←2013-08-07 2013-08-08 2013-08-09→ ↑2013 ↑all
00:00:02 <shachaf> Yes, "Word" is like "Int" except unsigned.
00:00:18 <yorick> oh, I didn't know that. in that case, weird.
00:00:50 <ion> λ> unsafeCoerce (256 :: Word16) :: Word8
00:00:52 <ion> 256
00:01:34 <ion> It’s almost as if unsafeCoerce is somehow unsafe.
00:01:42 <Bike> http://www.youtube.com/watch?v=SVoI3hB4IiM the way everything is recorded now is really weird O_o
00:02:03 <elliott> is it using the wrong instance dictionary or something?
00:03:10 <Bike> > -5 :: Word8
00:03:11 <lambdabot> 251
00:03:17 <shachaf> ion: That one is a bit less surprising.
00:03:21 <Fiora> Bike: @_@ are those like, compressed gas canisters?
00:03:21 <Bike> ok.
00:03:28 <ion> shachaf: I see what you did there.
00:03:28 <Bike> Fiora: yep
00:03:44 <Bike> Fiora: tagged 'lab safety'
00:04:01 <elliott> Bike: there's a thing where russian cars all have recorders in them for insurance stuff or something?
00:04:09 <yorick> Bike: it's nice how it's going "oooeeeoooeoeo" while everything is blowing up
00:04:14 <elliott> at least that's what I keep hearing when people ask why there are so many videos of russian traffic accidents
00:04:22 <yorick> elliott: I think it's called dashcam and yes that is the case
00:04:29 <Bike> elliott: yeah
00:04:35 <shachaf> elliott: is that "insurance" by the youtube mafia
00:04:53 <Bike> yorick: personally i like how the second half of the video is the driver backing up their car as far as they can
00:05:07 <kmc> oh god russian dashcams
00:05:07 <yorick> Bike: while the actual gas tanker is parking in front of it :P
00:05:11 <kmc> you know why this is a thing?
00:05:11 <elliott> love the soundtrack
00:05:15 <yorick> kmc: fraud
00:05:20 -!- |_ has quit (Remote host closed the connection).
00:05:25 <kmc> yeah, insurance fraud / scams
00:05:33 <ion> elliott: There are claims i haven’t verified that you don’t get insurance moneyz if you didn’t get a police officer to make a report about the accident, and which participant’s bribe is greater determines the content of the report if you don’t have evidence.
00:05:43 <Bike> elliott: stop drop and DANCE
00:06:07 <Fiora> Bike: http://www.youtube.com/watch?v=hkIZR-CE3iY summary of the last half of the video
00:06:15 <Bike> this is now my second favorite russian crash video
00:06:20 <Fiora> second?
00:06:34 <Bike> the favorite being the one where the truck falls over, releasing a bunch of cows, which stand up and moo at everything and wander around
00:07:03 <Bike> http://www.youtube.com/watch?v=n7CNR2bAh3Y
00:07:19 -!- |_ has joined.
00:07:49 <yorick> Bike: it's weird how they're all pretty much fine
00:07:55 <yorick> I mean they slide across the road quite a bit
00:08:07 <Bike> well they are cows.
00:08:36 <Bike> it's hard to hurt a cow. they're really quite solid.
00:09:06 -!- sacje has quit (Ping timeout: 264 seconds).
00:09:16 <ion> They’re partially gas, though.
00:09:28 -!- madbr has joined.
00:09:31 <oerjan> <shachaf> oerjan: is it a thing about ss(ssss......ssss)sss? <-- how should i know
00:09:57 <Fiora> Bike: geez
00:10:27 <elliott> I like how the cows only seem bothered for a second
00:10:32 <elliott> and then they're just like welp time to start walking
00:10:36 <madbr> hmm, I think I figured out how to do a cpu that does 1 opcode per cycle but has complex opcodes like add d2, [a0 + d0]
00:11:23 <madbr> (provided you have at least an instruction cache, even a really simple one like the 68020's 256 byte cache)
00:12:27 <madbr> separate address registers from data registers like on the 68000, and have address calculations use address registers only, or optionally an address register plus d0 (first data register)
00:12:43 <madbr> or possibly a choice of d0 or d1
00:12:50 <kmc> Bike: holy fuck
00:13:04 <Bike> at which one
00:13:07 <kmc> cows
00:13:10 <Fiora> cows.
00:13:18 <Bike> don't fuck with cows man
00:13:27 <kmc> i learned that the hard way
00:13:32 <Bike> used to be that every year or so the neighbors' cows would get out and wander all over my neighborhood
00:13:38 <Bike> i got memories man. memories. bad memories
00:13:41 -!- augur has joined.
00:18:42 -!- |_ has quit (Remote host closed the connection).
00:18:57 -!- |_ has joined.
00:22:44 <shachaf> "It can be argued that the term ‘categorification’ should be reserved for vertical categorification, since we can use ‘oidification’ for the horizontal concept."
00:24:40 <kmc> categorifornication
00:24:49 -!- yorick has quit (Remote host closed the connection).
00:27:07 <augur> shachaf: :|
00:27:10 <augur> where is that from
00:28:13 <madbr> it's the first time in my life that I've seen the term "oidification"
00:29:08 <shachaf> madbr: It's what you do when you go from "group" to "groupoid". h
00:29:10 <shachaf> th
00:32:58 -!- NihilistDandy has quit (Quit: NihilistDandy).
00:35:57 <oerjan> which one of these is the monoid -> monad thing
00:36:24 <elliott> monoids are misnamed :(
00:36:35 <elliott> they should be called cats, and categories should be called catoids.
00:36:40 <madbr> shachaf : what happens when your categories are not really vertical or horizontal?
00:36:50 <elliott> alternatively mons, and categories would be monoids
00:37:06 <Bike> what's the etymology of "monoid" again
00:37:26 <oerjan> mon, oi dunno
00:37:34 <Bike> forced
00:37:36 <ion> bravo
00:38:03 <oerjan> Bike: the mon is jamaican hth
00:38:36 <ion> and the oi is british hth hand
00:40:12 <shachaf> bravoid
00:40:31 <ion> bravod
00:40:37 <ion> bravad
00:48:52 <elliott> Bike: http://math.stackexchange.com/questions/156952/why-the-terminology-monoid
00:48:55 <elliott> (nobody knows)
00:50:09 <Bike> fuck
00:50:22 <Bike> `pastelogs 156952
00:50:24 <HackEgo> No.
00:50:28 <Bike> ok
00:50:34 <Bike> (what)
00:50:39 <elliott> `cat bin/pastelogs
00:50:40 <HackEgo> ​#!/bin/bash \ cd /var/irclogs/_esoteric \ \ pasterandom() { \ if [ "$1" -gt 150 ]; then \ echo "No." \ exit \ fi \ for i in $(seq "$1"); do \ file=$(shuf -en 1 ????-??-??.txt) \ echo "$file:$(shuf -n 1 $file)" \ done | paste \ } \ \ if [ "$1" ]; then \ if expr "$1" + 0 >/dev/null 2>&1; then \
00:50:50 <elliott> try ()ing the number
00:51:00 <Bike> `pastelogs (156952)
00:51:13 <HackEgo> http://codu.org/projects/hackbot/fshg/index.cgi/raw-file/tip/paste/paste.14019
00:51:54 <Bike> knew i'd seen it before
00:52:37 <Bike> http://retractionwatch.wordpress.com/2013/08/02/thats-not-plagiarism-its-an-administrative-error/ retractionwatch is so ridic
01:02:09 <nooodl^> good hackego
01:05:51 -!- sacje has joined.
01:07:42 -!- augur has quit (Remote host closed the connection).
01:09:40 -!- augur has joined.
01:30:40 -!- |_ has quit.
01:34:20 -!- augur has quit (Remote host closed the connection).
01:38:36 <Fiora> Bike: http://retractionwatch.wordpress.com/category/by-author/diederik-stapel/ oh gosh, they always use th same picture
01:39:44 -!- Bike_ has joined.
01:39:46 -!- Bike has quit (Ping timeout: 246 seconds).
01:43:27 <Fiora> 18:38 < Fiora> Bike:
01:43:28 <Fiora> http://retractionwatch.wordpress.com/category/by-author/diederik-stapel/ oh gosh, they always use th same picture
01:44:01 <Bike_> well this is the guy who nearly got his phd taken away, which is apparently a thing that can be conceived of
01:44:04 <Bike_> Serious Business
01:44:05 -!- Bike_ has changed nick to Bike.
01:45:07 <Fiora> gosh it's like he's smirking at me
01:46:57 <Bike> http://beheco.oxfordjournals.org/content/24/4/819 evopsych is one hell of a drug
01:57:58 <doesthiswork> I should try to coauther with him
02:06:46 <Bike> http://bogost.com/blog/oauth_of_fealty.shtml one of the more colorful "fuck your api" posts i've seen
02:21:44 <doesthiswork> the api deserves
02:21:46 <doesthiswork> it
02:23:19 -!- augur has joined.
02:31:21 <kmc> rocking the eee pc 900 crapbook
02:32:04 <kmc> the kbd is small enuf that i can hold the sides and type with my thumbs
02:32:57 <Bike> hey is anyone in #python
02:33:23 <Bike> or is it logged somewhere
02:34:24 <kmc> nobody is in #python
02:34:29 <kmc> python does not exist
02:34:33 <Bike> k
02:34:37 <kmc> you imagined it during one of your episodes
02:34:44 <kmc> but also, why?
02:35:18 <Bike> because i'm curious if someone is spamming #python in the same way they're doing in #lisp
02:35:41 <elliott> imo quotes
02:35:55 <elliott> also I just bit my tongue aaaaaaaaaaaaaargh
02:35:56 <kmc> what are they spamming
02:36:00 <kmc> elliott: D:
02:36:21 <elliott> not even on a manageable spot like the front or something it's right at the back to the side ow ow ow
02:36:27 <Bike> it's not funny or anything, they just came in and linked their super cool new web thing nobody cares about, and claimed it was in lisp when somebody pointed out this was dumb
02:36:39 <elliott> link
02:36:43 <Bike> said it's based on the implementation from the 80s when pressed
02:36:45 <Bike> what, to the thing?
02:36:48 <elliott> yes.
02:36:55 <elliott> I have a ~hunch~
02:37:04 <Bike> wow am i like second-order spamming here.
02:37:08 <Bike> http://www.leapfm.com/
02:37:23 <elliott> hunch was wrong
02:37:32 <elliott> this uh
02:37:36 <elliott> looks like the hacker news software
02:37:47 <Bike> yeah weird right
02:37:48 <Bike> " Sure, I chose CMUCL because it works well with cloud development and various unix systems that I work with. Has a good native code compiler but most importantly it's freely redistributable with full source code"
02:37:52 <elliott> ok but the html source doesn't
02:37:55 <Bike> vaguest description possible
02:37:56 <elliott> so I guess it's not actually Arc
02:38:41 <elliott> does Bike have arcpinions
02:39:13 <Bike> i stopped paying attention to arc when graham said about it that he was glad to have finally gotten rid of the "extra parentheses" in cond
02:39:24 <Bike> because that's the most banal thing i can imagine.
02:39:56 <kmc> itt Lisp doesn't have syntax
02:40:03 <kmc> getting rid of those parentheses was a deep semantic change
02:40:20 <Bike> like god CL is a fucking heap but you're not going to fix it if you seriously think that's important
02:40:37 <Bike> basically imagine me saying "imo uuuuuuugh" for two minutes straight.
02:42:59 -!- Bike_ has joined.
02:45:02 -!- Bike has quit (Ping timeout: 240 seconds).
02:45:21 <Bike_> so unrelated
02:45:30 <Bike_> i tried to print some medical records
02:45:38 <Bike_> and an #esoteric log came out? is coming out
02:45:39 <Bike_> help.
02:45:50 <elliott> we'er you're family now, Bike_
02:47:25 <Bike_> oh god they're still coming
02:47:33 <kmc> https://twitter.com/zanu_pf/status/363977016086765568 stupid future
02:47:37 <Bike_> what the FUCK did i do
02:47:49 <elliott> we want you to know EVERYTHING, Bike_
02:48:04 <elliott> ten years of #esoteric. they're all yours now
02:48:09 <Bike_> oh man i got a vintage MEANWHILE IN R/BITCOIN here
02:48:12 <Bike_> gonna frame this
02:48:30 <elliott> kmc: marriage between people who want babies, like, to collect them. baby enthusiasts
02:48:43 <elliott> does Mugabe still exist
02:48:48 <Bike_> yeah
02:48:51 -!- Bike_ has changed nick to Bike.
02:49:00 <Bike> why the fuck is this happening.
02:49:21 <elliott> what I like now is that you have to shred them
02:49:27 <elliott> because otherwise what if someone finds your printed #esoteric logs
02:49:30 <elliott> how the hell do you explain that??
02:49:52 -!- NihilistDandy has joined.
02:50:02 <Bike> i think i'll put them in a binder and donate them to the local library
02:54:33 -!- Bike_ has joined.
02:54:42 -!- Bike has quit (Ping timeout: 264 seconds).
02:56:00 -!- Bike_ has changed nick to Bike.
02:56:43 <doesthiswork> kmc: getting rid of those parenthesis is basically the difference between an alist and a plist
02:56:46 <Bike> http://24.media.tumblr.com/26c6ef77e28f105a837df9a3af56d7af/tumblr_mr6ysuGwKn1r7tprao1_1280.jpg
02:56:55 <Bike> i'm seriously, i can't even comprehend what just happened to me
02:57:25 <coppro> you printed out a log?
02:57:48 <coppro> no wait
02:57:52 <coppro> that's actually irssi
02:57:53 <coppro> lol
02:57:55 <elliott> looks like
02:57:55 <elliott> yeah
02:57:58 <elliott> an actual terminal screencap
02:58:13 <Bike> from like, weeks ago though
02:58:14 <elliott> Bike: please frame this
02:58:18 <Bike> of course
02:58:30 <coppro> a friend once attempted to print something
02:58:35 <coppro> the printer decided to turn off postscript
02:58:40 <coppro> and just printed the raw source as a text file
02:59:17 <elliott> Bike: did you maybe press a weird key at 17:42 weeks ago
02:59:53 <Bike> the other logs end at different times
03:00:20 <Bike> but uh is there some way i can destroy my print history easily to make it stop
03:04:19 -!- CADD has joined.
03:07:32 <doesthiswork> shhh don't let CADD know we're here
03:09:02 <elliott> ???
03:10:13 <kmc> this is some horror movie shit
03:10:38 <kmc> wait until it starts printing logs from the future talking about vaguely specified catastrophe
03:12:43 <elliott> this is horror movie stuff
03:12:48 <Bike> < elliott> wait < elliott> what happened to shachaf < FreeFull> idris........... < elliott> FUCK[log cuts off]
03:13:08 <kmc> <itidus21> I AM BECOME DEATH, THE DESTROYER OF WORLDS
03:13:30 <doesthiswork> actually the conventional translation is time rather than death
03:13:43 <kmc> but not the famous one
03:13:45 <Bike> at the end of the channel is elliott, alone
03:13:49 <Bike> trapped in his matrix of solidity
03:13:51 <Bike> with real dimensions
03:14:35 -!- oerjan has quit (Quit: Bubblewrap matrix).
