34 replies to this topic

[ArkhanTheBlack]

Okay, I've sent the first build of 3.1 to Thudo.

Changes:

- Updated to SS 1.10 patch
- Added Dark40k's CpuPrerequisite system to allow for Building Addons to now work properly with the AI
- Personalized all vanilla factions' tactics so each unit now has its own abilities' file
- Tactic: IsMoving() using new tweak by Larkin/Dreadnott
- Using new FoF code for Eldar's infantrytactic-- by Dreadnott
- Some general cleanup from here and there
- Fixed a bug with teleporting builders moving back to base after teleporting to a building
- Melee units get more time to end their attack
- Take & Hold should work more reliable since defense will be stopped now if they try to counter a victory counter


Know issues:

- Teleportation of units like WarpSpiders and Necron Lord -- they seem to not teleport around as much as they used to although this has to be substantiated.
=> Since compass move teleports got unstable in SS, this is the logical consequence! Can't be fixed at the moment!

- AI is a little slow capping outer points and sometimes forgets them entirely after early battles
=> I don't want to experiment here. Those problems have a lot to do with intentional careful capturing because of pathing problems (Moving through enemy base and/or army etc.)

- Melee units seem to be the biggest problem; they're easily confused. for example, with orks, their melee units work well when attached to the big mek. he teleports around causing havoc, etc. on their own BUT the individual squads tend to run away quickly if shot by a ranged squad. sometimes they run back and forth between 2 ranged squads never entering melee at all. something seems off here.
=> The increased 'play' time for melee units should reduce the risk of back and forth moves but can also result in some increased Kamikaze behaviours sometimes. There's no perfect solution here. You can only exchange solutions with each having their own advantages and disadvantages. Therefore, the current strategy is 'Better die fighting than die moving'.

[Inquisitor]

Good news, Arkhan! I will be testing it when thudo gives the green light. =)

[thudo]

Running some tests on Arkhan's submission then its off to the testers.

If you do not get an email within 24hrs, please PM or email me.

[thudo]

Notices have now been sent out -- go get em lads!

[Zenoth]

Alright, so far I've been focusing on the Orks' melee units and the Eldar's Fleet of Foot tactics (On and Off during battles).

But I'd like to understand how the FoF tactics work. I would presume that it has to turn Off while idle to increase the long-range accuracy (for ranged units of course) while being On while units are on the move. And from what I've observed so far it seems to work that way, indeed, but I think it may be a little slow in applying and in reaction to situations. I've observed the intended result, and I've seen FoF turning On and Off randomly while units are idle and on the move. I'd especially like to know under which conditions does the tactics apply if a unit switches between ranged to melee (and the other way around too). For instance is the Farseer remains ranged, and then enters melee, how does the code directs the tactics? Is the code not taking melee into consideration?

It doesn't happen often, to see it turning On and Off like that for no apparent reasons, mind you, but I've seen it. Maybe if I understand more how it works then I will be able to test it more properly and have a better judgment on it. At this point I'd say that if the intended effect is to indeed turn it Off while infantry is idle and turning it On while infantry is moving (with both ranged and melee) then it may not always be working, even if it does work most of the time. Maybe that the On and Off "randomization" is intended in the code itself, I don't know. When it works it's absolutely impressive, let me say, it has saved many lives, but when an infantry squad leaves if Off and moves away because it's being shot at then I'm sure that it's not supposed to be so.

Strangely the FoF On and Off tactics seem to work more with the Howling Banshees than with Warp Spiders for example (I.E melee Vs. ranged efficiency)... or maybe I'm imagining things. I still have to make sure that my claims are exact and that it wasn't just coincidence, but so far I've tested on Hard difficulty, with and without Heroes, on official maps, in 3 Vs. 3's and 4 Vs. 4's (in a total of four skirmishes) and from what I've seen then I must say that when it does work it's really great, but when it doesn't (or at least it seems like it right now) then it's just leaving the infantry without any real advantages of course. And three of the skirmishes were A.I Vs. A.I ones.

As for the Orks engaging in melee, it certainly works. I did notice that sometimes (it's not occurring often at all, but it can be observed in all skirmishes I've done) one or two squads will move back and forth between two threats without engaging, but to be honest even though we're talking about A.I here I myself wouldn't have known the priority under those situations. If a squad is being shot at on the left by the base's defenses they tried to assault and then being shot at from the right by incoming enemy units then what should we expect them to do exactly? Focus on the most damaged structures? Probably not because the A.I cannot target specific objects from what I can remember, nor even target specific units, but they can decide to run away because they're dying or in danger, right?

