--- Log opened Fri Jan 16 00:00:25 2004
01:59:01 -!- bvowk_ [~bvowk@h24-67-137-90.ed.shawcable.net] has joined #corewars
01:59:03 #corewars: < bvowk_> hrm
01:59:08 -!- bvowk_ [~bvowk@h24-67-137-90.ed.shawcable.net] has quit [using sirc version 2.211+ssfe]
04:59:22 -!- bvowk [~bvowk@h24-67-137-90.ed.shawcable.net] has quit [Remote closed the connection]
11:23:22 -!- Roy [~roy_van_r@guest-dhcp-164.si.hhs.nl] has joined #corewars
11:23:28 #corewars: < Roy> Howdy
13:26:13 -!- Roy [~roy_van_r@guest-dhcp-164.si.hhs.nl] has quit []
16:29:37 -!- CoreChild [~d432ba35@200.42.38.150] has joined #corewars
16:30:01 -!- CoreChild [~d432ba35@200.42.38.150] has quit [WebChat -> www.vIRCio.org  (EOF)]
16:39:15 -!- vIRCio602 [~d432ba35@200.42.38.150] has joined #corewars
16:39:30 #corewars: < vIRCio602> Hi
16:40:20 #corewars:  * vIRCio602 is CoreChild in disguise.
16:45:19 -!- CoreChild [~d432ba35@200.42.38.150] has joined #corewars
16:55:27 #corewars:  * CoreChild waves
16:56:01 -!- CoreChild [~d432ba35@200.42.38.150] has quit [WebChat -> www.vIRCio.org ]
16:56:12 -!- vIRCio602 [~d432ba35@200.42.38.150] has quit [WebChat -> www.vIRCio.org ]
17:57:06 -!- joonas [jpihlaja@kruuna.helsinki.fi] has joined #corewars
19:06:03 -!- michal [~michal@august.V-LO.krakow.pl] has joined #corewars
19:06:10 #corewars: < michal> hi
19:07:27 #corewars: < joonas> hi michal
19:07:57 #corewars: < joonas> still working on a mars?
19:08:46 #corewars: < michal> heh
19:09:13 #corewars: < michal> actually I've made a very smart or very stupid move, depending how you look on it
19:09:26 #corewars: < joonas> oh?
19:09:41 #corewars: < michal> I talked to my tutor, and he agreed to make it [and related stuff] my annual assigment
19:09:52 #corewars: < joonas> lol, good luck. :)
19:10:13 #corewars: < michal> I decided not to make it in asm, though
19:10:27 #corewars: < joonas> oh.  damn.
19:10:30 #corewars: < joonas> why not?
19:10:45 #corewars: < michal> in gcc, with some very bad style [lots of gotos etc.]
19:11:00 #corewars: < michal> just too much work for too little gain for me.
19:12:02 #corewars: < michal> I also think I found a bug in gcc, when doing 'research'
19:13:39 #corewars: < joonas> I'm not surprised.
19:13:49 #corewars: < joonas> what's the bug?
19:14:47 #corewars: < michal> gotos to pointers [that's gcc non-standard extension] are compiled wrong
19:15:13 #corewars: < joonas> really? which version of gcc?
19:15:39 #corewars: < joonas> that's a huge bug.
19:16:03 #corewars: < michal> hm. hard to say.. I saw one 2.95.x version working well
19:16:17 #corewars: < michal> and rest of them didn't
19:16:28 #corewars: < joonas> I know the guys writing gforth have been struggling with the 3.x series a lot.
19:16:43 #corewars: < michal> [including 3.x]
19:17:00 #corewars: < joonas> they've got it working on most of them and they do use computed gotos a lot.
19:17:59 #corewars: < michal> actually, I wasn't _very_ surprised, since I was doing things most sane people wouldn't do - jumping in and out of functions
19:18:14 #corewars: < joonas> oh.
19:18:25 #corewars: < joonas> hm. I think that's not supported at all actually.
19:18:57 #corewars: < joonas> IIRC, what is supported is doing computed gotos within a function.
19:19:22 #corewars: < joonas> but all the labels you're gotoing to indirectly have to be within the same functions.
19:20:56 #corewars: < michal> and if they aren't? ie. if I just decide to make up a pointer and jump to it?
19:21:08 #corewars: < joonas> from gcc.info:    You can use this mechanism to jump to code in a different function.
19:21:08 #corewars: < joonas> If you do that, totally unpredictable things will happen.  The best way
19:21:08 #corewars: < joonas> to avoid this is to store the label address only in automatic variables
19:21:08 #corewars: < joonas> and never pass it as an argument.
