00:05:51 <Phantom_Hoover> (i keep remembering stuff from x86 asm which would be insanely useful and then realising it could never work)
00:08:22 <int-e> there, 0x23 for the checkerboard
00:08:22 <shachaf> Oh, I looked it up and someone has a 4-instruction solution to Checkerboard.
00:08:31 <shachaf> That's much better than what I did.
00:08:37 <Phantom_Hoover> it is interesting how opcodes/addressing modes mean that even assembly is quite high-level
00:13:01 <int-e> I did some funny pipelining :)
00:13:19 <int-e> and I should be able to shave off one more cycle... let's see
00:15:16 -!- nitrix has changed nick to horse.
00:15:36 -!- horse has changed nick to nitrix.
00:18:23 -!- lambda-11235 has quit (Read error: Connection reset by peer).
00:28:33 -!- XorSwap has quit (Ping timeout: 240 seconds).
00:30:33 <boily> myname: mynamello. I think I prefer ahoily.
00:33:55 -!- oerjan has quit (Quit: Nite).
00:34:15 -!- XorSwap has joined.
00:41:57 <boily> GACK! just took a look at the outside world. weather was restored from months-old backups...
00:42:53 <fizzie> I think the weirdest part about that game was the term "memory depth" in the manual for the addressing modes.
00:46:53 <int-e> oh god, you can jump to odd addresses
00:47:14 <fizzie> I saw that in the manual as well.
00:47:29 <fizzie> But that's not odd from a real-hardware perspective.
00:48:56 <hppavilion[1]> Mapping data structures to control flow concepts, and vice versa
00:50:56 <hppavilion[1]> (Oh, and nesting data structures is equal to nesting control flow)
00:54:51 -!- lambda-11235 has joined.
00:55:43 <hppavilion[1]> lambda-11235: I'm figuring out a correspondence between data structures and control flow :)
00:57:51 <lambda-11235> hppavilion[1]: Hello. Have fun with that. I have to do English homework. :(
00:59:06 <int-e> seriously, 4 instructions... wow. I have 6, 5 if I allows the code to loop through nops instead of jumping back to the top...
01:07:11 -!- Guest772715 has joined.
01:07:54 -!- Guest772714 has quit (Read error: Connection reset by peer).
01:12:02 <impomatic> Oh, I missed all the BOX-256 chat :-(
01:12:46 <Phantom_Hoover> otherwise the guy might implement that shitty LODPC instruction
01:13:54 <impomatic> My smallest checkerboard is 6. I've been mostly optimizing for speed though.
01:15:16 <impomatic> I have the new Smiley Face record now :-)
01:17:17 <Phantom_Hoover> impomatic, https://www.reddit.com/r/box256/comments/4dfqlm/4_squares_4_threads_0x2e_cycles_using_a_beam/d1r7bi5?context=3
01:18:26 <int-e> is anybody collecting records without solutions?
01:18:28 <Phantom_Hoover> lets you get thread-specific behaviour at the cost of being an incredibly stupid way of doing so
01:20:47 <impomatic> int-e: I checked the reddit and hacker news threads for any records where the solution wasn't posted (for cycles only, no-one apart from me seemed interested in size)
01:22:51 <boily> int-e: int-ello. I can't stop binging QC.
01:23:16 <impomatic> Records are blue square 0x7, checkerboard 0x16, four squares 0x7, smiley face 0x2E
01:24:11 <impomatic> shachaf: I use the same nick everywhere :-)
01:24:13 <shachaf> Are those cycle records or line records?
01:24:14 <int-e> so my 0x22 checkerboard isn't too bad then
01:25:07 <int-e> is anybody trying for single-threaded cycles?
01:25:13 <shachaf> I was wondering about that.
01:25:32 <impomatic> int-e: that's pretty good. Only a couple of people have claimed less than 0x22.
01:26:11 <int-e> (singlethreaded I have 0x5E for the blue square, 0x1FC for checkerboard; multi-threaded it's 0x11 and 0x22 for now)
01:26:22 <impomatic> int-e: I wondered about collating single-thread cycle results, but everyone would just unroll the loops and fill all 63 available instructions.
01:27:15 <impomatic> Only 64 if the last instruction doesn't use the 4th byte.
01:27:20 <int-e> (if the last one is short)
01:27:53 <prooftechnique> "You must use a browser with WebGL. Consider downloading Firefox" > Running Firefox :|
01:28:58 <impomatic> There's a Windows version of BOX256
01:33:18 <Phantom_Hoover> if it's intelgrated that's probably why, especially if you're on linux
01:34:19 <HackEgo> /home/hackbot/hackbot.hg/multibot_cmds/lib/limits: line 5: exec: version: not found
01:34:31 <prooftechnique> Hanging Firefox, mainly, but maybe that'll figure itself out
01:34:37 <Phantom_Hoover> why do these shitty clients not have helpful /VERSIONs like xchat
01:36:03 <prooftechnique> Mine may be set weird because I'm connecting through ZNC
01:38:24 <Phantom_Hoover> -prooftechnique- VERSION WeeChat 1.4 (Mar 10 2016) via ZNC 1.6.2 - http://znc.in
01:38:41 -!- Phantom__Hoover has joined.
01:40:08 -!- Phantom__Hoover has quit (Client Quit).
01:42:04 -!- Guest772715 has quit (Ping timeout: 260 seconds).
