[03:58] Join: brx_ joined #corewars [04:00] MSG: Read error: Operation timed out [04:12] Nick Change: brx_ changed nick to brx [08:43] Join: Fluffy joined #corewars [08:43] :) [12:57] * Fluffy is away for about 2 hours ... [12:58] Nick Change: Fluffy changed nick to Fluffyafk [14:37] Join: Mizcu joined #corewars [14:49] Nick Change: Fluffyafk changed nick to Fluffy [14:49] Hi Mizcu :) [14:52] hi Fluff [14:53] hi fluff, h m [14:54] Hi brx :) I've compiled and installed Erlang [14:54] Fluffy: :) [14:55] have you had time to play with it a little? [14:55] Not yet. [14:55] I've updated some of my pages about the Score Surfaces. [14:56] And I've written a page about DiSSC: http://corewars.jgutzeit.de/score_surfaces/dissc/index.en.html [15:01] brx: About Erlang ... does include a way to handle proxys and firewalls nicely? [15:01] Fluffy: ? [15:02] You've said, that Erlang is about distributed applications. So you might run into problems, if there are firewalls ... [15:04] Fluffy: if there are firewalls not under your control? [15:04] for instance [15:04] Fluffy: I don't see how erlang should be able to handle non cooperative firewalls [15:04] Then you've answered the question [15:05] Fluffy: no, you have answered the question :) [15:05] :) [15:06] Fluffy: security is somewhat of an issue. [15:07] Fluffy: thank god the internet belongs to the more trustworthy networks [15:07] err [15:16] bleh, will start writing the parser now [15:18] for what? [15:18] EVE? [15:18] yes [15:18] Where do you need a paser? [15:19] Fluffy: for the final benchmarking against koenigstuhl [15:19] hmmmm [15:20] Fluffy: I have written a ruby script to preprocess *.red files using pmars [15:20] so it won't be too complicated :) [15:20] yes, that's what I've usually done, too [15:23] Fluffy: I don't know why the preprocessor doesn't mod away negative addresses [15:23] As far as I know it is for pretty-printing purposes. [15:23] not a real issue but if it did it the "parser" in erlang would be reduced to a maybe 4 lines of code [15:24] actually, I think it is worth it to make the ruby script handle it [15:26] or, not really, it's enough to remove all whitespace. [15:26] :) [15:27] :) [15:33] Fluffy: http://www.syberpunk.com/images/oolong/pancake3.jpg [15:34] Mizcu: heresy! I am going to burn down syberpunk's embassies [15:34] ? [15:34] Mizcu: Did you take this photo? [15:35] Fluffy: you are fluffy, it is fluffy. [15:35] Fluffy: the relevance is right in front of your eyes :) [15:35] brx: That's not, what I asked [15:35] Fluffy: I wanted to end it quickly [15:36] Fluff: nope [15:37] Mizcu: Nice nonetheless :) [15:37] anyway, depicting the holy messenger like that, intolerable! [15:38] * brx burns Mizcu on a pyre [15:38] i've got fire-resistant underwear on always that im on Irc or posting on usenet [15:39] or any other newsgroup [15:39] insha hamstah [15:39] Mizcu: Now I see, that this is a good idea :) [15:40] * brx prays to fluffy [15:41] I don't want to be prayed at :) [15:41] you are only fluff. [15:41] * Fluffy sighs [15:42] * brx is trying hard to evade writing the "parser" [15:43] I also wonder how to distribute the actual work across different erlang nodes [15:44] any ideas on that fluff? [15:44] maybe I should rephrase, not how to distribute but how to partition the "work" and what work [15:45] * Fluffy tries to think [15:47] You have two general tasks: benchmark a warrior and mutate a warrior [15:47] benchmarking should be on a per warrior basis, i.e. one warrior on one node against benchmark [15:47] mutating ... hmm [15:49] don't know [15:53] there are a lot of options. the simplest one is to not distribute at all, just evolve and send me an email with good warriors :) [15:54] No :) I'd pretend to have created a good warrior all by myself ;-) [15:54] other options involve classification of warriors or warrior snippets [15:54] and depending on that feed nodes with different snippets/warriors [15:56] and then of course, should all nodes benchmark their warriors or should I have dedicated nodes for that which update each other when new warriors are added to the bench [15:57] I can store warriors in a mnesia database, then all nodes have potential access to all information [15:57] I wish barkley was here [15:58] Why not make each noe download the benchmark from a website? [15:58] *node [15:58] Fluffy: because it would be easier to store the benchmark in a mnesia table which is automatically distributed to other mnesia nodes? [15:59] Maybe ... I don't know anything about mnesia [15:59] it's just a database management system which comes with erlang. [15:59] database updates are automatically distributed to other