Bug Bug Thread: Discussion - Page 248

thejeff · #**2471** January 18th, 2009, 11:16 AM

Quote:

Originally Posted by AreaOfEffect

If it is something a player is allowed to do, then it is valid for the AI to do it as well and is not a bug. Can a player equip a bow and script the unit to charge into melee? Yes. Is that a bug? No. Just remember that the AI is just suppose to simulate the actions of a player. Its not a bug if that player isn't especially bright, or decides to play the game in ways you yourself would not.

So you wouldn't consider it a bug for the AI to just do nothing? Sit in the starting castle and never recruit, research or try to expand. After all, a human player would be allowed to do that. Or maybe it could destroy it's own starting castle, lab & temple. Why not? A human player could.

The AI has enough trouble competing. It's worth pointing out especially egregious examples. Call it a design flaw rather than a bug, if you want. It's still an error of some kind.

Gandalf Parker · #**2472** January 18th, 2009, 01:03 PM

Interesting. Thanks for that.
In some of the most heated arguments Im not sure that the sequence of casting was considered. Some of the old "works for me / doesnt work for me" battles on the forum might be worth reconsidering.

Edi · #**2473** January 18th, 2009, 04:50 PM

Quote:

Originally Posted by thejeff

Quote:

Originally Posted by AreaOfEffect

If it is something a player is allowed to do, then it is valid for the AI to do it as well and is not a bug. Can a player equip a bow and script the unit to charge into melee? Yes. Is that a bug? No. Just remember that the AI is just suppose to simulate the actions of a player. Its not a bug if that player isn't especially bright, or decides to play the game in ways you yourself would not.

So you wouldn't consider it a bug for the AI to just do nothing? Sit in the starting castle and never recruit, research or try to expand. After all, a human player would be allowed to do that. Or maybe it could destroy it's own starting castle, lab & temple. Why not? A human player could.

The AI has enough trouble competing. It's worth pointing out especially egregious examples. Call it a design flaw rather than a bug, if you want. It's still an error of some kind.

The only way to fix that is to alter the AI decision trees and related algorithms to take this case into account. The real question is whether doing so will introduce more complications that make it more trouble than its worth to fix than leaving it as it is currently. Many of these more obscure issues are that way, even ones that are real bugs, such as the one cleveland reported by PM.

And believe me when I say you people don't want to get hit with the Stick of Unintended Consequences on stuff like this. With one of the recent patches, a batch of problems that had been around for a long while were fixed, but they produced a side effect that was a much greater headache and had to be fixed separately before the patch was released. If that patch had gone out in its initial form, it would have stopped every MP game in progress in its tracks and ruined many of them.

So if it's something as trivial as this particular scripting issue, think VERY hard before turning a molehill to a real hill. You might just get a Precision 100 Gifts from Heaven as a complimentary bonus if it's tackled.

AreaOfEffect · #**2474** January 18th, 2009, 04:54 PM

Quote:

Originally Posted by thejeff

So you wouldn't consider it a bug for the AI to just do nothing? Sit in the starting castle and never recruit, research or try to expand. After all, a human player would be allowed to do that. Or maybe it could destroy it's own starting castle, lab & temple. Why not? A human player could.

The AI has enough trouble competing. It's worth pointing out especially egregious examples. Call it a design flaw rather than a bug, if you want. It's still an error of some kind.

My statement assumes that the AI is capable of making judgments. If it were a case that the AI wasn't capable of action, or if it wasn't allowed to weigh it's options, I would say that would be a bug. This example clearly shows that the AI at least doesn't set all of its commanders to "set battle orders". If the AI is able to make decisions and doesn't do what you would do, that is not a bug.

In fact, I'm not sure I would fire off my bow of war if it were me. It might be 13 arrow shots, but it would be 13 short bow arrows, which frankly isn't that exciting. Compare that to the an 18 strength armor piercing life drain attack on an ethereal unit who flies and has fear. You can very likely deal no damage with the bow, but the melee attack is more then likely to deal damage, net bonus HP, and your looming presence is guaranteed to deal moral damage to 15 squares around you.

thejeff · #**2475** January 18th, 2009, 05:15 PM

Freely granted that this issue is fairly trivial and quite probably not worth fixing, even without worrying about unintended consequences.

I was reacting far more to the statements that:
If it is something a player is allowed to do, then it is valid for the AI to do it as well and is not a bug.

and

It works completely by the rules of the mechanics of the game as they are supposed to work. If scripted to attack, then it will attack. Therefore it is NOT a bug.

Both of which seem to not allow for the possibility of bugs in the AI logic as long as that logic doesn't permit impossible orders.

In focusing on those responses, not the original situation, I was also ignoring the possibility that attacking was at least a reasonable option despite the results as AoE suggested. That makes it even less likely to be an issue.

vfb · #**2476** January 18th, 2009, 06:57 PM

I see this entry in the short list:

Quote:

# BHV SPELL Globals Global enchantments of nations that have vanished stay in effect. [Edi's note: Presumably in cases where nation is defeated without the casting unit explicitly being killed (e.g. dominion death)]

I think this bug may be different:

In the World in Crisis MP game, TC had an immortal unit cast Arcane Nexus. That unit was killed in melee combat, in positive dominion. But TC's capitol was owned by another nation, so the immortal unit was permanently killed. However, the Arcane Nexus is still up.

Edi · #**2477** January 19th, 2009, 02:29 AM

That looks like a variant of the same bug. Immortality has all kinds of triggers associated with it and it probably just checks that when the unit is killed on the battlefield that immortality is ok, global stays up, but doesn't check whether the capital is in friendly hands.

In both cases the common factor is that the caster of the global is eliminated from the game without being explicitly killed on the battlefield or the global dispelled via overwriting or dispel. It becomes a case of "The Void ate my caster, but my global is still up! Nyanyannyan-naa-naa!"

chrispedersen · #**2478** January 19th, 2009, 02:55 PM

Yep, my low astral pretender has entered that strange little house more than once to avoid assassins, death.

Nothing like lost in space and time to keep a global up.

Edi · #**2479** January 19th, 2009, 04:35 PM

Quote:

Originally Posted by chrispedersen

Yep, my low astral pretender has entered that strange little house more than once to avoid assassins, death.

Nothing like lost in space and time to keep a global up.

Hmm, I could have said that better. The units of a defeated nation that disappear sort of vanish and can't come back until their ID numbers are recycled, at which point a new unit would have the "This unit has cast a global enchantment" icon and the global stays up because that unit had never "died", even though the nation vanished. That's a lot different from "Lost in Time & Space", but the effect is similar in practice.

chrispedersen · #**2480** January 20th, 2009, 12:31 AM

Three times now, I have gotten EA-Ctis to Dominion 0 (verified by scores, graphs, etc) - and the god will not die.

He finally dies when I take his citadel. This despite the fact that I have 2 candles on his citadel at the time I take it.