19:21:28 #corewars: < joonas> (that's from the info to gcc 2.95.4)
19:22:28 #corewars: < joonas> in asm you could by-pass all of this, of course.
19:24:17 #corewars: < michal> right.. but when I want to make a jump to a pointer at the end of a function, the compiler just picks one of the labels and makes a hardcoded goto
19:24:36 #corewars: < michal> [sometimes]
19:28:14 #corewars: < joonas> that's really odd.
19:29:51 #corewars: < joonas> ...and this is happening even when you're jumping within the function?
19:31:49 #corewars: < michal> I think it isn't
19:33:57 #corewars: < michal> no, it is :)
19:34:34 #corewars: < michal> the compiler is to smart for me..
19:34:39 #corewars: < joonas> I just tried it with gcc 3.0.4 and can't reproduce it. :-(
19:34:44 #corewars: < joonas> also with 2.95.4
19:35:36 #corewars: < joonas> what flags are you compiling with?
19:35:45 #corewars: < joonas> here's my test function btw:
19:35:49 #corewars: < joonas> void test(int i) {
19:35:50 #corewars: < joonas>     static void * const labs[] = { &&lab1, &&lab2 };
19:35:50 #corewars: < joonas>     void *ptr = labs[i];
19:35:50 #corewars: < joonas>  lab1: goto *ptr;
19:35:50 #corewars: < joonas>  lab2: goto *ptr;
19:35:52 #corewars: < joonas> }
19:36:12 #corewars: < joonas> and that compiles to: 
19:36:13 #corewars: < joonas> test:
19:36:13 #corewars: < joonas> movl4(%esp), %eax
19:36:13 #corewars: < joonas> movllabs.0(,%eax,4), %eax
19:36:13 #corewars: < joonas> .L2:
19:36:13 #corewars: < joonas> jmp*%eax
19:36:14 #corewars: < joonas> .p2align 2
19:36:16 #corewars: < joonas> .L3:
19:36:18 #corewars: < joonas> jmp*%eax
19:36:20 #corewars: < joonas> .Lfe1:
19:36:30 #corewars: < joonas> sorry about the tabs dissapearing.
19:38:28 #corewars: < michal> mhm. this one works for me too
19:39:48 #corewars: < michal> this seems to happen when I want it to jump via a global pointer
19:41:04 #corewars: < joonas> I changed the void *ptr; to be a global ptr but it still compiles find.
19:41:06 #corewars: < joonas> fine, even.
19:41:31 #corewars: < joonas> I get a "jmp *ptr" insn as the final insn.
19:43:29 #corewars: < joonas> do you use a global register assignment for your global ptr?
19:45:34 #corewars: < joonas> doesn't seem to provoke the bug here.
19:45:53 #corewars: < michal> ok. here's the code:
19:46:17 #corewars: < michal> #include <stdio.h>
19:46:17 #corewars: < michal>                                                                                                                                
19:46:17 #corewars: < michal> void *p = (void *) 1000;
19:46:17 #corewars: < michal>                                                                                                                                
19:46:17 #corewars: < michal> void *f ()
19:46:19 #corewars: < michal> {
19:46:21 #corewars: < michal>     printf ("calling f\n"); fflush (stdout);
19:46:23 #corewars: < michal>     if (p == (void *) 1000)
19:46:25 #corewars: < michal>         return &&secret;
19:46:27 #corewars: < michal>     printf ("examine2\n"); fflush (stdout);
19:46:29 #corewars: < michal> secret:
19:46:31 #corewars: < michal>     printf ("secret!\n"); fflush (stdout);
19:46:33 #corewars: < michal>     goto *p;
19:46:35 #corewars: < michal> }
19:46:37 #corewars: < michal>                                                                                                                                
19:46:40 #corewars: < michal> int main ()
19:46:42 #corewars: < michal> {
19:46:44 #corewars: < michal>     void *ptr = f ();
19:46:46 #corewars: < michal>     p = &&ret;
19:46:48 #corewars: < michal>     printf ("%p %p\n", ptr, p);
19:46:50 #corewars: < michal>     goto *ptr;
19:46:52 #corewars: < michal> ret:
19:46:54 #corewars: < michal>     printf ("done\n");
19:46:56 #corewars: < michal> }
19:47:00 #corewars: < michal> if (p == ... is only to prevent compiler from throwing away the rest of the code
19:50:04 #corewars: < joonas> hm.. I see.  the goto *ptr in main() is lost under 3.0.4
19:50:14 #corewars: < joonas> hm.. no, sorry.
