←2012-02-01 2012-02-02 2012-02-03→ ↑2012 ↑all
00:00:33 -!- pikhq_ has joined.
00:00:42 -!- pikhq has quit (Ping timeout: 244 seconds).
00:01:05 <fizzie> Current counts, FWIW: http://sprunge.us/aaSI
00:01:52 <fizzie> Also fi:aasi = en:donkey.
00:03:13 <oerjan> don't be an ass about it
00:04:06 <ion> fi:perse ≠ en:per se
00:04:17 <fizzie> ^ul (^.^)aaS
00:04:17 <fungot> ((^.^))
00:05:41 <ais523> fizzie: heh, I didn't really expet your script to be running that long
00:05:45 <ais523> *expect
00:05:54 <ais523> and, hmm, that's an interesting distribution
00:06:00 <ais523> not sure if I'd expect that from a uniform distribution or not
00:06:23 <fizzie> If I compute the chi2 statistic right, you probably shouldn't.
00:07:05 <fizzie> (You should expect only three out of each hundred uniform distributions to be that out of whack.)
00:09:08 <fizzie> Of course it's still not any sort of proof. But it's a larger difference that you'd expect (in a non-mathematical sense, I haven't calculated any numbers about this) from the bias inherent in rand()%11 with RAND_MAX of 2^31-1 or whatever.
00:11:54 <fizzie> Even with just 0..32767, you'd expect 0..9 to be just 1.0003 times more common than 10, 11, which is not a difference you'd see here.
00:19:03 -!- ais523 has quit (Remote host closed the connection).
00:38:17 -!- Phantom_Hoover has quit (Ping timeout: 240 seconds).
00:39:01 -!- augur has quit (Remote host closed the connection).
00:39:33 -!- augur has joined.
00:42:53 <kallisti> @hoogle freeHaskell
00:42:53 <lambdabot> Foreign.Ptr freeHaskellFunPtr :: FunPtr a -> IO ()
00:44:10 -!- augur has quit (Ping timeout: 248 seconds).
01:14:35 <oerjan> in Nagapattinam, Tamil Nadu, India, there is apparently a Sir Issac Newton College of education and (possibly distinct?) Sir Issac Newton Polytechnic College.
01:14:49 <oerjan> by all web evidence those are the official spellings.
01:16:44 <monqy> amazing best
01:17:45 <oerjan> i have this small personal task i occasionally take, to search wikipedia for that particular misspelling.
01:18:14 <oerjan> (you should have seen how common it was before the first time i did so :P)
01:18:45 <monqy> apparently issac is a real name and also a commune in france
01:19:01 <oerjan> yeah
01:25:20 <oerjan> http://www.sinc.in/
01:30:23 <oerjan> you don't happen to have marvel S.H.I.E.L.D. comics? i need to find out if Issac Newton is actually misspelled in those :P
01:30:50 <oerjan> or if i can change the two relevant wikipedia pages
01:31:04 * oerjan already asked on the Talk: page, mind
01:31:07 <monqy> I don't even know what those are :(
01:31:29 <oerjan> some kind of secret organization superhero comics
01:31:33 <oerjan> iiuc
01:32:45 <oerjan> also ancient, which is how Iss?aa?c Newton was a member
01:37:56 -!- cheater has quit (Read error: Operation timed out).
01:38:48 -!- cheater has joined.
01:42:19 -!- augur has joined.
01:42:43 -!- derdon has quit (Remote host closed the connection).
01:54:09 <kallisti> lolwat someone made a dupdog thing?
01:54:57 <oerjan> someone tried to. it's not exactly finished.
01:55:23 <oerjan> misinterpreting the spec doesn't help either
01:55:34 <kallisti> having a poorly written spec doesn't help either.
01:56:24 * kallisti hasn't looked at the source, but immediately recommends not using String as the data/program representation
01:56:53 <oerjan> i've thought for a while that dupdog is like made for using ropes
01:57:29 <oerjan> that way, you could interpret it usefully even if the size of the full expansion blows up exponentially
01:58:01 <oerjan> (ropes with a reversal flag, to be precise)
02:01:38 <kallisti> yes that would be good
02:01:51 <kallisti> Data.Sequence would work well I think.
02:02:32 <oerjan> well the thing is i'm not sure if Data.Sequence has enough sharing.
02:02:45 <oerjan> and it also needs reversal added.
02:03:20 <oerjan> you want duplication to be cheap
02:03:32 <kallisti> hm yes
02:03:51 <kallisti> reversal would simply be a matter of changing which view you use, right?
02:04:12 <oerjan> hm... actually yes
02:04:27 <oerjan> since you never concat things with mismatching reversal
02:04:33 <kallisti> believe <> is something like O(min(a,b))
02:04:44 <kallisti> which isn't too bad.
02:04:59 <kallisti> but not ideal.
02:05:28 <oerjan> er shouldn't there be a log in there
02:05:51 <kallisti> oh yes
02:05:54 <kallisti> and it's ><
02:06:38 * kallisti would like to expand on the idea of dupdog but isn't sure which direction to go.
02:07:10 <oerjan> still, a rope does that particular part in constant time. but i guess the logarithm then comes back when looking at the end instead.
02:07:20 <kallisti> right
02:07:42 -!- tswett has quit (Ping timeout: 252 seconds).
02:08:51 <kallisti> I think reversal is a useful operation
02:08:58 <kallisti> but duplication isn't, aside from looping infinitely
02:09:05 <kallisti> in a really clumsy manner.
