate loss statstics.
Sounds good, I guess this could be achieved by reducing the effect ageing has by 50%? I edit the divisions stats and costs if you want to go through with it.

The ai events would do that. Afaik most divisions of axis minors were trained and equipped in a poor way, so brigading them would not be very appropriate.
Imo the starting conditions in the scenario files should be more or less accurate while the behavior of the ai could be less historic.
Ok, fair enough.
 
works well just need someone to adjust the tech file for Mengkukou because it cant read the csv 100%

I think you deleted 2 ";". The number of them is constant undepended on how many specialties a team really has.

Pang: in ger_1936.ai garrison attachment prefs are set to none (to save manpower ?!?), please change this to 100% Police. That's what they used these kind of units for. DD are set up with 100% ASW, in reality german DD had a focus on surface combat.

1. I tend to believe that 5 Gar-none are better than 4 Gar-Pol. Especially as the AI seems to be relatively poor in handling those immobile division. Is there some evidence that POl brigades will do more good than harm to the AI?
2. Really? This explains those strange starting brigades. Will the AI be able to utilze other brigades in a proper way? :confused:
 
Pang Bingxun said:
What abut the starting divisions? Especially romania starts with a lot of divisons and only little manpower available for more. I could write some refit events using remove_division and replace them with new divisions with the same name but a new brigade.
well i actually like the historical OOBs, even if its not optimised for combat performance game wise. my idea was to simply give them the tech (supply dumps) or set the start year for supply dumps down to 1933.
my problem was just that i gave BUL ai orders to build ENG brigades, just it was too dumb to research the required tech up to 1940 or even later.
as for number of divisions i would rather try to reduce the overall number in game.
 
Yes, I was talking about a different parameter, that will prevent AI adding new units into production queue, for example production_limit = 2 , if units in queue more than 2 * base IC, nothing can be added into production any more.

As for your suggestion to have the production parameter behave as upgrading and reinforcement, this is not possible with the current system, because production is allocated last with everything remaining from the other five sliders.

I would like an event command that prevents the ai from building new units at all, most axis minors build that many divisions that they cannot reinforce them because they lack the manpower to do so.

Also, would it be possible to let the game continue unattended (= not the active window) if the cheat "handsoff" is entered? This would make modding easier, becaus you could run handsoff games and do other things with your PC.
 
Last edited:
A general thing about relative buildung scheme. The AI will not handle l.Arm and Arm as one unit type, will it? In that case the relative building scheme might have some serios downsides if l.Arms are build in the early game. Before they are upgraded the AI will try to match proper number of ARM but that leads to the effect of a too high number of Arm including l.Arm.. Germany will therefore build too many Armoured divisions till barbarossa. The Japanese AI might have an even more serios problem. If it upgrades all l.Arm to Arm it will try to rebuild l.Arm, causing an infinity loop. :(
The solution is to build Arm only via building scheme. Building l.Arm would require ai events. A length parameter for the build divisions command could help a lot. Right now one serials of length 1 are possible.

Pang Bingxun you can add those pictures. I have tested them and they works fine.

Ok, i overwrite the older pictures.
 
A question regarding SCW: Have you changed the way NatSpa receives its divs?

This is the second time in a row that NatSpa is almost crushed in the first few months.

I looked up the savegame of Jun 17 1936 and found out that NatSpa has only two(!) divs to start with. That´s somewhat strange IMO; even with the additional divs from ITA and GER they have no possibility to conquer RepSpa.
 
Here is a new set of changes.

Code:
\ai\afg_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\ast_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\aus_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\bel_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\bol_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\bra_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\bul_1936.ai
		cavalry techs are no longer ignored to be researched
		reinforcement = 0.5000 changed to 0.35
		
\ai\chi_1936.ai
		new endgoal 1220
		armor = 3 changed to 4
		aircraft = 1 changed to 2
		land_doctrines = 4 changed to 2
		
\ai\chl_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\col_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\cub_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\fin_1936.ai
		reinforcement = 0.5000 changed to 0.35
		
\ai\ger_1936.ai
		Mulhouse, Frankfurt am Main, Breslau, Königsberg, Oppeln, Dortmund and Innsbruck are now IC provs
		Königsberg is now infra prov
		garrison = 0 changed to 2
		hq = 3 changed to 1
		
\ai\gre_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\hun_1936.ai
		reinforcement = 0.5000 changed to 0.35
		
\ai\jap_1936.ai
		Infra provs fixed
		
\ai\man_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\men_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\mon_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\mex_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\omn_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\per_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\nzl_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\rom_1936.ai
		reinforcement = 0.5000 changed to 0.35
		
