00:00:02 i was first gonna written elliot, but then realized you're definitely a double geminate guy 00:00:38 heh 00:00:38 today, i heard of a person with my name 00:00:43 a guy said their teacher had said to the class she'd called for some tech assistance, and i had answered. 00:00:50 (happy tomorrow GMT) 00:01:04 (said "an old student" and my name) 00:01:18 (not sure it's me, but quite close, for i'm an old student of hers) 00:01:25 i didn't get a call though. 00:01:40 oklopol: they were going to ask for your help but then saw oklotalk 00:01:42 :) 00:04:22 -!- CakeProphet has quit (Connection timed out). 00:04:27 -!- CakeProphet__ has changed nick to CakeProphet. 00:04:35 -!- Jontte has quit ("Konversation terminated!"). 00:05:10 :D 00:05:14 QUIZ 00:05:20 find a language where this does what you'd expect 00:05:21 3=4 00:05:36 that one esolang, you mean? 00:05:45 i'd expect it's an always-failing proposition 00:05:58 i don't remember the name now, but the language is fun 00:07:51 no 00:07:56 it should actually redefine 3 as 4 00:08:04 x = 3 00:08:05 3 = 4 00:08:09 x // => 4 00:08:11 3 // => 4 00:08:17 3 + 1 // => 5 00:08:29 4 - 1 // => who knows? 00:08:47 4 - 1 // => 4, probably, since 3 is now 4 00:09:57 3 is now 4, but 4 is still 4 00:10:01 at least in the esolang 00:10:11 yes 00:10:15 3 // => 4 00:10:18 4 // => 4 00:10:25 yeah, so 4-1=3 00:10:26 4 - 1 // => this would be 3, but 3 is now 4, so: 4 00:10:27 oh. 00:10:29 4 of course :) 00:10:44 :D 00:13:45 esolangs aren't innovative anymore, just reinventions :( 00:16:56 -!- CakeProphet_ has quit (Connection timed out). 00:17:59 nah 00:18:30 i think i've invented quite a lot of new stuff. 00:18:43 for example... 00:18:57 have i mentioned GRAPHICA, HAVE I EVER MENTIONED GRAPHICA? 00:19:17 no. never. 00:21:26 :----) 00:24:06 fibonacci in graphica= 00:24:25 it's not tc yet. 00:24:32 or then it is, not sure 00:24:45 fibonacci is a simple rewrite rule 00:24:46 no TC needed 00:25:26 i guess you could just Fib m n :: m n <-> Fib m+n m 00:25:43 fibonacci's rewrite rules are very simple: 00:25:46 oh, wait 00:25:57 i guess you could just Fib m n :: m <-> Fib m+n m 00:25:57 (A → B), (B → AB) 00:26:00 start with A 00:26:04 then it's in unary. 00:26:37 or (B → BA). 00:26:52 mm 00:27:41 -!- CakeProphet__ has joined. 00:29:14 "A"->"B";"B"->"AB";"A" 00:29:19 doesn't work, though, in CRTL 00:29:29 because the rules get removed, and the A rule effects the "B"->"AB" 00:31:59 well, you can use my program though 00:32:47 also, you can attach a type tag there, to make sure rules apply in the right order and to the right data 00:39:17 A->B , B->AB | B->AB , AB->BAB ? 00:39:28 or A->B, B->BB? 00:41:26 A-B, B->AB 00:41:34 oh 00:41:35 you mean in my prog 00:41:41 "A"->"B";"B"->"AB";"A" 00:41:47 "B"->"BB";"B" 00:41:49 "BB" 00:41:56 so, "A"->"B";"B"->"AB";"A" outputs "BB" and exits 00:42:05 ehird`: what was wrong with my fibonacci? 00:42:22 oklopol: ? 00:42:30 the one i gave hours ago 00:42:35 show 00:43:37 (("fib"->a)->b)->(("fib"->b)->a~b),a;("fib"->"1")->"1" 00:43:50 let me step through that. 00:43:52 uh 00:43:53 you have , 00:43:57 , is not CRTL. 00:43:58 QED 00:44:00 well yeah, that was for output 00:44:03 you can forget about that one 00:44:11 show me a new one then 00:44:16 (("fib"->a)->b)->(("fib"->b)->a~b);("fib"->"1")->"1" 00:44:23 (("fib"->a)->b)->(("fib"->b)->a~b);("fib"->"1")->"1" 00:44:57 ("fib"->"1")->"11" 00:45:06 [end] 00:45:10 hmm 00:45:21 why doesn't rule 1 apply a second time? :\ 00:45:25 rules are one use? 00:45:30 because it's removed from play 00:45:33 didn't know that 00:45:42 then the language is pretty useless, huh? 00:45:57 i mean, you can do a fixed number of rewrites 00:46:37 nope, because you can do things like this: 00:46:49 -!