01:42:59 <catern> Are there any esoteric languages that don't parse to a tree, but rather parse to a DAG or a cyclic graph?
01:43:27 <shachaf> You can think of any lambda calculus expression as parsing to a directed graph.
01:43:46 <shachaf> Do you mean that it's observable from within the language?
01:45:37 <catern> Oh, well, I'm just asking because during an intro talk on compilers (that I am currently sitting in), the presenter said "Languages usually parse to a tree"
01:46:06 <catern> I was wondering if there was any way one could justify making that "usually" qualification
01:46:54 <catern> And I thought #esoteric might be helpful
01:47:35 <prooftechnique> Maybe this? https://en.wikipedia.org/wiki/Abstract_semantic_graph
01:47:45 <Phantom_Hoover> shachaf, what, because you merge identical subexpressions?
01:47:50 <prooftechnique> Looks like a compiler could use it for CSE and whatnot
01:48:03 <pikhq> I don't think Befunge parses into a tree typically.
01:48:12 <shachaf> Phantom_Hoover: I mean that every use of a variable has an edge pointing to the lambda introducing it.
01:48:21 <pikhq> At least, I don't see any sane way of representing a Befunge program as a tree in any meaningful sense. :)
01:48:41 <Phantom_Hoover> shachaf, 'use' in that context sounds like it's still a directed tree
01:49:07 <Phantom_Hoover> there's not even such a thing as a syntactically illegal befunge program, is there?
01:49:29 <pikhq> Well, there is in Befunge 93 at least.
01:50:26 -!- Guest772715 has joined.
01:51:00 <prooftechnique> catern: Andres Löh actually has a paper on using ASGs for DSLs, which might be of interest
01:51:21 <Phantom_Hoover> as specified i think befunge programs are just a grid of 'commands'
01:51:22 <pikhq> Phantom_Hoover: Being larger than 80x25
01:59:12 -!- Phantom_Hoover has quit (Remote host closed the connection).
02:01:57 <fizzie> Arguably, also files containing form feeds in {Une,Be}funge-98, and files containing newlines in Unefunge-98. (Although strictly speaking the spec does explicitly allow multi-line Unefunge-98 programs, the lines are just concatenated. And it doesn't specify what form feed does outside of Trefunge-98, so you could say by omission it's just a regular character. But still.)
02:02:36 <fizzie> Oh, and the Funge-98 spec says that Befunge-93 only allows "printable ASCII characters and the end-of-line controls".
02:12:35 <hppavilion[1]> fizzie: A fungeoid... based on Numerical Computing
02:12:51 <hppavilion[1]> fizzie: So you can calculate the product of two programs
02:13:16 <hppavilion[1]> (For individual instructions, + is parallel composition, * is sequential composition)
02:13:55 -!- lleu has quit (Quit: That's what she said).
02:35:45 <hppavilion[1]> If you have a multigroup (S, [a], [b], [c]), then there are infinitely many possible equations.
02:36:52 <hppavilion[1]> However, there are only (#S)^(n+2) possible n-ary functions
02:37:34 <hppavilion[1]> I conclude that there is more than one way to write some equations
02:37:48 <hppavilion[1]> And I conjecture that there are infinitely many ways to write any given equation
02:38:29 -!- Guest772714 has joined.
02:38:38 -!- Guest772715 has quit (Read error: No route to host).
02:39:53 <boily> hppavilion[1]: principle, not principal :P
02:41:49 <prooftechnique> Not in a way that I could talk at length about them without some reference material to back me up
02:42:30 <hppavilion[1]> prooftechnique: Do you understand them enough to transfer some understanding to someone who doesn't
02:42:45 <prooftechnique> Depends what you need to understand, I suppose. I could try
02:44:10 <hppavilion[1]> prooftechnique: What I want to do is make a tensor programming language
02:44:23 <hppavilion[1]> Not a language which uses tensors like an array programming language uses arrays
02:44:50 <hppavilion[1]> prooftechnique: I want to make a language for which the /programs/ are... tensory?
02:46:33 <hppavilion[1]> prooftechnique: I don't necessarily want to do it right now
02:46:45 <prooftechnique> What property of tensors would you want to apply to a program?
02:47:46 <hppavilion[1]> prooftechnique: I want to be have a language for which the programs form a tensor (wiki uses that phrase, so I think it works)
02:48:09 -!- boily has quit (Quit: MOVIE CHICKEN).
02:48:29 <prooftechnique> Well, I'd start by figuring out what that means. Maybe figure out what the Cartesian product of programs in a set language is, or what the space of linear transformations from programs to programs looks like
02:49:06 <prooftechnique> Assuming that programs are themselves vector spaces, that is
02:56:24 -!- earendel has quit (Quit: earendel).
03:02:27 <int-e> Oh finally, I see, the two square problems are badly designed... there's enough space to just draw all pixels explicitly.
03:30:54 -!- Kaynato has joined.
03:51:14 <hppavilion[1]> prooftechnique: Well, the cartprod of two programs {a, b, c} and {1, 2, 3} is {(a, 1), (a, 2), (a, 3), (b, 1), (b, 2), (b, 3), (c, 1), (c, 2), (c, 3)}
03:51:32 <hppavilion[1]> The set is parallel-composed and the tuples are sequential-composesd
03:52:33 -!- earendel has joined.
