zolobolo wrote:Seems like it is ignoring the retreat order as well as the 3X power disadvantage

Whoops, looks like I commented out something for testing and then forgot to uncomment it. Will fix that right away.

Edit: now fixed in e0c857f

harpy eagle wrote:Edit: now fixed in e0c857f

I've folded this mod into the "in_development" version of vanilla as well: all the content in e0c857f35 should now also be in r22260.

This is a lovely mod, and I've made no changes to it as yet; though watching various combat replays did inspire me to nerf the range of anti-missiles.

sven wrote: I've folded this mod into the "in_development" version of vanilla as well: all the content in e0c857f35 should now also be in r22260.

Great news!

Suggestion: remove all bonuses the AI receives on Normal difficulty. The AI will be able to send beginners running for the hills without it, and there should be a mode where all parties start on equal footing

Question:
Is the vanilla code conflicting with the strategics and tactical mod files or when using the mods they simply do not have any practical effect anymore?
Where would you prefer having future bugs reported, in the main issues thread or in the corresponding strategic and tactical AI mod threads?

@harpy eagle, congratulations on your code now being cannon

zolobolo wrote: Is the vanilla code conflicting with the strategics and tactical mod files or when using the mods they simply do not have any practical effect anymore?

Well, all these mods will still work as intended with the main version of the game that's live on steam, r22151. It's only the in_development BETA branch that has any of these mods integrated.

The strategic code in r22260 has become a little different than the latest version harpy eagle posted -- so trying to mix that mod with r22260 may now have weird side effects.

The research paths code is sortof branched -- in that there's currently a few updates to the path definitions harpy eagle has posted that I've yet to review and fold back into vanilla -- so that's a bit of a mess.

The tactical AI code should currently be at the same place in both this thread and r22260, so applying the mod to r22260 should have no practical effect.

zolobolo wrote: Where would you prefer having future bugs reported, in the main issues thread or in the corresponding strategic and tactical AI mod threads?

In principle, if the code is in vanilla, posting a bug report to the main bug report thread is probably most appropriate. That said, if you want to give detailed feedback on suboptimal strategies you've noticed the AI using (which is arguably not a 'bug', persay), sticking it in the related mod thread might not be a bad idea. While not every suboptimality can be fixed, I suspect that both harpy eagle and I will be interested in reading feedback on how the emergent behaviors might continue to be improved.

zolobolo wrote:Question:
Is the vanilla code conflicting with the strategics and tactical mod files or when using the mods they simply do not have any practical effect anymore?
Where would you prefer having future bugs reported, in the main issues thread or in the corresponding strategic and tactical AI mod threads?

Sven pretty much covered it for all the existing versions. If I do release any new updates or tweaks they will contain only the changes from the versions of the mod now in vanilla, and I'll probably label them as "v2.0" or something to be clear.

sven wrote:This is a lovely mod, and I've made no changes to it as yet; though watching various combat replays did inspire me to nerf the range of anti-missiles.

Yeah, I've found trackers tend to get obliterated pretty quickly, even without anti-missiles. I don't think it's a problem for missiles because ships can just relaunch waves after waves of them and usually enough get through to be balanced, I think. But I'm a little worried about fightercraft.

The max_sortie_dist calculation in CloseAndAttack.lua is actually purposefully larger than what it really should be, since a "smarter" calculation meant that carriers almost never launched their fighters unless at point blank range. While better for the AI, I thought it was kind of unfun since you never got to see things like tracker-tracker intercepts that I thought were pretty cool to watch in the earlier versions of the mod. But It does mean that fleets with carriers sometimes lose when they would have won, otherwise.

Privately I've been experimenting with increasing evasion just for fighters (I guess since they're manned there's some kind of explanation for that) and or reducing the PD/accurate weapon mod bonuses to +25%. But there could be better ways to approach it and that's probably a separate discussion.

Wouldn't worry about rockets and torpedoes: PD and fighters need to be hard counter to them otherwise, having rockets would be the only viable strategy. It makes sense that they are basically useless at long range. The ships either need to get close to the enemy or other fleet elements need to take out the PD escorts - perfect balance

Like the idea of fighters having good evasion - maybe even bombers to some extent
This would enable their losses to be kept at bay and thus they can be more freely utilized: the AI can be conservative in using them without being boring for players to watch/fight against.
Don't think there is any other way around this besides increasing flat survivability of the crafts (squadron size, HP, armor, giving them shields or evasion) or maybe if you enable regeneration of small crafts like 1 fighter regen /Turn, as this would also encourage a liberal (but still not careless) use of the squadrons).

Also this reminded me of something: observe the small turret on the Human bomber: would it be possible to have the bombers shoot PD beams against incoming anti missiles and fighters? This would at least increase bomber survivability against long range hard counters (not for direct fire PD) though this cool turret is only present for human fighters as far as I know - maybe a racial trait for them?

harpy eagle wrote: Privately I've been experimenting with increasing evasion just for fighters (I guess since they're manned there's some kind of explanation for that) and or reducing the PD/accurate weapon mod bonuses to +25%. But there could be better ways to approach it and that's probably a separate discussion.

One relatively minor diff that's been on my TODO for *years* is to modify fighters so that there's a chance that rather than being shot down by point defense, they'll just be forced to abort their attack run. I.e., in apply_generic_damage_to_tracker, there'd be a chance that instead of dying, a fighter would just lose it's right to attack that turn.

Um, there are currently 2 reasons why this isn't done yet. The first (and biggest) being that figuring out how to animate the 'dodge' mechanic in such a way that a player has reasonably odds of understanding what's supposed to be going on is hard. The second is that the fighter damage attack code is ancient, and a mess, and I think all the conceptual stuff underlying the 'attacks_this_turn' variable should likely be re-throught and re-written.

sven wrote:One relatively minor diff that's been on my TODO for *years* is to modify fighters so that there's a chance that rather than being shot down by point defense, they'll just be forced to abort their attack run. I.e., in apply_generic_damage_to_tracker, there'd be a chance that instead of dying, a fighter would just lose it's right to attack that turn.

Um, there are currently 2 reasons why this isn't done yet. The first (and biggest) being that figuring out how to animate the 'dodge' mechanic in such a way that a player has reasonably odds of understanding what's supposed to be going on is hard. The second is that the fighter damage attack code is ancient, and a mess, and I think all the conceptual stuff underlying the 'attacks_this_turn' variable should likely be re-throught and re-written.

That's a cool idea. Personally, I kind of think it might be enough just to put out floating text saying "Evade!" and explain somewhere that evading fighters can't attack that turn.

I'm assuming that the evade status doesn't carry over for fighters in flight over multiple turns. It might be interesting if the situation happened where all (surviving) fighters in a formation are forced to evade in a single turn, the entire formation is forced to break off and return to base. It might also help reinforce to players that taking more evasions can reduce the effectiveness of a fighter attack.

Anyways, on a completely unrelated note, I'm now wondering if I should have moved all the new combat AI source files to a subfolder inside AI...

harpy eagle wrote:On a completely unrelated note, I'm now wondering if I should have moved all the new combat AI source files to a subfolder inside AI...

Heh. Yeah, probably. I've moved them in AI/tactical_ai as of r37498.