00:00 (join) vipjun
00:09 (quit) zyoung: Remote host closed the connection
00:09 (join) bjz
00:11 (quit) yacks: Ping timeout: 276 seconds
00:14 (quit) Kaylin: Quit: Leaving.
00:24 (join) brum
00:31 (quit) brum: Remote host closed the connection
00:33 (join) jojo_
00:36 (join) noam__
00:39 (quit) noam_: Ping timeout: 256 seconds
00:54 (join) jao
00:54 (quit) jao: Changing host
00:54 (join) jao
00:56 (join) mithos28
01:01 (quit) mithos28: Quit: mithos28
01:12 (join) mithos28
01:12 (join) yacks
01:24 (quit) mceier: Quit: leaving
01:26 jojo_: Anyone know why I might be getting ptr-set!: given value does not fit primitive C type, C type: _double*, given value: #t, from the (send region% in-region x y) function?
01:28 (quit) Fare: Ping timeout: 260 seconds
01:41 (join) vkz
01:46 asumu: jojo_: that sounds possibly like a bug. Can you report it? (either via DrRacket or the web interface)
01:46 asumu: jojo_: you're not using the FFI right? Just the racket/draw library?
01:46 (join) basdirks
01:48 jojo_: yeah, just racket/draw
01:49 jojo_: I've been thinking its a problem with my drawing canvas but can't track it down, I'll report if its not my error
01:50 mithos28: asumu: you claimed me adding the type bound to the tvars is unsound. Is that the same issue that you reported with map?
01:51 mithos28: This line is doing it very similarly to the way I did it, https://github.com/plt/racket/blob/master/collects/typed-racket/typecheck/tc-lambda-unit.rkt#L173
01:51 rudybot: http://tinyurl.com/abdwmc3
01:53 asumu: mithos28: it's not exactly the same issue, but it's related. I used to think the map special case was sound, but while I was coming up with the counterexample for your line, I found that it was broken too.
01:54 asumu: That `rest-type` line is suspicious, but since it's not type-checking a body I think it may be fine.
01:55 asumu: Also, polydots is tricky because its scoping is currently still broken too. Unfortunately, I think type variables need some deeper fixes to work properly.
01:56 mithos28: Do we have a model of what correct scoping should be?
01:59 asumu: We don't. But at the very least, every (All (a) ...) should introduce a new lexical scope for the body it annotates.
01:59 asumu: (and similarly with define:, pdefine:, etc.)
01:59 asumu: Right now, this is not the case with polydots.
01:59 (join) karswell
02:01 mithos28: What does it mean for a type to annotate an expression? if it is used as an expected type for that expression?
02:03 (quit) vkz: Quit: vkz
02:04 asumu: To a first approximation. Though I think that's probably not a precise enough definition. For example, suppose there's (: f (All (a) (a -> a)))
02:04 asumu: and (define f (let () (: x (Option a)) (define x #f) ... some function ...))
02:04 asumu: The type variable `a` should not be bound in the let.
02:05 asumu: (it should be bound in the function though)
02:06 asumu: Anyhow, I've got to go to sleep. If you have any questions, I'll answer them later if you PM me them.
02:06 mithos28: k
02:07 mithos28: I'll think about that example, because I don't like that polymorphism is limited to functions
02:12 (quit) jao: Ping timeout: 256 seconds
02:17 (quit) basdirks: Ping timeout: 256 seconds
02:17 (quit) rohni: Quit: Leaving.
02:19 (quit) jonrafkind: Ping timeout: 248 seconds
02:23 (join) mceier
02:33 (join) rohni
02:39 (quit) ASau: Ping timeout: 264 seconds
02:51 (quit) racycle: Quit: racycle
02:55 (join) basdirks
03:09 (join) cdidd
03:15 (quit) mithos28: Quit: mithos28
03:20 (join) racycle
04:15 (quit) racycle: Quit: racycle
04:21 (join) anonymous97890
04:28 (join) noelw
04:29 (join) wsun
04:46 (quit) anonymous97890: Ping timeout: 240 seconds
04:54 (join) BakaKuna
04:55 (quit) TheGhostinaor: 
04:55 (join) TheBeerinator
04:59 (quit) tobih: Remote host closed the connection
05:00 (join) tobih
05:11 (quit) sirdancealo2: Ping timeout: 255 seconds
05:14 (join) bitonic
05:21 (quit) bitonic: Ping timeout: 250 seconds
05:32 (join) vkz
05:45 (join) bitonic
06:18 (join) Icarot
06:33 (join) soegaard
06:45 (quit) yacks: Ping timeout: 272 seconds
06:50 (join) jeapostrophe
06:50 (quit) jeapostrophe: Changing host
06:50 (join) jeapostrophe
07:07 (quit) soegaard: Quit: soegaard
07:17 (join) yacks
07:25 (quit) bjz: Read error: Connection reset by peer
07:25 (join) bjz_
07:51 (quit) BakaKuna: Ping timeout: 252 seconds
07:59 (join) Aune
08:07 (join) mizu_no_oto
08:11 (join) dzhus
08:12 (quit) yacks: Ping timeout: 260 seconds
08:14 (join) cnonim
08:15 cnonim: hi all
08:15 cnonim: how i can get arity or required args of some function?
08:15 (quit) mizu_no_oto: Quit: Computer has gone to sleep.
08:16 (join) sirdancealo2
08:18 (join) mizu_no_oto
08:23 (join) yacks
08:28 (join) RacketCommitBot
08:28 RacketCommitBot: [racket] plt pushed 3 new commits to master: http://git.io/wAyirw
08:28 RacketCommitBot: racket/master 9e0e2b9 Matthew Flatt: scribble/srcdoc: add `begin-for-doc'
08:28 RacketCommitBot: racket/master 81a0451 Matthew Flatt: ffi/unsafe/objc...