03:53:23 <hppavilion[1]> The programs {a, b, c} and {1, 2, 3} represent set programs, in which instructions are non-blocking
03:57:34 <hppavilion[1]> prooftechnique: Programs-as-vector-spaces sounds interesting, but I was thinking programs-as-vectors. But programs-as-vector-spaces sounds more interesting
04:07:50 -!- iconmaster has quit (Ping timeout: 244 seconds).
04:10:13 -!- Guest772714 has quit (Quit: Nettalk6 - www.ntalk.de).
04:15:32 -!- Reece` has quit (Quit: Leaving).
04:19:06 -!- Reece` has joined.
04:37:57 -!- Reece` has quit (Quit: Leaving).
04:38:58 -!- Reece` has joined.
04:39:07 -!- Reece` has quit (Read error: Connection reset by peer).
04:40:07 -!- Reece` has joined.
04:49:43 -!- Reece` has quit (Quit: Leaving).
04:50:48 -!- Reece` has joined.
04:53:16 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
05:05:33 -!- XorSwap has quit (Quit: Leaving).
05:09:09 -!- lambda-11235 has quit (Read error: Connection reset by peer).
05:10:23 -!- lambda-11235 has joined.
05:13:53 -!- Reece` has quit (Quit: Alsithyafturttararfunar).
05:14:58 -!- Reece` has joined.
05:18:12 -!- Reece` has quit (Client Quit).
05:19:13 -!- Reece` has joined.
05:25:00 -!- Reece` has quit (Quit: Alsithyafturttararfunar).
05:26:49 -!- Reece` has joined.
05:27:24 -!- Reece` has quit (Client Quit).
05:32:20 -!- Reece` has joined.
05:40:06 -!- Reece` has quit (Quit: Alsithyafturttararfunar).
06:30:24 -!- Kaynato has quit (Ping timeout: 244 seconds).
06:39:58 -!- hppavilion[1] has joined.
06:42:06 -!- lambda-11235 has quit (Quit: Bye).
06:54:46 <newsham> why not just make ont hat sucks a little bit?
07:07:17 -!- jaboja has joined.
07:07:37 <coppro> what's wrong with the old one?
07:08:00 <hppavilion[1]> coppro: Well, it's not suitable for operator-heavy development
07:08:13 <hppavilion[1]> coppro: I would like to be able to define new & arbitrarily-named operators at will
07:08:36 <hppavilion[1]> coppro: So I could have different operators for element-wise multiplication and traditional matrix multiplication
07:08:46 <coppro> s/new C++ compiler/new C++ language/
07:09:31 <hppavilion[1]> It prints out endless garbage when I try to compile a simple file
07:10:57 <hppavilion[1]> coppro: I'm thinking of calling the new language C⁂
07:13:16 <lifthrasiir> hppavilion[1]: then adopt a character ⁂. http://unicode.org/consortium/adopt-a-character.html
07:18:30 -!- pelegreno has quit (Read error: Connection reset by peer).
07:20:39 <hppavilion[1]> I had accidentally targetted the compilation of main.cpp to one of its dependencies... xD
07:22:48 -!- pelegreno has joined.
07:25:37 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
07:41:50 <HackEgo> [wiki] [[Kantate]] N https://esolangs.org/w/index.php?oldid=46758 * Keymaker * (+5598) Added my new low-level language based on a simple addition mechanism.
07:50:48 -!- AnotherTest has joined.
07:52:02 <HackEgo> [wiki] [[User:Keymaker]] https://esolangs.org/w/index.php?diff=46759&oldid=40487 * Keymaker * (+14) Added Kantate, some changes.
07:58:01 -!- tromp_ has joined.
08:02:24 -!- tromp_ has quit (Ping timeout: 246 seconds).
08:17:52 -!- J_Arcane has quit (Ping timeout: 244 seconds).
08:25:14 -!- acertain has quit (Read error: Connection reset by peer).
08:25:46 -!- acertain has joined.
08:30:56 -!- Sprocklem has quit (Quit: [).
08:37:35 -!- ais523 has joined.
08:45:30 -!- hipsterslapfight has left.
08:46:39 -!- mroman has joined.
08:46:51 <mroman> I was always getting shit for my irrational fear of dogs.
08:47:03 <mroman> Now people are scared of people not wearing bikinis.
08:47:55 <mroman> I'm pretty certain that getting attacked by a dog severly enough to see a doctor is far more likely than whatever a woman not wearing a bikini is likely to do to me.
08:48:09 <lambdabot> boily asked 3d 8h 26m 55s ago: hello? hello? hello?
08:48:22 <mroman> @tell boily HELLO FROM THE OTHER SIDE....
08:48:27 <mroman> IT MUST HAVE BEEN A THOUSAND TIMES
08:48:50 <mroman> that last one might be off an inch.
08:50:14 <mroman> (currently switzerland's top topic is if it's ok for muslims to not shake hands)
08:51:06 <mroman> (I'm guessing secretly everybody is happy if they don't have to shake a muslims hand but get enraged when a muslim chooses to not shake hands..)
08:51:55 <mroman> is it "other side" or "outside"?
08:53:00 -!- jaboja has quit (Ping timeout: 268 seconds).
09:02:41 -!- Sprocklem has joined.
09:06:47 -!- Sprocklem has quit (Client Quit).
09:07:08 -!- Sprocklem has joined.
09:22:40 -!- jaboja has joined.
09:24:33 -!- Frooxius has quit (Ping timeout: 244 seconds).