The problem itself is perhaps the back and forth movement, more so than the fact that they don't engage in melee. By that I mean that with the example I gave above, if the "confused" squad going back and forth could just go in one direction instead, for instance just retreating back to base or the nearest LP/CP then I believe that it would end our concerns right there. Indeed I think that the issue here is that they're doing a back and forth move, which is of course a danger because the enemy squads/vehicles will attack them, and the poor confused squad itself will never do anything, not even trying to weaken the attacking units, due to the back and forth confusion. But the fact that they don't actually engage in melee at that moment isn't that much of a problem in my book. If they're being shot at from two angles then the best they should be doing is to retreat in just one direction and never go back, that way they don't die needlessly just for the principle of engaging melee in a fight that would have been lost anyway.

So I propose to look at the "not engaging due to going back and forth" issue with a new perspective, one to allow them not to engage in melee and to just retreat while (if possible of course) fixing the back and forth movement. If you guys think differently and I would certainly respect that, and if you think that they absolutely have to engage in melee as well as fixing the back and forth movement issue then so shall it be. I just want to make it clear here that the back and forth issue is not occurring that often, at least not from what I've seen myself. I hope that it wasn't just luck.

I do notice however that the Orks tend to die often, by that I mean not just one squad of Orks but the entire team, and honestly I always thought that the Orks would last longer than the Eldar especially when facing Necrons (which is what I've done to allow the testing of the Necron Lord teleportation), but alas none of the skirmishes I've set resulted in a victory by the Eldar & Orks team. The thing is the Eldar put up a very nice fight, and the Orks always suffer from a serious lack of Requisition, and I'm still unsure of the cause of that, maybe it's due to the training of too many unique squads, or maybe too much researching early on, but in the end the Orks always accumulate tons upon tons of Power and cry for more Requisition, but I'm aware that the map itself and the number of LP's/CP's have to play a major role for that aspect of the A.I, of course.

Alright, I'll do some more extensive testing now.

[thudo]

- The Eldar FoF code was a collaboration with DreadNott back last April so it should be highly optimized BUT due to whatever reason can never ever be perfect.
- Engaging in Melee and the infamous Back-Forth movement: yeah that is a hard nut to crack.
- Orks running out of Req? I thought they are normally starved for power. I find the Orks got a boost in Soulstorm and generally fear them.

[Inquisitor]

I never got a pm or a message. Am I supposed to test?

What about patch 1.2 which is out now? Will you release a new version or do you want us to test this on 1.1?

- Orks running out of Req? I thought they are normally starved for power. I find the Orks got a boost in Soulstorm and generally fear them.

So, maybe look into it?

[thudo]

PM me your email, Inquisitor -- I sent to everyone who was a testers before but I gather your email may have changed?

Yes its stated the current 3.10 beta1 is meant for Soulstorm 1.10 + Hotfix.

As for Orks running out of Req -- thats still to be determined as this observations wasn't apparently a factor with the our 3.00 AI. Unsure why this would be now since the AI scripts for Orks are tight re: ensuring teching is curbed based on present ArmyStrength. More Ork battlefield analysis needs to happen.

[RobertW7928]

I'd just like to inquire about the Campaign any complications?, how doe's hero's work with the war gear, doe's it adjust accordingly? and i recollect that the AI/hero's feature was failing to load for the strongholds, if it is at all possible to enable them? I'm also curious if all units are functioning correctly with hero leveling in v3.1? and for a personal preference would you enable Massive battle for the Campaign? the decision is yours.

Keep up the Great work.

[Zenoth]

Alright, so I've done more extensive testing, and here are my observations and reports on the following points (taken by the "to do" list from Thudo's e-mail for this build):

o Dark Eldar Hellion Homing Device CTD

I confirm CTD's with that. I've experienced three so far, directly crashing to the Desktop without any messages. It occurred once during the Dark Eldar Campaign (which are basically skirmishes anyway), and two other times during skirmishes, one of which was a all Dark Eldars eight players deathmatch. The Homing Device is deployed regularly, but it has to be causing CTD's every now and then.

I would say that the CTD is still rare. I mean I've experienced it three times... but that's three times out of around twenty A.I Vs A.I skirmishes I've been doing since the past 48 hours. But the rarity of it isn't going to let it go into the final build. We can certainly live without the Homing Device, and the Dark Eldar won't be crippled by its absence at all, it hardly makes any differences.