02:09:46 <oerjan> i'm not sure. note that duplication and then running a few commands means the two parts are no longer equal.
02:10:00 <kallisti> yes that's true.
02:11:40 <kallisti> I think one parametric command would be of value.
02:11:55 <kallisti> perhaps to control the transliteration
02:13:24 <kallisti> the simplest one I can think of would be sab, which substitutes every a to b
02:13:32 <kallisti> this would consume the entire 3-character command
02:14:36 <kallisti> I think this would make dupdog quite a bit more powerful, though I'm not sure how to quantify how.
02:15:27 <kallisti> well
02:15:30 <kallisti> it's not dupdog anymore
02:15:34 <kallisti> because obviously the semantics are different
02:15:58 <kallisti> but the idea of advancing forward in a string, with each character modifying the source code in some simple way, remains.
02:58:50 -!- oerjan has quit (Quit: Goof night).
03:18:37 -!- ion_ has joined.
03:22:17 -!- ion has quit (Quit: switching to weechat for now).
03:22:17 -!- ion_ has changed nick to ion.
03:25:27 -!- pikhq has joined.
03:25:38 -!- pikhq_ has quit (Ping timeout: 248 seconds).
03:39:28 -!- Baco_ has joined.
03:39:48 -!- Baco_ has left ("Leaving").
04:03:19 <pikhq> Your professor might be laughably archaic if: they demand a hardcopy submission of code
04:08:16 <Sgeo> kallisti has been i[dated
04:08:18 <Sgeo> i[date
04:08:19 <Sgeo> d
04:08:21 <Sgeo> updatd
04:22:04 -!- GreaseMonkey has joined.
04:22:04 -!- GreaseMonkey has quit (Changing host).
04:22:04 -!- GreaseMonkey has joined.
06:19:17 -!- madbr has joined.
06:19:34 <madbr> Ho man
06:20:46 <madbr> Language with: +, -, ! (C style logical not), non-deterministic
06:20:54 <madbr> Is turing complete I think
06:21:44 <madbr> with non-deterministic = in the "prolog" kind of way
06:22:25 <madbr> program flow and arrays can be hacked from the non-deterministic logic
06:27:41 <quintopia> you did it?
06:28:09 <madbr> Well, I dunno how to write an interpreter
06:28:16 <quintopia> but
06:28:18 <quintopia> wiki?
06:28:20 <madbr> more precisely I dunno how to garbage collect it
06:28:33 <quintopia> do it the prolog way?
06:28:47 <quintopia> eh
06:28:50 <quintopia> just dont
06:28:54 <madbr> quintopia : essentially it's a numeric version of this: http://esolangs.org/wiki/Fatmouse
06:28:59 <quintopia> gc is an implementation detail
06:29:07 <quintopia> if you never free anything
06:29:12 <quintopia> the language will work
06:29:14 <quintopia> in theory
06:29:20 <madbr> quintopia : I'm exploring "grow only" languages
06:29:30 <quintopia> and esoteric languages are all about working in theory
06:29:37 <madbr> ie languages where you can't change the value of a variable once it's defined
06:30:36 -!- Sgeo has quit (Ping timeout: 276 seconds).
06:30:49 <madbr> so to do something like interpret a brainfuck program, you create a new copy of the BF program state on each cycle
06:31:11 <madbr> That's turing complete
06:31:48 <madbr> but if I want to do an interpreter, I'd like to be able to execute an infinite number of loops in finite memory at least
06:32:00 <madbr> that means I have to figure out how to garbage collect it
06:32:29 <madbr> problem: in a normal language, your variables eventually get dereferenced
06:32:37 <madbr> and then you can garbage collect them
06:33:10 -!- SgeoN1 has joined.
06:33:27 <SgeoN1> Can a full hard drive cause kernel panics?
06:33:28 <madbr> except in this case, how do you know a given variable can't generate any new values and can thus be collected?
06:35:44 <quintopia> gcing in fatmouse is easy: no data needs to ever be collected except the output buffer
06:36:35 <quintopia> in particular, every time a variable is consumed, all conditions in the program are checked. those that are satisfied are deleted.
06:36:38 <madbr> quintopia: then it's impossible to interpret it without leaking memory like crazy
06:36:49 <quintopia> if a variable has no conditions, it is deleted.
06:37:04 <quintopia> *consumed
06:37:37 -!- Sgeo has joined.
06:40:06 <quintopia> iterative conditionals just require you to keep around a short list of ranges that have been consumed. badly behaved programs could misuse this, but...garbage collection never works on badly behaved programs anyway
06:40:53 <madbr> hm
06:43:12 <madbr> hmmmm
06:43:23 <quintopia> the "array.x.x array.x-1.x-1" example might cause problems in a naive algorithm, but i can think of a way to handle such things efficiently
06:47:14 <madbr> I think I can see how it's possible
06:47:18 <quintopia> a particularly nasty one might be "a.x.y a.x-1.y-1;a.x.y a.x-2.y;a.x.y a.x.y-2;a.1.1"
06:47:26 <quintopia> a checkerboard pattern
06:48:10 <madbr> that's essentially a memory leak
06:48:19 <madbr> and an infinite loop
06:48:22 <quintopia> yep
06:48:55 <quintopia> what happens in fatmouse if the same variable is listed twice with different conditionals?
06:49:11 <quintopia> (as above)
06:50:10 <quintopia> i'm guessing it's a disjunction over the conditional sets
06:50:22 <madbr> yeah if
06:50:31 <madbr> b a
06:50:32 <madbr> b c
06:50:48 <madbr> if either a or c get defined, b gets defined too
06:50:55 <quintopia> yep makes sense
06:52:29 <quintopia> so a loop that stops looks like "a.x a.x-1 x<10;a.1"?
