←2011-02-14 2011-02-15 2011-02-16→ ↑2011 ↑all
00:04:04 -!- poiuy_qwert has joined.
00:05:55 <Gregor> OK, I've integrated animation into trunk egojsout.
00:05:58 <Gregor> The UI needs some work :P
00:06:55 <elliott> http://www.scribd.com/doc/48773692/Ignite-SV-Haskell-is-Better Oh snap, Haskell propaganda using Second Life. Somebody alert Sgeo__.
00:07:03 <elliott> Gregor: Can it SCRUB
00:07:34 <Gregor> elliott: Can your mom?
00:07:39 <elliott> Yes.
00:07:42 <elliott> ANY QUESTIONS
00:13:09 <Gregor> http://codu.org/eso/bfjoust/egojsout/?l=f2fbfce59c1af7376b794987b0b82b848cada986&r=47e874799fa9206bf40b7ef1e7cbf981187341aa&t=25 TADA
00:19:06 <elliott> Gregor: Suggest spacing below the tape length heading and below the animation :P
00:19:20 <elliott> Also some link to egojsout proper maybe
00:19:22 <elliott> But it's cool.
00:19:36 <Deewiant> Suggest being able to see the sources of the programs involved :-P
00:19:37 <elliott> Gregor: Oh, and I'd also put the trace in a scrollable div and have it automatically scroll down, so you can see the match evolve.
00:19:41 <elliott> Also, what Deewiant said.
00:19:45 <elliott> But the scrolling thing would be cool.
00:19:49 <quintopia> zzo38: watson is kicking their asses on jeopardy right now
00:20:09 <elliott> Deewiant: You actually do see it when the page loads :P
00:20:18 <elliott> Gregor: Suggest making re-ticking the animation box work :P
00:20:34 <Gregor> elliott: The animate box was just a hack, I'm getting rid of it entirely.
00:20:37 <Deewiant> elliott: Too quick, didn't even notice the first time :-P
00:21:29 <zzo38> quintopia: In my timezone Jeopardy is on later, not now
00:30:09 <quintopia> zzo38: then i will not discuss it any further
00:33:47 <Gregor> SliderSLIDERSLIDER
00:35:08 <elliott> Gregor: There's an HTML 5 for it.
00:35:09 <elliott> Just use dat :P
00:35:38 <elliott> Gregor: You should change all the "<b>foo</b> <a>foo</a>" links on [[BF Joust strategies]] to be just a bold link >_> I did one but oh god there are lots.
00:35:42 <elliott> PLURAL ones.
00:35:47 <Gregor> I am using it.
00:35:48 * Sgeo__ hits elliott with a DAT #0, #0
00:35:52 <Gregor> But I have a FF old enough not to have it :P
00:36:01 <Gregor> elliott: Regex dood
00:36:12 <elliott> Gregor: I'm using a FF crappy enough to not have built-in sed.
00:36:16 <elliott> *an
00:37:55 <Gregor> 's pretty sweet on Chrome :)
00:38:26 <elliott> Gregor: Did you make it continuous or discrete? The bar, that is.
00:38:35 <elliott> I'd prefer a continuous bar that's floor()'d in display, just because scrubbing would be smoother.
00:38:44 <Gregor> Continuous, floor'd in display.
00:38:47 * Sgeo__ installs a Microsoft game on WINE on Lubuntu under VirtualBox on Windows 7
00:39:05 <elliott> Gregor: *hi5*
00:42:25 <elliott> Gregor: Is it pushed?
00:42:31 -!- augur has quit (Ping timeout: 276 seconds).
00:42:33 <Gregor> Is now.
00:42:39 <Gregor> Would anybody mind if I put the program listings on the bottom when you permalink?
00:42:51 <Gregor> I just don't want them on top, I sorta want it to be a link to an animation essentially.
00:42:53 <elliott> Gregor: I'd have it be a link that expands them at the top.
00:43:00 <Gregor> That's not easy :(
00:43:03 <elliott> "Show sources" --> "Hide sources\n[two source textboxes]"
00:43:11 <elliott> Gregor: You mean hiding and showing an element isn't easy?
00:43:15 <elliott> Use jQuery or something if you're that lazy :P
00:48:44 <elliott> Gregor: Did you do that height-limited-scrolling-trace thing? >_>
00:49:07 <elliott> Next you just need to add LIVE PATCHING OF CODE and it'll be in like THOSE MOVIES ABOUT HACKERS gosh golly gosh darn.
00:51:34 <Sgeo__> WTF
00:51:42 <Sgeo__> I just Googled Core Wars vs VirtualBox
00:53:11 -!- pikhq has quit (Ping timeout: 240 seconds).
00:53:27 -!- pikhq has joined.
00:55:13 <Gregor> OK, permalinks are awesome now :)
00:55:13 <elliott> HAPPY.
00:55:23 <elliott> Gregor: Why do I always talk after long periods of silence at the same time as somebody else.
00:55:26 <elliott> Anyway link me up.
00:55:42 <Gregor> http://codu.org/eso/bfjoust/egojsout/?l=47e874799fa9206bf40b7ef1e7cbf981187341aa&r=ae4777f408850bad0cd6352d9b63a5d7db6ee5fb&t=25
00:56:11 <elliott> Gregor: No polarity specification in URL?
00:56:22 <Gregor> &p = polarity inverted, no &p = polarity normal
00:56:28 <zzo38> I used the algorithm for division in METAFONT to ensure it doesn't overflow
00:56:29 <elliott> Gregor: Is it, er, going slower?
00:56:34 <elliott> Or are the programs just slow?
00:56:41 <elliott> ...oh, my browser is lagged.
00:56:42 <Gregor> Same speed as always, these programs are just slowish :P
00:56:50 <elliott> Gregor: ANYHOW you need to implement that scrolling trace thing :q
00:56:51 <elliott> it would be awesome.
00:57:02 <Gregor> I refuse.
00:57:06 <elliott> Gregor: Why?
00:57:08 <Gregor> Patches welcome :P
00:58:00 <elliott> Gregor: s/<div/<div style="height:foo em; overflow:scroll"/; s//div contents=""/; s/next turn/next turn; div.append(line from full trace); div.scrollTo(whatever I forget the params)/
00:58:51 <Gregor> elliott: I refuse because I don't want it, though I'm insufficiently against it to refuse a patch.
00:59:03 <elliott> Gregor: I wanna know why you don't want it :P
00:59:15 -!- cal153 has quit (Ping timeout: 250 seconds).
00:59:18 <elliott> It would be helpful to see the "flat" tape's activity go by.
00:59:23 <Gregor> I find subscrollable areas annoying.
00:59:42 <Gregor> The alternative would be to just put it upside-down.
00:59:50 <elliott> Gregor: Use a frameset :P
00:59:57 <Gregor> ...
01:00:01 <Sgeo__> "VMware Player now creates virtual machines"
01:00:01 <elliott> See that ":P"?
01:00:02 <Sgeo__> What?
01:00:05 <Sgeo__> This sounds awesome
01:01:48 <pikhq> Sgeo__: There's even *less* reason to use VMware Server now.
01:02:10 <Sgeo__> pikhq, yes yes. But is it reason to switch away from VirtualBox? I think it may be
01:02:16 <pikhq> Probably not.
01:02:21 <Sgeo__> I do remember recently loathing VirtualBox
01:02:28 <elliott> Is Sgeo__ really going on about different VM products?
01:02:34 <elliott> Has a more debate ever been presented to the world?
01:02:34 <pikhq> *Especially* if you want to automate it.
01:02:41 <elliott> Has there ever been anything less impossible to get excited or interested about?
01:02:54 <pikhq> VMware is a bitch to automate compared to VirtualBox, believe it or not.
01:03:06 <elliott> Isn't VMWare slower than VirtualBox.
01:03:13 <pikhq> Probably.
01:03:13 <elliott> I don't see what's wrong with VirtualBox at all.
01:03:15 -!- poiuy_qwert has quit (Ping timeout: 246 seconds).
01:03:25 <pikhq> They're both reasonable as far as virtualisation goes, though.
01:03:33 <elliott> Sgeo is like a magpie except instead of hoarding shiny things he abandons his previous shiny thing and latches on to some other more shiny thing.
01:03:40 <elliott> Then it turns out it was just the sunlight making it look shinier.
01:03:45 <elliott> But hop he does.
01:03:50 <elliott> (I realise magpies don't actually do that.)
01:04:17 <zzo38> What is the correct pointer type for execv?
01:04:53 -!- augur has joined.
01:05:46 -!- poiuy_qwert has joined.
01:10:06 -!- Sgeo__ has changed nick to Sgeo.
01:14:42 <Sgeo> I like the VMware Workstation icon better than the VMware Player icon
01:15:20 <zzo38> I think the "texnicard.w" file is now complete as the first version will be. Can you check?
01:22:23 -!- Sgeo has quit (Read error: Connection reset by peer).
01:24:39 <Gregor> Anybody know how to do macros in MediaWiki?
01:26:40 <zzo38> Gregor: I know how to do templates, I don't know about macros, though.
01:27:24 <Gregor> Do templates require a separate node? All I really need is something like #define A(x) big_url_with_ ## x, then later A(bleh) A(florp) A(doink)
01:28:41 <zzo38> A template does require a separate node.
01:28:58 <zzo38> But it can have any number of parameters.
01:30:18 <zzo38> Parameters are used by typing {{{1}}} and {{{2}}} and so on.
01:30:57 <zzo38> You can also specify things that are used only when included in another page, or only when viewing the template itself.
01:31:16 <Gregor> I guess that's tolerable >_>
01:31:34 -!- Sgeo has joined.
01:32:04 <quintopia> !bfjoust space_elevator http://sprunge.us/YVZN
01:33:22 <EgoBot> Score for quintopia_space_elevator: 62.3
01:33:26 <pikhq> Time to get caught up on linear algebra homework.
01:33:37 <pikhq> Good *God* this is going to be bad...
01:33:51 <quintopia> huh. guess those two minuses are worth over 2 points
01:34:18 <Gregor> quintopia: DIEDIEDIE
01:34:27 <pikhq> All the homework for a portion of the class is due the day of the test.
01:34:34 <pikhq> Guess what I end up doing?
01:36:41 <pikhq> If you said "Put it off for the better part of a month", you'd be correct!
01:36:45 <pikhq> And your prize is nothing!
01:37:31 <quintopia> !bfjoust space_elevator http://sprunge.us/ciBB
01:37:39 <EgoBot> Score for quintopia_space_elevator: 65.0
01:37:58 <quintopia> okay, that was a definite improvement. saving that one
01:38:14 <quintopia> now let's see if those off-by-one errors are actually beneficial or no...
01:38:50 <zzo38> Really, can possbly anyone help with anything with TeXnicard? See if there are any obviously mistakes/omissions in texnicard.w file?
01:39:39 -!- Sgeo has quit (Read error: Connection reset by peer).
01:43:30 -!- Sgeo has joined.
01:43:32 -!- elliott has quit (Ping timeout: 272 seconds).
01:45:07 -!- cal153 has joined.
01:45:29 <quintopia> !bfjoust space_elevator http://sprunge.us/iKUL
01:45:42 <EgoBot> Score for quintopia_space_elevator: 64.8
01:45:47 <quintopia> wow
01:45:56 <quintopia> who'dathunk?
01:46:29 <quintopia> i suspect this has something to do with vibration
01:46:36 -!- Vorpal has quit (Read error: Operation timed out).
01:47:00 -!- Vorpal has joined.
01:55:49 <quintopia> !bfjoust space_elevator http://sprunge.us/Baid
01:55:58 <EgoBot> Score for quintopia_space_elevator: 60.5
01:58:17 <quintopia> !bfjoust space_elevator http://sprunge.us/LIHP
01:58:29 <EgoBot> Score for quintopia_space_elevator: 60.5
02:08:39 <Gregor> I added fancy animation links to a bunch of the strategies :)
02:09:33 -!- azaq23 has joined.
02:10:32 -!- zzo38 has quit (Remote host closed the connection).
02:19:19 <quintopia> neato
02:21:16 <Gregor> I made a program as the opponent designed to be simple enough that it's not a good strategy but complex enough that it would bring out interesting behavior. It beats some of the 2009 toplisters :P
02:22:40 -!- Mathnerd314 has joined.
02:23:36 <quintopia> i just thought up possibly the most difficult to counter defense program ever devised. basically the only way to counter it without making your program suck balls would be to 1) use ridiculously deep nesting to allow for more unpredictable behavior, or 2) do the saccharin_philip sort of thing :P
02:24:16 <Gregor> So, the only way to counter it is to do what's already done? :P
02:24:35 <Sgeo> What's a saccharin philip?
02:25:16 <quintopia> Gregor: saccharin_philip is pretty awesome dude. i mean really cool.
02:26:04 <Gregor> quintopia: It's a ripoff of poke + lead_acetate_philip X-P
02:26:55 <quintopia> Gregor: then lead_acetate_philip is pretty awesome
02:27:02 <Gregor> Heh
02:27:29 <quintopia> the idea of building 4 decoys at once slowly wreaks havoc on a decent clear
02:28:05 <quintopia> and the idea of treating the first 4 decoys carelessly before actually trying kills pretty much every defense program
02:28:29 <quintopia> and allows you to outpace a standard rusher
02:28:52 <Gregor> Ah, the simulbuild part is all saccharin, and yeah, I was originally going to build them one at a time, then tested it and came to the conclusion that it confused programs more to build them at once.
02:29:10 <Gregor> I still, however, can't do squat against shuddering >_<
02:29:17 <Gregor> If I could, maaaan things would be nice :P
02:29:39 <quintopia> which is why i was thinking that this ridiculously deeply nested clear needs to be written
02:30:15 <quintopia> i was thinking of making up a bf variant with subroutine calls to make writing it easier
02:30:29 <quintopia> it would compile by copying the subroutines into all the right places
02:30:46 <quintopia> since it's really quite impossible to do it just ({})% alone
02:31:12 <quintopia> actually
02:33:10 <quintopia> what would make it easiest is if we had (a?i{b}c)%i=32 type code, which would expand to: a32a31a30a29a28...bc...c
02:33:24 <quintopia> plus the subroutines
02:33:35 <Gregor> I was actually considering implementing that.
02:33:37 <quintopia> i'm seriously thinking about writing this compiler >_>
02:37:44 <quintopia> oh right. i was gonna see what happened if i used ridiculously large offsets for larger tape sizes :P
02:40:31 -!- augur has quit (Ping timeout: 264 seconds).
02:41:56 <quintopia> !bfjoust space_elevator http://sprunge.us/QDGa
02:42:10 <EgoBot> Score for quintopia_space_elevator: 60.7
02:42:22 <quintopia> a VERY slight improvement
02:44:27 <quintopia> results against definder2 did not change :/
02:45:05 <quintopia> but ais523 told me something that could very well change that...
02:46:07 * Gregor is waiting for things to shake out a bit before stepping back in the ring :P
02:48:39 <quintopia> !bfjoust space_elevator http://sprunge.us/BDCW
02:48:54 <EgoBot> Score for quintopia_space_elevator: 61.1
02:49:45 <quintopia> things aren't shaking very much right now gregor. also, space_elevator is still very vulnerable in a lot of ways, and using a good strategy that beats it could put you in 1st again
02:54:03 * pikhq is somewhat disappointed in Watson.
02:54:17 <pikhq> It didn't proclaim that the Singularity is here and proceed to demonstrate.
02:54:23 <pikhq> So disappointing. :P
02:58:10 <Sgeo> I should probably make a good faith effort towards my homework or something
02:59:54 <quintopia> !bfjoust space_elevator http://sprunge.us/BOBQ
03:00:13 <EgoBot> Score for quintopia_space_elevator: 60.7
03:00:18 <quintopia> pikhq: i was fucking amazed by watson.
03:00:46 <quintopia> this deepqa shit needs to be a search engine on the interwebs right now
03:00:53 <quintopia> i don't know if they can afford the hardware tho
03:00:58 <pikhq> On a more serious note, Watson is some pretty amazingly good natural-language processing, and I would like to thank IBM for actually doing R&D.
03:01:28 <quintopia> IBM research is huge main
03:01:45 <pikhq> Yeah, IBM takes research seriously.
03:02:05 <Gregor> !bfjoust aspartame_philip http://sprunge.us/OeJH
03:02:27 <EgoBot> Score for Gregor_aspartame_philip: 39.6
03:02:35 <Gregor> Wow, much worse than I expected X-P
03:03:05 <Gregor> !bfjoust aspartame_philip <
03:03:08 <EgoBot> Score for Gregor_aspartame_philip: 0.0
03:03:40 <Gregor> OHWAIT
03:03:41 <Gregor> I broke it :P
03:04:07 <Gregor> !bfjoust aspartame_philip http://sprunge.us/UUJX
03:04:12 <EgoBot> Score for Gregor_aspartame_philip: 43.1
03:04:22 <Gregor> Better
03:04:29 <quintopia> welcome back to the ring
03:05:00 <quintopia> egodeath is a really weird program that i need to look at and figure out
03:05:01 <Gregor> !bfjoust high_fructose_corn_philip <
03:05:09 <pikhq> ... Huh.
03:05:24 <pikhq> Deep Blue was a 11.38 GFLOPS supercomputer.
03:05:34 <EgoBot> Score for Gregor_high_fructose_corn_philip: 0.0
03:05:36 <quintopia> it was pretty poor in modern terms yes
03:06:06 <quintopia> if they were to set their sights on chess again, they'd own every grandmaster in the world
03:06:20 <quintopia> but that's not a profitable venture by any means
03:06:51 <quintopia> i think they should publish the deep blue algorithms
03:07:18 <quintopia> as i recall they had some rather innovative heuristics never seen before in chess AI
03:07:54 <pikhq> For comparison, the Cell processor hits 20.8 GFLOPS (for double-precision). In and of itself.
03:08:37 <pikhq> Or 211.2 GFLOPS (single-precision).
03:09:19 <quintopia> oh holy hell egodeath is a defender. :P
03:11:01 <quintopia> and it's my giant offsets that get me locked on it
03:13:46 -!- poiuy_qwert has quit (Read error: Connection reset by peer).
03:15:54 -!- poiuy_qwert has joined.
03:18:22 <Gregor> http://codu.org/eso/bfjoust/egojsout/?l=851f935e320b09c6ccf78337ab7f7c3c91583ce4&r=simple&t=25 This is a really elucidating poke :)
03:21:00 * pikhq ♥ Moore's
03:21:01 <pikhq> Law
03:21:13 <quintopia> !bfjoust space_elevator http://sprunge.us/YZca
03:21:14 <coppro> Gregor: <3 <# <#
03:21:28 <coppro> Gregor: I want those generally to see how coppro_suicide did so well
03:21:37 <EgoBot> Score for quintopia_space_elevator: 57.8
03:21:46 <quintopia> wow
03:22:21 <Gregor> THat's an improvement, I shook up the hill.
03:22:23 <Gregor> You were at 57.7
03:24:55 <quintopia> Gregor: is that two pipes as a pause? :P
03:25:24 <Gregor> quintopia: I made the opponent for all the examples be >(+)*64>(-)*64(>+>-)*3(>[-])*21
03:25:31 <Gregor> It's just enough to bring out behavior in almost everything.
03:25:40 <quintopia> i'm still not beating egodeath, even with smaller offsets :(
03:27:44 <quintopia> seriously though. where does philip come from? is it really just a pun on fill up?
03:30:53 <quintopia> if you pause at a particular frame and get a permalink, does it go ahead and run up to that frame and pause?
03:32:02 <Gregor> quintopia: No, it doesn't permalink on frames, only runs.
03:32:15 <Gregor> And philip is the name. But WHOSE name? :P
03:32:50 <Sgeo> Why is the more well known heart r=1-cos theta?