- CakeProphet has quit (Connection timed out). 00:47:05 hmm 00:47:08 ok then 00:47:11 let's apply it multiple times. 00:47:15 (("fib"->a)->b)->(("fib"->b)->a~b);("fib"->"1")->"1" 00:47:21 (("fib"->a)->b)->(("fib"->b)->a~b);("fib"->"1")->"11" 00:47:31 (("fib"->a)->b)->(("fib"->b)->a~b);("fib"->"11")->"111" 00:47:41 (("fib"->a)->b)->(("fib"->b)->a~b);("fib"->"111")->"11111" 00:47:42 yeah, that works 00:47:47 ok, now make it do output :P 00:48:00 well, only possible with "," 00:48:07 or something equivalent 00:48:17 nope, i'm pretty sure you can do it 00:48:22 let's try... 00:48:22 no way to make a new tuple -> no way to do infinite outputs 00:48:33 what about if we leave outputted strings in place. 00:49:02 well... then it's possible again, but perhaps a bit ugly, i'll try 00:49:55 actually... hmm 00:51:16 oklopol: ("fibo"~x)->... may help 00:53:22 (("fib"->a)->b)->(("printed"->b)->a~b); 00:53:22 (("printed"->a)->b)->a~"|"~b 00:53:22 a~"|"~b->(("fib"->a)->b) 00:53:22 ("fib"->"1")->"1"; 00:53:31 best i could think of 00:53:41 example output? 00:53:48 that'll print 1|1, 1|11, 11|111, 111|11111 00:53:50 i think. 00:53:56 so it generates statements like: 00:53:58 "1|11"; 00:54:04 you see i need to parse the values back after printing them 00:54:07 with no other expression in that statement? 00:54:30 generates what i said, but s/","/newline, i guess 00:54:58 i mean in the code. 00:55:07 hmm? 00:55:36 oh, printed is redundant... 00:56:04 (("fib"->a)->b)->b~"|"~a~b; 00:56:04 a~"|"~b->(("fib"->a)->b) 00:56:04 ("fib"->"1")->"1"; 00:56:42 now put that on the CRTL wiki and i'll update the spec so that it's true :P 00:56:46 :D 00:56:49 by the way, does that rely on any kind of evaluation order? 00:56:56 no, i don't think so 00:57:21 yay 00:57:45 but, the problem is, after you do output, you cannot (using my way of coding) output any less than *everything* you're gonna need in further computation 00:58:10 because printing is done by encoding the program state into a string, and then parsing the string into a tuple again. 00:58:11 oh 00:58:33 it actually relies on the fact a string is only output once, then tried the substitutions again 00:58:49 but this is a given, for otherwise *any* output you do could happen *any* number of times 00:58:57 and what's with the *emphasis* 00:59:34 yeah, that's right 00:59:39 http://esoteric.voxelperfect.net/wiki/CRTL#Evaluation updated 00:59:56 I'll add your fibonacci. 01:01:03 ok, added 01:01:06 I think it's turing complete. 01:01:26 ah... 01:01:38 ehird`: what's your lang called? 01:01:44 GreaseMonkey: CRTL... 01:01:57 Constantly Rewriting Term Language 01:02:18 oklopol: "," is not needed now yeah? 01:02:38 btw: a good excersize would be a dupdog interpreter. 01:03:54 well, "," would just be nice, since you would have tc output :-) 01:03:59 if you know what i mean 01:04:11 no, i don't 01:04:32 let's assume that CRTL is turing complete, are you saying it can't output every string as a result of a calculation? 01:04:43 yes. 01:04:49 ok. 01:04:54 in case there's no better way than what i'm using. 01:05:27 just like bf couldn't output every string if . printed "printing: " 01:05:32 it'd still be tc. 01:06:52 i guess a better example is, utm cannot do output, but i'm fairly sure it's tc! 01:09:05 yeah i get it 01:09:15 oklopol: got any more elegant solutions than ,? 01:10:18 hmm, i'm wishing a way to do without it would pop in my head... 01:11:15 hmmm... "," is non elegant why? 01:11:27 dunno 01:11:45 i mean, you can't pattern match on rules with free variables anyway, so there's a definite distinction between data and rewriting rules... 01:11:57 well 01:12:06 i mean, in some cases. 