Worked on Eylau for another couple of hours last night. I typed up my list of remaining tasks required to take the game Beta and sent them to Chris, along with the post-Origins 2004 baseline.
I then continued on with my effort to allocate Corps assignments down to the Corps' divisions. I am thinking now that the whole Corps assignment vs. Division assignment question should revolve around the allocation of Command Points. Leaving the assignment with the Corps means that the CP will be allocated (if available) to the Corps. Division assignments mean that the each Division "competes" individually for CPs. Given the way the CP and Manuever Unit rules work, I will probably keep the Corps assignment if all subcommands (divisions) of that Corps are in command w.r.t. the Corps Leader.
[some more thoughts...
It occurs to me as I write this that the number of assignments should correspond to the number of things I want the AI to be doing in a given turn. An assignment is just a unit of work for the AI; a "division" of labor. Right now I am characterizing plans using two criteria: success vs. failure, and coverage of all FEBA goals (victory locations) vs. incomplete coverage. The AI is set up to pick the best plan -- this is the most successful plan that covers all the goals (if possible). I think I can avoid breaking up Corps assignments into Division assignments as long at the number of Corps present is >= the size of the FEBA. I'll also need to add code to ensure that each engaged Corps may only be assigned to nearby goals, and that Divisional assignments can maintain Corps command span integrity (for offensive missions). The AI should allow Corps command span limits to be exceeded in the defense. And any Corps not in contact should be free to move to either flank. Thus the reserve becomes even more important.
I wonder if I should write the FEBA-selecting code to keep the number of goals <= the number of Corps (on the Imperial side) or Divisions/Detachments (on the Russian side)?
I'll have to look at the FEBA-selecting code for the larger scenarios (once I have written it), and revisit all my thinking here.
I also need to do some more thinking about kedging.
Thinking about surfaces and gaps (and salients) as they apply at a high level, say the level of characterizing the FEBA. I may be able to do something effective by determining whether segments of the FEBA are concave or convex.
]
In any event, I need to add code to pick the main effort division. I also need to add code to further characterize the divisional assignments. Of course the main effort division is probably going to have either attack or defend as its mission at any time. But the supporting divisions in the Corps need not stick to the basic mission. I can see the Corps using a supporting division to screen or defend a flank while the main effort division attacks. I imagine the cavalry brigade of an infantry Corps might be used for a spoiling attack if the Corps is defending.
I got the Imperial units in Assault of VII Corps to move last night. I had been moving Augereau and his aide while at the convention; last night I moved the divisions. Strange behavior: VII/2 moved out right away, while VII/1 did not move for the first few turns. They should both have moved immediately and continuously. Perhaps the line of advance limiting function is limiting based on the presence of friendly units as well as NME units.
[more thoughts...
I was wondering last night how to enforce the special rules of this game, from an AI perspective. The special rules require that the Imperial side advance for a while before it is allowed to retreat. The special rules are setup this way to establish the historical scenario. I think the way to do this is to change the Imperial victory conditions after the initial advance requirement has been satisfied.
]