06:53:06 <quintopia> with, uh, suitable parens where needed?
06:54:58 <madbr> yeah
06:58:49 <madbr> night
06:58:57 -!- madbr has quit (Quit: Page closed).
07:18:53 -!- SgeoN1 has quit (Quit: Bye).
07:43:15 -!- monqy has quit (Quit: hello).
08:32:33 -!- nooga has joined.
09:00:41 -!- ais523 has joined.
09:28:02 <ais523> wait that makes no sense
09:28:05 <ais523> I ssh -X to a computer
09:28:09 <ais523> try to open something in Firefox on it
09:28:14 <ais523> and it opens on my local Firefox install instead
09:28:43 <shachaf> Firefox is evil that way.
09:28:58 <ais523> it's more nonsensical than evil
09:29:52 <ais523> haha, it works the other way too
09:30:01 <ais523> if I close Firefox locally and open the remote Firefox
09:30:11 <ais523> then trying to open Firefox locally just opens a new window on the remote Firefox
09:30:44 <ion> :-)
09:31:04 <ais523> I guess the Firefox process, when started, looks for existing Firefox /windows/
09:31:10 <ais523> and when it finds them, signals them to do what it was trying to do
09:31:19 <ais523> rather than looking for existing Firefox /processes/, which would make more sense
09:32:03 <ais523> this is the first time I realised that sshing into the computer lab downstairs to print something made more sense than physically going there
09:32:18 <ais523> but I'm annoyed that I can't browse the web meanwhile
09:32:30 <fizzie> ISTR that there was a workaround, though. (I've used a remote Firefox sometimes to access our 'intranetty' things, and I don't recall having to close the local Firefox.)
09:32:51 <fizzie> -no-remote Open new instance, not a new window in running instance.
09:32:53 <fizzie> Maybe that one.
09:33:54 <fizzie> It's not possible to run two separate Firefoxes using the same profile even with that, but it should be okay to run something on a remote computer, since presumably it's not using the same profile.
09:34:12 <fizzie> (If it is, e.g. over NFS, I think it might not be the best of ideas.)
09:36:19 <fizzie> Also latest counts for 6797 samples: http://sprunge.us/GjLD -- and the chi2 score is 60, which corresponds to virtual certainty of rejecting the "uniform distribution" hypothesis (p=10^-9 or so).
09:36:25 -!- GreaseMonkey has quit (Quit: The Other Game).
09:36:29 <fizzie> Strange sort of bias, though.
09:38:21 <ais523> fizzie: it isn't in this case
09:38:37 <ais523> as for that distribution, I'm wondering if it's intended to be uniform but the randomizer is broken
09:42:46 <fizzie> Sounds possible, though I wonder how it is broken, and if it's time-seeded whether my ten-second polling interval has something to do with it.
09:47:19 <fizzie> Maybe after the 10k-sample run has finished I should make a forum post. They have forums, right? Though maybe it's a 'faux pas' to speak of the About page without pretending not to notice the randomness.
09:54:25 <ais523> mezzacotta has forums, I think they apply to everything hosted there
09:56:24 <fizzie> There seems to be a "Comments on a Postcard"... uh, group? topic? area? region? forum? shindig? hunting ground? ...in there.
09:58:32 <fizzie> OTOH, I've already download 40 megabytes' worth of data, I'm not entirely sure I want to admit that.
09:59:18 <fizzie> (The page sizes range from 2912 to 4360 bytes.)
10:11:15 -!- jix has quit (Quit: leaving).
10:14:10 -!- jix has joined.
10:44:28 -!- ion has quit (Remote host closed the connection).
10:45:10 -!- ion has joined.
11:18:04 -!- rodgort has quit (Ping timeout: 252 seconds).
11:21:03 <fizzie> Just in case, also fed the numbers to MATLAB's χ2-goodness-of-fit test, and it agrees on the unlikeliness: http://sprunge.us/SSQL
11:24:12 <fizzie> Sorry, http://sprunge.us/EjLY -- misread the documentation for 'nparams'.
11:24:57 <fizzie> Admittedly I *was* wondering about the p = NaN bit, thought it'd underflow to 0 instead, and that it wouldn't be *that* small quite yet.
11:30:07 -!- rodgort has joined.
11:31:59 <ais523> fizzie: seemed a bit weird to me too
11:32:06 <ais523> NaN means "no probability", rather than "no chance" :)
11:36:39 <fizzie> If I disregard the highest-numbered count, I get p=0.0596, which isn't yet quite unlikely enough.
11:37:58 <ais523> indeed
12:06:52 -!- variable has quit (Excess Flood).
12:08:57 -!- variable has joined.
12:51:07 -!- oerjan has joined.
13:22:33 -!- jix_ has joined.
13:23:19 -!- jix_ has quit (Client Quit).
13:29:31 -!- ais523 has quit (Ping timeout: 252 seconds).
13:34:48 -!- derdon has joined.
13:48:00 -!- ais523 has joined.
13:48:06 -!- ais523 has quit (Changing host).
13:48:06 -!- ais523 has joined.
14:09:29 -!- cheater has quit (Read error: Operation timed out).
14:09:41 -!- cheater_ has joined.
14:11:11 -!- Phantom_Hoover has joined.
14:25:21 -!- cheater has joined.
14:27:32 -!- cheater_ has quit (Ping timeout: 252 seconds).