If other testers can also confirm crashes, if it does occur on with them, then it's up to you guys, Thudo/Arkhan to see if there's anything else to try to leave that thing in. If not then let's disable its usage, no one will want even the rarest of the CTD's for such a mere addition. If it was indeed working, yes of course, let it in, but as of now it has to be still buggy, even if it doesn't happen often.

o Dark Eldar Wyche's Combat Drugs idleness when ability is used

I've yet to notice them being idle when the ability is used. I don't have much to elaborate on it really, since as I said, I simply haven't noticed any idleness under that condition, at least not so far. Maybe I will notice it if I focus on that longer, but I did just that already in a few skirmishes, and replays, but nothing to say about it, at least not right now, it's certainly a good sign to say the least.

o Necron/WarpSpider teleportation (apparently a bug introduced in
Soulstorm which we cannot control)

Both the Necron Lord and the Warp Spiders, along with regular Necron units all teleport. Now obviously they don't always teleport, but they seem to do it when necessary. I haven't seen the Necron Lord teleporting just because it's fun, but because it was reacting to a structure being under attack (probably) or to get out of danger. Was the bug causing any confusion about the teleportation? What is the bug actually causing Thudo/Arkhan? Care to elaborate more on what I should be focusing on?

If the bug causes those units not to teleport at all then it's definitely gone and it's now working. If the bug forces those units to sometimes teleporting in seemingly "random" manners without any apparent reasons then I've yet to see it happen. Again, I think I'll need to focus on it a little more, and I'll be able to comment more precisely and in better details.

o Take&Hold aggressiveness -- ensure AI goes on the offense to take
each point away from the opponent without compromise.

That definitely works. There are of course situations under which the LP/CP/Relic capture process has to be aborted due to nearby enemy presence, but when the capturing unit(s) have their surroundings controlled and without any enemies then there's no compromise.

But there is one thing I've noticed with this build about capturing that I haven't before, and curiously it occurs only when capturing Critical Points, I have yet to see it happen for Relics or LP's. The "issue" if I can call it that way is that whenever a squad seem to decide it wants to capture a Critical Point then it looks like the whole infantry in the team receives the order and goes there too.

Indeed, I've seen gathering of infantry units around CP's very often, on many different (official) maps, it applies to all nine factions. Basically what happens is that one or two squads get to a CP, whether it's early on or late in a game, it doesn't matter. So, one or two squads get there, one of them stays idle, and the other captures it. Then a couple of seconds later the next thing you know is the whole infantry family is gathered there talking about their favorite hobbies during the week-ends.

But yes, seriously, I've noticed units gathering around CP's under this build that I hadn't noticed before. And I do know that it's not a new "issue", we've all seen and discussed it many times before and we can trace such discussions all the way back to the Winter Assault days when I first started testing. If it was fixed before then maybe this time around you guys will need to take a look at it. And more precisely, by "gathering" I do mean like four or five infantry squads around the same CP, and one or two of them are sometimes even moving back and forth a little just around the point itself, as if they were confused...

Squad #1: "Isn't it grand, we're capturing a CP!"
Squad #2: "Hmmm, hey sup guys, we'll stay here to guard you, mmk?"
Squad #1: "Yeah sure, thanks."
Squad #3: "...what the... hey we were sent to capture this point guys!"
Squad #1: "Suck it down man, this point is ours."
Squad #2: "Yeah guys, leave 'em alone and go kill those green skins instead."
Squad #3: "No way, we'll stay here anyway, it was our order."
Squad #4: "Hey guys, look, it's break time! They're almost all here. Let's have a stop here."
Squad #5: "Hey guys we've just received the order to capt..."
Squad #1/#2/#3: "...oh not this again..."
Squad #4/#5: *simultaneously looking between themselves, confused*

Ahh yes, good ol' times.

o Watch for Melee units like Ork StormBoyz jumping in and engaging in
CC effectively -- such units will now stay engaged to the death rather
than run away

If the goal with that was indeed to let them engage to the death, then yes, it works. There's not much else to say here. They properly engage in CC, it's done effectively, and they indeed often die in the process since they try holding their ground the best they can, often until death. I will try to spot perhaps one or two weakness/malfunction about that, but I'm pretty sure that's done properly.

o Eldar Bonesinger teleporting to remote buildings to begin
construction -- should be fixed as I've seen Eldar AI use builders
properly now.

Confirmed fixed. I've seen them not completing buildings before, indeed, but now, not anymore. At least not so far, but it does seem entirely fixed, I haven't noticed an incomplete building once in since I started testing this build. Good thing.

o Eldar Fleet of Foot On and Off tactics -- a gem of testing and
scripting from Dreadnott, one of our gracious DoW AI testers, so
enjoy!

As mentioned in my first post, it does work most of the time, and when it does it's very impressive to look at, it's effective and it does help the Eldar getting kills more often and getting out of danger simply faster and saving lives in the process, it's a good thing to have. And as pointed by Thudo, it cannot be perfect anyway, so seeing it sometimes turning On and Off for no apparent reasons, and that if it happens very rarely as I've seen it, then no big deal, it's so rare as to be a very negligible point to even consider looking for.

o Anything else that comes to mind!