08:28 RacketCommitBot: racket/master 454f4c3 Matthew Flatt: slideshow: add `interactive'...
08:28 (part) RacketCommitBot
08:37 (quit) dzhus: Ping timeout: 250 seconds
08:40 (quit) Icarot: Ping timeout: 276 seconds
08:43 (quit) Aune: Quit: Hath Deprated
08:47 (quit) cored: Ping timeout: 248 seconds
08:49 (join) cored
08:49 (quit) cored: Changing host
08:49 (join) cored
08:49 (quit) bitonic: Ping timeout: 250 seconds
08:54 (join) zyoung
09:05 (join) didi
09:13 (join) serhart
09:17 (join) Fare
09:23 (quit) mizu_no_oto: Quit: Computer has gone to sleep.
09:27 (join) bitonic
09:32 Nisstyre: cnonim: procedure-arity
09:36 (join) mizu_no_oto
09:58 (quit) bitonic: Remote host closed the connection
09:59 (join) bitonic
10:00 (join) BakaKuna
10:05 (join) racycle
10:13 (join) ASau
10:20 (join) sstrickl
10:20 (quit) sstrickl: Changing host
10:20 (join) sstrickl
10:21 (join) soegaard
10:27 cnonim: how i can mark some stuff and check if this stuff marked then do some work with this stuff?
10:30 cnonim: example: i have list of some values
10:32 cnonim: and i do (map (lambda (v) (if (foo? v) v (mark v)) (list ...values))
10:33 cnonim: then id do (map (lambda (v) (if (marked? v) (bar v) v) (list ...marked-values))
10:34 didi: cnonim: Why don't you use `filter'?
10:34 (quit) sirdancealo2: Ping timeout: 255 seconds
10:34 cnonim: what should be 'mark' and 'marked?' ?
10:36 cnonim: didi: there's really not a list? i have some function which should return simple value or marked value
10:37 didi: rudybot: (map sqrt (filter even? (list 1 2 3 4)))
10:37 rudybot: didi: your sandbox is ready
10:37 rudybot: didi: ; Value: (1.4142135623730951 2)
10:37 cnonim: and then i should check which value get
10:40 cnonim: question is not how to filter values
10:41 cnonim: the question is how to mark the value to add some attribute which can then be checked
10:44 cnonim: i can create some struct like 'Marked' whick hold actually value and then check if i got this struct or anyghing else
10:45 cnonim: but thought maybe there is a better solution to this problem
10:47 didi: cnonim: Well, if you insist, you might want to create a vector, with the same size of your list, and mark the individual positions.
10:48 (quit) jeapostrophe: Ping timeout: 256 seconds
10:49 didi: This reminds me of file systems.
10:50 (quit) noelw: Quit: noelw
10:51 cnonim: yes I can, but the best way to mark, structure, or may be a pair of value and a some symbol or any other solution
10:52 (quit) ambrosebs: Remote host closed the connection
10:55 (quit) bitonic: Ping timeout: 276 seconds
11:00 (join) mithos28
11:00 (quit) basdirks: Ping timeout: 240 seconds
11:01 cnonim: i have functions which parses input and store some result in list
11:01 cnonim: these functions processed sequentially
11:02 cnonim: but some function doesn't return anything
11:03 cnonim: what should store in list for them?
11:03 (quit) rohni: Quit: Leaving.
11:06 cnonim: i thought storing last result it possible solution problem
11:07 cnonim: but in the list have duplicates
11:07 soegaard: cnonim: Store the marked values in an eq-hashtable. Then mark stores an element in the hash table and marked? simply looks it up.
11:07 soegaard: If you are worried about garbage collection use a weak hashtable.
11:09 cnonim: exactly
11:09 (quit) mceier: Quit: leaving
11:10 cnonim: i can store value in hashtable with random created key and store this key in list
11:11 soegaard: Btw. You mentioned parsing. If your objects are syntax-objects you can use syntax-properties to store extra information.
11:13 (join) bitonic
11:14 cnonim: example of parsing sequence
11:15 cnonim: (((get-sequence (get-string "hello") (get-char #\ ) (get-string "world")) get-true) '() '() input 0)
11:15 cnonim: it's return
11:16 cnonim: '("hello" #\space "world")
11:16 soegaard: asumu: https://gist.github.com/soegaard/5117604
11:16 cnonim: if input it's "hello world"
11:17 (join) anRch
11:17 cnonim: and if add epsilon to sequence
11:17 soegaard: You can use '("hello" #\space "world") as the key in the hash table.
11:17 cnonim: (((get-sequence (get-string "hello") get-epsilon (get-char #\ ) (get-string "world")) get-true) '() '() input 0)
11:17 cnonim: then result it's
11:18 cnonim: '("hello" "hello" #\space "world")
11:18 soegaard: Actually if you only want to know whether it is marked or not, you can store the marked values in a set.
11:19 cnonim: but i need
11:19 cnonim: '("hello" (marked "hello") #\space "world")
11:19 (join) jbclements
11:19 (join) Kaylin
11:19 cnonim: and them i can replace value before marked with market
11:20 cnonim: *marked
11:20 cnonim: and got
11:20 cnonim: '("hello" #\space "world")
11:20 asumu: soegaard: (add1 soegaard)
11:22 (quit) BakaKuna: Ping timeout: 245 seconds
11:22 soegaard: asumu: It's just, well, it is a solution I would have chosen, if it weren't for the task description.
11:22 soegaard: "it is not"
11:22 cnonim: soegaard: i understand you idea? i store "hello" in dictionary with random created key and store in list what key
11:23 cnonim: '("hello" keyRND #\space "world")
11:23 soegaard: Why do you need the random key?