09:52:27 -!- sebbu has quit (Ping timeout: 244 seconds).
09:58:39 -!- sebbu has joined.
10:05:52 -!- zadock has joined.
10:17:07 -!- zadock has quit (Quit: Leaving).
10:51:59 -!- earendel has quit (Quit: earendel).
11:08:31 -!- Frooxius has joined.
11:26:01 -!- lleu has joined.
11:26:01 -!- lleu has quit (Changing host).
11:26:01 -!- lleu has joined.
11:28:46 -!- boily has joined.
11:36:14 -!- Phantom_Hoover has joined.
11:49:54 -!- Phantom_Hoover has quit (Remote host closed the connection).
11:51:33 -!- Phantom_Hoover has joined.
12:01:19 -!- boily has quit (Quit: RESOLVED CHICKEN).
12:32:05 -!- J_Arcane has joined.
12:50:10 -!- tromp_ has joined.
13:00:55 -!- tromp_ has quit (Remote host closed the connection).
13:02:34 -!- jaboja has quit (Ping timeout: 244 seconds).
13:13:03 -!- Reece` has joined.
13:19:08 -!- jaboja has joined.
13:23:59 -!- oerjan has joined.
13:26:14 -!- jaboja has quit (Ping timeout: 250 seconds).
13:42:50 <J_Arcane> does anyone else know of any accidentally TC tabletop games besides MtG?
13:49:15 <oerjan> i know NP-complete and PSPACE-complete games, but TC requires a complexity beyond normal games, i think.
13:49:25 <oerjan> because unbounded state.
13:50:32 <oerjan> well, you need to generalize the size for the former, usually. so maybe they don't count in this sense either.
13:52:09 <ais523> you can do things like infinite generalizations of Sokoban and Minesweeper
13:52:34 <ais523> finding a game that naturally has infinite storage is hard, though
13:52:55 <ais523> and the ones that do, like Monopoly, don't have any obvious way to use it or do computations based on it
13:53:13 <ais523> hmm, what about Monopoly where the program is a repeating sequence of dice rolls?
13:58:43 <b_jonas> ais523: IMO what makes it hard to find TC-ness is that in the typical PSPACE-completeness proof of a game like sokoban or mario, you built the memory cells in the polynomial time preprocessing, and they're represented in-game as something like immutable in-game rooms. For TC you have to find a way to dig an unbounded sequence of rooms in-game, which is more difficult than just setting up a fixed number of rooms.
13:58:58 <b_jonas> \ you built the memory cells in the polynomial time preprocessing, and they're represented in-game as something like immutable in-game rooms. For TC you have to find a way to dig an unbounded sequence of rooms in-game, which is more difficult than just setting up a fixed number of rooms.
14:01:24 -!- tromp_ has joined.
14:05:49 -!- tromp_ has quit (Ping timeout: 260 seconds).
14:07:01 <oerjan> b_jonas: nothing was cut off hth
14:17:38 -!- jaboja has joined.
14:24:59 -!- spiette has joined.
14:32:13 <Taneb> Apparently in 2005 my uni's CS department held a Vim vs Emacs paintball tournemant
14:32:19 <Taneb> Paintball is probably not turing complete
14:33:44 <HackEgo> 1271) <ais523> editor flame wars are fun, I typically take the side of emacs and vim versus everything else <ais523> normally I can get most of the emacs /and/ vim users round to my side, thus catching out all the other-editor-users who thought they were safe
14:39:14 -!- Guest95899 has joined.
14:45:33 -!- Sgeo has quit (Ping timeout: 240 seconds).
15:06:59 -!- Kaynato has joined.
15:21:49 -!- lambda-11235 has joined.
15:32:35 -!- jaboja has quit (Ping timeout: 248 seconds).
15:39:27 <oerjan> @tell mroman <mroman> is it "other side" or "outside"? <-- that depends on whether or not you're a ghost hth
15:52:51 -!- Gwir has joined.
15:56:21 -!- lambda-11235 has quit (Quit: Bye).
16:01:21 -!- Gwir has left ("Leaving").
16:01:37 -!- tromp_ has joined.
16:05:58 -!- tromp_ has quit (Ping timeout: 244 seconds).
16:07:16 -!- mroman has quit (Quit: Lost terminal).
16:11:03 -!- J_Arcane has quit (Ping timeout: 240 seconds).
16:11:33 -!- Kaynato has quit (Ping timeout: 240 seconds).
16:14:03 -!- J_Arcane has joined.
16:19:15 -!- Kaynato has joined.
16:49:13 -!- XorSwap has joined.
16:50:15 -!- XorSwap has quit (Client Quit).
16:58:08 -!- ais523 has quit (Ping timeout: 250 seconds).
16:58:55 -!- jaboja has joined.
17:04:45 -!- jaboja has quit (Ping timeout: 248 seconds).
17:08:44 -!- oerjan has quit (Quit: Later).
17:16:30 -!- Guest772714 has joined.
17:22:55 -!- J_Arcane has quit (Ping timeout: 252 seconds).
17:27:19 -!- Guest95899 has quit (Ping timeout: 252 seconds).
17:27:45 -!- Guest95899 has joined.
17:30:49 -!- Guest772714 has quit (Quit: Nettalk6 - www.ntalk.de).
17:35:58 -!- Guest772714 has joined.
17:47:15 -!- Guest95899 has quit (Ping timeout: 264 seconds).