\ai\spa_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\spr_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\tur_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\ven_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\yem_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\yug_1936.ai
		cavalry techs are no longer ignored to be researched
		
\ai\switch\ger_build_land.ai
		relative_build_scheme = no changed to yes
		infantry = 32 changed to 22
		armor = 10 changed to 9
		garrison = 0 changed to 6
		hq = 7 changed to 2
		submarine = 0 changed to 10
		
\ai\switch\ger_build_mix.ai
		relative_build_scheme = no changed to yes
		armor = 10 changed to 9
		garrison = 0 changed to 6
		hq = 7 changed to 2
		
\ai\switch\ger_build_navy.ai
		relative_build_scheme = no changed to yes
		garrison = 0 changed ro 4
		hq = 5 changed to 1
		
\ai\switch\JAP_Build_Fort.ai
		max_factor = 0.02 changed to 0.40
		
\ai\switch\JAP_Build_Fort_Stop.ai
		max_factor = 0.00 changed to 0.40
		
\ai\switch\jap_naval_pacific.ai
		max_factor = 0.10 changed to 0.40
		
\ai\switch\JAP_Pacific.ai
		max_factor = 0.10 changed to 0.40
		
\ai\switch\JAP_Pacific_war_Phase4.ai
		max_factor = 0.10 changed to 0.40
		
\ai\switch\max_factor_050.ai
		max_factor = 0.45 changed to 0.5
		
\config\Boostertext.csv
		file by Denniss
		
\config\event_text.csv
		file by Denniss
		
\config\extra_text.csv
		file by Denniss
		
\config\new_text.csv
		file by Denniss
		
\config\province_names.csv
		file by Denniss
		
\db\revolt.txt
		repaired line 1097(RUS)
		
\db\events\AI_BRA.txt
		33060: event now reduce max_factor to 0.5 only
		
\db\events\AI_CAN.txt
		36173: event now reduce max_factor to 0.5 only
		
\db\events\AI_CHI.txt
		27590-27593: Changed chain of events
		file by Denniss
		27031: Domestic value = 9 changed to 10
		file by Denniss
		
\db\events\AI_CHL.txt
		35360: event now reduce max_factor to 0.5 only
		
\db\events\ai_disbanding.txt
		file completely overhauled
		
\db\events\ai_fin.txt
		32064-32071: new ai events to give control of finnland cores to finnland
		
\db\events\ai_ger.txt
		20007: Barbarossa should now start historical if historical outcome option enabled, reduced may chances
		20556-20560: changed chain of events
		file by Deniss
		file by Deniss
		20555: Redelayed to november
		
\db\events\AI_Inherit_Chi.txt
		7005-7010: Inherit events are now possible for human CHI
		7012: 7006 leads to 10 dissent
		7013: 7007 leads to 10 dissent
		7014: 7008 leads to 10 dissent
		7015: 7009 leads to 10 dissent
		file by Denniss
		
\db\events\AI_ITA.txt
		26002: italy may now attack greece if sardinia lost
		26521: minister is now in the correct section
		26550: event outcommmented
		26551: event outcommmented
		26552: event outcommmented
		
\db\events\ai_jap.txt
		22554-22556: changed chain of event
		file by Deniss
		
\db\events\ai_ministers.txt
		11054: headofgovernment which = 21056 changed to 21025
		11055: foreignminister which = 21069 changed to 21001
		11056: foreignminister which = 21071 changed to 21003
		
\db\events\AI_MEX.txt
		32962: event now reduce max_factor to 0.5 only
		
\db\events\AI_POL.txt
		file by Deniss
		
\db\events\ai_sov.txt
		25005: soviet gear up reverted to original state
		25552: new event for some dissent in early barbarossa
		25553: new event for some dissentreduction in mid barbarossa
		
\db\events\AI_SWE.txt
		32660: event now reduce max_factor to 0.5 only
		
\db\events\death_events.txt
		364: Rudolf Hess flees to the United Kingdom only when still controlled by ENG, picture changed
		391: Reinhard Heydrich assassinated only when London still controlled by ENG
		
\db\events\hoi2.txt
		90: new UnitedFrontFlag
		91: new UnitedFrontFlag
		
\db\events\italy.txt
		2900: new cores are seceded to VIC again
		
\db\events\japan.txt
		3894: Garrisons will now occur in Hanoi instead of being obliterated in Haiphong
		3836: new event to unlock garrisons
		file by Denniss
		