03:33:20 <Sgeo> I like r = |theta - pi/2| -pi/2 <= theta <= 3pi/2
03:34:04 <coppro> x^2 + (y - (|x|)^(1/2))^2
03:34:15 <coppro> (try it on WA)
03:35:23 <quintopia> Gregor: you have just proved that large decoys are sometimes better than small decoys :P
03:35:38 -!- poiuy_qwert has quit (Ping timeout: 240 seconds).
03:36:07 -!- poiuy_qwert has joined.
03:38:18 -!- Ilari has quit (Ping timeout: 245 seconds).
03:38:26 <Sgeo> http://www.wolframalpha.com/input/?i=r+%3D+|theta+-+pi/2|+from+-pi/2+to+3pi/2
03:38:27 <Sgeo> WTF
03:38:39 <Sgeo> Input interpretation says "r from" but graph has theta from
03:40:00 -!- Ilari has joined.
03:41:09 <coppro> Sgeo: not fan honestly
03:41:35 <Sgeo> I'm kind of partial to it since I stumbled on it by accident when I was younger
03:42:05 <quintopia> !bfjoust space_elevator http://sprunge.us/BhhN
03:42:18 <EgoBot> Score for quintopia_space_elevator: 58.0
03:43:12 <coppro> Gregor: can I use the input at the top to run an alternate program?
03:43:41 <Gregor> coppro: http://codu.org/eso/bfjoust/egojsout/
03:44:40 <coppro> awesome
03:44:50 <coppro> Gregor: what happens if you run off the end of your program; are you restarted?
03:45:09 <Gregor> pooppy: Eternal nopping.
03:45:56 <coppro> ok, that's what I thought
03:48:08 <Sgeo> coppro, you have to admit though, 1-cos(theta) sucks worse
03:52:40 <coppro> how does double_tripwire_avoider work? it just seems to run off the end
03:53:17 <quintopia> !bfjoust http://sprunge.us/EFPA
03:53:17 <EgoBot> Use: !bfjoust <program name> <program> . Scoreboard, programs, and a description of score calculation are at http://codu.org/eso/bfjoust/
03:53:23 <quintopia> !bfjoust space_elevator http://sprunge.us/EFPA
03:53:26 -!- pikhq has quit (Ping timeout: 240 seconds).
03:53:32 -!- pikhq has joined.
03:53:32 <EgoBot> Score for quintopia_space_elevator: 57.6
03:53:49 <quintopia> hmm
03:55:56 <quintopia> !bfjoust space_elevator http://sprunge.us/BhhN
03:56:05 <EgoBot> Score for quintopia_space_elevator: 58.0
03:56:14 <quintopia> coppro: it doesn't if you lay decoys fast enough
03:56:20 <quintopia> (against it)
03:56:32 <coppro> lol
04:05:00 <Gregor> !bfjoust aspartame_philip http://sprunge.us/KTWQ
04:05:07 <EgoBot> Score for Gregor_aspartame_philip: 39.7
04:05:12 <Gregor> ...wtf
04:05:22 <Gregor> Oh, I broke it again X_X
04:05:37 * Sgeo pulls a Phantom_Hoover and fills Gregor with benzene
04:05:37 <Gregor> !bfjoust aspartame_philip http://sprunge.us/RDhU
04:05:53 <EgoBot> Score for Gregor_aspartame_philip: 41.3
04:06:19 <Gregor> !bfjoust aspartame_philip http://sprunge.us/jRVO
04:06:30 <EgoBot> Score for Gregor_aspartame_philip: 41.3
04:06:30 <Gregor> (Trying again ... )
04:06:52 -!- azaq23 has quit (Ping timeout: 264 seconds).
04:06:57 <Gregor> !bfjoust aspartame_philip http://sprunge.us/TOYj
04:07:02 <EgoBot> Score for Gregor_aspartame_philip: 43.5
04:07:05 <Gregor> Hm
04:07:34 <quintopia> whatcha wigglin'?
04:07:43 <Gregor> Just the rush timing.
04:07:56 <Gregor> Slightly slower did better against defend13, but seems to do worse against most other things :P
04:08:07 <Gregor> (Which is unsurprising, but was worth a look)
04:08:22 -!- azaq23 has joined.
04:09:19 <Gregor> OHWAIT, I missed something ...
04:09:45 <Gregor> This will either do better or fail epically
04:09:51 <Gregor> !bfjoust aspartame_philip http://sprunge.us/fBdR
04:09:54 -!- augur has joined.
04:10:04 <EgoBot> Score for Gregor_aspartame_philip: 38.8
04:10:10 <Gregor> Hyuk
04:10:18 <Gregor> !bfjoust aspartame_philip http://sprunge.us/TOYj
04:10:30 <EgoBot> Score for Gregor_aspartame_philip: 43.5
04:11:12 <Gregor> Once more ...
04:11:23 <Gregor> !bfjoust aspartame_philip http://sprunge.us/NCdP
04:11:35 <EgoBot> Score for Gregor_aspartame_philip: 40.4
04:11:37 <Gregor> !bfjoust aspartame_philip http://sprunge.us/TOYj
04:11:43 <Gregor> Hmm hmm hmm.
04:11:45 <EgoBot> Score for Gregor_aspartame_philip: 43.5
04:16:53 <quintopia> inorite?
04:18:11 <quintopia> it's sort of like the way wireless is still doing so damn well somehow, while wireless_frownie slides slowly away
04:18:17 -!- sftp has quit (Remote host closed the connection).
04:20:15 <quintopia> !bfjoust poke http://sprunge.us/feYZ
04:20:23 <EgoBot> Score for quintopia_poke: 45.5
04:20:44 <Gregor> Wow, resubmitting poke?
04:21:13 <quintopia> now it draws with space_elevator XD
04:21:29 <quintopia> space_elevator just dropped 2 points in score
04:28:18 -!- pikhq_ has joined.
04:29:07 -!- pikhq has quit (Ping timeout: 276 seconds).
04:31:09 <quintopia> !bfjoust space_elevator http://sprunge.us/KHTd
04:31:47 <EgoBot> Score for quintopia_space_elevator: 58.9
04:31:53 <quintopia> :D
04:33:35 <Gregor> How can I fight it when it changes every two minutes :P
04:33:44 <Gregor> I should make non-suicide resubmissions illegal :P
04:33:54 <Gregor> So you always have to rename and a name always refers to one program.
04:34:04 <Gregor> (Note: I should not do this)
04:38:25 -!- calamari has joined.
04:39:19 -!- Mathnerd314 has quit (Remote host closed the connection).
04:41:18 -!- augur has quit (Read error: Connection reset by peer).
04:41:36 -!- asiekierka has joined.
05:02:46 -!- asiekierka has quit (Read error: Connection reset by peer).
05:18:28 -!- calamari has quit (Quit: Leaving).
05:25:35 <quintopia> Gregor: how to specify tape length in the wiki style permalink? can i just stick a &t=15 in there?
05:25:57 <Gregor> I didn't arrange the template to be able to :P
05:26:00 <Gregor> Gimme a sec, I'll set it up.
05:27:08 -!- pikhq_ has changed nick to pikhq.
05:27:23 <Gregor> OK, you should be able to add |t=15 now.
05:27:37 <Gregor> (|, not &, yay MediaWiki?)
05:34:15 <Sgeo> ....you're using MediaWiki in Codu?
05:34:21 * Sgeo is utterly confused right now
05:34:51 <Sgeo> So, my Systems Analysis and Design professor has background music on her page
05:34:59 <Sgeo> At least she's not the web design professor
05:38:32 <quintopia> Gregor: don't forget to make the template links generated on codu use the t= thing now too.
05:38:56 <Gregor> quintopia: Bleh :P
05:45:36 <quintopia> Gregor: p works too?
05:45:57 <Gregor> quintopia: Oh come on, do you honestly need to quicklink to different polarities? X-P
05:46:05 <Gregor> For the time being, just use |t=whatever&p
05:46:06 <quintopia> Gregor: yes
05:46:10 <quintopia> oh okay
05:46:16 <Gregor> It's actually a bit of a pain to work that one out ...
05:47:45 -!- VastOne has joined.
05:47:59 <VastOne> test
05:48:09 <Sgeo> F++
05:48:45 -!- pikhq has quit (Ping timeout: 240 seconds).
05:48:50 -!- pikhq has joined.
05:49:05 <Sgeo> Ping of death?
05:49:23 <pikhq> Seems that this router just goes "Fuck you, I'm going to drop all connections" every now and then.
05:49:25 <VastOne> me Sgeo?
05:49:39 <pikhq> Induced depending on load.
05:49:51 * pikhq suspects a damned shitty NAT implementation.
05:49:59 <pikhq> Sgeo: What about it?
05:50:14 <Sgeo> pikhq, just that you pinged out and I wanted to make a joke, is all
05:50:33 <Sgeo> Convert the world to IPv6 so you don't have to deal with crappy NAT
05:50:49 <pikhq> I'm afraid my ISP is moronic.
05:50:51 <pikhq> Like most are.
05:51:02 <VastOne> Arnt they all?
05:51:09 <pikhq> Though I *do* have an IPv6 tunnel, that ends up going through crappy NAT as well.
05:53:13 -!- VastOne has left (?).
05:57:34 -!- poiuy_qwert has quit (Ping timeout: 240 seconds).
05:58:02 -!- poiuy_qwert has joined.
05:59:49 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/GaFK
06:00:32 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 48.1
06:00:45 <Gregor> Why hallo thar
06:02:07 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/XGNV
06:02:23 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 47.0
06:02:36 <Gregor> Yeah, didn't think that'd be better.
06:03:13 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/JfLW
06:03:29 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 48.7
06:04:40 <quintopia> that's the best FFBG you've fielded, innit?
06:07:44 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/RjSA
06:07:56 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 48.7
06:07:59 <Gregor> I've only used FFBG for wildly-experimental things, nothing has worked before :P
06:09:16 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/bQDh
06:09:23 <Gregor> (Probably not better)
06:09:33 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 48.7
06:10:05 <Sgeo> Since when does BF Joust accept Perl? Or is it just a polygot?
06:10:15 <Sgeo> Or worse, not actually Perl kin any way
06:10:32 <Gregor> It's a polyglot.
06:11:57 * Sgeo sees i now
06:12:00 <Sgeo> it too
06:12:09 <Sgeo> I am not an it, prior statements to the contrary.
06:12:37 <Sgeo> I think I forgot a notwithstanding
06:12:51 <pikhq> And it appears to be the Perl code that generates the BF Joust program. :)
06:13:18 -!- azaq23 has quit (Quit: Leaving.).
06:13:19 <Sgeo> I don't think I want to imagine myself drunk
06:13:38 <pikhq> In addition to being a minorly cheating quine. :P
06:14:26 <Gregor> pikhq: It's only a pseudoquine, that's just for convenience.
06:14:40 <Gregor> Also, it's a MAJORLY cheating quine if I called it a quine :P
06:15:31 <pikhq> Gregor: Okay, okay, so "just reading the entire code in and printing it out" is a more major cheat.
06:17:01 <Sgeo> If I were drunk, I don't know if I'd be more of a "makes stupid jokes" person or a "kills everyone in sight" person
06:18:27 <pikhq> Sgeo: If it makes you feel better, getting drunk is kinda the province of the stupid.
06:18:38 <pikhq> ... Province? Odd word choice there.
06:18:43 <pikhq> Maybe I should sleep.
06:42:12 -!- aloril has quit (Ping timeout: 240 seconds).
06:48:10 -!- pikhq_ has joined.
06:48:48 -!- pikhq has quit (Ping timeout: 276 seconds).
06:51:40 <quintopia> Gregor: oicwutudidthar. you're doing what i did except more efficiently :P
06:51:56 -!- poiuy_qwert has quit (Ping timeout: 240 seconds).
06:52:01 <Gregor> Stealing is the sincerest form of flattery ;)
06:56:22 -!- aloril has joined.
06:56:28 <Gregor> I've still got more up my sleeve, if I don't decide I need to sleep first.
06:57:06 -!- poiuy_qwert has joined.
06:57:44 <Gregor> Naw, sleep first.
06:57:48 <quintopia> hmmm is what say to you good sir
06:57:52 <quintopia> i should sleep also :/
07:07:59 <quintopia> !bfjoust space_elevator http://sprunge.us/eagG
07:12:02 -!- FireFly has joined.
07:15:44 <EgoBot> Score for quintopia_space_elevator: 56.7
07:16:39 <quintopia> that was not an improve
07:28:22 -!- fizzie has quit (Read error: Operation timed out).
07:28:27 -!- fizzie has joined.
07:31:20 <fizzie> Gregor: Heh, your furry_furry_bondage_girls fails with cranklance with a "parse error: bad char after (): ;" due to "my ($fh, $line, $i, $j);" -- it does prune out (..)*0 comments, but only in a later stage; the count is read immediately after ). Maybe I should make it just treat ") followed by garbage" as ")*0".
07:34:41 -!- poiuy_qwert has quit (Ping timeout: 250 seconds).
07:35:23 -!- poiuy_qwert has joined.
07:38:42 <quintopia> !bfjoust space_elevator http://sprunge.us/ZXWS
07:39:30 <EgoBot> Score for quintopia_space_elevator: 55.3
07:43:48 <quintopia> oic
07:48:55 <quintopia> !bfjoust space_elevator http://sprunge.us/dJBH
07:49:05 -!- pikhq_ has quit (Remote host closed the connection).
07:49:25 -!- pikhq has joined.
07:52:23 <fizzie> http://zem.fi/~fis/dpoints15.png → http://zem.fi/~fis/cycles15.png where the latter shows (in log10 scale) the average number of cycles simulated in a particular game (all the way up to pure red = 10^5).
07:52:43 <fizzie> Not sure how useful that is, but at least definder2 stands out.
07:53:51 <quintopia> yes. definder2 is p much guaranteed to take forever when it's working.
07:55:09 <EgoBot> Score for quintopia_space_elevator: 55.4
07:55:19 <quintopia> ha
07:58:54 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+>-(>[-([+{[(+)*8[-].[.+]]}])%3])*21
07:59:59 -!- clog has quit (ended).
08:00:00 -!- clog has joined.
08:01:33 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+>-(>[-([+{[(+)*8[-].[.+]]}])%3])*21
08:05:37 <quintopia> !bfjoust space_elevator http://sprunge.us/SOaH
08:08:50 <quintopia> Deewiant: why did you submit the same identical program twice?
08:08:59 <Deewiant> Because breakdown.txt wasn't being generated
08:09:22 <quintopia> so you assumed it wasn't working?
08:09:26 <Deewiant> Yep
08:09:40 <quintopia> waiting usually works for me
08:10:16 <Deewiant> Nothing at all in almost three minutes made it seem unlikely
08:10:31 <Deewiant> The .bfjoust did go up, but breakdown wasn't even emptied
08:11:56 <EgoBot> Score for quintopia_space_elevator: 54.7
08:11:56 <EgoBot> Score for Deewiant_pendolino: 51.5
08:11:57 <EgoBot> Score for Deewiant_pendolino: 51.5
08:12:18 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
08:12:43 <EgoBot> Score for Deewiant_pendolino: 53.3
08:14:17 <fizzie> Also match lengths over all 68880 program/tape-length/polarity configurations: http://zem.fi/~fis/cycleh15.png (histogram in log-scale)
08:14:54 <quintopia> every change i make just makes it worse :P
08:15:25 <fizzie> Maybe I should try a logarithmically-spaced-bins histogram, but I don't see an option for that.
08:16:55 <quintopia> fizzie: is that spike at 60000 due to definder?
08:17:30 <Deewiant> I suspect it's sloth
08:17:33 <fizzie> Well, http://zem.fi/~fis/cycleh15b.png with X axis in log-scale by just applying log to the data.
08:23:07 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*10(>[-([+{[(+)*8[-.].[.+]]}])%3])*10>[-([+{[(+)*8[-].[.+]]}])%3]
08:23:17 <EgoBot> Score for Deewiant_pendolino: 49.6
08:24:42 <fizzie> Yes, it's mostly due to sloth.
08:25:18 <fizzie> http://p.zem.fi/1k46 ← number of matches in the [50000, 70000] range for everyone.
08:26:39 <Deewiant> !bfjoust -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3]+)*21
08:26:39 <EgoBot> Use: !bfjoust <program name> <program> . Scoreboard, programs, and a description of score calculation are at http://codu.org/eso/bfjoust/
08:26:45 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3]+)*21
08:26:50 <EgoBot> Score for Deewiant_pendolino: 50.9
08:36:14 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
08:36:23 <EgoBot> Score for Deewiant_pendolino: 53.3
08:41:23 <quintopia> prepare for suck
08:41:30 <quintopia> !bfjoust space_elevator http://sprunge.us/OEIj
08:41:41 <EgoBot> Score for quintopia_space_elevator: 53.7
08:47:11 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12<(-)*30<(-)*30<(+)*30<(+)*30(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
08:47:30 <EgoBot> Score for Deewiant_pendolino: 52.2
08:48:01 <Deewiant> !bfjoust pendolino -(>)*6(-)*12<(+)*12<(-)*30<(-)*30<(+)*30<(+)*30(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
08:48:06 <EgoBot> Score for Deewiant_pendolino: 52.2
08:48:23 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12<(+)*30<(-)*30<(-)*30<(+)*30(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
08:48:29 <EgoBot> Score for Deewiant_pendolino: 53.3
08:48:57 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
08:49:17 <EgoBot> Score for Deewiant_pendolino: 53.3
08:52:51 <quintopia> !bfjoust space_elevator http://sprunge.us/KHTd
08:53:01 <EgoBot> Score for quintopia_space_elevator: 56.7
08:58:25 -!- GreaseMonkey has quit (Quit: The Other Game).
09:08:13 <fizzie> http://zem.fi/~fis/clust15.png → http://zem.fi/~fis/clust15b.png if you just flip the input from the score table to the cycle-count table.
09:09:09 -!- MigoMipo has joined.
09:11:29 <fizzie> (And http://zem.fi/~fis/clust15c.png if you multiply them. :p )
09:12:08 -!- poiuy_qwert has quit (Ping timeout: 250 seconds).
09:13:56 -!- Leonidas has quit (Read error: Operation timed out).
09:14:01 -!- Leonidas has joined.
09:17:00 -!- poiuy_qwert has joined.
09:20:50 -!- Leonidas has quit (Ping timeout: 276 seconds).
09:21:09 -!- Leonidas has joined.
09:43:21 -!- MigoMipo has quit (Remote host closed the connection).
10:15:55 -!- Tritonio has joined.
10:26:47 -!- Tritonio has quit (Quit: Leaving).
10:29:15 -!- ais523 has joined.
10:32:12 -!- cheater99 has joined.
10:33:36 -!- cheater99 has quit (Client Quit).
10:33:53 -!- cheater99 has joined.
10:36:20 -!- cheater99 has quit (Client Quit).
10:36:35 -!- cheater99 has joined.
10:46:37 -!- cheater99 has quit (Ping timeout: 240 seconds).
10:47:30 -!- cheater99 has joined.
10:56:44 -!- cheater00 has joined.
10:57:01 -!- cheater99 has quit (Ping timeout: 240 seconds).
11:11:48 -!- cheater00 has quit (Ping timeout: 240 seconds).
11:13:22 -!- cheater00 has joined.
11:41:16 -!- FireFly has quit (Quit: swatted to death).
12:08:08 -!- variable has quit (Read error: Connection reset by peer).
12:09:34 -!- variable has joined.
12:28:07 -!- pikhq_ has joined.
12:28:35 -!- pikhq has quit (Ping timeout: 276 seconds).
12:29:17 -!- copumpkin has quit (Quit: Computer has gone to sleep.).
12:52:33 -!- poiuy_qwert has quit (Ping timeout: 240 seconds).
13:06:36 -!- roconnor has quit (Remote host closed the connection).