18:01:13 -!- Kaynato has quit (Ping timeout: 268 seconds).
18:02:11 -!- ais523 has joined.
18:03:15 -!- Kaynato has joined.
18:06:04 -!- Caesura has joined.
18:07:05 -!- Zoroaster has joined.
18:09:51 -!- Kaynato has quit (Ping timeout: 268 seconds).
18:11:05 -!- Caesura has quit (Ping timeout: 268 seconds).
18:26:15 -!- Frooxius has quit (Ping timeout: 264 seconds).
18:31:54 -!- jaboja has joined.
18:42:29 -!- iconmaster has joined.
18:42:33 -!- Frooxius has joined.
19:09:40 -!- Zoroaster has quit (Ping timeout: 268 seconds).
19:20:04 -!- Guest772714 has quit (Ping timeout: 252 seconds).
19:30:45 -!- hppavilion[1] has joined.
19:32:53 -!- Reece` has quit (Quit: Alsithyafturttararfunar).
19:32:56 -!- hppavilion[2] has joined.
19:33:16 -!- Reece` has joined.
19:34:11 -!- jaboja has quit (Ping timeout: 248 seconds).
19:35:28 -!- hppavilion[1] has quit (Ping timeout: 252 seconds).
19:36:38 -!- lambda-11235 has joined.
19:45:31 -!- Kaynato has joined.
19:54:11 -!- lambdabot has quit (Quit: brb).
19:55:50 <hppavilion[2]> The goal is to use it as an interactive Vi in-browser
19:58:58 -!- lambdabot has joined.
20:01:48 -!- Guest772714 has joined.
20:11:46 -!- hppavilion[2] has quit (Ping timeout: 252 seconds).
20:35:08 -!- p34k has joined.
20:36:02 -!- ais523 has quit (Read error: Connection reset by peer).
20:37:15 -!- ais523 has joined.
20:51:03 -!- hppavilion[2] has joined.
20:53:43 -!- rzfjwyxq has joined.
20:54:00 -!- rzfjwyxq has quit (Client Quit).
20:57:52 -!- lambda-11235 has quit (Quit: Bye).
21:04:54 <hppavilion[2]> I think I've discovered a correspondence between data structures and control flow
21:06:39 <hppavilion[2]> Of course, like the CHI, not all things on either side have a common partner
21:11:21 <shachaf> Maybe this channel would be more appropriate for my question.
21:11:50 <shachaf> If all function application is left-associated, what set of primitives would you need to be able to write any function?
21:12:09 <shachaf> Something like unlambda without `
21:12:21 <Taneb> So "f $ g $ h" = "(f $ g) $ h"?
21:13:27 <shachaf> I don't think that's sufficient.
21:18:20 <b_jonas> shachaf: stack operation primitives, I think
21:18:34 <b_jonas> shachaf: no wait, you said left-associated, not right-associated
21:19:05 <b_jonas> in that case you need to start form a primitive that's not well-typed in HM, right?
21:19:26 <b_jonas> wait, which way is left-associated and which way is right-associated?
21:19:44 <b_jonas> and from which side are you applying functions?
21:22:40 <hppavilion[2]> shachaf: Wait, does left associative imply that types can only by (((Q -> P) -> S) -> T) -> U and such?
21:22:52 <shachaf> OK, then show how to translate an arbitrary lambda expression.
21:22:53 <b_jonas> shachaf: in that case, you need a non-well-typed function
21:23:19 <b_jonas> shachaf: you can't apply arbitrarily many arguments to a well-typed function, right?
21:27:05 <int-e> fungot: can you factor eleven?
21:27:06 <fungot> int-e: commissioner, the council of ministers, several former prime ministers, this institution is, after all, this was also true of the police and judiciary, long regarded as the most comprehensive legal statement of children' s safety may be a case of double hulls. with regard to amendments nos 67 and 68 would complicate the system rather than trying to do this both in conjunction with countries that have applied to join the
21:27:56 <myname> how is stacl based not left assoziative?
21:27:58 -!- bb010g has joined.
21:29:02 <int-e> myname: in what context does that question make sense?
21:29:21 <fungot> Available: agora alice c64 ct darwin discworld enron europarl* ff7 fisher fungot homestuck ic irc iwcs jargon lovecraft nethack oots pa qwantz sms speeches ss wp youtube
21:29:47 <myname> i am refering zu b_jonas idea of stack operations
21:31:31 -!- hppavilion[2] has quit (Ping timeout: 252 seconds).
21:33:54 <fizzie> Double hulls, *that's* how we'll keep the children safe.
21:35:07 <fizzie> fungot: Do you have other parenting tips?
21:35:08 <fungot> fizzie: from a logical point of view on the subject of this own-initiative report commits various cardinal sins. the first deadline to be set at a higher level of investment and, thus, cannot accept amendments nos 30, 38 and amendments nos 5 and 6. as for the inspections, very much for your invitation, mrs kinnock, for pointing this out because i was looking to a new trend in the member states and applicant countries, whereas p
21:37:09 -!- J_Arcane has joined.
21:38:40 -!- Guest772714 has quit (Ping timeout: 252 seconds).