01:12:21 hmm 01:12:23 i can add ' 01:12:24 quote 01:12:31 to pattern match on free vars. 01:12:41 yeah, metalevels ftw! 01:13:28 but, i still think "," would be nice, otherwise you'll just have a fixed set of rules... and that's just too static! 01:13:42 ok. 01:45:27 -!- Corun has quit ("This computer has gone to sleep"). 02:05:19 back 02:05:42 oklopol: i will add quoted = "'" name | "'" quoted 02:05:48 oklopol: but this makes programs = data 02:05:54 so propose a , that doesn't differenciate 02:06:01 -!- CakeProphet__ has quit (Read error: 110 (Connection timed out)). 02:08:35 oklopol: ping 02:09:17 o 02:09:44 so yeah, do that :P 02:09:48 :) 02:10:11 well, there's one that i can think of at least, but it's a bit complex. 02:10:16 tell me. 02:11:14 a rule A -> B, C is considered two separate rules A -> B; A -> C when it's pattern matched on, the "," just means the two rules are always applied both 02:11:41 then it could be pattern matched on 02:12:28 A -> B; A -> C is pretty nonsensical 02:12:43 if A->B is run first the program becomes A -> B; B -> C 02:12:50 the other way around, C -> B; A -> C 02:13:27 so, suggest an elegant solution that makes sense :D 02:13:48 that was not a way to convert a rule with "," to one without it 02:13:57 that's impossible 02:14:08 okay 02:14:10 then explain 02:14:12 it'd be A->B; A->C; when pattern matched on. 02:14:34 when it's used as the pattern, it'd be a special pattern thatmakes two tuples from one input tuple. 02:14:37 *that makes 02:14:55 so pattern matching doesn't reflect what you see in the source... 02:14:59 no: i don't like that. 02:15:43 then you can just add the new operator "," 02:16:09 the problem is, that makes it useless to have data and program flow integrated when the feature is not used 02:16:29 sorry i think you'll have to come up with something a lot nicer :) 02:16:30 actually.... 02:16:31 i suggest: 02:16:35 a,b 02:16:37 err 02:16:38 ok 02:16:46 foo->(bar,baz) 02:17:04 foo->bar, but appends "baz" to the end of the program and maybe-outputs it 02:17:19 actually i don't think "," as an operator would be that bad 02:17:43 so 02:18:22 "a","b"->"c" would consume an "a" and a "b" and create a "c" 02:18:34 then it'd be a bit more generic 02:18:54 ok, write a fibonacci-outputs-in-unary-without-weird-extra-state 02:18:56 using that 02:18:57 :-) 02:19:06 umm... 02:19:15 it'd work with what i originally had 02:19:23 i can find it again 02:19:36 (("fib"->a)->b)->(("fib"->b)->a~b),a;("fib"->"1")->"1" 02:19:49 oh 02:19:52 so it still does the weird thing 02:19:56 ok, how about 02:20:08 ok wait 02:20:19 in your example, oklopol 02:20:21 the weird thing? 02:20:21 the a 02:20:23 where does it go? 02:20:27 after the match? or at the end of the program 02:20:36 if at the end of the program, then that's my foo->(bar,baz) semantics 02:21:01 the way i see it, ";":s in the program code separate entries in a tuplespace, which is unordered 02:21:11 so i don't really think "where a goes in the code" 02:21:28 but in the case the tuplespace can be assumed unordered, in the end? doesn't matter 02:21:40 so 02:21:44 x->(y,z) 02:21:55 replaces x with y, and introduces z into the program space. 02:22:04 well, with your semantics, yes 02:22:13 what's your semantics for x->(y,z) 02:22:23 oh, well, actually, that's my semantics, sorry :) 02:22:25 also oklopol you keep missing out the final ; 02:22:32 indeed i do, but i know it's there. 02:22:37 noticed myself too 02:22:45 should i remove the need for it? 02:23:10 hmm... perhaps, ";" seems more like a separator than a delimiter to me 02:23:27 it's used as both, so i don't think it matters. 