13:08:18 <Gregor> <fizzie> Gregor: Heh, your furry_furry_bondage_girls fails with cranklance with a "parse error: bad char after (): ;" due to "my ($fh, $line, $i, $j);" -- it does prune out (..)*0 comments, but only in a later stage; the count is read immediately after ). Maybe I should make it just treat ") followed by garbage" as ")*0".
13:08:36 <Gregor> fizzie: I was considering requesting a spec change to allow proper comments, since I also had to do stupid crap like:
13:08:39 <Gregor> #(
13:08:43 <Gregor> if (...) { ... }
13:08:44 <Gregor> #)%0
13:10:09 <Gregor> In fact,
13:11:51 <Gregor> ais523, elliott: Real nested comments. I want them. Any suggestions? I was thinking maybe a "bare" {} (outside of any ()s) would be considered a comment, since we're running out of pairable characters :P
13:12:40 <ais523> (comment (nested comment)*0 more comment)*0
13:13:05 <ais523> what you mean is, you want comments that allow arbitrary characters inside
13:13:23 <Gregor> ais523: Yes. See furry_furry_bondage_girls. It is a polyglot. It would be good to allow such honesty easily.
13:15:21 <ais523> you found a program befitting the name?
13:15:36 <Gregor> Sure, why not :P
13:17:27 <ais523> ah, it's a turtle?
13:17:34 <ais523> just with a crazy decoy setup?
13:17:38 <Gregor> It's just a cleverer philip.
13:17:55 <Gregor> (And a philip is just a turtle with some small-decoy avoidance)
13:18:15 <Gregor> I've got more up my sleeve, but that's for later.
13:18:24 <ais523> I was just verifying that definder2 won the way it was meant to against it, and it does
13:19:41 <Gregor> Anyway, any thoughts on comments :P
13:20:06 <Gregor> I know you generate some programs, and IM(Not-at-all)HO, it would be good for anybody who used code generators to present them as part of their solution.
13:21:31 <ais523> I don't generally use them for the whole thing
13:21:54 <ais523> I just use them in order to generate things like complex lock patterns
13:22:01 <Gregor> Mmmmmmm
13:22:06 <ais523> definder and definder2 were generated via the same program, for instance
13:22:10 <Gregor> Even so :P
13:22:11 <ais523> even though they have basically no code in common
13:22:36 <Gregor> Even so, an arbitrary comment would allow you (encourage you?) to include the relevant code in the relevant part.
13:22:47 <Gregor> (If not be a polyglot pseudoquine :P )
13:24:17 <ais523> sometimes the generation backfires horribly, like with defend9
13:24:21 <ais523> I've thought up a new strategy, anyway
13:27:57 -!- poiuy_qwert has joined.
13:32:55 -!- ais523_ has joined.
13:36:51 -!- FireFly has joined.
13:44:07 -!- miekko has changed nick to Zwaarddijk.
13:44:56 -!- cheater00 has quit (Ping timeout: 240 seconds).
13:47:49 -!- cheater00 has joined.
13:48:35 -!- Phantom_Hoover has joined.
13:48:39 -!- pikhq has joined.
13:49:01 -!- pikhq_ has quit (Ping timeout: 255 seconds).
13:49:41 -!- Sgeo_ has joined.
13:49:59 <Phantom_Hoover> What are the haps my friends
13:52:31 -!- Sgeo has quit (Ping timeout: 240 seconds).
14:16:55 <ais523_> Phantom_Hoover: we're busy documenting BF Joust strategies on the wiki
14:17:00 -!- poiuy_qwert has quit (Quit: This computer has gone to sleep).
14:17:03 <Phantom_Hoover> O god
14:31:22 <ais523_> why is decoytuner in the notable programs section? surely it was never at the top of the hill, given how bad it is?
14:31:27 <ais523_> decoytuner/2/ I can believe
14:33:21 <Phantom_Hoover> Great, this connection has slowed to a halt for no apparent reason.
14:33:31 <Phantom_Hoover> I can't get HTTP connections to *anything*.
14:33:47 <Phantom_Hoover> fungot, wait, does IRC still work?
14:33:48 <fungot> Phantom_Hoover: a role or position to which the player
14:34:35 <Phantom_Hoover> I can't get through to elliott's logs, which is highly distressing because there's nothing actually there.
14:34:40 <Phantom_Hoover> Hmm, maybe the DNS is down.
14:35:13 <Phantom_Hoover> Nope.
14:35:49 <Phantom_Hoover> pings to Google still go through, with times of ~50ms, which isn't strikingly long.
14:36:39 <Phantom_Hoover> ^style
14:36:40 <fungot> Available: agora* alice c64 ct darwin discworld europarl ff7 fisher ic irc jargon lovecraft nethack pa speeches ss wp youtube
14:36:44 <Phantom_Hoover> ^style youtube
14:36:44 <fungot> Selected style: youtube (Some YouTube comments)
14:37:03 <Phantom_Hoover> fungot, are you any good as a YouTube substitute?
14:37:04 <fungot> Phantom_Hoover: it doesn't mean they do! lol he was a discussion between those who speak a different way he could make a fly by
14:38:19 <Phantom_Hoover> "Packet filtered".
14:38:26 <Phantom_Hoover> Someone tell me what that means.
14:38:53 <fizzie> Sounds context-dependent.
14:39:03 <Phantom_Hoover> It was in a ping to irc.freenode.net.
14:39:22 <Phantom_Hoover> "From corv-car1-gw.nero.net ( icmp_seq=1 Packet filtered"
14:40:52 <fizzie> Hm. Maybe it prints that when the other end sends one of the weirder ICMP status codes, the "administratively prohibited" ones or so.
14:42:17 <Phantom_Hoover> I can't make another connection to Freenode through XChat, either.
14:43:18 <fizzie> That's weird; they have billion servers, after all.
14:43:29 <fizzie> PING ( 56(84) bytes of data.
14:43:29 <fizzie> From icmp_seq=2 Packet filtered
14:43:34 <fizzie> I get that too from one of them.
14:43:40 <Phantom_Hoover> Hmm, OK.
14:43:47 <Phantom_Hoover> But I still can't connect to anything.
14:44:07 <Phantom_Hoover> Pinging them works fine, but actually trying to do something inevitably fails.
14:45:03 <Phantom_Hoover> It happened suddenly, too; the connection worked fine for a while, then suddenly cut out.
14:45:43 <Phantom_Hoover> Plan B: see if this is happening at the router.
14:46:49 <Phantom_Hoover> ...or not, because it seems to be both modem and router.
14:51:34 -!- augur has joined.
14:51:47 <Phantom_Hoover> nmap says everything is down as well.
15:00:19 <ais523_> I've had the "existing IRC connections work but nothing else does" problem before now, too
15:01:16 <Phantom_Hoover> How did it resolve?
15:01:33 <Phantom_Hoover> And any way to see where it's going weird?
15:01:54 * Phantom_Hoover traceroutes.
15:03:00 -!- pikhq has quit (Ping timeout: 240 seconds).
15:03:04 -!- pikhq has joined.
15:03:46 <Phantom_Hoover> It seems to be able to get straight through to Google without a hitch, but very slowly.
15:05:17 -!- Sgeo_ has quit (Ping timeout: 240 seconds).
15:15:33 -!- ais523_ has quit (Quit: Page closed).
15:16:31 -!- sftp has joined.
15:16:56 <Phantom_Hoover> Uh-oh.
15:17:10 <Phantom_Hoover> `echo echo
15:17:16 <Phantom_Hoover> fungot, hello?
15:17:17 <fungot> Phantom_Hoover: nah theres australia mmmmmmmmate!!!! causes hallucinations, and you didn't reply directly to you kz 2!
15:17:32 <Phantom_Hoover> Thought I had lost this connection as well for a moment...
15:17:38 <HackEgo> echo
15:27:58 <ais523> Phantom_Hoover: you're still here
15:32:41 <Phantom_Hoover> I know.
15:32:54 <Phantom_Hoover> Ah, it fixed itself.
15:36:54 -!- azaq23 has joined.
15:43:13 -!- augur has quit (Remote host closed the connection).
15:52:01 -!- MigoMipo has joined.
15:54:36 -!- BeholdMyGlory has joined.
15:59:09 <Phantom_Hoover> http://i.min.us/il4KQQ.jpg
15:59:19 <Phantom_Hoover> I don't think this person realises that this already exists and it's hideous.
16:16:29 <Gregor> $ du -h furry_furry_dominance_girls.bfjoust
16:16:29 <Gregor> 71Mfurry_furry_dominance_girls.bfjoust
16:16:32 <Gregor> Hahawhoops
16:20:40 -!- Behold has joined.
16:23:37 <ais523> that could take a while to run
16:23:40 <ais523> even in lance
16:23:57 -!- BeholdMyGlory has quit (Ping timeout: 250 seconds).
16:24:12 <ais523> what do you think of [[e:BF Joust strategies]] now, btw?
16:24:19 <ais523> umm, http://esolangs.org/wiki/BF_Joust_strategies
16:24:25 <ais523> messed up my text replace there
16:28:23 -!- pikhq has quit (Ping timeout: 240 seconds).
16:28:43 -!- pikhq has joined.
16:29:22 <Phantom_Hoover> Wait, what/
16:29:27 <Phantom_Hoover> *?
16:30:27 <Gregor> We should set a size limit :P
16:30:29 <Gregor> Maybe 640K
16:30:53 <ais523> some programs need to go exponential to really work properly
16:30:58 <ais523> I like the idea of a size limit, though
16:31:23 <ais523> to force people to write at least vaguely Brainfucky code rather than just using [] as if and unrolling all control structures into a sequence of ifs
16:32:25 <Gregor> Yeah, exactly.
16:32:29 <Gregor> Preventing exactly what I just did :P
16:32:34 <Gregor> I can only unroll to depth 1 now.
16:33:02 <Gregor> (depth 1 is still superuseful though)
16:33:16 <Gregor> (Or would be if I didn't have an off-by-two(wtf) error)
16:34:24 <ais523> I unrolled let_sleeping_tripwires_lie to level 4 by hand, I think
16:36:23 <Gregor> Hahaha, watching full-tape clear to the finish would be soooo boring :P
16:36:47 <Gregor> (On any longer length)
16:37:52 <ais523> and incredibly CPU-intensive, too
16:38:09 <ais523> that's why I shortened the tape length for the defend7 example, as it uses a full-tape clear too
16:38:16 <Gregor> Oh cool, there are descriptions now.
16:38:28 <ais523> I added descriptions for my own programs
16:38:33 <ais523> because I knew what they did
16:39:03 <Gregor> That is a sensible reason :P
16:39:08 <Gregor> I should add descriptions for mine I suppose.
16:39:27 <Gregor> defend7 at this tape length is very elucidating.
16:39:53 <cheater00> look what digging up my browser history brought
16:40:00 <cheater00> an old paste with vagrant on it
16:50:21 <Gregor> I should adjust EgoJSout so that the parser function yields :P
16:52:35 <ais523> ah, is that what causes the freeze just after loading a large program, before running it?
16:56:30 -!- augur has joined.
16:56:47 -!- Sgeo has joined.
16:58:28 -!- pikhq_ has joined.
16:58:40 -!- pikhq has quit (Ping timeout: 260 seconds).
17:09:43 <ais523> wow defend9 is buggy, in all sorts of ways
17:09:48 <ais523> I only noticed checking it in egojsout
17:10:01 <ais523> but the detection is all screwed up
17:10:17 <ais523> I should definitely make a remake of it
17:16:49 <ais523> it detects simple, modified to be three-cycle, with the wrong polarity
17:20:11 <Gregor> Aren't'cha glad egojsout exists ^^
17:20:16 <Gregor> Although it's kinda borken right now :P
17:27:50 <Gregor> And now it's fixed, and you'll never know how it was broken.
17:31:35 -!- yorick has quit (*.net *.split).
17:31:35 -!- shachaf has quit (*.net *.split).
17:31:36 -!- aloril has quit (*.net *.split).
17:31:36 -!- rodgort has quit (*.net *.split).
17:31:36 -!- Zwaarddijk has quit (*.net *.split).
17:31:36 -!- Behold has quit (*.net *.split).
17:31:36 -!- variable has quit (*.net *.split).
17:31:37 -!- dbc has quit (*.net *.split).
17:31:38 -!- SimonRC has quit (*.net *.split).
17:31:38 -!- Deewiant has quit (*.net *.split).
17:31:38 -!- mtve has quit (*.net *.split).
17:31:38 -!- coppro has quit (*.net *.split).
17:31:39 -!- ais523 has quit (*.net *.split).
17:31:39 -!- ineiros has quit (*.net *.split).
17:31:39 -!- Phantom_Hoover has quit (*.net *.split).
17:31:40 -!- cal153 has quit (*.net *.split).
17:31:40 -!- sebbu has quit (*.net *.split).
17:31:40 -!- Mannerisky has quit (*.net *.split).
17:36:15 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/XeXB
17:36:53 <Gregor> Oh shoot, screwed it up :P
17:36:55 <Gregor> That's a suicider.
17:38:40 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 0.0
17:38:43 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/VUYG
17:38:55 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 0.0
17:39:13 <Gregor> wtf
17:39:46 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/VUYG
17:39:52 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 0.0
17:40:12 <Gregor> Doblewtf, the actual code is identical to what it was before >_<
17:40:34 <Gregor> Unless ... does egojoust not handle *0 correctly at all? :P
17:40:42 -!- dbc has joined.
17:41:43 -!- cal153 has joined.
17:41:43 -!- sebbu has joined.
17:41:43 -!- Mannerisky has joined.
17:41:53 <Gregor> !bfjoust (<)*0>(+)*64>(-)*64(>+>-)*3(>[-])*21
17:41:54 <EgoBot> Use: !bfjoust <program name> <program> . Scoreboard, programs, and a description of score calculation are at http://codu.org/eso/bfjoust/
17:42:07 <Gregor> !bfjoust testing_testing_one_two_three (<)*0>(+)*64>(-)*64(>+>-)*3(>[-])*21
17:42:11 <EgoBot> Score for Gregor_testing_testing_one_two_three: 0.0
17:42:14 <Sgeo> !bfjoust nop-or-suicide (<)*0
17:42:25 <EgoBot> Score for Sgeo_nop-or-suicide: 0.0
17:42:35 -!- aloril has joined.
17:42:36 -!- rodgort has joined.
17:42:36 -!- Zwaarddijk has joined.
17:42:49 <Gregor> OK, so with *0 not working, I'm honestly not sure what FFBG was even doing :P
17:43:15 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/NHgV
17:43:25 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 49.0
17:45:40 -!- Deewiant has joined.
17:45:40 -!- shachaf has joined.
17:45:40 -!- SimonRC has joined.
17:45:40 -!- Behold has joined.
17:45:40 -!- variable has joined.
17:45:40 -!- ais523 has joined.
17:45:40 -!- ineiros has joined.
17:45:40 -!- coppro has joined.
17:46:12 -!- coppro has quit (*.net *.split).
17:46:12 -!- shachaf has quit (*.net *.split).
17:46:12 -!- SimonRC has quit (*.net *.split).
17:46:12 -!- ais523 has quit (*.net *.split).
17:46:15 -!- ineiros has quit (*.net *.split).
17:48:23 -!- shachaf has joined.
17:49:02 -!- Sgeo has quit (Ping timeout: 240 seconds).
17:54:07 -!- sftp has quit (Remote host closed the connection).
17:54:34 -!- sftp has joined.
17:58:24 -!- SimonRC has joined.
17:58:24 -!- ais523 has joined.
17:58:24 -!- ineiros has joined.
17:58:24 -!- coppro has joined.
18:01:49 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/HXMB
18:02:16 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 49.0
18:07:03 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/OaYa
18:07:27 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 49.0
18:08:07 <Gregor> There we go, now I include the original, unwtfed Perl code by means of some very wtf Perl code which includes Base64-encoded code X-P
18:08:24 <ais523> Gregor: does % work correctly in egojoust atm?
18:08:35 <Gregor> ais523: Yes
18:08:43 <Gregor> ais523: So long as you don't nest as ({{}})
18:08:52 <ais523> ah, thanks
18:08:58 <ais523> I'm just testing defend9.5 locally atm
18:09:30 <ais523> it'll go onto the hill soon, and I have high hopes for it
18:09:49 <ais523> then I'll see if I can modify it to beat turtles and defence and see if it does even better
18:11:28 -!- SimonRC has quit (Ping timeout: 265 seconds).
18:11:36 -!- SimonRC has joined.
18:13:46 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/VGSh
18:13:52 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 51.0
18:14:03 <ais523> does egojsout show the code position debug data any more?
18:14:13 <ais523> and if so, how do I obtain it?
18:19:29 <Gregor> ais523: Oh darn, you found what I just fixed :P
18:19:37 <Gregor> If you haven't reloaded since I said I fixed something mysterious, you need to.
18:19:42 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/JPeQ
18:19:51 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 50.9
18:20:08 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/ZeaT
18:20:20 <Gregor> Probably neither of these will do better...
18:20:21 <ais523> I figured out the bug in my program a different way
18:20:25 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 50.9
18:20:40 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/hePS
18:20:48 <EgoBot> Score for Gregor_furry_furry_bondage_girls: 51.0
18:21:01 <ais523> !bfjoust defend9.5 http://sprunge.us/VQDW
18:21:28 <EgoBot> Score for ais523_defend9_5: 23.2
18:21:36 <ais523> hmm, that low?
18:22:58 -!- sebbu has quit (Changing host).
18:22:59 -!- sebbu has joined.
18:23:13 -!- hirnstein has joined.
18:23:14 <ais523> it seems it's beating everything it's meant to, it's just that not countering defence means you do badly nowadays
18:24:48 <hirnstein> moin, what doo u think about tyche,nibiru, nemesis...
18:29:35 <ais523> hirnstein: are you here to talk about programming?
18:29:41 <ais523> if not, this is likely the wrong channel, and the wrong server
18:29:41 -!- SimonRC has quit (Ping timeout: 265 seconds).
18:29:46 -!- ais523 has quit (*.net *.split).
18:29:46 -!- ineiros has quit (*.net *.split).
18:30:04 -!- coppro has quit (*.net *.split).
18:30:07 -!- coppro has joined.
18:30:10 -!- ineiros has joined.
18:30:34 -!- SimonRC has joined.
18:31:18 -!- ais523 has joined.
18:31:55 -!- hirnstein has left (?).
18:34:01 <ais523> hmm, I know what I /want/ to write here, but I fear it would kill egojoust
18:34:30 <ais523> Gregor: will egojoust hate me if I put a huge program (several kilobytes) inside a ()%1000 loop? or can it do that efficiently nowadays?
18:34:37 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+----]]}])%3])*21
18:34:43 <EgoBot> Score for Deewiant_pendolino: 50.6
18:34:46 <Gregor> ais523: It can't do that efficiently :P
18:34:56 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.++-------]]}])%3])*21
18:34:59 <ais523> OK, looks like I'll have to think up a different way to do this
18:35:17 <EgoBot> Score for Deewiant_pendolino: 51.0
18:35:47 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+++------]]}])%3])*21
18:36:09 <EgoBot> Score for Deewiant_pendolino: 51.9
18:36:14 -!- Sgeo_ has joined.
18:36:42 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
18:36:50 <EgoBot> Score for Deewiant_pendolino: 52.7
18:38:16 -!- pikhq_ has quit (Ping timeout: 245 seconds).
18:38:22 -!- pikhq has joined.
18:40:13 -!- cheater00 has quit (Ping timeout: 240 seconds).
18:40:50 -!- variable has quit (Remote host closed the connection).
18:41:22 -!- cheater00 has joined.
18:42:30 -!- variable has joined.