11:24 cnonim: how to distinguish between "hello" and (marked "hello")?
11:28 (join) yeboot
11:29 yeboot: So the other day I tried making racket code to accept or  call a function pointer returned by C code
11:29 yeboot: that was fun
11:30 stamourv: yeboot: Did it work?
11:31 stamourv never used the Racket FFI.
11:31 yeboot: no, although I got it to see the pointer after a lot of hacking
11:31 yeboot: then again I was doing it on purpose to stretch the limits of the ffi
11:34 (quit) Kaylin: Ping timeout: 264 seconds
11:34 soegaard: cored: If the strings are uninterned, then the are not eq? . So if you are using an eq-hash the hashtable can tell the difference.
11:34 cored: soegaard: what?
11:34 soegaard: rudybot: eval (eq? (string #\f #\o #\o) (string #\f #\o #\o))
11:34 rudybot: soegaard: your sandbox is ready
11:34 rudybot: soegaard: error: : undefined; <NEWLINE>  cannot reference an identifier before its definition <NEWLINE>   in module: 'program
11:35 soegaard: cored: sorry that was for cnonim
11:36 soegaard: rudybot: eval (eq? (string #\f #\o #\o) (string #\f #\o #\o))
11:36 rudybot: soegaard: ; Value: #f
11:36 cored: soegaard: ok
11:36 (quit) mizu_no_oto: Quit: ["Textual IRC Client: www.textualapp.com"]
11:37 yeboot: rudybot: "hello world"
11:37 rudybot: yeboot: your sandbox is ready
11:37 rudybot: yeboot: ; Value: "hello world"
11:40 cnonim: racycle: (define end-of-input (lambda () #f)
11:40 cnonim: ups
11:40 cnonim: rudybot: (define end-of-input (lambda () #f))
11:40 rudybot: cnonim: your sandbox is ready
11:40 rudybot: cnonim: Done.
11:41 didi: rudybot: (let ((h "hello, world") (e "hello, world")) (eq? h e))
11:41 rudybot: didi: If you're talking about getting a *different* binding in, then the answer to samth is "no, they're not `eq?'", and that runs into other problems: specifically, requiring two modules that provide different `+'s.
11:41 didi: rudybot: eval (let ((h "hello, world") (e "hello, world")) (eq? h e))
11:41 rudybot: didi: your sandbox is ready
11:41 rudybot: didi: ; Value: #t
11:41 didi: That's weird.
11:43 cnonim: (define input (let* ([in (string->list "hello world")] [sz (length in)]) (lambda (i) (if (> i sz) end-of-input (list-ref in i)))))
11:43 cnonim: rudybot: (define input (let* ([in (string->list "hello world")] [sz (length in)]) (lambda (i) (if (> i sz) end-of-input (list-ref in i)))))
11:43 rudybot: cnonim: Done.
11:44 cnonim: (define get-true (lambda ($last input position) $last))
11:44 cnonim: rudybot: (define get-true (lambda ($last input position) $last))
11:44 rudybot: cnonim: Done.
11:44 cnonim: ((get-true) '() input 0)
11:44 cnonim: rudybot: ((get-true) '() input 0)
11:44 rudybot: cnonim: error: get-true: arity mismatch; <NEWLINE>  the expected number of arguments does not match the given number <NEWLINE>   expected: 3 <NEWLINE>   given: 0
11:44 bartbes: yeboot: why would that be a problem?
11:44 stamourv: didi: String literals are interned. IIRC that's a recent change.
11:44 bartbes: I imagine the ffi uses function pointers anyway
11:45 cnonim: rudybot: (get-true '() input 0)
11:45 rudybot: cnonim: ; Value: ()
11:45 yeboot: bartbes why would what be a problem
11:45 yeboot: calling function pointers
11:45 yeboot: ?
11:45 yeboot: there's no type for them
11:46 bartbes: that's awkward
11:46 yeboot: they're normally bound with a call
11:46 yeboot: functions I mean
11:46 didi: stamourv: oic. I was revisiting my prior understanding of `eq?' and friends.
11:46 stamourv: didi: `eq?' is a very risky thing to rely on.
11:47 cnonim: rudybot: (define get-any (lambda (next) (lambda ($last input position) (let ([$last (input position)]) (if (eq? $last end-of-input) #f (next $last input (add1 position)))))))
11:47 rudybot: cnonim: Done.
11:47 stamourv: It's usually ok to use it as a shortcut, e.g. when checking for something in a cache, and if you don't find it, compute the result.
11:47 cnonim: ((get-any get-true) '() input 0)
11:48 cnonim: rudybot: ((get-any get-true) '() input 0)
11:48 rudybot: cnonim: ; Value: #\h
11:48 stamourv: But depending on it to distinguish two objects is risky, except for objects whose identity is part of their interface, e.g. boxes.
11:48 cnonim: rudybot: ((get-any (get-any get-true)) '() input 0)
11:48 rudybot: cnonim: ; Value: #\e
11:48 asumu: (eq? #&5 #&5)
11:48 asumu: rudybot: (eq? #&5 #&5)
11:48 rudybot: asumu: your sandbox is ready
11:48 rudybot: asumu: ; Value: #f
11:49 didi: stamourv: Well, I find it useful for checking if two objects are the same. Or at least, I used to find.
11:49 stamourv: didi: "The same" is a very tricky concept.
11:49 asumu sort of expected #&5 to be eq? to #&5, but nope
11:49 didi: stamourv: True.
11:49 stamourv: Do you want intensional equality? Extensional equality?
11:49 stamourv: Something else entirely?
11:50 didi: stamourv: "The same" as in pointer "the same".
11:50 stamourv: `eq?' will reliably give you that. The
11:51 didi: stamourv: That's why I was surprised with my `let' example.