19:51:11 #corewars: < michal> I think it is.. and inside the function the goto is hardcoded to secret
19:51:45 #corewars: < michal> this code actually works in gcc version 2.95.4 20011002 (Debian prerelease)
19:51:50 #corewars: < michal> [and only there]
19:52:21 #corewars: < michal> I remember I somehow managed to get it to work in gcc 3+ by putting some extra garbage in both functions
19:55:35 #corewars: < joonas> ok, with 3.0.4 the goto inside f() is ok, but the goto in main() isn't even there.
19:57:02 #corewars: < joonas> in any case, what you're trying to do is so not supported. :)
19:57:59 #corewars: < michal> but ordinary function calls are so slow :)
19:58:29 #corewars: < joonas> if you're bent on doing it in C, just stuff everything you want to goto to into one function.
19:59:55 #corewars: < michal> that was the plan..
20:00:17 #corewars: < joonas> oh, and goto to those places from within that function, of course.
20:00:29 #corewars:  * joonas wonders whether "goto to" is grammatical
20:00:46 #corewars: < michal> I just wanted to check if I could make compiler's life  a little easier..
20:02:41 #corewars: < michal> no big loss, this way the alignment would be broken anyway
20:17:12 -!- Roy [~roy_van_r@kf-mss-cb01-0965.dial.kabelfoon.nl] has joined #corewars
20:17:16 #corewars: < Roy> Hello
20:17:20 #corewars: < michal> hi
20:17:22 #corewars: < joonas> hi
20:17:34 #corewars: < Roy> Wow, haven't seen you guys around for a while
20:18:27 #corewars: < joonas> how have you been?
20:18:51 #corewars: < Roy> I'm just fine, not so busy at school, time for fun things :)
20:30:24 #corewars: < joonas> anything in particular?
20:32:17 #corewars: < Roy> Animating pictures, I made Devilstick, juggling, girlfriend, lots of stuff
20:35:08 #corewars: < grabek> yohoho and the bottle of rum
20:36:36 #corewars: < Roy> And drinking too yes, thanks for reminding Lukasz :)
20:37:06 #corewars: < joonas> hi lukasz
20:37:17 #corewars: < grabek> hi Roy & joonas 
20:41:36 #corewars: < Roy> So, when can we expect your names back on the hill with new warriors?
20:41:48 #corewars: < Roy> Its been very quiet lately...
20:41:56 #corewars: < grabek> Yeah, right, joonas, when this could be?
20:42:15 #corewars: < joonas> when pong falls off?
20:42:28 #corewars: < joonas> within 2 weeks of that?
20:42:59 #corewars: < Roy> Good point...
20:43:01 -!- grabek is now known as pong
20:43:03 #corewars: < pong> I have not
20:43:27 #corewars: < Roy> But you only one warrior on the hill isn't much, we need at least one challenger a day
20:43:45 -!- pong is now known as grabun
20:43:56 #corewars: < grabun> I don't know, really.
20:43:57 #corewars: < joonas> I found out later that 'pong' was actually the name of an early version Tangle Trap.
20:44:00 #corewars: < Roy> I want to see how long Devilstick will survive on the hill, and at this rate it'll have age 100 at the end of the year...
20:44:27 #corewars: < grabun> Not much time for CW, for pals, for beer. We spent plenty of time preparing the wedding, our place, things
20:44:28 #corewars: < joonas> I could try to quicken it up for you, if you like. ;)
20:45:27 #corewars: < grabun> Which one?
20:45:46 #corewars: < grabun> Can you paint walls and glue (?) tiles?
20:46:12 #corewars: < joonas> grabun: don't worry... I'm told by married friends that the wedding plannings are eventually hijacked by the bride.
20:47:03 #corewars: < grabun> Yeah, I've heard that too. I didn't say: my plans, no?
20:49:18 #corewars: < Roy> Congratulations Lukasz, didn't know you where getting married :)
20:49:27 #corewars: < grabun> Thank you
20:50:38 #corewars: < Roy> God, I don't even want to think about it yet
20:51:00 #corewars: < Roy> You guys are all getting way too old :P
20:51:28 #corewars: < joonas> heh
20:51:37 #corewars:  * grabun is not that old
20:51:55 #corewars: < joonas> bvowk's having a 25-year-crisis.
20:52:01 #corewars: < Roy> Lol :)
20:52:06 #corewars: < Roy> He is?
20:52:21 #corewars: < joonas> yes.  he's turning 25 real soon now.
20:52:34 #corewars: < joonas> I think the 15th of Feb.
20:52:37 #corewars: < Roy> Argh, I'm turning 21 soon, and nobody ever told me there is a crisis ahead!
