--- Log opened Wed Nov 21 00:00:24 2012
--- Day changed Wed Nov 21 2012
20121121 00:00:24-!- fendrin [~fabi@88-134-20-241-dynip.superkabel.de] has joined #wesnoth-umc-dev
20121121 00:00:24-!- fendrin [~fabi@88-134-20-241-dynip.superkabel.de] has quit [Changing host]
20121121 00:00:24-!- fendrin [~fabi@wesnoth/developer/fendrin] has joined #wesnoth-umc-dev
20121121 00:00:37-!- fendrin [~fabi@wesnoth/developer/fendrin] has quit [Client Quit]
20121121 00:15:15-!- BTreeGA [~BTreeGA@moose.CS.McGill.CA] has joined #wesnoth-umc-dev
20121121 01:38:53-!- irker760 [~irker@ai0867.net] has joined #wesnoth-umc-dev
20121121 01:38:53< irker760> wesnoth-umc-dev: espreon * r16202 /trunk/Archaic_Resources/ (6 files in 3 dirs): 
20121121 01:38:54< irker760> wesnoth-umc-dev: AR: ran umcpropfix.
20121121 01:39:13< irker760> wesnoth-umc-dev: espreon * r16203 /trunk/Bad_Moon_Rising/utils/enemies-list.cfg: 
20121121 01:39:13< irker760> wesnoth-umc-dev: BMR: ran umcpropfix.
20121121 01:42:07< irker760> wesnoth-umc-dev: espreon * r16204 /trunk/Tales_of_the_Setting_Sun/ (178 files in 15 dirs): 
20121121 01:42:07< irker760> wesnoth-umc-dev: TotSS: ran umcpropfix.
20121121 01:42:27< irker760> wesnoth-umc-dev: espreon * r16205 /trunk/Trinity/maps/F5b3.map: 
20121121 01:42:27< irker760> wesnoth-umc-dev: T: ran umcpropfix.
20121121 01:46:09< irker760> wesnoth-umc-dev: espreon * r16206 /trunk/Nightmares_of_Meloen/po/de.po: 
20121121 01:46:09< irker760> wesnoth-umc-dev: NoM: ran umcpropfix.
20121121 02:11:10-!- faabumc [~vcr@wesnoth/developer/faabumc] has joined #wesnoth-umc-dev
20121121 02:31:15-!- Crendgrim [~crend@port-92-204-47-107.dynamic.qsc.de] has quit [Ping timeout: 260 seconds]
20121121 02:37:44< irker760> wesnoth-umc-dev: faabumc * r16207 /trunk/Nightmares_of_Meloen/utils/jinx.cfg: 
20121121 02:37:44< irker760> wesnoth-umc-dev: NoM: fix 'cursed' label when jinx is used during defense.
20121121 03:54:15-!- loonybot [~loonybot@wesnoth/bot/loonybot] has quit [Remote host closed the connection]
20121121 04:07:03< irker760> wesnoth-umc-dev: doofus-01 * r16208 /trunk/Bad_Moon_Rising/utils/ (skirmish.cfg enemies-list.cfg): 
20121121 04:07:03< irker760> wesnoth-umc-dev: enemies list for despair faction skirmishes
20121121 04:08:18< irker760> wesnoth-umc-dev: doofus-01 * r16209 /trunk/Bad_Moon_Rising/scenarios2/1_02_World.cfg: 
20121121 04:08:18< irker760> wesnoth-umc-dev: made it not possible to go to scenario Camp without going through Rescue
20121121 04:18:23-!- faabumc [~vcr@wesnoth/developer/faabumc] has quit [Quit: Ex-Chat]
20121121 04:40:11< irker760> AI-Demos: mattsc master * r4c4aaf4 / lua/grunt-rush-Freelands-S1_engine.lua : Fred: retreat injured units to occupied villages - http://git.io/c0zYlQ
20121121 04:48:27-!- Vultreon [~chatzilla@wesnoth/developer/vultraz] has quit [Ping timeout: 276 seconds]
20121121 04:49:30-!- Vultreon_ [~chatzilla@124.109.10.167] has joined #wesnoth-umc-dev
20121121 04:49:45-!- Vultreon_ is now known as Vultreon
20121121 04:53:34-!- Vultreon is now known as vultraz
20121121 04:53:40-!- vultraz [~chatzilla@124.109.10.167] has quit [Changing host]
20121121 04:53:40-!- vultraz [~chatzilla@wesnoth/developer/vultraz] has joined #wesnoth-umc-dev