03:15:29 <doesthiswork> I thought it was pi dimesions
03:16:52 <Bike> pi is real. pi is all around us.
03:19:53 <shachaf> Bike: help
03:20:17 <Bike> sorry shachaf i future logged it
03:20:21 <Bike> in the future you're already dead
03:20:43 <shachaf> p. sure this channel has a no future logging policy
03:21:03 <Bike> if it did that would have to be indicated in the topic
03:21:05 <Bike> freenode rules
03:21:42 <shachaf> Bike: it sure does
03:21:57 <Bike> yes
03:22:22 <CADD> Bike: interesting about pi, it was almost legislated to be a finite fraction at one point.. imagine what would have happened to math if it passed?
03:22:51 <Bike> nothing, because everyone would ignore a dumbassed state law in the same way botanists don't give a fuck what the SCOTUS said tomatos were
03:23:19 <CADD> cant find the reference, but it was something like 237/74
03:24:07 <Bike> > 237/74 :: Float
03:24:08 <lambdabot> 3.2027028
03:24:30 <CADD> im pretty sure its not exacty that fraction. but it was something close
03:24:34 <CADD> exactly*
03:24:46 <Bike> i'm pretty sure that pi is exactly that fraction
03:24:58 <CADD> exactly
03:25:08 <CADD> well, i was just reading the wiki article on pi
03:25:17 <Bike> this reminds me that i don't actually have an intuitive idea of why pi is irrational
03:25:20 <Bike> sucks imo
03:25:23 <CADD> newton came up with an infinite sum of fractions that approaches pi
03:25:40 <CADD> im pretty sure the wiki article could shed some light on that
03:26:23 <madbr> irrationals are crazy
03:26:28 <CADD> ikr?
03:26:38 <madbr> transcentenals even more so
03:26:40 <Bike> ew, pun
03:26:44 <CADD> yup
03:26:44 <Bike> oh, you were serious
03:26:45 <CADD> lol
03:26:54 <CADD> i didnt even see that as a pun
03:26:54 <madbr> oh heh
03:26:57 <CADD> nice catch Bike
03:27:57 <madbr> between every two numbers there's an infinity of transcendentals that are uncalculatable
03:28:13 <CADD> yup
03:28:19 <CADD> surreal numbers are amazing
03:28:20 <madbr> even if you could do any infinite series or solve anything
03:28:23 <Bike> Between any two reals there is an infinite number of rationals.
03:28:54 <madbr> bike: yes but that's a countable infinite
03:28:59 <doesthiswork> unless you include infinitesimals
03:29:32 <madbr> even the infinitesimals are infinitely small compared to the reals
03:29:49 <madbr> oh wait
03:29:51 <madbr> nevermind
03:30:41 <madbr> most numbers you will never ever be able to reach
03:31:28 <Bike> Can you think of an uncomputable number that isn't the limit of a sequence of computable numbers?
03:32:00 <doesthiswork> yes w
03:32:22 <CADD> and w^2
03:32:51 <Fiora> BB(1000)?
03:33:24 <madbr> bike: has to exist
03:33:40 <madbr> sequences of computable numbers are countable
03:34:17 <doesthiswork> fiora: wouldn't any busy beaver be just as uncomputable any other?
03:34:19 <madbr> so numbers that you can reach by sequence are only an infinitely small part of the reals
03:34:28 <Fiora> doesthiswork: BB(4) is computed, though, I think?
03:34:57 <Fiora> like, the halting problem only says "some programs can't be proven to halt", not "all programs can't be proven to halt"
03:35:03 <shachaf> can Bike think of a number that even he could not compute?
03:35:12 <Bike> I meant an example.
03:35:15 <doesthiswork> fiora: got it
03:35:25 <Fiora> and like, the small busy beaver ones are pretty easy or something
03:35:41 <Bike> It's pretty easy to construct a computable sequence that has a busy beaver number as its limit.
03:35:58 <Bike> also they're all, like. natural numbers.
03:36:42 <Fiora> but what if BB(1000) is uncomputable?
03:36:52 <Fiora> like, if you don't know what it is, howcould you converge to it?
03:37:12 <Bike> Dovetail running every relevant Turing machine
03:37:15 <doesthiswork> bike: smallest number that is part of a nonterminating collatz sequence
03:37:36 <Bike> What makes you think that?
03:37:58 <Fiora> but like.... dovetailing turing machines can only find you a lower bound, right?
03:38:15 <Bike> A sequence of lower bounds.
03:38:46 -!- Nisstyre has quit (Quit: Leaving).
03:38:59 <Fiora> that makes sense... I guess...
03:39:11 <doesthiswork> untill you have proved that every one of those machines that hasn't halted, will not halt it isn't computed
03:39:24 <Bike> I asked for a sequence.
03:39:32 <madbr> hmm
03:39:39 <doesthiswork> I thought you asked for a number
03:39:56 <Bike> I asked for a number not the limit of a sequence.
03:39:58 <madbr> a number that can't be reached through sequence
03:40:00 <elliott> Fiora: but taking you the limit gets you a perfect lower bound
03:40:05 <elliott> since the lower bounds are constantly improving
03:40:16 <elliott> that's vague but. anyway.
03:40:25 <elliott> I bet Bike would explain it better.
03:40:32 -!- CADD has changed nick to CADD|away.
03:40:38 <madbr> not sure I can think of any
03:40:42 <madbr> it's pretty hard
03:41:50 <doesthiswork> busy beaver is pretty much the prototype of uncomputable
03:41:51 <madbr> infinite sums are out ofc, same for solutions to anything (since you can infinite-newton-raphson it)
03:42:00 <Bike> http://en.wikipedia.org/wiki/Specker_sequence Another nonexample.
03:42:30 <madbr> (though maybe a solution to something discontinuous and nonconverging)
03:43:30 <doesthiswork> transcendental numbers?
03:43:38 <Fiora> Bike: wow, that's an evil sequence
03:43:42 <Bike> It is.
03:44:17 <Bike> It also means the computable reals aren't compact, so, that kinda blows.
03:45:27 <madbr> bike: woa this is remarkably hard
03:45:46 <Bike> Isn't it?
03:46:03 <Bike> How about another question: An uncomputable function that is continuous.
03:46:26 <madbr> hm
03:46:35 <coppro> Bike: trivial.
03:46:42 <Bike> coppro: Yeah?
03:46:45 <doesthiswork> busy beaver with linear interpolation
03:46:46 <coppro> Bike: take the busy beaver function, connect the dots
03:46:51 <madbr> maybe just the fourier transform of an uncomputable series
03:47:26 <coppro> incidentally, the best proof I've been taught in a class was
03:47:35 <coppro> "Proof. Draw the graph on sufficiently fine grid paper."
03:48:09 <doesthiswork> any 3 points make a straight line give a wide enough pen
03:48:17 <Bike> How about differentiable?
03:48:31 <doesthiswork> the integral of the previous answer
03:48:43 <Bike> boring.
03:49:34 <coppro> doesthiswork: this was a serious proof
03:49:42 <coppro> I'll let you puzzle over what theorem
03:50:55 <doesthiswork> coppro: yes, I can imagine things that would be proven by that
04:20:57 <shachaf> "We write α ∶ F ≅ G, if α is a natural isomorphism. As an example, the identity is a natural isomorphism of type F ≅ F."
04:21:07 <shachaf> is this the worst example
04:21:32 <Bike> Maybe.
04:37:59 <Bike> http://arxiv.org/abs/1303.0371 So like, searching in constant time.
04:38:01 <Bike> What the fuck?
04:40:47 <lifthrasiir> in the classical algorithm?
04:40:55 <lifthrasiir> ah, quantum one.
04:41:23 <Bike> Not that i understand what "nonlinear quantum model" means here in a nontrivial way, but
04:41:31 <lifthrasiir> Bike: and it's not the constant time, it's rather like O(n^(1/4)).
04:42:11 <Bike> i read that but wikipedia summarized it as "constant" so i dunno
04:42:42 <lifthrasiir> it is a long known fact that the quantum computer can do unstructured searching in sublinear time (Grover's algorithm, O(n^(1/2))) so it is not that crazy one
04:42:56 <lifthrasiir> though I'm not aware of details
04:43:12 <Bike> well grover's is also crazy.
04:43:27 <lifthrasiir> every quantum algorithm is crazy ;)
04:43:32 <madbr> they figured how to make some part of it constant
04:44:02 <madbr> but some other part has grown in requirements so the overall growth is O(n^(1/4))
04:44:02 <lifthrasiir> it seems that Grover's algorithm is the optimal one of that kind though
04:44:12 <madbr> that's what I can read from what they're saying
04:44:45 <lifthrasiir> (the optimality of the number of unitary operators U_\omega?)
04:45:01 <lifthrasiir> the number of applications of unitary operator*
04:45:16 -!- carado has quit (Ping timeout: 246 seconds).
04:45:50 <Bike> lifthrasiir: yeah grover's is the fastest "linear"
04:46:02 <Bike> again i have no real idea what this means
04:59:08 -!- mnoqy has joined.
05:09:43 <shachaf> mnoqy: do you find that you check super mega less now that it has a regular update schedule
05:10:23 <mnoqy> i actually probably check it the same amount
05:10:34 <mnoqy> what if it updates out of schedule! what then
05:10:41 <mnoqy> i have to be on top of things like this
05:12:06 -!- Bike_ has joined.
05:12:16 <shachaf> mnoqy: JohnnySmash wouldn't do that would he
05:12:33 <mnoqy> but what if he did...........
05:13:11 <mnoqy> what if someone held johnnysmash at gunpoint to make comics on wednesdays too without telling anyone about it
05:13:29 -!- Bike has quit (Disconnected by services).
05:13:31 <shachaf> hang on that could actually happen
05:13:37 <shachaf> maybe i better check on wednesdays too
05:13:41 <shachaf> and on weekends
05:13:43 -!- Bike_ has changed nick to Bike.
05:13:53 <shachaf> and even mondays
05:14:02 <Bike> I assume johnnysmash is an intelligence agent, and Super Mega is coded messages.
05:18:53 -!- sacje has quit (Quit: sacje).
05:19:16 -!- CADD|away has quit (Read error: Connection reset by peer).
05:22:10 <oklopol> "<Bike> Can you think of an uncomputable number that isn't the limit of a sequence of computable numbers?" even the rationals (which are computable) are dense among reals, so there are no such numbers
05:23:30 <oklopol> "<Fiora> BB(1000)?" BB(1000) is a computable number, as it is an integer. this particular representation of it might be impossible to evaluate, but that is irrelevant.
05:24:16 <Bike> take all my fun why doncha
05:24:30 <madbr> oklopol : "dense" ?
05:25:08 <Bike> http://en.wikipedia.org/wiki/Dense_set doot
05:25:55 <comex> so just pack BB(n) in order into a real
05:25:57 <Bike> Since any real has a rational of any distance you want by it.
05:26:38 <madbr> mhm
05:26:48 <Bike> So for any real we can construct a sequence etc etc converges to etc etc etc and so on etc
05:27:09 <madbr> which means that... yeah if you can get an approximation
05:27:29 <Bike> computables are also dense, so, bla bla analogy etc left as an exercise etc bla blatc
05:27:30 <madbr> but that's a contradiction
05:27:35 <Bike> What?
05:27:52 <madbr> hmm
05:27:59 <madbr> or I might be wrong
05:28:23 <madbr> but I have this intuition that sequences approaching a given limit are countable
05:28:37 -!- copumpkin has quit (Ping timeout: 248 seconds).
05:28:42 <Bike> That the set of all sequences approaching a given limit is countable
05:28:43 <Bike> *?
05:29:06 -!- copumpkin has joined.
05:29:22 <madbr> hm
05:29:39 <madbr> the set of all countable sequences approaching a given limit
05:29:44 <madbr> uh
05:29:49 <Bike> Because that's not the case. Take any sequence approaching the limit that isn't bla bla, multiply it elementwise by some real, you get another sequence
05:29:51 <madbr> the set of all computable sequences approaching a given limit sorry
05:30:51 <elliott> nobody said the sequence had to be computable itself, though.
05:31:06 <Bike> admittedly i did intend that, but yeah.
05:32:57 <oklopol> by definition, they usually are
05:33:04 <oklopol> oh okay i was scrolled up
05:33:12 <comex> oh, duh
05:33:53 <madbr> because rationals are dense over reals, that meach every real must be reachable by successive iteration
05:34:42 <madbr> except for numbers involving non-computable parts, this means this successive iteration can be defined by a computer program
05:35:02 <madbr> computer programs are countable
05:35:39 <Bike> Computable numbers are countable, yes, but (countably) infinite sequences of members of a countable set are well, not.
05:36:14 -!- mnoqy has quit (Quit: hello).
05:36:36 <elliott> my proof: real numbers are iso to digit sequences, then think a lil
05:36:44 <oklopol> why did you say that the specker sequence is a nonexample?
05:37:00 <oklopol> well
05:37:09 <Bike> Because it's a sequence of computable numbers with an uncomputable limit?
05:37:42 <oklopol> i may have changed your question
05:37:59 <Bike> It wasn't a very good question.
05:39:02 <madbr> are there numbers that can't be computably successively approximated?
05:39:47 <Bike> well like oklopol said there's always a sequence of rationals or computables or whatever the hell.
05:40:14 <madbr> oh I see
05:40:14 <Bike> there are like, functions that aren't semicomputable, though? probably
05:40:18 <oklopol> if a single program needs to output the sequence though, this is more interesting
05:40:30 <Bike> yeah.
05:40:35 <madbr> I reread the specker sequence thing
05:41:19 <madbr> essentially some digits of the maximum of the sequence are uncomputable
05:41:44 <madbr> so the maximum of the sequence can't be successively approximated
05:41:49 <madbr> clever
05:49:14 <oklopol> if you do specker with a subset X of naturals, and you have a program that outputs a sequence of rationals that converges to it, then X is necessarily recursively enumerable, presumably by the proof in the specker article
05:49:57 <oklopol> so if you do specker with a set that is not recursively enumerable, then that number cannot be approximated by a computable sequence of rationals, in a somewhat interesting sense
05:52:00 <madbr> you sorta can't force it to converge
05:52:18 <madbr> interesting
05:52:42 <oklopol> well not sure i get your intuition there
05:53:14 <madbr> well, if you have a rule that you must always converge to at least distance N of the real number
05:53:26 <madbr> with N being divided by 2 on each step
05:53:41 <madbr> then a given step might loop infinitely of course
05:54:24 <oklopol> here, the property of a real a we are interested in is "is there a turing machine that, given n, outputs a rational r_n, such that r_1, r_2, ... converges to a"
05:55:13 <madbr> right
05:56:06 <oklopol> hmm
05:56:57 <oklopol> okay actually perhaps "is there a turing machine that, given n, outputs a rational r_n, such that r_1, r_2, ... converges to a, and |r_i - a| is decreasing"
05:57:17 <oklopol> because otherwise i don't know if X needs to be r.e.