20:53:20 #corewars: < Roy> I only knew about the midlife crisis thingy, but I'm not planning on doing that until I'm 40+ :)
20:53:21 #corewars: < grabun> joonas: you're pretty young. You sound too serious for *that* age. Ever though about getting out somewhere?
20:53:47 #corewars: < joonas> no, not really. :)
20:54:06 #corewars:  * joonas is 25 also
20:54:21 #corewars: < Roy> 4 years to go...have to spend them well!
20:54:44 #corewars: < joonas> don't worry though.  I think the crisis at 25 is optional.
20:55:33 #corewars: < grabun> optimal, you mean.
20:55:59 #corewars: < joonas> surely not.  better to have it earlier.
20:56:01 #corewars: < Roy> Ah, I'll just see how life goes, no worry, just live with it
20:56:22 #corewars: < joonas> or not at all.
20:56:53 #corewars: < Roy> If a beer comes my way I'll drink it, if a girl comes in my way I'll marry her, just go with the flow
20:57:06 #corewars: < joonas> :)
20:57:30 #corewars: < grabun> heh
20:57:40 #corewars: < grabun> I would not compare girls and beer :>
20:58:29 #corewars: < Roy> Well, I'm going great now, I have my current girlfriend for 3 months now! (That really beats my previous record of a whopping 2 weeks)
20:59:46 -!- michal [~michal@august.V-LO.krakow.pl] has quit [BitchX: for distribution only with a new PC]
21:00:11 #corewars: < grabun> A name for a warrior, please
21:00:41 #corewars: < joonas> test #!
21:01:13 #corewars: < Roy> Mazzeltoff
21:01:47 #corewars: < grabun> Oh, just a good name for a stone/imp
21:03:13 #corewars: < Roy> Xena?
21:03:16 #corewars: < grabun> LOL
21:03:17 #corewars: < Roy> warrior princess :)
21:04:16 #corewars:  * Roy likes the idea of LG sending a s/i....devilstick likes si's, it may increase its lead :D
21:05:27 #corewars: < Roy> http://tinyurl.com/399d6
21:05:32 #corewars: < Roy> (generate a name!)
21:07:13 #corewars:  * grabun thinks
21:07:40 #corewars: < joonas> try an anagram creator
21:09:16 #corewars: < grabun> A mighty name
21:09:49 #corewars: < Roy> Lizard mark use -> (Anagram of something... :))
21:12:03 #corewars: < grabun> OK
21:12:11 #corewars: < grabun> I've submitted ti
21:12:17 #corewars: < grabun> s/ti/it/
21:12:20 #corewars: < grabun> With a stolen name
21:13:12 #corewars: < Roy> Thats an old name :)
21:13:42 #corewars: < grabun> Yeah
21:13:53 #corewars: < grabun> But a warrior with such a name is obliged to last long
21:14:00 #corewars: < Roy> Rainstorm Power (Stone Imp Warrior)
21:14:07 #corewars: < grabun> I don't think it will survive >100 fights :-(
21:14:11 #corewars: < Roy> (Argh, misses one I)
21:14:26 #corewars: < Roy> Will it get on the hill?
21:14:50 #corewars: < grabun> I don't know
21:14:56 #corewars: < Roy> Well, it isn't slaughtered by Devilstick
21:15:37 #corewars: < Roy> Beats Recon and Dracula really nicely
21:16:08 #corewars: < Roy> Yes, its going to get in the top 5 I recon
21:16:10 #corewars: < grabun> Is it sucky?
21:16:31 #corewars: < Roy> Loses from Hazy Lazy, thats a shame
21:16:55 #corewars: < Roy> Ah, its second, grats
21:17:00 #corewars: < grabun> ???
21:17:20 #corewars: < grabun> Fuckingly odd!
21:17:25 #corewars: < grabun> Thank you
21:17:47 #corewars: < joonas> yes, congratulations.
21:17:51 #corewars: < joonas> what's odd?
21:18:14 #corewars: < Roy> You closed the gap a bit, thats a pitty :)
21:18:17 #corewars: < grabun> It sets on a second place
21:18:24 #corewars: < grabun> That's odd
21:18:50 #corewars: < Roy> Odd in what way...? Mine is in first place, THATs odd!
21:19:14 #corewars: < grabun> Eh, why?
21:19:56 #corewars: < Roy> Well, I'm still a bit newbie, I still can't make a normal stone, and I had a 7/8 point lead on the koth.org hill, thats odd
21:20:23 #corewars: < Roy> You have been leading the hill more then a dozen times, so second is less odd
21:20:47 #corewars: < grabun> Practise makes a master, is it the way one says?