02:23:34 actually 02:23:40 if you have it be a separator 02:24:00 oklopol: on the LHS, what does (a,b) do? 02:24:03 e.g. in (a,b)->c 02:24:04 then a ";" in the end of a program means the string "", possibly, which is printed, as a nop 02:24:23 it finds something matching a and something matching b, and returns c in the tuplespace 02:24:37 a and b will share variables in pattern matchin 02:24:38 *g 02:25:10 oklopol: ...? 02:25:19 hmm? 02:25:22 (a,b)->c "find both a and b, and replace them with c?" 02:25:32 yes. 02:25:39 but that's totally different from your RHS , semantics 02:25:43 no 02:25:44 it's not related 02:25:47 ok then i guess 02:26:09 wait, i'll clear my thoughts a bit 02:26:23 oklopol: remember that it would find ALL as and Bs 02:26:24 it's really annoying to think when the text appears a minute after i've written it 02:26:28 because replacements replace EVERYTHING they can find 02:27:05 you can think of the "replace all" thing as "replace one, repeat" 02:27:19 in which case matching multiple tuples at the same time is okay. 02:27:44 -!- helios24_ has joined. 02:28:28 -!- helios24 has quit (Read error: 113 (No route to host)). 02:28:52 basically, a rule (comma-separated tuples)->(comma-separated tuples) means "repeatedly find a set of tuples that match left side, put their free variables to corresponding vars in the right side, remove all left-side tuples, add all right-side tuples" 02:29:19 this works like expected for comma-separated lists with just one element == like before. 02:29:59 that was not a fully formal explanation for i am very sleepy, i can maybe give you an example if you didn't get it 02:30:26 "a"->"b"; "a"; ==> "a"->"b"; "b"; 02:30:33 the case where only one tuple on both sides 02:30:45 oklopol: OH! how do you match a,b pairs???? 02:30:54 left side ("a") was removed from the tuplespace 02:30:54 (OH DEAR YOU LOSE.) 02:31:01 right side was added 02:31:04 Also you can't nest ,s 02:31:09 So blahhh your semantics suck. 02:31:18 k 02:31:23 :P 02:31:58 sure you can nest them, actually 02:32:15 how 02:33:10 ... 02:33:50 (a,b->"c")->("d"->a,b); "a","b"->"c"; "d"; => (a,b->"c")->("d"->a,b); "d"->"a","b"; "d"; => (a,b->"c")->("d"->a,b); "d"->"a","b"; "a"; "b"; and modify semantics so that strings are gc'd (and thus printed) once nothing matches on them 02:34:09 sorry, lag addes a few minutes to my response time 02:34:11 *adds 02:34:27 umm that makes no sense :D 02:34:35 also i don't want to remove strings once nothing matches on them 02:34:37 :( 02:35:04 well, your language ;-) 02:35:15 i'm going to remove , for now until we can think of a sane way to do this stuff :P 02:35:30 (my semantics are perfect though :)) 02:35:39 (uhhh, touch them) 02:35:49 ? 02:36:09 (parens are there so you know to ignore it) 02:37:24 yeah 02:37:25 but 02:37:28 (uhhh, touch them) 02:38:13 a comma-separated list is not hard to match. it's a linear rise in complexity, if you allow patterns in the left side to match the same tuple 02:38:38 a,b->"b";"a" ==> a,b->"b";"b" 02:39:14 oklopol: yes but nesting them makes not a lot of sense 02:39:38 I think what we need is an implementation of the current spec first 02:39:41 and then think about it 02:39:49 what language would be best suited for it though? 02:40:04 we need to do parsing, loads of mutability on the program, and lots of pattern matching 02:40:12 well, "," would only have the special property on the topmost level, otherwise it'd just be a separator 02:40:19 which already is how "->" works. 02:41:14 what about "(uhhh, touch them)"? a request to touch the semantics to feel how soft they are. 02:41:18 hmm 02:41:19 not soft 02:41:40 something. 02:41:50 anyway! 02:41:51 implementation. 02:42:07 imple pimple! 02:42:22 i can make one quite quickly in python, but i'm trying to change language 02:42:31 to scheme 02:42:40 wise choice :-) 02:42:46 yeah python would be good for the mutability, but yeck 02:42:50 haskell might actually be good 02:42:54 but parsing will be a bitch. 