What is the status on the Haemonculus' squad attachment capability? I'm still seeing him wandering around alone, always, I haven't seen him attached at all so far. Was it established that it cannot be done due to coding restrictions? Maybe I wasn't aware of it if it is indeed the case. If not then it's certainly something to look for, to make it work simply, because right now, at least in my tests, that guy is a solitaire, and not very useful alone.

I'd like to conclude this report about the Campaign. Basically the Campaign works without problems expect of course if playing with and/or against the Dark Eldar, which will potentially result in a CTD due to the Homing Device bug. But other than that it works great. Also the Fortress, Massive Battles and Heroes extensions work in Campaign. And finally, I've also tested this build both in skirmishes and in the Campaign while being merged with the Bugfix Mod 1.33 and it works like a charm. I can see that there won't be many builds before the final version, since as mentioned before nothing much was needed to make this mod compatible with the first patch.

So, I'll keep on testing and come back with another report.

Edited by Zenoth, 05 November 2008 - 07:24 AM.

[thudo]

o Dark Eldar Hellion Homing Device CTD
I confirm CTD's with that.

Yeah crap. I know if you have a 3vs3 All Dark Eldar game it will likely CTD during the game sometime. I think it best we disable this in the DE's loader.ai, Arkhan. I do not want CTDs to be remotely a factor even though I do like the Hellion's ability.

o Dark Eldar Wyche's Combat Drugs idleness when ability is used
I've yet to notice them being idle when the ability is used.

Thats good news -- hopefully other testers will report the same.

o Necron/WarpSpider teleportation (apparently a bug introduced in
Soulstorm which we cannot control)
Both the Necron Lord and the Warp Spiders, along with regular Necron units all teleport.

Yes this still needs to be observed. Just curious if the testers noticing such units teleporting more or less the same way as in the past. We believe its a little less due to a yet-unknown Soulstorm bug. However, it is NOT a show stopper.

o Take&Hold aggressiveness -- ensure AI goes on the offense to take
each point away from the opponent without compromise.
The "issue" if I can call it that way is that whenever a squad seem to decide it wants to capture a Critical Point then it looks like the whole infantry in the team receives the order and goes there too.

Hmm.. I never really noticed that per se except late game AND not only limited to Critical Points. Its a rare occurrence. I have noticed sometimes near later Tiers the AI will congregate his entire army on one side of the map and wait and wait wait.. but again it doesn't happen often.

What is the status on the Haemonculus' squad attachment capability? I'm still seeing him wandering around alone, always, I haven't seen him attached at all so far. Was it established that it cannot be done due to coding restrictions?

I think we need to re-evaluate the Haemonculus attachment code in its tactic file -- thats the problem and its quite likely super-easy to fix. We'll take a look at it.

I'd like to conclude this report about the Campaign. Basically the Campaign works without problems expect of course if playing with and/or against the Dark Eldar, which will potentially result in a CTD due to the Homing Device bug. But other than that it works great.

Wow fantastic -- I was a little concerned about the Campaign integration -- we really need to ensure it works well especially with the game modes. This is good news so far!

[Inquisitor]

AI Version: 3.1b1
Matches: 2v2 AI
Factions: Necron and SM vs Sisters and DE
Game Modes Used: Annihilate
Map: Biffys Peril
AI Difficulty: Hard
Resources Rate: Normal
Resources Sharing: On
Starting Location: Random
Winner: Necron and SM

Comments: First mp game with 3.1. Played via Hamachi.

Observations of interest:

• Great game. Been a long time since we played, so we were a bit rusty.
• No problems.
• AI played a nice game where it assisted each other when we where pressing one of them too hard.

Suggestions:
I haven't had time to review the replay yet, but it was a hard fight.

Replay: Available upon request.

++ Bless the Machine god for he has given us tanks in plenty ++

AI Version: 3.1b1
Matches: 1+1AI vs 1+1AI
Factions: Necron and DE vs SM and Sisters
Game Modes Used: Annihilate
Map: Biffys Peril
AI Difficulty: Hard
Resources Rate: Normal
Resources Sharing: Off
Starting Location: Random
Winner: Necron and DE (i.e. me)

Comments: Second mp game with 3.1. Played via Hamachi.

Observations of interest:

• Great game. Our second game in a long time. It was great fun and we might play again tomorrow.
• No problems.
• Necrons had a tough battle vs SM but gained the upperhand by good ol' NW spamming.
• AI helped building SM base in their own camp when SM base was overrun.
• Both AI's played a nice game where they assisted their ally, when we were being attacked (almost) in our base.