05:57:49 <madbr> I was thinking of having the difference exponentially decreasing
05:58:30 <oklopol> well isn't the number computable then
05:58:43 <madbr> hm yes
05:59:22 <oklopol> i think mine is a weaker notion
06:00:30 <oklopol> there's a lot of literature on this stuff, but i haven't read any of it
06:01:44 <oklopol> actually on second thought i'm not sure even my latter definition works
06:02:07 <oklopol> or is that a third thought
06:03:48 <Bike> on nth thought, for some unknown n
06:07:06 <oklopol> ah okay i think r_i itself needs to be increasing: then you can tell n \in X when the nth digit flips to a 4, so X is r.e.
06:08:39 <elliott> oklopol: the definition I know is just forall m n, |r_m - r_n| <= m + n
06:08:44 <elliott> or am I talking nonsense
06:09:01 <oklopol> are you sure about the m+n?
06:09:08 <oklopol> you know less and less about the numbers?
06:09:24 -!- conehead has quit (Ping timeout: 256 seconds).
06:09:26 <elliott> well I looked it up in the paper I remember reading so as to look slightly less stupid.
06:09:43 <Bike> it's ok elliott, i'm here for you, to look stupidest.
06:09:49 <elliott> ok wait that wasn't the exact paper
06:09:51 <oklopol> elliott: i'm not trying to define computable numbers, i'm trying to define a larger class
06:09:55 <elliott> oh.
06:10:05 <elliott> thanks, now I know I'm talking nonsense
06:10:23 <Bike> The important question is, are these numbers Lyapunov stable
06:10:42 <oklopol> if in yours, you replace m+n by 1/(m+n), then that's computable numbers, since the nth approximation is at most 2/n away from the real
06:11:12 <oklopol> or i guess even 1/n away
06:11:35 <elliott> For example, I will represent the real numbers (see Definition 9.16 and
06:11:35 <elliott> Definition 11.1) by the type
06:11:36 <elliott> ∃f: Q+ ⇒ Q.∀ε1 ε2.|f(ε1) − f(ε2)| ≤ ε1 + ε2.
06:12:03 <elliott> I guess it's possible that roconnor made the exact same mistake in the definition in both his Ph.D. thesis and this other paper
06:12:15 <Bike> rookie mistake
06:12:33 <oklopol> are those epsilons natural numbers though?
06:12:34 <Bike> wait should i even ask how that's a type
06:12:36 <shachaf> Bike: did someone castle agaain
06:12:41 <elliott> it copied kind of badly
06:12:44 <Bike> yes
06:12:46 <elliott> oklopol: oh, I guess they're Q+ there.
06:12:49 <Bike> what's a castle
06:12:50 <elliott> so, sure.
06:12:55 <oklopol> looking at the type, that seems to work
06:13:10 <oklopol> idea is that given epsilon, f computes an epsilon approximation
06:13:18 <oklopol> for positive rational epsilon
06:13:22 <elliott> Bike: it's (f : Q+ -> Q, prf : forall e1 e2, abs (f e1 - f e2) <= e1 + e2)
06:13:34 <elliott> that ∃ is meant to be an \exists
06:13:44 <elliott> oh.
06:13:45 <elliott> it is.
06:13:49 <elliott> Windows just sucks.
06:15:15 <oklopol> i see them just fine
06:17:19 <madbr> I see...
06:17:26 <madbr> so it's like...
06:18:36 <oklopol> can i see the paper?
06:18:48 <madbr> - any real can be reached by the limit of an infinite series or rationals, since rationals are dense over reals
06:18:50 <oklopol> because i'm also wondering how that's a type
06:19:23 <madbr> - however that series can't converger exponentially or else the number would be computable (and thus, countable)
06:20:00 <Bike> i mean it's a type in that it defines a set of stuff but
06:20:31 <elliott> oklopol: http://r6.ca/thesis.rev.fluorine.pdf
06:20:32 <oklopol> it doesn't really define anything though, it's just a true statement about functions :P
06:20:37 <elliott> but I mean it's pretty standard?
06:20:40 <oklopol> but there might be something implicit going on
06:20:43 <elliott> it's a dependent tuple with a function and a proof object
06:20:47 <madbr> - which means that 'most' reals must be approached by a series that you can't determine if it converges or not
06:20:55 <elliott> where the proof object is a dependent function from two Q+s to a proof obejct that blah blah blah.
06:21:16 <madbr> (or at least doesn't converge exponentially)
06:21:48 <oklopol> madbr: the second one is slightly wrong, the series can converge as fast as you like... it's the enumerability by a turing machine that's important
06:22:19 <madbr> hm, true
06:22:45 <madbr> so it can't... converge exponentially in an enumerable way
06:23:21 -!- Bike_ has joined.
06:23:37 <oklopol> yes, or otherwise the number is computable (and thus, belongs to the countable set of computable numbers)
06:24:24 <madbr> so either it must converge exponentially in a non computable/enumerable way
06:24:38 -!- Bike has quit (Ping timeout: 240 seconds).
06:25:20 <madbr> or if you force your successive steps to be computable/enumerable, then it can't converge exponentially anymore?
06:25:30 -!- Bike_ has changed nick to Bike.
06:25:31 <oklopol> ah okay if you have an existentially quantified formula, then it's the type of values the existential quantifier can take
06:26:40 -!- CADD has joined.
06:27:02 <oklopol> madbr: something like that i guess.
06:28:28 <CADD> sooo... why arent there more people talking about egison?: http://egison.pira.jp/
06:28:35 <elliott> oklopol: well uh, you have to bundle the proof object.
06:28:57 <elliott> a value of the type would look like (f, \e1 e2 -> ...proof of the <=...)
06:29:03 <oklopol> yeah but i consider that part of the logic we're working in
06:29:04 <elliott> given appropriate constructors for <= or whatever on Q.
06:29:10 <elliott> well, sure
06:29:10 -!- madbr has quit (Quit: Radiateur).
06:29:30 <elliott> but it isn't really like that in type theory, which is the point of the thesis.
06:29:37 <oklopol> okay
06:29:41 <elliott> like, proofs of (m <= n) aren't necessarily unique depending on how you define <=.
06:29:47 <elliott> so the proof term can be relevant.
06:30:08 <Bike> Algorithms are expressed intuitively and will be auto-parallelized with Egison!!
06:32:05 <oklopol> oklotalk-- did pattern matching of objects without normals forms too
06:32:24 <oklopol> Egison is the world's first programming language that makes pattern-matching with unfree data types practical.
06:32:40 <oklopol> okay perhaps that's different
06:32:43 <Bike> so, we can conclude that oklotalk-- wasn't real and youjust hallucinated
06:32:59 <elliott> I remember oklotalk--
06:33:03 <oklopol> i wonder what's "practical"
06:33:05 <Bike> no you don't
06:33:20 <elliott> oklopol: remind me how oklotalk-- was great. I remember it being great
06:33:22 <elliott> @karma oklotalk
06:33:22 <lambdabot> oklotalk has a karma of -4
06:33:35 <oklopol> well it was great for example because it mixed static and dynamic scoping
06:33:48 <Bike> wow great.
06:33:55 <oklopol> every variable was first checked in the static scope, then the dynamic scope
06:34:01 <oklopol> how great is that? very.
06:34:42 <oklopol> also pattern matching was pretty advanced
06:34:46 <oklopol> dunno if there was anything else :P
06:35:10 <oklopol> gotta go see a doctor now.
06:35:18 <elliott> oklopol: don't die :(
06:35:39 <Bike> advanced and practical
06:35:47 <oklopol> i'll try my best
06:45:09 -!- Bike has quit (Quit: leaving).
06:55:20 -!- btiffin has joined.
07:34:58 -!- epicmonkey has joined.
07:37:15 -!- Taneb has joined.
07:38:04 -!- MindlessDrone has joined.
07:53:36 <Taneb> Aaaaaaaaaaaaaah
07:53:43 <Taneb> I can watch Airplane WHENEVER I WANT TO
07:57:17 -!- epicmonkey has quit (Read error: Operation timed out).
07:57:37 <Taneb> No I can't
08:11:06 -!- epicmonkey has joined.
08:16:05 -!- epicmonkey has quit (Ping timeout: 248 seconds).
08:16:06 <Taneb> In other news, my copy of To Be Or Not To Be: That Is The Adventure has shipped!
08:16:28 -!- doesthiswork has quit (Quit: Leaving.).
08:50:52 -!- Taneb has quit (Ping timeout: 264 seconds).
09:04:38 -!- btiffin has left.
09:07:00 -!- Taneb has joined.
09:18:03 -!- katla has joined.
09:26:47 -!- epicmonkey has joined.
09:34:10 -!- Guest18414 has quit (Ping timeout: 245 seconds).
09:34:28 -!- Guest18414 has joined.
09:35:52 -!- epicmonkey has quit (Ping timeout: 264 seconds).
09:47:50 -!- epicmonkey has joined.
10:05:00 -!- aloril has quit (Ping timeout: 245 seconds).
10:06:15 -!- Sgeo has quit (Read error: Connection reset by peer).
10:07:50 -!- Sgeo has joined.
10:18:16 -!- aloril has joined.
11:22:12 <Taneb> Well, that's the best I've ever done at Brogue
11:24:07 <elliott> yay you're playing brogue
11:24:13 <elliott> are you playing the latest version? it is kind of bad
11:24:16 <elliott> you mightw ant to try the previous one
11:25:35 <Taneb> I think I am playing the latest version
11:25:37 <Taneb> It's a different version to that which I was previously playing
11:25:51 -!- CADD has quit (Quit: Lost terminal).
11:26:03 <elliott> it has a kind of terrible food clock with some serious issues (food basically only ever spawns on certain fixed levels due to some bad design and food is really tight so it's easy to starve because of this)
11:26:08 <elliott> if you do starve maybe roll back a version :p
11:26:15 <Taneb> I haven't been starving
11:26:28 <Taneb> So maybe I am not on that version
11:29:42 <Taneb> Huh, I am on the latest version
11:29:57 <Taneb> Maybe I just suck so much that it hasn't became an issue
11:30:06 <elliott> what did you die to?
11:30:08 <Taneb> Centaur
11:30:11 <Taneb> Level 19
11:30:16 <elliott> ahh centaurs
11:30:19 <elliott> were you a melee dude
11:30:23 <Taneb> Yeah
11:30:25 <elliott> okay well
11:30:28 <elliott> here is my centaur tip
11:30:35 <elliott> walk through a door and wait for it to come for you
11:30:44 <elliott> this is like the only way to kill centaurs as a melee character without wanting to kill yourself
11:30:56 <elliott> it's kind of silly that you have to use bad AI in this way but it's so intolerable doing anything else
11:31:00 <Taneb> I had a +11 rapier of multiplicity and it was awesome
11:31:06 <elliott> wow +11
11:31:10 <elliott> did you just sink every enchant into it
11:31:17 <Taneb> p. much
11:31:24 <Taneb> When I found it it was +4
11:31:54 <Taneb> I would hit something and I'd have an army of floaty blades
11:40:16 -!- nooodl has joined.
11:41:02 <Jafet> Enchantments go to 11?
11:45:53 <Taneb> At the very least!
11:46:40 <Jafet> That sounds almost as unbalanced as nethack
11:47:32 -!- Sgeo has quit (Read error: Connection reset by peer).
11:47:40 <Taneb> Scrolls of enchantments aren't that easy to come across
11:47:53 <fizzie> Enchantments go to +127, presumably.
11:48:42 <Jafet> Not 2147483647?
11:49:17 <nooodl> in nethack it's +127
11:49:51 <nooodl> however this also applies to your to-hit bonus, which is also a signed char, or something, and it'll overflow and you'll have -118 to-hit
11:50:07 <elliott> brogue is roughly infinitely more balanced than nethack
11:50:16 <fizzie> But think of all the sexy foocubi you'll get by showing off your +127 whatever.
11:50:27 <Roujo> `? foocubi
11:50:29 <HackEgo> foocubi? ¯\(°_o)/¯
11:50:29 <myndzi> |
11:50:29 <myndzi> º¯`\o
11:50:35 <fizzie> ^style nethack
11:50:35 <fungot> Selected style: nethack (NetHack 3.4.3 data.base, rumors.tru, rumors.fal)
11:50:39 <fizzie> fungot: Tell me about foocubi.
11:50:39 <fungot> fizzie: tiger: 1. a check of my attire showed no obvious holes or damage. i once saw a hacker named beaver. weak and ailing, they have eight legs and filling his lungs with air.
11:50:40 <Jafet> You cannot enchant meat sticks
11:50:42 <elliott> I have a +127 whatever, if you know what I etc. etc. etc.
11:50:55 <fizzie> fungot: That doesn't sound terribly sexy.
11:50:55 <fungot> fizzie: pyrolisk: at first. he appreciated the priceless wine that much that he is not surprisingly, are large, elephantlike mammal of the damned to hell.
11:51:01 <Roujo> I should try nethack sometime
11:51:08 * Roujo dodges the dirty looks
11:51:25 <Roujo> Anything I should know?
11:51:33 <Roujo> Before I download it and start?
11:51:48 <fizzie> Play it on one of the servers instead, they're fancy.
11:51:50 <nooodl> "play brogue instead"
11:51:53 <Roujo> And die horribly from a goblin with a wand of death or some other stuff?
11:52:07 <Roujo> Nethack... servers?
11:52:07 <Jafet> Play a game that has been developed in the last decade
11:52:15 <fizzie> You get your embarrassing deaths typically announced on an IRC channel.
11:52:40 <Taneb> Doesn't nethack have like 1000 ways to die or something
11:52:51 <nooodl> Roujo: `telnet nethack.alt.org`
11:53:10 <fizzie> Taneb: NAO's "top types of death" list has 16697 unique entries.
11:53:17 <nooodl> Taneb: killed by a goblin called A, killed by a goblin called B, killed by a goblin called...
11:54:07 <Taneb> :P
11:54:15 <fizzie> Taneb: Granted, it's probably kind of a Zipfian distribution. But even the rank-1000 entry ("killed by a fire ant, while frozen by a monster's gaze") has happened 55 times, so it's not quite unique.
11:54:29 <Taneb> I think I'll stick with Pokemon Mystery Dungeon Red Rescue Team
11:54:33 <Jafet> http://alt.org/nethack/player-endings.php?player=DeathRobin
11:54:37 <fizzie> Entries 6048..16697 all have a count of 1, though.
11:55:01 <Jafet> Hmm, there have been 101 trickeries.
11:55:28 <fizzie> "killed by a gnome lord called 104" thanks, saiph.
11:55:34 <Roujo> Jafet: Well, the goal here is to actually play nethack, so playing something else instead doesn't quite fit the bill =P
11:55:43 <Roujo> nooodl: Alright, I'll telnet there.
11:55:48 <Roujo> I'll telnet SO HARD
11:55:59 <fizzie> There's quite a few entries where people have been frustrated.
11:56:03 <nooodl> Roujo: if you really wanna play nethack maybe play acehack instead
11:56:06 <fizzie> E.g. "killed by a pony called KILL ME I FAIL AT ILLITERACY".
11:56:07 <nooodl> which is like
11:56:13 <nooodl> nethack without the terrible terrible ui
11:56:18 <Roujo> That might be nice
11:56:24 <Taneb> "Was killed by an Oddish using Tackle while paralyzed"
11:56:29 <Roujo> I tried to play DF without a graphics pack
11:56:31 <fizzie> Or "killed by a rock troll called OH FUCK YOU TROLL SERIOUSLY".