14:32:02 -!- ais523 has quit (Ping timeout: 252 seconds).
14:33:16 -!- cheater_ has joined.
14:35:11 <Sgeo> WHY did my HD just go read-only?
14:35:35 -!- cheater has quit (Ping timeout: 272 seconds).
14:37:48 <fizzie> Remount-as-read-only is one common response to errors.
14:38:11 <fizzie> The "errors=remount-ro" mount option.
14:38:48 <ion> sgeo: dmesg | tail
14:39:33 <Sgeo> Sorry in advance for incoming flod
14:39:36 <Sgeo> sgeo@sgeo-Satellite-A105:~$ dmesg | tail
14:39:36 <Sgeo> [28984.216422] ecryptfs_write_inode_size_to_header: Error writing file size to header; rc = [-30]
14:39:36 <Sgeo> [28984.216429] Error writing inode size to metadata; rc = [-30]
14:39:36 <Sgeo> [28984.216612] ecryptfs_write_inode_size_to_header: Error writing file size to header; rc = [-30]
14:39:36 <Sgeo> [28984.216619] Error writing inode size to metadata; rc = [-30]
14:39:36 <Sgeo> [28984.216633] ecryptfs_write_inode_size_to_header: Error writing file size to header; rc = [-30]
14:39:38 <Sgeo> [28984.216640] Error writing inode size to metadata; rc = [-30]
14:39:40 <Sgeo> [28991.831985] ecryptfs_write_inode_size_to_header: Error writing file size to header; rc = [-30]
14:39:42 <Sgeo> [28991.831996] Error writing inode size to metadata; rc = [-30]
14:39:44 <Sgeo> [28991.832082] ecryptfs_write_inode_size_to_header: Error writing file size to header; rc = [-30]
14:39:46 <Sgeo> [28991.832089] Error writing inode size to metadata; rc = [-30]
14:40:01 <fizzie> You could've just said "repeats N times".
14:40:16 <Sgeo> Maybe it isn't remounted RO, but .. something else?
14:40:16 <ion> Ok, the last 10 lines weren’t enough, you’ll need to look at more of it. dmesg | less and scroll around. What happened before those errors?
14:40:45 <Sgeo> [29068.269545] ecryptfs_encrypt_page: Error attempting to write lower page; rc =
14:40:45 <Sgeo> [-30]
14:40:45 <Sgeo> [29068.269553] ecryptfs_writepage: Error encrypting page (upper index [0x0000000
14:40:45 <Sgeo> 000001cd3])
14:40:59 <Sgeo> That occurs a bunch of times
14:41:06 <Sgeo> But there's the earlier errors I pasted before then
14:41:30 <fizzie> Well, it's very confused right now. The first error is often the most relevant.
14:41:38 <ion> Anything about problems with the actual device (that might be a hardware issue) or just ecryptfs problems (that, if alone, looks like a software bug)?
14:41:40 <fizzie> Especially if it comes from some lower level than ecryptfs.
14:41:53 <Sgeo> Just hit the top of dmesg
14:41:57 <Sgeo> More encryptfs stuff
14:42:08 <Sgeo> No
14:42:15 <Sgeo> I confused PageUp and PageDown
14:42:20 <ion> Well, it could be a symptom of bad RAM or other faulty hardware, too, but probably not e.g. a broken HDD.
14:42:41 <Sgeo> encryptfs errors fill dmesg
14:43:03 <Sgeo> I had a similar issue last night too. I needed to restart in recovery, let that take place, then booted up
14:43:26 <ion> less /var/log/syslog, something useful might have managed to go into it.
14:44:04 <Sgeo> DY ERR }
14:44:05 <Sgeo> Feb 2 01:59:33 sgeo-Satellite-A105 kernel: [ 1458.247678] ata1.00: error: { ICRC ABRT }
14:44:13 <Sgeo> Feb 2 01:59:33 sgeo-Satellite-A105 kernel: [ 1458.247639] ata1.00: failed command: WRITE DMA
14:44:27 <Sgeo> (Just posting interesting looking ones. And the last one I posted was before the first)
14:44:50 <Sgeo> Feb 2 01:59:33 sgeo-Satellite-A105 kernel: [ 1457.858737] ata1.00: exception Em
14:44:50 <Sgeo> ask 0x0 SAct 0x0 SErr 0x0 action 0x6
14:45:03 <Sgeo> Feb 2 01:59:33 sgeo-Satellite-A105 kernel: [ 1457.489310] res 51/84:18
14:45:03 <Sgeo> :a8:01:c5/84:01:06:00:00/e6 Emask 0x30 (host bus error)
14:45:19 <Sgeo> A whole bunch of these:
14:45:20 <Sgeo> Feb 2 01:59:32 sgeo-Satellite-A105 kernel: [ 1457.473124] Buffer I/O error on d
14:45:21 <Sgeo> evice sda1, logical block 13147229
14:45:44 <ion> Ok, that does look like a broken HDD.
14:45:55 <ion> And the ecryptfs problem is just a symptom of it.
14:46:01 <Sgeo> :(
14:46:04 <fizzie> Or the SATA port, that can be flaky too.
14:46:26 <Sgeo> My HD is a bit loose in the thing, is there a nice way to tell if it just sort of lost connection or if there's damage?
14:46:47 <ion> Yeah. But every HDD breaks sooner or later, SATA controllers are quite a bit less likely to just break by themselves.
14:47:02 <fizzie> You can run some sort of a 'badblocks' scan over it. If it's just intermittently flaky, it shouldn't find errors consistently at the same location.