20121121 04:59:45-!- faabumc [~vcr@wesnoth/developer/faabumc] has joined #wesnoth-umc-dev
20121121 05:06:53< irker760> AI-Demos: mattsc master * r4bf821e / lua/grunt-rush-Freelands-S1_engine.lua : Fred: retreat villages more than 1 move away don't need to be unoccupied - http://git.io/U7rpiw
20121121 06:37:47-!- Cookiee [~quassel@60-240-54-150.tpgi.com.au] has joined #wesnoth-umc-dev
20121121 06:37:47-!- Cookiee [~quassel@60-240-54-150.tpgi.com.au] has quit [Changing host]
20121121 06:37:47-!- Cookiee [~quassel@unaffiliated/cookiee] has joined #wesnoth-umc-dev
20121121 08:14:41-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-umc-dev
20121121 08:53:12-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has joined #wesnoth-umc-dev
20121121 09:30:28-!- irker760 [~irker@ai0867.net] has quit [Quit: transmission timeout]
20121121 09:59:32-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20121121 10:10:58-!- Crendgrim [~crend@port-92-204-84-161.dynamic.qsc.de] has joined #wesnoth-umc-dev
20121121 10:37:39-!- Crendgrim [~crend@port-92-204-84-161.dynamic.qsc.de] has quit [Quit: Konversation terminated!]
20121121 10:37:45-!- Crendgrim_ [~crend@port-92-204-84-161.dynamic.qsc.de] has joined #wesnoth-umc-dev
20121121 10:47:19-!- mnewton1 [~mnewton1@c-76-99-170-226.hsd1.de.comcast.net] has quit [Quit: Rice and Corn]
20121121 10:47:48-!- Crendgrim_ [~crend@port-92-204-84-161.dynamic.qsc.de] has quit [Quit: Konversation terminated!]
20121121 10:47:56-!- Crendgrim_ [~crend@port-92-204-84-161.dynamic.qsc.de] has joined #wesnoth-umc-dev
20121121 11:18:39-!- Cookiee [~quassel@unaffiliated/cookiee] has quit [Ping timeout: 265 seconds]
20121121 11:30:22-!- Cookiee [~quassel@unaffiliated/cookiee] has joined #wesnoth-umc-dev
20121121 11:39:49-!- faabumc [~vcr@wesnoth/developer/faabumc] has quit [Quit: Ex-Chat]
20121121 11:56:24-!- Cookiee [~quassel@unaffiliated/cookiee] has quit [Ping timeout: 252 seconds]
20121121 12:33:38-!- Cookiee [~quassel@unaffiliated/cookiee] has joined #wesnoth-umc-dev
20121121 12:40:27-!- Cookiee [~quassel@unaffiliated/cookiee] has quit [Remote host closed the connection]
20121121 13:09:09-!- loonybot [~loonybot@46.138.73.254] has joined #wesnoth-umc-dev
20121121 13:09:09-!- loonybot [~loonybot@46.138.73.254] has quit [Changing host]
20121121 13:09:09-!- loonybot [~loonybot@wesnoth/bot/loonybot] has joined #wesnoth-umc-dev
20121121 13:09:09-!- mode/#wesnoth-umc-dev [+v loonybot] by ChanServ
20121121 14:22:26-!- skyfaller [~skyfaller@wikipedia/Skyfaller] has quit [Remote host closed the connection]
20121121 14:28:44-!- loonycyborg_ [~loonycybo@wesnoth/developer/loonycyborg] has joined #wesnoth-umc-dev
20121121 14:34:32-!- Netsplit *.net <-> *.split quits: iwaim_, irker625, loonycyborg
20121121 14:34:32-!- loonycyborg_ is now known as loonycyborg
20121121 14:43:50-!- Netsplit over, joins: iwaim_
20121121 14:49:13-!- mnewton1 [~mnewton1@c-76-99-170-226.hsd1.de.comcast.net] has joined #wesnoth-umc-dev
20121121 14:58:01-!- mattsc [~mattsc@d154-20-32-241.bchsia.telus.net] has quit [Quit: I'm asleep]
20121121 15:03:52-!- Alarantalara [~Adium@CPEc0c1c09e8055-CM00252eac6d62.cpe.net.cable.rogers.com] has quit [Quit: Leaving.]
