00:00:56 <oerjan> also, that i set it to warn me. it's still slightly annoying that it sets a scheduled time by itself.
00:03:32 <oerjan> <int-e> is it one of those "the only way to win is not to play" deals? <-- i simply interpret the Game as "the only way to win is not to care" and that solves all the problems hth
00:03:49 <int-e> Well, it's Microsoft's computer; you should be happy that they let you use it!
00:04:42 <int-e> (only half kidding. sigh.)
00:05:29 -!- boily has joined.
00:05:31 <lambdabot> CYUL 092300Z 24014KT 15SM BKN057 OVC090 10/05 A2976 RMK SC7AC1 SLP082
00:07:26 <oerjan> int-e: it is now evening
00:07:47 <oerjan> it seems that they were at least prepared for the poison
00:08:29 <oerjan> have i mentioned that ENVA is pretty close to HELL
00:12:28 <int-e> work, bridges, go match, new spaceship... who has times for comics with so much else going on ;)
00:13:08 <oerjan> https://www.google.no/maps/@63.45201,10.9148617,14z?hl=no
00:13:24 <boily> today's headlines: work bridges go match new spaceships.
00:13:46 <oerjan> boily: the airport is ENVA, then look a little below to the left hth
00:14:22 <int-e> oerjan: http://www.conwaylife.com/wiki/Copperhead https://niginsblog.wordpress.com/2016/03/07/new-spaceship-speed-in-conways-game-of-life/
00:14:46 -!- mihow has quit (Quit: mihow).
00:15:27 <int-e> okay, so, yeah, they know about the poison. nothing much going on, a bit of exposition... moving on :)
00:16:00 <int-e> well, exposition... more of a flashback without a flashback. a reminder of what happened back then, perhaps.
00:18:03 <boily> southwest from the word Trondheim on Google Maps.
00:18:19 <boily> but then I searched for the airport. it's not in Trondheim.
00:18:34 <boily> ah! Hell. it is found.
00:18:38 <oerjan> boily: i tried to link you above. did it not give the right place?
00:19:02 <boily> I didn't see the link.
00:19:12 * boily fails spot checks regularly
00:21:56 <boily> . o O ( quick, a believable excuse! anything! )
00:22:06 <boily> uh... uhm... I... I'm eating a sandwich!
00:22:26 -!- p34k has quit.
00:22:44 <int-e> I have 21 lines of scrollback...
00:23:24 <boily> I think screen/tmux each have their own scrollback buffers.
00:23:40 <boily> they may involve interesting shortcuts to invoke, though.
00:23:54 <oerjan> sandwiches are well known to affect spot checks, indeed
00:24:17 <oerjan> boily: ^B PgUp in tmux
00:24:22 <int-e> well, it's irssi which has its own backlog as well... but I need to know to look for something first.
00:24:41 <oerjan> also, you need to press q to leave the mode
00:24:51 <int-e> anyway, I was actually trying to suggest a line of excusing for boily, not for myself...
00:25:14 * int-e isn't sure how to nounify that verb.
00:25:20 <oerjan> excuse is a perfectly cromulent noun
00:25:52 <int-e> but it's not the process of making up an excuse.
00:25:53 <boily> those are sane shortcuts. I'm more and more tempted to switch over to tmux...
00:28:03 <HackEgo> entschuldigung? ¯\(°_o)/¯
00:28:31 <oerjan> `learn Entschuldigung is the German word for blaming something on trees moving by themselves.
00:28:38 <HackEgo> Learned 'entschuldigung': Entschuldigung is the German word for blaming something on trees moving by themselves.
00:28:44 <int-e> yorick: I don't know, did Shakespeare write about you?
00:29:10 <int-e> oerjan: an "ent" pun?
00:29:28 * yorick resets the alas counter
00:30:39 <int-e> yorick: so I expected that fungot's shakespeare style might know you... unfortunately, fungot doesn't really take context into account, so it's hard to find out in practice
00:31:48 <int-e> yorick: and to be honest, I was at that moment unaware that it's also a (namely your) nickname here :P
00:32:56 -!- fungot has joined.
00:33:13 <oerjan> `le/rn yorick/We know nothing about yorick, alas.
00:33:15 <fungot> fizzie: any slight tips this time?
00:33:21 <fizzie> That server's being real slow.
00:33:31 <fizzie> fungot: Here's a pro tip: don't eat the yellow snow.
00:33:31 <fungot> fizzie: a person would connect to a +5v dc power supply, just need to hope the kernel kills that
00:33:41 <boily> fungot: o hai. are you eating a sandwich? as a matter of fact, do you eat?
00:33:42 <fungot> boily: good evening :)
00:34:13 <fungot> Selected style: ss (Shakespeare's writings)
00:34:13 <fungot> boily: val. no valentine indeed, for i am thee; thee will i loue, i gaue thee villaine? the pigge quoth i, this generall applause, and chearefull showt, argues your wisdome, and your royall court, by heauen
00:34:28 <boily> I thought I had smothered him...
00:34:36 <fizzie> I think something's up with wolfe.freenode.net.
00:34:54 <oerjan> fungot: is there someone you knew well?
00:35:54 <prooftechnique> I'm just thrilled we get to add hellorick to the repertoire
00:36:02 <fizzie> Well that's just ricudilous.
00:36:53 <fizzie> It hasn't even seen 'rjan's question yet, if the debug logging is to be believed.
00:37:37 <boily> prooftechellonique.
00:37:52 <fizzie> I think I'll swap them servers.
00:37:53 -!- fungot has quit (Remote host closed the connection).
00:38:02 <fizzie> That happened instantaneously. Weird.
00:38:34 -!- fungot has joined.
00:38:38 <fungot> fizzie: boy. if to doe were as easie as a downe bed would affoord it
00:38:55 -!- ais523 has joined.
00:38:56 <fungot> int-e: banks. take me vp, and they are pale; she takes him by the nose
00:39:08 <ais523> !bfjoust is_this_working_yet <
00:39:11 <EgoBot> Score for ais523_is_this_working_yet: 0.0
00:39:26 <oerjan> fungot: so any old acquaintances?
00:39:26 <fungot> oerjan: duke. be it known, as the glasses where they view themselues, which in their throng and press to that last hold, confound themselves. ' tis most meet you should wear it in my arm.
00:39:41 <oerjan> ais523: Lymia managed to fix it with shellshock
00:40:14 <boily> wait. Lymia shellshocked one of freenode's servers???
00:40:26 <oerjan> boily: no, just EgoBot
00:40:26 <ais523> if the bot is still shellshock-vulnerable, it probably shouldn't be on IRC
00:41:14 <oerjan> ais523: the trick is to get Gregor to awake and update both bash and gearlance
00:41:42 <oerjan> because it seems the basic problem was a bug in old gearlance
00:42:13 * boily mapoles Gregor a few times to rouse him from his slumber
00:42:32 <oerjan> Gregor is not known to slumber lightly
00:44:10 <int-e> In his house at R'lyeh, Gregor waits dreaming.
00:44:16 <fizzie> That is not dead which can ... COME ON
00:46:12 <oerjan> since he recommended that on a previous occasion
00:46:23 -!- staffehn_ has joined.
00:48:19 -!- deltab_ has joined.
00:49:04 -!- Vorpal has quit (Ping timeout: 248 seconds).
00:49:05 -!- clog has quit (Ping timeout: 248 seconds).
00:49:05 -!- deltab has quit (Ping timeout: 248 seconds).
00:49:05 -!- feliks has quit (Ping timeout: 248 seconds).
00:49:05 -!- carado has quit (Ping timeout: 248 seconds).
00:49:06 -!- staffehn has quit (Ping timeout: 248 seconds).
00:49:14 -!- Vorpal has joined.
00:49:14 -!- Vorpal has quit (Changing host).
00:49:14 -!- Vorpal has joined.
00:50:05 -!- carado has joined.
00:51:14 * oerjan assumes that split was wolfe croaking...
00:53:09 <zzo38> I added a few more cursor shapes, now there is arrow with ... written on it, and there is magnification (plain, minus, and plus), and also a stop sign icon.
00:53:52 <zzo38> What other mouse cursor icons do you expect should be needed (not counting the ones in the standard X cursor font)?