14:47:04 <ion> Connect it to another computer.
14:48:07 <Sgeo> Can I try some sort of SMART thing?
14:48:11 <oerjan> > deleteBy (const (== 3)) undefined [1,2,3,4,5,3,6,7]
14:48:12 <lambdabot> [1,2,4,5,3,6,7]
14:48:38 <Sgeo> Although I don't know if I have any such thing already installed
14:48:51 <ion> smartctl -a /dev/sda | less
14:49:17 <Sgeo> The program 'smartctl' is currently not installed. You can install it by typing:
14:49:17 <Sgeo> sudo apt-get install smartmontools
14:49:36 * Sgeo needs to get ready to head for school
14:49:41 <oerjan> Sgeo: didn't you have a broken HD before (and got everyone annoyed by not following their salvaging advice...)
14:49:50 <fizzie> oerjan: Did you hear about my latest CHI^2 SCORES.
14:49:52 <ion> udisks --dump | less
14:49:54 <Sgeo> oerjan, this is a different HD
14:50:43 <ion> It doesn’t seem he’s losing any data. He hasn’t mentioned a lack of backups, which probably would be on the top of one’s mind if there are HDD problems *and* no backups. :-)
14:50:48 * oerjan cackles sadistically
14:50:57 <Sgeo> "backups"?
14:50:59 <Sgeo> >.>
14:51:15 <Sgeo> Although I don't have much of anything important on here. I think.
14:51:20 <oerjan> fizzie: i saw something about almost certainly not uniform
14:51:33 <Sgeo> Except my logs
14:51:36 <Sgeo> I love my logs
14:51:50 <ion> So… udisks --dump | less
14:52:15 <Sgeo> The status column is filled with good and n/a
14:52:29 <ion> ok
14:52:32 <fizzie> oerjan: It's like p=10^-11 likely get a this nonuniform sample if it actually were uniform.
14:53:02 <Sgeo> http://pastie.org/private/7eaa3bfslg7sdekiroftg
14:53:49 <oerjan> fizzie: ideally with enough data you should get the real fractions of probabilities; if they are close to simple rationals that should give a hint...
14:54:02 <ion> I’d connect the disk to another computer and see what happens.
14:54:11 <oerjan> *approximations to
14:54:26 <fizzie> "udma-crc-error-count" *sounds* like it might be an issue somewhere between the processor and the disk.
14:55:13 <ion> aye
14:55:27 <Sgeo> fizzie, could that be caused by a loose connection?
14:55:33 <ion> It could.
14:55:55 <oerjan> like, if there are two tiers, one with probability p and one with probability q, as you'd expect if it was (x % 11) applied to something that _is_ uniform in some integer interval
14:56:00 <Sgeo> I should really get a mount for this thing
14:56:13 <Sgeo> In the meantime, I should probably resist the temptation to take it off the table
14:56:41 <Sgeo> Going to restart
14:56:47 <ion> Try fixing it with XML.
14:56:52 <fizzie> oerjan: Yes, though normally you'd expect that x to be uniform in [0, 2^31-1] or something, in which case the difference from that would be quite small indeed.
14:56:58 -!- Sgeo has quit (Read error: Connection reset by peer).
14:57:13 <ion> Any bets whether it’ll boot? :-P
14:57:55 <fizzie> oerjan: "Normalized" counts -- as in, count/expected, which would give [1 1 ... 1], are: 1.28168 1.09694 1.02814 0.97209 0.96826 0.96699 0.95807 0.94915 0.93896 0.93514 0.90456.
14:58:22 <fizzie> oerjan: Last time I tried after ignoring the largest one, the rest weren't quite yet abnormal enough to reject the null hypothesis.
15:00:39 <fizzie> Currently MATLAB's saying p=0.01 even for the "other ten" ones.
15:01:29 <oerjan> mind you you _shouldn't_ normally be getting 0.05 from something that is uniform, even if that's not sufficient for discarding a scientific null hypothesis
15:02:08 <oerjan> so, hm
15:02:14 <fizzie> Well, you should be getting 0.05 from about every 20th experiment, I suppose.
15:02:41 -!- Vorpal has joined.
15:02:50 <oerjan> which isn't quite the normal case, is it
15:03:59 <oerjan> now for there to be only a largest one that has higher probability, the interval before x%11 should have length 11*y+1
15:04:29 <oerjan> where 1 + 1/y is the excess in probability
15:04:42 <fizzie> Yes, and for 1.3 y is kinda small.
15:05:35 <oerjan> y = 3 or 4, giving 34 or 45, which are strange numbers to choose
15:06:05 <fizzie> I was hoping for the interval of 16, implying two iterations of the Befunge '?'.
15:06:36 <oerjan> 32 would have been perfect if there was precisely one item _less_ likely than the rest
15:06:41 -!- SgeoN1 has joined.
15:06:45 <fizzie> Still, there are all those stories about patterns in bad RNGs.
15:07:03 <SgeoN1> The recovery thing worked last night...
15:07:13 <oerjan> fizzie: it could be he's generating it from a time stamp, as well.
15:07:54 <oerjan> the main mezzacotta comic uses a date (although not the _present_ date)
15:08:05 <SgeoN1> Oh, it was just being slow
15:08:11 <fizzie> Seeding from the current second is rather popular too.
15:08:43 <fizzie> I was wondering if my ten-second sampling interval is relevant. Or actually it's "fetch, delay of at least 10 seconds, fetch, delay ...", so the rate is in fact something close to 11 seconds.