20121121 15:10:16-!- mattsc [~mattsc@207.230.251.234] has joined #wesnoth-umc-dev
20121121 15:15:27-!- mnewton1 [~mnewton1@c-76-99-170-226.hsd1.de.comcast.net] has quit [Quit: Rice and Corn]
20121121 15:41:33-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-umc-dev
20121121 15:42:27-!- irker724 [~irker@ai0867.net] has joined #wesnoth-umc-dev
20121121 15:42:27< irker724> AI-Demos: mattsc master * rf44e821 / lua/grunt-rush-Freelands-S1_engine.lua : Fred: retreat only one unit toward each out-of-reach village - http://git.io/esRK4g
20121121 15:42:55-!- noy_ [~Noy@wesnoth/developer/noy] has joined #wesnoth-umc-dev
20121121 15:45:56-!- noy [~Noy@wesnoth/developer/noy] has quit [Ping timeout: 255 seconds]
20121121 15:45:56-!- noy_ is now known as noy
20121121 15:58:37< irker724> AI-Demos: mattsc master * r296b297 / lua/grunt-rush-Freelands-S1_engine.lua : Make Fred more aggressive in the west - http://git.io/VJm_3A
20121121 16:15:35-!- mattsc [~mattsc@207.230.251.234] has quit [Quit: I'm asleep]
20121121 16:23:29-!- skyfaller [~skyfaller@wikipedia/Skyfaller] has joined #wesnoth-umc-dev
20121121 16:34:36-!- mattsc [~mattsc@fw.hia.nrc.ca] has joined #wesnoth-umc-dev
20121121 16:56:19< irker724> AI-Demos: mattsc master * rb4d2451 / lua/grunt-rush-Freelands-S1_engine.lua : Fred: make sure the zones cover the entire map - http://git.io/qdjYgA
20121121 17:05:26< irker724> AI-Demos: mattsc master * r249c499 / lua/grunt-rush-Freelands-S1_engine.lua : Fred: deal with left before center - http://git.io/_uwzgw
20121121 17:27:18-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20121121 17:33:31< irker724> AI-Demos: mattsc master * rde845c2 / lua/grunt-rush-Freelands-S1_engine.lua : Fred: add min_y parameter for zone holding - http://git.io/A2QAmw
20121121 17:34:20< mattsc> skyfaller: here's another one: http://imagebin.org/236713
20121121 17:34:53< mattsc> Should Fred try to get rid of that assassin, or grab the two villages in the south that he can get to?
20121121 17:35:51< mattsc> Also, would the answer be different if Fred's unit were not poisoned and the blue unit were, say, an archer?
20121121 17:51:00-!- BTreeGA [~BTreeGA@moose.CS.McGill.CA] has quit [Ping timeout: 248 seconds]
20121121 18:01:54< skyfaller> mattsc: what's the ctk? I can't tell without seeing HP etc.
20121121 18:02:58< mattsc> Don't know without looking it up, but I'd like to know the answer not just for this specific situation, but a bit more generally.
20121121 18:03:18< skyfaller> also, are any reinforcements on the way? If there are reinforcing units on the way I would lean more towards killing the assassin, b/c then your odds of holding the village for the long term go up
20121121 18:03:21< mattsc> So, "if CFT > X then ... otherwise ..."
20121121 18:03:47< skyfaller> I wouldn't trust either of those units to hold a village after another turn of fighting + being poisoned
20121121 18:04:12< skyfaller> so if these are all the forces you have, then I would heal them both right away
20121121 18:04:22< skyfaller> and take two villages
20121121 18:04:33< mattsc> In this case, the other units are busy holding off the attack in the center, so no reinforcements at least for a turn or 3.
20121121 18:04:58< skyfaller> ok, I would lean towards grabbing 2 villages then
20121121 18:05:07< skyfaller> also, can the enemy grunt reach?