00:55:33 -!- feliks has joined.
00:59:35 <boily> hezzo38. left hand cursor?
01:00:46 <HackEgo> U+270A RAISED FIST \ UTF-8: e2 9c 8a UTF-16BE: 270a Decimal: ✊ \ ✊ \ Category: So (Symbol, Other) \ Bidi: ON (Other Neutrals) \ \ U+270B RAISED HAND \ UTF-8: e2 9c 8b UTF-16BE: 270b Decimal: ✋ \ ✋ \ Category: So (Symbol, Other) \ Bidi: ON (Other Neutrals) \ \ U+270C VICTORY HAND \ UTF-8: e2 9c 8c UTF-16BE: 270c Decimal: c
01:01:33 <zzo38> boily: There is already hand cursor included in the standard cursor font, in both direction
01:01:42 <HackEgo> U+1F595 - No such unicode character name in database \ UTF-8: f0 9f 96 95 UTF-16BE: d83ddd95 Decimal: 🖕 \ 🖕 (🖕) \ Uppercase: U+1F595 \ Category: Cn (Other, Not Assigned)
01:03:05 <int-e> HackEgo doesn't know unicode 8.0 yet.
01:04:15 <int-e> oh well it wasn't a very polite one anyway. "REVERSED HAND WITH MIDDLE FINGER EXTENDED"
01:07:29 <\oren\> I added a sodium memory chip. hopefully now I have 12 GB of memory
01:08:30 <\oren\> related, I figured out how to use a metal file to turn a big screwdriver into a compuetr-sized one
01:08:49 -!- mihow has joined.
01:10:08 -!- adu has joined.
01:10:09 -!- pikhq has joined.
01:17:11 <boily> he\\oren\. sodium memory chip? you salted your machine?
01:17:55 -!- deltab_ has changed nick to deltab.
01:19:24 <ais523> when's the next alphago match?
01:20:32 -!- mad has joined.
01:21:32 -!- adu has quit (Quit: adu).
01:23:16 <\oren\> for some reason, laptops use lithium batteries and sodium memeory
01:24:43 <\oren\> I bet kerbal space program will be faster now
01:29:28 <boily> meanwhile, damned teasing abyssal rune. "You are suddenly pulled into a different region of the Abyss!".
01:29:51 <boily> mon cul que t'as le droit de me faire ça juste quand chu pour arriver dessus.
01:30:15 -!- Phantom_Hoover has quit (Read error: Connection reset by peer).
01:31:37 <\oren\> actually, ksp isn't that memory taxing. I should really be testing dwarf fortress!
01:32:06 <\oren\> heh. or, why not both!
01:34:04 -!- nisstyre has quit (Ping timeout: 264 seconds).
01:38:41 <HackEgo> [wiki] [[Vitsy]] N https://esolangs.org/w/index.php?oldid=46566 * VTCAKAVSMoACE * (+19745) Vitsy Esoteric Programming Language
01:45:07 <HackEgo> [wiki] [[Vitsy]] https://esolangs.org/w/index.php?diff=46567&oldid=46566 * VTCAKAVSMoACE * (+570)
01:46:23 <HackEgo> [wiki] [[Vitsy]] M https://esolangs.org/w/index.php?diff=46568&oldid=46567 * VTCAKAVSMoACE * (+65)
01:48:13 -!- XorSwap has quit (Quit: Leaving).
01:52:48 <oerjan> @tell tswett <tswett> My neural net thinks there's a historical language called Middle Worse. <-- PLAUSIBLE
01:54:19 <oerjan> <ais523> when's the next alphago match? <-- iirc in 2 hours?
01:54:45 <ais523> wasn't sure if it was a matter of hours, days, or weeks
01:54:58 <ais523> the previous match was really worth watching
01:56:44 <mad> now we need a new game that we can beat computers at
01:57:03 <ais523> well humans still beat computers at BF Joust
01:57:20 <ais523> I only know of one computerized Nomic player and it was terrible
01:57:37 <oerjan> fungot: is that what you want us to think?
01:57:37 <fungot> oerjan: enter oswald the steward. how faine would i goe to, carry this vnto her graue? be buried quicke with her, and fnord thee after supper, of a weak and silly mind fnord to wail his death.
01:57:44 <ais523> that said, it worked on the same principle as fungot's babble output, so it's perhaps not surprising that it did badly
01:57:44 <fungot> ais523: lys. one turfe shall serue as pillow for vs both, and for that i have not the book of honour razed quite, and laid mine honour too fnord on't: the fool will be hang'd else
01:57:55 <fungot> Selected style: agora (a large selection of Agora rules, both current and historical)
01:58:07 <ais523> fungot: how would you do as a Nomic player?
01:58:07 <fungot> ais523: all penalties shall be the set of valid ballots on an entity has an initiator, or otherwise defaults to 1.0.
01:58:07 <oerjan> fungot: what does graue have to do with this
01:58:07 <fungot> oerjan: ( a) scamster, which must be the entire day of each color of mark is a person is reduced by one or more win conditions are satisfied. such a notification to that creditor according to their category.
01:58:30 <ais523> hmm, we need to get one of those new neural net things to look at Agora proposals
01:58:33 <ais523> and come up with some of its own
02:00:47 <mad> grr... this is frustrating... I have the knowledge to reinvent the synthesizer (music instrument)... but I don't know in what direction to reinvent it
02:01:34 <zzo38> If up doesn't work, then sideways
02:02:08 -!- adu has joined.
02:02:16 <prooftechnique> Failing that, consider gaining the knowledge to invent a new direction, and use that one
02:04:06 <mad> once you accept the standard 12-key per octave keyboard, velocity, pitch bender and mod wheel setup it kindof all turns into that kind of 90's mix-of-every-pop-instrument thing that doesn't bring anyting new to the table
02:04:45 <zzo38> Then use microtonal instruments
02:05:04 <mad> I've looked into that as well
02:05:09 <mad> it's kinda futile
02:05:40 <prooftechnique> The keytar already happened, so consider melding it with another instrument. A theremin with a keyboard could change the musical landscape
02:05:52 <HackEgo> [wiki] [[Vitsy]] M https://esolangs.org/w/index.php?diff=46569&oldid=46568 * VTCAKAVSMoACE * (+3) Fixed table break
02:06:08 <mad> prooftechnique : you mean like an ondes martenot?
02:07:21 <zzo38> There is analog and digital synthesizer, there is hardware and software synthesizer, so which would be? Or something that doesn't match?
02:09:30 <mad> that's a much easier proposition than hardware and the only thing it's missing is input specific to that particular synth
02:10:42 <prooftechnique> I find the 80s German industrial scene did a lot of good work when it came to finding specific inputs for things
02:11:01 <prooftechnique> Maybe your synthesizer should interpret input from pieces of metal being beaten on
02:11:57 <mad> isn't that called a vibraphone? :D
02:12:30 <HackEgo> [wiki] [[Vitsy]] https://esolangs.org/w/index.php?diff=46570&oldid=46569 * VTCAKAVSMoACE * (+5) Fixed a problem with the table.
02:12:34 <prooftechnique> It is practically impossible to invent a new instrument, it seems
02:13:03 <HackEgo> The pidgin hole principle states that if n+1 messages are sent over n protocols, then at least two messages are sent over the same protocol.
02:13:49 <\oren\> I made a terrible program to write music with once
02:14:05 <zzo38> \oren\: What program is that?
02:14:41 <mad> prooftechnique : well, it's easy to invent a new _irrelevant_ instrument :D
02:14:43 <\oren\> zzo38: it was a set of perl and C programs that transformed a weird language into a wav file
02:15:22 <\oren\> nayway, it used exact fracitons for tuning, instead of letters
02:15:38 <mad> prooftechnique : when they make it easy to play in tune
02:15:39 <zzo38> O, so it is writing just intonation?