15:09:07 <fizzie> Or maybe not close to 11, but over 10, anyway.
15:09:32 <SgeoN1> Not the present date?
15:09:44 <SgeoN1> Even for the present day's comic?
15:12:22 <fizzie> I'd say it's always just the date you specify, which for the present day's comic happens to coincide with the present date.
15:15:05 <fizzie> Anyway, here's the counts graphically: http://users.ics.tkk.fi/htkallas/mezza.png -- make of that what you will. But it doesn't quite look two-tiered.
15:17:58 -!- Sgeo has joined.
15:18:02 <Sgeo> I think
15:18:12 <Sgeo> I'm not going to rest this laptop on my lap for a while
15:19:45 -!- pikhq_ has joined.
15:20:18 -!- pikhq has quit (Ping timeout: 260 seconds).
15:20:33 <oerjan> fizzie: if you remove both the two largest items, what then?
15:22:17 <oerjan> > 9*6 + 7 + 8
15:22:17 * oerjan stares sternly at lambdabot
15:22:18 <lambdabot> 69
15:23:15 <oerjan> dmm wouldn't choose that number, would he.
15:25:07 <fizzie> What's that number about?
15:25:33 <fizzie> Oh, right.
15:25:51 <oerjan> apparently fizzie wouldn't either.
15:26:15 <fizzie> No, I mean, I know the result, I was just wondering about the 9*6+7+8 briefly.
15:26:54 <oerjan> i was guessing what if the two top numbers are special
15:27:15 <oerjan> then that seemed like a close approximation
15:27:36 -!- augur has quit (Remote host closed the connection).
15:28:08 <fizzie> Right, right. Though I'm not sure if my current results match that terribly well. Though the smaller the differences, the harder it is to say.
15:29:05 <oerjan> statistically there should be _some_ outliers
15:32:42 <fizzie> p-values for selecting the sets 1..11, 2..11, 3..11, 4..11 and 4..10, respectively: 7*10^-5, 0.0129, 0.4662, 0.8690, 0.9666.
15:33:01 <fizzie> So, indeed, after discarding the top two it's not distinguishable from uniform.
15:33:25 <fizzie> (And the 4..10 set is quite a lot more uniform-ish that you'd expect.)
15:33:37 <oerjan> heh
15:34:29 <fizzie> It's done 8861 samples now, so it'll be finished this evening.
15:34:45 <fizzie> Three more hours or so.
15:38:56 <fizzie> Uh, posthumous s/-5/-15/ too.
15:47:42 -!- oerjan has quit (Quit: Later).
15:48:07 -!- SgeoN1 has quit (Quit: Bye).
15:54:37 <Phantom_Hoover> helo
15:54:37 <lambdabot> Phantom_Hoover: You have 5 new messages. '/msg lambdabot @messages' to read them.
16:03:25 -!- cheater_ has quit (Ping timeout: 252 seconds).
16:10:05 -!- augur has joined.
16:16:54 -!- cswords has quit (Ping timeout: 276 seconds).
16:22:32 <Phantom_Hoover> fungot!
16:22:33 <fungot> Phantom_Hoover: there's a defined format for data:// is reasonable. i concede that there may be
16:59:23 -!- myndzi has quit (Ping timeout: 245 seconds).
17:01:43 -!- ais523 has joined.
17:02:16 -!- myndzi has joined.
17:13:55 <Gregor> Foop
17:14:56 <ais523> Gregor: is that the opposite of a sudden vanish?
17:16:36 <Gregor> Indeed!
17:49:57 -!- ais523_ has joined.
17:51:58 -!- ais523 has quit (Ping timeout: 240 seconds).
17:52:52 -!- ais523_ has changed nick to ais523.
18:01:08 -!- Taneb has joined.
18:01:46 <Taneb> Hello!
18:01:46 <lambdabot> Taneb: You have 3 new messages. '/msg lambdabot @messages' to read them.
18:01:53 <Taneb> @clear
18:01:53 <lambdabot> Messages cleared.
18:02:15 <Taneb> I've been thinking about Haskell Obfuscation
18:02:22 <Taneb> :t null . guard
18:02:23 <lambdabot> Bool -> Bool
18:02:33 <Taneb> > null $ guard True
18:02:34 <lambdabot> False
18:02:37 <Taneb> > null $ guard False
18:02:38 <lambdabot> True
18:02:51 <Taneb> Less clear version of "not"
18:03:43 <ion> :-)
18:04:37 <Taneb> If you need that
18:05:18 <Taneb> Also, point-free style seems to have a lot of point symbols
18:05:56 <Taneb> @pl \x y z -> y x (z x)
18:05:56 <lambdabot> ap (flip . ((.) .) . flip id) (flip id)
18:06:02 <Taneb> 4 in that
18:06:30 <Taneb> Also, I don't have a decent IRC client on this computer
18:08:41 <fizzie> Those are just dots; and note that it's full of nice round ()s instead of pointy ->s. Not that it's relevant, but still.
18:14:48 -!- Vorpal has quit (Ping timeout: 245 seconds).
18:54:38 -!- pikhq_ has quit (Ping timeout: 240 seconds).
18:54:55 -!- pikhq has joined.
19:00:52 <fizzie> @tell oerjan Final counts: http://sprunge.us/CejV
19:00:53 <lambdabot> Consider it noted.
19:15:18 -!- Taneb has quit (Ping timeout: 240 seconds).
19:24:51 -!- monqy has joined.
19:39:37 -!- pikhq has quit (Read error: Operation timed out).
