[00:03] MSG: Read error: Operation timed out [00:34] MSG: [07:31] MSG: Read error: Operation timed out [12:08] Join: datagram joined #corewars [12:41] Join: Mizcu joined #corewars [12:49] Join: inversed joined #corewars [12:50] Hi Miz and Data ! [12:50] hi inversed [12:51] how's it going? [12:51] Quite good. Today my provider gives a present - free inet, so I can chat all the day :p [12:52] nice [12:59] Why don't you publish static? [13:02] i dont have the source anymore [13:02] i know the stone and its stepsize, and implaunchers relative distance to stone, but the rest is blank [13:03] That's why I backup regulary [13:05] blownup HD.. [13:05] Exploded? [13:06] head crash [13:15] Join: Core29 joined #corewars [14:00] Join: Judodudo joined #corewars [14:00] Hi [14:00] hi Judo [14:00] hi JoodDood ;) [14:04] ? Loads of people dropped off 94b. There's only 22 warriors! :-/ [14:05] well, more space for warriors [14:05] I've killed many my warriors (~5) [14:06] Blotter was doing good on 94nop, some other need to be improved, other just suck [14:11] Myabe I'll kill some more [14:11] *Maybe [14:18] ahh [14:18] I thought maybe the age limit has been imposed [14:34] MSG: Ping timeout: 252 seconds [15:00] Join: Fluffy joined #corewars [15:00] :) [15:00] * Fluffy is just quickvisiting ... [15:05] PyCorewar 0.1.0 seems to be ready ... I'm uploading it now [15:05] Join: pak21 joined #corewars [15:06] Hi pak21! [15:06] With source? [15:06] of PyCorewar [15:06] * pak21 waves [15:06] you meant the URL, right? [15:06] (Just logging on again after a power cut killed my server) [15:07] killed = dead forever? [15:07] I mean, will you upload .exe and python source code? [15:08] Dunno. It's not owned by me, but I guess it will be back sometime soon when the admins get physical access. I suspect it's just sitting at a fsck prompt. [15:08] of course the source code [15:08] I have one very cool idea that can be used in evolvers and optimizers [15:09] URL of PyCorewar is: http://corewars.jgutzeit.de/pycorewar/pycorewar-0.1.0.tar.bz2 [15:09] Documentation and a proper homepage will be uploaded tomorrow [15:11] I have a pascal-written optimizer that uses this idea, but it uses pMars; it executes it too often so it's quite slow [15:11] Yes, I know that problem from the score surfaces. It was one of the reasons to write a fast MARS with a simple Python interface [15:12] The current version on support ICWS '88, but is at least faster than pMARS [15:12] about 2.1-2.4 times the speed of pMARS [15:12] still not as fast as fmars [15:13] What about jmp 0, nprocs and ncells surfaces? [15:13] not yet :-( [15:13] I can't multitask properly [15:14] So there is only one active project at the time (which is PyCorewar) [15:14] org Fluff Fluff: spl ScoreSurf jmp PyCorewar [15:15] hehe [15:16] (input error: Not a valid program. Labels ScoreSurf and PyCorewar are missing) [15:17] Can you make a prog that runs set of warriors against testset using your PyCorewar? I really need it [15:20] sth. like benchmark(warrior, (benchwarrir1, benchmarwarrior2, benchwarrior3, ...), rounds, fvalue) [15:20] ? [15:20] with or without threads? [15:21] benchmark((testwarr1, testwarr2, ...), (benchwarr1, benchwarr2, ...)) [15:23] but with a twist: continious pruning; that is: if some warrior is surely worse than others, don't benchmark it further [15:24] surely worse means that it's score is less than (maxscore-2*delta), where delta=130/sqrt(nrounds) [15:25] This technique speeds up everything a lot if you only need to know what is the best warrior [15:28] hmm [15:29] So it should run a few rounds, see what warriors are surely bad and continue with others [15:30] I'm still not sure, that I completely understand, what you want [15:30] In any case today's time is over [15:30] (sorry) [15:30] * Fluffy waves [15:30] Part: Fluffy left #corewars [15:30] bye [16:24] MSG: Ping timeout: 252 seconds [17:14] Join: pkhuong