02:16:01 <mad> you can write just intonation with letters
02:16:42 <mad> basically you have to use 3:2 just fifths and violate the usual rules of writing
02:17:03 <mad> Like using C Fb G as a major chord gives you just intervals
02:17:07 <\oren\> so on each note, you write the fraction that it makes with the previous note
02:18:41 <\oren\> I forget the syntax I used for that, but I got it to play a simple tune
02:18:55 <\oren\> I'll try to find it sometime
02:19:26 <zzo38> I have done just intonation before with DATA commands in BASIC
02:21:28 <mad> \oren\ : it's not very hard to translate note names to fractions
02:22:13 <mad> you just have to decide for every 3rd if you want it to be just (5/4) or made out of a superposition of 2 whole tones (81/64)
02:26:04 <mad> For western music the only real intervals are the octave (2/1), the fifth (3/2), whole tone (9/8) and just major 3rd (5/4)
02:26:12 <mad> everything else is a byproduct of those
02:27:12 <mad> the semitone is the difference between 3 whole tones and a fifth for instance
02:38:26 <HackEgo> [wiki] [[Vitsy]] M https://esolangs.org/w/index.php?diff=46571&oldid=46570 * VTCAKAVSMoACE * (+0) Fix typo
02:41:23 -!- boily has quit (Quit: SPEEDY CHICKEN).
02:43:00 -!- clog has joined.
02:58:06 -!- Alejandro15 has joined.
02:58:16 -!- Alejandro15 has left.
03:00:29 <Lymia> !sh stat /bin/bash
03:00:30 <EgoBot> File: `/bin/bash' \.Size: 907184..Blocks: 1784. IO Block: 4096 regular file \ Device: ca01h/51713d Inode: 359597. Links: 1 \ Access: (0755/-rwxr-xr-x) Uid: ( 0/ UNKNOWN). Gid: ( 0/ UNKNOWN) \ Access: 2016-03-10 00:39:08.000000000 +0000 \ Modify: 2009-08-23 17:18:23.000000000 +0000 \ Change: 2009-11-05 19:51:30.000000000 +0000
03:00:56 <ais523> Lymia: what am I meant to deduce from this?
03:04:46 <Lymia> I'm pretty sure it's a chroot that Gregor didn't think to update. ;c
03:09:49 <\oren\> I've sent a email to the unicode mailing list about the gaps in the Mathematical Alphanumeric Symbols block
03:10:33 <\oren\> I see no reason why these couldn't be defined as decomposing to the equivalent letters in Letterlike Symbols
03:12:05 -!- XorSwap has joined.
03:14:20 <\oren\> having gaps in contiguous ranges is the flaw that made programmers hate EBCDIC
03:17:22 <izabera> "we're very sorry for this bug. this will be changed immediately and everyone will have to adapt"
03:19:43 <prooftechnique> \oren\: Did you drop the cirled Latin letters from your font?
03:20:21 <mad> ebcdic caused C digraphs
03:20:41 -!- puckipedia has quit (Read error: Connection reset by peer).
03:20:41 <mad> so ebcdic is a net negative in my life no matter its merits or flaws
03:21:00 -!- puckipedia has joined.
03:21:56 <mad> if they added BCD support to C then ebcdic would be easier to process
03:22:16 <\oren\> prooftechnique: no not yet. I'm going to move those characters to the script math
03:23:07 <prooftechnique> \oren\: Ah, okay. I was hoping for caps versions, but more math scripts can't hurt :)
03:23:26 <\oren\> I'm planning on making the capitals too of course
03:23:55 <prooftechnique> This font has taken over my entire setup :D Emacs, terminal, Firefox
03:24:32 <mad> craziest idea
03:24:58 <mad> computer that uses BCD with 10 bit words, and each word encodes 0..999
03:25:15 <mad> double word would encode 0.999999
03:25:51 <\oren\> that would be way more efficient than 8 bit BCD
03:26:55 <\oren\> 1000/1024 versus 100/256. night and day
03:27:47 <\oren\> wait I think you have to divide the logarithms? or something?
03:28:17 <lambdabot> arising from a use of ‘show_M90262971364701015908444’
03:28:24 -!- mihow has quit (Quit: mihow).
03:28:42 <lifthrasiir> \oren\: you should look at https://en.wikipedia.org/wiki/Chen%E2%80%93Ho_encoding
03:28:52 <lifthrasiir> and also https://en.wikipedia.org/wiki/Densely_packed_decimal
03:29:17 <\oren\> `` perl -e 'print (log 1000)/(log 2)'
03:30:28 <\oren\> `` perl -e 'print (log 1000)'
03:30:42 <\oren\> `` perl -e 'print ((log 1000)/(log 2))'
03:31:09 <\oren\> `` perl -e 'print ((log 100)/(log 2))'
03:31:26 <\oren\> `` perl -e 'print ((log 1000)/(log 1024))'
03:31:34 <\oren\> `` perl -e 'print ((log 100)/(log 256))'
03:34:15 <mad> http://sites.ieee.org/scv-cs/files/2013/03/Right-SizingPrecision1.pdf <- paper on variable sized floating point numbers
03:34:35 <mad> (and suggesting that it could be the future of computation)
03:47:11 <izabera> mad: i thought digraphs/trigraphs were added to support keyboards that lacked those keys
03:48:27 <pikhq> No, they're specifically for charsets missing them.
03:48:46 <pikhq> Specifically, *some* EBCDIC code pages.
03:49:30 <mad> it's because IBM has a seat on the standards comittee
03:49:34 <pikhq> (EBCDIC code pages all have a common subset, but that common subset does *not* include the trigraph/digraph characters. Worse, the trigraph/digraph characters are encoded differently in the code pages where they exist!)
03:50:16 <mad> also keyboards are purposely designed to have all 32 ascii punctuation signs
03:50:23 <mad> sometimes in kinda roundabout ways
03:50:30 <pikhq> It's even sillier when you consider that you can only use C on IBM mainframes from within the POSIX environment, which by necessity must have the characters in question.
03:50:47 <mad> like ` (accent deadkey for typing èàù) + space
03:50:51 <pikhq> (POSIX has a hard requirement for them)
03:50:56 <mad> same for ^
03:51:17 <pikhq> That might be true of some keyboards.
03:51:43 <pikhq> On US keyboards, all of the printable ASCII characters are available as either a key press or shift and a key press.
03:52:01 <pikhq> ASCII is one hell of a US-centric standard. :)
03:52:03 <mad> yes because english doesn't have accents so they didn't need any hacks
03:52:21 <pikhq> English has (had) them but infrequently.
03:52:39 <mad> that's more like the situation of dutch
03:52:49 <pikhq> And they weren't in use in the US with typewriters, so it was deemed reasonable to omit them.
03:53:03 <pikhq> They've also become even more infrequent because of that.
03:53:34 <pikhq> Hence why it's typically spelled "resume", rather than "résumé" these days.
03:53:52 <mad> the root cause is that spelling evolved other ways of writing the extra vowels that you didn't have in latin
03:53:56 <pikhq> And "naive" instead of "naïve".
03:55:00 <pikhq> I'm pretty specifically referring to diactric usage that was normative in US Engliah circa 1900 but is not circa 2000. :)
03:55:31 <mad> kindof like how dutch doesn't need 'ü' and 'ö' because it moved 'u' to 'oe', and 'ö' to 'eu'
03:57:42 <mad> ï makes little sense in english because english has so few vowel-vowel sequences inside words
03:58:35 <pikhq> It was a systemic pattern then: every time you had a vowel-vowel sequence you put a diaresis.
03:58:53 <pikhq> "Zoë", "coöperate", "reëlect", etc.
03:59:14 <pikhq> I grant that it's not *that* significant, and so it makes sense it got dropped. :)
03:59:15 <mad> most of the time when you do have vowel-vowel, it's a french loan
03:59:41 <mad> so it's likely do have had a diaresis from the get go
03:59:45 <pikhq> Uh, I'm pretty sure that's "boolean" even with diaresis usage.
04:03:13 <mad> for most languages, diacritics are overdesign anyways
04:06:16 <mad> yeah, I'm sure the poor indian tribe appreciates having ẍ in their romanization
04:06:40 <mad> making it impossible to type so they don't write anything down
04:06:51 <mad> then their kids all learn english only
04:06:52 <oerjan> https://www.youtube.com/watch?v=l-GsfyVCBu0&src_vid=vFr3K2DORc8&feature=cards&annotation_id=68c52597-c5c6-4484-9007-74c98feeaa85
04:06:54 <mad> problem solved
04:07:03 <\oren\> Huh? I thought it was for the second derivative of x by t?