18:46:50 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[.+---]]}])%3])*4(>[-([+{[(+)*8[-].[.+]]}])%3])*17
18:46:55 <EgoBot> Score for Deewiant_pendolino: 36.6
18:48:51 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+>[-([+{[(+)*8[.+---]]}])%3](>[-([+{[(+)*8[-].[.+]]}])%3])*20
18:48:59 <EgoBot> Score for Deewiant_pendolino: 44.8
18:50:41 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-][.+]]}])%3])*21
18:50:47 <EgoBot> Score for Deewiant_pendolino: 49.7
18:51:33 <Deewiant> !bfjoust pendolino -(>)*6(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
18:51:39 <EgoBot> Score for Deewiant_pendolino: 52.7
18:55:42 <ais523> !bfjoust defend9.75 http://sprunge.us/BMhZ
18:55:49 <ais523> this has been beating all but the weirdest programs in my testing
18:56:02 <ais523> although it's moderately easy to counter by targeting it specifically
18:57:21 <ais523> hmm, it hasn't even managed to run even one program yet
18:57:39 <ais523> I didn't use a % loop, but nested * loops, which egojoust isn't that bad at
18:58:35 <ais523> OK, I'll make it easier for egojoust to handle
18:58:57 <ais523> !bfjoust defend9.75 http://sprunge.us/KeYI
19:00:55 <ais523> bleh, still not running
19:01:12 <ais523> let me see if I can make the loop even shorter without losing to something I should beat
19:02:46 <Gregor> I have a nifty idea (stolen of course) which is presently not working at all :P
19:03:53 <ais523> !bfjoust defend9.75 http://sprunge.us/badj
19:03:56 <EgoBot> Score for ais523_defend9_75: 0.5
19:03:57 <EgoBot> Score for ais523_defend9_75: 0.5
19:03:58 <EgoBot> Score for ais523_defend9_75: 0.5
19:04:04 <ais523> that's... wrong
19:04:10 <Gregor> lawlehcoptahs
19:04:30 <ais523> it beats space_elevator convincingly on egojsout
19:04:34 <ais523> so I'm going to assume it's an egojoust bug
19:04:36 <Deewiant> It only played against maglev?
19:04:45 <Deewiant> And report.txt is broken
19:04:47 <Gregor> I'm thinkin' if elliott isn't done with lance by the weekend, I'm just gonna write egojoust2. Probably that just runs into some stupid egojoust bug >_<
19:04:53 <ais523> breakdown shows it beating maglev
19:05:01 <ais523> let me rerun it and see what happens
19:05:02 <fizzie> Gregor: You can always start cranking cranklance. :p
19:05:14 <ais523> !bfjoust defend9.75 http://sprunge.us/badj
19:05:16 <Gregor> fizzie: I keep forgetting about it because you're not loud enough :P
19:05:26 <Gregor> fizzie: I'll look at it tonight, relink me the source in PM?
19:05:35 <fizzie> I don't like to "tootle my own horn", knowwhatImean.
19:05:51 <fizzie> I'm not sure "relink" is the right word, considering I haven't shown it to anyone yet.
19:06:48 <fizzie> It's also not in egojoust-compatible output format, since I've added cycle-count dumps and such for those visualizations, so it's not such a turnkey solution as you get from elliott incorporated.
19:07:01 <EgoBot> Score for ais523_defend9_75: 36.6
19:07:06 <ais523> Gregor: it's definitely hitting an egojoust bug; it beats saccharine_philip on all 42 runs on egojsout, but loses most of them on egojoust
19:07:13 -!- Sgeo_ has quit (Ping timeout: 240 seconds).
19:08:39 <ais523> ah, no
19:08:55 <ais523> it's my shortening of the main loop so that egojoust can tolerate it that's making it lose to saccharine_philip
19:09:25 -!- augur has quit (Read error: Connection reset by peer).
19:09:25 -!- augur_ has joined.
19:09:25 <Gregor> Don't bother, just let egojoust suffer.
19:09:30 -!- augur_ has changed nick to augur.
19:10:37 <ais523> hmm, no, not that either
19:10:41 <ais523> it's something else that incidentally changed
19:11:54 <ais523> I must have put the wrong program into egojsout
19:11:59 <ais523> that 36.6 seems correct, now
19:12:24 <fizzie> http://zem.fi/~fis/cranklance-beta.tar.gz if someone likes to sneak a peek; not exactly my finest work ever, since it was supposed to be just a throwaway experiment. (Also I want to add a duplicate set of op_X's that it can switch to after the other program dies, for faster running of the live one.)
19:13:02 <Deewiant> !bfjoust sm3 http://sprunge.us/hMUX
19:13:16 <EgoBot> Score for Deewiant_sm3: 48.6
19:13:16 <fizzie> Deewiant: You're such a train.
19:13:22 <Deewiant> I know.
19:13:37 <fizzie> Are you an official train enthusiast person guy, in fact?
19:13:41 <Deewiant> No.
19:14:10 <fizzie> Aw; could've fool'd me.
19:14:21 <Deewiant> !bfjoust sm3 http://sprunge.us/BfBV
19:14:40 <EgoBot> Score for Deewiant_sm3: 13.7
19:15:32 <Deewiant> !bfjoust sm3 http://sprunge.us/MEWB
19:15:58 <EgoBot> Score for Deewiant_sm3: 50.7
19:16:39 <Deewiant> !bfjoust sm3 http://sprunge.us/FCFO
19:16:53 <EgoBot> Score for Deewiant_sm3: 49.2
19:18:16 -!- pikhq has quit (Ping timeout: 245 seconds).
19:18:16 -!- pikhq_ has joined.
19:18:27 -!- augur has quit (Read error: Connection reset by peer).
19:18:43 -!- augur has joined.
19:22:21 -!- augur_ has joined.
19:22:53 -!- augur has quit (Disconnected by services).
19:22:58 -!- augur_ has changed nick to augur.
19:24:07 <Deewiant> !bfjoust sm3 http://sprunge.us/biVY
19:24:32 <EgoBot> Score for Deewiant_sm3: 50.7
19:26:17 -!- FlyingTortilla has joined.
19:26:19 <Deewiant> !bfjoust sm3 http://sprunge.us/FFci
19:26:44 <Gregor> ... goddamnit.
19:26:49 <EgoBot> Score for Deewiant_sm3: 44.7
19:27:04 <ais523> !bfjoust defend9.75 http://sprunge.us/EjhJ
19:27:06 <ais523> Gregor: ?
19:27:13 <Gregor> 50.7
19:27:27 <ais523> oh, does that beat one of your programs?
19:27:59 <ais523> hmm, there seems to be a threshold between "egojoust runs quickly" and "egojoust doesn't run at all"
19:28:05 <ais523> and it's somewhere between 9 and 16 iterations
19:28:15 * ais523 tries 12
19:28:20 <Gregor> lawl
19:28:33 <Deewiant> !bfjoust sm3 http://sprunge.us/HSjE
19:28:41 <ais523> !bfjoust defend9.75 http://sprunge.us/NKhG
19:28:48 <Gregor> $ du -h furry_furry_leather_discipline_girls.bfjoust
19:28:49 <Gregor> 1.6Mfurry_furry_leather_discipline_girls.bfjoust
19:28:57 <Gregor> I am doing a TERRIBLE job of not writing exponential programs >_<
19:29:42 -!- augur has quit (Read error: Connection reset by peer).
19:29:48 <EgoBot> Score for Deewiant_sm3: 44.9
19:29:49 <EgoBot> Score for ais523_defend9_75: 0.6
19:29:50 <EgoBot> Score for ais523_defend9_75: 0.6
19:30:00 -!- augur has joined.
19:30:04 <ais523> !bfjoust defend9.75 http://sprunge.us/NKhG
19:30:16 <ais523> the low scores seem a symptom of replacing a hard-to-run program with an easy-to-run program
19:30:23 <ais523> and can be fixed just by rerunning it
19:30:45 <ais523> 23 | + + + + + + + + + + + + + + + + + + + + + + + - - - - - - - - - - 0 0 - - - - - - - - - - - | 0.6 | 0.0 | 23 | ais523_defend9_75.bfjoust
19:30:47 <ais523> hahahaha
19:30:47 <Gregor> Probably more like replacing an in-progress program with another program.
19:30:53 <ais523> yep
19:31:06 <Deewiant> !bfjoust sm3 <
19:31:34 <ais523> !bfjoust defend9.75 http://sprunge.us/ILYM
19:31:42 <ais523> back to 9 iterations
19:31:49 <ais523> which is /just/ enough to beat out the slowest programs on the hill
19:31:53 <ais523> it should be much higher, though
19:32:11 <ais523> I shouldn't have to tweak constants just to get egojoust to work
19:32:41 <Gregor> Actually FFLDG isn't even exponential, just high-polynomial :P
19:32:57 <Gregor> ais523: egojoust needs to go, this is well-established :P
19:33:56 -!- augur_ has joined.
19:33:57 <ais523> I'm not sure if 9.75 will hit the top of the leaderboard when it's properly tweaked
19:34:01 <ais523> but it looks like it'll be high up
19:34:04 -!- augur has quit (Read error: Connection reset by peer).
19:34:40 <Deewiant> It's just 9 with bugs fixed?
19:34:52 <EgoBot> Score for ais523_defend9_75: 19.6
19:34:52 <EgoBot> Score for Deewiant_sm3: 43.4
19:34:53 <EgoBot> Score for ais523_defend9_75: 19.6
19:34:59 <ais523> !bfjoust defend9.75 http://sprunge.us/ILYM
19:35:01 <ais523> Deewiant: not any more
19:35:08 <ais523> 9.5 is basically a bugfixed 9
19:35:18 -!- augur_ has changed nick to augur.
19:35:21 <Deewiant> Whyy did < get 43.4 :-P
19:35:24 <ais523> 9.75 is 9.5, plus a bunch of modifications that have improved every defence program I've put them on
19:35:33 <ais523> Deewiant: haha, that's amazing
19:36:02 <ais523> `addquote <Deewiant> !bfjoust sm3 < <EgoBot> Score for Deewiant_sm3: 43.4
19:36:03 <Deewiant> Running more than one program simultaneously breaks stuff quite a bit
19:36:05 <ais523> yep
19:36:25 <Deewiant> It'd be nice if the server handled it so we don't have to worry about it :-P
19:36:50 <HackEgo> 306) <Deewiant> !bfjoust sm3 < <EgoBot> Score for Deewiant_sm3: 43.4
19:37:11 <ais523> that must be a record for the self-suiciding program
19:37:29 <Gregor> lawl, after reducing the depth to produce shorter code, I have created the world's worst tripwire: when tripped, it does nothing, and I continue on my merry way.
19:37:52 <ais523> could still be useful for synchronization
19:37:57 <EgoBot> Score for ais523_defend9_75: 45.0
19:38:03 <ais523> unless it's a reverse tripwire which you keep checking even when tripped
19:38:53 <ais523> the loss to slowrush looks like a bug, btw
19:39:06 <ais523> so I think it can go higher
19:43:34 <ais523> yay, found the bug
19:44:30 <ais523> I wonder if slowrush's offset of 22 was chosen because 22*3 > 64?
19:44:43 <ais523> because it beats certain defensive algos quite well for that reason
19:45:05 <ais523> !bfjoust defend9.75 http://sprunge.us/OdLA
19:45:52 -!- FlyingTortilla has quit (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203074432]).
19:47:27 <EgoBot> Score for ais523_defend9_75: 48.0
19:52:53 <ais523> hmm, superfast_jellyfish uses a 7-cycle clear
19:52:54 <ais523> I think that's detectable too
19:54:06 <Gregor> Bleh, without an else branch, it is so difficult to make sure you're always where you think you are without unrolling 'til you have 1.6MB programs :P
19:56:05 <Deewiant> You can have elses by nesting 'til you have 2.1MB programs
19:58:38 <quintopia> ais523: defend9_75 looks exactly like something i was thinking of implementing. stop reading my mind.
19:59:13 -!- comex_ has joined.
19:59:45 -!- ineiros_ has joined.
19:59:47 <quintopia> although, you didn't quite get all my ideas
19:59:54 <quintopia> i think i can surprise
19:59:58 <ais523> !bfjoust 9.75 http://sprunge.us/bdUL
20:00:04 <ais523> did you get all mine?
20:01:01 <ais523> I suppose it's impossible for us to collaborate on a "perfect" BF Joust program, because everything's beatable
20:01:11 <ais523> whoops, got the name wrong
20:01:12 <quintopia> yes
20:01:14 <ais523> I'll delete the old version once the new version processes
20:01:24 -!- Ilari_ has joined.
20:01:30 <quintopia> i'm not quite sure because i haven't quite figured out what you're doing yet
20:01:52 <ais523> actually, I'm not convinced the new version's better, the fix I tried to use doesn't actually work
20:01:52 -!- elliott has joined.
20:01:56 <ais523> hi elliott!
20:02:11 <EgoBot> Score for ais523_9_75: 45.4
20:02:16 <elliott> hi
20:02:19 <elliott> huh
20:02:20 -!- ineiros has quit (*.net *.split).
20:02:20 -!- Ilari has quit (*.net *.split).
20:02:20 -!- comex has quit (*.net *.split).
20:02:20 -!- fungot has quit (*.net *.split).
20:02:28 <ais523> !bfjoust 9.75 <
20:02:32 <ais523> wrong name
20:02:34 <EgoBot> Score for ais523_9_75: 0.0
20:02:41 <quintopia> my idea was to first detect polarity and offset size using the ([{}])% style tripwires, two if necessary, and then detect cycle length with another regular tripwire.
20:02:46 <quintopia> is that what you're doing?
20:02:49 <Deewiant> !bfjoust choloepus >+++<(++----+)*8192(>)*8(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:03:06 <EgoBot> Score for Deewiant_choloepus: 23.9
20:03:09 <ais523> quintopia: no, I'm detecting polarity and cycle length with one
20:03:24 <ais523> and mathmaticising around the offset size
20:03:39 <Deewiant> !bfjoust choloepus >+++<(++----+)*768(>)*8(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:03:49 <EgoBot> Score for Deewiant_choloepus: 25.3
20:03:51 <elliott> 22:09:16 <Gregor> !bfjoust furry_furry_bondage_girls http://sprunge.us/bQDh
20:03:53 <elliott> wow @ polyglot
20:04:03 <Deewiant> !bfjoust choloepus >+++<(++----+)*256(>)*8(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:04:10 <quintopia> ais523: are you ensuring that even large offsets/careless clears never pass zero?
20:04:15 <ais523> anyway, defend9.75 beats everything above it on the leaderboard
20:04:16 <EgoBot> Score for Deewiant_choloepus: 25.1
20:04:24 <quintopia> (and by pass, i mean see)
20:04:26 <elliott> Gregor: since when do you code perl?
20:04:30 <Gregor> elliott: 'twas a borken polyglot due to egojoust, but also due to having no real comments as parens still require matching ({}) within them.
20:04:35 <elliott> Gregor: since when do you code perl?
20:04:37 <Gregor> elliott: It's part of my vast lexicon :P
20:04:39 <Deewiant> !bfjoust choloepus >+++<(++--)*256(>)*8(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:04:58 <ais523> quintopia: careless clears are dealt with a different way; for large offsets I use a two-cycle tripwire
20:05:04 <EgoBot> Score for Deewiant_choloepus: 15.9
20:05:25 <elliott> 23:31:20 <fizzie> Gregor: Heh, your furry_furry_bondage_girls fails with cranklance with a "parse error: bad char after (): ;" due to "my ($fh, $line, $i, $j);" -- it does prune out (..)*0 comments, but only in a later stage; the count is read immediately after ). Maybe I should make it just treat ") followed by garbage" as ")*0".
20:05:27 <elliott> no, that's indeed an error
20:05:32 <Deewiant> !bfjoust choloepus <
20:05:33 <quintopia> ais523: try my idea then :P i have to study today
20:05:44 <EgoBot> Score for Deewiant_choloepus: 0.0
20:05:45 <ais523> well, I have to work too
20:05:51 <ais523> and you might run out of decoy space
20:05:59 <ais523> defend9.75 more or less autoloses on short tapes as it is
20:06:09 <ais523> due to being rule-of-nined into oblivion
20:06:48 <elliott> 05:11:51 <Gregor> ais523, elliott: Real nested comments. I want them. Any suggestions? I was thinking maybe a "bare" {} (outside of any ()s) would be considered a comment, since we're running out of pairable characters :P
20:06:56 <ais523> it beats everything above interior_crocodile_alligator, at least
20:06:56 <elliott> Gregor: No comments in BF-esques kthx
20:07:07 <elliott> ()*0 is for notes, not #if 0 :P
20:07:29 <quintopia> ais523: yeah my idea would suffer from the same problem, but that's fine
20:07:30 <elliott> Gregor: It's hardly "honesty". You could just do
20:07:33 <elliott> ( Generator: http://sprunge.us/... )*0
20:07:35 <elliott> at the start.
20:07:36 <elliott> :P
20:07:43 <elliott> Gregor: Polyglots are meant to be difficult, yo.
20:07:53 <Deewiant> !bfjoust sm3 -(>)*8(>[-([+{[(+)*8[-].[.+]]}])%3])*2(<)*4(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:08:01 <EgoBot> Score for Deewiant_sm3: 47.3
20:08:02 <ais523> wait, why does defend9.75 beat superfast_jellyfish on kettle polarity?
20:08:05 <ais523> that makes no sense
20:08:09 * ais523 investigates
20:08:11 <elliott> ais523: hTF did interior_crocodile_alligator get so high up?
20:08:22 <quintopia> elliott: he just said why
20:08:25 <ais523> I don't know
20:08:25 <Deewiant> Woot, I bumped pendolino to #1
20:08:28 <elliott> no he didn't
20:08:29 <Deewiant> That wasn't intentional
20:08:45 <elliott> Gregor: http://codu.org/eso/bfjoust/egojsout/
20:08:45 <elliott> What's "simple"
20:08:48 <ais523> it's doing something that my defenders dislike, I need to look into it
20:08:50 <ais523> elliott: it's the example program used on the wiki
20:08:57 <ais523> to demonstrate how other programs work
20:09:10 <elliott> Gregor: Ehm, whenever I use egojsout, the animation plays and starts playing without me even asking for an animation.
20:09:11 <Gregor> elliott: It's just complicated enough to bring out behavior in (most) other programs, without actually being very good.
20:09:13 <ais523> it's a stupid rush program that contains very few countermeasures, and does things itself that are interesting to show off other programs against
20:09:24 <Gregor> elliott: I no longer have a do-or-don't-animate checkbox. Deal with it.
20:09:35 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*2(<)*4(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:09:46 <EgoBot> Score for Deewiant_sm3: 48.5
20:10:09 <elliott> Gregor: That's incredibly irritating since I mostly read traces.
20:10:10 <elliott> (And I think so do most other people)
20:10:12 <ais523> Deewiant: still trying to revive the era of one-liners?
20:10:25 <quintopia> he's doing a fine job of it :P
20:10:28 <Deewiant> Hey, half of the top 6 are one-liners
20:10:29 <Gregor> elliott: So what? Animation should chew up very little CPU unless your browser sucks foot.
20:10:51 <ais523> Gregor: it chews up a lot in both Firefox and Chromium generating the traces
20:10:54 <quintopia> ^
20:10:56 <elliott> Gregor: Dude, I don't care about CPU, I care about it being distracting and having to pause it explicitly every single time I look at a trace, and also the fact that it wastes screen space.
20:11:00 <ais523> but presumably it's the traces that are problematic, not the animation
20:11:08 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*2(<)*4(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:11:13 <ais523> the screen space doesn't bother me because I can scroll
20:11:15 <elliott> (I only have 900 vertical pixels.)
20:11:16 <Gregor> ais523, quintopia: Animation only starts after the trace is finished being collected.
20:11:17 <quintopia> Gregor: i agree that it should not autoplay.
20:11:17 <EgoBot> Score for Deewiant_sm3: 51.2
20:11:19 <ais523> starting it paused by default would make sense, though
20:11:20 <elliott> ais523: Still sucks for analysing early games.
20:11:31 <elliott> Gregor: How about this: Float the animation right, and pause it by default.