21:39:32 <int-e> oh you mean something like this
21:39:35 <int-e> > let push x a c = c (a:x); add (a:b:x) c = c (a+b:x); end x = x; start c = c [] in start push 1 push 2 add push 3 add end
21:41:11 <myname> end (x:_) = x would be nicer
21:42:11 <myname> if you only want to calculate scalars, that is
21:42:46 <int-e> oh, Oleg attributes this to Okasaki, interesting. http://okmij.org/ftp/Haskell/polyvariadic.html
21:43:10 <shachaf> myname: But you don't know at compile-time what will be in the stack.
21:43:18 <shachaf> Of course you could type-check it if you wanted to.
21:43:23 <shachaf> > let { push x a c = c (a,x); add (a,(b,x)) c = c (a+b,x); end (x,()) = x; start c = c () } in start push 1 push 2 add push 3 add end
21:43:33 <shachaf> > let { push x a c = c (a,x); add (a,(b,x)) c = c (a+b,x); end (x,()) = x; start c = c () } in start push 1 push 2 add push 3 add add end
21:43:34 <lambdabot> Couldn't match type ‘(t, ())’ with ‘()’
21:43:34 <lambdabot> Expected type: (t, ()) -> ((t, ()) -> t) -> t
21:43:35 <lambdabot> Actual type: (t, (t, ())) -> ((t, ()) -> t) -> t
21:45:06 <int-e> such a readable type error :)
21:45:23 <shachaf> @let data HList :: [*] -> * where { HNil :: HList '[]; (:!) :: x -> HList xs -> HList (x ': xs) }
21:45:43 <int-e> fungot: do we want to talk to quintopia?
21:45:44 <fungot> int-e: commissioner diamantopoulou, mr katiforis presents a report on one very important element of which is the year of marine safety, which will require the mobilisation of european engineers and support teams could, in fact, the dogma of the uncontrolled interpretative powers granted to the commission proposal for extending magp iv and it is there that we have balanced all the points we fought to improve.
21:45:56 <int-e> politicians just can't say no.
21:48:00 <myname> shachaf: why do i need to know?
21:48:57 <myname> just throw the error, it's fine
21:49:36 <shachaf> gotta type-check everything
21:49:47 <shachaf> Also I thought I'd use HList and maybe get a better error message.
21:50:08 <shachaf> But the joke's on me because it's not even working and I don't know why.
21:50:57 <int-e> does it work in ghci?
21:50:59 <shachaf> Just another day in HListland.
21:51:03 <shachaf> I don't know, I don't have ghci.
21:51:10 <shachaf> Well, the other day I wrote HListy code in C++. That was worse.
21:51:16 <int-e> how can you not have ghci!
21:51:31 <shachaf> > let { push x a c = c (a :! x); add (a :! (b :! x)) c = c ((a+b) :! x); end (x :! HNil) = x; start c = c HNil } in start push 1 push 2 add push 3 add end
21:51:33 <lambdabot> Couldn't match expected type ‘HList (x : xs1) -> t2’
21:51:33 <lambdabot> because type variable ‘xs1’ would escape its scope
21:53:14 -!- Kaynato has quit (Ping timeout: 250 seconds).
21:53:30 <int-e> :t let { push x a c = c (a :! x); add (a :! (b :! x)) c = c ((a+b) :! x); end (x :! HNil) = x; start c = c HNil } in start push
21:53:31 <lambdabot> Couldn't match expected type ‘HList (x1 : xs1) -> t2’
21:53:31 <lambdabot> because type variable ‘xs1’ would escape its scope
21:53:51 <int-e> I guess not being able to type this prefix could be a problem
21:54:28 <shachaf> > let { push x a c = c (a :! x); add (a :! (b :! x)) c = c ((a+b) :! x); end (x :! HNil) = x; start c = c HNil } in () -- The issue is in the definitions.
21:54:30 <lambdabot> Couldn't match expected type ‘HList (x : xs1) -> t1’
21:54:34 <lambdabot> because type variable ‘xs1’ would escape its scope
21:58:10 -!- Kaynato has joined.
21:58:10 <shachaf> > let { push x a c = c (a :! x); add :: HList (Int ': Int ': xs) -> (HList (Int ': xs) -> r) -> r; add (a :! (b :! x)) c = c ((a+b::Int) :! x); end :: HList '[x] -> x; end (x :! HNil) = x; start c = c HNil } in start push 1 push 2 add push 3 add end
21:58:18 <shachaf> It was just an inference issue?
21:58:52 <shachaf> > let { push x a c = c (a :! x); add :: Num x => HList (x ': x ': xs) -> (HList (x ': xs) -> r) -> r; add (a :! (b :! x)) c = c ((a+b) :! x); end :: HList '[x] -> x; end (x :! HNil) = x; start c = c HNil } in start push 1 push 2 add push 3 add end
21:59:08 <int-e> will it work in ghc 8?
21:59:11 <shachaf> Why does that have an inference issue?
22:01:11 <shachaf> :t let { end (x :! HNil) = x } in end
22:01:13 <lambdabot> Couldn't match expected type ‘t2’ with actual type ‘x’
22:05:31 <int-e> :t let end xs = case xs of { x :! HNil -> () } in 42
22:05:33 <lambdabot> Not in scope: data constructor ‘:!’
22:05:33 <lambdabot> ‘:+’ (imported from Data.Complex),
22:05:58 <int-e> :t let end xs = case xs of { x :! HNil -> () } in 42
22:05:59 <lambdabot> Couldn't match expected type ‘t’ with actual type ‘()’
22:06:53 <int-e> so it has learned nothing about the return type... and then there comes the GADT pattern match that freezes the type variables outside? dunnno. the ways of the type checker are mysterious and convoluted.