04:08:05 * oerjan thinks maybe there must be a better way to copy the link without all that junk
04:08:40 <\oren\> just scrool down to the share part
04:08:50 <\oren\> they have a copyable minilink
04:11:02 -!- bender| has joined.
04:11:03 <oerjan> prooftechnique: they did that yesterday too, after a while.
04:12:34 -!- bender| has quit (Read error: Connection reset by peer).
04:13:15 <\oren\> shit no matter if I run Df, ksp and a youtube video all at once my memry still isn't full
04:13:31 <\oren\> wtf do people even do with 16GB?
04:14:18 <prooftechnique> And also just keeeping more things in memory for better load times, I guess
04:16:11 <\oren\> I have 12GB and I can't fill it, I guess when I get 16 (another 8GB chip is in the mail) I'll run a VM or something
04:18:02 <prooftechnique> Yeah, I've usually got a couple of vagrant boxes, a browser, emacs, terminal, and other assorted nonsense. I usually hover around 50-60% utilization
04:18:11 <deltab> yeah, my office just received a server with 20GiB that used to run VMs
04:18:36 <prooftechnique> But memory's cheap and I'd rather have it and not need it than the other way :D
04:19:17 <deltab> linux is good at using memory for filesystem cache
04:21:22 <\oren\> windows is currently keeping a bunch of memory on "standby" not sure what that means
04:23:02 <\oren\> it seems to mean "not in use, but not empty either"
04:23:14 <\oren\> so I assume it's the file cache
04:23:42 <deltab> I expect sysinternals has an article explaining it
04:23:48 -!- ais523 has quit (Read error: Connection reset by peer).
04:24:47 -!- idris-bot has quit (Quit: Terminated).
04:25:07 -!- idris-bot has joined.
04:26:45 <\oren\> well, maybe it was overkill, but at least ksp runs super smooth now
04:28:58 -!- ais523 has joined.
04:33:08 <zzo38> The ! command in vim can be useful to do stuff that is not built-in to vim, such as !tac to reverse lines and so on
04:33:32 -!- llue has quit (Quit: That's what she said).
04:37:14 -!- treaki__ has joined.
04:37:58 -!- XorSwap has quit (Quit: Leaving).
04:40:15 <zzo38> Such a command is also moving it to the top of the file (if used with a selected area); now I can know that too, although I meant only part of the file, of course other commands such as cut, awk, sed, rev, xclip, grep, etc can also be used; many of those things can be done with built-in commands too somehow though
04:40:35 -!- treaki_ has quit (Ping timeout: 248 seconds).
04:40:49 <zzo38> But if you use ! command then you can use the same commands as the UNIX shell rather than always using different one
04:52:11 -!- prooftechnique has quit (Ping timeout: 250 seconds).
04:52:12 -!- kragniz has quit (Ping timeout: 246 seconds).
04:52:46 <oerjan> alphago made a surprising move, they say
04:54:20 -!- kragniz has joined.
04:55:54 -!- prooftechnique has joined.
04:59:12 <mad> what's the move?
05:00:01 -!- hppavilion[1] has quit (Read error: Connection reset by peer).
05:02:14 <ais523> mad: it replied to P9 with O10
05:02:52 <ais523> it's the most recent move, Lee Sedol has spent a lot of time (10 minutes?) thinking about it
05:05:11 -!- kragniz has quit (Ping timeout: 250 seconds).
05:05:44 -!- kragniz has joined.
05:19:37 <zzo38> If a implementation of Famizork mapper with CPLD were made, how many cells would it require?
05:32:20 <\oren\> https://youtu.be/kImJgUN5Y9s
05:32:57 -!- adu has quit (Quit: adu).
05:36:15 <\oren\> also, preview: http://www.orenwatson.be/script.png
05:37:15 <\oren\> based on cursive writing the way I learned it in primary school
05:38:25 <prooftechnique> I always forget how incomprehensible the D'Nealian G is
05:39:04 <\oren\> yah. In school I always just drew a random scribble for G
05:40:25 -!- puckipedia has quit (*.net *.split).
05:41:43 <prooftechnique> Russian cursive has more pathological cases, I think, though
05:42:11 -!- dingbat has quit (Quit: Connection closed for inactivity).
05:42:56 <prooftechnique> Which looks like this: https://i.imgur.com/4Ux8lgg.jpg
05:44:02 <\oren\> my dad has a friend named shchakhmatov, whose signature looks like "Max", so my dad always called him Max
05:46:11 -!- puckipedia has joined.
05:47:34 <\oren\> well that and shch is a ridiculous consonant cluster
05:51:50 <\oren\> yeah, well in russian they apparently say shch a lot so it's one letter
05:52:35 <prooftechnique> I think only in Ukrainian and such is it really like a shch sound. In Russian, anyway, it's just a... I dunno. Longer sh?
05:54:42 <\oren\> well he's been in japan for 30 years, so I think his name is probably now officialy シャッハマトブ
05:55:56 <\oren\> apparently a lot of russian mathematicians ended up at Osaka university somehow
05:59:11 <\oren\> Ok, I found a article that cites him as D. Shakhmatov
06:00:04 <\oren\> so maybe Щ is basically a sh
06:01:32 <\oren\> whatever, when I met him in Barcelona my dad just yelled "hey Max!"
06:01:57 -!- adu has joined.
06:02:57 -!- puckipedia has quit (*.net *.split).
06:03:37 <\oren\> thy does Unicode have separate codepoints for latin, greek and chinese cursive, but not cyrillic?
06:05:08 <\oren\> I think they are on the tertiary ideographic plane
06:05:47 <\oren\> oh, that's just what's planned
06:05:58 <\oren\> they havent assigned any codepoints yet
06:07:07 <\oren\> but they are planning on having seal script, bronze script, and oracle bone script
06:07:51 <lifthrasiir> \oren\: yeah, planned. also I think they are essentially different enough scripts to the modern han anyway
06:07:55 <\oren\> well I guess technically, modern chinese is a cursivized version of oracle bone characters
06:08:05 <lifthrasiir> no one would unify Linear B with modern Greek ;)
06:08:39 -!- puckipedia has joined.
06:08:54 <\oren\> well linear B is an alternate writing system entirely IIRC
06:10:23 <lifthrasiir> hmm, it is true. how about runes and modern Latin alphabets then?
06:10:39 <\oren\> yeah, runes hav their own
06:11:29 <\oren\> so they should have cyrillic cursive since it barely resembles printed cyrillic at all
06:14:08 -!- lambda-11235 has quit (Quit: Bye).
06:23:32 -!- hppavilion[1] has joined.
06:41:02 -!- hppavilion[1] has quit (Ping timeout: 244 seconds).
06:43:33 <ais523> prooftechnique: it's pretty interesting
06:43:51 <ais523> lee sedol has a lot of sure but small territories; alphago's territory is larger but shakier
06:44:15 <ais523> so it's basically up to how much lee sedol can reduce alphago's territory by attacking it
06:58:44 -!- MoALTz_ has changed nick to MoALTz.
07:06:28 <zzo38> I do have some idea about a new kind of inter-client protocol for X, other than _SEND_RESOURCE, can be MEDIA_PLAYER protocol. The user should be allowed to enable/disable use of this protocol (the default is up to the individual program). It is done by the playback window gaining ownership of the MEDIA_PLAYER selection; it can pause playback if it ever loses ownership.
07:07:26 <b_jonas> What's the result of the second game?
07:07:27 <zzo38> It can then be controlled by sending selection requests to it, and can be based on what the selection target is; some selection targets used with it might be: FILE_NAME, STRING, TARGETS, PLAY, PAUSE, STOP, RECORD, EJECT, NEXT_TRACK, PREV_TRACK, REWIND, FAST_FORWARD
07:07:50 <oerjan> b_jonas: it's not finished
07:08:22 <zzo38> Does these thing make sense to you?
07:08:22 <b_jonas> but it's eight in the morning
07:08:43 <zzo38> Don't a game of Go to take a long time though?