\db\events\nationalist_china.txt
		29032: new UnitedFrontFlag
		
\db\events\UK.txt
		3408: new event to unlock garrisons
		file by Denniss
		
\db\events\USA.txt
		3709: construct which = ic where = -1 changed to -4
		3710: construct which = ic where = -1 changed to -4
		2729: new event to unlock garrisons
		file by Denniss
		
\db\events\USSR.txt
		2653: Third GPW event may now fire if YUG destroyed
		
\db\events\vichy_new.txt
		100003: does no longer triger 100004 directly
		100004: now triggered indirectly
		
\db\ideas\Ideas.txt
		file by Denniss
		
\db\leaders\brazil.csv
		file by Pioniere
		
\db\leaders\romania.csv
		file by Pioniere
		
\db\leaders\leadersU73.csv
		file by Denniss
		
\db\minister\ministers_bra.csv
		file by Pioniere
		
\db\minister\ministers_rom.csv
		file by Pioniere
		
\db\tech\teams\teams_afg.csv
		file by Pioniere
		
\db\tech\teams\teams_csx.csv
		file by Pioniere
		
\db\tech\teams\teams_fla.csv
		file by Pioniere
		
\db\tech\teams\teams_men.csv
		file by Pioniere
		
\db\tech\teams\teams_sco.csv
		file by Pioniere
		
\db\tech\teams\teams_spa.csv
		file by Pioniere
		
\db\tech\teams\teams_spr.csv
		file by Pioniere
		file by Denniss
		
\db\tech\teams\teams_vic.csv
		file by Pioniere
		
\gfx\events_pics\Austria_Alliance.bmp
		file from Doomsday

\gfx\events_pics\Fail_Hitler_Assasination.bmp
		file from Doomsday

\gfx\events_pics\last_flight.bmp
		file by Pioniere

\gfx\interface\ill_prov_31.bmp
		file by Pioniere

\gfx\interface\Ill_prov_397.bmp
		file by Pioniere

\gfx\interface\ill_prov_453.bmp
		file by Pioniere

\gfx\interface\ill_prov_501.bmp
		file by Pioniere

\gfx\interface\ill_prov_804.bmp
		file by Pioniere

\gfx\interface\ill_prov_900.bmp
		file by Pioniere

\gfx\interface\ill_prov_1121.bmp
		file by Pioniere

\gfx\interface\ill_prov_1422.bmp
		file by Pioniere

\gfx\interface\ill_prov_1457.bmp
		file by Pioniere

\gfx\interface\Ill_prov_2430.bmp
		file by Pioniere

\gfx\interface\Ill_prov_2453.bmp
		file by Pioniere

\gfx\interface\Ill_prov_2575.bmp
		file by Pioniere

\gfx\interface\pics\Alexiu_Paul_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Antonescu_Victor_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Atanasescu_Constantin_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Avramescu_Gheorghe_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Barzotescu_Emanoil_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Beldiceanu_Ioan_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Bengliu_Ioan_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Corbuleanu_Mihail_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Cornicioiu_Grigore_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Cretulescu_Ilie_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Garbea_Titus_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Georgescu_Pion_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Ilasievici_Constantin_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_Castro.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_CostaEuclides.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_Farias.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_Freire.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_GomesdeLima.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_Paula_Cidade.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_RegueraIsauro.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_SouzaLima.bmp
		file by Pioniere
		
\gfx\interface\pics\L_BRA_XavierMario.bmp
		file by Pioniere
		
\gfx\interface\pics\M22012.bmp
		file by Pioniere
		
\gfx\interface\pics\M22106.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_AugustoRademaker.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_Ernesto.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_GóesMonteiro.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_JoãoAmazonasde.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_MárcioMelo.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_MaurícioGrabois.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_Médici.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_Morais.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRA_Silva.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRAFranciscoG.C.B.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRALeonelBrizola.bmp
		file by Pioniere
		
\gfx\interface\pics\M_BRAPintoGuedes.bmp
		file by Pioniere
		
\gfx\interface\pics\M_ROM_GheorgheTatarescu.bmp
		file by Pioniere
		
\gfx\interface\pics\M_ROM_PrezanConstantin.bmp
		file by Pioniere
		
\gfx\interface\pics\Manoliu_Gheorghe_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Marinescu_D_Gheorghe_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Negulescu_Ion_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Papadopol_Constantin_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Paraschivescu_Ermil_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Partenie_Ilie_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Petrescu-Comnen_Nicolae_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Popescu_I_Dumitru_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Serb_Teodor_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Sidorovici_Teofil_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\T13013.bmp
		file by Pioniere
		