11:56:34 <Roujo> Didn't really work out
11:56:39 <Taneb> Roujo, I've never played DF with a graphics pack
11:56:48 -!- dessos has quit (Ping timeout: 245 seconds).
11:56:52 <Roujo> Taneb: Then you are a greater person than I am
11:56:53 * Roujo bows
11:57:03 <Taneb> Nah, I'm just lazy and stubborn
11:57:14 <Roujo> There's the Lazy Newb Pack for that =P
11:57:15 * nooodl wants to make an ascii pokemon roguelike based on actual pokedex data
11:57:43 <Taneb> nooodl, with the whole Alakazam has an IQ of over 3000 crap?
11:57:47 <nooodl> that'd be rad. it'd be like mystery dungeon except not rguhhghggg and with more stuff i guess
11:57:55 <Taneb> Roujo, well it's too late now, I'm not a newb
11:57:56 <nooodl> oh no, i just mean, accurate movesets etc.
11:58:03 <Taneb> Oh
11:58:13 <fizzie> http://sprunge.us/eGWP NetHack players have dirty mouths.
11:58:15 <Taneb> So, basing it on one of the main series games?
11:58:39 <nooodl> elliott: 9877 1 0.000 killed by a djinni called Where's my fucking wish?
11:58:43 <nooodl> guess who :')
11:58:59 <nooodl> Taneb: yeah
11:59:26 <fizzie> "housecat called Kitty with a huge gaping jaw the size of a fucking hallway" has killed two people, I see.
11:59:33 <nooodl> Roujo: well acehack still has the text graphics
11:59:43 <nooodl> they're not as 100% insane as dwarf fortress's though
11:59:58 <Jafet> http://www.rarecandytreatment.com/comics/1019152/brink-of-mentality/
12:00:08 <nooodl> fizzie: must be someone's petname
12:00:41 <Roujo> I'm watching a game right now
12:00:54 <Roujo> And I'm having trouble following =P
12:02:21 <fizzie> http://sprunge.us/gSiR "it keeps happening"
12:03:55 <nooodl> fizzie: poetic
12:06:47 <Jafet> Do not call watch captains shitty minetown bones.
12:11:05 -!- dessos has joined.
12:12:25 -!- yorick has joined.
12:22:10 -!- oerjan has joined.
12:25:48 <elliott> graphical DF is impossible to play I think
12:28:20 <Jafet> You mean the one where an opengl program pretends to be a terminal emulator
12:28:28 <Jafet> That was weird yes
12:29:15 <elliott> I mean the one where you have to keep track of >100 tiny little smudgy pictures instead of simple geometric shapes you already know
12:41:33 <elliott> fizzie: whoa, tilaa have SSDs now.
12:42:35 <fizzie> "A lot of our competitors who offer SSD VPS actually only do SSD caching. That means that on the backend your data is still stored on "spinning rust" and the SSD's are used to offload the HDD's somewhat." I love the "spinning rust" expression, it's so disparaging.
12:43:48 <elliott> do I want 1 gigabyte of RAM and 20 GB of SSD or do I want 2 gigabytes of RAM and 30 GB of spinning rust?
12:43:52 <elliott> I guess the latter also has like twice the CPU.
12:44:05 <elliott> these prices were so much better when I thought 20 dollars was 20 pounds instead.
12:44:07 <oerjan> @tell Bike <Bike> Can you think of an uncomputable number that isn't the limit of a sequence of computable numbers? <-- not if the sequence itself can be uncomputable. after all every real number is a limit of a sequence of rationals.
12:44:07 <lambdabot> Consider it noted.
12:44:21 <elliott> oerjan: we went over thiiiiiiiiiiiiiiiiiiiiiiiiiis
12:44:25 <oerjan> OKAY
12:45:01 <oerjan> hm, in fact you can even define explicitly a unique sequence...
12:45:26 <elliott> alternatively do I want 32 gigabytes of RAM, 19.2 GHz (???) of CPU, and 1280 gigabytes of hard disk (241.40 EUR/mo), or the same but with 240 gigabytes of SSD (243.80 EUR/mo)?
12:45:36 <oerjan> just take your favorite ... oh wait, just take the continued fraction cutoffs.
12:45:55 <elliott> wait, that's pretty incredibly cheap.
12:46:20 <elliott> 32 gigabytes of RAM and eight cores and a terabyte of disk for two hundred pounds a month?
12:46:31 <elliott> would anyone like to give me some money.
12:47:10 <Jafet> I can give my two cents
12:47:11 <fizzie> Aw, the SSD slider doesn't go above 240 gigs.
12:47:41 <oerjan> elliott: here's a nickel, boy. buy yourself a better computer.
12:48:07 <elliott> fizzie: yes. that was my one disappointment from accidentally hitting "end" on the storage meter and then seeing what would happen if I did it to all the other sliders too.
12:48:19 <Jafet> The funny thing is that you can actually get a computer for a nickel these days
12:48:21 <fizzie> I think the 19.2 GHz is like 8*2.4 GHz.
12:48:31 <elliott> right.
12:48:34 <oerjan> Jafet: something embedded stuff?
12:48:35 <Roujo> Jafet: Well, you can get anything for a nickel, really
12:48:44 <Roujo> It just depends on the nickle
12:49:26 <fizzie> If you ask for 32 gigs of RAM, the CPU slider also doesn't go below 9.6 GHz.
12:50:00 <Roujo> That's a lot of hertz o0
12:50:21 <oerjan> presumably that's sum of cores?
12:50:26 <fizzie> Yes. Sadly.
12:50:43 <Roujo> Could you imagine?
12:50:57 <fizzie> I guess I have a... 13.6 GHz computer, according to their math.
12:51:00 <Roujo> A single-core processor SO FAST that it makes parallel computing irrelevant
12:51:22 <oerjan> Roujo: that's what people imagined until about 2003 hth
12:51:29 <Roujo> http://valid.canardpc.com/show_oc.php?id=2587625
12:51:56 <Roujo> 8.7 GHz on an AMD FX-8350
12:52:14 <Jafet> Is memory faster than 133 MHz yet
12:52:18 <oerjan> 283.69 is quite precise
12:52:41 -!- carado has joined.
12:53:07 <Roujo> oerjan: My guess is that the last time he went to 283.70, the universe SEGFAULTed or something
12:53:14 <oerjan> okay
12:53:24 <Roujo> So we're actually a restored-from-backup copy
12:53:36 <oerjan> > 1/283.69
12:53:39 <lambdabot> 3.5249744439352815e-3
12:53:48 <oerjan> hm that doesn't help
12:54:07 <Jafet> http://totl.net/Eunuch/
12:54:12 <Jafet> (obligatory)
12:54:20 <oerjan> i wondered if that was simply the result of two other numbers divided with unreasonable precision or something
12:54:48 <fizzie> Jafet: PS4 has (will have) eight gigs of GDDR5 (clock speed 1.375 GHz, data rate of 5.5Gbps) as its unified CPU/GPU RAM.
12:55:00 <Jafet> > approxRational 0.01 (1/283.69)
12:55:01 <lambdabot> 1 % 74
12:56:16 <oerjan> > 1/74 - 1/283.69 -- i don't think that's a very good use of approxRational
12:56:18 <lambdabot> 9.988539069578233e-3
12:56:47 <oerjan> > approxRational 0.000001 (1/283.69)
12:56:48 <lambdabot> 0 % 1
12:56:56 <oerjan> wat
12:56:58 <fizzie> Heh.
12:57:03 <Jafet> > flip approxRational 0.01 (1/283.69)
12:57:04 <lambdabot> 0 % 1
12:57:23 <oerjan> :t approxRational
12:57:25 <lambdabot> RealFrac a => a -> a -> Rational
12:57:28 <Jafet> > flip approxRational 0.000001 (1/283.69)
12:57:29 <lambdabot> 3 % 851
12:58:03 <fizzie> How about, you know, just changing the arguments instead of flipping the function?
12:58:04 <oerjan> > flip approxRational 0.00001 (1/283.69)
12:58:05 <lambdabot> 1 % 283
12:58:06 <elliott> fizzie: hmm, why don't computers do RAM like that?
12:58:15 <oerjan> fizzie: now that's crazy talk
12:58:24 <oerjan> > flip approxRational 0.0000001 (1/283.69)
12:58:25 <lambdabot> 13 % 3688
12:58:36 <oerjan> ok i guess not.
12:59:03 <fizzie> elliott: Because it costs money, I suppose. (If you mean "why not use GDDR5 also for system RAM instead of GPU only", and not "why not unified memory architectures".)
12:59:22 <elliott> fizzie: I think I mean both, though I can think of reasons for the latter quicker.
12:59:35 <elliott> surely there are lots of people willing to pay money for computers.
12:59:39 <Roujo> We should just put the processor in the cloud, really
12:59:46 <Roujo> Outsource your motherboard
13:00:11 -!- boily has joined.
13:00:14 -!- metasepia has joined.
13:00:30 <fizzie> elliott: I think you need something different in the memory controller for that, and Intel et al. might not be willing to cater for the presumably-not-too-large market willing to pay for GDDR5 RAM. (Or I could be talking out of my ass; I don't know about hardware.)
13:00:35 <Roujo> Although I guess the cloud would be too mainstream
13:00:37 <elliott> Roujo: they call that OnLive.
13:00:52 <Roujo> I store all my files in the water table
13:00:53 <elliott> fizzie: isn't that market, like, all gamery and overclocky type weirdos.
13:00:59 <elliott> those people have lots of money.
13:01:02 <Deewiant> fizzie: GDDR5 has worse latency
13:01:06 -!- ais523 has joined.
13:01:10 <elliott> thank you Deewiant, that sounds like a much more realistic reason.
13:01:11 <Roujo> It's like the clound, only more underground
13:01:18 <Roujo> cloud*
13:01:19 <ais523> whoops, wrong nick
13:01:22 -!- ais523 has changed nick to ais523_.
13:01:27 <ais523_> hi #esoteric, btw
13:01:31 <Roujo> Heya ais523_
13:01:35 <elliott> Your Linode, solidity, has exceeded the notification threshold (1000) for disk io rate by averaging 1243.39 for the last 2 hours. The dashboard for this specific Linode is located at: <https://manager.linode.com/linodes/dashboard/solidity>
13:01:41 <elliott> linode email me every time I do an upgrade
13:02:01 <Roujo> `relcome ais523_
13:02:03 <HackEgo> ais523_: Welcome to the international hub for esoteric programming language design and deployment! For more information, check out our wiki: http://esolangs.org/wiki/Main_Page. (For the other kind of esoterica, try #esoteric on irc.dal.net.)
13:02:05 <elliott> ais523_: I like your thesis.
13:02:05 <Deewiant> And DDR4 is coming out anyway so if you want better/more expensive desktop RAM you'll get it
13:02:11 <ais523_> something else came up when I was writing my research, which is incredibly minor and probably not worth mentioning in the research
13:02:22 <ais523_> but it's arguably an esoprogramming problem, and I'm having trouble seeing the trick needed to get the answer
13:03:19 <ais523_> let's see... it's very easy to write a regular expression (I'm talking mathematical regular expression here, not regexp) to match strings of the form "ababababababab" and so on
13:03:30 <ais523_> and likewise, it's very easy to write one for "cdcdcdcdcdcdcd" and so on
13:03:39 <ais523_> but I'm trying to write one that matches any interleaving of those strings
13:03:49 <ais523_> as in, if you look at just the a's and b's, the a's and b's alternate
13:03:56 <ais523_> and if you look at just the c's and d's, the c's and d's alternate
13:04:08 <ais523_> but there's no restriction on how the a's and c's can line up against each other, for instance
13:04:12 <boily> ais523_: long time no see!
13:04:20 <ais523_> so something like abacabdc is fine
13:04:48 <ais523_> in theory, it should be possible, because it's clearly quite easy to write an FSM to do that
13:04:54 <ais523_> and regular expressions and FSMs are equally powerful
13:04:56 <ais523_> I just can't see how
13:05:12 <Jafet> Alternatives, lots of alternatives
13:05:13 <Deewiant> Just write the FSM and convert it to a regex
13:06:41 <ais523_> hmm... the FSM is {from, input, to} {0,a,1},{0,c,2},{2,a,3},{1,c,3},{3,b,1},{3,d,2},{1,b,0},{2,c,0}
13:06:58 <ais523_> I'm not familiar with methods of regexifying those, though
13:07:26 <Deewiant> http://cs.stackexchange.com/questions/2016/how-to-convert-finite-automata-to-regular-expressions
13:07:39 <ais523_> thanks for the link
13:08:12 <ais523_> (I'm currently writing about models of hardware, and have concluded that delay-insensitive asynchronous circuits are regular expressions and synchronous circuits are recurrence relations)
13:08:41 <Taneb> Sounds fun
13:09:13 <fizzie> I remember a FSM-to-regular-expression conversion routine from the "intro to theoretical computer science" course.
13:09:25 <fizzie> It was all about merging edges and whatnot.
13:10:07 <fizzie> Presumably the same as the "state removal method" mentioned in Deewiant's link.
13:10:49 <Deewiant> It looks familiar so they're probably the same, yes.
13:12:10 <ais523_> so let's see... if I remove 3, we get {0,a,1},{0,c,2},{1,b,0},{2,c,0},{2,ab,2},{2,ad,1},{1,cb,2},{1,cd,1}
13:12:59 <ais523_> not quite, there are typos there, but I'll mentally correct them
13:13:00 <fizzie> ais523_: Are all your states final, or just the one that has both b and d last?
13:13:10 <ais523_> fizzie: they're all final
13:13:11 <Deewiant> Doing this by hand without drawing it is probably a good way of making mistakes and getting it a bit wrong
13:13:37 <fizzie> ais523_: In that case, I think you need to add epsilon edges from all to a dummy final state, since the algorithm only removes states that aren't final.
13:13:56 <fizzie> (Disclaimer: it might still work out in practice, in your case.)
13:14:02 <ais523_> fizzie: right; it actually doesn't really matter for my application
13:14:16 <ais523_> especially because I'm just curious as to what it looks like, more than anything else
13:15:24 <ais523_> then removing 2: {0,a,1},{1,b,0},{1,cd,1},{0,c(ab)*d,0},{0,c(ab)*ad,1},{1,cb(ab)*d,0},{1,cb(ab*)ad,1}
13:16:22 <ais523_> which is {0,c(ab)*d,0},{0,a+c(ab)*ad,1},{1,b+cb(ab)*d,0},{1,cd+cb(ab*)ad,1}
13:17:03 <elliott> Deewiant: so why is the PS4 using higher-latency RAM then?
13:17:48 <Jafet> Presumably they believe in caches
13:17:50 <ais523_> then finally, removing 0: (c(ab)*d+((a+c(ab)*ad)(cd+cb(ab)*ad)*(b+cb(ab)*d)))*
13:17:57 <fizzie> elliott: From what I recall about the PS4 seminar, their developers were really hankering for a unified memory architecture. Plus they were focused a lot on doing non-graphics ("compute shaders") stuff on the GPU.
13:18:13 <Jafet> That regular expression is smaller than I thought
13:18:14 <elliott> fair enough
13:18:17 <fizzie> elliott: (Parallelized GPU work is of course more about throughput than latencies.)