mnesia nodes [16:04] obviously the main advantage of distributing work is to be able breed more / benchmark more in a period of time. I am contemplating what other gains I could get out of it by isolating pools of warriors or balancing them out [16:05] at least the latter already implies warrior classification [16:34] Join: Roy joined #corewars [16:34] Hi Roy! [16:35] brx: It would be cool to have nodes with their own benchmarks that are a bit seperated [16:35] Then you get the islands effect, species evolving in different directions [16:35] Hi fluff [16:35] hrmmm [16:36] Another advantage is that they will never specify too much, beating only the benchmark warriors if people all have different benchmarks [16:37] Will produce better warriors IMO [16:37] so now there is a http://drawmohammed.com/ too -_- [16:38] Roy: I am only going to use static benchmarks, as in koenigstuhl, for a final test anyway [16:38] Well, you have got to have a final benchmark to compare how good the warriors are [16:38] Else you can't really compare.. [16:39] Roy: that is not entirely true. [16:39] Roy: actually, it is not true at all :) [16:40] Why is that..? [16:40] Roy: because as long as you have at least two warriors, you can compare. no matter their origin. [16:41] Roy: you only need a static benchmark like koenig to see a warrior in a specific context [16:41] Thats what I meant, you can easily compare one-on-one but it doesn't say much [16:42] Roy: yes it says very much [16:42] And mizcu: That link... the world is about to lose it... not good [16:42] it says: "this warrior is better than the other" [16:42] now if you have a dynamic bench of total garbage warriors, the best ones will obviously be the ones which survive. [16:42] Yes but it says nothing about if it could make it on the hill or something :) [16:43] over time the bench will be composed mostly of warriors which survive and the better ones are those that also kill [16:44] yes. as I said, it provides a specific context. in the case of koenigstuhl it is the mean raelity [16:44] But if you end up with a whole bunch of very good HSA scanners, and suddenly one stone evolves, the very good HSA scanners will slowly be overpowered by that single stone [16:44] Roy: which is why the pool needs to be balanced out [16:46] That will happen automagicly [16:46] Or never actualy.. the hill for example is also never balanced [16:47] the question is, do I want it to happen automatically for every pool [16:51] Roy: the hill is not a good example for what the pools should be like [16:52] Why not? [16:54] I don't see a reason why it should. it's just human made warriors which can stay on in the current topology. the topology itself serves no purpose but to arbitrarily keep certain warriors from entering [16:54] Join: Metcalf joined #corewars [16:54] Hi [16:54] hi [16:54] Hi JM [16:54] Hi John! [16:55] Have I missed anything? [16:55] not really [16:55] brx: hey [16:55] Fluffy: hey [16:55] Metcalf: I've started round 2 for calculating score surfaces [16:55] http://corewars.jgutzeit.de/score_surfaces/dissc/index.en.html [16:56] brx: I think its pretty good to have in for a pool, it balances itself out, when a new warrior similair to a old one enters the least good one will fall off sooner [16:57] Roy: no. [16:57] Roy: there is no guarantee that it will. [16:57] There is.. the hill is a very good example, after all these years its still balanced [16:58] 17:46 < Roy> Or never actualy.. the hill for example is also never balanced [16:58] :) [16:58] Never perfectly balanced, there is always a strategy least supported [16:59] But it isn't like the hill only contains scanners (or any other strategy) [17:00] Roy: no, but it certainly could if people only sent scanners. which is why the hill has no inherent quality as a model for evolver pools [17:00] Jusr [17:00] ? [17:01] Sorry, wrong window! [17:04] Have you noticed the new Wikipedia Redcode page? [17:04] http://en.wikipedia.org/wiki/Redcode [17:04] ooh [17:05] :-o [17:07] Metcalf: Have you downloaded/bought the CW-mentioning-book yet? [17:07] * Roy is afk, dinner [17:08] I've been away Roy, no chance yet :-( [17:18] Every Corewar website seems to have a different version of Intro to Art in '88 [17:55] back, cool, never noticed the difference [17:56] The first 25 patches are done :) [17:56] They're all broken as well, due to Redcode containing lots of < addressing modes [17:57] I've combined all the versions and fixed the result, and put it at http://corewar.co.uk/mintard [18:18] MSG: Ping timeout: 252 seconds [19:28] Join: lgrabun joined #corewars [19:30] MSG: Client Quit [19:31] damn! missed him :( [19:39] Join: Metcalf joined #corewars [19:39] Roy: who wrote the Exploring Stones article? [19:42] * Metcalf slaps Roy around a bit with a TCL powered popup [19:44] heh [19:44] No idea who wrote that... [19:44] I don't even know what article you are talking about :-) [19:44] Well, wasn't it either you or Nenad? [19:44] Core Explorer issue 2 :-P [19:45] Definitly wasn't me.. then its gotta be nenad :) [19:45] Maybe somebody else, a couple of people helped [19:48] So is Core Explorer 3 coming soon ;-) [20:42] MSG: Ping timeout: 252 seconds [20:45] Join: Fluffy joined #corewars [21:43] hmm [21:51] hrm [21:51] Hmmm? [21:51] Hey, you are all still alive :) [21:52] Yes, I've been busy writing since I arrived [21:52] I'm not sure, what I'll do ... [21:53] Probably I've expected too much and thought, that there are maybe two or three out there running the score surface client ... [21:55] MSG: [22:00] ... [22:00] * brx starts up dissc [22:01] yippie :) [22:01] test version okay? [22:01] yes [22:01] The "official" version at http://corewars.jgutzeit.de/score_surfaces/dissc/index.en.html is the same as the test version [22:02] Quote from that page: "Currently it is worked on 3 patches." :-) [22:02] heh [22:02] you know what's funny [22:02] what? [22:03] I feel strangely coerced into running the client [22:03] and your one comment did it [22:03] I've already mentioned, that I own supernatural powers ;-) [22:04] Tomorrow afternoon I'll probably publish a partial image of the until then calculated data. [22:05] not really. it's just that, although I don't believe in anything anymore, have no hope, no expectations, I just can't take it when people around me mention "losing belief" or realizing that their expectations were moot even if they are only joking :( [22:05] I then feel strangely compelled to do whatever I can to prevent it [22:05] I didn't joke [22:06] coerced actually. [22:06] :) [22:06] so now you all know to avoid being depressed like that, or you will make me feel bad and go out of my way to help. [22:07] brx: How many computers do you own? [22:07] Fluffy: is that a question or an offer? :) [22:07] question [22:07] I officially own 2. though one is currently broken. I have access to a few more. [22:07] In case you answer with a number greater than 1, I probably start to feel very depressed again, because you're only running one client [22:08] * Fluffy is VERRRY depressed now [22:08] ;-) [22:08] Fluffy: I will buy another one in time. and maybe I will fix the old one too. [22:09] buy computers ... what a strange notion [22:10] yes, I know, an absolutely alien concept to you [22:10] Just because I feel, that it is the right moment to mention it ... my brother thinks, that his computer is quite old now [22:11] he's got some "old" Athlon [22:12] * Fluffy has started to compile NetHack ... [22:13] nethack is a nice game. [22:14] Yes, I know. It's been about 5 years since I've played it the last time [22:14] (more or less) [22:14] I wonder why they haven't use its idea in commercial games yet [22:14] +d [22:14] You cannot count those short games in between [22:17] Hehe ... now I've remebered, why my database stores the start AND end time for a patch. [22:18] After a surface was finished I wanted to be calculate stupid things like average calculation speed, ... [22:19] http://bash.org/?609178 :-) [22:30] :) [22:30] average calc speed :] [22:30] Just to see how slow my computer is ... [22:31] At the moment it is about 0.5 ... 0.6 pairs/sec :( [22:31] mine is similar [22:31] ok ... 1.0 .. 1.2 pairs/sec because I've got two processors [22:31] but it is bogged down by other heavy processes [22:31] kill those other processes ;-) [22:31] /quit [22:32] ha ha [22:34] hey ... 32 patches completed :) [22:34] bvowk should come back :) [22:35] Maybe he can spare 100 or 200 computers :) [22:36] probably not :) [22:36] hmm 10 or 20 would nice nonetheless [22:40] Join: Roy joined #corewars [22:40] * Roy gives up... I've lost the ability to make warriors [22:41] Roy: you are probably just burnt out [22:42] If you run THE CLIENT, you'll probably get enough mana to make new warriors :) [22:42] ... good new warriors [22:43] Heh, I might pick up the evolving again if brx does 'it' :) [22:43] does> makes ;-) [22:46] Time to get some sleep ... [22:46] * Fluffy waves [22:46] * Roy waves [22:46] Part: Fluffy left #corewars [22:48] /me waves [22:49] * brx waves [22:50] Roy: I can't give you exactly what you want [22:51] Something close is welcome too :) [22:51] Which part isn't going to work? [22:52] The multilingual [22:52] part? [23:11] yes, fro one thing. [23:41] MSG: Ping timeout: 252 seconds [23:56] Join: Roy joined #corewars