joined #corewars [17:17] MSG: [17:28] MSG: Ping timeout: 252 seconds [18:08] Join: inversed joined #corewars [18:20] MSG: Quit: Trillian (http://www.ceruleanstudios.com [19:10] MSG: Quit: Changing host [19:13] Join: pak21 joined #corewars [19:18] Join: fiz_afk joined #corewars [19:18] hi [19:18] Nick Change: fiz_afk changed nick to Fizmo [19:18] hi inversed. A happy new year [19:20] still in progress giving the fsh94nop0.3 the final shape [19:24] will be back soon [19:24] MSG: [20:17] Join: Fizmo joined #corewars [20:17] back [20:18] Will I see qutrum in scn? [20:18] Join: datagram joined #corewars [20:19] hi inversed [20:19] maybe. I am still working on pws [20:19] hi Fiz [20:20] you're in pap with PhatPaper [20:20] omghax [20:20] I return. [20:20] hi data [20:21] Fiz: maybe you should add PhatPaper's parted silk description to Paper manual [20:21] yes, it's really a nice idea [20:22] I really ought update my paper manual as well :-/ [20:22] you still have a easy to notice typo in it [20:24] I guess there are many typos in :( [20:24] I was optimizing PhatPaper against KS top 50 and result is somewhat overfitted [20:25] It's N4 in ks top 50 but it wasn't succesfull on the hill [20:25] but feel free to give me a notice, I'll eliminate the typo then [20:26] inversed: I think that is bad luck. Maybe the wrong time for this strategy [20:26] the commentary of Shrinking paper has "HOLE paper" [20:29] ups, yes [20:33] PhatPaper scores 132.01 against the 94nop0.3 [20:34] so it's in the upper half of all pap [20:34] :) [20:36] the best paper in the benchmark is The Human Machine which scores incredible 147.62 [20:36] Hullabaloo's score is with 146.48 lower than that paper [20:37] Never heard of it, what is it? [20:37] there will be a lot of unpublished warriors in it :-P [20:39] it's bvowks evolved paper with some handcoded add-ons by me. BTW, it looks very similar to Jens' Harmless Fluffy Puffy [20:39] Join: Philb joined #corewars [20:40] hi Phil :) [20:40] A Happy New Year \o/ [20:40] Hi Fizmo, sleepers, Happy New Core [20:40] * Philb was hoping to find you here [20:40] Topic Change: Mizcu sets topic: www.koth.org - www.corewar.info - corewar.co.uk [20:41] now you catched me ;-) [20:41] Is there a R25 of Frenzy? [20:42] I was updating some things and saw R24 had finished from the newsgroup [20:42] well, maybe. [20:42] There are some who want to organize RF25 [20:43] I'm not agitiating for a round to begin, just wanted to be sure I was up-to-date :) [20:43] :) [20:43] You missed a really nice CSEC2005 [20:44] I have much to catch-up. Just seen the score-surfact/probablity based threads from Aug. Nice [20:44] I hope'll participate on the next one. I'll surely do :-) [20:45] hoping that somebody else will host it [20:45] I think we could do with a better way to categorize warriors. [20:46] So as to more easily distinguish between variations on a theme. [20:46] And find the current state-of-the-art of various types. [20:47] This is partly motivated by seeing Reepicheep is at #15 on NOP top 50. It's better than that but everyone has optimised or cloned it. [20:47] Ok, I am paying attention ... yes, I hope you get to compete in the next round. Should be mandatory! [20:48] I don't just mean state-of-the-art now, I mean at any momenet in time. [20:49] yeah, categorize warriors is a pretty difficult thing. It takes me a long time to do that for optimax [20:50] Optimax has differnt goals though becuase it is looking for most representative warriors at the current time. [20:50] Googling for HFB gives http://tarshaan.moonlit-eyrie.com/stories/bunny1.html [20:50] The number of distinct variants/techniques around is much larger and some although clever aren't competetive. [20:51] Um, too much too read. [20:52] yes. I once proposed a kind of trading card game but instead of cards with different strength one use cw warriors [20:53] clever non-allround warrior would've been