13:18:23 <ais523_> that's not in lowest terms, though
13:18:26 <ais523_> I think I can simplify it
13:18:52 <elliott> ais523_: that regular expression is basically what I was on the path to writing (very slowly)
13:19:01 <ais523_> then finally, removing 0: (c(ab)*d+((a+c(ab)*ad)(c(ba)*d)*(b+cb(ab)*d)))*
13:19:09 <ais523_> * (c(ab)*d+((a+c(ab)*ad)(c(ba)*d)*(b+cb(ab)*d)))*
13:19:13 <katla> what does it match
13:19:15 <elliott> ais523_: I just started writing out all the permutations in the knowledge that since {a,b,c,d} is finite, eventually there'll be a synchronisation point where you can just * the whole thing
13:19:19 <elliott> sort of thing
13:19:25 <elliott> but then you found a non-stupid way to do it before I could be bothered
13:19:32 <ais523_> katla: strings where a and b alternate (with potential c and d in between) and also c and d alternate (with potential a and b in between)
13:19:38 <Jafet> Shouldn't the expression be symmetrical
13:19:47 <ais523_> elliott: that's what I started doing but I ended up failing due to lack of a GOTO :(
13:19:53 <ais523_> Jafet: it is
13:19:57 <katla> hmm
13:20:00 <ais523_> just in a nonobvious way
13:20:10 <katla> eyeah you can just write an NFA for it
13:20:23 <ais523_> katla: that's what we just did in-channel
13:20:54 <ais523_> conclusion: state machines are a better model for asynchronous digital circuits than regular expressions
13:22:17 <ais523_> this must be why electronic engineers like them so much
13:22:24 <Jafet> You need less time to type regular expressions than to draw state machines though
13:22:38 <Jafet> Think about this as the paper deadline nears
13:23:44 <fizzie> On the third hand, you don't get to use a TeX package for drawing state machines.
13:24:05 <ais523_> Jafet: yeah, that's actually why I was using regular expressions so far
13:26:09 <Jafet> The simple regular expressions might also correspond to some interesting subset of circuits
13:26:27 <ais523_> does "simple" have a technical meaning here?
13:27:00 <ais523_> anyway, the reason why regular expressions are appropriate for this is that not only do these sorts of circuit (like most circuits) not produce certain outputs, they also refuse certain inputs
13:27:05 <ais523_> as in, those inputs can't be given to it
13:27:10 <boily> ~duck simple
13:27:10 <metasepia> simple definition: free from guile.
13:27:45 <ais523_> it's not even in the sense of bottom/fail, either; it's in the sense of undefined behaviour
13:27:56 <ais523_> for some reason, I find it reassuring seeing UB come up in formal mathematics
13:28:17 <Jafet> I thought asynchronous circuits accepted all (appropriately spaced) inputs
13:29:10 <ais523_> this is delay-insensitive circuits, so the potential appropriate spacing could be infinite
13:29:15 <ais523_> and you can only determine it via observing their output
13:29:39 <ais523_> e.g. for a simple circuit like XOR, AQ+BQ, once you give A you can't send A again or B until you get the Q
13:30:21 <ais523_> or for a circuit like C, ABQ+BAQ, once you've sent A, you have to send B next, you're never allowed to send A twice without a Q in between (and the Q won't be sent until it sees the B)
13:30:39 <ais523_> also, something that confuses me a lot is that the person who invented the C element is apparently called Muller, no umlaut
13:31:46 <Deewiant> Pronounce it /mʌlər/ to be safe
13:33:01 <Jafet> Mahler
13:33:56 <elliott> does anyone know what decides what Debian packages are installed as part of a "base system"?
13:34:05 <elliott> there doesn't seem to be a metapackage corresponding to something like ubuntu-desktop
13:34:12 <elliott> and the packages don't seem to all be marked as automatically installed
13:34:23 <Jafet> gnome-desktop-task
13:34:35 <elliott> (I'm trying to get a list of all packages I explicitly installed; aptitude search '~i !~M !~E' does okay but includes a lot of stuff that comes with Debian)
13:34:39 <elliott> Jafet: but it's a server :p
13:34:52 <Jafet> Oh, the actual base system.
13:35:25 <ais523_> elliott: I remember reading something in Debian policy about that
13:35:33 <ais523_> (what does it say about me that I read Debian policy for fun?)
13:36:13 <Jafet> You are well-informed about relevant items.
13:36:32 <ais523_> elliott: "all packages with priority `required' or `important'"
13:36:50 <ais523_> so basically anything with a lower priority, you installed yourself
13:37:21 <ais523_> that's actually a simpler definition than I was expecting
13:39:17 <ais523_> fwiw, the breakdown seems to be: "required" = enough functionality to boot the system and install more software; "important" = minimal amount needed to be able to reasonably administer the system without having to install extra packages to do so; "standard" = reasonable defaults for a small character-mode system; "optional" = pretty much everything goes here; "extra" = packages that conflict with more useful packages or are only us
13:39:30 <ais523_> a base system is required+important, a default install is required+important+standard
13:39:46 <ais523_> and required+important+standard+optional is intended to be theoretically possible but you'd have to be mad to actually do ti
13:39:47 <ais523_> *it
13:40:13 <Deewiant> ais523_: Cut off: '"extra" = packages that conflict with more useful packages or are only us'
13:40:23 <ais523_> ...useful in specialized cases
13:40:32 <fizzie> Filtering based on priority might help in "list explicit installations".
13:40:34 <ais523_> like debug symbols, that sort of thing
13:40:51 <Jafet> Pretty sure there would be people that test *+optional systems
13:40:59 <ais523_> yeah, I think there's at least one
13:41:01 <fizzie> !~prequired or some-such.
13:41:28 <ais523_> well, "required" is pretty much "if this isn't installed, your system is doomed"
13:42:01 <ais523_> whereas "essential", a different measurement is "things that are installed so widely that you don't need to bother declaring dependencies on them"
13:42:08 <ais523_> *, a different measurement,
13:42:49 <fizzie> ais523_: There seems to be (possibly hardware-detect-related?) things in the default install that don't have those priorities, yet aren't marked with a high priority.
13:43:07 <ais523_> fizzie: hmm
13:43:12 <fizzie> ais523_: E.g. "acpi" has Priority: optional, is not marked as automatically installed, is installed, and I certainly haven't installed it myself.
13:43:24 <fizzie> ais523_: (And this was just a base system with no tasksel tasks, IIRC.)
13:43:27 <ais523_> yeah, it seems possible that the installer would add in drivers specific to your hardware config
13:43:41 <ais523_> policy doesn't mention that, but OTOH I don't see a reason why it would
13:44:02 <fizzie> Similarly other such things like pciutils, pm-utils seem to be here.
13:44:55 <fizzie> (Okay, pciutils is Priority: standard. But pciutils is optional.)
13:45:05 <fizzie> Er, pm-utils is optional.
13:46:08 <fizzie> Huh, how have I ended up with a non-automatic installation of 'libfreeimage3'?
13:46:22 <fizzie> I don't even have a single package that'd depend on it.
13:46:24 <elliott> ais523_: huh, python and perl match even with !~prequired !~pimportant
13:46:29 <elliott> do those raelly not come with debian?
13:46:40 <elliott> *really
13:47:09 <ais523_> elliott: there's -minimal versions that are shipped with debian
13:47:22 <fizzie> ais523_: Curiously enough, both python and python-minimal are Priority: standard.
13:47:22 <ais523_> IIRC, the full version of Perl is flagged build-essential, but not anything higher than that
13:47:38 <ais523_> (which means that Perl doesn't have to be specified as a build dependency of anything, and thus needs to be installed to build packages)
13:47:42 <Jafet> I like how prequired includes awk, sed, perl, and two shells.
13:47:47 <fizzie> (perl-base is 'required', perl is 'standard'.)
13:48:04 <ais523_> Jafet: that's because they're often used in init scripts and the like, I guess
13:48:17 <ais523_> also I have a suspicion that apt is written at least partly in Perl
13:48:31 <Jafet> But apt is pimportant.
13:48:37 <elliott> $ aptitude search '~pimportant !~i'
13:48:38 <elliott> c libboost-iostreams1.46.1 - Boost.Iostreams Library
13:48:41 <elliott> c libboost-iostreams1.48.0 - Boost.Iostreams Library
13:48:41 <fizzie> Also called "p. important".
13:48:44 <elliott> p libboost-iostreams1.53.0 - Boost.Iostreams Library
13:48:47 <elliott> p libboost-iostreams1.54.0 - Boost.Iostreams Library
13:48:50 <elliott> c libept1 - High-level library for managing Debian package information
13:48:53 <elliott> c libsysfs2 - interface library to sysfs
13:48:56 <elliott> p nfacct - netfilter accounting object tool
13:48:59 <elliott> why are there a bunch of versions of a boost library there
13:49:03 <Jafet> That dirty ortant.
13:49:09 <fizzie> elliott: Because of aptitude.
13:49:20 <elliott> but four versions?
13:49:21 <ais523_> hmm, what does required use as the package manger?
13:49:25 <fizzie> elliott: (It depends on libboost-iostreams.)
13:49:25 <ais523_> dpkg + wget?
13:49:29 <elliott> but four versions?
13:49:35 <elliott> maybe the query is bad
13:49:45 <elliott> is it showing other versions that could satisfy the requirement for libboost-iostreams and hence are important
13:49:52 <elliott> despite the fact that I'll have some other version of libboost-iostreams installed?
13:50:00 <ais523_> yeah, that makes sense, I guess
13:50:12 <Jafet> There is no wget. Presumably you need to load debs off a floppy.
13:50:13 <fizzie> elliott: Those two "c" entries are some kind of conflicts, I think.
13:50:22 <elliott> debian is hard :(
13:50:32 <ais523_> hmm, is this a version of debian before or after they invented "breaks"?
13:50:40 <elliott> this is jessie
13:50:43 <elliott> but it started as squeeze or something
13:50:48 <elliott> maybe even the one before squeeze
13:50:59 <ais523_> probably a new version, then, because I don't recognise that codename
13:51:06 <ais523_> (except that it comes from toy story)
13:51:12 <elliott> $ aptitude search '~i !~M !~E !~prequired !~pimportant !~Rbuild-essential' | less
13:51:15 <elliott> seems pretty good
13:51:21 <elliott> ais523_: jessie is testing right now
13:51:22 <fizzie> ais523_: Both gcc-4.7-base and gcc-4.8-base are in priority: required.
13:51:25 <ais523_> elliott: right
13:51:40 <elliott> btw, is the official Debian package manager apt-get or aptitude this week?
13:51:42 <ais523_> fizzie: that makes sense, I think
13:52:18 <fizzie> Also some rather curious things that presumably are dependencies of something; like libtext-charwidth-perl "get display widths of characters on the terminal".
13:52:42 <elliott> heh, I had gcc 4.4 installed
13:52:44 <Jafet> libncurses is prequired.
13:52:46 <elliott> now I don't
13:52:55 <elliott> I like "prequired", it looks like a word
13:53:05 <elliott> huh, I have busybox installed explicitly
13:53:12 <elliott> but why would I have installed busybox?
13:53:16 <Jafet> I think bash is the only program that needs it.
13:53:32 <elliott> hmm... I wonder if there's a way to phrase "not depended on by any other installed packages"
13:53:39 <elliott> to find the leaves
13:53:57 <Jafet> You mean the roots?
13:54:01 <elliott> er. yes.
13:54:01 <quintopia> no
13:54:05 <elliott> I'm no good at plants.
13:54:16 <quintopia> the ones that are not depended on are leaves
13:54:29 <quintopia> the ones with no dependencies are roots
13:54:53 <quintopia> and those seem like as good names as any for those things
13:54:56 <Jafet> `words 20
13:55:00 <HackEgo> teiwerke ben kiim men khi jan earra tam sacripe dugo maged traft distile elson flama vasci crawa word unho sou
13:55:03 <quintopia> dependency leaf/dependency root
13:55:06 <quintopia> use those
13:55:25 <Jafet> Word.
13:55:26 <fizzie> Jafet: util-linux PreDepends on libncurses5 too.
13:55:27 <elliott> and why do I have tcpd installed?
13:55:31 <elliott> I don't remember installing half these packages
13:55:49 <fizzie> Jafet: (And a couple of pimportant things -- procps, bsdmainutils -- but nothing else prequired.)
13:56:21 <fizzie> In fact, bash doesn't; just libtinfo5.
13:56:25 <elliott> wtf
13:56:29 <elliott> why does my server have mplayer installed???
13:56:31 <elliott> oh
13:56:32 <elliott> for mencoder
13:56:44 <elliott> why does my server have mingetty installed...
13:57:15 <fizzie> For the virtual terminal provided by your VPS system's VNC console thingamajick?
13:57:30 <elliott> but it's explicitly installed and stuff.
13:57:37 <elliott> I guess that just means the VPS installation process might have explicitly installed it.
13:57:53 <elliott> a victory: learned that mailman was installed; uninstalled it.
14:00:48 <elliott> i console-setup - console font and keymap setup program
14:00:52 <elliott> come on, why is this installed
14:01:17 <fizzie> It gets done real early in the setup, automatically, by the installer.
14:01:29 <elliott> what does?
14:01:32 <fizzie> console-setup.
14:01:39 <Jafet> Are you trying to uninstall things until debian stops working
14:01:43 <elliott> Jafet: yes
14:01:48 <fizzie> I think everything that's done as a particular step by the installer ends up in your query.
14:01:57 <elliott> that's annoying
14:02:02 <elliott> they should mark these things better :(
14:02:11 <fizzie> E.g. I've got exim4 there, and I certainly didn't install it explicitly myself, but the installer did to get the default mail system in place.
14:02:32 <Jafet> I have never attempted to use the automatically-installed indicator manually.
14:02:46 <elliott> i eject - ejects CDs and operates CD-Changers under
14:02:59 <ais523_> eject's description is misleading
14:03:04 <elliott> I want to believe that I can cause the CD tray of some physical server sitting in a data centre somewhere to pop out
14:03:09 <elliott> that would make my day, I think
14:03:24 <Jafet> You want to believe that your rack mount server has a CD tray
14:03:29 <elliott> it's a VPS.
14:03:33 <ais523_> it unmounts arbitrary removable media, regardless of whether it's a CD or a USB stick or whatever
14:03:34 <elliott> can't I dream?
14:03:46 <Jafet> Oh ok, it could be a virtual CD tray.
14:03:49 <ais523_> (and thus is necessary because umount only works as root)
14:04:03 <elliott> fizzie: what do you want if you just want a sendmail executable these days on Debian?
14:04:11 <elliott> install exim and just accept that you'll have a daemon running?
14:04:49 <fizzie> elliott: I've used 'scmail' for a thing where I just wanted a sendmail-binary-compatible replacement that always sends everything to a SMTP server.
14:05:28 <elliott> is that ssmtp?
14:05:43 <elliott> oh, that wants a proper mail server
14:05:53 <elliott> my use-case is "MediaWiki wants to be able to send emails"
14:06:14 <fizzie> Well, do you want people to be able to receive the emails it wants to send?