21:21:41 #corewars: < Roy> So its now proven :) Now if only someone makes a new scanner and pushes 38835-2156 etc off
21:21:53 #corewars: < Roy> Then I'm happy again :)
21:23:37 #corewars: < grabun> Ok, I will post it.
21:23:55 #corewars: < grabun> joonas: I didn't have time to use your optimizer
21:24:03 #corewars: < Roy> Just send it to Koenigstuhl (Thats what I always do)
21:24:24 #corewars: < Roy> Or make write a little piece and send it to JM for the next Core Warrior
21:24:33 #corewars: < joonas> grabun: that's ok.  it's a mess anyway.
21:25:49 #corewars:  * grabun hesitates.
21:25:56 #corewars: < joonas> and many years old too.
21:26:10 #corewars:  * grabun remember Joonas said something about not publishing warriors the way they're
21:26:29 #corewars: < Roy> I always do (that IS a bad thing though)
21:26:48 #corewars: < Roy> People always put those warriors in benchmarks and optimize against them
21:26:58 #corewars: < Roy> You'll lose precious points
21:27:05 #corewars: < joonas> exactly.
21:27:28 #corewars: < Roy> But I'm always too curious to see what I'll score in FULL strenght
21:27:35 #corewars: < Roy> So I just send them the way they are
21:28:27 #corewars: < joonas> well the answer to that is to create two ~equally strong warriors.
21:29:01 #corewars: < Roy> Heh loon :)
21:29:03 #corewars: < joonas> and since they're not going to be exactly equal, I'd publish the slighly weaker one.
21:29:24 #corewars: < joonas> :)
21:29:34 #corewars: < Roy> I'm going to change the version on the hill anyway, I have a slightly better version
21:29:57 #corewars: < Roy> (Better spread qscan and paper distibution)
21:30:01 #corewars: < joonas> well, there you go then, loon. :)
21:31:31 #corewars: < Roy> I'm not loon...superbear is! http://tinyurl.com/2fggn
21:43:31 #corewars:  * Roy is afk, just letting the optimizer make me an even bigger loon
21:51:05 #corewars: < grabun> OK, new idea for a warrior
21:52:17 #corewars: < grabun> What do you think about HL series?
21:58:00 #corewars: < joonas> it's great
21:59:58 #corewars:  * grabun bows
22:00:08 #corewars: < grabun> thank you for very detailed answer :>
22:00:51 #corewars: < joonas> sorry, I'm not really paying attention.
22:00:54 #corewars: < joonas> well wasn't.
22:00:56 #corewars: < joonas> HL.
22:01:16 #corewars: < grabun> Yeah, HL
22:01:45 #corewars: < Roy> What is HL?
22:01:46 #corewars: < joonas> well, firstly, I think Steve is being overly modest when he says/has said that HL is "just a little tweak of The Machine"
22:01:56 #corewars: < Roy> Hazy Lazy...duh :)
22:02:27 #corewars: < grabun> What is its secret, anyway?
22:02:28 #corewars: < joonas> The Machine is already a really powerful warrior and is very delicately balanced.
22:03:15 #corewars: < Roy> The only big weakness of HL I know it that its vulnarable against bootdistances, its easy to optimize against
22:03:27 #corewars: < grabun> Point, Roy
22:03:28 #corewars: < Roy> (IMHO)
22:04:04 #corewars: < Roy> Thats why I love Herbal Avanger, its also a bit stronger
22:04:39 #corewars: < grabun> The secret. The thing that's behind it.
22:04:43 #corewars: < grabun> Quick boot?
22:04:47 #corewars: < grabun> Nasty bomb?
22:04:51 #corewars: < grabun> Step size?
22:05:27 #corewars: < Roy> Maybe a boot that turns into a small bombing routine?
22:06:00 #corewars: < joonas> I think a version of HL is published somewhere.
22:06:02 #corewars: < Roy> In HA its definetly the Step size, that one is great
22:06:15 #corewars: < Roy> http://www.ociw.edu/~birk/COREWAR/94/HILL/hazylazy.red
22:06:51 #corewars: < joonas> can't say for HL, but in The Machine, one great thing you really have to admire is how the scan loop is organised to detect dat 0 bombs.
22:07:13 #corewars: < grabun> ???
22:07:31 #corewars: < grabun> Yeah, I remember you stating that in CW#80
22:07:35 #corewars: < grabun> I don't get it
22:08:19 #corewars: < Roy> Well, no scanners really find dat 0,0 bombs do they?
22:08:35 #corewars: < Roy> I don't get it either :)
22:09:32 #corewars: < joonas> hrmph.. maybe I really am getting too old... I distincly remember going through a week or two deciphering the scan loop to see how it drops through to the clean when hit by a dat 0,0 bomb.