20121121 18:05:21< skyfaller> if it is a 6 move grunt, fighting becomes more risky
20121121 18:05:52< skyfaller> if it is a 5-move grunt, you can arrange your units so that it can't reach, while still trapping the assassin (you have to surrender some defense though)
20121121 18:06:18< mattsc> Ok.
20121121 18:06:48< skyfaller> if the ctk on the assassin were above 60%, I might try to kill it
20121121 18:06:54< mattsc> Let me ask a slightly different question though: why would you not take the two villages under (essentially) all circumstances
20121121 18:07:30< mattsc> It'll make a huge difference in income and unpoison both units.
20121121 18:07:36< skyfaller> b/c it has to be a pretty good ctk for me to give up on village grabbing in favor of a possible kill + XP
20121121 18:08:32< mattsc> That's the argument the other way around, I think...
20121121 18:08:41< skyfaller> yeah, that's true, but kills + XP are still good even for Northerners
20121121 18:09:07< skyfaller> a lvl2 assassin is hard to deal with
20121121 18:10:13< mattsc> So, as usual, there's no answer that can be put into an easy equation.  :)
20121121 18:10:17< skyfaller> yes :)
20121121 18:10:31< skyfaller> that enemy assassin looks like it might be intelligent, so you'd rather it didn't live to level
20121121 18:10:31< mattsc> However, it seems that grabbing the villages is better more often than not.
20121121 18:10:48< skyfaller> yeah, with no reinforcements coming, I would lean towards village grabbing
20121121 18:10:58< skyfaller> the ctk probably isn't that high
20121121 18:11:03< mattsc> Ok.  That's what Fred does at the moment.  :)
20121121 18:11:51< skyfaller> if Fred's units were not poisoned, and it is a less dodgy archer with a correspondingly higher ctk, it would make me more likely to try the kill
20121121 18:12:08< skyfaller> but on the other hand archers hurt the enemy less to lose
20121121 18:12:48< skyfaller> part of why I want the kill is because assassins are expensive and that one looks intelligent
20121121 18:13:28< mattsc> Right...  Now put this into an equation that works under all circumstances...
20121121 18:13:31< mattsc> Sigh ...
20121121 18:13:41< skyfaller> a non-intelligent archer I'd be less excited about killing
20121121 18:14:02< mattsc> Btw, when you're talking CTK, do you mean CTK due to attack by one of the two units, or the combination of the two?
20121121 18:14:07< skyfaller> combination
20121121 18:14:32< mattsc> Ok.
20121121 18:14:35< skyfaller> if it will probably die this turn I can live with having to retreat and not getting a village
20121121 18:15:39< skyfaller> you know what, I probably wouldn't try to kill the unit unless one assassin's throwing knife could do the trick
20121121 18:16:06< skyfaller> that would give me 5 strikes on the unit, with three of them using marksmanship
20121121 18:16:22< skyfaller> if the unit can survive that then it is just leading a charmed life
20121121 18:16:34< mattsc> :)
20121121 18:16:58< skyfaller> it probably is carrying the Teela Brown gene and I would fear engaging it in combat
20121121 18:17:38< mattsc> DNA testing is beyond the Fred's scope, so we might just have to risk it.
20121121 18:17:59< mattsc> Well, thanks for the comments.  Very frustrating...
20121121 18:18:03< skyfaller> lol
20121121 18:18:44< mattsc> It just shows once more why no AI will ever beat a semi-experienced human.
20121121 18:19:08< mattsc> Not consistently, at least, there's always the luck or surprise factor.
20121121 18:20:05< skyfaller> never is a long time
20121121 18:20:13< skyfaller> but I tend to agree
20121121 18:20:34< skyfaller> I think our goal should just be for Fred to annihilate all of the other AIs ;-)
20121121 18:21:04< mattsc> If that's the only goal, we're done (except for Ron with Undead).  ;)
20121121 18:21:35< skyfaller> well, we're done with that goal except Fred only works on orcs freelands side 1
20121121 18:21:50< mattsc> true
20121121 18:22:05< skyfaller> let's get Fred to beat Ron with Undead, and then start generalizing him
20121121 18:22:12< mattsc> Anyways, I don't know if you've been following the commits over the last couple days, I've actually started to do the tuning of Fred's parameters.  It's fun, but it also shows some of the limitations again.