22:10:56 <ais523> :! doesn't look like an operator, ': might be one though
22:11:04 <int-e> The :! is part of the HList definition. ': is a type-level cons operator.
22:11:21 <int-e> well, type-literal level
22:11:33 -!- Guest772714 has joined.
22:12:35 <ais523> is ' reserved for type-level operators?
22:13:00 <int-e> (okay, that was stupid)
22:13:11 <ais523> it got all the way up to the ] before getting confused
22:13:24 <ais523> I take it '[' was parsed as a character
22:13:26 <int-e> > '['() -- is just a type error
22:13:27 <lambdabot> Couldn't match expected type ‘() -> t’ with actual type ‘Char’
22:13:27 <lambdabot> The function ‘'['’ is applied to one argument,
22:13:37 <b_jonas> ais523: I think you need to write it as '[]()
22:13:39 <ais523> and only failed to parse after that
22:13:51 <int-e> b_jonas: I think ais523 didn't make the mistake
22:13:59 <int-e> b_jonas: but anyway :t parses expressions
22:14:39 -!- spiette has quit (Ping timeout: 260 seconds).
22:14:55 <ais523> what does k mean in a kind?
22:15:08 <int-e> it's a polymorphic kind
22:15:12 <int-e> so a kind variable
22:15:31 <ais523> I didn't realise we had kind-level polymorphism
22:15:50 <int-e> ghc has grown some fancy type system extensions in the last three (I think) years
22:16:00 * ais523 wonders if it's possible to identify types with kinds without becoming dependently typed
22:20:59 <shachaf> ais523: That's sort of what GHC does.
22:21:08 <shachaf> Depending on what you mean by "identify".
22:21:14 <ais523> I'm not sure what I mean
22:22:15 <shachaf> When you define a type Maybe a with values Nothing and Just a, GHC also defines a kind Maybe a with types Nothing and Just a
22:22:44 <ais523> is it possible to construct a value of type Nothing?
22:22:59 <shachaf> No. By "type" I just mean that it exists on the type level.
22:23:16 <coppro> ais523: well, even in a dependently-typed language you can't create a object of type Nothing
22:23:24 <coppro> at least, not any one I've seen
22:23:30 <ais523> hmm, it doesn't seem that the types have much in common with the kinds apart from having the same name
22:23:58 <int-e> but anyway the type-level 'Nothing and the data-level Nothing are not the same.
22:24:25 <shachaf> What if value-level functions were also automatically lifted to type-level functions?
22:25:22 <int-e> . o O ( my_head _ = unsafePerformIO ... )
22:25:41 <shachaf> ski: see also the conversation in this channel a few screens ago
22:25:50 <ais523> shachaf: is that even possible in general?
22:25:56 <int-e> but I suspect that then you get all the bad aspects of dependent type (undecidable type checking) without any benefits.
22:26:10 <ais523> is there a type-level fix? that feels like the largest problem
22:26:47 <shachaf> ais523: Well, you have isorecursive types. Not equirecursive types like I think you can get in ocaml.
22:27:04 <int-e> but you can already do funny things (like constructing a type satisfying x ~ [x]) with type families.
22:27:53 <b_jonas> shachaf: I think haskell has those too, but only if you enable an extension flag, and you normally don't want that, because it makes typechecker errors REALLY confusing because type errors that lead to recursive types get caught too late
22:27:54 <int-e> sorry... didn't read what shachaf wrote before pressing return
22:28:16 <shachaf> b_jonas: I've never seen Haskell have those.
22:28:29 <shachaf> It would be interesting to try.
22:32:45 <ski> b_jonas : it's the OCaml implementation which has them (`-rectypes'). they already need them for (cyclic) object types. the option just removes the restriction that the cycle has to go through an object type
22:33:28 <shachaf> I remember https://mail.haskell.org/pipermail/haskell-cafe/2006-December/020072.html
22:36:59 -!- XorSwap has joined.
22:38:43 <izabera> https://github.com/Rochester-NRT/AlphaG
22:38:48 <izabera> https://github.com/Rochester-NRT/AlphaGo
22:42:38 <ski> (and you need cyclic object types for binary methods and "clone" methods. see <http://caml.inria.fr/pub/docs/manual-ocaml/objectexamples.html#ss%3Acloning-objects>,<http://caml.inria.fr/pub/docs/manual-ocaml/objectexamples.html#ss%3Abinary-methods>. also <http://caml.inria.fr/pub/docs/manual-ocaml/objectexamples.html#ss%3Afunctional-objects>)
23:16:39 -!- gremlins has joined.
23:18:27 -!- Reece` has quit (Ping timeout: 244 seconds).
23:19:29 -!- oerjan has joined.
23:28:36 -!- Guest772714 has quit (Quit: Nettalk6 - www.ntalk.de).
23:35:37 -!- boily has joined.
23:38:32 <oerjan> <shachaf> If all function application is left-associated, what set of primitives would you need to be able to write any function? <-- hm sounds familiar.
23:39:01 <shachaf> oerjan: I think I've asked this before.
23:39:09 <shachaf> And someone even sent me a paper or something about it.
23:39:19 <shachaf> It was called something like "bracketing combinators" but I can't find it now.
23:39:20 <ais523> oerjan: shachaf: isn't that just Underload?