19:40:32 -!- pikhq has joined.
20:04:11 -!- Ngevd has joined.
20:14:23 -!- GreaseMonkey has joined.
20:14:24 -!- GreaseMonkey has quit (Changing host).
20:14:24 -!- GreaseMonkey has joined.
20:21:48 <Ngevd> :)
20:29:54 <itidus21> so.. it occurs to me that ideas of generalizations only hold true alongside things like platonic ideals
20:30:16 -!- KingOfKarlsruhe has joined.
20:31:46 <itidus21> on that note it has never been true i think that humans need anything more than a simple roof over their heads, a bit of food, clean water, exercize, and plenty of medicine in order to achieve maximum potential lifespan
20:32:20 <itidus21> and why on that note? well i mean humans do not need plato for anything but medicine
20:35:46 <Ngevd> Humans need those things to survive; to live, though?
20:37:27 <itidus21> i mean i dont think plato is vital for medicine but the idea of looking at things as platonic solids is fairly necessary probably for the weird science stuff which enables the creation of medicine i am guessing
20:37:47 <itidus21> Ngevd: oh yeah.. hmm you're right
20:39:15 <itidus21> Ngevd: so if someone is walking down the street with a gun i couldn't give him the benefit of the doubt that he was just going to hunt some boars... but i would be very tempted to give him that benefit of the doubt
20:40:41 <Ngevd> itidus21, do you live in an area where boars exist and can be hunted legally?
20:42:14 <Ngevd> But still look at him very carefully and make careful note of his appearance
20:45:16 -!- GreaseMonkey has quit (Quit: The Other Game).
20:57:36 -!- azaq23 has joined.
20:57:42 -!- oerjan has joined.
21:00:08 -!- oerjan has set topic: {M[m(_o)O!"elliott sacked as bearer of Element of Loyalty, seeking pegasus replacement | http://codu.org/logs/_esoteric/ | Now slightly on-topic | Now failing to construct an esolang in THE. WORST. POSSIBLE. WAY."(_o)o.?]}.
21:01:26 <oerjan> ais523: heaps of spam
21:01:27 <lambdabot> oerjan: You have 1 new message. '/msg lambdabot @messages' to read it.
21:01:38 <ais523> oerjan: I know
21:02:10 <ais523> when it's going this fast I prefer to take care of it in batches, it's faster
21:02:32 <ais523> [[LifeScript]] is nonspam, right?
21:04:59 -!- kwertii has joined.
21:05:19 <oerjan> ais523: i'd say :P
21:05:50 <oerjan> in fact, unusually much non-spam today, too
21:05:52 <oerjan> sadly not as much as the spam
21:06:05 <quintopia> ais523: is there a standard way of repeatedly deleting values from a list comprehension without adding a constraint for every value deleted?
21:06:24 <ais523> quintopia: err, I fear I'm missing context
21:06:39 <oerjan> which language, for a start
21:06:52 <ais523> it's like, I understand the individual words, and phrases like "list comprehension"
21:06:58 <ais523> but can't fit them together into a sentence
21:07:01 * oerjan would be most helpful if it's haskell, probably
21:07:36 <quintopia> i'm speaking purely abstractly
21:07:44 <oerjan> > [x | x <- [1..], not (x `mod` 3 == 0)]
21:07:45 <lambdabot> [1,2,4,5,7,8,10,11,13,14,16,17,19,20,22,23,25,26,28,29,31,32,34,35,37,38,40...
21:07:50 <oerjan> like this?
21:07:51 <quintopia> just wondering if *any* language has a way of doing it
21:08:37 <quintopia> oerjan: yes, but what if i wanted to just eliminate the values returned by my random number generator, but didn't want the comprehension description to grow linearly with number of values eliminated?
21:09:29 <Ngevd> Could you first zip [1..] and use that?
21:10:09 <ais523> quintopia: isn't that effectively asking "how do I get a list of all integers /not/ in a given lazy list"?
21:10:13 -!- nooga has quit (Ping timeout: 245 seconds).
21:10:16 <ais523> I don't think that's mathematically possible
21:10:24 <oerjan> i am not sure "comprehension description" is the term you really want to use, here
21:11:16 <oerjan> indeed, unless the list of things to remove is either finite or predictably ordered, you may _never_ be able to conclude that a value is in the final list
21:11:46 <quintopia> another example: let's say i have a 100x100 binary matrix, and i set the bits on the main diagonal. a dense matrix requires 10000 bits to represent this. a (standard) sparse matrix requires 1400 bits. but it could be done with a single expression like "set if x=y". is there a well-known algorithm that heuristically encodes things this way?
21:12:06 <quintopia> i know compression in general is AI-hard, but i thought maybe there might be an incomplete standard solution
21:12:16 <oerjan> > [1..] \\ [2, 4 .. 10]
21:12:17 <oerjan> argh
21:12:17 <oerjan> @ping
21:12:17 <lambdabot> pong
21:12:17 <lambdabot> [1,3,5,7,9,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,3...
21:12:45 <ais523> \\ is list difference?
21:12:48 <oerjan> yes
21:13:00 <ais523> does it require a consistent sort order between the lists?
21:13:04 <oerjan> but it won't return anything unless the second argument is finite
21:13:07 <quintopia> oerjan: you can assume the number of excluded values is "n"
21:13:12 <ais523> ah, requires finite second list
21:13:12 <quintopia> a finite value
21:13:29 <quintopia> i like the way you did it there
21:13:32 <oerjan> also i think it only removes one instance if there are duplicates
21:14:27 <oerjan> which on the positive side means it should have no overhead for the remainder of the list after all the subtracted elements have been found
21:15:16 <oerjan> quintopia: for your "another example", haskell does not remove from you the obligation to choose efficient data structures for representing things.