14:06:23 <elliott> in a perfect world
14:06:58 <elliott> ii g++ 4:4.7.2-1 i386 GNU C++ compiler
14:07:01 <elliott> ii g++-4.7 4.7.3-4 i386 GNU C++ compiler
14:07:04 <elliott> ii gcc 4:4.7.2-1 i386 GNU C compiler
14:07:07 <elliott> ii gcc-4.7 4.7.3-4 i386 GNU C compiler
14:07:10 <elliott> ii gcc-4.7-base:i386 4.7.3-4 i386 GCC, the GNU Compiler Collection (base package)
14:07:13 <elliott> ii gcc-4.8-base:i386 4.8.1-2 i386 GCC, the GNU Compiler Collection (base package)
14:07:16 <elliott> I do wonder why gcc-4.8-base is all on its own there.
14:07:16 <Jafet> You're running mediawiki, and you don't like the exim daemon?
14:07:26 <Jafet> 4.8 is the hipster version
14:07:30 <fizzie> scmail, ssmtp and one-other-I-forget-which are reasonable replacements for the case where you have a nearby real (SMTP-accessible) mail server and you want to use that for all outgoing mail.
14:08:00 <ais523_> doesn't sending mail require a daemon, in general, for the case where there's a temporary fault on the next server in the chain?
14:08:21 <elliott> Jafet: I don't mind exim.
14:08:25 <fizzie> ais523_: Not if you can live with the fact that the 'sendmail' binary just returns an error in that case.
14:08:26 <elliott> Jafet: it has postfix right now and I don't know why.
14:08:37 <elliott> ideally I'd just install the simplest "Debian Way"(tm) thing for it. maybe that is just exim
14:08:56 <Jafet> I thought the typical mail chain is of length 2 these days
14:09:00 -!- coppro has quit (Remote host closed the connection).
14:09:02 <fizzie> ais523_: (If you want to talk directly to other people's mail servers, then.)
14:09:16 <elliott> I wonder if there is a way to install mencoder that doesn't hurt me by having packages named e.g. libgl1-mesa-glx and libxinerama1 on my server.
14:09:36 <fizzie> Jafet: My most recent message in the inbox has 11 Received: headers.
14:10:02 <ais523_> fizzie: mailing list?
14:10:24 <fizzie> ais523_: Well, yes. Let's see a most recent non-list post.
14:10:44 <fizzie> 7 Received:-headers.
14:11:05 <oklopol> ais523_: it's very often the case that a closure property of regular languages has a natural implementation by FSA, but there is really no way to do it with regular expressions
14:11:14 <Jafet> Heh, mencoder depends on mplayer.
14:11:15 <oklopol> i think
14:11:40 <ais523_> oklopol: yeah, that makes sense
14:11:44 <ais523_> also vice versa
14:11:54 <oklopol> can you give an example of the vice versa
14:11:58 <oklopol> i can't
14:12:24 <oklopol> trivial things like reversal & homomorphisms are slightly easier with regexps, but they are also trivial for FSA.
14:12:30 <elliott> I think I need ghostscript for mediawiki
14:12:41 <ais523_> there's one on the Wikipedia page: (a+b)*a(a+b)(a+b)(a+b)(a+b)(a+b)
14:12:42 <elliott> which is annoying beacuse it also wants me to have libxinerama1
14:12:43 <oklopol> well maybe there are such examples, dunno
14:12:55 <elliott> should just install ubuntu
14:12:57 <elliott> and forget I have packages
14:12:57 <fizzie> scia-ca-demic.org (first source), smtp-in-02.aalto.fi (ingress SMTP here), "localhost (Email Security Appliance)", marconi.ics.hut.fi (departmental SMTP inside the local network), localhost (amavisd), marconi.ics.hut.fi (again), marconi.ics.hut.fi (Cyrus LMTP).
14:13:07 <ais523_> you get an exponential blowup of state because you need to remember which of the most recent 6 characters have been a
14:13:33 <ais523_> fizzie: is "again" a hostname, there?
14:13:33 <oklopol> okay NFA
14:13:49 <ais523_> yeah, you can't get a combinatorial explosion converting a regex to an NFA
14:13:55 <fizzie> ais523_: No, it was a description.
14:14:20 <ais523_> I almost want to call a computer "again", now, but it'd violate my naming convention
14:14:26 <ais523_> where all my computers are named after biomes
14:14:40 <elliott> does ais523_ have a computer named "nether" yet
14:15:01 <Jafet> fizzie: but isn't all of .fi just one host or something
14:15:08 <ais523_> elliott: no, also I don't think that's a real biome
14:15:21 <elliott> it is fsvo real
14:15:36 <boily> there should be one machine named "disco", for the new 1.6 cañons.
14:15:55 <fizzie> Checked another chain for the personal email; that one went: mailsend.netposti.fi - mailsend.netposti.fi - ikiaikainen.iki.fi - jatkuu.iki.fi - fe23.mail.saunalahti.fi - be18.mail.saunalahti.fi - "tmail-2006i.2007-sau", whatever that means.
14:16:12 <elliott> I should just pay ais523_ to run my server for me
14:16:17 <elliott> ais523_: how much do you want
14:16:18 <fizzie> So the chains aren't quite of length 2 yet.
14:16:23 <ais523_> but that'd mean having to administer MediaWiki
14:16:30 <elliott> ais523_: no, it's OK
14:16:32 <elliott> I don't do that either
14:16:46 <elliott> I just ignore it until the version number gets sufficiently far behind that I feel really guilty
14:16:47 <fizzie> Ooh, a DKIM signature, so fancy.
14:16:51 <ais523_> elliott: fwiw, I've sometimes taken days to notice that the nethackd on nethack4.org is down
14:16:57 <fizzie> (Mail headers are such a joy.)
14:17:14 <elliott> ais523_: I guess your community is scared of you
14:17:15 <elliott> or non-existent
14:17:21 <elliott> or they'd point it out
14:17:24 <ais523_> elliott: it's more that the user-visible error message is really misleading
14:17:28 <ais523_> "IPv6 error"
14:17:32 <elliott> heh
14:17:33 <Jafet> When I installed my computer, I named it test
14:17:39 <ais523_> regardless of what protocol they connected via
14:17:41 <fizzie> Both http://pulseaudio.org/ and http://alsa.opensrc.org/ have been down the last couple of days, right when I was configuring those two things.
14:17:46 <Jafet> This leads to weird pauses when looking at network logs
14:17:58 <fizzie> I've had to use some sort of freedesktop.org copy of PA documentation and whatnot.
14:18:28 <elliott> ais523_: it actually really annoys me that I have PHP and MySQL installed just for one abominable program :)
14:18:54 <fizzie> It annoys me that I have PHP and MySQL installed for just one abominable program (Gallery2).
14:18:55 <elliott> too bad there is not really any viable alternative I could get away with
14:18:56 <ais523_> I think MySQL is a dependency of something important in Ubuntu. Somehow.
14:19:12 <Jafet> That's if you consider anything in ubuntu important
14:19:13 <boily> is mariadb ubuntu-ready?
14:19:20 <elliott> it might be MariaDB I have
14:19:31 <elliott> I assume it is still just horrific
14:20:08 <fizzie> Qt4 brings in the MySQL client, but at least not the server.
14:20:57 <elliott> ais523_: oh, the worst thing about MW is that Debian's package is really old
14:21:02 <elliott> so it's a real pain to upgrade
14:21:11 <fizzie> (Via libqtgui4 -> libqt4-declarative -> libqt4-sql -> libqt4-sql-mysql -> libmysqlclient18.)
14:33:03 -!- Bike has joined.
14:41:48 <boily> ~yi
14:41:49 <metasepia> Your divination: "Obstruction" to "Obstruction"
14:43:30 <Bike> uh oh
14:45:05 <boily> I think elliott is in a serious pinch.
14:48:06 <Roujo> ~yi Hats
14:48:06 <metasepia> Your divination: "Retiring" to "Radiance"
14:50:23 <boily> Roujo: ~yi ignores its parameters. you have to think about the question with your mind (or soul, if you prefer. in an emergency, you can also use a rental soul, but results aren't as good usually. ymmv. hth.) hth.
14:50:43 <boily> I guess you could also tag your yises to remember what were the questions asked.
14:53:49 -!- Phantom_Hoover has joined.
14:54:41 <boily> Phantom_Helloover.
14:55:15 <Roujo> Thanks, boily, I'll keep that in mind =)
14:55:16 <Roujo> Also
14:55:20 <Roujo> `relcome Phantom_Hoover
14:55:22 <HackEgo> Phantom_Hoover: Welcome to the international hub for esoteric programming language design and deployment! For more information, check out our wiki: http://esolangs.org/wiki/Main_Page. (For the other kind of esoterica, try #esoteric on irc.dal.net.)
14:55:47 <Phantom_Hoover> rello
14:57:45 <ais523_> hmm, I never knew what relcome looked like, until I viewed it in a webclient
14:57:49 <ais523_> (I filter colors on my usual client)
14:58:37 <Bike> http://leonarnott.neocities.org/ awesome
14:58:42 <Bike> oh hey, it'a ais. hais
14:58:45 <ais523_> "neocities"?
14:58:50 <ais523_> is that like geocities and reocities?
14:58:51 <Bike> beats me
14:59:06 <Bike> hm, looks like it
14:59:16 <ais523_> sorry I was missing for a while, I was trying to track down a copy of a specific paper book
14:59:20 <Bike> " Designed as a 21st century reincarnation of GeoCities, NeoCities lets you make your own site for free. And it just might spark a renaissance of creativity online. "
14:59:27 <Bike> what book
14:59:32 <ais523_> because it seems that everyone has been copying citations of it without looking at the actual book itself
14:59:47 <boily> Bike: why. why did you link to that.
14:59:49 <ais523_> turns out the library here has a copy but it's buried in some secret vault somewhere that takes a while to open, or something
15:00:03 <ais523_> "Switching theory. Vol. 2, Sequential circuits and machines", fwiw
15:00:16 <ais523_> it's what everyone cites as the source of the Muller C-element
15:00:24 <Bike> invigorating
15:00:33 <elliott> hey Bike, want a job
15:00:40 <ais523_> (interestingly, Muller appears to be Muller not Müller, and yet the book is by... Miller)
15:01:00 <ais523_> anyway, I'm wondering if it really is the original source, or if there's an older one
15:01:07 <Bike> elliott: sure what is it
15:01:39 <elliott> Bike: write a mediawiki replacement in any language other than PHP with any storage mechanism other than MySQL.
15:01:45 <elliott> I'll give you $300.
15:01:54 <ais523_> there used to be an experimental postgresql branch of mediawiki
15:02:05 <elliott> it supports postgres
15:02:21 <elliott> but I don't see much point converting the database to only get (a) worse support, (b) still have to use PHP, (c) etc.
15:02:22 <ais523_> so that's half of the problem solved :)
15:02:27 <ais523_> and yeah
15:02:35 <ais523_> I wonder if you could compile the PHP into something else
15:02:51 <elliott> well, sure.
15:03:01 <elliott> but it'd still have the architecture of a PHP application
15:03:03 <Bike> i'll use http://esolangs.org/wiki/Cbrain
15:03:39 <elliott> you're the one missing out on $300 here
15:05:47 <Bike> i don't think $300 is enough to justify learning web programming
15:06:54 <boily> I'm a professional developer. I could do that for you, starting rate at 100 CAD/hour, negociable.
15:07:23 <elliott> boily: that is okay as long as you do it in three hours.
15:08:32 <Bike> wait is it just "300 dollars" independent of currency
15:08:39 <elliott> sure.
15:08:45 <elliott> as long as they're dollars
15:09:50 <Phantom_Hoover> what's the highest-valued dollar
15:09:52 <Bike> http://en.wikipedia.org/wiki/Dollar#Economies_that_use_a_dollar hm
15:09:54 <boily> I'll go with NZD, then :D
15:09:56 <Bike> need some exchange rates here
15:10:24 <Phantom_Hoover> i like how for antigua and barbuda it's just XCD
15:10:38 <Phantom_Hoover> "fuck if i know what letter to use"
15:12:07 <Phantom_Hoover> http://en.wikipedia.org/wiki/Highest-valued_currency_unit
15:12:43 <Phantom_Hoover> cayman islands dollar looks like the best
15:13:44 <Bike> so you'd get like $360 USD, nice
15:23:29 <Phantom_Hoover> oh fucking hell
15:24:14 <Phantom_Hoover> warwick have decided to replace their outdated student management webpage thing with an absolutely horrible one
15:24:41 <boily> is it based on Banner?
15:25:11 <Phantom_Hoover> i don't know
15:25:15 <Phantom_Hoover> i don't want to know
15:26:04 <boily> yes you do. http://www.ellucian.com/Solutions/Banner-Student/
15:26:41 <boily> that thing is pure absolute certified crap. we had a undemocratic totalitarian forced transition pushed on us when I was at university.
15:29:16 <boily> Roujo: si quelqu'un de louche avec un trenchcoat dans la rue te propose d'aller à l'université laval, dit non.
15:30:05 <Roujo> boily: Considérant que c'est à 3h d'ici et que je dois prendre le train pour 16h00... Okay =P
15:32:35 -!- coppro has joined.
15:34:46 <elliott> ais523_: ooh, maybe Esolang could run on ikiwiki instead
15:35:15 <ais523_> I'm not sure wiki migrations from one codebase to another are massively easy or successful
15:35:21 <Phantom_Hoover> boily, I DIDN'T WANT TO KNOW
15:35:30 <elliott> ais523_: yes, but MediaWiki is just that awful...
15:35:37 <Roujo> How is it awful?
15:35:41 <ais523_> elliott: imagine how difficult migrating from it will be, then!
15:35:46 <elliott> ais523_: ideally I would write something with the exact same markup/template/etc. stuff as MediaWiki
15:36:04 <ais523_> Roujo: the most obvious reason for a server admin is that it uses up mindbogglingly high server resources for no obvious reason
15:36:12 <Roujo> ais523_: Alright. =P
15:36:25 <Roujo> I use MediaWiki and never noticed anything wrong, so I was curious
15:36:35 <ais523_> it's quite nice from a user's point of view
15:36:43 <elliott> the server resources are OK
15:36:45 <ais523_> it doesn't fall into the trap of trying to use markdown for markup, for instance
15:36:46 <Roujo> I mean that I admin a MediaWiki site
15:36:47 <elliott> except like, MySQL itself takes up a ton
15:36:55 <elliott> assuming you use caching etc.
15:37:01 <elliott> but if you don't then you're really crazy of course
15:37:13 <Roujo> Depends on the number of hits you get, I'd guess
15:37:15 <ais523_> well it depends on how often the site is accessed
15:37:24 <Roujo> I'd be surprised if I got more then 5/day =P
15:37:45 <elliott> well it also takes something like three lines of configuration to set up caching
15:37:57 <ais523_> yeah but what if you get slashdotted? or reddited? or linked from the home page of Google?
15:38:28 <Roujo> Considering that all that's on the Wiki is the ruleset for a Nomic game that's in a coma and some docs for a Super Mario World ROM Conversion that my brother is doing
15:38:52 <Roujo> Then this happens: http://wiki.psychose.ca/
15:38:58 <Roujo> (The wiki isn't public =P)
15:39:15 <Roujo> Heck, I was supposed to put in a $wgLogo, but never did
15:42:12 <elliott> hmm, ikiwki has templates
15:42:41 -!- oerjan has quit (Quit: leaving).