Suggestions:
I haven't had time to review the replay yet, but it was a great game and confidence in the Skirmish AI capabilities remain in 3.1. Good job!

Replay: Available upon request.

++ Cease and repent. ++

[thudo]

Hmmm.. glowing review but problem is the opposing AI team didn't win. :( Crush the weak fleshling damn it!

[Inquisitor]

Hmmm.. glowing review but problem is the opposing AI team didn't win. :( Crush the weak fleshling damn it!

Hmm, it was on the winning side in the second game, but I guess I did the hard work...

[ArkhanTheBlack]

Orks always suffer from a serious lack of Requisition

That's usually map dependant. In addition Eldar can use a lot more available power than orcs.

o Dark Eldar Hellion Homing Device CTD

Yeah, we'll deactivate that again.

Was the bug causing any confusion about the teleportation? What is the bug actually causing Thudo/Arkhan? Care to elaborate more on what I should be focusing on?

Before SS, it was possible to just let a unit jump x meters in any direction by just checking the terrain position. However, in SS the game can crash by just checking a terrain position. Therefore we now have to search for secure positions like LPs, other units, etc. Those positions are always 'valid' and the game never crashes. But the problem is now, that if none of those save positions is in range the unit can't jump anymore because we don't have a valid position available.

But there is one thing I've noticed with this build about capturing that I haven't before, and curiously it occurs only when capturing Critical Points, I have yet to see it happen for Relics or LP's. The "issue" if I can call it that way is that whenever a squad seem to decide it wants to capture a Critical Point then it looks like the whole infantry in the team receives the order and goes there too.

Well, there is a reason for this behaviour. The Take and Hold behaviour was critised for not always working. According to the code, this could only happen when the AI was defending. If the critical counter runs down to about 3 mins, the AI creates an attack plan and tries to capture the closest critical point which belongs to the enemy. The problem occurs if the AI is defending. In this case, the units are assigned for defending the base and the attack plan for capturing an enemy LP doesn't get any units. I've changed that now that all defend plans are stopped if the AI tries to stop a victory counter. However, this is only the emergency counter if the timer is critical. If there's still enough time, the AI will just send a free capturing squad.

Basically the Campaign works without problems expect of course if playing with and/or against the Dark Eldar, which will potentially result in a CTD due to the Homing Device bug.

Perfect!


Very well, good testing so far Zenoth and Inquisitor.

Edited by ArkhanTheBlack, 05 November 2008 - 11:44 PM.

[dreddnott]

Hey, Zenoth, I'm still around, sorta, just lurking a bit. You've put quite a bit of thought in your posts...

Alright, so far I've been focusing on the Orks' melee units and the Eldar's Fleet of Foot tactics (On and Off during battles).

But I'd like to understand how the FoF tactics work. I would presume that it has to turn Off while idle to increase the long-range accuracy (for ranged units of course) while being On while units are on the move. And from what I've observed so far it seems to work that way, indeed, but I think it may be a little slow in applying and in reaction to situations. I've observed the intended result, and I've seen FoF turning On and Off randomly while units are idle and on the move. I'd especially like to know under which conditions does the tactics apply if a unit switches between ranged to melee (and the other way around too). For instance is the Farseer remains ranged, and then enters melee, how does the code directs the tactics? Is the code not taking melee into consideration?

As far as the timing, the issue is how often my Fleet of Foot code loop in eldarinfantrytactic.ai gets executed (not very often). From the conversations I had at the time with the crew, there was no real way around this.
Also, the code doesn't take melee into consideration, but is hard-coded for each unit type. Default-melee units like Banshees, Seer Council, and Farseer turn FoF off when they are within 15 units of an enemy, even when moving, while ranged units will turn it off at a longer range (a default of 35, shorter for Fire Dragons and longer for Dark Reapers). This is done in order to compensate for the time delays - Eldar ranged infantry performing an attack move will turn FoF off just before they stop moving and start shooting, in an ideal situation. This code does not execute when infantry are making normal moves or retreat moves, of course - otherwise they'd never escape from nearby enemies.

It doesn't happen often, to see it turning On and Off like that for no apparent reasons, mind you, but I've seen it. Maybe if I understand more how it works then I will be able to test it more properly and have a better judgment on it. At this point I'd say that if the intended effect is to indeed turn it Off while infantry is idle and turning it On while infantry is moving (with both ranged and melee) then it may not always be working, even if it does work most of the time. Maybe that the On and Off "randomization" is intended in the code itself, I don't know. When it works it's absolutely impressive, let me say, it has saved many lives, but when an infantry squad leaves if Off and moves away because it's being shot at then I'm sure that it's not supposed to be so.

