[02:56] Join: Caelian_ joined #corewars [02:59] MSG: Ping timeout: 240 seconds [04:07] MSG: Ping timeout: 240 seconds [04:07] Join: Caelian|w joined #corewars [05:03] MSG: Ping timeout: 240 seconds [05:04] Join: Caelian|w joined #corewars [05:58] Nick Change: Caelian_ changed nick to Caelian [07:08] bvowk: i thought about it some more and i think the primary reason for the initial hill high-scores to steadily decline during the first 10s of cycles is not so much warriors starting to defend against each other ... but more a sign that the "mad suiciders" are gradually being eliminated from the initial breeding pool :) [07:45] bvowk: ping [11:19] Join: fiveop joined #corewars [12:58] eh? [13:07] mornin? [13:18] bvowk: hehehe ... my typical way of saying "get back to me when you're there" :) [13:20] bvowk: i was wondering ... i know Ruby has the concept of "distributed ruby" .. allowing one to use objects provided by a remote system as if they were local .... i wonder (assuming python has something similar) if it would be worth creating a "remote valhalla" that several evolvers could test against as if it were local .. thereby directly profiting from the evolving work from other evolvers testing against the same "remote valhalla" :) [13:27] I just use a tcp socket and send the warriors through, pycorewar parses them on the other end.. [13:27] didn't make much sense with nano, because one computer seemed to be enough for the most part [13:30] ok :) [14:29] bvowk: btw .. i am going to try and see if i can actually write a "ruby-corewar" extension based on the PyCorewar code [14:29] should make for an interesting project :) [14:47] I think there are more marses than corewarriors ;) [14:47] hehehehe [14:47] I wrote a scheme and a CL extension w/ exhaust [14:50] I'm amazed there isn't a perl one personally [14:51] ~ [14:51] given that I wrote a pile of corewars related perl code.. [14:51] HWMNBM had a huge wad of perl for SoV [14:57] well .. what are you waiting for ... go write one :)) [14:57] * Caelian|w ducks for cover [14:57] heh [14:57] I don't write anything other than one liners in perl anymore really. [14:57] anything I want to use more than once.. [14:58] too many times I looked at a perl file and said.. "WTF! what idiot wrote this!" and then realized by the CVS stamp that it was me. [14:59] doesn't happen quite so much with python ;) [15:00] lol [15:00] don't get me wrong tho, I still emit an awful lot of perl. [15:00] because its everywhere, and just works for the most part.. [15:01] hmmm ... just looking at some of the header files in the PyCorewar code makes me think i should do better .... a lot of the definitions feel like they're rather hacked together and could be done much more elegantly [15:01] I'm sure fluffy wouldn't mind the patches.. [15:01] which version you using? [15:02] the jnsl1.1.tar,gz you send me [15:02] PyCorewar-0.2.0.28 is the newest.. [15:02] so that would be 0.2.0.28 [15:02] ok.. [15:02] good :) [15:03] just checking, I remembered a memory leak in the previous version when I was hacking nano.. [15:03] well ... i'll most likely be using his code as a reference as i can't use it verbatim in any way because of his code is intended for python .. mine will be for ruby [15:04] i'll see if i can get anything significant going there before starting to suggest somebody else to do things differently though :)) [15:13] one of the things i think could be done more elegantly e.g. is instead of using the whole bunch of INSN and related macros .... it should be doable to simply use bitfields instead [15:13] time will tell though :) [15:14] hmmm .. .i should probably just shut up and see what i can get done first ... because i think i already see why i would probably need those macros any way [15:28] what we need is a extremly optmised C/ASM mars! [15:29] Write a QEMU module :-) [15:29] Free JIT. [15:31] there is Qmars, which is pretty goddamn fast [15:31] go ahead and make something partially based on it [15:31] qemu? [15:32] Mizcu: brx tried that, iirc even with michal [15:32] at least he wrote some lines of scheme to generate asm :P [15:32] and optimise them and what not [15:34] http://fabrice.bellard.free.fr/qemu/ [15:51] Join: Core_new joined #corewars [15:58] fmars is basicly an asm mars [15:58] iirc it takes a scheme script and compiles a custom mars based on the settings.. [15:58] uses mmx/sse [15:58] its stupidly fast, but you can't make it support *ALL* possible instructions [16:15] looks like the current warrior in Yace hasnt changed composure for 1 hour [16:15] time to kill and restart [16:16] no wonder, old settings in [16:28] bvowk: because of gcc limitations? [16:28] yeah, it makes a huge wad of unrolled code [17:02] Join: sf_ghoul joined #corewars [17:02] * bvowk throws a water balloon at sf! [17:02] * sf_ghoul catches :) [17:04] hrm. [17:04] MSG: Quit: Trillian (http://www.ceruleanstudios.com [17:04] * bvowk uses a garden hose on sf instead then. [17:04] Join: Core_new joined #corewars [17:05] Thanks, I needed cooling down :) [17:09] no problem. [17:19] an imp-pump. Really creative from the evolver. Blah. [18:51] MSG: Quit: Ex-Chat [19:04] MSG: Quit: Get out of that boring IRC client! It's no good for you. Bersirc 2.2 is your answer! [ http://www.bersirc.org/ [19:10] hrm. [19:10] rhm [20:43] Join: Caelian joined #corewars [21:40] MSG: Quit: Trillian (http://www.ceruleanstudios.com [22:15] ok ... i am slowly crawling towards the bottom of the nano hill :) [22:15] latest submission scored: 51 39.7 53.2 7.1 Eyeless Moon caelian 126.1 0 [22:16] heh [22:16] how [22:16] how many computrons? [22:16] 1 [22:17] assuming you're referring to the number of computers used [22:17] yeah, what is it? [22:18] Athlon64 3800+ 2GB RAM [22:18] cause if I say one, it could be an 8way opteron or a 1Ghz duron ;) [22:18] 754 or 939? [22:18] or M [22:19] Socket AM2 [22:19] not sure about the pin count [22:23] http://pastebin.ca/643602 <--- that's the code belonging to my last submission btw ... in case you're interested [22:26] :) [22:26] looks like something I've evolved :) [22:26] * bvowk claims prior art and credit ;) [22:27] hehehe [22:27] i think with limited understanding that it probably is trying to be some sort of "paper" like construct [22:27] they usually don't score that well without the djn [22:27] actually ... this one scores better than the number 2 ... which Does include a DJN [22:32] Join: Selene joined #corewars [22:32] MSG: Client Quit [22:36] an interesting difference between this one .. and the one including the DJN as the final instruction is that the DJN version uses SPL.B ... instead of SPL.X [22:36] don't really if it's anything significant though :) [22:36] * Caelian adds in a "know" in that sentence