22:09:53 #corewars: < joonas> I'm *almost* sure I didn't make that up in a dream.
22:10:09 #corewars: < joonas> clean -> clear
22:10:44 #corewars: < grabun> jmn line?
22:10:50 #corewars: < joonas> yes. and indirections
22:11:47 #corewars: < grabun> But it protects just one line
22:12:25 #corewars:  * joonas goes and takes a look
22:14:26 #corewars: < Roy> Maybe I'm going to make a scanner like HA, with other constants, maybe even tiny tweaks, a qscan and a boot
22:14:40 #corewars: < joonas> the jmn top, @top protects both top and @top
22:14:55 #corewars: < grabun> Funny
22:15:02 #corewars: < grabun> I was to say that two lines are protected
22:15:19 #corewars: < grabun> Then I thought I forgot why the other is protected too
22:15:26 #corewars:  * grabun slaps his forehead
22:16:46 #corewars: < joonas> also, look at how it's b-decrement protected.
22:16:59 #corewars: < joonas> ptr: no effect.
22:17:12 #corewars: < joonas> top: fall to clear
22:17:19 #corewars: < joonas> scan: ??
22:17:35 #corewars: < joonas> scan+1: no effect
22:17:52 #corewars: < joonas> a: fall to clear, I think.
22:18:26 #corewars: < joonas> the jmn line: fall to clear, I think.
22:18:28 #corewars: < grabun> please explain the top line
22:18:58 #corewars: < Roy> The it points to ptr-1 -> thats empty so falls to clear..?
22:19:01 #corewars: < Roy> the-then
22:19:05 #corewars: < joonas> yes.
22:19:10 #corewars: < Roy> Woo! :)
22:19:53 #corewars: < joonas> well b-decrementing scan+1 could mess it up a little.
22:20:36 #corewars: < Roy> Well, there is a big change ptr+step is also empty
22:21:05 #corewars: < joonas> ah, yes.
22:21:16 #corewars: < Roy> That JMN line is pretty safe
22:21:28 #corewars: < joonas> yup
22:22:39 #corewars: < joonas> so 33% of the main loop is dat 0,0 protected, and most of it is b-decr. protected.
22:23:13 #corewars: < joonas> that's pretty cool I think.
22:23:29 #corewars: < Roy> But there aren't many dat 0,0 bombers these days are there?
22:23:45 #corewars: < joonas> well that's true.
22:24:28 #corewars: < joonas> but then, if the bomb is of the form dat xx, hop with hop not too small, then the top line at least is protected.
22:25:51 #corewars: < Roy> Yes..
22:26:43 #corewars: < joonas> anyway, one big thing about The Machine is the 1,1 decoy left behind.
22:26:57 #corewars:  * grabun is listening
22:27:48 #corewars: < joonas> if you take out the decoy, then the anti-djn'ess of TM goes away since it has to create more 1,1 decoy.
22:28:10 #corewars:  * grabun has created cool version of boot for HL
22:28:26 #corewars: < joonas> what's cool about it?
22:29:24 #corewars: < joonas> oh, and re: the boot, in TM, you can't touch it, or it'll go all to hell.  don't know why, but it just is.
22:29:34 #corewars: < grabun> lol
22:29:35 #corewars: < joonas> magic, that is.
22:29:58 #corewars: < grabun> The cool twist is that now only the boot code is triggered by the scan loop
22:30:25 #corewars: < grabun> The warrior body (left behind) is left unnoticed
22:30:48 #corewars: < joonas> ah, nice!
22:30:57 #corewars: < joonas> do you use a mirror insn or something?
22:32:22 #corewars: < grabun> Not mirror (unless mirror is something different that I know)
22:32:40 #corewars: < grabun> I'll put on my site
22:33:08 #corewars: < grabun> http://www.astercity.net/~grabek/df.red
22:33:16 #corewars: < grabun> I don't know whether this work and it's dirty!
22:33:43 #corewars: < joonas> ah, one more thing in TM: if a multi-pt imp comes crashing into the scanner, it'll be gated immediately.
22:34:22 #corewars: < joonas> well, at least when ptr+IMPSTEP is 0,0.
22:34:55 #corewars: < joonas> so, that's, what, 2/3 (?) of the time for a 3-pt imp?
22:36:48 #corewars:  * joonas is being stupid.
22:36:54 #corewars: < joonas> all the time, of course.
22:37:21 #corewars: < joonas> grabun: I don't see it
22:37:49 #corewars: < grabun> the file or the "trick"
22:38:27 #corewars: < joonas> the trick... but I see it now.
