08:33:47: -!- Doxin has joined #corewars.
08:33:47: -!- Doxin has quit (Changing host).
08:33:47: -!- Doxin has joined #corewars.
12:06:00: -!- Doxin has quit (Remote host closed the connection).
12:51:43: -!- Doxin has joined #corewars.
12:51:43: -!- Doxin has quit (Changing host).
12:51:43: -!- Doxin has joined #corewars.
15:44:55: -!- Manji_ has joined #corewars.
15:45:11: <Manji_> hi
15:45:30: <Manji_> can anyone give me a few hints about tweaking this warrior?
15:48:22: <Manji_> http://www.corewar.eu/corepaste/id.php?id=11
15:50:32: <Manji_> btw the imp is shamelessly copied from bluefunk
15:58:40: <Manji_> I would like to kill the bombing engine's spl but I have no idea how.
15:59:40: <john_metcalf> Hi :-)
16:00:22: <Manji_> hi
16:00:24: <john_metcalf> I'm just about to eat, will take a look later.
16:00:41: <Manji_> I dont force you to help me ;D
16:13:18: -!- Doxin has quit (Ping timeout: 264 seconds).
16:34:18: -!- madsara has joined #corewars.
16:43:38: <Manji_> hi
16:53:09: <Manji_> does anyone see a major improvement for my warrior? I dont want to fine tune it, it was intended to be a proof of concept. but with a score under 100 against wilfiz its doesnt proof anything ;D
17:22:04: <john_metcalf> Hi Madsara
17:33:32: <john_metcalf> Manji_: some ideas 1) a gap between the launcher and imp can help against certain scanners 2) put the clear closer to the gate than the stone so more of the core is cleared (careful not to SPL wipe your stone though) 3) try a better clear, e.g. this one from Flux http://corewar.co.uk/infitiny/FLUX.RED 4) add another bombing line MOV XXX, @SPY. It won't be a spy bomb, but it might help
17:40:14: <madsara> Hey john_metcalf
17:40:20: <Manji_> thx
17:40:31: <madsara> So hey, quick question...
17:40:48: <madsara> Some of these warriors copy themselves into memory, are they spawning multiple instances of themselves?
17:40:56: <Manji_> yes
17:41:08: <madsara> And you have to kill each and every one of them, right?
17:41:36: <Manji_> yes. with the spl command
17:41:43: <Manji_> a new process is queued
17:41:51: <Manji_> AFTER all others
17:42:09: <Manji_> if you want to win, you have to kill all these processes
17:42:47: <Manji_> but since there are more they usually take longer to do something. so you have in a sense a speed advantage
17:42:53: <madsara> I've never played, but this is really interesting (I've worked in robocode and ATrobots2)
17:43:30: <Manji_> check out this site
17:43:31: <Manji_> http://corewar.co.uk
17:43:46: <madsara> And DAT kills a process, so if you copy a DAT instruction to a memory address, and your opponent jumps to that address it causes it to kill itself, is that the gest of the bombers?
17:44:18: <Manji_> basically yes. although some prefer to call this technique stone
17:44:24: <madsara> Yeah.
17:44:28: <Manji_> note that executing the dat
17:44:32: <Manji_> takes one cycle
17:44:48: <Manji_> and pre and postincrements get executed as well
17:44:59: <madsara> So they could potentially spawn another instance, then kill it's original self?
17:45:09: <Manji_> yes.
17:46:02: <madsara> Nice.
17:46:06: <madsara> Yeah, this seems like a fun game
17:46:13: <Manji_> this is quite easy to do. the advantage is that you are a lot faster and it is difficult to catch you. on the other hand you need enough instances to survive (using papers)
17:47:01: <Manji_> although I have to say papers is one of the more difficult topics.
17:47:44: <Manji_> scanners and bombers are easier to understand and also easier to write some better warriors. papers take a lot of tuning
17:48:25: <Manji_> at http://sal.discontinuity.info/hill.php?key=94b you find a hill for beginners :) feel free to submit!
