|
|
#31 | |
|
User
Respected User
Join Date: Oct 2002
Posts: 1,296
|
Quote:
The key problem is that the display natives can deal with UTF-8 and interpret it properly but SubString can not (and maybe StringHash not either?). To replace part of the text with the right race and similar you have to parse the text and therefore read it character by character and that requires SubString. So what the character table does is that it associates with the answers SubString provides the original character string (before the alterations SubString does). I guess SubString adds an additional character to the "weird" single byte characters but to assemble the UTF-8 for the display functions you cannot use that. Btw, do the special characters like ä, ö and similar still work? Last edited by AIAndy : 09-18-2009 at 10:56 AM. |
|
|
|
|
| Sponsored Links - Login to hide this ad! |
|
|
|
|
#32 |
|
Great Sage
Project Leader: AMAI
Join Date: Jun 2003
Posts: 688
![]() ![]()
|
Ah now it all makes sense. Yeah StringHash has apparently the same problems as SubString.
__________________As to the special characters they are saved correctly in the script so i presume they should still work, but I should probrably check that up in game to be sure. Some languages only working on certain language specific versions of warcraft I should add to the readmes and things though. I then really need to a rundown of which language versions of warcraft / OS people are using that work with a particular language. Last edited by Strategy Master : 09-27-2009 at 12:27 PM. |
|
|
|
|
|
#33 |
|
User
Respected User
Join Date: Oct 2002
Posts: 1,296
|
What might work to fix it is to try if a 2-character substring survives correctly where a 1-character one does not (and progress in the parsing of the string accordingly).
That would replace the char table stuff. So you compare SubString(i,i+1) + SubString(i+1,i+2) with SubString(i,i+2). If it is the same, then the character at the current position is a normal ASCII character. Otherwise either the character starting at i or the character at i+1 is a 2-character long UTF-8 character. Compare SubString(i,i+2) + SubString(i+2,i+3) with SubString(i,i+3) to find out which. A SubString that is 2 characters long starting at the start of such an UTF-8 character might work out correctly (while the 1 character long does not). Last edited by AIAndy : 09-28-2009 at 04:49 PM. |
|
|
|
|
|
#35 |
|
User
Join Date: Dec 2008
Posts: 11
|
Hello, Strategy Master.
As you see, i and my friends(high school boys and girls)enjoy teaming against AMAIs on the net. It's a pity that 1.24 killed those bots. We're now having 1.21b-----a version old but nice and peaceful Anyway, those chatters aside, we liked your AI productions a lot, and we really hope to see the new available ones for 1.24 soon. Take care and good luck ![]() Last edited by ericwu101 : 10-10-2009 at 03:15 PM. |
|
|
|
|
|
#36 |
|
User
Join Date: Oct 2009
Posts: 10
|
@ericwu101
The 1.24b EXE posted in this thread seems to work fine for me. Page 2 post 26. |
|
|
|
|
|
#37 |
|
User
Join Date: Oct 2009
Posts: 10
|
@Strategy Master
Can we make strategies for 1.24b, or do your developer tools need to be updated? I see dev tools for 2.54 and 3.00? Which version should i get started in for editing the strategies/profiles? I've read through the beginner tutorial and am in interested in tinkering with strategies. I used to code AI scripts for Starcraft, so some of the process looks familiar. The Warcraft scripts look simpler and the tools/syntax much better. Is there a blank replay to test AI's at 8x speed? In starcraft we had blank replays that would run 45 minutes to an hour and the replays were just AI vs AI, so you could test your AI by just loading the replay. |
|
|
|
|
|
#38 |
|
Great Sage
Project Leader: AMAI
Join Date: Jun 2003
Posts: 688
![]() ![]()
|
@Peppe: There is no official released version for 1.24b of warcraft available yet and so no developer edition available that support this version. Strategies can be easily exported and imported across different versions of AMAI though so you can start developing on one of the current versions, but you will not be able to test it out. Its best to make strategies with the 2.54 developer tools at present.