07:08:59 <oerjan> b_jonas: neither of them has used up their 2 hour base time
07:09:36 <oerjan> but then, they didn't yesterday either
07:10:16 <b_jonas> I'll check back later then
07:10:35 <ais523> it's an interesting game, I'm enjoying it
07:18:34 -!- AnotherTest has joined.
07:23:18 -!- AnotherTest has quit (Ping timeout: 276 seconds).
07:42:42 <oerjan> and lee is on extra time
07:49:21 <ais523> ooh, lee sedol spent one of his extra minutes
07:49:47 <b_jonas> are they in the endgame, just doing their captures and whatever?
07:50:34 <coppro> there's still a little open territory to fight
07:50:39 <coppro> but it's down to a few points here and there
07:52:49 <ais523> lee sedol's on one minute per move from now on
07:53:31 <b_jonas> well, he's a professional, I guess he knows what he's doing using his time
08:08:04 <ais523> so go programs have a tendency to play bad moves if they'll win anyway, as they're programmed to take a sure win if they see one
08:08:10 <ais523> the commentators think that that might be happening
08:09:45 <ais523> it still might not be endgame yet, though
08:09:50 <ais523> not all the territory is settled
08:11:07 <ais523> alphago's running low on time too
08:11:29 <ais523> it has 4 minutes of regular time + overtime (which is 1 minute per move, plus 2 minutes that can be added to specific individual moves)
08:12:22 -!- J_Arcane has joined.
08:14:27 <oerjan> michael redmond thinks alphago is winning again
08:16:55 -!- earendel has quit (Ping timeout: 268 seconds).
08:18:34 <ais523> alphago in overtime too now
08:30:52 -!- oerjan has quit (Quit: Nite).
08:33:22 <lifthrasiir> all the people says that Lee had no significant mistake and still lost
08:33:50 <coppro> someone ping me ith tanything interesting coming out of the press conference please
08:36:03 -!- incomprehensibly has quit (Ping timeout: 268 seconds).
08:36:39 -!- ocharles_ has quit (Ping timeout: 268 seconds).
08:40:04 -!- ocharles_ has joined.
08:42:25 -!- incomprehensibly has joined.
08:45:04 <ais523> lifthrasiir: apparently on one of the streams, one of the 9 dan commentators identified a move that he thinks lost Lee the game
08:45:26 <ais523> but unfortunately, my source is a livethread and there's not enough context to figure out which :-(
08:54:22 -!- heroux has quit (Ping timeout: 244 seconds).
08:56:57 -!- J_Arcane has quit (Ping timeout: 244 seconds).
09:00:05 -!- mad has quit (Quit: Pics or it didn't happen).
09:10:00 -!- earendel has joined.
09:16:27 -!- earendel has quit (Ping timeout: 246 seconds).
09:17:26 -!- ais523 has quit.
09:38:00 -!- heroux has joined.
09:50:19 -!- AnotherTest has joined.
09:52:09 -!- adu has quit (Quit: adu).
10:22:50 <fizzie> fungot: What do you think of your brethren beating us meatbags in Go? (The game, not the language.)
10:22:50 <fungot> fizzie: i just want a nice pastime, go to the grocery store door i thought " let-values"
10:30:54 -!- jaboja has joined.
10:45:44 <HackEgo> [wiki] [[List of ideas]] M https://esolangs.org/w/index.php?diff=46572&oldid=46358 * YoYoYonnY * (+132) /* Joke/Silly Ideas */
11:22:00 <myname> what was the language whose programs build a ring (i guess?) called?
11:26:43 -!- boily has joined.
11:51:27 <Riviera> myname: You don't mean a turning tarpit with ring, right?
12:03:07 -!- earendel has joined.
12:05:57 <boily> @tell Gregor your bot, it is tremendously slow hth
12:06:13 <HackEgo> [wiki] [[Stack]] M https://esolangs.org/w/index.php?diff=46573&oldid=45264 * VTCAKAVSMoACE * (+84) Added another example esolang to the list.
12:06:56 <HackEgo> Welcome to the international hub for esoteric programming language design and deployment! For more information, check out our wiki: <http://esolangs.org/>. (For the other kind of esoterica, try #esoteric on EFnet or DALnet.)
12:07:14 <boily> earendel: please refer to the rainbow welcome message up there ↑
12:12:05 -!- boily has quit (Quit: SLUG CHICKEN).
12:51:51 -!- bender| has joined.
13:13:22 -!- contrapumpkin has quit (Quit: Textual IRC Client: www.textualapp.com).
13:13:45 -!- copumpkin has joined.
13:24:42 -!- clog has quit (Ping timeout: 276 seconds).
13:24:49 -!- clog has joined.
13:35:28 <HackEgo> [wiki] [[Vitsy]] https://esolangs.org/w/index.php?diff=46574&oldid=46571 * VTCAKAVSMoACE * (+8908)
13:39:24 -!- bender| has quit (Remote host closed the connection).
13:46:03 -!- jaboja has quit (Ping timeout: 240 seconds).
13:53:48 -!- hkgit03 has joined.
14:10:03 -!- heroux has quit (Ping timeout: 244 seconds).
14:10:21 -!- heroux has joined.
14:14:53 -!- J_Arcane has joined.
14:20:29 -!- spiette has joined.
14:30:15 -!- heroux has quit (Ping timeout: 250 seconds).
14:31:51 -!- heroux has joined.
14:32:13 -!- hkgit03 has quit.
14:37:44 -!- prooftechnique has quit (Quit: ZNC http://znc.in).
14:38:33 -!- prooftechnique has joined.
14:44:57 -!- jaboja has joined.
14:45:55 <prooftechnique> Anybody want a keybase invite? I have 8 or 9, I think.
14:51:29 -!- AnotherTest has quit (Ping timeout: 250 seconds).
14:56:21 -!- lambda-11235 has joined.
15:05:03 -!- clog has quit (Ping timeout: 246 seconds).
15:13:56 <b_jonas> fungot, do you want a keybase invite from prooftechnique?
15:13:57 <fungot> b_jonas: i'm a cl programmer :) you fnord your code? it makes my eyes bleed
15:14:45 <fungot> prooftechnique: how do i use to make similar links to r5rs from scheme pastes that aren't of the form " fnord". pl/ 1, and -0 is a 0 0 5 0 0
15:16:06 <b_jonas> and he still doesn't understand how quotation marks work
15:17:55 <fungot> prooftechnique: now that i think about it? when it comes to punctuation, i end up liking it better. or worse yet horror of horrors there is a mandelbrot for some esolang?
15:40:17 <int-e> Yes. Impressive. I'm now worried that Lee may not win any game after all.
15:41:15 <prooftechnique> I only watched the first hour or two. What was the result?
15:41:41 <int-e> Lee resigned again... and he was never clearly ahead from what I've read.
15:42:07 <int-e> And in the end I think he was losing by more than 10 points, but that doesn't really mean much.
15:46:55 <int-e> Okay I can't count. It was more like 6 to 8 points.
15:51:22 <prooftechnique> On one hand, this is pretty groundbreaking stuff. On the other hand, I don't know how impressed I should be that a server farm beat a single brain at a computational task.
16:00:13 -!- lambda-11235 has quit (Quit: Bye).
16:00:48 -!- adu has joined.
16:19:44 -!- heroux has quit (Ping timeout: 244 seconds).
16:28:37 -!- heroux has joined.
16:56:59 -!- adu has quit (Quit: adu).
17:03:51 -!- clog has joined.
17:05:03 -!- jaboja has quit (Ping timeout: 276 seconds).
17:08:11 <izabera> i'm watching alphago's 2nd match
17:08:24 <izabera> the commenter said "oh this is a very weird move from alphago"
17:08:33 <izabera> and lee sedol left the room
17:08:52 <izabera> they didn't show him but i imagine him cursing and kicking his chair in rage
17:12:36 -!- AnotherTest has joined.
17:15:50 -!- jaboja has joined.
17:16:03 -!- J_Arcane has quit (Ping timeout: 244 seconds).
17:23:10 -!- heroux has quit (Ping timeout: 248 seconds).
17:29:29 -!- heroux has joined.
