2.45 beta 1 ? Life Fountain and Defeat Conditions bugs.

I've played a single game. TFT 1.18 at the “market square� 2x2x2x2. I didn't change anything neither in the code nor in the profiles.

Very small groups of units are send to the Life Fountain, as in the previous versions. So the blood never dries away from the central square (heh, 8 players game, what do u whant? ^_^ ).
I suggest, to simplify as much as possible: in “random� cases the whole armies are send to Fountain for healing (healthy units cover/excort wounded).
In addition, Life Fountain may be counted as a perspective target for an attack (or simpler as a waypoint!). And may be it worth to add a new variable to the AMAI profiles – love for Fountains (attack and/or heal).

Defeat Conditions works worse than in the previous version. Now all AMAI players stay in the game till:
FoodUsed() < 6 and hall_count == 0 and worker_count == 0
or
FoodUsed() == 0 and hall_count == 0

The first informal defeated AMAI was sea-green. It's near 31 minute.

In that game almost nobody reached 100 food. AFAIK except one AMAI. I don't know if it is a randomness or a something else? I'll pay attention at it in the next games.

I've thought here. May be testers would be of more help, if we will know more about debugging?
Please explain how to output different debug information (in details so we can understand). (My tries hadn't succeeded.)

I've saw certain improvements!

– AMAI now retreat faster, and counts dangerous situations more precisely. But may be the first is a consequence of the second. In any case the real progress is beyond question.

– I saw two or three (?) separate groups/armies with the Heroes (for one/each AMAI). They act completely independent.
Unfortunately such small groups were easily beaten in the most cases.
IMHO in team games such tactics are fatal. Where there isn't enough place to maneuver – the brute force rules (I don't mean totally narrow passages).

– AMAI armies have got slightly increased movement speed?

PS
Sorry for my English.
Attachment's real size 570 KB :)

1) We are testing a feature with the fountain control which has not been introduced yet. At moment all that is implemented is control to prevent units going to fountain when enemies are healing there. (hmmm i going to make a known issues post i feel)

2)Are you sure all AMAI players leave at only those conditions. You do realise each profile has a different leave levels. Some profiles give up easily like "chick3n" while others aggressario give up using last versions defeat. chick3n can be defeated if any of the 12 defeat conditions are true so chick3n really is a chick3n. Aggressario can only be defeated with 1 single condition.

If you using custom profiles or something check the developer option in the profiles section that they not all set at 1.

3)Its possibly randomness but there may be a problem. Tell us if it continues to become a problem

4)Debugging

If you are referring to debugging the actual code, then thats a complicated matter and all sorts of techniques. But you can write following statements

- write messages on screen when that point of script is run.

- displays contents of the variable if integer

So you could place "display to alls" in each defeat condition. If it displays the text you know which defeat condition the computer used to give up and you can really see if amai always uses that defeat condition.

These can be placed anyway. The common.eai, job scripts etc.


For developer work
The manual has lots of info about using the DE edition more so than the old version of the manual and is easier to navigate. To ease helping us do not repeat issues we know about in the known issues of the manual. To test out our new features for DE they are just all newly added global settings. They are "version_heroes" , "disable_retreat", "fixedcomputername"

AMAI vs Ai
For some who wanted a better tool for debugging AMAI vs Ai matches we do have a specially made debug tool where humans are just made observers. It stores results in game cache displaying on multiboard while you observer (and can use cheats if you desire). The first half of computer players are AMAI and second half are standard ai. Games just repeat forever till you quit and allows to see if mistakes cost AMAI the match. After let say 20 games you can see how many are amai victories or ai victories and u can do a print screen or whatever to show us. I have attached this here so people can download.

human.ai is amai script human2.ai is referring to normal ai scripts.




5)hmmm. Separate armies you say. Thats a very interesting result and its either for worse or for better or random occurence.




AMAI_Debug_Version (For Ai vs AMAI debug matches)
464.2 KB

Defeat Conditions works! I’ve played two times – everything fine. That problems were a strange bug (but we caught it!).

As for not entering high upkeep. Last game one AMAI had reached 81 food. :) Others had less. I'll keep watching.

AMAI enters high upkeep. There is no bug.

http://www.hornes.pwp.blueyonder.co....tegymaster.jpg

What' this?

Tell me PLZ

It's StrategyMaster's project of implementing Warcraft II using the Warcraft III engine. Units, Spells, and even the attack/armor system and scales are all modded to be very very close to Warcraft II. That also uses a special version of AMAI.

@iamxiaot: I see the same post in three threads, and all of them look like Troll Posts because you also did not explain in detail your opinion.

Is the debug tool self-contained, or do I need the beta files? Do I need to install the scripts onto the maps and then use those in conjunction with the debug tool, or do I simply run the tool and use the standard maps?

BTW, the tool didn't run game after game for me, when the first game finished it told me Game Over, I clicked OK and it took me back to the lobby. Was this because I made myself an observer (so I could use an appropriately sized map) rather than let the tool do it?

Self contained. No. No. Yes. :)

Not sure, but think so. Pick a 3 player map and up, once started you won't appear as a player. BTW the leaderboard scores are stored under your current profile as Campaigns.w3v and accumulated from session to session. To reset the board delete this file. If you have started a Warcraft campaign previously and wish to return to it later, create a new profile for AMAI testing.

I tried both network and custom games as a human player, but the game never ended. Once one AI had won it would just sit in its base and the game would keep on running...

I'm having problems after running the debug exe. After a debug "tourney", I fully exit WC3. Next time I run WC3, AMAI isn't there, Computer players are named "Computer", no status or chat messages from AMAI, and the game stutters once per second. This if I run any of the debug, SE exe's, or run WC3 by itself and load a map with AMAI embedded. WC3 must be loading other persistent data from the campaign file?

If I delete the player profile the debug games were run under, problem goes away everything back to normal.

So try creating a new player profile each time you run debug and deleting it afterward. This is necessary to reset the scores anyway.

Try setting yourself as random on a 3rd team apart from the two ai's.

I notice AMAI still chats to me, refers to my color/race, even though I'm not in the game :)