22:38:30 #corewars: < joonas> nice.
22:39:04 #corewars: < joonas> I thought that 'gap' was the separation between scans, but it's not.
22:39:40 #corewars: < joonas> hm...
22:39:45 #corewars: < joonas> I should go.
22:39:51 #corewars: < joonas> bye
22:39:57 -!- joonas [jpihlaja@kruuna.helsinki.fi] has quit [ircII2.8.2-EPIC3.004+Kasi --- Bloatware at its finest.]
22:40:02 #corewars: < grabun> Aw, it's still triggered
22:40:04 #corewars: < grabun> ups
22:40:15 #corewars: < Roy> Heh, too bad
22:40:29 #corewars: < Roy> I wasn't paying attention (doing some new research myself)
22:40:48 #corewars: < Roy> I'm trying to implement Zooms speed trick in the Machines layout
22:48:04 #corewars: < grabun> Good luck
22:48:08 #corewars: < Roy> Hmm, yay, I'm happy with this! looks like it works better
22:48:21 #corewars: < Roy> I added one line, but it scans faster now
22:48:32 #corewars: < Roy> (until it finds the first thingy)
22:49:06 #corewars: < Roy> I'll upload it
22:49:59 #corewars: < Roy> http://www.geocities.com/roy-v-rijn/zoom.txt
22:51:53 #corewars: < Roy> Now you can even do something nice with the first JMN line (maybe make it JMZ and add one little scan)
22:54:49 #corewars: < grabun> Now my scanner is not triggered in *any* way
22:55:07 #corewars: < Roy> Thats not right either :)
22:56:27 #corewars: < grabun> Does not work?
22:56:29 #corewars: < grabun> Erm
22:56:34 #corewars: < grabun> I mean
22:56:34 #corewars: < grabun> does it work?
22:56:37 #corewars: < grabun> Your warrior?
22:56:51 #corewars: < grabun> I don't get, but I don't zoom trick anyway
22:56:56 #corewars: < Roy> Can't really tell, I think it works...but scores lower (must be the extra line
22:57:47 #corewars: < Roy> You don't get Zooom? Zooom is easy, it first has a small scanning loop, until the first instruction is scanned -> then it { the loop so that it now also clears
22:58:37 #corewars: < Roy> first sub/jmz, then mov/sub/jmz (so no clearing is spoiled when nothing is found)
22:58:57 #corewars: < Roy> That way you could find the enemy faster
22:59:22 #corewars: < grabun> How is decremented line fixed?
22:59:32 #corewars: < grabun> I mean
22:59:40 #corewars: < Roy> sGat:mov   }m,       *bDis   ; multi-function
22:59:40 #corewars: < Roy>      mov   sBmb,     >sGat
22:59:40 #corewars: < Roy> sLoo:sub   #step,    #step*numb
22:59:40 #corewars: < Roy> m:   jmz.f sLoo,     @sLoo   ; a-field mutates
22:59:40 #corewars: < Roy>      mov.b sLoo,     sGat
22:59:40 #corewars: < Roy>      jmn.a {m,       *sGat
22:59:41 #corewars: < grabun> Is jmn {a not executed any more?
23:00:08 #corewars: < Roy> This is the mean part, if its decremented one extra time -> sGat is executed and fixes itself
23:00:20 #corewars: < Roy> *main (evil english language!)
23:00:26 #corewars: < grabun> When is it executed?
23:01:33 #corewars: < Roy> First m points at sLoo, small scanloop, then after one thing it found it points to sLoo-1 (clearing) and if the {m is executed again it points to sGat, fixing itself with }m
23:02:14 #corewars: < grabun> oh
23:02:19 #corewars: < grabun> so the loop runs slower, no?
23:02:33 #corewars: < Roy> Only the time it finds enemy code
23:03:03 #corewars: < grabun> Yeah
23:03:06 #corewars: < grabun> Cool, indeed
23:03:07 #corewars: < Roy> other times it is fixed and stays pointed at sLoo-1 (usual mov/sub/jmz)
23:03:08 #corewars: < grabun> Thanks
23:03:28 #corewars: < Roy> Thats what I implemented in HA in my warrior just now
23:03:44 #corewars: < grabun> Yeah, I get it
23:06:22 #corewars: < grabun> Ok, I am proud of my version of HL
23:06:41 #corewars: < Roy> Does it work now?