17:29:36 -!- zadock has joined.
17:32:39 <HackEgo> [wiki] [[Special:Log/newusers]] create * ETHproductions * New user account
17:33:49 <b_jonas> prooftechnique: the first time computers started winning against humans in chess, they required a whole server farm to win too. But then computers got faster.
17:35:02 <prooftechnique> I think my tongue was somewhat firmly in my cheek, but I take your point.
17:55:15 <HackEgo> [wiki] [[Japt]] N https://esolangs.org/w/index.php?oldid=46575 * ETHproductions * (+1175) Created page with "'''Japt''' is a language designed for ''code-golfing''. == Overview == Japt is a compiled language, or specifically, ''transpiled''. Many functions have been shortened to a..."
17:56:23 <HackEgo> [wiki] [[Language list]] https://esolangs.org/w/index.php?diff=46576&oldid=46547 * ETHproductions * (+11) /* J */ Added Japt to language list
17:59:17 <myname> just realized df will turn 10 this year
18:02:32 <b_jonas> myname: um, do you mean dwarven fortress?
18:06:56 -!- jaboja64 has joined.
18:10:46 -!- jaboja has quit (Ping timeout: 252 seconds).
18:21:24 <HackEgo> [wiki] [[Japt]] M https://esolangs.org/w/index.php?diff=46577&oldid=46575 * ETHproductions * (+72) Added categories
18:33:45 -!- mihow has joined.
18:39:03 -!- lleu has joined.
18:40:27 -!- mihow has quit (Quit: mihow).
18:52:15 -!- lynn_ has joined.
18:57:11 -!- jaboja64 has quit (Ping timeout: 250 seconds).
19:01:03 -!- lambda-11235 has joined.
19:02:15 -!- p34k has joined.
19:13:52 -!- Treio has joined.
19:17:17 -!- LexiciScriptor has joined.
19:30:05 -!- lynn_ has changed nick to lynn.
19:30:18 -!- hppavilion[1] has joined.
19:33:48 -!- LexiciScriptor has quit (Quit: LexiciScriptor).
19:36:48 -!- lambda-11235 has quit (Quit: Bye).
19:40:00 -!- jaboja64 has joined.
19:58:30 -!- LexiciScriptor has joined.
19:58:37 -!- LexiciScriptor has quit (Client Quit).
19:59:12 -!- lynn has quit (Read error: Connection reset by peer).
20:05:15 -!- Phantom_Hoover has joined.
20:07:51 -!- gde33 has left.
20:20:56 <fizzie> izabera: My wife was worried that he might get sufficiently frustrated to do something drastic.
20:21:18 <fizzie> I got the impression from some random earlier headlines that he was all "I'll win 5-0 or 4-1, no question about that" beforehand.
20:23:32 <fizzie> "For me, I'm not interested in playing against an easy opponent, especially for an easy-to-handle computer, --" says Ke Jie. Sounds suspiciously similar.
20:24:25 -!- lambda-11235 has joined.
20:24:58 <HackEgo> lambda-11235: Welcome to the international hub for esoteric programming language design and deployment! For more information, check out our wiki: <http://esolangs.org/>. (For the other kind of esoterica, try #esoteric on EFnet or DALnet.)
20:25:26 <lambda-11235> prooftechnique: Thanks. I've already been welcomed.
20:32:01 <b_jonas> ARGH! the algorithm I'm trying to implement can't work. it's broken as designed. damn.
20:32:10 <b_jonas> I'll have to rethink this whole thing.
20:34:54 -!- ais523 has joined.
20:34:59 <b_jonas> Now I understand why I didn't implement it like that before I started to rewrite it.
20:37:31 <b_jonas> I started to rewrite a code to use what I thought was a simpler algorithm. It turns out the simpler algorithm can't work. The complication was necessary.
20:54:00 -!- mihow has joined.
20:57:29 -!- zadock has quit (Quit: Leaving).
21:02:07 <zzo38> I think I read in newspaper the Go player expected to win against the computer the first time but that after the program is corrected then next time they probably cannot win
21:02:30 <izabera> fizzie: "my wife [...]" oh you're old
21:02:47 -!- lambda-11235 has quit (Quit: Bye).
21:11:01 <HackEgo> [wiki] [[Vitsy]] https://esolangs.org/w/index.php?diff=46578&oldid=46574 * VTCAKAVSMoACE * (+15038) Links fo' days, fixed some typos
21:17:07 <HackEgo> [wiki] [[Vitsy]] M https://esolangs.org/w/index.php?diff=46579&oldid=46578 * VTCAKAVSMoACE * (+91) Added some wikitrivia.
21:17:35 <HackEgo> [wiki] [[Vitsy]] M https://esolangs.org/w/index.php?diff=46581&oldid=46579 * VTCAKAVSMoACE * (+4) Fix italicization.
21:24:00 <HackEgo> [wiki] [[Language list]] M https://esolangs.org/w/index.php?diff=46582&oldid=46576 * VTCAKAVSMoACE * (+12) Added Vitsy to language list.
21:25:54 <zzo38> Do you know if there is any way with Xlib that you could figure out the clipping rectangle boundaries of a GC?
21:30:58 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
21:31:24 <b_jonas> ais523, zzo38: I have some vague thought about cpu architecture that probably doesn't even have a truth value, but I want to run it through you
21:32:23 <b_jonas> You know how much of the problem of the x86 arch is that it's really old, and at each step they updated it incrementally such that the same cpu can both run existing programs faster but also new programs can be written better but the chip implementing these two are practically the same.
21:32:55 <b_jonas> Although there were a few specific mistakes in these updates, most of the problem is just wanting to repeat such small incremental changes in the arch to make this possible, many times.
21:34:03 -!- hppavilion[1] has joined.
21:35:10 <b_jonas> And some of the particular decisions that made sense in the earlier stages of x86 but no longer are well known, like how the instruction encoding is really hard to decode quickly because it's very variable width, or that the 8 and 16 bit instructions have to update parts of registers,
21:36:17 <b_jonas> But there's a particular thing where I only now realized why the x86 design made sense originally but no longer.
21:36:56 <b_jonas> You know how despite the z80 being bigger, the 6502 instruction set turns out to be way more powerful than the z808 instruction set in practice?
21:37:27 <b_jonas> And I think one of you mentioned the reason for that, which was really not obvious to me since I didn't actually program those cpus:
21:38:38 <b_jonas> that the 6502 has only one accumulator, but it has instructions addressing the zero page with an 8 bit offset, and those instructions are really short, all of them being 2 bytes long, so you effectively use those 256 bytes as registers.
21:39:31 <b_jonas> Whereas on the z80, you have somewhat more general registers, but still not enough of them of course, and if you run out of them, you're really screwed, because the z80 doesn't have any sane way to address memory.
21:39:45 <b_jonas> It doesn't even have instructions to address memory with immediate offsets.
21:41:10 <b_jonas> Those RISC chips that have three-argument instructions on 15 to 32 independent registers 4 bytes long each and instructions with three register operands could get away with not having memory access instructions with encoded offsets,
21:41:25 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
21:42:11 <b_jonas> but the z80 is not like that, it has few registers, they're overlapping, and the arithmetic instructions only work on A or LH as one of the sources and destination, so you effectively have register-to-register moves and ONE-OPERAND instructions, not even two-operand ones..
21:42:52 <b_jonas> This might not be entirely accurate, since I'm a bit fuzzy about how the z80 works, and surely it has some tricks I'm missing, plus it has the IX and IY registers to help out.
21:43:43 <b_jonas> Anyway, you know how the x86 has more registers than the z80, and two-operand instructions to access that. But the trick is that not only it has those, but it also has what is effectively the equivalent of 6502's zero-page access:
21:44:30 <b_jonas> all the general instructions can use 8 byte offsets around four of the registers, so you can access 128 bytes above and 128 bytes below the stack base pointer, plus three other similar spaces based on the other registers.
21:45:16 <b_jonas> AND even though the general instructions are two-operand, they work both ways: the memory operand can be either source and destination, encoded by a bit in the opcode.
21:46:11 <b_jonas> So the x86 is a hybrid cpu that tried to have the advantages of both a larger instruction set risc cpu and the in-memory registers of 6502.