\gfx\interface\pics\T13014.bmp
		file by Pioniere
		
\gfx\interface\pics\T13015.bmp
		file by Pioniere
		
\gfx\interface\pics\T_AFG_Kabul.bmp
		file by Pioniere
		
\gfx\interface\pics\T_AFG_MohammedDaoud.bmp
		file by Pioniere
		
\gfx\interface\pics\T_BEL_Ilya_Prigogine.bmp
		file by Pioniere
		
\gfx\interface\pics\T_CSX_Railway.bmp
		file by Pioniere
		
\gfx\interface\pics\T_FLA_Leon_Gillis.bmp
		file by Pioniere
		
\gfx\interface\pics\T_MEN_Hohhot_uni.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SCO_John_Francis_Ashley_Erskine.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SCO_Sheila_Scott_Macintyre.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SCO_Sir_Robert_A_Watson-Watt.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SCO_WilliamBeardmore.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SCO_WS.Douglas.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SPA_AgustinGrandes.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SPA_Eduardo_Gonzalez-Gallarza.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SPR_FelipeSandino.bmp
		file by Pioniere
		
\gfx\interface\pics\T_SPR_Juan_Modesto.bmp
		file by Pioniere
		
\gfx\interface\pics\T_VIC_Andre_Bayle.bmp
		file by Pioniere
		
\gfx\interface\pics\T_VIC_Jean-MarieCAbrial.bmp
		file by Pioniere
		
\gfx\interface\pics\Teclu_Iacob_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Teodorescu_Iosif_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Teodorini_Corneliu_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Trestioreanu_Constantin_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Vasilescu_Cristea_rom.bmp
		file by Pioniere
		
\gfx\interface\pics\Vasiliu-Rascanu_Constantin_rom.bmp
		file by Pioniere
		
\map\adj-defs.csv
		file by Denniss

\scenarios\0_gc_1936.eug
		db\events\ai_disbanding.txt is no longer included
		trade 325: increased supplies to CGX
		
\scenarios\1936\cgx_36.inc
		file by Denniss
		
\scenarios\1936\chc_36.inc
		file by Denniss
		
\scenarios\1936\chi_36.inc
		file by Denniss
		new blueprints 1200 1210 1220
		
\scenarios\1936\csx_36.inc
		file by Denniss
		
\scenarios\1936\cxb_36.inc
		file by Denniss
		
\scenarios\1936\cyn_36.inc
		file by Denniss
		
\scenarios\1936\japan.inc
		-6x12 manpower
		added 6x12 Inf-Art in the building queue
		file by Denniss
		
\scenarios\1936\man.inc
		new tech 1020
		
\scenarios\1936\saudi.inc
		-1 manpower
		added 1x4 Infrastructure for Riyadh in the building queue
		file by Denniss
		
\scenarios\1936\uk.inc
		file by Denniss

A question regarding SCW: Have you changed the way NatSpa receives its divs?

This is the second time in a row that NatSpa is almost crushed in the first few months.

I looked up the savegame of Jun 17 1936 and found out that NatSpa has only two(!) divs to start with. That´s somewhat strange IMO; even with the additional divs from ITA and GER they have no possibility to conquer RepSpa.

The SCW is very randomized. About 50% of SPR troops change allegiance to SPA. I donnot know the exact conditions but those starting troops and their locations tend to be the most important factor for the SCW.
 
German ai events should switch to a ger_build_xxx.ai switch file once at war or in 1940 and there's no L-Arm anymore.
I just had at look at the CORe files regarding build shemes and build lengths - all AI seem to use relative build shemes. Lengths vary but GER has a lot with 1100 day while ENG is at ~700 in the initial ai files. Length at war/home front are usually, but not always, shorter.
 
German ai events should switch to a ger_build_xxx.ai switch file once at war or in 1940 and there's no L-Arm anymore.

That is true, but then there might be 10+10 Arm divisions, so after upgrading there might be 20 Arm. Is this a bit too much? Assuming germany loses all Arm+l.Arm than afterwards the Ai would attempt to build only 10 Arm. This is a bit suboptimal. But ther real problem would be japan anyway. A possible solution still using Arm would be 3 building schemes. One before 1939 Arm, one after it and one after Arm 1943. In the first l.Arm is allowed to be build. In the second neither l.Arm nor Arm are allowed to be build. In the last one only Arm is allowed to be build.
I might try to change all building schemes to relative ones. Are there any known downsides of relative building schemes?
 
Is there a way to make the AI keep a manpower reserve? Say reserve = 50 makes the ai not build units that would decrease the manpower below 50. Even better would be a relative value. reserve = 0.1 would make the ai keep a reserve of 10 manpower per every 100 manpower in the military.