__________________It may be possible to use the world editors own ai editor to run a test run at high speed. Would require some tinkering to get it working for AMAI though. The related AMAI scripts would first have to be imported into the world editor map first. |
|
|
|
|
|
#39 |
|
User
Join Date: Oct 2009
Posts: 10
|
Thanks for the info.
I did some tinkering last night. I reverted the game back to 1.23. And MPDraft worked to create the exe for that version with your AI's without issue. I start trying to make my own strat in AMAI 3.00. I haven't seen any crashing or too terribly bad behavior. Is there some 3.00 issue i should be looking out for? The replay stuff didn't require any replay editing in warcraft, just had to create a lan game with me as the observer and save that replay.. I made an AI for each race i wanted to test that does nothing. Locked AMAI to them and loaded up a 1vs1 map and set them to play each other. Came back 45 minutes later ended the game and saved the replay. Now i can test that matchup at 8x speed using the replay of that game. Just have to lock it to the strategy and load the replay. I found where to change what skills heroes pick. Is there a place to change how they use the skills? Shadow Hunter with healing wave will blow his mana on the first engagement and uses the skill on any damage. If it could only heal at like 50% unit health should improve the usage. |
|
|
|
|
|
#40 |
|
Great Sage
Project Leader: AMAI
Join Date: Jun 2003
Posts: 688
![]() ![]()
|
AMAI3.00 is a technical alpha so it could easily do strange things. The code base is pretty stable though. Technically the current build of 3.00 is the same as 2.54 just uses slightly different methods in its strategy generation.
__________________Thats a reasonable replay method you can do. Note though it will be using the same random seed so it wont be as perfectly random as playing on a new game each time but yes it will at least be 8x speed for a basic strategy test. Hardcoded unit micro is not overwritten due to that requires completly taking over all the unit control over and hard programming in commands for everything. Although unit micro has been made smarter with some abilities. As ability usage extends over time there would be more call to do this. Otherwise for now its best to a least have some default intelligence instead of nothing at all :D Last edited by Strategy Master : 10-12-2009 at 08:31 PM. |
|
|
|
|
|
#41 |
|
Great Sage
Project Leader: AMAI
Join Date: Jun 2003
Posts: 688
![]() ![]()
|
Right over the next week stay tuned as i will hopefully have the update for 1.24 out officially (AiAndy's idea i will have to come back to for future versions). Finally re-synched to the website again too and got the web developer environment up (man what a load of rubbish i've left on there :O). So i'm nearly prepared, should be a matter of finalizing the new AMAI2.54 build and readme's and then its all go.
__________________Last edited by Strategy Master : 10-13-2009 at 09:13 PM. |
|
|
|
|
|
#42 |
|
User
Join Date: Oct 2009
Posts: 10
|
Sounds good. I have been working on mimicking the current strats people use. Working on ORC and Undead at the moment.
I play undead, so first worked on an orc strategy to play against (BM + TC, Grunts, Raiders, Walkers). Open Altar -> Burrow -> Barracks. Tried SH second, but the AI micro for SH is rubbish. AI plays pretty good with TC. Doesn't harass with the BM like a human, but does get the red numbers flying for the big battles. Made an undead version to matchup against the orc above (DK + Lich + DL, Mass Fiends, Statue/Destroyer, Upgrades). Graveyard -> Crypt -> Zig -> Altar -> Shop. AI plays pretty good and creeps with skeles well (could be better if it considered how long it has the skeles and moved quickly to use them before they pop, and if it made skeles at graveyard before first creep). I'll probably leave diversifying strategies till your new version is out and make some new replays to test with. I figure getting a replay on 3-4 common maps for each match-up should add back some of the randomness that is lost in a single replay. Last edited by Peppe : 10-13-2009 at 10:52 PM. |
|
|
|
|
|
#43 |
|
User
Join Date: Dec 2008
Posts: 11
|
hum.
All seems fine,Can't wait for it. I'll stay tuned. |
|
|
|
![]() |
| Thread Tools | Search this Thread |
|
|
|
Donate |