23:39:28 <oerjan> i know i showed once that you can refactor everything as a big blob of id and (.) followed by your original terms, but i'm not sure if the big blob itself can then be split up.
23:39:36 <ais523> or, well, concatenative languages generally
23:39:56 <oerjan> ais523: that would be right-associated.
23:40:14 <oerjan> without a lot of tweaking, anyway.
23:40:16 <ais523> all functions are associative in Underload
23:40:23 <ais523> so the association doesn't really matter
23:40:34 <shachaf> oerjan: You mean the big blob is a tree?
23:40:36 <oerjan> ais523: um we're assuming the application is the usual lambda one.
23:41:02 <shachaf> I'm talking about "unlambda without `"
23:41:03 <ais523> I guess it is right-associative then?
23:41:22 <ais523> shachaf: where you basically have enough `s at the start of the program to balance it?
23:41:25 <shachaf> In Haskell "f x y" is "(f x) y"
23:41:34 <oerjan> shachaf: a (b (c d) e) = (.) a (b (c d)) e = (.) ((.) a) b (c d) e etc.
23:42:03 -!- AnotherTest has quit (Quit: ZNC - http://znc.in).
23:42:37 <oerjan> = (.) ((.) ((.) a) b)) c d e
23:43:01 <shachaf> OK, but your blob isn't standalone there.
23:43:16 <oerjan> you eventually get all the rest out
23:43:42 <shachaf> Do you? When I've done it in the past I ran into cases where it didn't happen with the natural (.)ification.
23:43:52 <oerjan> * = (.) ((.) ((.) a) b) c d e
23:44:04 <shachaf> You run into (.)^6 = (.)^10 and the expression keeps growing.
23:44:13 <oerjan> = (.) (.) ((.) ((.) a) b c d e
23:45:10 <oerjan> * = (.) (.) ((.) ((.) a)) b c d e
23:45:24 <shachaf> it's easier if you give (.) a non-infix name hth
23:45:41 <oerjan> = (.) ((.) (.)) (.) ((.) a) b c d e
23:45:47 <shachaf> copumpkin: Maybe you were the one who sent me that paper?
23:46:12 <oerjan> = fmap (fmap fmap) fmap (fmap a) b c d e
23:46:13 <shachaf> oerjan: "bracketing combinators" or something
23:46:32 <oerjan> = fmap (fmap (fmap fmap) fmap) fmap a b c d e
23:46:37 <oerjan> there, one initial blob
23:46:50 <shachaf> OK, I believe that it works in this case.
23:47:02 <oerjan> i'm pretty sure it works in general.
23:47:02 <shachaf> I don't remember whether the issue I ran into was trying to deblob the fmaps themselves.
23:47:15 <oerjan> right, i don't know if that works.
23:47:44 <lambdabot> mroman said 14h 59m 22s ago: HELLO FROM THE OTHER SIDE....
23:48:06 <boily> @tell mroman *gasp*! you're not dead! well... if you are, you communicate!
23:48:23 <oerjan> = fmap fmap (fmap (fmap fmap)) fmap fmap a b c d e
23:48:44 <oerjan> perhaps it starts getting harder now
23:48:45 -!- Sgeo has joined.
23:51:32 <oerjan> <shachaf> You run into (.)^6 = (.)^10 and the expression keeps growing. <-- in that case surely you can just shortcut back >:)
23:51:33 <shachaf> myname: just search the #haskell logs for "fmap fmap fmap fmap". i've done plenty of this.
23:52:07 <shachaf> oerjan: Anyway, you're obviously not going to be able to encode an arbitrary binary tree structure in unary.
23:52:26 <oerjan> shachaf: well ok right, you need more than just fmap.
23:52:45 <oerjan> so the question is, how little do you need to be able to decode a fmap blob
23:53:44 <oerjan> boily: from other things mroman said, i think he may be confused about the connotation of "THE OTHER SIDE"
23:54:01 <shachaf> oerjan did say "fmap and id"
23:54:02 <oerjan> (dammit, forgot to portmantell again)
23:54:46 <oerjan> <myname> you are crazy <-- wait, anyone in particular, or all of us twh
23:55:24 <oerjan> shachaf: i think you only need id for the trivial case.
23:55:43 <boily> hellørjan. there are multiple other sides? as far as I know, there's the Other Side, Our Side, and that's about it.
23:55:45 <oerjan> consider it the empty blob, or something.
23:58:19 <lambdabot> Functor f => (a -> b) -> f a -> f b
23:58:24 <lambdabot> (Functor f, Functor f1) => f (a -> b) -> f (f1 a -> f1 b)
23:58:31 <lambdabot> Functor f => (a -> b) -> f a -> f b
23:58:38 <ais523> :t fmap fmap fmap fmap fmap fmap
23:58:39 <lambdabot> (Functor f, Functor f1) => (a -> a1 -> b) -> f a -> f (f1 a1 -> f1 b)
23:59:00 <shachaf> :t dot dot dot dot dot dot
23:59:01 <lambdabot> (b -> b1 -> c) -> (a -> b) -> a -> (a1 -> b1) -> a1 -> c
23:59:23 <ais523> I actually think the fmap version might be clearer, it's surprising that it only uses two functors
23:59:45 <oerjan> :t fmap fmap fmap fmap fmap fmap fmap fmap fmap fmap
23:59:45 <lambdabot> (Functor f, Functor f1) => (a1 -> a -> b) -> f a1 -> f (f1 a -> f1 b)