20121121 18:22:25< skyfaller> .... don't tell me you're considering another refactoring ;-)
20121121 18:22:48< mattsc> No.  Not for a while at least.
20121121 18:23:13< mattsc> I was just going to write that this tuning is much easier now than it was before the refactoring.
20121121 18:23:26< skyfaller> are there no other AIs in the field? just the RCA and our own Ron?
20121121 18:23:56< mattsc> No complete AIs, no.
20121121 18:25:00< skyfaller> well, maybe once Fred is generalized we can have him included in Wesnoth, alongside Ron
20121121 18:25:19< skyfaller> then players could choose easy, medium, hard (RCA, Ron, Fred)
20121121 18:25:39< mattsc> Yeah, let's see where we're getting with this.
20121121 18:25:53< mattsc> I am actually somewhat encouraged by recent progress/results.
20121121 18:25:58< skyfaller> yay!
20121121 18:26:13< mattsc> Fred's already winning against Ron already while making some _really_ stupid moves.
20121121 18:26:26< mattsc> And fiddling with things has become much easier.
20121121 18:27:20< mattsc> So currently I am watching a game of Fred vs. Ron, try to fix the most blatant move that I see, then repeat.
20121121 18:27:39< mattsc> *most blatant mistake
20121121 18:28:07< skyfaller> and Ron consistently clobbers the RCA AI, right?
20121121 18:28:14< mattsc> Oh yeah
20121121 18:29:00< mattsc> On even terms, the RCA AI never wins against either Fred or Ron any more, unless in cases of extreme luck or really stupid mistakes.
20121121 18:29:08-!- BTreeGA [~BTreeGA@moose.CS.McGill.CA] has joined #wesnoth-umc-dev
20121121 18:30:01< skyfaller> so simply not making horrendous errors should be enough for Fred and Ron to defeat the RCA in the future
20121121 18:30:20< mattsc> Oh, there's BTreeGA.  He might have an alternative full AI at some point ...
20121121 18:30:31< mattsc> I think we're there already.
20121121 18:31:04< mattsc> If I test against the RCA AI, I always give it at least 150 gold, otherwise it's just no contest.
20121121 18:31:46< skyfaller> BTreeGA: you're working on an AI too? What's it like / how is it different?
20121121 18:32:18< mattsc> The problem with playing against humans is that they always very quickly identify the weaknesses and exploit those to the maximum extent possible.
20121121 18:32:39< mattsc> It's very annoying actually  :D
20121121 18:32:53< skyfaller> :D
20121121 18:32:58< mattsc> (because it emphasizes all the weaknesses and doesn't show off what the AI is actually good at)
20121121 18:33:00< AI0867> on any map with any faction?
20121121 18:33:01< skyfaller> yeah, that's why I'd like to build surprises into Fred
20121121 18:33:44< mattsc> AI0867: for Ron that's pretty close to true (although I have to admit that I have not tested it recently)
20121121 18:34:20< mattsc> Alarantalara has done a lot of tests on other maps, so he might be able to answer that.
20121121 18:34:26< skyfaller> I'd like Fred to have unusual strategies that he uses some tiny percentage of the time, so that the human always has to be on their toes and can't assume anything
20121121 18:34:57< mattsc> skyfaller: yes, I know you want that.  And that's actually one of the reasons why I've done all that refactoring recently.
20121121 18:35:58< mattsc> Having Fred do the main assault through the center or left rather than on the right is a matter of changing less than 10 lines now.
20121121 18:36:19< mattsc> And maybe 20-30 lines to set it up to do it randomly.
20121121 18:36:24< skyfaller> yay!
20121121 18:36:50< skyfaller> I'd also like other surprises like BURs (Bizarre and Unusual Recruits, like HODOR)
20121121 18:37:08< skyfaller> and ambushes
20121121 18:37:18< mattsc> Sure.
20121121 18:37:58< mattsc> Let's still get "default Fred" to work a bit better though, I've only been at that for a couple days.
20121121 18:38:01< skyfaller> certainly :)
20121121 18:38:35< mattsc> But as I said, I am pleased already that we've gone from always losing to Ron to winning against Ron most of the time.  