20:11:34 <quintopia> Gregor: it wasn't doing that yesterday. sometimes it was getting confused and playing sooner
20:11:36 <Gregor> Bleh, you people and your extremely minor concerns.
20:11:39 <elliott> Then it takes up no vertical screen space and isn't annoying.
20:11:45 <Deewiant> Thus falls space_elevator
20:11:59 <Gregor> I hate you all so much.
20:12:02 <Gregor> You have no idea how much I hate you all.
20:12:03 <ais523> defend9.75 beats it pretty convincingly
20:12:10 <Gregor> It is a burning, firey hatred.
20:12:21 <ais523> customers are never satisfied, eh?
20:12:22 <Deewiant> Some others do too, none of mine though
20:12:31 <quintopia> i decided last night i was done with space_elevator and ready to move on to other things
20:12:32 <ais523> egojsout's pretty useful, though, I went and added examples using it to the wiki, in addition to the ones already there
20:12:33 <Gregor> ais523: PAY ME, THEN YOU WILL BE A CUSTOMER.
20:12:36 <elliott> 06:34:35 <Phantom_Hoover> I can't get through to elliott's logs, which is highly distressing because there's nothing actually there.
20:12:46 <elliott> Herobrine is dead or whatever.
20:12:55 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[.++-------]]}])%3])*2(<)*4(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:13:04 <elliott> ais523: I like egojsout, I just don't like the only thing I use egojsout for made awkward and distracting :P
20:13:13 <elliott> The animation is cool but not useful most of the time IMO
20:13:19 <EgoBot> Score for Deewiant_sm3: 43.4
20:13:20 <ais523> I find it useful
20:13:32 <ais523> and use the traces when I need more detail
20:13:36 <quintopia> sometimes you can see what happens faster using the animation than the trace
20:13:38 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[.-]]}])%3])*2(<)*4(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:13:40 <ais523> it's excellent for visualising strategies, too
20:13:46 <elliott> ais523: If it was floated-right and automatically paused, you could see it just as well with one click.
20:13:49 <EgoBot> Score for Deewiant_sm3: 48.2
20:13:51 <elliott> And also not bother me in the slightest.
20:13:54 <elliott> Everyone wins. :p
20:14:03 <ais523> yep
20:14:11 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*4(<)*6(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:14:12 <ais523> or if the traces were floated right, I suppose
20:14:17 <elliott> ais523: that would look weird
20:14:17 <quintopia> wot
20:14:23 <elliott> there'd be an awful lot of blank space on the left for most of the page
20:14:23 <ais523> Deewiant: tweaking constants?
20:14:29 <ais523> perhaps
20:14:29 <Deewiant> As usual
20:14:30 <EgoBot> Score for Deewiant_sm3: 48.4
20:14:58 <Deewiant> !bfjoust pendolino -(>)*6(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*5>->+(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:15:13 <EgoBot> Score for Deewiant_pendolino: 52.2
20:15:23 <elliott> Simple beats interior crocodile alligator :(
20:15:31 <ais523> still, the issue remains: htf is defend9 beating a 7-cycle clear? (I wonder if it's triggering the "stupidly slow attacker, become a fast rush" codepath?)
20:15:38 <ais523> elliott: how does interior crocodile alligator work?
20:15:38 <elliott> Gregor: Also why are there two permalink buttons :P
20:15:46 <elliott> ais523: with syntax errors AFAICT
20:15:46 <ais523> elliott: one's a permalink to the programs, the other to the individual run
20:15:55 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*4(<)*6(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*11(>[-([+{[(+)*8[-].[.+]]}])%3])*17
20:15:57 <elliott> hmm, i don't actually get syntax errors now from egojsout on it
20:15:58 <elliott> how odd
20:16:10 <EgoBot> Score for Deewiant_sm3: 50.7
20:16:12 <quintopia> fizzie: i don't know how hard it would be to do in cranklance+scipy, but i'd like to see a graph of average absolute decoy structure (aka absolute value of decoy sizes assuming [-127,128] as a function of distance from flag)
20:16:21 -!- Sgeo has joined.
20:16:24 <ais523> Deewiant: well done at breaking 50
20:16:27 <ais523> that's becoming harder and harder nowadays
20:16:36 <ais523> because the hill is more varied and more evenly matched
20:16:41 <Deewiant> I'm almost pushing space_elevator back to #1, that's annoying :-P
20:16:41 <elliott> ais523: basically, it's a standard (>)*9(clear>)*20clear program
20:16:46 <elliott> ais523: but with a really confusing clear
20:16:50 <ais523> ah, OK
20:16:54 <ais523> no wonder it beats defenders
20:17:01 <ais523> and presumably loses to fast rushes
20:17:04 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*2(<)*8(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*9(>[-([+{[(+)*8[-].[.+]]}])%3])*19
20:17:07 <elliott> ais523: first, it goes into a loop, and if the value is non-zero, it decreases it by one
20:17:11 <elliott> to get 1-decoys
20:17:16 <elliott> then it enters another loop, which is skipped if it was 1 ofc
20:17:18 <elliott> and tries ++
20:17:19 <elliott> because
20:17:22 <ais523> yep, that's standard
20:17:24 <elliott> FF -> FE because of the first - -> FF -> 00
20:17:28 <EgoBot> Score for Deewiant_sm3: 0.0
20:17:31 <ais523> I've done that in some of my own programs too
20:17:34 <ais523> as has Gregor
20:17:37 <Deewiant> Woops
20:17:39 <ais523> and probably other people as well
20:17:43 <elliott> ais523: but /then/, it guesses that it's a flag, and takes 128 off (despite the fact that this actually turns out to be (-)*127 because of the -++)
20:17:49 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*2(<)*4(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*9(>[-([+{[(+)*8[-].[.+]]}])%3])*19
20:17:52 <elliott> ais523: _then_ it does [-[-[-(64 of those)
20:17:56 <EgoBot> Score for Deewiant_sm3: 52.8
20:17:57 <elliott> ais523: then [+[+[+(64 of those)
20:17:57 <Sgeo> So, what happened with the ()*0 issue and the mysterious furries?
20:17:59 <elliott> ais523: then [-]
20:18:02 <elliott> and then so, so many ]s
20:18:08 <ais523> ah, no wonder, it's too slow to detect
20:18:14 <quintopia> and it is the ]s that give it its power
20:18:15 <Gregor> Sgeo: I did it a different way.
20:18:36 <ais523> elliott: does it even set decoys?
20:18:37 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*7(<)*9(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*14(>[-([+{[(+)*8[-].[.+]]}])%3])*14
20:18:38 <Sgeo> Do tell?
20:18:40 <elliott> ais523: so basically, it aims for 1, then FF, then... 0x81 :P, and then it goes down for a while, up for a while, then down
20:18:42 <quintopia> what is an sm3?
20:18:49 <EgoBot> Score for Deewiant_sm3: 40.0
20:18:52 <elliott> ais523: and, depending on how far it gets into the loops, then sits around for _ages_ after clearing a cel
20:18:55 <elliott> which seems to help confuse defenders
20:18:55 <Sgeo> Also: Did we ever find out what furries with false comments does?
20:18:56 <elliott> *cell
20:19:00 <ais523> elliott: it would do
20:19:01 <elliott> ais523: no decoys, nope
20:19:02 <elliott> I might try adding some
20:19:05 <ais523> confuses tripwires
20:19:20 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*1(<)*3(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*8(>[-([+{[(+)*8[-].[.+]]}])%3])*20
20:19:38 <Gregor> Sgeo: Just some small amount of random flag nonsense before moving on.
20:19:39 <EgoBot> Score for Deewiant_sm3: 52.6
20:20:08 <quintopia> Deewiant: what is an sm3?
20:20:18 <Deewiant> http://en.wikipedia.org/wiki/VR_Class_Sm3
20:20:47 * Sgeo trains a train on Deewiant
20:21:11 <ais523> how the hell does superfast_jellyfish beat anything but defence?
20:21:16 <quintopia> Deewiant: has bf joust caused you to have to start studying trains?
20:21:18 <elliott> Deewiant: Have you ever made a program that isn't like a train? :P
20:21:19 <Gregor> There, it doesn't autoplay.
20:21:21 <Gregor> Now STFU.
20:21:25 <Deewiant> elliott: sloth
20:21:27 <elliott> Gregor: Is it floated right?
20:21:33 <Deewiant> quintopia: Not really
20:21:33 <Gregor> elliott: FUCK
20:21:34 <Gregor> elliott: YOU
20:21:42 <elliott> Gregor: The Authority (ais523) demanded that it be floated right.
20:21:44 <elliott> He might exile you.
20:21:45 <ais523> Gregor: he's underage, that's illegal
20:21:47 <ais523> elliott: no I haven't
20:21:51 <elliott> ais523: Close enough.
20:21:56 <elliott> You agreed it would create happiness!
20:22:04 <ais523> I don't mind it being floated right
20:22:10 <ais523> but I don't mind it at its current location either
20:22:12 <Gregor> I can't float it right, then it could cause line-breaking in the trace for no good reason (screen width is fine, just line breaking for idiocy)
20:22:15 <Sgeo> ....Paul the Hermit day
20:22:15 <Sgeo> WTF
20:22:32 <elliott> Gregor: ...how would it cause line-breaking?
20:22:46 <elliott> Even /my/ screen can fit the trace and video on maximum tape length.
20:22:46 <Gregor> elliott: By the trace plus the animation being wider than the screen durpadurp
20:22:50 <ais523> elliott: on a narrow screen, the trace and video wouldn't both fit
20:22:50 <elliott> <elliott> Even /my/ screen can fit the trace and video on maximum tape length.
20:22:53 <elliott> And it's only 1440 wide :P
20:22:57 <ais523> that's pretty wide
20:23:01 <elliott> ais523: Not really.
20:23:07 <elliott> Nobody uses 1024x768 any more (except maybe you)
20:23:07 <Gregor> Yes. Really.
20:23:08 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*3(<)*4(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*10(>[-([+{[(+)*8[-].[.+]]}])%3])*18
20:23:23 <ais523> mine's 1366
20:23:29 <EgoBot> Score for Deewiant_sm3: 48.1
20:23:33 <Gregor> elliott: Why looka this, it's almost as if I'm not making the change because fuck you!
20:23:35 <ais523> my last laptop was 1280 wide
20:23:50 <Deewiant> !bfjoust sm3 -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*2(<)*4(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*9(>[-([+{[(+)*8[-].[.+]]}])%3])*19
20:23:55 <quintopia> Deewiant: that article makes me wish USA had an extensive and reasonably priced rail network. amtrak is comfortable, but expensive compared to flying, and doesn't go many places...
20:23:56 <elliott> Gregor: Can you at least wrap the whole player in a div so that I can hide it?
20:23:59 <EgoBot> Score for Deewiant_sm3: 52.8
20:24:08 <Gregor> elliott: That is so completely absurd, but fine.
20:24:20 <Gregor> Honestly, is a black, non-animating box that distracting to you?
20:24:40 <ais523> his complaint is that it pushes things down the screen
20:24:44 <elliott> Gregor: No. Having to scroll down to see more than three lines of trace is.
20:24:55 <elliott> Which is exactly what I do constantly when developing warriors.
20:24:57 <Gregor> The top however-many lines of trace are useless anyway X_X
20:25:03 <elliott> No, they're not.
20:25:07 <elliott> I can see how decoys are built, etc.
20:25:17 <Gregor> Oh look, I moved over and started to build a decoy
20:25:21 <Gregor> Woohoo
20:25:34 <ais523> elliott still lives in a world of small decoys, I think
20:25:40 <elliott> Gregor: Um, yes, because I tweak warriors.
20:25:41 <elliott> ais523: Whaddya mean
20:25:55 <elliott> Gregor: I like to see what decoys other warriors build so that I can figure out how to effectively counter them.
20:25:58 <quintopia> Gregor: i don't think he'll be happy until the animation opens in a separate window when you click a button
20:26:00 <elliott> And it lets me see if they're setting up a tripwire, etc.
20:26:02 <elliott> So yes, the earlygame is useful.
20:26:02 <Gregor> I tweak warriors too. With the animation playing. And I don't go "OMG THE ANIMATION NOOOOOOSE"
20:26:07 <elliott> quintopia: No, I just want him to wrap it in a div.
20:26:17 <Deewiant> !bfjoust pendolino -(>)*6->+>(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:26:20 <Gregor> Bug closed. E_DONTCARE
20:26:20 <elliott> That's exactly one (1) unit of effort.
20:26:26 <elliott> X_X
20:26:28 <ais523> most programs are still building their second or third decoys by the time it scrolls off the screen
20:26:35 <Gregor> Ohyeah, was going to wrap it in a div :P
20:26:40 <EgoBot> Score for Deewiant_pendolino: 40.8
20:26:42 <elliott> ais523: Not fast ones.
20:27:01 <ais523> indeed, but they're rare nowadays, that's what I meant by "still lives in a world of small decoys"
20:27:12 <Deewiant> !bfjoust pendolino -(>)*6->+>(<)*2(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:27:12 <ais523> anyway, what do you think of http://esolangs.org/wiki/BF_Joust_strategies?
20:27:20 <EgoBot> Score for Deewiant_pendolino: 50.6
20:27:42 <Deewiant> !bfjoust pendolino -(>)*6->+<(+)*12<(-)*12(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:27:49 <EgoBot> Score for Deewiant_pendolino: 50.8
20:27:50 <elliott> ais523: Someone did my link suggestion, so I'm happy. But "Trace and animation" looks kinda ugly with its sorta lopsided alignment.
20:27:58 <elliott> ais523: I think I'll create a template that expands to:
20:28:09 <ais523> that's templated already, just change the template
20:28:10 <Deewiant> !bfjoust pendolino -(>)*6->+<(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*7(>[-([+{[(+)*8[-].[.+]]}])%3])*21
20:28:12 <elliott> ''''program header thing''' &bull; '''trace link'''
20:28:12 <Sgeo> ais523, oooooo
20:28:13 <elliott> :Description
20:28:23 <EgoBot> Score for Deewiant_pendolino: 51.7
20:28:24 <ais523> what do you think of the /content/ of the page?
20:28:25 <elliott> (OK, I should use a definition list, but I don't know the wiki syntax for that.)
20:28:29 <elliott> ais523: It's good :P
20:28:30 <quintopia> ais523: did you figure out why superfast_jellyfish beats the speedies yet? :P
20:28:34 <ais523> ; for <dt>, : for <dd>
20:28:41 <elliott> ais523: But using {} for examples is problematic.
20:28:48 <ais523> quintopia: no; I figured out why defend9.5 almost ties, though
20:28:58 <ais523> elliott: ?
20:29:00 <elliott> ais523: May I suggest «...»?
20:29:04 <elliott> ais523: Rather than {tripwire} and the like.
20:29:05 <ais523> elliott: I didn't write that bit
20:29:12 <ais523> the page isn't all mine, it's a collaborative effort
20:29:15 <elliott> ais523: I'm not psychic; you asked what I thought of the page. :p
20:29:26 <ais523> feel free to tinker with the page
20:29:32 <elliott> ais523: you're wrong, it isn't templated
20:29:32 <Gregor> There, the animation is now in a div, o annoying one.
20:29:33 <ais523> and yes, the {} confused me too
20:29:35 <elliott> [http://codu.org/eso/bfjoust/in_egobot.hg/index.cgi/raw-file/89da92dc7a23/Gregor_sucralose_philip.bfjoust '''Gregor_sucralose_philip''']
20:29:35 <elliott> {{BFJAnim|f8c9e50f46cb1efa5723f43717ba87be36f7cd99}}
20:29:35 <elliott> Description: Same as high_fructose_corn_philip, but after finding four decoys, switches to a simple two-cycle clear strategy to avoid spending too many cycles on closing loops.
20:29:42 <ais523> elliott: BFJAnim is templated
20:29:46 <ais523> and that's the bit you were talking about
20:29:46 <elliott> ais523: that's not what i want to change
20:29:51 <ais523> ah
20:29:54 <elliott> <elliott> ''''program header thing''' &bull; '''trace link'''
20:29:55 <elliott> <elliott> :Description
20:30:00 <Deewiant> !bfjoust sm3 <
20:30:01 <ais523> yep, that could do with templating too
20:30:07 <elliott> I'll do it in a minute.
20:30:08 <EgoBot> Score for Deewiant_sm3: 0.0
20:30:10 <Deewiant> !bfjoust pendolino -(>)*6->+>(>[-([+{[(+)*8[-].[.+]]}])%3])*2(<)*4(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*9(>[-([+{[(+)*8[-].[.+]]}])%3])*19
20:30:18 <EgoBot> Score for Deewiant_pendolino: 54.9
20:30:22 <ais523> Deewiant: did you just delete the program at the top of the leaderboard?
20:30:27 <ais523> and if so, why?
20:30:32 <Deewiant> Yes, and replaced it with pendolino
20:30:39 <Gregor> Stop pushing my wonderful furry_furry_bondage_girls down :(
20:30:40 <ais523> ah, are they basically the same?
20:30:49 <Deewiant> So close that I'd rather not have them both
20:30:50 <ais523> I liked them both being there, they pushed defend9.75 above wireless
20:30:52 <Gregor> ais523: I think he uses the name sm3 for testing, then swaps that into the "real" name.
20:31:05 <Deewiant> No, actually I was going for something a bit different with sm3
20:31:07 <elliott> what's the mediawiki for "if this parameter exists, ..., else nothing"?
20:31:12 <ais523> elliott: ouch
20:31:16 <elliott> ais523: X_X
20:31:22 <elliott> don't tell me that's hard
20:31:27 <ais523> the "correct" way to do it is {{#if:}} but that's an extension that isn't installed on Esolang
20:31:32 <ais523> and there's a crazy hacky way to do it too
20:31:32 <elliott> Right.
20:31:34 <elliott> What's the hacky way.
20:31:34 <Gregor> !bfjoust saccharin_philip <
20:31:35 <ais523> let me dig it up
20:31:42 <EgoBot> Score for Gregor_saccharin_philip: 0.0
20:32:07 <Sgeo> It ocnly now occurs to me that programs can safely try to clear to the end, since if they correctly clear before then, battle's over
20:32:09 <Gregor> How evil would it be to establish a warrior limit on the hill >: )
20:32:17 <Gregor> That is, per-player warrior limit.
20:32:21 <ais523> EgoBot: I'm trying to find it
20:32:25 <elliott> Gregor: Too evil.
20:32:30 <elliott> <EgoBot> ais523: OK
20:32:34 <Gregor> elliott: Esp. since it'd be so easy to work around.
20:32:35 <ais523> *elliott:
20:32:47 <Gregor> EgoBot needs love too.
20:32:58 <elliott> in "trace and animation", what is the other program? simple?
20:33:04 <Sgeo> EgoBot, will you be my Valentine?
20:33:06 <Gregor> elliott: If unspecified.
20:33:22 <Gregor> elliott: You can also specify two unnamed arguments and then the second will be, err, the second.
20:33:45 <elliott> ais523: in "{{foo|x|abc=def|y}}", is y {{{2}}} or {{{3}}}?
20:33:47 <elliott> or is that invalid?
20:33:50 <quintopia> Gregor: i was considering the idea of a hill where people couldn't play their warriors against their other warriors. i rejected it because people seem to like to come up with ideas that can beat their old ideas (i know i do)
20:33:50 <ais523> elliott: http://en.wikipedia.org/wiki/Template:Qif
20:33:56 <ais523> elliott: I think it's {{{2}}}, but I'm not sure
20:34:07 <Gregor> quintopia: How would you accomplish such a concept?
20:34:12 <ais523> it's certainly confusing, at least, and would better be written with named params everywhere or all the anonymous templates first
20:34:24 <elliott> ais523: but the named one is the optional one
20:34:30 <elliott> {{BFJProg|nescience_creep|
20:34:35 <elliott> Abcdefdescription.}}
20:34:35 <elliott> vs.