11:51 stamourv: "lossiness" comes from the fact that Racket often doesn't offer guarantees about what things can share pointers.
11:51 stamourv: IMO, that's the right thing because it frees up the implementation to translate your code more efficiently.
11:53 (quit) soegaard: Quit: soegaard
11:54 (join) shriphani
11:57 cnonim: guys, can you try run with https://gist.github.com/cNoNim/bf426f3c27bfa12ed53d
11:58 cnonim: and tellmy how i can rewrite get-epsilon to remove duplicates
11:58 cnonim: *tell me
11:59 stamourv: cnonim: Have you heard of the design recipe? (from HTDP)
12:00 cnonim: yes :)
12:00 stamourv: Great!
12:00 stamourv: So, where are your contracts, purpose statements and tests?
12:00 (join) jeapostrophe
12:00 (quit) jeapostrophe: Changing host
12:00 (join) jeapostrophe
12:01 stamourv: Without those, figuring out what your code is doing is very hard.
12:02 (quit) didi: Ping timeout: 248 seconds
12:13 (join) RacketCommitBot
12:13 RacketCommitBot: [racket] plt pushed 3 new commits to master: http://git.io/DMuppA
12:13 RacketCommitBot: racket/master df581c3 Eric Dobson: Remove unused argument to tc/rec-lambda/check.
12:13 RacketCommitBot: racket/master 1fd21ef Eric Dobson: Avoid TR internal error, by checking shape of args.
12:13 RacketCommitBot: racket/master 1334e8d Eric Dobson: Reduce usage of check-below....
12:13 (part) RacketCommitBot
12:14 (join) didi
12:14 (quit) yeboot: Quit: bye
12:16 cnonim: how write contract for variadic function?
12:16 stamourv: rudybot: doc ->*
12:16 rudybot: stamourv: your sandbox is ready
12:16 rudybot: stamourv: http://docs.racket-lang.org/reference/function-contracts.html#(form._((lib._racket%2Fcontract%2Fbase..rkt)._-~3e*))
12:19 (join) jonrafkind
12:22 (join) mceier
12:29 cnonim: is that https://gist.github.com/cNoNim/bf426f3c27bfa12ed53d better?
12:30 (join) basdirks
12:31 (join) dauterive
12:31 cnonim: stamourv: how i can use template of contract for reduce contract size?
12:33 (quit) sizz: Read error: Operation timed out
12:34 (join) sizz
12:36 stamourv: cnonim: That's better already. But where are your purpose statements?
12:37 stamourv: Re contract templates: Contracts are values, you can bind them to variables.
12:37 stamourv: rudybot: (define my/c (-> integer? char?))
12:37 rudybot: stamourv: Done.
12:38 cnonim: what is a purpose statements?
12:39 stamourv: They're the second step of the design recipe.
12:40 stamourv: cnonim: http://www.ccs.neu.edu/home/matthias/HtDP2e/index.html
12:40 (quit) jojo_: Ping timeout: 245 seconds
12:41 (quit) anRch: Quit: anRch
12:55 (join) jao
12:55 (quit) jao: Changing host
12:55 (join) jao
13:00 (quit) bitonic: Ping timeout: 248 seconds
13:01 (quit) didi: Ping timeout: 272 seconds
13:03 cnonim: stamourv: https://gist.github.com/cNoNim/bf426f3c27bfa12ed53d now?
13:06 (join) rohni
13:19 (quit) jbclements: Quit: jbclements
13:21 (quit) jeapostrophe: Ping timeout: 255 seconds
13:22 (join) anRch
13:25 (join) didi
13:33 (quit) rohni: Quit: Leaving.
13:40 sstrickl: stamourv: going to look at implementing that solution I forwarded you this afternoon, if you haven't already been working on it yourself
13:40 sstrickl: stamourv: (want to make sure I'm not duplicating work if you've already started)
13:44 (join) rohni
13:48 stamourv: cnonim: Looking at it now.
13:48 stamourv: sstrickl: I haven't really started. Focusing on an ICFP submission.
13:48 sstrickl: stamourv: okay, just making sure.  I'll see what I can do this afternoon, then let you know where that stands if I finish or have to stop for now.
13:49 stamourv: sstrickl: Sounds good.
13:49 stamourv: Thanks!
13:49 (join) jeapostrophe
13:49 (quit) jeapostrophe: Changing host
13:49 (join) jeapostrophe
13:50 stamourv: cnonim: Looks much better, yes. :)
13:50 stamourv: Even better would be to have a purpose statement for the whole module, too.
13:51 stamourv: I infer that you're implementing finite state automata, but it would be nice to say so explicitly.
13:58 (join) mizu_no_oto
14:00 (join) sirdancealo2
14:07 greghendershott: eli: asumu: Weird timing. Just yesterday I used span and break from srfi/1. First time ever using anything from that.
14:08 (quit) jonrafkind: Ping timeout: 256 seconds
14:16 eli: greghendershott: Nice timing indeed.
14:20 asumu: greghendershott: yeah, the goal is that you don't have to require SRFIs.
14:20 asumu wants to replace srfi/19 in the distant future
14:23 (quit) anRch: Quit: anRch
14:23 eli: asumu: BTW, since I'm waiting for something else, I'm hacking more on it now.
14:23 asumu: eli: on `takef` and friends?
14:24 (join) dyoo
14:24 eli: Yes.
14:25 asumu: Ok, sounds good. I'm assuming from the radio silence that there are no objections.
14:27 (join) jonrafkind
14:34 greghendershott: asumu: That sounds great. (assuming they're not renamed rrfi's)
14:35 greghendershott: I've skimmed just a very little Clojure code lately, and what I've read seems idiomatic to use srfi/1-sh funcs quite a bit.
14:36 greghendershott: I think that's what prompted me to stop starting to write span from scratch, and think to look for equivalent.
14:37 greghendershott: As much as I feel like I've learned, so far, I'm still peeling the Racket onion
14:38 eli: greghendershott: In this case you're getting into fresh onion layers...
14:39 greghendershott: ^ In a good way. Can get things done, but still having happy discoveries.
14:39 greghendershott: eli: I like fresh onions. Thank you.
14:40 eli: :)
14:45 (quit) dyoo: Quit: dyoo
14:47 (quit) sirdancealo2: Ping timeout: 260 seconds
14:54 (join) dyoo
15:12 (quit) jonrafkind: Ping timeout: 260 seconds
15:13 (quit) rohni: Quit: Leaving.
15:30 (quit) Fare: Ping timeout: 260 seconds
15:31 (quit) dauterive: Read error: Connection reset by peer
15:33 (join) rohni
15:43 (join) Fare
15:48 (join) dauterive
15:53 (quit) Nisstyre: Quit: Leaving
15:55 (quit) mario-goulart: Read error: Connection reset by peer
15:56 (join) mario-goulart
15:58 (quit) mizu_no_oto: Quit: ["Textual IRC Client: www.textualapp.com"]
16:01 (quit) yacks: Quit: Leaving
16:02 (join) rich1
16:24 (quit) jeapostrophe: Ping timeout: 252 seconds
16:24 (join) Icarot
16:38 (nick) pono -> pono_yolosl
16:39 (quit) rohni: Quit: Leaving.
16:42 rich1: as a learning exercise would it be a lang reader or macro to enable mathematica like function chaining...
16:42 rich1: foo[a, b, c] // bar => bar[foo[a, b, c]]
16:43 rich1: so in racket (foo a b c) // bar => (bar (foo a b c))
16:43 (join) rohni
16:48 stamourv: rich1: I'd say it's probably easier to implement as a macro.
16:48 stamourv: You'd need to wrap the region that uses that notation in an invocation of the macro, though.
16:49 stamourv: If you're looking for something more challenging, but probably nicer to use, you could implement a custom reader.
16:49 didi: rich1: For the functionality, you can take a look into `compose'.
16:49 stamourv: If you go down that road, good examples would be dyoo's excellent brainfudge tutorial, the at-exp syntax that comes with Racket, and asumu's sweet expressions package on PLaneT.
16:50 rich1: stamourv: just trying to keep things simple at the moment while i'm getting to grips with racket
16:50 stamourv: rich1: Ok, then the first option is probably easiest.
16:50 rich1: cool.
16:51 stamourv: You'd have a macro `mathematica', that you could use like that:
16:51 stamourv: `(mathematica (foo a b c) // bar)'
16:51 stamourv: It would traverse its input, looking for `//', and rearrange the pieces.
16:52 stamourv: For bonus points, you could allow multiple `//' and build a pipeline.
16:52 rich1: sounds like a plan - here comes my 1st macro (hopefully)
16:53 stamourv: :D
16:53 stamourv: Have you read greghendershott's Fear of Macros guide?
16:53 stamourv: I recommend it.
16:53 rich1: no, searching...
16:53 (join) mizu_no_oto
16:54 stamourv: http://www.greghendershott.com/fear-of-macros/
16:54 rich1: thanks. i just remember reading in the style guide use procedures instead of macros if possible
16:55 stamourv: Yep, that's a good rule of thumb.
16:55 stamourv: Macros are harder to debug than functions, and they can't be passed around like functions can.
16:56 stamourv: But, there are many things that functions can't do, that only macros can.
16:56 rich1: like, if ?
16:56 stamourv: So basically, unless you're trying to do one of these things, you're better off sticking with functions.
16:56 stamourv: Yes, `if' is an example.
16:56 stamourv: More generally, anything that introduces an exception to the rule "arguments must be evaluated".
16:57 stamourv: There are other things that only macros can do, like abstracting over binding of variables, introducing new syntax (which is what you're doing, with your infix composition syntax), etc.
16:58 rich1: i always add (require rackunit) 1st and do my tests - for my tests will i have to compare a quoted expression against the macro application?
16:58 stamourv: But in general, before you write a macro, it's always a good idea to ask yourself "could I do this with a function instead?". Often, the answer is yes, especially with higher-order functions.
16:58 (quit) mizu_no_oto: Ping timeout: 276 seconds
16:59 stamourv: Testing macros is hard, and we don't have a definitive solution yet.
16:59 stamourv: ryanc's been working on it, though, and I think some of that work is available in the `unstable' collection.
16:59 rich1: eg. (check-equals? '(bar (foo a)) (mathematica (foo a) // bar))
17:00 stamourv: That won't work, macros operate on richer data structures than s-expressions.
17:00 stamourv: For simple macros, testing the macro by testing its expansion usually works.
17:01 (join) mizu_no_oto
17:01 stamourv: So, in that case, you'd check that, if you apply them to the same argument, `(bar (foo a))' and `(mathematica (foo a) // bar)' give the same answer.
17:01 rich1: right ok
17:02 stamourv: For more complicated macros, there are other more complicated testing methods that give you more guarantees, but don't worry about it for now.
17:03 stamourv: For example, Typed Racket is implemented as 5k+ lines of macros, and we want some pretty strong guarantees about how it expands, so we need to do more.
17:03 stamourv: But I recommend starting simple. :)
17:05 rich1: i haven't got to Typed Racket yet - how does it compare to racket with contracts?
17:05 rich1: i assume contracts are still good in typed racket
17:07 stamourv: Typed Racket checks types at compile time, unlike contracts. It also catches errors even if they wouldn't actually happen when running the program. Contracts only catch bugs if you do end up calling the function with erroneous inputs (or if it does produce erroneous output).
17:08 stamourv: When you write contracts, you can use arbitrary Racket functions, which means that you can check pretty complicated properties, e.g. if a list is sorted. Types in TR are a lot more expressive than in most typed languages, but still have limits.
17:09 stamourv: As for mixing contracts and types, there's not a good way to do it yet. That's on the to-do list. :)
17:09 (quit) cored: Remote host closed the connection
17:09 (join) cored
17:11 cnonim: I first write the solution using a function, then rewrite using macros, for better performance at runtime
17:11 rich1: stamourv: cool
17:12 stamourv: cnonim: Be very careful with that. It's *really* easy to introduce bugs when you do that.
17:12 cnonim: exactly )
17:12 rich1: cnonim: what sort of scale improvement? 1%? i was happy using the optimisation coach in drracket that it inlines stuff on its own
17:13 stamourv: I recommend using the Optimization Coach to make sure that manual inlining is really necessary, i.e. the compiler doesn't automatically inline already.
17:13 stamourv: rich1: Re improvements: They can be significant. I've seen 60+%.
17:13 rich1: s/1%/1%,10%,50%?
17:13 rich1: ah ok
17:14 stamourv: rich1: Re Optimization Coach: it doesn't actually do the inlining, it only reports what the Racket compiler does.
17:14 stamourv: And recommends manual transformations you can do yourself it you want even more performance.
17:14 stamourv: cnonim: Re using macros for performance: Do you know about `define/inline'?
17:15 cnonim: stamourv: it's not only inline
17:15 stamourv: It does the job of writing the macro for you. It has the exact same interface as `define', but defines a macro instead.
17:16 stamourv: cnonim: Right, you can also do optimization as part of the expansion, but often the main benefit of using macros is the inlining.
17:16 stamourv: Inlining is the mother of all optimizations, it allows other passes to do a better job.
17:17 rich1: can or does racket do any algebraic optimisations. a simple example (not (not a)) => a?
17:18 stamourv: It can't do that one, because it's not correct:
17:18 stamourv: rudybot: (not (not 3))
17:18 rudybot: stamourv: ; Value: #t
17:19 rich1: quite correct :) i meant for booleans
17:19 stamourv: But the Racket compiler does constant folding, if that's what you're asking.
17:20 stamourv: If you want to see if it's happening, you can use `raco decompile', which shows you decompiled bytecode. It's not pretty, but still somewhat readable.
17:20 rich1: yeh
17:21 (join) anRch
17:22 cnonim: which difference between define-inline and define-syntax-rule?
17:22 dented42: in http://paste.lisp.org/+2WV7 , shouldn't the type checker see that node-left returns a tree, and because of the if testing if (node-left tree) is null, shouldn't it realize that it can't be null and therefore must be a node, and therefore shouldn't the snippet typecheck properly?
17:22 stamourv: cnonim: Here's an example:
17:22 dented42: probably could have asked that question better…
17:22 stamourv: rudybot: (define-syntax-rule (f x) (+ x x))
17:22 rudybot: stamourv: Done.
17:23 dented42: tl;dr, shouldn't that typecheck?
17:23 stamourv: rudybot: (f (begin (displayln "here!") 2))
17:23 rudybot: stamourv: ; Value: 4
17:23 rudybot: stamourv: ; stdout: "here!\nhere!\n"
17:23 stamourv: rudybot: (require racket/performance-hint)
17:23 rudybot: stamourv: Done.
17:23 stamourv: rudybot: (define-inline (f2 x) (+ x x))
17:23 rudybot: stamourv: Done.
17:23 stamourv: (f2 (begin (displayln "here!") 2))
17:23 stamourv: rudybot: (f (begin (displayln "here!") 2))
17:23 rudybot: stamourv: ; Value: 4
17:23 rudybot: stamourv: ; stdout: "here!\nhere!\n"
17:23 stamourv: rudybot: (f2 (begin (displayln "here!") 2))
17:23 rudybot: stamourv: ; Value: 4
17:23 rudybot: stamourv: ; stdout: "here!\n"
17:23 stamourv: rudybot: (define (f3 x) (+ x x))
17:23 rudybot: stamourv: Done.
17:24 stamourv: rudybot: (f3 (begin (displayln "here!") 2))
17:24 rudybot: stamourv: ; Value: 4
17:24 rudybot: stamourv: ; stdout: "here!\n"
17:24 cnonim: understood
17:24 stamourv: cnonim: That's one of the tricky things.
17:24 cnonim: what about my code? :) which difference between define-inline and define-syntax-rule?
17:24 cnonim: ups
17:24 cnonim: https://gist.github.com/cNoNim/bf426f3c27bfa12ed53d
17:24 cnonim: :)
17:24 stamourv: I recommend using define-inline, just in case you cange your code.
17:25 stamourv: You don't want to silently introduce bugs, do you. :)
17:26 stamourv: dented42: Does binding `(node-left tree)' to a variable work?
17:26 dented42: yes it does
17:26 stamourv: I would expect occurence typing paths to work here, though.
17:26 (quit) basdirks: Ping timeout: 256 seconds
17:26 dented42: but I'm curious as to why it doesn't realize that both calls to node-left should have the same value.
17:26 dented42: yes, exactly.
17:27 stamourv: I forget how paths work for structs, lemme look up the implementation.
17:27 (quit) serhart: Quit: Leaving.
17:28 (join) jonrafkind
17:28 (quit) jonrafkind: Changing host
17:28 (join) jonrafkind
17:29 stamourv: dented42: My guess is that it fails because the typechecker doesn't unfold `Tree' when it gets to that point.
17:29 stamourv: I don't know if there's a reason for that. And if there is, it really should be documented.
17:29 stamourv: Could you submit a bug report?
17:29 dented42: right
17:29 dented42: I'd be happy to
17:29 stamourv: Thanks!
17:30 dented42: should I link to the paste? or just paste the code into the bug report?
17:34 (quit) vkz: Quit: vkz
17:36 stamourv: I recommending putting the code in the bug report. It's pretty short, and who know how long the paste will be up.
17:37 cnonim: rudybot: (require typed/racket)
17:37 rudybot: cnonim: Done.
17:38 cnonim: rudybot: (list 1 '1 #\1 "1")
17:38 rudybot: cnonim: ; Value: (1 1 #\1 "1")
17:38 rudybot: cnonim: ; stdout: "- : (Listof (U Char One String)) [generalized from (List One One Char String)]\n"
17:38 cnonim: typed racket havn't tuple?
17:38 stamourv: It does. That's `List'.
17:39 stamourv: TR generalizes types before printing them, which is why you see both types.
17:40 dented42: ok
17:40 dented42: 'tis done
17:40 stamourv: Thanks!
17:41 cnonim: rudybot: (list #\  " ")
17:41 rudybot: cnonim: ; Value: (#\space " ")
17:41 rudybot: cnonim: ; stdout: "- : (Listof (U Char String)) [generalized from (List Char String)]\n"
17:41 cnonim: rudybot: (car (list #\  " "))
17:41 rudybot: cnonim: ; Value: #\space
17:41 rudybot: cnonim: ; stdout: "- : Char\n"
17:42 cnonim: rudybot: (cdar (list #\  " "))
17:42 rudybot: cnonim: error: #<undefined>:1:0: Type Checker: Polymorphic function cdar could not be applied to arguments: <NEWLINE> Domains: (Listof (Listof a)) <NEWLINE>          (Pairof (Listof a) b) <NEWLINE>          (Listof (Pairof a b)) <NEWLINE>          (Pairof (Pairof a b) c) <NEWLINE> Arguments: (List Char String) <NEWLINE>  <NEWLINE>   in: (#%app cdar (#%app list (quote #\space) (quote " ")))
17:42 cnonim: rudybot: (cdr (list #\  " "))
17:42 rudybot: cnonim: ; Value: (" ")
17:42 rudybot: cnonim: ; stdout: "- : (Listof String) [generalized from (List String)]\n"
17:42 cnonim: rudybot: (cadr (list #\  " "))
17:42 rudybot: cnonim: ; Value: " "
17:42 rudybot: cnonim: ; stdout: "- : String\n"
17:44 cnonim: i'm don't understand it's internally have type list of union or like tuple of several type?
17:44 cnonim: imho tuple it's like struct not list of union
17:45 stamourv: cnonim: Internally, the type is `(List Char String)'.
17:45 stamourv: `(Listof (U Char String))' is only there because TR generalizes types when showing them to you.
17:45 stamourv: In this case, it's not very useful, but when the internal type is big and complicated, generalization makes types more user-friendly.
17:47 cnonim: rudybot: (make-list 100 'foo)
17:47 rudybot: cnonim: ; Value: (foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo)
17:47 rudybot: cnonim: ; stdout: "- : (Listof 'foo)\n"
17:49 cnonim: rudybot: (cons 'bar (make-list 100 'foo))
17:49 rudybot: cnonim: ; Value: (bar foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo foo)
17:49 rudybot: cnonim: ; stdout: "- : (Listof (U 'foo 'bar))\n"
17:50 cnonim: rudybot: (cons 'bar (list 'foo))
17:50 rudybot: cnonim: ; Value: (bar foo)
17:50 rudybot: cnonim: ; stdout: "- : (Listof (U 'foo 'bar))\n"
17:50 cnonim: rudybot: (cons 'bar (list 'foo 'boo))
17:50 rudybot: cnonim: ; Value: (bar foo boo)
17:50 rudybot: cnonim: ; stdout: "- : (Listof (U 'foo 'bar 'boo))\n"
17:51 stamourv: cnonim: You're trying to get rudybot kicked for flooding?
17:51 stamourv: :)
17:52 cnonim: :)
17:55 stamourv heads out for the weekend.
17:55 dyoo: have a good weekend
17:57 (join) bitonic
18:01 (quit) Fare: Ping timeout: 250 seconds
18:06 (quit) Icarot: Ping timeout: 260 seconds
18:07 (join) Icarot
18:10 (join) zyoung_
18:12 (join) RacketCommitBot
18:12 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/bC6Ohg
18:12 RacketCommitBot: racket/master 539c25b Stevie Strickland: Fix with-contract so that contracts are only applied per mutation....
18:12 (part) RacketCommitBot
18:13 (quit) zyoung: Ping timeout: 250 seconds
18:14 (join) Fare
18:15 (quit) anRch: Quit: anRch
18:15 (quit) zyoung_: Ping timeout: 252 seconds
18:16 (join) zyoung
18:16 (quit) zyoung: Remote host closed the connection
18:17 (join) zyoung
18:19 (quit) zyoung: Remote host closed the connection
18:27 (quit) rich1: Quit: Leaving.
18:28 (join) soegaard
18:29 (quit) Icarot: Ping timeout: 248 seconds
18:38 (join) Icarot
18:45 (quit) Icarot: Remote host closed the connection
18:46 (join) Icarot
18:59 (quit) soegaard: Quit: soegaard
19:01 (quit) mizu_no_oto: Quit: ["Textual IRC Client: www.textualapp.com"]
19:01 (quit) sstrickl: Quit: sstrickl
19:15 (quit) vipjun: Ping timeout: 264 seconds
19:25 (quit) Icarot: Ping timeout: 250 seconds
19:31 (quit) Mattykins: Ping timeout: 276 seconds
19:43 (join) Nisstyre-laptop
19:43 (join) Mattykins
20:00 (quit) asumu: Ping timeout: 245 seconds
20:08 (join) asumu
20:09 (quit) dyoo: Quit: dyoo
20:13 (quit) bitonic: Read error: Operation timed out
20:16 (join) codeblack
20:18 (join) ambrosebs
20:23 (quit) ambrosebs: Remote host closed the connection
20:27 (join) ambrosebs
20:32 (join) dyoo
20:32 (quit) dyoo: Client Quit
20:37 (join) adu
20:41 offby1: rudybot: why you say foo so much?
20:41 rudybot: *offby1: ,'""foo""
20:48 dented42: awww, he's so cute!
20:52 offby1: that's why I wrote him!
20:53 rudybot blushes
20:55 (quit) didi: Ping timeout: 272 seconds
20:57 (quit) jonrafkind: Ping timeout: 256 seconds
21:02 (quit) dauterive: Quit: Leaving
21:16 (quit) cdidd: Ping timeout: 276 seconds
21:18 cnonim: rudybot: ((lambda (char) (and (char>=? #\a char) (char<=? #\z char))) #\h)
21:18 rudybot: cnonim: ; Value: #f
21:18 rudybot: cnonim: ; stdout: "- : Boolean\n"
21:18 cnonim: rudybot: ((lambda (char) (and (char<=? #\a char) (char>=? #\z char))) #\h)
21:18 rudybot: cnonim: ; Value: #t
21:18 rudybot: cnonim: ; stdout: "- : Boolean\n"
21:20 cnonim: rudybot: ((and (curry char<=? #\a)  (curry char>=? #\z)) #\h)
21:20 rudybot: cnonim: error: #<undefined>:1:0: Type Checker: Cannot apply expression of type (U False (Char -> Boolean)), since it is not a function type <NEWLINE>   in: (#%app (if (#%app curry char<=? (quote #\a)) (#%expression (#%app curry char>=? (quote #\z))) (quote #f)) (quote #\h))
21:21 cnonim: rudybot: ((and (curry char<=? #\a)  (curry char>=? #\z)) #\h)
21:21 rudybot: cnonim: error: #<undefined>:1:0: Type Checker: Cannot apply expression of type (U False (Char -> Boolean)), since it is not a function type <NEWLINE>   in: (#%app (if (#%app curry char<=? (quote #\a)) (#%expression (#%app curry char>=? (quote #\z))) (quote #f)) (quote #\h))
21:21 cnonim: rudybot: init
21:21 rudybot: cnonim: your sandbox is ready
21:21 cnonim: rudybot: ((and (curry char<=? #\a)  (curry char>=? #\z)) #\h)
21:21 rudybot: cnonim: ; Value: #t
21:21 cnonim: why it's doesn't work in typed/racket?
21:24 bremner: cnonim: I _think_ because the and might return #f, as far as the type checker knows
21:24 cnonim: this is bug? or how i can rewrite?
21:26 bremner: I'm not sure the "point-free" way; you could use an explicit lambda
21:28 bremner: I think the untyped way kindof "works" by fluke: last argument to and is returned because neither is #f
21:30 cnonim: rudybot: ((and (curry char<=? #\a)  (curry char>=? #\z)) #\0)
21:30 rudybot: cnonim: ; Value: #t
21:30 cnonim: :)
21:30 cnonim: exactly
21:30 cnonim: thx
21:33 cnonim: how i can rewrite (lambda ($) (foo $) $) in "point free" way? or it's not possible?
21:35 (join) Icarot
21:37 (quit) adu: Quit: adu
21:54 (join) adu
22:02 (join) mizu_no_oto
22:15 eli: asumu, stamourv: ping
22:17 (join) sstrickl
22:17 (quit) sstrickl: Changing host
22:17 (join) sstrickl
22:26 (quit) Icarot: Remote host closed the connection
22:29 (join) yacks
22:52 (quit) cnonim: Read error: Connection reset by peer
22:53 offby1: multi-cast ping!
22:54 (join) lemald
23:06 (join) RacketCommitBot
23:06 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/c1u8_Q
23:06 RacketCommitBot: racket/master 57e9001 Stevie Strickland: Fix a couple of identifer references that were overlooked.
23:06 (part) RacketCommitBot
23:07 mithos28: offby1: pong
23:15 offby1 's arena is corrupt!
23:15 offby1: I got a pong without a matching ping!
23:16 mithos28: you did a multi cast
23:16 offby1: no, _eli_ did a multicast.  I was just observing that fact.
23:16 offby1: Or did I screw up the use/mention distinction again?
23:16 mithos28: it wasn't obvious what you were doing
23:19 offby1 is a master of subtlety
23:19 offby1: that's why
23:20 offby1: crap, now I need to arrange to expect pongs from ... everybody.
23:23 (quit) mithos28: Quit: mithos28
23:34 (join) mithos28
23:37 (quit) mithos28: Client Quit
23:38 (join) RacketCommitBot
23:38 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/QMr9YA
23:38 RacketCommitBot: racket/master 71a64bf Stevie Strickland: Remove additional cruft.
23:38 (part) RacketCommitBot
23:40 (quit) ambrosebs: Remote host closed the connection
23:40 offby1: "Mario Bros Cruft Removal" -- since 1922
23:43 (quit) mizu_no_oto: Quit: Computer has gone to sleep.
23:58 asumu: eli: pong
23:58 eli: Two things
23:59 offby1: An apple, and a banana.
23:59 offby1: That's two
23:59 eli: 1. The name should be `splitf-at' or `split-atf'.  I'm going for the first.
23:59 eli: 2. There's also a problem with the argument order