20121121 18:38:45< mattsc> It's the little things that make me happy.  :)
20121121 18:38:51< skyfaller> :D
20121121 18:39:32< skyfaller> If Ron already consistently clobbers the RCA AI, I wonder if we should ask to have him included in Wesnoth, alongside the ML recruiter
20121121 18:40:12< mattsc> I have mixed feelings about that
20121121 18:40:21< skyfaller> why is that?
20121121 18:40:27< mattsc> In my opinion, Ron's a "temporary AI".
20121121 18:40:37< mattsc> He's a placeholder that we use until Fred can play any map.
20121121 18:40:47< skyfaller> I don't see Ron as a temporary AI, I think there is a place for Ron indefinitely
20121121 18:41:16< mattsc> But in the end, Fred will do anything that Ron does - and more if appropriate.
20121121 18:41:23< skyfaller> I like the idea of having an AI be "hard" or "easy" based on its brains, not just how much gold you give it
20121121 18:41:44< skyfaller> if Ron is a simple-minded cousin of Fred, that's fine, some humans are simple-minded too ;-)
20121121 18:41:59< mattsc> :)
20121121 18:42:05< mattsc> Agreed, but ...
20121121 18:42:17< skyfaller> maybe instead of easy, medium, hard we could label the AIs normal, hard, and insane :D
20121121 18:42:43< mattsc> I don't think we'll manage insane, at least not against people like you  :)
20121121 18:42:48< skyfaller> fair enough
20121121 18:43:12< mattsc> ... in my mind, Fred will be a configurable AI
20121121 18:43:27< skyfaller> why would you configure Fred to be dumber?
20121121 18:43:47< skyfaller> I saw Fred as intended to be as good as possible
20121121 18:43:59< mattsc> You just argued for having an easier AI.  Why would you make it two AIs, rather than one with a switch?
20121121 18:44:19< skyfaller> fair enough, maybe I just like the idea of them being written differently
20121121 18:44:34< mattsc> Yeah ...
20121121 18:44:37< skyfaller> I think we should go for simplicity with Ron
20121121 18:44:47< skyfaller> like for his code to actually be as simple and concise as possible
20121121 18:45:05< skyfaller> although with all of the work Alarantalara has done on him, I wonder if that is still feasible
20121121 18:45:06< mattsc> Well, let's just see how it goes.
20121121 18:45:43< skyfaller> basically I'd like Ron to be the simplest possible AI that can still clobber the RCA
20121121 18:46:03< mattsc> I don't have a very strong opinion on this, just saying how I am expecting it to go.
20121121 18:46:22< skyfaller> fair enough, but I argue that Ron should have an independent existence
20121121 18:46:42< skyfaller> that Fred should be designed to be map-specific and have specific strategies coded in
20121121 18:47:20< skyfaller> while Ron doesn't have any human help like that
20121121 18:47:50< mattsc> Well, we're pretty much thinking about it the same way then
20121121 18:48:30< skyfaller> ok, Ron has been rescued from being temporary? :)
20121121 18:48:30< mattsc> Just that in your case, you're thinking about it as implemented in 2 AIs, while I am thinking of: Fred without additional parameters = Ron
20121121 18:48:39< skyfaller> aw, fair enough
20121121 18:49:01< mattsc> We'll certainly have a Ron mode of Fred.  :)
20121121 18:49:26< mattsc> As I said, let's see how it goes.  We're far from having Fred working on all maps.
20121121 18:49:32< skyfaller> but if Ron is already better than the RCA, why not put him in Wesnoth already?
20121121 18:50:08< skyfaller> Fred will get better over time, that's no reason not to put earlier versions of him into Wesnoth if he is better than the default AI
20121121 18:50:31< skyfaller> if they had waited until Wesnoth was perfect before releasing it... would we even be here talking about this?
20121121 18:51:06< mattsc> I think that's a question of whether it's worth the effort right now, while both Ron and Fred still change significantly all the time.
20121121 18:51:13< mattsc> It's also a question for the developers.
20121121 18:51:49< skyfaller> well, I don't think it's too early to discuss it with them
20121121 18:51:57< mattsc> So I'll let you take that up with Alarantalara later, I have to do some work now.  :)
20121121 18:52:06< skyfaller> ha, ok
20121121 18:52:27< skyfaller> I mean, it might be worth knowing what level of stability they would want before including it in unstable, for example
20121121 18:52:39< mattsc> I'll continue trying to eliminate Fred's worst mistakes.  I'm expecting to ask you to test things pretty soon.