important as well [20:53] Fiz: that was mine idea too! [20:53] cool :) [20:54] The problem is how to realize this.... [20:55] Join: pkhuong joined #corewars [20:55] because it should be continuing an online game [20:56] "i tap koth and summon HFB" sounds bad to me [20:56] BTW, there exists pure online kinds of trading card games [20:56] I play my magic card reducing the max. process by half [20:57] I also had an idea to cross corewar with decker [20:58] * Philb watches usenet to see if anyone proposes Joonas's SDL version [20:58] Mary Decker? [20:59] this decker: http://www.caro.net/dsi/decker/ [20:59] ugh, that decker [21:00] Sorry, it was the 2nd i thought of after Black & Decker [21:00] lol [21:00] heh. Isn't there a generic card game? [21:01] the games mechanics have been ripped off shadowrun [21:02] Yes, http://www.caro.net/dsi/decker/ . It'd be cool to write progs yourself in redcode [21:03] generic card game? [21:03] Hi Paul [21:03] hi [21:04] Fizmo: yes. Like nomics, but for cards [21:07] and WotC now sells MTG cards solely for net play :/ [21:07] hmm, with corewar you would need only programming skills [21:08] and koen ;-) [21:09] And I still want to make corewar/decker-style HL mod [21:09] (IMHO, of course) even worse idea [21:10] but, you do what you want [21:10] it would give cw a totally new aspect, beside the hills and tournaments [21:11] the problem is to make a software/online-platform for that [21:15] * Philb waves and off to listen to neighbourhood car alarms [21:16] have fun [21:16] Part: Philb left #corewars [21:17] I know that distopya HL2 mod has some kind of cyberspace battles, but I never saw it [21:18] I saw it on HL2 and could dl the demo, but it looks somehow a bit strange to me [21:19] Tell me if you try it [21:20] Well, maybe I'll give it a try the coming week [21:32] Now I'm working on scn [21:32] evaluating Qutrum [21:36] inversed: Qutrum is in scn of the fsh94nop0.3 [21:40] I have scanner that scores ~145 fsh0.2 but it can't enter the hill [21:42] cds with original zoom trick [21:43] hmm, that's strange [21:43] there are too much pap/pwi/pws on the hill which nicely beats cds [21:44] should be a good time for Qutrum ;-) [21:45] I guess you would join me in a top spot on the 94nop hill :-P [21:47] Maybe, but I don't like the idea of simply resubmitting pushed off warriors [21:48] There is some space for an improvement [21:48] add a decoymaker. That will give you some few extrapoints [21:49] that one from HSA would be perfect but you have to adjust it to your scanner [21:50] Phew, done. 94nop0.3 is ready now [21:50] I've tryed and it doesn't helped (but I was a noob). I'll try again [21:53] wow, 12 so far unpublished warriors in the new benchmark.... [21:54] Where can I download it? [21:54] Wonder if i should recreate Static.. though there is nothing to add without making notable changes [21:55] in some few minutes on the optimax homepage [22:07] ok, ready to download :) [22:08] http://www.corewar.info/optimax/dload/ [22:09] but some of the warriors I'll publish on r.g.c. because they need some more explainations [22:09] because they are quite new [22:09] and unique [22:13] You've revealed password of Grabun [22:14] in which warrior? [22:14] Cracking Ice [22:17] uhh. That's strange I am pretty sure that he published this version on r.g.c. but I didn't find it at koen now... [22:17] BTW, check out in the pwi folder S.D.N. and tell me what you think about this idea [22:17] ;-) [22:17] ooh, i get to see SDN too [22:18] YAY, SDN! [22:18] Mandragora: I think I know better way to not to stun imp launcher: make it suicidal [22:18] e.g. boot the spiral not too far behind the launcher. [22:19] Not necesary, you can have djn imp, #time/mov kill, implaunch [22:21] if the launcher is multi threaded, that'll not work as expected. [22:21] mov.i #0, -1 [22:22] It'll work ok: one process will fall through djn and after some time it