21:47:04 <b_jonas> And I think this was a good idea at the 8086, and probably still a mostly sane idea at the 386, but turned out to be a bad idea later.
21:47:41 <b_jonas> Mostly because you can't run instructions parallel and out of order and speculatively if they all reference main memory and any main memory write could affect any main memory read.
21:48:03 <b_jonas> The cpu can work out the dependencies between register accesses, but not between memory writes.
21:49:11 <ais523> hmm, the major "zero page"ish use I've seen on x86 is offsets from bp or sp
21:49:12 <b_jonas> So we ended up with a CPU that we are effectively trying to program like a RISC machine with everything stored in registers, sometimes even to the point that if you don't have enough general registers, it can be worth to save a value to an xmm register instead of the stack.
21:49:18 <ais523> it's like having a sliding window zero page
21:49:37 <ais523> perhaps we could have a stack that's /only/ addressible like that
21:49:43 <ais523> to solve the aliasing issues
21:49:48 <b_jonas> ais523: yes, and the MMIX has such a sliding window register stack.
21:50:15 <b_jonas> Now, there are some things in MMIX I don't like, but the sliding window local register stack is brilliant.
21:50:32 <b_jonas> (It also has global registers that don't move when the window slides.)
21:52:02 <b_jonas> They at least did away with the two-direction instructions with the vector instructions: those only read from memory, never write, but they were still all two-operand before AVX. And even AVX uses the mod-r/m encoding so there are two bits in every instruction just to say that it's accessing registers only, not memory.
21:52:41 <b_jonas> Mind you, at that point they have worse problems with the instruction encoding, but still.
21:53:20 <ais523> actually one thought I've had
21:53:23 <b_jonas> And even with all the great vector instructions, you can't do away with the general register instructions, you still have to use them for conditionals etc.
21:53:32 <ais523> is a CPU where the /only/ memory accesses are pushes and pops on a set of stacks
21:53:42 <ais523> each of which is independent
21:53:50 <ais523> this solves all cacheing and aliasing problems
21:54:11 <ais523> although, of course, it's not so great for programming as you don't have any random-access memory
21:54:35 <b_jonas> ais523: yes, that would be very inefficient for programming
21:54:52 <ais523> I wonder what sort of programs can be written efficiently on a multiple-stack machine
21:54:57 <ais523> could you write a compiler, for example?
21:55:27 <ais523> let's say, a compiler that runs in O(n log n) in the size of the original program, and does optimizations on an AST
21:56:32 <b_jonas> ais523: doesn't forth sort of work like this, where there's a data stack, a jump stack, and even a limited size for loop stack? It also has main memory, and one of the stacks are accessible through main memory, but I think a variant where none of those stacks are acessible in the main memory would look very similar to forth.
21:56:57 <b_jonas> ais523: so there would be no constant-time reversal of a stack?
21:57:25 <ais523> no, also the stacks are conceptually infinite anyway
21:58:06 <b_jonas> ais523: and do you allow enough stacks? eg., can a program use 256 stacks?
21:58:09 <ais523> hmm, let's add a slow random-access permanent storage, like a hard disk or SSD
21:58:14 <b_jonas> I know that doesn't matter for the asymptotics, but still
21:58:18 <ais523> b_jonas: there are a lot but a finite number
21:58:40 <b_jonas> 16 is too few unless you also have registers (or stacks limited to a small fixed depth)
21:58:52 <ais523> the main problem I have here is that this model can do program spawn-and-wait trivially
21:59:03 <ais523> but can't do multiprocessing at all well
21:59:27 <b_jonas> ais523: with the 16 stacks, would you allow directly reading and writing any of the top 16 elements? or just the top element?
22:00:12 <ais523> there's a reasonably large window at the top of the stack that's all readable and writable
22:00:24 <b_jonas> in that case 16 stacks are enough
22:00:35 <b_jonas> because you can put multiple registers on the same stack
22:00:40 <ais523> actually maybe we should have fewer stacks, it might be more efficient
22:02:00 <b_jonas> Totally different, but there's a crazy home-made eso-hardware that tries to masquarade as forth-based, but actually has a stack of, I kid you not, two elements max depth.
22:02:31 <b_jonas> The docs at first tells it has four elements, but later admits that the bottom two are reserved for interrupt handlers so you can only use the top two if you don't want to lose the value.
22:02:42 <b_jonas> At that point, I don't understand what makes this stack-based.
22:03:50 <b_jonas> It's really an accumulator-based machine with zero-operand instructions only, since you can keep one thing (the accumulator) on the stack for a few instructions, but have to load a "register" from memory as the second operand, then run one instruction that uses the two values together.
22:04:04 <b_jonas> (Arithmetic, indirect store, or conditional jump.)
22:04:32 <b_jonas> I can buy that it was easy to implement it in hardware this way (since it also has very few instructions), but I totally don't buy that it's stack-based or in any way related to forth.
22:05:16 <b_jonas> This one is the opposite: you say it's stack-based, but actually it has enough general purpose registers for a risc machine.
22:05:51 <ais523> it's stack-based in that it uses stacks for /storage/, not in that it uses them for calculation
22:06:16 <b_jonas> Anyway, it's an interesting question. It's not obvious what it could do.
22:06:46 <ais523> your "L1 cache" is the top of the stacks
22:06:47 <b_jonas> It would have to be very different from normal random-access programs.
22:07:12 <ais523> they get swapped out if you push too much onto them to fit into the cache, but the swapping is very predictable and can be done in the background; in particular you can always do a linear load
22:08:21 <b_jonas> By the way, the next M:tG set seems to be graveyard-themed like Odyssey: they bring back madness (which is why zzo38 was talking about madness, I didn't understand why), and a new threshold variant
22:10:41 <b_jonas> I'm surprised. It seems strange for them to bring back BOTH madness and threshold. Two such strange mechanics in one set? In Odyssey, madness was at least in a later set of the block.
22:10:46 <ais523> madness (if you discard this card, discard it into exile, then cast it for its madness cost or put it into your graveyard), delirium (if you have four or more card types in your graveyard, …), skulk (~ can't be blocked by creatures with higher power), investigate (place a Clue artifact token onto the battlefield with "{2}, sacrifice ~: draw a card")
22:11:04 <ais523> also apparently they tested threshold but didn't like the way it played
22:11:36 <b_jonas> Also, I think madness and threshold and other graveyard counting are mechanics I don't really enjoy playing with, because you have to build around them.
22:11:47 <b_jonas> Build the kind of decks I don't want to build around them.
22:11:54 <b_jonas> Decks that put cards to the graveyard.
22:12:10 <b_jonas> Can work if it's a whole set of course, but not too well in isolation.
22:14:23 <ais523> come to think of it, SoI only has four mechanics
22:14:24 <b_jonas> "let's say, a compiler that runs in O(n log n) in the size of the original program, and does optimizations on an AST" -- um, a compiler from what to what? I think you can do peephole matching on a tree in O(n log n) if you serialize the tree in a way that it's pre/postorder and the smaller weight child is closer to the parent than the larger weight child.
22:14:33 <ais523> Khans and Dragons each had six
22:15:01 <ais523> b_jonas: I'm thinking in general
22:15:22 <ais523> basically you want to be able to do a recursive tree pattern match
22:15:50 <ais523> hmm, for a specific example, let's say we want something that matches sexps against patterns, which allow * to mean "any sexp here"
22:15:54 <b_jonas> The problem is that a compiler also needs a symbol table, which is a dictionary, and I think that requires random access.
22:15:59 <ais523> and converts them to other patterns
22:16:31 <ais523> b_jonas: you could convert variables to lambdas and lambdas to stack references, Underload-style
22:16:34 <b_jonas> So the trees are no problem, but referring to O(n) symbols O(n) times in an arbitrary way is effectively random access.
22:17:02 <ais523> hmm, even more specific concrete example
22:17:13 <b_jonas> ais523: if the program is arbitrary, then it can refer to any symbol, even older ones
22:17:16 <ais523> s/(a $x $y)/(b $y $x)/g
22:17:28 <ais523> where sexps not starting with a and b are unchanged
22:17:36 <ais523> can a stack machine do this in O(n log n) time?