20121121 18:52:45< skyfaller> ok, sounds good
20121121 18:56:26< mattsc> Oh, hey, skyfaller, did you see my questions from ... yesterday sometime, I think:
20121121 18:56:42< mattsc> do you know, in general, how the zone_control mechanism works
20121121 18:56:55< skyfaller> oh, not really
20121121 18:57:12< skyfaller> is it documented anywhere? :)
20121121 18:57:13< BTreeGA> Hey sorry, I was afk. Work is still going slow. I'm setting up the behaviour trees and genetic algorithm still.
20121121 18:57:16-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-umc-dev
20121121 18:57:22< mattsc> I think getting yourself familiar with the general principle would help with testing things and figuring out what is (relatively) easy to change and what is hard.
20121121 18:57:56< mattsc> skyfaller: no.  But have a look at the following functions (in Fred):
20121121 18:59:13< mattsc> get_zone_cfgs(), zone_control_eval(), get_zone_action()
20121121 18:59:55< BTreeGA> What I'm aiming for is seeing how good of an AI can be automatically created, without doing actual AI programming by me. Theoretically, I could get an 'optimal' AI, but I expect it to fail miserably. Such is research...
20121121 19:00:22< skyfaller> BTreeGA: oh, that sounds fun! I hope it evolves well.
20121121 19:01:46< mattsc> BTreeGA: looking forward to what you find.  And hopefully it won't fail too badly.  :)
20121121 19:01:57< BTreeGA> Thank you. I hope to have it actually evolving on its own next week sometimes.
20121121 19:02:58< mattsc> skyfaller: don't worry too much about the details (unless you want to), but I think having you understand how Fred analyzes things would help you understand why he does things the way he does, and what we might be able to do about it.
20121121 19:03:11< skyfaller> ok, sounds good
20121121 19:03:35< skyfaller> that's another reason I'd like Ron to be independent, I'd like to have an AI that has a slightly different structure
20121121 19:03:37< mattsc> Essentially, it's set up so that we can evaluate CAs either just for a zone or for the map as a whole, depending on how the parameters are set.
20121121 19:44:13-!- mnewton1 [~mnewton1@c-76-99-170-226.hsd1.de.comcast.net] has joined #wesnoth-umc-dev
20121121 20:07:21-!- mattsc [~mattsc@fw.hia.nrc.ca] has quit [Quit: bye]
20121121 20:36:34-!- BTreeGA [~BTreeGA@moose.CS.McGill.CA] has quit [Read error: Connection reset by peer]
20121121 20:55:09-!- Alarantalara [~Adium@CPEc0c1c09e8055-CM00252eac6d62.cpe.net.cable.rogers.com] has joined #wesnoth-umc-dev
20121121 21:21:00-!- mattsc [~mattsc@fw.hia.nrc.ca] has joined #wesnoth-umc-dev
20121121 22:00:32< irker724> AI-Demos: mattsc master * r8035d58 / (5 files): Change AI descriptions slightly - http://git.io/TKiRdA
20121121 22:35:47< irker724> AI-Demos: mattsc master * r0a8fe0b / lua/grunt-rush-Freelands-S1_engine.lua : Fred: retreat_injured bug fix - http://git.io/6mLl0g
20121121 22:39:57< irker724> AI-Demos: mattsc master * ra0811cf / lua/grunt-rush-Freelands-S1_engine.lua : Fred: another retreat_injured bug fix - http://git.io/xoKriQ
20121121 22:55:45< irker724> AI-Demos: Alarantalara master * r3549418 / lua/generic-recruit_engine.lua : Replace all instances of helper with H - http://git.io/nFak7Q
20121121 22:58:12< mattsc> Alarantalara: do you have an estimate of the average win % of Ron vs. RCA AI (same starting gold) across all maps, all factions?
20121121 22:59:12< Alarantalara> Not really. I've been mostly testing recruitment in isolation, which is somewhat different
20121121 22:59:54< Alarantalara> I've watched Ron win in 4 player free for alls and other less normal situations
20121121 23:00:34< Alarantalara> Ron+RCA vs 2 RCA on Siege Castles has gone to Ron 4 of 4 games
20121121 23:00:43< Alarantalara> things like that
20121121 23:01:07< mattsc> Ok.  I really just meant 1v1, but this is good to know too.