15:43:40 <ais523_> hmm, I think I'll stick to the plan of joining #esoteric if and only if I have something esoprogramming-related to say
15:43:45 <ais523_> time to go home, anyway
15:43:47 <ais523_> night!
15:44:21 -!- ais523_ has quit (Quit: Page closed).
15:44:22 <Deewiant> Why is markdown a trap
15:44:29 <elliott> we will never know
15:44:41 <Taneb> I've got to stop these damnded naps
15:48:33 <Phantom_Hoover> stay alert at all times
15:48:50 <Phantom_Hoover> you never know when the six pigs will begin their forage
15:50:44 <Roujo> Yup
15:50:48 <Roujo> Darn pigs these days
15:50:53 <Roujo> Foraging and
15:50:54 <Roujo> and
15:50:55 <Roujo> stuff
15:58:36 -!- ka7la has joined.
15:58:37 -!- coppro has quit (Quit: Lost terminal).
15:59:28 <Phantom_Hoover> oh, i forgot the capitals
15:59:41 <Phantom_Hoover> you never know when the Six Pigs will begin their Forage
15:59:49 <Phantom_Hoover> or `Dread Forage' as it is oft known
16:01:18 -!- katla has quit (Ping timeout: 264 seconds).
16:28:59 <Taneb> Phantom_Hoover, we don't talk about the Dread Forage, lest we invoke Its Happening
16:29:25 <Phantom_Hoover> don't go overboard with the capitals
16:36:32 <Roujo> I am the Gate
16:36:34 <Roujo> I am the Key
16:43:00 -!- Administrator1 has joined.
16:43:18 <Administrator1> sup
16:44:57 <elliott> `relcome Administrator1
16:44:59 <HackEgo> Administrator1: Welcome to the international hub for esoteric programming language design and deployment! For more information, check out our wiki: http://esolangs.org/wiki/Main_Page. (For the other kind of esoterica, try #esoteric on irc.dal.net.)
16:48:19 <Administrator1> hi elliott
16:48:27 <Administrator1> shoot how do I change my name in here?
16:48:32 <Phantom_Hoover> /nick
16:48:49 -!- Administrator1 has changed nick to fox___.
16:48:51 <fox___> woot
16:48:52 <fox___> ty
16:49:56 <fox___> sup phantom_hoover
16:50:26 <Phantom_Hoover> i dunno
16:50:31 <Phantom_Hoover> i only deal with what's left
16:50:41 <fox___> what's left?
16:50:50 <Phantom_Hoover> lamp
16:51:48 <fox___> lol
16:51:49 <fox___> moving?
16:52:05 <Phantom_Hoover> there's no such thing as a moving lamp
16:52:17 <fox___> :P
16:52:28 <fox___> so what do you guys talk about on here?
16:53:09 <Phantom_Hoover> post-structural marxism
16:53:38 <fox___> really?
16:53:46 <fox___> weird
16:53:47 <Phantom_Hoover> yes
16:53:53 <fox___> what is that btw
16:53:53 <Phantom_Hoover> so elliott
16:53:58 <fox___> post structural I mean
16:54:04 <Phantom_Hoover> what's your opinion on the dielectric
16:54:17 <elliott> it's a bit too electric
16:54:22 <elliott> not too keen on the dying, either
16:55:46 <Phantom_Hoover> very true
16:55:58 <fox___> dying?
16:56:39 <fox___> whatcha talking about?
16:56:43 <Phantom_Hoover> fox___, (we did tell you what the channel was about in rainbow colours)
16:56:56 <fox___> ?
16:57:12 <fox___> I know not much of political theory but was confused by the mention of electric-dielectric
16:58:21 <Phantom_Hoover> there's a marxist dialectic, and i live in constant hope that i will one day be able to make a decent pun on it with dielectric
16:58:35 <fox___> lol
17:00:10 <Phantom_Hoover> (this channel is actually about esolangs, which mostly manifests in the form of the offtopic tangents averaging out to being centred on esolangs)
17:00:51 <fox___> right..reading about them now
17:01:15 <fox___> so church of the subgenius applied to extra cryptic programming...nice
17:03:23 <fox___> http://www.subgenius.com/bigfist/pics13/devivals/2007/Balto-07poster-legume.jpg
17:04:00 <Phantom_Hoover> doesn't work as well in high-res, loading from the top-down
17:04:20 <ka7la> so is ICFP competition out yet?
17:05:42 -!- ka7la has quit (Quit: Leaving).
17:08:06 -!- iamfishhead has joined.
17:15:29 -!- mnoqy has joined.
17:31:07 -!- Taneb has quit (Quit: Leaving).
17:31:47 -!- ineiros has joined.
17:42:00 -!- conehead has joined.
17:47:53 -!- zzo38 has joined.
17:51:20 -!- Administrator1_ has joined.
17:52:37 -!- jconn has quit (Ping timeout: 248 seconds).
17:54:48 -!- fox___ has quit (Ping timeout: 260 seconds).
17:56:43 -!- coppro has joined.
18:01:00 -!- Administrator1_ has quit (Quit: Leaving).
18:01:08 -!- Administrator1_ has joined.
18:01:11 -!- Administrator1_ has changed nick to fox___.
18:07:52 -!- epicmonkey has quit (Ping timeout: 276 seconds).
18:08:32 -!- Administrator1_ has joined.
18:08:48 -!- fox___ has quit (Ping timeout: 260 seconds).
18:16:01 <fizzie> Welp, I "fixed" that USB audio stick issue by the ludicrously low-tech manner of just writing a fixed default.pa config that doesn't use the usual approach (module-udev-detect loading module-alsa-card*N) but just configures module-alsa-sinks and module-alsa-sources in a predefined manner; that way it doesn't seem to really touch mixer controls.
18:18:14 * boily stands far, far away from fizzie's PA aura
18:18:14 -!- Taneb has joined.
18:18:44 <fizzie> I tried out the JACK thing, and it... worked, but it wasn't really the best fit for a desktop audio config like this.
18:22:01 <boily> jack is good. jack is your friend.
18:22:58 <fizzie> I don't really know. I need volume control for my headphones, and I don't have any proper audio hardware with knobs and such.
18:23:02 <boily> besides, if you eventually manage to get PA correctly running with multiple audio devices (and some USB ones), ping me please.
18:23:20 <fizzie> Wiring things through jack-mixer "worked", but was very clunky.
18:23:20 <boily> oh. well. then maybe JACK is perhaps a little bit maybe too overkill.
18:23:42 <boily> apart from the pretty splines you get from that, it can get nightmarish enough quickly.
18:25:26 <fizzie> This PA setup seems okay now that I convinced it to not really do too much. When it tried to manage "cards" it was kind of crummy, but now it just has a list of ALSA PCM playback devices as sinks, and capture devices as sources, and that's about it.
18:25:54 <fizzie> Though without the udev autoloading thingie in place, I need to run a pa script after plugging in the stick; perhaps not a great chore.
18:26:28 <olsner> why not just use plain alsa instead of pulseaudio?
18:26:34 <fizzie> Curiously enough, the onboard audio actually has passable quality in this system. On the previous one, with headphones, there was horrible amounts of noise e.g. when moving the mouse around, or esp. windows.
18:27:02 -!- FreeFull has quit.
18:28:33 <fizzie> olsner: Because it's so awkward when I want to move some audio stream from USB-stick headphones to soapbox mini-speakers to S/PDIF out to the big speakers. Every app does its configuration differently; and changes to .asoundrc take place only when they feel like it.
18:29:29 -!- FreeFull has joined.
18:29:35 -!- lifthrasiir has quit (Ping timeout: 245 seconds).
18:30:38 -!- lifthrasiir has joined.
18:30:52 <fizzie> I'd probably go with Alsa and dmix if I just had that one audio device that'd have all these modalities.
18:31:15 <olsner> sounds like you've found an actual usecase for pulseaudio then, congratulations
18:32:02 -!- sacje has joined.
18:32:44 <fizzie> There's some sort of a thing on the onboard sound chip that it can make a copy of the "Default PCM" stream into the S/PDIF hole, so that sort of merges two of the three, but there's still the headphones. (Unless I switch to using the headphone port on the onboard sound, which wasn't an alternative before.)
18:34:35 <boily> or, you could, you know, use a windows machine :P
18:34:47 <boily> (sez the linux guys who works for a FOSS company...)
18:36:06 <fizzie> This machine does in fact have a Windows nowadays, since I had some leftover disks.
18:36:10 <zzo38> Use the designated audio output on the sound card; that ought to work better than using USB, because then you don't get the noise when moving the mouse and whatever.
18:36:53 <fizzie> zzo38: In the old system I had, it was exactly the other way around. The USB device had very good audio quality, while the onboard audio out was noisy as anything, especially when something happened on-screen.
18:37:15 <fizzie> (In this new setup, both seem fine.)
18:38:50 <zzo38> Yes, I have actually heard of both ways in different computers, but maybe the onboard audio was a defective sound card? It would seem that if it is implemented properly, the analog output of the sound card should work. (The operating system might not be designed for any of this, though, necessarily. But I think there is a variant of Linux which is designed for this?)
18:45:57 <kmc> boily: who do you work for?
18:46:09 <kmc> streaming audio over the network is another actual use case for pulseaudio
18:47:24 <kmc> I've enjoyed using this (or another audio daemon? i forgot) to have video playing from the media server on the big projection screen, with audio sent to my headphones via laptop (so I don't disturb housemates)
18:47:45 <kmc> still I think it's absurd that Pulse is the default even for people with the common case of one audio device
18:50:50 <Taneb> huh
18:51:02 <Taneb> In Scandinavia and the World, Finland's shirt is a lot more centred than I thought
18:51:21 <Bike> hi somebody tell me something interesting that has nothing to do with dawkins
18:52:13 <Taneb> Bike, in Scandinavia and the World, characters who would have Nordic crosses on their shirts have them way too centred
18:53:03 <Taneb> For reference, http://satwcomic.com/poster
18:53:26 <Bike> that is weird.
18:53:47 <Bike> what's the black catlooking thing
18:54:27 <fizzie> I kind of like the idea of switching to plain ALSA, and the sound quality seems fine... but the computer's headphone port is pretty much on the floor, while the USB stick is in a hub on the desk, and the cable's more comfortable that way.
18:55:27 <Taneb> Bike, an Icelandic Hellspawn
18:55:38 <Taneb> Because Iceland is right next to hell, don'tchaknow
18:55:48 <Bike> sensible
18:56:30 <Taneb> The thing on Denmark's head is a beer-drinking ghost
18:56:31 <Phantom_Hoover> http://satwcomic.com/worthless
18:56:44 <Phantom_Hoover> you can't assemble the royal seal from 1p coins alone, idiot
18:57:12 <Bike> stop ruining my immersion, comic
18:57:16 <Taneb> Phantom_Hoover, it's written by a Dane. The author slapped the Brit who was explaining it to her then ran away
18:57:52 <Phantom_Hoover> she even says she played around with them when she was in the uk! this is just sloppy work
18:58:45 <kmc> Rust kind of has the same problem as C++ where you write everything twice, in a mutable version and an immutable version
18:58:59 <kmc> except that in C++ you can avoid this just by ignoring const correctness
19:00:31 -!- copumpkin has quit (Ping timeout: 276 seconds).
19:00:40 <Phantom_Hoover> http://en.wikipedia.org/wiki/File:New_British_Coinage_2008.jpg
19:00:42 <Phantom_Hoover> see!
19:00:49 <Phantom_Hoover> also wow those coins look really shiny
19:00:54 <Bike> wow phantom i didn't know you were a HUGE NERD
19:00:59 -!- copumpkin has joined.
19:01:08 <boily> kmc: savoirfairelinux.com.
19:01:13 <Phantom_Hoover> look this is just basic knowledge
19:01:23 * boily slaps Bike with a very nerdish penguin.
19:01:37 <kmc> boily: well you are a Microsoft Certified Partner too
19:02:21 <kmc> I like how I can kind of read French for no clear reason
19:02:26 <kmc> maybe due to knowing some Spanish
19:02:44 <Phantom_Hoover> and the fact that it's a romance language, i mean come on
19:02:57 <kmc> or maybe because computer shit is the same in every language
19:03:09 <kmc> drupal est un blog de GPL en PHP
19:03:19 <Phantom_Hoover> that also
19:03:25 <kmc> PHP est merde
19:04:05 -!- Bike_ has joined.
19:04:07 -!- Bike_ has quit (Client Quit).
19:04:20 -!- Bike_ has joined.
19:04:38 <Bike_> kmc: from practice reading "segfault" in every language
19:04:41 -!- Bike has quit (Disconnected by services).
19:04:42 <kmc> oui
19:04:46 -!- Bike_ has changed nick to Bike.
19:07:06 -!- Bike_ has joined.
19:09:42 -!- Bike has quit (Ping timeout: 256 seconds).
19:10:29 -!- Bike_ has changed nick to Bike.
19:12:34 -!- MindlessDrone has quit (Quit: MindlessDrone).
19:14:14 -!- |_ has joined.
19:15:47 <boily> kmc: d'accord pour le php. la certification microsoft est un détail technique.
19:18:12 -!- clog has quit (Ping timeout: 256 seconds).
19:18:37 -!- clog has joined.
19:19:47 -!- epicmonkey has joined.
19:25:50 -!- yorick has quit (Read error: Connection reset by peer).
19:28:58 -!- clog has quit (Ping timeout: 256 seconds).
19:29:54 -!- apex has joined.
19:32:55 <Taneb> I wonder if I have enough coins to do the coiny thing
19:34:55 <Taneb> Haha I can
19:35:07 <Phantom_Hoover> [S] Taneb: do the coiny thing
19:35:38 <Taneb> :P
19:35:51 <fizzie> IDGI. A keybinding to spawn "scrot -z ..." for a full-screen screenshot works fine; a keybinding to spawn "scrot -s -z ..." just results in "giblib error: couldn't grab keyboard:Resource temporarily unavailable" in .xsession-errors; "scrot -s -z ..." from the command line works just fine and lets me select a rectangle.
19:36:37 <zzo38> kmc: If you can make the mutable and immutable version in Rust, is there some kind of way to use macros to make it do that automatically in some cases?
19:36:57 <fizzie> (And running scrot -z -s from a spawn "dmenu_run" also works.)
19:37:20 -!- clog has joined.
19:38:05 <Taneb> Phantom_Hoover, would that make me the Heir of Cash
19:38:22 <Phantom_Hoover> yes
19:39:04 <Bike> deep
19:39:18 -!- yorick has joined.
19:39:22 <Phantom_Hoover> aaaargh i can't handle all this bureaucracy
19:39:46 <Taneb> I told you dog. I warned you about the bureaucracy.
19:39:50 <fizzie> Well, that's stupid; apparently spawn "sleep 0.1; scrot -s -z ..." works; the select-a-rectangle mode just doesn't like working if it's started when the hotkey is still down.
19:41:23 <Phantom_Hoover> at least i know my national insurance number this time
19:43:30 <kmc> zzo38: sometimes
19:44:37 -!- clog has quit (Ping timeout: 248 seconds).
19:47:01 -!- clog has joined.
19:47:45 -!- apex has quit (Quit: Leaving).
19:49:44 -!- ap3x has joined.