20:34:41 <elliott> {{BFJProg|nescience_creep|anim=sdfg09ug98t98rtue|
20:34:42 <elliott> Abcdefdescription.}}
20:34:42 <ais523> ah, I see
20:34:48 <quintopia> Gregor: just assume that people always submit their warriors using the same nick. but that was all hypothetical and rejected.
20:34:49 <ais523> btw, it's best for long params like the description to be named
20:34:56 <ais523> because otherwise the template breaks if they have an = sign in anywhere
20:35:09 <elliott> ais523: heh, OK, I'll name it desc then.
20:35:33 <ais523> hmm, I wonder if {{qif}} counts as public-domain?
20:35:42 <ais523> if not, we might not be allowed to use it on Esolang, which would be kind-of ridiculous
20:35:49 <elliott> ais523: tell me how it works
20:35:52 <elliott> and I'll reimplement it
20:36:21 <quintopia> obtw, i never asked before but i've always wondered. why limes?
20:36:28 <Sgeo> "A few programs may benefit from not setting up decoys. This normally happens with bad programs, where setting up decoys may not aid their strategy because it was a bad idea in the first place, whereas not setting up decoys can trick tripwire avoiders off the end of the tape;"
20:36:38 <Sgeo> So, I should never write a program that makes a decoy. Got it.
20:36:42 <elliott> ais523: wtf@qif
20:36:46 <ais523> umm, that would involve knowing how it works
20:36:54 <elliott> ais523: I'm going to consider this public domain, because insanity isn't copyrightable
20:37:06 <ais523> quintopia: I think it's just a picture Graue liked
20:37:11 <ais523> there isn't any inherent reason as far as I know
20:37:27 <elliott> has anyone ever NOT asked why the logo is limes :)
20:37:42 <ais523> I never asked
20:37:43 <quintopia> ais523: so. do you think he'd let us change it to a complete and cartoonified version of careless philip the turtle? and make him our mascot? :P
20:37:48 <elliott> so, err, qif is worked like
20:37:55 <elliott> {{qif|param|ifitsthere|ifitsnot}}
20:37:56 <elliott> right?
20:37:57 <ais523> I think I know how qif works
20:38:11 <ais523> no, it's {{qif|test=maybenull|then=one|else=two}}
20:38:16 <elliott> ah
20:38:18 <ais523> the param names are needed, the template doesn't work otherwise
20:38:26 <elliott> can I omit else?
20:38:35 <Sgeo> According to the docs, yes
20:38:36 <ais523> no, but you can omit then if you like
20:38:43 <elliott> haha
20:38:46 <ais523> ah no
20:38:50 <ais523> they changed it to the sane way round
20:38:53 <elliott> ais523: argh, "then= foo" strips the space?
20:39:01 <ais523> I think so
20:39:08 <elliott> ais523: what to do? &nbsp;? :P
20:39:11 <elliott> &#32;?
20:39:12 <ais523> use nbsp if you can, or perhaps <nowiki></nowiki> before the space
20:39:21 <ais523> or nowiki the space itself
20:39:26 <elliott> ah, thanks
20:39:36 <elliott> yep, <nowiki></nowiki> works
20:39:49 <elliott> Gregor: What does {{BFJAnim|abc|}} do?
20:39:52 <elliott> With the empty last parameter?
20:40:00 <elliott> PLEASE say "works perfectly as if you didn't have an empty last parameter".
20:40:16 <Gregor> elliott: I have no idea, depends on how MediaWiki handles empty parameters.
20:40:22 <elliott> ais523: {{{2|x}}} where 2 is empty =?
20:40:39 <ais523> that gives x
20:40:43 <ais523> whereas it gives the value of 2 if 2 isn't empty
20:40:47 <elliott> Gregor: hmm, what does the second parameter to BFJAnim do?
20:41:05 <elliott> I need to know what to call it :P
20:41:06 <Gregor> elliott: It is the second program to be used.
20:41:07 <ais523> elliott: second param is the program to play against, "simple" or a hash in storage
20:41:09 <elliott> Oh, right.
20:41:13 <Gregor> elliott: "r"
20:41:15 <elliott> {{anim2}} then.
20:41:19 <elliott> Gregor: No, in BFJProg.
20:41:21 <Gregor> Oh
20:41:25 <ais523> I suggest you take a call to BFJAnim as input, btw
20:41:27 <Gregor> opponent?
20:41:32 <ais523> rather than its params individually
20:41:32 <elliott> ais523: oh, probably the best idea
20:41:36 <ais523> because it has several optional params
20:41:38 <elliott> I'll do that
20:41:52 * elliott converts the page to use the new template
20:42:14 <ais523> Gregor: decoytuner was never top of the leaderboard, was it? I removed it from the page because it was awful and I can't believe it was #1
20:42:20 <ais523> decoytuner2 might have been, though
20:42:45 * elliott ponders whether to have |code= as a param to BFJProg
20:42:51 <elliott> or whether that should be part of the descriptoin
20:42:53 <elliott> *descrption
20:42:54 <elliott> *descrption
20:42:57 <elliott> *description
20:42:59 <elliott> I'll make it a param
20:43:00 <elliott> for flexibility
20:43:03 <Gregor> ais523: I only added things that were #1.
20:43:06 <Sgeo> Gregor, the animation thing is awesome
20:43:08 <Sgeo> <3
20:43:13 <Gregor> Sgeo: Why thank you :P
20:43:15 <ais523> the thing is, some have code because they're short, some have links because they're long
20:43:26 <ais523> Gregor: I think you might have made a mistake
20:43:31 <elliott> ais523: right; well, I will have |code= and |codeurl= then
20:43:35 <ais523> decoytuner, the original, never went anywhere
20:43:42 <Gregor> ais523: Impossible!: P
20:43:51 <elliott> maybe |url= instead of |codeurl=
20:44:29 <Deewiant> !bfjoust pendolino -(>)*7->+(>[-([+{[(+)*8[-].[.+]]}])%3])*2(<)*4(+)*23<(-)*23(<(-)*30<(+)*30)*2(>)*10(>[-([+{[(+)*8[-].[.+]]}])%3])*19
20:45:16 <EgoBot> Score for Deewiant_pendolino: 54.7
20:45:33 <elliott> OK, nobody edit [[BF Joust strategies]]
20:45:40 <ais523> to avoid conflicts?
20:45:50 <elliott> indeed
20:45:56 <elliott> ouch, the code of these programs aren't in <nowiki>
20:46:04 <elliott> but have <s in them
20:46:30 <elliott> {{BFJProg|nescience_creep|code=<nowiki> >+>->+>->+>->+(>-++-(.)*132[+]++>-++-(.)*132[-]--)*15 </nowiki>
20:46:31 <elliott> |desc= A very slow rush that leaves a trail of small decoys, using a simple clear.}}
20:46:32 <elliott> looks nice enough to use
20:46:42 <ais523> they're in verbatim blocks, that's what the leading space does
20:46:43 * elliott puts the code on a separate line
20:46:50 <elliott> ais523: ah
20:46:56 <elliott> ais523: need nowiki with the template, though
20:47:00 <elliott> since it handles the display itself
20:47:00 <ais523> you should enclose them in <pre> if you're using the template, I think
20:47:05 <ais523> it's a different sort of verbatim block
20:47:09 <ais523> and has an implicit nowiki
20:47:18 <elliott> ais523: the template might want to display it outside of a pre, though!
20:47:19 <ais523> (note: this is not what <pre> actually means in HTML)
20:47:23 <elliott> (it uses <tt> right now)
20:47:25 <ais523> and hmm, OK
20:47:34 <elliott> woo, the template doesn't work
20:47:38 <elliott> probably because qif doesn't work
20:48:03 <elliott> hmm
20:48:04 <elliott> how odd
20:48:13 <elliott> the invocation looks OK to me
20:48:51 <ais523> {{{anim}}} should probably be {{{anim|}}} in case an animation isn't available or can't easily be produced
20:49:05 <elliott> ;{{qif|test={{{url}}}|then=[{{{url}}} {{{1}}}]|else={{{1}}}}}{{qif|test={{{anim}}}|then=<nowiki></nowiki> &bull; {{{anim}}}}}{{qif|test={{{code}}}|then=<nowiki></nowiki> &bull; <tt>{{{code}}}</tt>}}
20:49:09 <elliott> ais523: ah
20:49:13 <elliott> ais523: so I need to do that even with qif?
20:49:14 <elliott> nice :P
20:49:15 <ais523> (I was going to add ones for some of the early winners, but it turns out that they lose to even simple programs; I can't think back far enough to create something they beat and still looks plausible)
20:49:29 <ais523> elliott: test param of qif definitely needs an empty param default
20:49:39 <ais523> if anim doesn't have a value, {{{anim}}} is literally {{{anim}}}
20:49:51 <ais523> and {{{anim|}}} is the null string, which is what you probably wanted
20:50:02 <elliott> heh
20:50:19 * elliott wtfs
20:50:23 <elliott> "Code:" in a qif parameter ->
20:50:24 <elliott> Code
20:50:25 <elliott> :foo
20:50:27 <elliott> behaviour
20:50:32 * elliott nowikis
20:50:44 <ais523> elliott: : doesn't just trigger at the start of the line, that's correct
20:50:45 -!- Sgeo_ has joined.
20:50:49 -!- sebbu2 has joined.
20:50:49 <ais523> instead of going
20:50:49 -!- sebbu2 has quit (Changing host).
20:50:49 -!- sebbu2 has joined.
20:50:52 <ais523> ;title
20:50:54 <ais523> :definition
20:50:56 <ais523> you can also go
20:50:59 <ais523> ;title:definition
20:51:05 <elliott> heh
20:51:09 <ais523> that's probably a misfeature, but it's how it parses nonetheless
20:51:20 <ais523> now, there's a ; at the start of that line, so...
20:52:05 <Sgeo_> Why do I feel dizzy?
20:52:11 <Sgeo_> Not dizzy
20:52:13 <ais523> wow the philips have fallen down the leaderboard a lot
20:52:14 <elliott> yay, it's working now
20:52:14 <Sgeo_> Just.. weird
20:52:34 <ais523> what happened there? people engaging in anti-careless strategies?
20:52:48 <ais523> careless itself is halfway down, it was near the top yesterday
20:53:08 <Gregor> ais523: http://codu.org/eso/bfjoust/in_egobot/reports/report-2011-02-11-05-07.txt
20:53:37 <Gregor> ais523: Since most of my strategies have careless cores, anticareless has done well :P
20:53:41 <ais523> Gregor: that was a different version of decoytuner than the one you put on the wiki, I think
20:53:57 <Gregor> ais523: Ah; I just checked what was the top, then grabbed the latest thing with that name.
20:54:11 <elliott> yay, another bug in BFJProg
20:54:12 <ais523> yep, that version was a turtle, the one you put on the wiki wasn't
20:55:00 <elliott> I'm quite amazed that we have creative strategy names :P
20:55:28 <elliott> that's why we need people who aren't ais523 to do well :P
20:55:47 <elliott> argh, I see the problem
20:55:51 <elliott> {{BFJProg|foo
20:55:51 <ais523> I use creative names for my /strategies/ (lock, etc); just not for my /programs/
20:55:54 <elliott> | abc = def}}
20:56:02 <elliott> [{{{abc}}} {{{1}}}]
20:56:05 <elliott> {{{1}}} ends with a newline
20:56:08 <elliott> Is there a "strip" template? :P
20:56:19 <ais523> there's no way to remove newlines from template params
20:56:23 <ais523> you can do foo
20:56:25 <ais523> =param
20:56:30 <ais523> I think, not sure
20:56:36 <ais523> or maybe even foo= on a line and the param on the next
20:56:36 <elliott> ais523: eh?
20:56:37 <Gregor> I use creative names for my programs (ridiculous and scary is creative, right?), but don't use creative names for my strategies since i just steal.
20:56:45 <ais523> mediawiki template newline handling is insane
20:56:46 <elliott> Gregor: "turtle"
20:57:01 <ais523> well, it's called "careless clear" on the wiki after the first program to be successful with it
20:57:04 <Gregor> elliott: I accidentally reinvented careless with turtle :P
20:57:13 <elliott> turtle is a better name, though
20:57:18 <ais523> in fact, I think the terminology seems to be that "turtle" is the program type and "careless" the strategy it uses
20:57:24 <elliott> haha
20:57:27 <ais523> i.e. turtles are careless
20:58:06 -!- iamcal has joined.
20:58:29 <Sgeo_> !bfjoust dumb_turtle (>[(-)*128])*-1
20:58:41 <Sgeo_> Hmm, that's a really bad idea, come to think of it
20:58:54 <ais523> Sgeo_: you might want to put a + after the > so that the loop is actually entered
20:59:05 <ais523> and might also want to rule-of-nine it
20:59:14 <elliott> "An obvious modification of slowrush, possibly of an earlier version."
20:59:17 <Sgeo_> I don't want the loop to be entered until I'm on some value
20:59:22 <elliott> IIRC "reversi" = "reversed" = flipped polarity
20:59:24 <ais523> Sgeo_: ah, that makes sense
20:59:28 <elliott> of slowrush
20:59:38 <ais523> put > after the 128 then
20:59:39 <elliott> it might have been of an old version, but IIRC after stealing programs I used to tweak them
20:59:41 <Sgeo_> Ah, wait, even if I win, I'm going to rush off the edge, aren't I?
20:59:44 <ais523> so you don't end up oscillating on a 1-valued decoy
20:59:48 <ais523> umm, .> after the 128
20:59:55 <Sgeo_> I'm going to go home now
20:59:56 <ais523> you won't fall off the end if you win, ] takes a tick
21:00:06 -!- Sgeo has quit (*.net *.split).
21:00:06 -!- cal153 has quit (*.net *.split).
21:00:07 -!- sebbu has quit (*.net *.split).
21:00:07 -!- Mannerisky has quit (*.net *.split).
21:00:08 -!- sebbu2 has changed nick to sebbu.
21:00:12 <ais523> but if the first thing you hit isn't a flag, you'll go into an infinite loop
21:00:17 <Sgeo_> lolwhat at Sgeo
21:00:29 <Sgeo_> Hence the dumb
21:00:35 <ais523> `addquote <Sgeo_> I'm going to go home now <-- Sgeo has left this server (*.net *.split).
21:00:37 <ais523> such perfect timing
21:00:49 <ais523> if only it was the other nick, it'd have been even better
21:01:13 <Sgeo_> Bye
21:01:19 <ais523> bye
21:01:24 <elliott> "An alternating polarity offset clear that leaves no decoys and takes forever to reach the other flag. It's a wonder this thing ever won."
21:01:27 <elliott> :D
21:01:32 <elliott> ais523: bye?
21:01:36 <ais523> to sgeo
21:01:38 <elliott> ah
21:01:40 <ais523> *Sgeo
21:01:53 <HackEgo> 307) <Sgeo_> I'm going to go home now <-- Sgeo has left this server (*.net *.split).
21:02:11 <elliott> btw, I'm pretty sure that Patashu_rushpolarity is slowrush polarity'd
21:02:15 <elliott> or at least some rush polarity'd
21:02:16 <EgoBot> Score for Sgeo__dumb_turtle: 16.6
21:02:31 <ais523> elliott: it alternates polarity on adjacent tape elements, I think
21:02:35 <elliott> ah
21:02:36 <ais523> in order to avoid its polarity being detected
21:02:40 <elliott> is it still on the hill?
21:03:01 <ais523> yes, in fact it's sixth
21:03:27 <ais523> because it's a pain to beat when writing defense programs, so people normally don't bother
21:03:31 * elliott saves the template conversion so far
21:03:33 <ais523> and it does quite well against the prevalent strategies nowadays
21:03:35 <elliott> now to do the rest
21:04:05 <ais523> it is beatable though, e.g. defend9.75 beats it and space_elevator for the same reason
21:04:17 <elliott> what does 9.75 do?
21:04:31 <ais523> it's basically defend9 updated to modern strategies
21:04:42 <ais523> it detects enemy polarity and cycle length
21:04:50 <ais523> then uses a lock algorithm matched to what the enemy does
21:05:08 <ais523> it also sets seven decoys, three of which are also used as tripwires, and switches to a five-cycle clear versus defence
21:05:29 <ais523> and has all sorts of tricks to beat other common strategies
21:05:39 <elliott> nice
21:06:49 <quintopia> guys. i just realized the current top 3 are in a cycle with each other. pendolino>sm3>space_elevator>pendolino
21:07:05 -!- augur has quit (Read error: Connection reset by peer).
21:07:17 -!- Sgeo_ has quit (Ping timeout: 255 seconds).
21:07:25 <elliott> quintopia: heh
21:07:30 -!- augur has joined.
21:07:31 <ais523> and the fourth beats all three of them
21:08:12 <ais523> 9.75 is a limited strategy in a way; there are countermeasures against it you take that it can't be modified to beat without changing the whole concept or crippling it heavily against most other programs
21:08:41 -!- rodgort has quit (Quit: Coyote finally caught me).
21:08:57 -!- rodgort has joined.
21:09:47 <elliott> there's no description for space_elevator
21:09:48 * elliott adds one
21:09:55 <elliott> | desc = Utterly insane.}}
21:10:12 <ais523> really, the desc should be by the author and explain exactly how it works
21:10:27 <ais523> I have a pretty good idea of how it works, as I've studied a whole bunch of methods for countering it, from the specific to the generic
21:10:39 <elliott> But until such a description is written, it stays like that :P
21:10:45 <elliott> ais523: have you studied ICA's? :P
21:11:09 <elliott> template conversion done
21:11:33 <ais523> I haven't studied ICA's yet, but they're the sort of thing I don't like to try to beat
21:11:41 <elliott> ais523: *it's
21:11:47 <elliott> ais523: and why not?
21:11:51 <elliott> it's not a specific program targeter
21:11:53 <elliott> it's a confuser
21:12:10 <elliott> perhaps a defence program should try and get multiple readings of cycle length in different situations?
21:12:16 <ais523> it's because I'm bad at beating confusers
21:12:23 <ais523> defence programs lose to rush if they do that
21:13:03 <quintopia> gregor: i've got a fixed point algorithm for scoring. it's so simple you won't believe it
21:13:13 <elliott> quintopia: Is it mine? :P
21:13:15 <ais523> I like Gregor's, and how it works with the hill
21:13:23 <elliott> ais523: Gregor wants it to be fixed-point
21:13:25 <elliott> ais523: it would work the same way
21:13:27 <elliott> ais523: except be recursive
21:13:29 <quintopia> elliott: yes, except you divide by the number of programs
21:13:34 <elliott> quintopia: That's what I said ...
21:13:38 <quintopia> *don't divide
21:13:40 <elliott> Oh.
21:13:43 <quintopia> you just leave it
21:13:44 <elliott> quintopia: Surely that saturates?!
21:13:45 <quintopia> and it works
21:13:48 <elliott> O KAY
21:13:51 <elliott> And differs from the points?
21:13:55 <quintopia> it does
21:13:57 <ais523> ooh, looks like I have both highest negative-points and lowest positive-points programs on the hill
21:14:20 <elliott> Gregor: S(p) := sum S(q) for all q that p beats, where S(x) is initially taken as P(p).
21:14:25 <elliott> Gregor: quintopia says this is fixed-point, as unlikely as that sounds.
21:14:57 <quintopia> according to this algorithm 9.75 is easily the best program in the top 5 (which is exactly what we'd expect)
21:15:06 <elliott> sounds good
21:15:45 <elliott> quintopia: got a closed-form way to calculate it? :P
21:16:24 <quintopia> elliott: actually, it's not fixed point. i was mistaken XD. it cycles because the top 3 are cyclical
21:16:30 <quintopia> i don't know how to resolve that
21:16:42 <elliott> quintopia: If you visit the same program twice, the second time, return score 0.