21:15:24 <quintopia> what if i start with a lazy list like that, and i want to write a function delete_with_minimal_overhead(list,value)?
21:15:48 <oerjan> > delete 5 [1..]
21:15:49 <lambdabot> [1,2,3,4,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,2...
21:15:49 <quintopia> what would an efficient data structure for that be?
21:16:33 <oerjan> um you've already said "if i start with a lazy list", which sort of ties your hands a bit
21:17:07 <quintopia> for instance, i start calling it with (list,2),(list,4),(list,6),(list,8). It would be nice if it figured out that a bunch of even numbers had been deleted, and compressed {2,4,6,8,10,12,..} to "even numbers in this range"
21:18:00 <quintopia> i don't mean a particular representation of a list comprehension. i mean "whatever data structure that works like one but also does this"
21:19:05 <quintopia> maybe it just hasn't been done. i have to go :(
21:19:22 <oerjan> quintopia: well haskell doesn't do it, anyhow.
21:19:42 -!- pikhq_ has joined.
21:19:47 -!- pikhq has quit (Ping timeout: 252 seconds).
21:21:56 <oerjan> ais523: ok LifeScript isn't spam, but i'm not convinced it isn't trolling :D
21:22:58 <ais523> heh
21:23:26 <oerjan> ais523: ok scratch some of that about unusually much non-spam, some of the edits that looked like they could be genuine were spams anyhow
21:39:28 <Ngevd> Etymology is sometimes interesting
21:39:36 <Ngevd> Scotland means "land of the Irish"
21:43:12 <quintopia> wat
21:43:24 <quintopia> so..."scots-irish" means "irish irish"?
21:55:19 <Ngevd> Ish
21:55:39 <Ngevd> The Romans called the Irish "Scotii"
21:55:53 <Ngevd> Which was then used to mean Gaelic
21:56:28 <Ngevd> And as Gaels (?) from Ireland went to Scotland (or "Caledonia" or "Pictland"), some stuff happened, and we're in a mess now
21:57:55 <oerjan> we can only assume the picts got a raw deal.
21:58:16 -!- Ngevd has quit (Quit: Goodnight).
21:58:44 <oerjan> clearly this is a matter of which Ngevd does not wish to speak.
22:01:18 <fizzie> Picts or didn't happen, like they say.
22:06:45 <fizzie> Bah, can't decide whether to make that biased-randomness post since it'd mean having to admit to wasting 34.65 MiB's worth of bandwidth.
22:07:12 <fizzie> Also incidentally it's the "Wow! signal" one that appears "too" often. Coincidence?
22:07:23 -!- nooga has joined.
22:07:57 <oerjan> just make a donation to the Jane Goodall Institute mentioning mezzacotta first.
22:08:32 <oerjan> or something like that, it's where they ask people to donate instead of to them.
22:09:13 * oerjan so reasonable suggestions
22:16:30 <oerjan> apparently Scott Aaronson is a betting guy
22:16:45 <oerjan> http://rjlipton.wordpress.com/2012/01/30/perpetual-motion-of-the-21st-century/#comment-17911
22:17:08 <oerjan> QC = quantum computing
22:19:53 <oerjan> (he did a similar bet against the P!=NP proof that was floating around a year and a half ago)
22:25:19 <oerjan> oops, my window with top -d 30 in it has died
22:26:22 <oerjan> that was the first disconnect since i started putting that up. thus fails my theory that the problem had either stopped permanently, or that the windows somehow kept each other alive.
22:27:18 <oerjan> ooh wait
22:27:39 <oerjan> false alarm. it had just died last time i turned off my computer XD
22:29:20 * oerjan usually hibernates when possible
22:54:17 -!- augur has quit (Remote host closed the connection).
23:02:32 <kallisti> > map unwords $ (`replicateM` ["player", "hater"]) =<< [1..]
23:02:33 <lambdabot> ["player","hater","player player","player hater","hater player","hater hate...
23:03:05 <monqy> hi
23:12:40 -!- ais523 has quit (Remote host closed the connection).
23:22:28 -!- azaq23 has quit (Ping timeout: 248 seconds).
23:23:33 <kallisti> monqy: are you a player hater hater or a hater player hater or a hater player player?
23:23:43 <kallisti> or just a hater hater hater
23:24:13 <monqy> I don't
23:24:14 <monqy> know???
23:24:39 -!- azaq23 has joined.
23:24:51 -!- azaq23 has quit (Max SendQ exceeded).
23:25:56 <kallisti> not cool
23:27:53 -!- derdon has quit (Remote host closed the connection).
23:31:18 -!- Phantom_Hoover has quit (Ping timeout: 240 seconds).
23:45:08 -!- Phantom_Hoover has joined.
23:48:25 -!- KingOfKarlsruhe has quit (Quit: ChatZilla 0.9.88 [Firefox 10.0/20120129021758]).
23:51:33 -!- cheater has joined.
23:51:38 -!- azaq23 has joined.
23:54:00 -!- pikhq has joined.
23:54:15 -!- pikhq_ has quit (Read error: Operation timed out).
23:55:16 -!- augur has joined.
23:59:11 -!- Phantom_Hoover has quit (Quit: Leaving).
←2012-02-01 2012-02-02 2012-02-03→ ↑2012 ↑all