19:56:54 <Phantom_Hoover> my emergency contacts list somehow has my father listed in both slots, in one of them as my daughter
19:58:35 <Taneb> ...
19:58:57 <Taneb> Time travel and future technology gender reassignment surgery?#
19:59:12 <Taneb> Phantom_Hoover, you're your own grandfather#
19:59:22 <Taneb> Why do I keep hitting hash before enter#
19:59:46 <Phantom_Hoover> apparently
20:00:37 <boily> your keyboard are weird.
20:00:53 <Phantom_Hoover> Taneb, 'widefingeritis'
20:01:09 -!- katla has joined.
20:01:19 <Taneb> I think it's inaccurateandslippyfingeritis
20:01:21 <Phantom_Hoover> oh shit, they want to know where i'm living next year
20:02:08 <Taneb> Hexham
20:02:10 <Taneb> CA
20:02:12 <Taneb> Finland
20:02:14 <Taneb> Aaaah
20:02:36 <katla> is ICFP out yet
20:03:42 <boily> Phantom_Hoover: don't worry, I want to know where you live this year.
20:03:52 <boily> katla: when the Infamous Interview will be out too.
20:04:09 <Phantom_Hoover> boily, so do I
20:04:20 <boily> `quote FN35EM
20:04:22 <HackEgo> No output.
20:04:35 <boily> hm. looks like I had never mentioned where in montreal I live.
20:04:37 <olsner> `pastelogs FN35EM
20:04:58 * boily can hear HackEgo churning and grinding...
20:04:58 <katla> what?
20:05:23 <HackEgo> http://codu.org/projects/hackbot/fshg/index.cgi/raw-file/tip/paste/paste.7170
20:05:44 <boily> oh. I did.
20:33:37 -!- oerjan has joined.
20:35:32 <oerjan> <Phantom_Hoover> you never know when the six pigs will begin their forage <-- wat
20:35:54 <Phantom_Hoover> shh! do not speak of the porcine ones too loud
20:36:39 <oerjan> are they related to these http://www.girlgeniusonline.com/comic.php?date=20130805
20:37:35 <Phantom_Hoover> no
20:37:44 <oerjan> fiendish
20:37:50 <Phantom_Hoover> hint: where does Taneb live
20:38:01 <oerjan> ooh
20:38:26 <oerjan> the hexham apocalypse, i take?
20:40:33 <Bike> six pigs
20:40:57 <oerjan> also by wikipedia forage is either plant material or fish, so i think we should be safe.
20:49:03 <fizzie> Huh, the "really" thing (of postgresql 9.3+142really9.1+146) seems to be a full-blown Debian-testing/unstable convention; here's yafaray 0.1.2+really0.1.2~beta5-5+b1.
20:50:27 <Phantom_Hoover> oerjan, to Them Of Whom The Jews Do Not Eat we are but fish
20:51:31 <oerjan> i suppose i _do_ recall some stories of pigs eating people.
20:53:14 -!- Administrator1_ has quit (Quit: Leaving).
20:58:05 -!- Bike_ has joined.
20:58:14 -!- Bike has quit (Ping timeout: 240 seconds).
21:01:23 <oerjan> `addquote <Phantom_Hoover> my emergency contacts list somehow has my father listed in both slots, in one of them as my daughter
21:01:27 <HackEgo> 1084) <Phantom_Hoover> my emergency contacts list somehow has my father listed in both slots, in one of them as my daughter
21:01:31 <oerjan> a moment in #esoteric.
21:01:38 -!- Bike_ has changed nick to Bike.
21:09:29 -!- boily has quit (Quit: Poulet!).
21:09:34 -!- metasepia has quit (Remote host closed the connection).
21:50:19 -!- katla has quit (Quit: Leaving).
22:00:07 -!- |_ has quit.
22:03:53 <Bike> http://betabeat.com/2013/08/app-will-ping-you-when-its-time-to-bang/ cool
22:04:54 -!- epicmonkey has quit (Ping timeout: 264 seconds).
22:06:58 <mnoqy> kill earth
22:07:34 <zzo38> Why? How?
22:07:57 <oerjan> mnoqy is angry because he bumped his toe on it hth
22:08:22 <zzo38> Then you have to be more careful next time.
22:09:21 <kmc> Bike: wow
22:22:17 -!- Taneb has quit (Quit: Leaving).
22:23:55 -!- doesthiswork has joined.
22:27:51 -!- sebbu has quit (Ping timeout: 240 seconds).
22:32:53 -!- sebbu has joined.
22:36:31 -!- sacje has quit (Ping timeout: 240 seconds).
22:45:25 -!- copumpkin has quit (Ping timeout: 248 seconds).
22:46:02 -!- copumpkin has joined.
22:50:38 -!- nooodl has quit (Ping timeout: 240 seconds).
23:00:48 -!- sacje has joined.
23:05:16 <shachaf> kmc: exciting new rust runtime
23:06:10 <kmc> yes!
23:06:15 <kmc> as of today, Rust's runtime is written in Rust
23:06:20 <kmc> not many languages can do that
23:06:29 <Bike> oh that is nice.
23:06:29 <Lumpio-> Except most compiled languages
23:06:30 <Fiora> is gcc's runtime written in C?
23:06:37 <Bike> libc is in C, yes
23:06:44 <Bike> well, probably
23:06:44 <Lumpio-> Most actually usable compiled languages at least.
23:06:45 <kmc> Lumpio-: I don't know about that. Not Haskell or OCaml, for sure
23:07:03 <kmc> I wouldn't call libc the "runtime", that would be like crt*.o and libgcc.a
23:07:07 <Fiora> and C++ is in C++ and I guess java is written in java?
23:07:24 <Lumpio-> Java isn't compiled to native code so I doubt the JVM is written in Java
23:07:25 <Bike> i... actually don't know what jvms are written in usually
23:07:27 <kmc> you can build a C program without libc and ~all of the "language features" still work
23:07:29 <Lumpio-> It's possible if you JIT it in advance but
23:07:31 <Lumpio-> Doubtful.
23:07:38 <Bike> kmc: i don't even know what libgcc is
23:07:42 <kmc> sometimes Java is compiled to native code
23:07:43 <kmc> but not usually
23:07:59 <Lumpio-> What's libgcc do anyways
23:08:01 <kmc> Bike: well for example, if you use uint64_t on a 32-bit machine, it needs to call some function to add double-size integers
23:08:06 <Bike> oh. that stuff.
23:08:13 <kmc> so there are things like that which look like primitives but actually call into functions
23:08:13 <Fiora> __udivdi3 and friends
23:08:14 <Bike> the not-in-libc-but-needs-software-support-stuff.
23:08:30 <Lumpio-> Such as?
23:08:32 <shachaf> what counts as the "runtime" is a bit of a definitions game
23:08:37 <Bike> such as what kmc just said?
23:08:56 <Lumpio-> But what primitives are there if you don't use libc nor the magic builtins
23:09:10 <Lumpio-> To make it simpler, say you only call functions you wrote yourself.
23:09:12 <Bike> 64-bit ints
23:09:16 <Lumpio-> oh
23:09:17 -!- doesthiswork has quit (Quit: Leaving.).
23:09:18 <Lumpio-> hum.
23:09:19 <Bike> on a 32-bit machine
23:09:22 <Bike> like kmc said
23:09:36 <Fiora> Bike: that's an implementation detail though, I guess? like, gcc could inline it if it wanted, right?
23:09:37 <Lumpio-> I guess that's a point
23:09:46 <Lumpio-> It could and it probably does
23:10:26 <Bike> it could, there are other routines though sometimes
23:10:38 <kmc> it does sometimes
23:10:45 <Fiora> complex number support, right?
23:10:56 <Bike> plus inlining can lead to code bloat bla bla bla
23:10:58 <kmc> on some machines floating point will be a library call
23:11:01 <kmc> or integer multiply even
23:11:01 <Lumpio-> On an actual computer is it the job of the operating system to zero out .bss and whatnot
23:11:10 <kmc> yes
23:11:16 <Fiora> softfp~
23:11:21 <Lumpio-> I've only looked at full disassemblies of executables for microcontrollers
23:11:22 <kmc> or rather, anonymous mappings are zeroed by default
23:11:33 <Lumpio-> Where GCC emits a small piece of code to do that because there's no OS.
23:11:37 <kmc> for security -- you can't allow leftover data from some other program
23:11:38 <Lumpio-> ...or I guess that could come from some sort of library.
23:12:03 <zzo38> How does it zero out the BSS? Is there hardware support for doing that?
23:12:15 <Lumpio-> memset() ¬u¬
23:12:16 <zzo38> Or does it have to write every byte to zero individually?
23:12:23 <Bike> doesn't it just map a pre-zeroed page if possible, or whatever
23:12:30 <Lumpio-> memset() and friends are usually optimized to ridiculous levels
23:12:40 -!- carado has quit (Ping timeout: 246 seconds).
23:12:45 <Lumpio-> There's probably an assembly implementation for all popular platforms.
23:12:57 -!- conehead has quit (Quit: Computer has gone to sleep.).
23:12:59 <Lumpio-> It rarely goes byte by byte, on a 64-bit machine you can do at least 8 bytes at a time
23:13:13 <Lumpio-> And I wouldn't be surprised if x86 had some sort of block fill extension.
23:13:30 -!- conehead has joined.
23:13:31 <kmc> Lumpio-: you'll find various unrolled SSE memsets and such
23:13:38 <Lumpio-> yes
23:13:40 <zzo38> Is there hardware support for the entire RAM at once in something outside of the CPU?
23:13:50 <Lumpio-> Entire RAM at once?
23:13:52 <Lumpio-> That would be suicidal
23:13:52 <Bike> why would you want to zero all of ram at once
23:13:52 <kmc> I think Ivy Bridge has a special optimization such that the lowly REP STOSB is actually really fast
23:14:04 <kmc> I know they have it for memcpy() / REP MOVSB, but with some wacky restrictions
23:14:08 <zzo38> Lumpio-: Yes, or a very large block such as 1GB, or something like that
23:14:21 <kmc> Bike: yeah, you can map a copy-on-write zeroed page
23:14:40 <kmc> then if processes reserve lots of .bss and never use it, they don't take up physical memory
23:14:43 <kmc> I think that's typically done
23:14:48 <Bike> right
23:15:03 <zzo38> For example, if the loader is in ROM and the data it keeps track of is stored in registers, then you might want to clear the entire RAM at once before the program is loaded.
23:15:20 <Lumpio-> zzo38: On what kind of 1980 single-tasking computer?
23:15:20 <kmc> (well, you could do that anyway, but this way they can allocate a lot of .bss and read zeroes from it and still not take up physical memory)
23:17:07 <Bike> Lumpio-: are you not used to zzo
23:17:19 <zzo38> Lumpio-: I am not really thinking of a specific kind of computer, but rather if there is a DRAM or something that has a pin for such a function, so that a computer can be made having a BIOS that can do this.
23:17:47 -!- carado has joined.
23:19:27 <Lumpio-> Bike: But I like talking with him!
23:19:39 <Lumpio-> zzo38: Yes, I'm sure it's possible to do.
23:19:56 <Bike> well sure
23:19:58 <zzo38> Do you know which ones have such things?
23:20:01 <Bike> you just occasionally get this
23:20:08 <Lumpio-> I didn't say I knew of any implementation
23:20:15 <Lumpio-> Just that it's possible.
23:23:15 <zzo38> I would like to know which RAM chips do this so that I can make something using this.
23:23:53 <Lumpio-> Not sure if any existing chips have this feature.
23:24:06 <Lumpio-> All you really need is a thing to pull all select lines high at once
23:24:40 <Lumpio-> Usually there's no direct access to the select lines.
23:26:41 <Lumpio-> Hmm, although is the DRAM controller itself on the motherboard on modern computers
23:26:47 <Lumpio-> Instead of the chips themselves
23:26:57 <Lumpio-> If that's so all you need is a soldering iron and a steady hand!
23:27:08 <Bike> i feel like zzo should get a few fpgas to mess with
23:27:12 <Lumpio-> ...and probably magic wire that doesn't introduce extra inductance/capacitance to the really high frequency bus
23:28:13 <shachaf> `olist (909)
23:28:14 <HackEgo> olist (909): shachaf oerjan Sgeo FireFly
23:29:33 <Fiora> I think RAM controllers have been on cpus for a while now?
23:32:31 <kmc> yeah for a long time
23:33:16 <pikhq> Lumpio-: The RAM bus is not high frequency.
23:33:25 <pikhq> It's 200 MHz.
23:34:12 <pikhq> (or less)
23:34:29 <Lumpio-> oh right
23:34:32 <Lumpio-> It's parallel
23:34:47 <Lumpio-> Fiora: Well, outside the RAM modules themselves at any rate.
23:35:39 -!- Sgeo has joined.
23:36:22 <pikhq> "PC3-6400" is a mere 100 MHz, it just has the bandwidth of a hypothetical PC-100 module clocked at 6.4 GHz.
23:36:36 <Sgeo> `olist 909
23:36:38 <HackEgo> olist 909: shachaf oerjan Sgeo FireFly
23:37:29 <pikhq> Oh, sorry, the RAM *bus* is at 400MHz.
23:37:42 <pikhq> But that's feeding into 100MHz chips.
23:38:08 <pikhq> Still, not the highest frequency bus.
23:39:22 <shachaf> Sgeo..........................................
23:39:53 <Bike> hey, he tried, shachaf.
23:39:58 <pikhq> I think in terms of clock rate USB 2.0 is higher freq?
23:40:49 <Sgeo> At least I indicated which one so you didn't think there were two updates within 8 minutes of each other
23:46:39 <Phantom_Hoover> humble bundle: now dropping the pay what you want component
23:47:44 <Phantom_Hoover> or... apparently not
23:47:47 <Bike> so we only pay what we don't want?
23:51:27 <shachaf> Bike: true
23:51:35 * oerjan feels thoroughly `olisted
23:51:37 <shachaf> Sgeo: true
23:52:12 <kmc> i will pay what Phantom_Hoover wants
23:53:33 <Phantom_Hoover> $300 (you choose the dollar)
23:53:39 <Bike> the elliott system
23:53:46 * oerjan is wondering what they'll do when the protection spell runs out (and the spare in the staff)
23:53:50 <Bike> also: it's weird to realize that horse_ebooks is actually a very successful spam marketer
23:54:07 -!- doesthiswork has joined.
23:54:07 -!- iamfishhead has quit (Ping timeout: 260 seconds).
23:54:26 <shachaf> imo someone should make nlab_ebooks
23:55:02 <Bike> kmc: http://24.media.tumblr.com/tumblr_m92seuPm161r9hh3yo1_400.jpg update re: gay cuttlefish
23:55:20 <mnoqy> those aren't cuttlefish
23:55:42 -!- yorick has quit (Read error: Connection reset by peer).
23:55:44 <shachaf> mnoqy: do you happen to know when super mega updates usually happen
23:55:49 <shachaf> when is the evening
23:55:55 <shachaf> what time zone is johnny smash in
23:56:01 <Bike> about nine pounds
23:56:05 <mnoqy> i think he lives in arizona?
23:56:18 * oerjan swats Bike, disturbingly gently -----###
23:56:35 <Bike> hi
←2013-08-07 2013-08-08 2013-08-09→ ↑2013 ↑all