21:16:56 <elliott> quintopia: Try with that
21:17:14 <elliott> quintopia: Basically for any call of S(p) keep a bag for that call and its sub-calls :P
21:17:25 <elliott> And if you ever recurse on a program that you've already seen, return 0.
21:17:37 <quintopia> elliott: that requires work. right now i'm just multiplying a matrix by a vector repeatedly.
21:17:47 <quintopia> you want me to write an actual *algorithm*
21:17:56 <elliott> quintopia: A dumb recursive algorithm is not much of an algorithm :P
21:18:19 -!- augur_ has joined.
21:18:21 <quintopia> it's not dumb if you're making it dynamic programming
21:18:31 <elliott> quintopia: Don't then.
21:18:48 -!- augur has quit (Read error: Connection reset by peer).
21:19:15 <ais523> I don't really like being at #1, it makes people target me
21:19:16 <elliott> quintopia: Or, even better idea: If you get called upon to return the score of a program you've already seen, return its score per the last iteration.
21:19:22 <ais523> but I suppose it's nice to be there occasionally
21:20:33 <pikhq_> Grawr.
21:20:35 -!- pikhq_ has changed nick to pikhq.
21:20:53 <ais523> hmm, did pikhq ever joust? I can't remember
21:21:00 <pikhq> Nope!
21:21:19 <elliott> write a jouster in PEBBLE :P
21:21:27 <elliott> hideous inefficiency awaits
21:21:35 <pikhq> That'd involve writing a new backend.
21:21:47 <elliott> ORLY?
21:21:50 <elliott> Valid BF = valid BF Joust.
21:21:55 <pikhq> Oh, never mind.
21:22:03 <pikhq> elliott: Funny thing is, PEBBLE is one of the more efficient high-level languages targetting Brainfuck.
21:22:06 <elliott> pikhq: ()* and ()% are just conveniences :P
21:22:24 <pikhq> ()* and ()% could probably be added with ease.
21:22:36 <ais523> OTOH, BF Joust doesn't follow normal programming physics
21:22:58 <ais523> assigning to a variable then checking several times that its value is still the same would normally be considered insane
21:23:04 <ais523> but it's par for the course when jousting
21:23:13 <pikhq> Probably helps a lot that PEBBLE is higher-level than Brainfuck but still *insanely* low-leve.
21:23:26 <pikhq> Low-level, rather.
21:23:38 <pikhq> It doesn't even have normal conditionals!
21:23:58 <elliott> pikhq: Yeah, but BF Joust makes assumptions like ais523 said :P
21:24:15 <pikhq> elliott: Maybe a Joust-specific set of macros would be handy.
21:24:22 <ais523> 1cnis would make a good BF Joust preprocessor, IMO
21:24:24 <elliott> ais523: also, sure it makes sense, some people actually think that writing code like that is _sane_
21:24:31 <elliott> ais523: see: people who think shared memory is a good idea (crazy people)
21:24:32 <quintopia> elliott: how about iterating until it sees a scoreboard its seen before and then averaging the entire cycle?
21:24:39 <pikhq> Thing is, PEBBLE is really little more than a macro system for Brainfuck.
21:24:44 <elliott> shared memory concurrency, that is
21:24:48 <elliott> quintopia: if that works, then sure
21:24:50 <elliott> quintopia: try it
21:24:52 <ais523> macro system would be nice too
21:24:57 <ais523> e.g. defend9.25 has two almost identical halves
21:24:58 <pikhq> ... With optional variable support.
21:25:08 <ais523> and one bug I came across was changing things in one half but not the other
21:25:15 <ais523> double reverse tripwire programs are like that
21:25:24 <ais523> (and using one reverse tripwire tends not to work well IME)
21:25:32 <elliott> quintopia: but will that work if only the top four are cyclic? it might be thousands of scoreboards before you see the exact same scoreboard
21:25:39 <elliott> because of the butterfly effect on the programs below
21:25:56 <elliott> maybe even millions
21:26:04 <elliott> ais523: well, ()% is already pretty powerful
21:26:08 <quintopia> elliott: it would be a very unusual scoreboard with a cycle that long
21:26:17 <elliott> ais523: and a macro language would probably end up insane, vs. using a program to generate
21:26:20 <ais523> elliott: indeed
21:26:24 <elliott> quintopia: but only the four top cycles
21:26:26 <elliott> quintopia: the rest don't
21:26:32 <elliott> so while the first four could go around in a tight cycle
21:26:36 <elliott> the programs below could change constantly
21:26:37 <elliott> because of those changes
21:26:44 <quintopia> no i don't think so
21:26:59 <quintopia> i think everything that isn't actually in a cycle will settle down
21:27:29 <quintopia> the cyclical ones just trade fractions of points with one another after that
21:27:51 <elliott> quintopia: yeah, but the other programs' scores are based on the top four
21:27:57 <elliott> so when they trade, the ones below trade too
21:28:03 <elliott> and possibly in a way that takes longer to cycle
21:30:23 <Gregor> SodaStream's Pete's Choice and Diet Pete's Choice may be better than Dr. Pepper and Diet Dr. Pepper, respectively.
21:30:49 <elliott> Really?
21:31:01 <elliott> Gregor: *Dr, btw
21:31:01 <elliott> no .
21:31:06 <elliott> Gregor: Oh wait, over in the US of A your soft drinks are all HFCS.
21:31:12 <elliott> Try sugar Dr Pepper, I hear it's a rarity there.
21:31:16 <quintopia> Gregor: have you tried making a Coke-alike?
21:31:30 <variable> elliott: tis called silly marketing
21:31:30 <Gregor> quintopia: Not myself, Cola is really complicated.
21:31:32 <quintopia> Gregor: i mean this recipe: http://news.yahoo.com/s/yblog_thelookout/20110215/ts_yblog_thelookout/did-nprs-this-american-life-discover-cokes-secret-formulahttp://news.yahoo.com/s/yblog_thelookout/20110215/ts_yblog_thelookout/did-nprs-this-american-life-discover-cokes-secret-formula
21:31:33 <elliott> Yeah.
21:31:35 <elliott> I was gonna say.
21:31:38 <elliott> Cola is fucking complicated :P
21:31:41 <quintopia> whoops
21:31:43 <quintopia> doublelink
21:31:43 <elliott> Even OpenCola looks like a pain.
21:31:46 <quintopia> sry
21:31:51 <Gregor> quintopia: In spite of the "article", that's not new news :P
21:31:54 <elliott> quintopia: Did they actually reveal the recipe or just the ingredients?
21:31:59 <elliott> The ingredients are irrelevant.
21:32:13 <elliott> http://www.colawp.com/colas/400/cola467_recipe.html = OpenCola
21:32:16 <quintopia> Gregor: it's not news, it's just a place on the internet you can find a cokealike recipe
21:32:16 <variable> elliott: Article not found or expired on Yahoo! News
21:32:22 <elliott> I want to try and make OpenCola sometime
21:32:24 <elliott> variable: not my link...
21:32:26 <elliott> quintopia's
21:32:29 <elliott> quintopia: http://www.colawp.com/colas/400/cola467_recipe.html
21:32:32 <elliott> OpenCola ftw
21:32:33 <Gregor> There are lots of colas *shrugs*
21:32:35 <elliott> ( http://en.wikipedia.org/wiki/OpenCola_(drink) )
21:32:47 <quintopia> elliott: don't click the link. copy and paste the first half of the link
21:32:54 <elliott> ...
21:33:06 <elliott> BUY AN IRC CLIENT YOU GUYS
21:33:11 <elliott> One that distinguishes names :P
21:33:23 <quintopia> http://news.yahoo.com/s/yblog_thelookout/20110215/ts_yblog_thelookout/did-nprs-this-american-life-discover-cokes-secret-formula
21:33:26 <quintopia> there
21:33:28 <quintopia> that's it not doubled
21:33:29 <quintopia> geez
21:33:44 -!- pikhq_ has joined.
21:33:45 <elliott> quintopia: ALSO: Making Coca-Cola is illegal.
21:33:52 <elliott> 'Cuz good fucking luck buying coca leaves, dude.
21:33:56 -!- pikhq has quit (Ping timeout: 264 seconds).
21:33:57 <quintopia> ...
21:34:05 <variable> elliott: how about I preface every statement in this channel with your name?
21:34:06 <quintopia> coca-cola uses cocaine free coca leaves
21:34:10 <elliott> quintopia: Yes.
21:34:11 <quintopia> and it's perfectly legal to
21:34:13 <elliott> quintopia: Which aren't sold to anyone else.
21:34:34 <quintopia> elliott: my client makes you and variable the same color. everyone else is different
21:34:35 <elliott> Which kinda puts a stopper on the idea.
21:35:03 <quintopia> are you sure it's impossible to acquire them?
21:35:12 <elliott> quintopia: Pretty damn sure.
21:35:33 <quintopia> how do they do that?
21:35:36 <quintopia> money?
21:35:37 <elliott> quintopia: This, quite understandably, riles some people up.
21:35:42 <elliott> quintopia: Um, yes, money.
21:35:47 <elliott> Coca-Cola is a friggin' huge corporation.
21:35:52 <elliott> [[Because cocaine is naturally present in coca leaves, today's Coca-Cola uses "spent," or treated, coca leaves, those that have been through a cocaine extraction process, to flavor the beverage. The coca leaves are imported from countries like Peru and Bolivia, and they are treated by chemical company Stepan, which then sells the de-cocainized residue to Coca-Cola.[3] Some contend that this process cannot extract all of the cocaine alkaloids at a
21:35:52 <elliott> molecular level, and so the drink still contains trace amounts of the stimulant.[1][4] The Coca-Cola Company currently refuses to comment on the continued presence of coca leaf in Coca-Cola.[5][6]]]
21:36:01 <elliott> tl;dr all they have to do is tell Stepan not to sell them to anyone else.
21:36:11 <Gregor> ais523: My client doesn't color people, and yet I never seem to get them confused.
21:36:13 <Gregor> *ba-dum ching*
21:36:28 <ais523> Gregor: that was a strange choice of nick to ping
21:36:29 <quintopia> elliott: bah. what's stopping a different company from doing that extraction process?
21:36:33 <elliott> quintopia: I'm pretty sure you need Special Permission(TM) to treat the coca leaves too, considering they, you know, have cocaine in them.
21:36:42 <variable> funny note about Coca-Cola - the branch is worth so much that if their factories burnt down they could borrow against their name to rebuild
21:36:50 <elliott> quintopia: (This is reddit-accuracy-level information, but the logic checks out.)
21:36:58 <elliott> quintopia: ((I literally read it on reddit :P))
21:37:35 <quintopia> elliott: why not buy spent coca leaves from drug cartels? then you haven't bought illegal coca leaves or done the illegal extraction yourself
21:37:46 <elliott> DEALING WITH DRUG CARTELS: Totally legal
21:37:55 <quintopia> is it not? :P
21:38:02 <elliott> I don't... think so X-D
21:38:03 <variable> elliott: considering that you didn't buy anything illegal...
21:38:11 <elliott> *so, unless you have tons of money, like everything.
21:38:11 <quintopia> ^
21:38:16 <variable> you would just be financing them
21:38:18 <elliott> Yeah, but are the drug cartels doing business legally?
21:38:29 <pikhq_> Deal with the Yakuza.
21:38:30 <elliott> I think if you knowingly did business with drug cartels who aren't doing business legally, that would be illegal.
21:38:34 <quintopia> it doesn't matter what they do with your money. that's them being illegal, not you
21:38:38 <pikhq_> They *are* in fact perfectly legitimate businessmen..
21:38:39 <pikhq_> :P
21:38:53 <elliott> ANYWAY, point is that an _individual_ is never gonna be able to legally make Coca-Cola because of this :P
21:39:01 <elliott> quintopia: I'm sure knowingly funding illegal activities is illegal, dude.
21:39:08 <pikhq_> elliott: In certain jurisdictions.
21:39:09 <elliott> Considering that you're actually BUYING the results of an illegal activity.
21:39:12 <elliott> Rather than it just being indirect.
21:39:27 <variable> elliott: its not illegal to pay for something in cash if you know the store owner won't pay taxes on it
21:40:01 <variable> (IANAL)
21:40:18 <elliott> variable: Drug cartel = people who don't pay taxes! :P
21:40:47 <variable> elliott: note the word _not_
21:40:56 -!- MigoMipo has quit (Read error: Connection reset by peer).
21:41:05 <elliott> variable: right
21:41:13 <elliott> variable: I'm saying that I'm not sure the situation is comparable :P
21:41:18 <variable> hehe
21:41:34 * variable has to go for a bit
21:42:18 <Gregor> I have a T-Rex is Lonely comic queued that is ... so bad. So, so bad. Good bad, that is. So bad.
21:43:06 <pikhq_> elliott: In many South American countries it is actually entirely legal to grow and consume coca leaves.
21:43:10 -!- pikhq_ has changed nick to pikhq.
21:43:12 <elliott> Right.
21:43:20 <elliott> So to make Coca-Cola, move to South America :P
21:43:37 <elliott> Gregor: Link :P
21:43:44 <Gregor> elliott: I shouldn't! It's queued!
21:43:50 <elliott> Gregor: WE WON'T TELL.
21:43:51 <quintopia> this discussion makes me want to illegally make some coca-cola
21:44:02 <Gregor> elliott: Maybe I'll PM if you convince me :P
21:44:02 <elliott> 07:17:17 <fungot> Phantom_Hoover: nah theres australia mmmmmmmmate!!!! causes hallucinations, and you didn't reply directly to you kz 2!
21:44:03 <elliott> Australia: Causes hallucinations, mmmmmmmmate!!!!
21:44:18 <elliott> Gregor: I would tell absolutely ZERO people, which is ONE LESS THAN ONE (but ONE MORE THAN NEGATIVE ONE).
21:45:23 <pikhq> Inexplicably, coca leaves in *most* countries are treated identically to cocaine.
21:47:11 <quintopia> gregor: i think only putting "champion" programs on the wiki makes it so that we'll never see any of ais523's awesome defenders there since they can't really be champions.
21:47:33 <ais523> quintopia: several of my defenders are there already
21:47:34 <Gregor> quintopia: #1s was just a baseline.
21:47:40 <ais523> defend13 was champion for a time, IIRC
21:47:42 <pikhq> Which seems a bit like banning rye bread because LSD can be synthesized from ergot.
21:48:00 <quintopia> ergotamine -> lysergic acid? really?
21:48:10 <quintopia> never paid much attention to that
21:48:13 <pikhq> quintopia: That is how it was originally synthesized, actually.
21:48:22 <quintopia> i probably knew that at one point
21:48:28 <quintopia> when i was studying it
21:48:36 <Deewiant> Rye bread might have relatively more legal uses than cocaine
21:48:44 <quintopia> pikhq: or banning hemp >.<
21:48:58 <elliott> Deewiant: *than coca leaves
21:49:03 <Deewiant> Right :-D
21:49:06 <elliott> Deewiant: Does making cola count?
21:49:09 <elliott> :p
21:49:19 <elliott> Deewiant: Ooh, but what about getting high?
21:49:20 <elliott> ...wait...
21:49:57 <quintopia> i wonder if it's possible to breed a coca plant that tastes the same but doesn't produce cocaine
21:50:29 <Deewiant> Coca-Cola might contain coca but I don't think other colas do
21:50:44 <elliott> quintopia: Wouldn't it be simpler to legalise cocaine :P
21:50:45 <pikhq> Deewiant: Before extracting cocaine from the leaves became possible/common, consuming the leaves straight was a very common and not-very-notable thing...
21:50:56 <quintopia> it would be false advertising if coca-cola did not contain coca!
21:51:06 <elliott> Yeah isn't there that stuff about all the KULCHURES complaining about people wanting to ban chewing coca leaves
21:51:20 <pikhq> In the quantities of cocaine one gets from there, it's really no different than caffeine.
21:51:49 <quintopia> we should try to do cocaine extraction from coca-cola...
21:52:20 <Deewiant> pikhq: So was slavery but we still prefer to not do it these days
21:52:26 <pikhq> Apparently there's been no documented addictions to coca leaves.
21:52:33 <elliott> "In the United States, Stepan Company is the only manufacturing plant authorized by the Federal Government to import and process the coca plant"
21:52:35 <elliott> TOLD YOU
21:52:54 <elliott> US: Bullshit, or total bullshit?? You decide
21:52:55 <pikhq> Deewiant: Sure we do; we just call it different things.
21:52:59 <pikhq> Like "minimum wage".
21:53:15 <elliott> Minimum wage is ... slavery? :wat:
21:53:16 <quintopia> and any other company that would try would be stopped, of course, because coca-cola would pay to have it stopped
21:53:24 <elliott> Sounds like the kind of thing a libertarian crackpot would say :P
21:53:26 <elliott> quintopia: "Some contend that this process cannot extract all of the cocaine alkaloids at a molecular level, and so the drink still contains trace amounts of the stimulant.[1][4] The Coca-Cola Company currently refuses to comment on the continued presence of coca leaf in Coca-Cola.[5][6]"
21:53:31 <pikhq> elliott: Minimum wage in the US is not a living wage.
21:53:35 <elliott> quintopia: Depending on how crackpotty Mr. Some is, you could possibly do it.
21:53:35 <Deewiant> pikhq: We don't have that in this country :-P
21:53:36 <elliott> pikhq: Right.
21:53:52 <elliott> Deewiant: Ha ha, Finland, you crazy.
21:54:21 <pikhq> Deewiant: Yeah, you just have strong unions.
21:54:39 <pikhq> Deewiant: Which gets the same effect and makes the "small government" wackos' heads explode.
21:54:47 <quintopia> it is in fact impossible to set minimum wage to a living wage as all increases in minimum wage result in increased cost of living
21:54:54 <quintopia> in this country i mean
21:55:16 <quintopia> with changes in other laws it might could be done
21:55:19 <pikhq> quintopia: That suggests some major economic brokenness.
21:55:30 <quintopia> pikhq: it is broken. you didn't know this?
21:56:03 <pikhq> quintopia: Well, I *am* actually aware that the US economy is really broken by now...
21:56:09 <elliott> pikhq: Unions don't make small government wackos' head explode.
21:56:15 <elliott> pikhq: How can something they've outlawed confuse them? :)
21:56:21 <pikhq> elliott: Unions *instead of* a government, however?
21:56:52 <elliott> pikhq: Are you proposing some kind of unionocracy? :P
21:56:55 <pikhq> elliott: "The government is not involved but there's unions... MY HEAD ASPLODE!"
21:57:08 <elliott> As bad ideas go, unionocracy sounds pretty bad.
21:57:12 <pikhq> elliott: I believe the proper term for that is "democracy", actually. :P
21:57:28 <elliott> Unless I'm mistaken, democracy means one thing, and what it means is not what unionocracy would mean were it in the dictionary.
21:57:55 -!- ais523 has quit (Remote host closed the connection).
21:58:11 -!- ais523 has joined.
22:00:33 <ais523> !bfjoust vibration_fool_faster >>>++++<----<++++<(-)*127.(-+)*50000
22:00:57 <EgoBot> Score for ais523_vibration_fool_faster: 23.8
22:01:04 <ais523> not bad for a winner from ages ago
22:01:10 <quintopia> nice job
22:01:16 -!- augur_ has quit (Ping timeout: 255 seconds).
22:01:18 <quintopia> you knocked our last shudderer off :P
22:01:21 <ais523> heh, it beats defend9.5
22:01:26 <pikhq> elliott: THOU ART MISTAKEN.
22:01:31 <pikhq> elliott: FOR I HATH DECREED IT.
22:01:38 <ais523> good_vibrations is still there; is it a shudderer or a vibrator?