If you've noticed FoF cycling rapidly on stationary units, try and describe what was going on - was a member of the squad stuck somewhere, were they capping a point, etc. Screencap might even help. When I fired up the new beta the other day I ran a couple Eldar matches to make sure my code still worked and it seemed to be all right. I had the cycling problem a while back but I thought I licked it.

Strangely the FoF On and Off tactics seem to work more with the Howling Banshees than with Warp Spiders for example (I.E melee Vs. ranged efficiency)... or maybe I'm imagining things. I still have to make sure that my claims are exact and that it wasn't just coincidence, but so far I've tested on Hard difficulty, with and without Heroes, on official maps, in 3 Vs. 3's and 4 Vs. 4's (in a total of four skirmishes) and from what I've seen then I must say that when it does work it's really great, but when it doesn't (or at least it seems like it right now) then it's just leaving the infantry without any real advantages of course. And three of the skirmishes were A.I Vs. A.I ones.

Warp Spiders? They don't have FoF, perhaps you mean Dark Reapers...

[Zenoth]

Alright, let's start by answering things by quotes.

Me: Orks always suffer from a serious lack of Requisition

Arkhan: That's usually map dependent. In addition Eldar can use a lot more available power than orcs.

Indeed, as I suspected, it's map dependent. And according to more recent tests I've observed the same thing with other factions, not just Orks, so I guess that there's nothing to worry about here, case closed.

Me: Was the bug causing any confusion about the teleportation? What is the bug actually causing Thudo/Arkhan? Care to elaborate more on what I should be focusing on?

Arkhan: Before SS, it was possible to just let a unit jump x meters in any direction by just checking the terrain position. However, in SS the game can crash by just checking a terrain position. Therefore we now have to search for secure positions like LPs, other units, etc. Those positions are always 'valid' and the game never crashes. But the problem is now, that if none of those save positions is in range the unit can't jump anymore because we don't have a valid position available.

Ah, I see, now it's quite clear. And it explains why I indeed often see teleportations occurring near or literally at LP's and very close to friendly units. Now that you mention how it works I can confirm that I don't think I've ever seen a Necron Lord (for example) teleporting just in the middle of "nowhere", that could crash the game in SS, according to what I now understand. Those safe locations such as LP's and groups of units is in effect exactly where I see the teleportations happen all the time. Thanks for elaborating on this Arkhan.

So I guess that it works the way it can in Soulstorm. It's probably not much of a problem either. I mean it affects only units with teleportation capabilities, and the actual reason (not the destination) of the teleportation is working as intended so far, I'd say so at least from what I've seen. Even if various infantry squads and the Necron Lord doesn't teleport in the middle of nowhere but still teleports "somewhere safe" for a good reason, such as because he's in danger or because he needs to catch up with a grouping of squads at 'x' location in preparation of an assault, then there's no issues.

The irony in this is that the bug forces the units who can teleport to do so at a safe location whenever it's possible, which by itself is certainly a good thing. I do remember that in Dark Crusade, sometimes, I would see a mass melee between many groups of infantry, in which a Necron Lord was involved, and badly hurt, only to teleport just six or seven meters "away" from the mess, but was soon killed anyway by a passing vehicle or by another nearby enemy squad simply because he wasn't any "safer" now that he was alone than he was when he at least had one or two squads left around him within the melee itself. If the SS bug makes things safer for the teleporting units then... well, nice bug? Hehe.

Me: But there is one thing I've noticed with this build about capturing that I haven't before, and curiously it occurs only when capturing Critical Points, I have yet to see it happen for Relics or LP's. The "issue" if I can call it that way is that whenever a squad seem to decide it wants to capture a Critical Point then it looks like the whole infantry in the team receives the order and goes there too.

Arkhan: Well, there is a reason for this behaviour. The Take and Hold behavior was criticized for not always working. According to the code, this could only happen when the AI was defending. If the critical counter runs down to about 3 mins, the AI creates an attack plan and tries to capture the closest critical point which belongs to the enemy. The problem occurs if the AI is defending. In this case, the units are assigned for defending the base and the attack plan for capturing an enemy LP doesn't get any units. I've changed that now that all defend plans are stopped if the AI tries to stop a victory counter. However, this is only the emergency counter if the timer is critical. If there's still enough time, the AI will just send a free capturing squad.

