[AI] performance cost of mutliple CAs

[white_haired_uncle]

I'm looking at creating a few units whose mission is to target a human-controlled leader. At the moment, using the simple_attack micro_ai. Each time I create one of these units, I create a CA for it, so I end up with something like this:

Code: Select all

        [candidate_action]
                enabled=yes
                engine="lua"
                id="mai_simple_attack_move"
                location="ai/micro_ais/cas/ca_simple_attack.lua"
                max_score=110050
                name="mai_simple_attack_move"
                score=0
                type=""
                [args]
                        ai_id="mai_simple_attack"
                        ca_score=110050
                        [filter_second]
                                id="Efraim"
                        [/filter_second]
                        [filter]
                                id="hates_Efraim-1"
                        [/filter]
                [/args]
                [filter_own]
                [/filter_own]
        [/candidate_action]
        [candidate_action]
                enabled=yes
                engine="lua"
                id="mai_simple_attack1_move"
                location="ai/micro_ais/cas/ca_simple_attack.lua"
                max_score=110050
                name="mai_simple_attack1_move"
                score=0
                type=""
                [args]
                        ai_id="mai_simple_attack1"
                        ca_score=110050
                        [filter_second]
                                id="Efraim"
                        [/filter_second]
                        [filter]
                                id="hates_Efraim-3"
                        [/filter]
                [/args]
                [filter_own]
                [/filter_own]
        [/candidate_action]

In this particular case, I doubt I have to worry about performance, but I want to learn. Is there a point when adding a CA for each unit would be a bad idea, when I could, with a bit of work if I want to do it right, instead create one CA that filters on "hates_Efraim-*" (assuming that is valid, if not I can probably come up with something else to filter on if I work at it)?

If I had 100 units, would there be a difference between having one CA to match them all and having 100 CAs each matching one unit?

TIA

[lhybrideur]

I now nothing about ai, but could this work with a role?
like role="hates-Efraim"
Then you would give this role to all the units you want to hate Efraim