22:01:54 <quintopia> oh
22:01:57 <elliott> ais523: I believe I readded vibration_fool_faster ages ago
22:01:58 <elliott> or Gregor
22:02:01 <quintopia> it was at the bottom last i checked
22:02:05 <quintopia> it'll be gone soon
22:02:17 <ais523> elliott: you didn't re-add an exact copy
22:02:20 <quintopia> pikhq: hath is present tense. you mean "did decree"
22:02:25 <Deewiant> elliott: elliott_return_of_vibration_fool_fasting_because_it_is_lent_mathematica_edition.bfjoust
22:02:27 <ais523> because it gets a different score from the actual VFF
22:02:30 <elliott> ais523: I did; must have been a different version
22:02:31 <elliott> oh
22:02:32 <elliott> Deewiant: that's not VFF
22:02:35 <Gregor> Gregor_julius_freezer is a shudderer
22:02:40 <elliott> that's my modification of VFF resubmitted
22:02:42 <elliott> it was literally called that
22:02:48 <Deewiant> I know
22:02:51 <ais523> elliott: how did you modify it?
22:02:56 <elliott> ais523: Randomly, probably
22:02:59 <Deewiant> I just think that's the only related one you've added
22:02:59 <ais523> heh
22:03:00 <elliott> I have no idea how it was Mathematica edition
22:03:12 <elliott> I think it's still on the hill, though, relatively high up too
22:03:21 <elliott> 42 31 23.36 -7.24 elliott_return_of_vibration_fool_fasting_because_it_is_lent_mathematica_edition.bfjoust
22:03:23 <elliott> OK, not so high.
22:03:34 <elliott> 41 22 23.70 -3.55 ais523_defend9_5.bfjoust
22:03:38 <elliott> wow, 9.5 must have been pretty bad
22:03:45 <ais523> it isn't /bad/, but it's passive
22:03:51 <ais523> which tends to make programs score badly
22:04:00 <ais523> it's also vulnerable to turtles, and there are a bunch of those around
22:04:12 <ais523> 9.75 is just 9.5 with those weaknesses fixed, and look how much better it's doing
22:04:45 <ais523> heh, strangely 9.5 beats decoybooster2 and 9.75 doesn't, but that's the only exception to 9.75 beating or tying 9.5
22:04:46 <elliott> so the next Debian release will be called wheezy
22:04:49 <elliott> which is a pretty bad name
22:05:07 <ais523> (also, 9.5 beats 9.75, but I can manipulate that result trivially)
22:05:12 <elliott> 08:16:29 <Gregor> $ du -h furry_furry_dominance_girls.bfjoust
22:05:12 <elliott> 08:16:29 <Gregor> 71Mfurry_furry_dominance_girls.bfjoust
22:05:13 <elliott> 08:16:32 <Gregor> Hahawhoops
22:05:13 <elliott> 08:23:37 <ais523> that could take a while to run
22:05:13 <elliott> 08:23:40 <ais523> even in lance
22:05:17 <elliott> Um, lance would tell Gregor that program is too big X-D
22:05:29 <quintopia> ais523: does the definder sequence ever go to zero if the opponent is not changing the flag?
22:05:35 <elliott> 07:59:09 <Phantom_Hoover> http://i.min.us/il4KQQ.jpg
22:05:35 <elliott> 07:59:19 <Phantom_Hoover> I don't think this person realises that this already exists and it's hideous.
22:05:36 <elliott> heh, I use to use that user javascript
22:05:38 <elliott> but it was horrible and annoying
22:05:40 <elliott> so i removed it
22:05:48 <quintopia> aka, go to zero and stop there
22:05:48 <ais523> (and I know why it happens, I think I might even have done it that way round deliberately)
22:05:51 <Gregor> "<elliott> heh, I use to use that user javascript" wtf
22:05:51 <elliott> (ais523: that one that gave an orange box of any page link you hovered over)
22:05:56 <elliott> <elliott> (ais523: that one that gave an orange box of any page link you hovered over)
22:06:01 <elliott> You put it in your monobook.js.
22:06:10 <ais523> quintopia: the original doesn't; definder2 does, IIRC
22:06:18 <elliott> 08:30:27 <Gregor> We should set a size limit :P
22:06:18 <elliott> 08:30:29 <Gregor> Maybe 640K
22:06:25 <elliott> Gregor: lance's is 1 Mio I think.
22:06:28 <elliott> (ha, stole the o stuff from Deewiant)
22:06:32 <Deewiant> Thief
22:06:34 <Gregor> 640K instruction characters, not input characters.
22:07:16 <ais523> elliott: mibioctet?
22:07:21 <ais523> *mebioctet
22:07:22 <elliott> ais523: yep
22:07:22 <Deewiant> mebi*
22:07:25 <elliott> yep
22:07:31 <ais523> a mibicotet would be, umm, 1/1024th of an octet?
22:07:32 <elliott> Deewiant: But do you say kio or Kio?
22:07:40 <elliott> 08:39:53 <cheater00> look what digging up my browser history brought
22:07:40 <elliott> 08:40:00 <cheater00> an old paste with vagrant on it
22:07:40 <elliott> rolleyes
22:07:41 <Deewiant> kibi is Ki
22:07:47 <elliott> Deewiant: But that's Inconsistent.
22:07:52 <Deewiant> kibi is Ki.
22:07:55 <elliott> Deewiant: (And not SI standard anyway, so it's irrelevant,)
22:07:57 <elliott> *irrelevant.
22:08:00 <quintopia> ais523: did you ever test the original definder without the tripwire, egodeath style?
22:08:12 <elliott> Deewiant: Who cares what the IEC says :P
22:08:15 <ais523> quintopia: *Nyarlathotep style?
22:08:17 <elliott> ki is far more consistent.
22:08:20 <ais523> and no, I don't think I did
22:08:26 <quintopia> egodeath did it too iirc
22:08:31 <Deewiant> k is inconsistent in the first place :-P
22:08:39 <elliott> Deewiant: But it's SI, and SI standards are unquestionable.
22:08:44 <ais523> but the timings would be off if the opponent took more than 98 turns to turn up
22:08:46 <ais523> which happens quite a lot
22:09:15 <elliott> I love how I can say "34 kb" to mean 4.15 kio :P
22:09:20 <elliott> SUCH CONFUSE
22:09:25 <elliott> (kilobits)
22:09:26 <ais523> elliott: kilobels?
22:09:30 <ais523> oh, that's B
22:09:33 <elliott> ais523: kilobits, bits is lowercase b
22:09:39 <quintopia> ais523: you can determine the most likely moment for the opponent to show up by examining timings on the current hill and just wait that long
22:09:55 <ais523> quintopia: wow that'd be overspecific to the current hill
22:09:59 <ais523> and such programs tend to do badly
22:11:04 <elliott> I'm going to use "1 do" to mean 0.8 b.
22:11:10 <elliott> (decioctet)
22:11:28 <elliott> "Oh yeah, that's about 100 dos" = "Oh yeah, that's about 10 bits".
22:11:29 <elliott> SO CONFUSING
22:11:31 <quintopia> ais523: i doubt it would be actually. i suspect an average of timings over the current hill would be well-suited to the future as well. however, i expect that it would do badly in any case, i was just wondering how badly
22:12:10 -!- augur has joined.
22:12:18 <elliott> Hey Gregor
22:12:20 <elliott> Your template is broken
22:12:21 <elliott> "http://codu.org/eso/bfjoust/egojsout/?ap&l=5d19f8ddd438f31e319b3e7e33b7f06cd4d444d7&r=simple&t=25{{{p}}}"
22:12:27 <elliott> You need {{{p|}}}
22:12:36 <elliott> Or, preferably, {{qif|test={{{p|}}}|then=&p}}
22:12:44 <elliott> Then you can just do |p=1 in the template invocation.
22:12:47 <quintopia> elliott: just use |t=32&p
22:12:50 <elliott> Permission to fix?
22:12:52 <Gregor> Uhhh ... I don't think I put {{{p}}} there ...
22:12:52 <quintopia> it works i promise
22:12:55 <elliott> Gregor: You totally did.
22:12:57 <elliott> Or someone did.
22:12:58 <elliott> Not me.
22:13:10 <elliott> I'll add {{{p}}} support anyway :P
22:13:12 <elliott> Or rather, fix it.
22:13:18 <elliott> (cur) (last) 17:38, 15 February 2011 Ais523 (Talk | contribs) (add option to allow polarity invert)
22:13:19 <elliott> Blame found
22:13:31 <ais523> elliott: I was just going to fix that, because I just noticed it unrelatedly to you telling me
22:13:37 <elliott> TOO BAD ALREADY FIXED
22:13:47 <elliott> erm
22:13:49 <elliott> failed edit
22:13:56 <elliott> argh
22:13:59 <elliott> mediawiki is so slow
22:14:13 <ais523> looks like I got there first
22:14:22 <elliott> ais523: you fixed it wrong, though :P
22:14:25 * elliott fixes properly
22:14:38 <ais523> you're not going to use qif, are you?
22:14:40 -!- poiuy_qwert has joined.
22:14:49 <elliott> ais523: ALREADY DID
22:14:54 <elliott> It's like Christmas in a template.
22:15:12 <elliott> Fixed the article too.
22:15:14 <elliott> (removed &p stuff)
22:15:19 -!- ais523 has quit (Read error: Connection reset by peer).
22:15:23 <elliott> Gregor: Can't you just make the p parameter take either "sieve" or "kettle"? :P
22:15:26 <elliott> That would be far, you know, saner.
22:15:35 <elliott> Then we could just do &p={{{p|sieve}}}
22:15:36 <Gregor> elliott: I could. I choose not to. To make you hurt.
22:16:09 <elliott> argh )
22:18:48 -!- pikhq_ has joined.
22:18:51 -!- pikhq has quit (Ping timeout: 260 seconds).
22:27:08 -!- Ilari_ has changed nick to Ilari.
22:30:13 -!- augur_ has joined.
22:31:21 -!- augur has quit (Read error: Connection reset by peer).
22:36:41 -!- Sgeo has joined.
22:37:19 -!- augur_ has quit (Remote host closed the connection).
22:38:34 -!- poiuy_qwert has quit (Quit: This computer has gone to sleep).
22:42:01 -!- BeholdMyGlory has joined.
22:45:54 -!- Behold has quit (Ping timeout: 264 seconds).
22:50:51 <elliott> Debian sid is currently completely uninstallable.
22:51:04 <elliott> "linux-image-2.6-686 : Depends: linux-image-2.6.37-1-686 but it is not installable"
22:52:37 <quintopia> why do you need sid? what's it got that squeeze ain't got?
22:53:23 <pikhq_> elliott: Debian sid is always uninstallable.
22:53:25 <elliott> quintopia: *wheezy, surely.
22:53:27 <elliott> pikhq_: Untrue.
22:53:27 -!- pikhq_ has changed nick to pikhq.
22:53:39 <elliott> pikhq: The debian-installer nightly businesscard images offer sid installation.
22:53:42 <elliott> But it's not guaranteed to ever work :P
22:53:46 <pikhq> elliott: Oh, never mind.
22:53:52 <elliott> quintopia: squeeze is the current stable release (i.e. will rapidly become a dinosaur).
22:53:59 <elliott> quintopia: wheezy is the current testing release.
22:54:04 <elliott> quintopia: I want to install sid because I'm ker-azy.
22:54:09 <quintopia> O KAY
22:54:10 <pikhq> squeeze is *slightly* out-of-date as-is.
22:54:13 <elliott> And because wheezy isn't even security-team-supported right now.
22:54:16 <elliott> So it'd be a waste of time :P
22:55:08 <elliott> "Warning: Failure when configuring base packages."
22:55:17 <elliott> sid installed yesterday, but wouldn't boot :-D
22:55:19 <elliott> now it won't even install
22:56:16 <pikhq> Sid is probably all sorts of crazy right now.
22:56:25 <elliott> pikhq: The technical term is "fun".
22:56:39 * elliott tries linux-image-2.6.32-5-686. Specificity will set you free!
22:56:55 <elliott> Woop, seems to be working.
22:57:05 <pikhq> As they're probably putting in every single package that couldn't make it in during the Squeeze testing freeze.
22:58:07 <elliott> Happy tenth birthday, All Your Base.
22:58:58 <elliott> pikhq: The freeze model is a bit broken.
22:59:36 <pikhq> elliott: Yeaaah.
23:00:42 <pikhq> It'd make a bit more sense, IMO, to have testing as a rolling release model, fork for a stable release, and do bug fixes on that fork.
23:01:14 <elliott> Or just have a stable rolling release :P ... nah, that would be impossible with Modern Linux.
23:01:36 <elliott> Unless you're deluded enough to think that every software will support udeviceHALKit the day it comes out.
23:01:46 <elliott> (Or, indeed, before its predecessor removed entirely.)
23:01:49 <elliott> *is removed
23:04:03 -!- Behold has joined.
23:07:17 -!- BeholdMyGlory has quit (Ping timeout: 246 seconds).
23:08:23 -!- poiuy_qwert has joined.
23:08:57 -!- zzo38 has joined.
23:10:42 <Gregor> All the strategies I've tried for furry_furry_leather_discipline_girls have gone awry >_>
23:11:04 <elliott> So what comes after furry_furry_leather_discipline_girls?
23:11:10 <elliott> I'm scared.
23:11:16 -!- FireFly has quit (Quit: swatted to death).
23:12:10 <Gregor> >: )
23:12:10 -!- augur has joined.
23:12:27 <Gregor> I've got a list, but then I start to run low on femdom-inspired sexual fetishes ...
23:12:43 <elliott> X-D
23:12:51 <zzo38> Gregor: Just select words at random, then?
23:13:07 <elliott> furry_furry_exegesis_girls
23:13:10 <elliott> Gregor: DO IT
23:15:21 <pikhq> Gregor: Shouldn't run out of fetishes in general, though.
23:15:42 <Gregor> The furry furry girls are dominatrixes :P
23:15:58 <Gregor> (They're not the ones in bondage, you are)
23:17:19 <elliott> Are they just furry, or furry furry?
23:17:41 <pikhq> They're furry furries.
23:20:33 <Gregor> I don't think you can comprehend how furry they are.
23:21:04 <pikhq> I was stating that they are furry (as in very, very hairy) and furry (as in the subculture).
23:22:02 <elliott> Gregor: Comment on the hairiness of the aforementioned girls?
23:22:16 <Gregor> Only the latter.
23:22:31 <Gregor> They have a normal, human amount of hair.
23:22:42 <Gregor> Which they may or may not do less with than the average western female.
23:23:06 <pikhq> *Awww*.
23:23:39 <pikhq> I was imagining near-apes in fursuits.
23:27:23 <elliott> Gregor: But they have reaaaaally long armpit hair, right?
23:27:35 <elliott> Like, they could fashion a cushion out of it and sit on it.
23:27:40 <elliott> That would be pretty darn furry.
23:27:42 <Gregor> pikhq: Hey, keep your fetishes to yourself.
23:27:45 <Gregor> pikhq: They've got their own.
23:27:51 <pikhq> Gregor: I never said it was my fetish.
23:28:05 <pikhq> Gregor: I simply state that furry people in fursuits would be funny.
23:28:22 <zzo38>
23:28:31 <elliott>
23:28:50 <zzo38> Oops, sorry, I think I wrote it wrong by mistake
23:30:12 <zzo38> Now answer my question
23:31:09 <elliott> What question
23:31:24 <zzo38> This question?
23:31:32 <elliott> wat
23:32:44 <zzo38> Actually that isn't the question. The question is, how to figure out what other people that are interested with TeXnicard to comment/help?
23:34:55 <zzo38> And how do I fix the lights in the keyboard?
23:38:08 <quintopia> i've tried finding other people for you zzo38 but it means finding someone like you, and so no such person exists
23:39:42 <zzo38> quintopia: At least you tried. However, maybe there are people interested in these kind of things if you want to make high-quality card game... . If you cannot answer that, however, can you answer the other question about how fix the lights in the keyboard?
23:41:15 <zzo38> How am I supposed to find the people interested in this cards, by myself?
23:41:54 <quintopia> ask on forums where geeky people frequent
23:42:47 <quintopia> and if you disassemble your keyboard you can maybefigure out what's wrong and fix it
23:43:53 <Gregor> Hahah, I've upped my fail and generated 400MB :P
23:44:29 <elliott> Gregor: So if lance supports 400 meg programs...
23:46:24 <elliott> Y'know, maybe I should actually work on lance :P
23:46:30 <elliott> I'll... do that now.
23:47:53 <Gregor> No, we should set a hard instruction limit (unexpanded)
23:48:35 <elliott> Gregor: Yah, but what? I don't want to make it a hard forever rule :P
23:48:36 <Deewiant> Unexpanded 64K fits everything we have currently
23:48:41 <elliott> I think a 700K program would be fine, for instance.
23:48:44 <elliott> I don't mind generation.
23:49:00 <elliott> "Anything that can feasibly be downloaded from the Internet in 30 seconds from EgoBot", say :P
23:49:12 <quintopia> we should make a new instruction set that makes these kind of constructs more compact and set the limit on that one
23:49:12 <Gregor> Hard, /static/ limit X-P
23:49:18 <Gregor> Noooo
23:49:23 <Gregor> No exponential programs.
23:49:28 <Gregor> Unrolling is a lame strategy.
23:51:11 <elliott> Gregor: We already have exponential programs with lance (which differentiates between * and %).
23:51:47 <Gregor> I'm using a rather silly definition of "exponential programs" :P
23:51:56 <Gregor> Basically I don't want you to be able to go too far just by always knowing exactly where you are.
23:52:13 <quintopia> why not?
23:52:15 <quintopia> :/
23:52:31 <elliott> Gregor: It would be nice to have something that lets us transform [a[b[c[d]]]]x into [a[b[c[d]x]x]x]x :P
23:52:36 <elliott> (To "break out of a loop".)
23:52:42 <elliott> Sure, it's "cheating", but... a hard limit irks me.
23:52:48 <elliott> space_elevator is awesome. But not unbeatable.
23:52:49 <quintopia> elliott: my subroutine idea frex
23:53:00 <elliott> quintopia: I prefer things that are essentially string rewriting, but optimised :P
23:53:05 <elliott> Gregor: I don't think a hard limit will prevent any god strategies.
23:53:10 <elliott> I don't think there _are_ any unbeatable programs.
23:53:31 <Gregor> No, of course not, but things start getting totally impossible to understand once the program is just a loop-unrolled mess of nonsense.
23:53:37 <quintopia> elliott: that's what it was. a construct that implements subroutines by string rewriting
23:53:54 <elliott> Gregor: The rules aren't based on your pitiful mind's tiny capacity :P
23:54:03 <Gregor> elliott: I'm preventing MYSELF here.
23:54:06 <Gregor> elliott: I'M writing programs that do this.
23:54:12 <elliott> Gregor: I approve!
23:54:15 <Gregor> elliott: I'M trying to make sure that this retard strategy doesn't work.
23:54:18 <elliott> I do think open-sourcing the generators for such program is a good thing, though.
23:54:23 <elliott> Gregor: It won't work, we'll beat it :P
23:54:48 <Gregor> They're still size-infeasible, so if they're a legit strategy then we should either have a macro system or disallow it *shrugs*
23:55:21 <elliott> Gregor: Does EgoBot refuse to download files over a certain size?
23:55:22 <quintopia> yes
23:55:25 <quintopia> macro system
23:55:34 <Gregor> elliott: I don't recall :P
23:56:08 <elliott> Gregor: If it does, let's just rely on that... I don't like adding such things at the battle layer :P
23:56:17 <elliott> Especially since, y'know, trying programs like that locally sounds like fun.
23:56:25 <Gregor> Whoahwhoahwhoah
23:56:29 <Gregor> I was talking about for the HILL, not the language.
23:56:34 <elliott> Gregor: Yah, I know.
23:56:47 <elliott> Gregor: I'm fine with a limit in !bfjoust, but I have a feeling that EgoBot already handles a sane enough limit.
23:56:54 <elliott> I don't want a limit that is treated as unexpandable :P
←2011-02-14 2011-02-15 2011-02-16→ ↑2011 ↑all