Ah, then good to see that you'll try another approach to that code. I'll certainly want to test the results of that in the next build. Also about that I'd like to confirm what Thudo mentioned, namely that the behavior is especially (but not exclusively) noticeable during late Tiers, mostly from Tier 3 indeed, here's screen-shot I took during one of my tests (note that the team's name reads 'Zenoth' but it was an A.I Vs. A.I game):



When I took the picture the units started moving away from the CP, but they gathered there around one minute or a little more before and remained there, idle, of course until they moved. You can also notice vehicles (in this case a Wraithlord for instance), which surprised me, because they (the vehicles, even air units) gather at LP's or sometimes at Relics, but I really honestly had never seen an infantry squads plus vehicles gathering at a CP, hence why I took the picture.

Also I want to be specific about something in relation to this. When I say gather I do refer to the "issue" which implies that during a capturing process then what looks like the rest of the team gets there as well, and idles until the point is captured or until enemy units approach close enough to stop the capturing process and start a fight. I'm saying this because there's also another gathering behavior, but that one isn't the issue. That other gathering behavior is when any unit types, and in any numbers will gather at either LP's or Relics (or both given enough units) before any assault(s) is ordered. That type of behavior is very common and I am aware of it, and I'm not referring to that one in particular.

Me: Basically the Campaign works without problems expect of course if playing with and/or against the Dark Eldar, which will potentially result in a CTD due to the Homing Device bug.

Arkhan: Perfect!

Well, I do have "bad" news about the Campaign, however I don't know with perfect exactitude if the cause of the problem was me forcing things up or because there was indeed a bug (perhaps even a known SS bug which wasn't related to the A.I at all).

What I decided to do to test the Campaign "properly" was to set the A.I to play it at my place, as if I was testing an A.I Vs. A.I skirmish, since the Campaign is a collection of skirmishes in the first place. So I did that, and it worked. Each "missions" was played by the A.I, and after each skirmishes I set things up as if I was playing the whole thing myself, such as reinforcing the provinces I controlled and giving wargear to my hero.

I played the Eldar (Craftworld) Campaign on Normal difficulty, against the Necrons. I played until I reached their Stronghold. So from the first skirmish to up to their Stronghold no bugs were seen, good thing. But at some point during the Stronghold mission it simply crashed to the Desktop. Now I am aware that the Stronghold missions for each factions in the Campaign do have scripted events and cut-scenes that interfere with game-play (or perhaps I shouldn't say interfere but instead say influence).

Now I just wonder if letting the A.I play Stronghold missions from any factions would also result in a CTD, or if it was just Necron-related or Eldar-related, or even SS-related (excluding the A.I completely) or just map-related... I'm at a loss. But maybe I forced things up as well. The A.I mod isn't meant to understand the Campaign and react to the cut-scenes and so forth. In fact, more precisely, the CTD occurred during a cut-scene (the second one after the introduction). So what I will do will be to play that mission myself instead and see if it crashes, quite simply. And I will come back on that later.

Edited by Zenoth, 06 November 2008 - 10:50 AM.

[Zenoth]

Hey, Zenoth, I'm still around, sorta, just lurking a bit. You've put quite a bit of thought in your posts...

Alright, so far I've been focusing on the Orks' melee units and the Eldar's Fleet of Foot tactics (On and Off during battles).

But I'd like to understand how the FoF tactics work. I would presume that it has to turn Off while idle to increase the long-range accuracy (for ranged units of course) while being On while units are on the move. And from what I've observed so far it seems to work that way, indeed, but I think it may be a little slow in applying and in reaction to situations. I've observed the intended result, and I've seen FoF turning On and Off randomly while units are idle and on the move. I'd especially like to know under which conditions does the tactics apply if a unit switches between ranged to melee (and the other way around too). For instance is the Farseer remains ranged, and then enters melee, how does the code directs the tactics? Is the code not taking melee into consideration?

Point #1: As far as the timing, the issue is how often my Fleet of Foot code loop in eldarinfantrytactic.ai gets executed (not very often). From the conversations I had at the time with the crew, there was no real way around this.
Also, the code doesn't take melee into consideration, but is hard-coded for each unit type. Default-melee units like Banshees, Seer Council, and Farseer turn FoF off when they are within 15 units of an enemy, even when moving, while ranged units will turn it off at a longer range (a default of 35, shorter for Fire Dragons and longer for Dark Reapers). This is done in order to compensate for the time delays - Eldar ranged infantry performing an attack move will turn FoF off just before they stop moving and start shooting, in an ideal situation. This code does not execute when infantry are making normal moves or retreat moves, of course - otherwise they'd never escape from nearby enemies.