17:48:43: <madsara> So, a scanner... what does it do exactly? It doesn't actually read addresses does it/
17:49:14: <madsara> Actually I found the webpages on it, I'll look at that. brb
17:49:22: <Manji_> seq and sne can compare 2 "cells"
17:50:35: <Manji_> but the highly depend on the modifiers
17:50:40: <Manji_> like .x .i .f
17:53:31: <Manji_> @john: I cant see the advantage of the clear from flux
17:54:06: <Manji_> because there is no real "gate"?
17:59:11: <Manji_> mm......looks like I dont get the clear ;D
18:00:44: <Manji_> so its a stream directed clear
18:01:02: <Manji_> quite interesting, havent seen a clear using div yet
18:04:12: <madsara> Hey Manji_ --what's SNE.X do vs SNE ?
18:04:27: <madsara> I see the modifier on the opcodes page, just don't understand that offhand.
18:06:10: <Manji_> w8
18:10:07: <Manji_> this is kind of confusing I have to admit and im not sure if I can properly tell. so it compares the a value of adress a with the b value of adress b and the b value of adress a with the a value of adress b
18:10:46: <Manji_> but I could be wrong here. hopefully john joins the discussion to help :)
18:14:04: <Manji_> so for example sne.x 2,3 jmp 0 dat 1337,
18:14:13: <Manji_> >.< dat 1337,7331
18:14:18: <Manji_> dat 7331,1337
18:16:12: <Manji_> http://www.corewar.eu/corepaste/id.php?id=12
18:21:37: <Manji_> note the result is the same if you change one dat to nop for example
18:23:51: <madsara> Why sne.x vs sne though... in both instances it skips the jump, rihgt?
18:24:40: <madsara> Oh, oh oh..
18:24:43: <madsara> No I see...
18:24:52: <madsara> .x A->B AND B->A
18:28:26: <Manji_> better write that out. sne is usually sne.i
18:28:34: <Manji_> and .i also takes into account
18:28:42: <Manji_> the command, not only the values
18:29:37: <Manji_> and sne.i and sne.f both just do their a: a==a? b: b==b? thing
18:32:32: <Manji_> do you use corewin?
18:32:47: <Manji_> using pmars can be quite a hassle
18:40:48: <madsara> I was working with pmars, but only for a second.
19:06:04: <Manji_> ,D
19:06:16: <Manji_> I think corewin also runs in whine
19:13:12: <john_metcalf> Manji_: you could make the clear faster and add a proper gate like this http://www.corewar.eu/corepaste/id.php?id=13
19:13:30: <john_metcalf> That scores 100+ points on 94nop on it's own
19:34:14: <john_metcalf> Each memory location contains a complete Redcode instruction. .I uses/compares the entire instruction. .F uses/compares the two operands only. .X uses/compares the two operands with the first set of operands reversed.
19:57:28: -!- Doxin has joined #corewars.
20:21:43: <Manji_> ah I remember now. :D
20:21:54: <Manji_> thx john
20:21:57: <Manji_> hi doxin
20:22:05: <Doxin> ohai Manji_
20:22:31: -!- Manji_ is now known as lain2.
21:09:13: <john_metcalf> Hmmm... I need a copy of Redcoder 1.0 by Alex MacAulay to complete the set. (I have 2.0, 2.1 and 2.1.1!)
21:13:03: <madsara> john_metcalf: What does your copy pase do exactly?
21:13:13: <madsara> imp gate?
21:16:33: <john_metcalf> Briefly, it clears the core sequentially. First with SPL which slows down the opponent then with DAT which kills the opponent. The clear pointer is located just above the code where it can double as an imp gate.
21:26:04: <lain2> also, its stream directed
22:03:18: <lain2> thx john, the change to the .5 bombing engine scores over 10 pts more
22:16:48: <john_metcalf> :-)
22:16:57: <john_metcalf> Have you tried it on the multiwarrior hill?
22:29:59: <lain2> not yet :) well, nevertheless the warrior scores worse than the clear alone. guess I have to boot away
22:34:23: <lain2> no multiwarrior entry too
22:34:51: <lain2> probably it comes down now to a different imp and tweaking constants?
23:10:42: -!- Doxin has quit (Remote host closed the connection).