22:17:43 <ais523> clearly you can do it without the /g
22:18:52 <ais523> (also this regex syntax is highly dubious)
22:20:03 <b_jonas> ais523: I have the feeling that you can do that in O(n log n) time with a couple of stacks, let me try to figure out how exactly
22:20:14 <b_jonas> the number of stacks matter of course, because this is something you can't do with one tape
22:22:33 <b_jonas> ais523: first pass, annotate each node with the length (number of recursive nodes) of itself and children and grandchildren etc to a sufficient fixed depth.
22:22:40 -!- AnotherTest has quit (Ping timeout: 252 seconds).
22:23:02 <ais523> oh, you're assuming unbounded integers there
22:23:07 <ais523> admittedly this isn't a problem in practice
22:23:24 <b_jonas> ais523: no, only O(log n) size integers, which you can handle I think
22:23:34 <ais523> right, because of that
22:24:46 <b_jonas> In fact, let's say annotate each node with the length and head of itself and each close relative (including parent and nephew)
22:25:34 <b_jonas> Second pass: read the tree in sequence, and figure out where in the output sequence each symbol will move. For this, you keep track of the offset of how much each node moves, and when you encounter an "(a" or one of its relatives, you add or subtract the right weights.
22:25:47 <b_jonas> Third pass: sort everything, in O(log n) passes
22:25:54 <b_jonas> Fourth pass: fix it up to the output format
22:26:15 <b_jonas> I'm not sure this works, but I think it might.
22:26:43 <b_jonas> And there's probably a much easier way.
22:28:31 <ais523> hmm, actually that's a good point: can we sort a stack in O(n log n)?
22:28:42 <ais523> it's surely got to be possible but I'm not sure if any of the standard algorithms work
22:29:12 <b_jonas> ais523: yes, because you can sort in O(n log n) time with two or three _tapes_, that is, three to six stacks.
22:29:34 <ais523> hmm, which algo? quicksort seems promising
22:30:06 <b_jonas> ais523: much of Knuth vol 3 is about external sorting with tapes. In short, mergesort or its inverse.
22:30:32 <ais523> was wondering about mergesort
22:31:25 <b_jonas> ais523: however, radix sort also works here because you're keyed with numbers
22:32:16 <b_jonas> also takes O(log n) passes
22:32:20 <ais523> haha, that's really dirty :-)
22:32:28 -!- lynn has joined.
22:32:57 <ais523> I tend to mentally react to anything that assumes integers are finitely large
22:33:08 <ais523> integers don't work like that in theory! only in practice
22:33:47 -!- Treio has quit (Quit: Leaving).
22:33:56 <b_jonas> ais523: my favourite algorithmic problem can be solved in O(n) time despite that it requires a sorting step, because the sorting is on keys O(log n) large
22:34:28 <ais523> err, doesn't that give you a total O(n log log n)?
22:34:33 <ais523> radix sort is O(n log k)
22:37:17 -!- AnotherTest has joined.
22:38:11 -!- AnotherTest has quit (Client Quit).
22:38:11 <b_jonas> ais523: it's in the ram model with, integers large enough to address O(n), so you use bucket sort
22:38:41 <b_jonas> ais523: in the ram model, even radix sort is O(n log n) time
22:39:01 <ais523> as I said, radix sort is O(n log k), where k is the maximum value of the things you're sorting
22:39:08 <ais523> this doesn't require assuming that integers are bounded
22:39:23 <ais523> arguably in the ram model, all algorithms are O(1)
22:39:32 <ais523> because you can only store a finite amount of data
22:39:39 <ais523> sure there's a large constant factor, but you don't count that :-P
22:39:45 <b_jonas> ais523: bucket sort _doesn't_ work on arbitrarily large values
22:39:56 <b_jonas> it needs O(n+k) times if k is your largest possible integer
22:40:53 <b_jonas> Although I'm not sure if my algorithmic problem actually requires sorting, maybe that's just the lazy way to prove it works in O(n)
22:40:59 <b_jonas> I'll have to think about it
22:42:01 <quintopia> ais523: i'm not convinced. a simple counter could run forever as long as you keep adding memory, and it takes so long to step 64 gigs through all of its states that you have plenty of time to upgrade while it's running
22:42:29 <ais523> quintopia: well the RAM model b_jonas is talking about includes finitely large addresses
22:42:49 <ais523> although I guess you could pause the program, increase all the address sizes, then continue
22:43:46 <quintopia> ais523: then we'll have to count up in increments of *however many rams we've used up ach time we ran out of addresses*
22:44:15 -!- spiette has quit (Quit: :qa!).
22:44:27 <quintopia> and just remember what they had when we pulled them
22:45:16 <b_jonas> ais523: I haven't really thought about how fast it could be done on a real pointer machine (depends on the definition of the problem at that point). The interesting part of that problem to me is that it's possible but not trvial to do it in O(n**(1+epsilon)) time rather than O(n**(2+epsilon))
22:45:42 <b_jonas> And that I could use it as a great _educational_ example on what algorithm theory is about.
22:47:01 <b_jonas> Since the O(n**3) algorithm is obvious, and the O(n**2) algorithm is simple enough too, and I can show that the latter is faster, and can tell that there's an O(n**(1+epsilon)) time algorithm but that one is difficult.
22:48:16 <b_jonas> Except that I'm a bit confused.
22:48:36 <b_jonas> In that variant, the linear algorithm is actually O(n) and doesn't involve sorting.
22:49:19 <b_jonas> There's another variant where I'll have to think whether the quasi-linear algorithm can be done linearly, but in that case the exponents of the easy algorithms might be different, I'm not sure.
22:52:17 <b_jonas> Ok, let me try to hastily assemble an M:tG deck or two since I both have to work and then will play M:tG tomorrow.
23:16:06 -!- hppavilion[1] has joined.
23:24:07 -!- Sgeo_ has joined.
23:26:03 -!- Sgeo has quit (Ping timeout: 240 seconds).
23:30:44 -!- J_Arcane has joined.
23:34:43 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
23:35:59 -!- oerjan has joined.
23:47:37 -!- hppavilion[1] has joined.
23:47:42 -!- lynn has quit (Ping timeout: 260 seconds).
23:49:13 <myname> what is fnord? what is fnord? fnord is the space between your pixels at your monitor. fnord is the reason lisp has that many parantheses. fnord is the color only the blind can see. fnord is the sound of a single clapping hand.
23:49:16 <hppavilion[1]> λ-calculus is the basis of normal purely-functional programming
23:49:49 <hppavilion[1]> `le/rn inory/Inorically, inory is when you say something is irony that really isn't
23:50:42 <hppavilion[1]> SKI (or another CL) is the basis of normal epically-functional programming
23:51:10 <hppavilion[1]> And Turing machines^1 are^2 the basis^3 of imperative^4 programming^2
23:52:08 <hppavilion[1]> But really, there's no machine that corresponds to logic programming
23:52:17 <hppavilion[1]> I mean, there's actual logic, but that's pretty different
23:54:40 * oerjan swats hppavilion[1] for missing punctuation -----###
23:54:57 <hppavilion[1]> `le/rn inory/Inorically, inory is when you say something is irony that really isn't. Moron.
23:55:44 <oerjan> hppavilion[1]: it only applies to wisdoms hth
23:56:09 <oerjan> also, because `learn_append only works when things are punctuated.
23:56:48 <oerjan> `` sed -i 's/Moron./Someone who does this is an inorite./' wisdom/inory
23:57:31 <HackEgo> sep="/"; [[ "$0" == *//* ]] && sep="//"; [[ "$1" == ?*"$sep"* ]] || exit 1; key="$(echo "${1%%$sep*}" | lowercase)"; value="${1#*$sep}"; [ -e "wisdom/$key" ] && verb="Relearned" || verb="Learned"; echo "$value" > "$(echo-p "wisdom/$key")" && echo "$verb «$key»"
23:57:48 <oerjan> what's that supposed to mean?
23:58:34 <hppavilion[1]> oerjan: It'd be shorthand for `` sed -i 's/from/to/' wisdom/name
23:59:48 <oerjan> sounds a little limited.