02:43:07 parsec doesn't like paren := (expr) where expr has an option of paren 02:43:11 can't figure out why oh well 02:43:35 hmm, i think parsing is fairly straightforward 02:43:51 i guess 02:44:01 in haskell you mean 02:44:01 ? 02:44:07 sorry 02:44:11 yeah. 02:44:13 when i asked you hadn't answered. 02:44:17 parsec - haskell parsing library 02:44:20 and yeah. i had. :P 02:44:22 yeah 02:44:24 i know you have lag though 02:44:35 parsec rules! 02:44:36 yep, "lag", my xchat is broken. 02:44:47 oklopol: really the perfect language would be CRTL heh :| 02:44:54 haha, indeed :P 02:45:14 it's like, when haskell was first formalized 02:45:25 'shit guys implementing this is gonna be HELL' 02:45:38 'well you could use monads and polymorphic type classes generalizing over pattern matchers...' 02:45:45 'yeah but that's what we're trying to implement' 02:46:00 :) 02:46:44 most of crtl is fairly straightforward, the problem is if you want efficiency, you need to keep a conceptual separation of data and program 02:47:10 not that anything could ever be hard in python... god i miss python :-) 02:47:15 stopped using it earlier today 02:47:30 yes it is very unelegant but has that extra ZAM which scheme lacks 02:47:48 indeed, and i kinda like the ZAM 02:47:50 oklopol: for mutable things like this (i.e. not elegant, pattern-weaving and folding algorithms) maybe you should check out haskell 02:47:58 i've actually found its mutability to be nice. 02:48:06 i know haskell 02:48:08 somewhat. 02:48:21 i never fully understood monads :-) 02:48:42 * oklopol is a rare case! 02:48:53 ZAM = expressive type system? 02:49:12 no, python has that not. 02:49:13 faxlore: python has one of them? 02:49:18 no.. 02:49:23 neither does scheme 02:49:24 oklopol: i figured out monads a while ago 02:49:26 well, you know... ZAM 02:49:30 oklopol: forget about 'do' 02:49:33 and think about it like this 02:49:44 i never actually tried that hard 02:49:54 (do a <- b; c) is (b >>= (\a -> c)) 02:49:58 hoped i'd just get them eventually... 02:50:01 (do a; b) is (a >> b) 02:50:05 yeah 02:50:06 then, what happens is 02:50:12 "a" recieves the monad as an implicit perameter 02:50:15 and returns the 'new monad' 02:50:18 yeah 02:50:30 in >>, that just means the monad b gets is what a 'returned' as the monad 02:50:51 in >>= the RESULT that a returns is passed to the function, which also gets its implicit monad changed to what a's monad became 02:51:01 so monads have an implicit part and an explicit part 02:51:08 yeah, kind of 02:51:11 and the way e.g. IO works 02:51:23 in m >>= l, the explicit one is what is given to l 02:51:25 as a param 02:51:36 return lifts it back on the implicit param 02:51:37 (>>=) = flip concatMap -- :D 02:51:40 on the "monad" 02:51:42 is that the monadic values 'bubble up' into functions all of which must have the IO monad themselves (let's ignore unsafePerformIO for now) and then the runtime system at the top does the evil 02:51:57 oklopol: yeah, remember though that when you do "return x" in a monadic function that's one of two return values 02:52:00 that's the "real return value" 02:52:04 the monad is "implicitly" passed on 02:52:22 yeah 02:52:29 and now you understand how monads can be stateful: they're implicitly passed around (obviously the implementation is a hell of a lot more efficient than that, haha ;)) 02:52:53 see now that wasn't hard was it :-) 02:52:59 i guess i kinda understood that already 02:53:08 i just feel there's still something missing :-) 02:53:10 oh, right 02:53:17 should read the monad transformers 02:53:17 can you formalize a>>b again? 02:53:35 http://uebb.cs.tu-berlin.de/~magr/pub/Transformers.en.html 02:53:35