will move dat over spl of launcher. [22:25] back from smoking [22:27] well, Mandragore seems working ok on the hill but I am sure that it can be further improved [22:29] all constants are heavily optimized, so the actuall positioning of the components and the spiral seems almost perfect [22:30] but you're right maybe it would be better to suicid the imp-launcher, like in Digitalis 2002 [22:33] It only makes partial spl wipe, so there may be some active paper copies when dat-wipe starts. [22:34] So it can't beat some papers (PhatPaper for example), while disharmonious can [22:34] Slime still part of the benchmark!? [22:37] Similar results for NPaper2: Mandragora 132 Disharmonious 166 [22:38] Probably same results for any timescape-style paper [22:38] Maybe soon you'll need sabi dir [22:39] yes, it know that, but on the other side it is less vulnerable against scanner [22:39] Why? [22:41] the clear is smaller and less visible, and I optimized it especially against scanner [22:42] maybe there are better constants against papers [22:42] You can have any clear with suicidal implauncher [22:46] that's true. Well, I've choosen this one which seems to do a good job. But as I mentioned I am sure that there is plenty of possibilities to further improve it [22:46] It would be nice to see some more warriors of this kind on the hill ;-) [22:47] I think I'll make a remix of mandragora with suicidal implauncher (poor p/s ;) [22:48] :-P [22:53] posted S.D.N. on r.g.c. with a little comment [22:54] Join: Core29 joined #corewars [22:56] I had this idea when I was a newb, but never managed to make it work properly. Now expect SDN clones [22:58] It has small nprocs (relative to papers), I'm surprised that it works good against scanners [22:58] go ahead ;-) Other player surely have other ideas and further development starts [22:59] it works good against scanner [23:00] but I was a bit dissappointed that it was pushed off after a while, because it got good scores against all papers and stone/papers because of the fast boot [23:01] and I expected that it stays much longer on the hill [23:02] At what age it died? [23:02] just a score against Qutrum [23:02] 25.20 [23:02] 58.00 [23:02] 16.80 [23:02] 92.40 [23:02] 190.80 [23:02] erhm, SDN wins 58% [23:03] I can't remember, but it was somewhere between 100 and 200 [23:04] against paper and stone/paper it scores usually around 125-150 [23:05] ugh, 30 mins cleaning character sheet and writing stuff there [23:05] maybe it was a bad time with some paper/imp killing scanner on the hill [23:05] And now i have to put it to paper [23:06] Fiz: arr*gough* [23:06] ;-) [23:07] inversed: Also have a look to sai: The Utterer. A very funny idea which surprisingly seems to work ok [23:08] especially my '88 code versions [23:09] Wow, only Recon2 can beat SDN [23:12] really? Here SDN wins 40% and lose only 34% [23:13] 250 rounds with exmars [23:13] BTW, I forgot to mention that the imp-launcher in Mandragora splits its processes back to clear which makes the clear running faster than usual [23:14] A snippet I am using in most of my stone/imps [23:14] I am still wondering why nobody is using this snippet beside me [23:15] It's only usefull with vortex-launcher IMO [23:15] yes it is [23:16] SDN vs Recon2 1000 rounds SND wins 36.4% losses 40.3% [23:16] *SDN [23:17] Anyway it's surprising that only one scanner can beat it [23:17] ok, but it's still ok ;-) [23:19] Oneshots with anti-imp clear can beat it too [23:19] yes [23:20] also Crackling Ice and Fuse [23:22] No, Cracking Ice doesn't beats it [23:23] I wonder if SDN can be adapted for LP [23:24] hmm, I tried it only for the tiny hill without success [23:28] I'm also happy to see other unpublished warriors [23:30] :) [23:35] crude version 112 wilfiz.. [23:35] *sigh* [23:37] Reconstruction of static? [23:38] y [23:59] ok, time to go for me [23:59] * Fizmo waves [23:59] bye [23:59] bye [23:59] MSG: