Talk:OcnirpSNG

Shall I proud with this bot? I own only half of it. » Nat | Talk » 12:11, 1 March 2009 (UTC)

Good bot! I'm trying to put random movement to Scytodes, but I can't make it less than 266 bytes... Maybe Simonton's gun costs less codesize? --HUNRobar 12:58, 1 March 2009 (UTC)

It the smallest PM I can found on the wiki right now :) Note that Simonton are excellent in codesize reduction. You might read the CodeSize/WritingSmallCode on the old wiki. Note that with his trick on variable declaration, I can take 14 bytes away from Ocnirp and have enough room for StopNGo, anyway. » Nat | Talk » 14:14, 1 March 2009 (UTC)

Note for you, if you don't mine to turn while you are moving, you can reduce codesize to 227, if not, 237.

• The e.getDistance() called twice, declare a variable.
• Note that instead of degrees, use radians. The turn code will look like setTurnRightRadians((Math.cos(absB = e.getBearingRadians()) + direction * -0.174532925D));
• As you will see above, use Math.cos will move in perpendicular automatically. (I don't know why, but ripped from WeekendObsession)
• Move the turn code before the gun code, assign the e.getBearingRadians() to absB first, then in gun, use += getHeadingRadians();

That all I squeeze the Scytodes, note that we use the same gun :) » Nat | Talk » 14:28, 1 March 2009 (UTC)

In fact Scytodes 0.3 is awfully inoptimized in codesize. Just look at Prestige's code. I even managed to make direction unnecessary! Okay, in stop and go it wouldn't work, but now I could be sqeeze further any more, maybe 1-2 bytes. Anyway, I deal with nanos so much that I discover newer and newer tricks every week. --HUNRobar 15:11, 1 March 2009 (UTC)

Come on, Robar, share your secret with us. » Nat | Talk » 15:35, 1 March 2009 (UTC)

Start to fear, I managed to finish the first version of BlackWidow, the worst enemy of WeekendObsession and OcnirpSNG! :P --HUNRobar 19:38, 2 March 2009 (UTC)

Oh oh... I'll wait. With Ocnirp, I win WeekendObsession ans several PM bot. Try your bot with RaikoNano, GFNano or something else, if you win, I'll spent more time on it =D » Nat | Talk » 02:52, 3 March 2009 (UTC)

Nat, it is a very bad idea to put underscores in your versionnumber. Roborumble can not handle them very well. When listed as first in a battle, everything goes ok, but when listed as second contender, roborumble gives an exception. This is an old bug and maybe it is solved in a new version, but not in 1.5.4 that I still use. --GrubbmGait 22:10, 3 March 2009 (UTC)

• Indeed, alphanumeric versions numbers are best. On an unrelated topic, why are you limited to 1.5.4 GrubbmGait? Any reason why not 1.6.0.2? --Rednaxela 00:16, 4 March 2009 (UTC)
• No real reason, I shortly used 1.6.0 but that gave problems with SilverSurfer, and after that I just did not take the time to upgrade to a newer version. --GrubbmGait 07:04, 4 March 2009 (UTC)
  • Btw, what is the latest stable version to use with RoboRumble? --GrubbmGait 08:19, 4 March 2009 (UTC)

I just understand right now why these test bots doesn't get score =) I agree that 1.6 and above get error with silversurfer, 1.6.2 and above get error with SandboxDT =) If the author of these bot doesn't fix it, we should removed them from roborumble. » Nat | Talk » 12:02, 4 March 2009 (UTC)

SilverSurfer's current version with 'fix' in the version should work fine with all versions of Robocode to my knowledge. As far as my experience the versions known to be stable for rumble and reasonably current are just 1.5.4 and 1.6.1.4. Also, by the looks of things 1.7.1 Beta looks like it will be stable for rumble with the exception of a couple bots (SandboxDT and Stampede2 for instance). As far as removing them? Only if we determine with complete certainty that the old behavior they used was actually wrong. At least in the case of SandboxDT and Stampede2 I think there's a reasonably high chance the bug is not the bot, but that needs to be investigated further. --Rednaxela 14:48, 4 March 2009 (UTC)

Hi, when will you delete your testbots from rumble? I'm quite confused about real rankings. Everything is pushed down because of the invasion of OcnirpSNG's. :) --HUNRobar 19:14, 4 March 2009 (UTC)

• I was kind of wondering that myself. I've seen 3 versions of a bot there myself before (Were mine actually), and seen 2 a few times from various people... and usually they put extra computers on rumble-duty when doing that. but 5 versions? That just makes all those versions take forever to get the battle count up. --Rednaxela 21:00, 4 March 2009 (UTC)

I'll removed it right now, because it doesn't get score! But they may re-enter to check what is best again =) » Nat | Talk » 23:56, 4 March 2009 (UTC)

It's time to do something on your bot, it's only 6th and BlackWidow 1.3 has just occupied the 3rd place. ;) --HUNRobar 19:48, 5 April 2009 (UTC)

No, it isn't time yet. I still get 2nd PL. Actually, I've test a lot of thing with this bot, one is distance controller but it perform worse and can't fit into nano. My aim is now point at megabot, nano is just a bot when I want to write a small code within 10 minutes ;) If you get my 2nd PL, I may do more on this bot. » Nat | Talk » 20:02, 5 April 2009 (UTC)

Yes, PL is the only thing where OcnirpSNG does better. :P Maybe in 1.35 or 1.4... ;) --HUNRobar 07:16, 6 April 2009 (UTC)

Hi, don't you think it's time to update Ocnirp? ;) Just look at Pugio 1.3 and note that it's just a little 'rethink'. New versions are under construction. --HUNRobar 17:34, 14 May 2009 (UTC);)

Yeah, I'm trying to update. But anyway, I can't find anyway to improve it right now =( » Nat | Talk » 23:28, 14 May 2009 (UTC)

I think you have to move on. I have more than a dozen bytes of free space in BlackWidow and I can't make it better. Think something new and original. Try to implement more advanced techniques into nanos, which I even don't know. For example WaveSurfing, GuessFactor Targeting etc.. --HUNRobar 14:30, 15 May 2009 (UTC)