20121121 23:01:42< Alarantalara> I usually pick the odd maps, since they;re the ones with something to show
20121121 23:01:56< Alarantalara> Ron has a 100% win on Arcanclave Citadel
20121121 23:02:00< mattsc> Makes sense.
20121121 23:02:27< mattsc> I claimed earlier that Ron almost always wins against RCA, and AI0867 asked whether that's true across all maps.
20121121 23:02:46< mattsc> So that's the context of the question.  I didn't know the answer to that.
20121121 23:03:37< Alarantalara> I would imagine Ron is at least as good as the ML recruiter alone since the recruiting is comparable.
20121121 23:04:18< mattsc> True
20121121 23:04:19< Alarantalara> I could run a set of games using SeattleDad's test script of RCA vs Ron
20121121 23:05:01< Alarantalara> It would take only 5 minutes to set up and I could give you a definite answer tomorrow
20121121 23:05:10< mattsc> That would be interesting, but only if it's no extra effort.
20121121 23:05:20< mattsc> too slow again ...
20121121 23:05:45< mattsc> If it's just 5 min, and you don't mind, it would be interesting to know
20121121 23:08:44< Alarantalara> In that case, I've started it. Let's see what happens.
20121121 23:08:58< mattsc> Great.  Thanks!
20121121 23:14:03< Alarantalara> Apparently a sticking point is going to be vs Drakes on Caves of the Basilisk
20121121 23:14:29< mattsc> ok
20121121 23:14:47< Alarantalara> Everyone but drakes goes around the cave in the centre, so RCA drakes is going to get a high number of leader kills there
20121121 23:15:09< mattsc> I'll try to watch a couple games tonight or so.
20121121 23:25:22< mattsc> Alarantalara: I just watched one game of Ron Loyalists vs. RCA Drakes on Basilisk.
20121121 23:25:47< mattsc> Ron won that pretty handily by grabbing villages and attrition.  Of course, one game doesn't mean anything.
20121121 23:26:17< Alarantalara> You'll notice there were a large number of units running through the middle?
20121121 23:26:45< Alarantalara> Sometimes Ron can win by sending a couple extra units through the center and getting lucky
20121121 23:26:56< mattsc> No, not really.  I think most of them went around.
20121121 23:29:20< Alarantalara> In most human games, it's unusual for more than 1-2 units to go through the center
20121121 23:29:35< Alarantalara> It's generally a bad place to be fighting
20121121 23:29:58< Alarantalara> The AI sends an unusually large number through, and the units on the sides can't get in to stop them easily
20121121 23:30:15< Alarantalara> large = 3
20121121 23:30:53< mattsc> Ah, ok, if that's the definition of large, then yes, Ron sent a large number of units in.
20121121 23:32:16< mattsc> Anyways, it was only a single game, so the statistical error bar is 100%...
20121121 23:32:24< Alarantalara> Of course
20121121 23:33:29< mattsc> On a different note, Fred takes quite a long time for some moves at the moment (esp. when losing toward the end), but I want to make sure that my recent changes work before I do something about that.
20121121 23:35:23< Alarantalara> You just reminded me that I need to finish optimizing the recruit code
20121121 23:35:46< Alarantalara> Evaluating was fast, but it then took about a second to actually pick a unit and hex
20121121 23:35:46< wesbot> Alarantalara: Sometimes we are fast
20121121 23:36:20< mattsc> wesbot likes you
20121121 23:36:25< Alarantalara> Apparently
20121121 23:36:57< Alarantalara> Ron is 18/21 so far
20121121 23:37:10< mattsc> nice
20121121 23:37:12< Alarantalara> since I have to leave for a bit
20121121 23:37:41< mattsc> I really have to do some work anyway.  ttyl.
20121121 23:47:32-!- Crendgrim_ is now known as Crendgrim
20121121 23:50:42-!- mnewton1 [~mnewton1@c-76-99-170-226.hsd1.de.comcast.net] has quit [Quit: Rice and Corn]
20121121 23:51:54-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has quit [Ping timeout: 255 seconds]
--- Log closed Thu Nov 22 00:00:11 2012