23:07:19 #corewars: < grabun> I works
23:07:21 #corewars: < grabun> Perfectly
23:07:34 #corewars: < grabun> The boot code is left behind and does not trigger the scanner
23:07:47 #corewars: < grabun> Now I have to optimize boot distance and step size
23:08:12 #corewars: < Roy> And implement Zooom :P
23:09:06 #corewars: < grabun> heck why not :-)
23:09:31 #corewars: < Roy> Its a nice practise and pretty damn hard to do (and not scoring points extra here :( )
23:09:43 #corewars: < Roy> So why bother :)
23:10:11 #corewars: < grabun> aw
23:10:20 #corewars: < grabun> I though your version works :-(
23:10:31 #corewars: < Roy> Well, it works (in theory)
23:10:38 #corewars: < Roy> But isn't better :)
23:10:57 #corewars: < grabun> oic
23:11:00 #corewars: < grabun> *sigh*
23:11:44 #corewars: < grabun> What is best mod 16 step size for a scanner?
23:12:14 #corewars: < Roy> How in heavenly gots name must I know, I still don't know what those mod thingies stand for!
23:12:30 #corewars: < Roy> gots...gods :-S thats my worse spelling error ever
23:13:57 #corewars: < grabun> lol
23:14:20 #corewars: < grabun> Wanna see it?
23:14:30 #corewars: < Roy> Sure (won't understand it though)
23:14:34 #corewars: < Roy> But I'll try!
23:14:52 #corewars: < grabun> The addres is the same as I gave few minutes ago
23:15:19 #corewars: < Roy> k
23:16:25 #corewars: < Roy> And why exacly was the unbooted warrior unsorted? :)
23:16:42 #corewars: < grabun> Watch it in xpmars :-)
23:16:48 #corewars: < grabun> use sk 50
23:16:50 #corewars: < grabun> :-)
23:17:04 #corewars: < Roy> xpmars...whats that!? :-S
23:17:09 #corewars: < Roy> I only have corewin working here
23:17:10 #corewars: < grabun> ?
23:17:12 #corewars: < grabun> oh
23:17:16 #corewars: < grabun> might work as well :>
23:20:39 #corewars: < Roy> Still don't see it sorry :( what must I look for? :)
23:21:27 #corewars: < grabun> Lines are unsorted as not to trigger the seq line
23:21:49 #corewars: < Roy> yes...? I'll take another look
23:23:04 #corewars: < Roy> spl 1 spl 1 mov -1 mov -1, nice
23:23:47 #corewars: < grabun> :-)
23:24:39 #corewars: < grabun> I kill oneshots as it was a stone
23:24:57 #corewars: < Roy> And thats how its supposed to be :P
23:25:33 #corewars: < grabun> Surely, I lose to paper and paper with imps :-)
23:27:59 #corewars: < Roy> ok...I'm going to try my version with decoy on the hill (you should too, scores pretty good DFx)
23:28:17 #corewars: < grabun> I will
23:28:23 #corewars: < grabun> I am benchmarking it
23:29:52 #corewars: < Roy> In the queue now...
23:34:40 #corewars: < grabun> Wow, close one
23:34:40 #corewars: < Roy> Almost pushed pong off, then Joonas had some new work to do :P
23:35:12 #corewars: < Roy> According to my bench yours will be almost as close, scoring almost the same scores in the bench
23:37:36 #corewars: < grabun> Submitted
23:39:17 #corewars: < Roy> Argh! just maybe found a MAJOR improvement in Devilstick (testing now!)
23:40:11 #corewars: < grabun> wow
23:40:20 #corewars: < Roy> (notice the maybe)
23:41:38 #corewars: < Roy> Wow! pretty sweet!
23:42:23 #corewars: < grabun> and?
23:42:49 #corewars: < Roy> Your scores are great, mine are not as good as hoped for
23:43:18 #corewars: < Roy> Devilstick normal: 100 / DevChanged: 129 -> but not better in the benchmark :(
23:43:58 #corewars: < grabun> 129 > 100 !
23:44:08 #corewars: < Roy> Yes, that surprised me too
23:44:10 #corewars: < grabun> What is that 100 and 129 score?
23:44:24 #corewars: < Roy> Just one on one 100 rounds
23:44:34 #corewars: < grabun> oic
23:45:21 #corewars: < grabun> ok, time I went to bed
23:45:22 #corewars: < grabun> *yawn*
23:45:27 #corewars: < grabun> pretty effective day
23:45:27 #corewars:  * Roy waves!
23:45:38 #corewars: < Roy> Yes, two nice warriors on the hill, grats ;-)
23:45:55 #corewars:  * grabun waves
23:45:58 #corewars:  * grabun bows, thank you
23:46:18 #corewars: < Roy> (but I'm still king, na na na naa naaa!)
--- Log closed Sat Jan 17 00:00:39 2004