It doesn't happen often, to see it turning On and Off like that for no apparent reasons, mind you, but I've seen it. Maybe if I understand more how it works then I will be able to test it more properly and have a better judgment on it. At this point I'd say that if the intended effect is to indeed turn it Off while infantry is idle and turning it On while infantry is moving (with both ranged and melee) then it may not always be working, even if it does work most of the time. Maybe that the On and Off "randomization" is intended in the code itself, I don't know. When it works it's absolutely impressive, let me say, it has saved many lives, but when an infantry squad leaves if Off and moves away because it's being shot at then I'm sure that it's not supposed to be so.

Point #2: If you've noticed FoF cycling rapidly on stationary units, try and describe what was going on - was a member of the squad stuck somewhere, were they capping a point, etc. Screencap might even help. When I fired up the new beta the other day I ran a couple Eldar matches to make sure my code still worked and it seemed to be all right. I had the cycling problem a while back but I thought I licked it.

Strangely the FoF On and Off tactics seem to work more with the Howling Banshees than with Warp Spiders for example (I.E melee Vs. ranged efficiency)... or maybe I'm imagining things. I still have to make sure that my claims are exact and that it wasn't just coincidence, but so far I've tested on Hard difficulty, with and without Heroes, on official maps, in 3 Vs. 3's and 4 Vs. 4's (in a total of four skirmishes) and from what I've seen then I must say that when it does work it's really great, but when it doesn't (or at least it seems like it right now) then it's just leaving the infantry without any real advantages of course. And three of the skirmishes were A.I Vs. A.I ones.

Point #3: Warp Spiders? They don't have FoF, perhaps you mean Dark Reapers...

Hey dreddnott! Thanks for posting here and giving all that information, it's much appreciated

I will answer you by points (your own replies).

Point #1

Alright, well it's simple and complex at the same time. I do honestly understand what you're telling me and how the FoF code and conditions work, that's the good news. But the "lesser good" ones is that I do notice FoF turning itself constantly On and Off whether or not a ranged infantry squad is idle or moving, and whether or not it's engaged in combat (shooting). It's not happening often, as I mentioned, it's quite rare, but I've seen it.

The one situation under which it can often be observed is when many units/squads gather at a LP/Relic before an assault is ordered, and the infantry just idles there for sometimes as long as two or three minutes. Now under such a condition, presuming of course that the FoF research was done, then what can be seen is that while the units are are idle they still turn FoF On and Off constantly, and they're idle, doing nothing, just waiting for their numbers to grow and go on with their desired task later on.

The frequency at which it turns On and Off seem constant to me, I may be wrong but at first glance (I haven't counted seconds between each activations and deactivations, maybe I should to be more precise) it seems to be the case, and the frequency, by estimation, is around four or five seconds. So let's say that FoF has just been researched. Now one of the already-gathered-and-idle squad turns it On (not all squads do it at the same time by the way), then let's say around four or five seconds later, it turns Off... and so on, in a loop. Now you can imagine just how much a Christmas Tree it looks like when many squads are present and turn it Off and On at different intervals, it's a cacophony of lights, hehe!

But again, I will repeat myself here, I don't see that happen often (but when it can be observed then as I mentioned the one condition under which is can be seen is especially when they idle during gathering before an assault is given, but it also happens during combat, although very rarely).

Point #2

Well I can certainly say with certainty that the cycling "problem" is never happening during an actual point capturing process, be it for a LP, CP or Relic. From what I myself have observed (if other testers could show up on that particular point too maybe we would see it another way but right now I'm alone on this, so guys please bear with me) it happens under the following conditions:

1) When units are stationary/idle (mostly during gathering around LP's or Relics or CP's, but the capturing squad itself, if any, will not have that behavior).

2) When units are moving (be it just getting outside of a base, without any enemies around, just going to "point B" or during fights, simply said whenever they move, and it includes fleeing due to morale loss or just because the unit is in danger, such as having to fall back due to having low health even with high morale, which is quite unlikely to happen anyway, but I'm just saying).

Point #3

My bad... yes I meant Dark Reapers, sorry, hehe!

So I hope it helps better understanding how it happens and why.

Let me test a little more and I will come back with a replay soon about that.

[RobertW7928]

I'm glad to see progress on the AI mod and I was curious if you would include option/s? either via AI configuration menu or other means - Enable/Disable - AI High Speed, Fortress, Massive Battle & Hero's for the Campaign, because most would prefer some options while not others and delivers a more preferred and pleasurable experience, as for the Stronghold conflicting with the AI I hope you guys sort out a way to resolve that dilemma.

Keep It Up Guys

[troubadour]

I have nothing to say better than Zenoth did but just wanted to confirm point 1 and point 2 reg FoF, even for stationnary unit it turns on and off and a regular pace even when no ennemy around (well i have to perform more testing to check how far 'around' is...)