00:04 bamaung: hi asumu i m new to racket and i m using emacs. is there any library like slime for racket?
00:06 (quit) wtetzner: Remote host closed the connection
00:08 asumu: bamaung: Try Geiser http://www.nongnu.org/geiser/
00:12 bamaung: asumu: thanks
00:12 asumu: NP.
00:17 (quit) zyoung: Remote host closed the connection
00:43 (part) bamaung: "ERC Version 5.3 (IRC client for Emacs)"
00:46 (quit) jao: Ping timeout: 272 seconds
00:48 (quit) getpwnam: Ping timeout: 265 seconds
00:59 (quit) rahul_: Ping timeout: 248 seconds
00:59 (quit) __rahul__: Ping timeout: 265 seconds
01:03 (join) rahul_
01:03 (join) __rahul__
01:05 Shambles_: I found a proofreading error on http://docs.racket-lang.org/guide/module-basics.html the diagram that shows how modules could be organized for a candy sorter has the contents of the database and machine blocks swapped.  The text is right, but the diagram is wrong.
01:07 asumu: Shambles_: Thanks for reporting it. It's been fixed in the pre-release though: http://pre.racket-lang.org/docs/html/guide/module-basics.html
01:07 Shambles_: asumu: Sorry.  Just thought I'd mention it.
01:12 asumu: No need for apology, please report doc errors zealously. :)
01:31 (quit) yoklov: Quit: bye!
02:14 (quit) jeapostrophe: Ping timeout: 252 seconds
02:18 (join) plobzik
02:21 (quit) plobzik: Read error: Connection reset by peer
02:23 (join) nilyaK
02:24 (quit) cdidd: Read error: Operation timed out
02:37 (quit) jonrafkind: Ping timeout: 245 seconds
02:50 (join) tfb
03:07 (quit) tfb: Quit: gone
03:20 (quit) rahul_: Ping timeout: 260 seconds
03:20 (quit) __rahul__: Ping timeout: 260 seconds
03:24 (join) antithesis
03:32 (join) __rahul__
03:33 (quit) jacius: Quit: Leaving
03:40 (join) rahul_
03:44 (quit) kvda: Quit: Computer has gone to sleep.
03:56 (join) Shvillr
03:57 (quit) Shviller: Ping timeout: 245 seconds
04:16 (join) bluezenix
04:19 (join) noelw
04:34 (join) tfb
04:50 (join) ambrosebs
04:52 (quit) ambrosebs: Client Quit
04:52 (join) ambrosebs
04:53 (join) cdidd
04:53 ambrosebs: How do I get a value out of a Vector in Typed Racket?
04:53 ambrosebs: I want to use vector-ref
04:53 ambrosebs: but it wants a (Vectorof a)
04:54 ambrosebs: do I convert the vector somehow?
04:59 (join) nilyaK1
05:01 noelw: What is the error message?
05:03 (quit) nilyaK: Ping timeout: 248 seconds
05:05 (quit) antithesis: Quit: yes leaving
05:15 (join) mceier
05:19 ambrosebs: Type Checker: Polymorphic function vector-ref could not be applied to arguments:
05:19 ambrosebs: Argument 1:
05:19 ambrosebs: Expected: (Vectorof a)
05:19 ambrosebs: Given:    Vector
05:20 ambrosebs: Argument 2:
05:20 ambrosebs: Expected: Integer
05:20 ambrosebs: Given:    One
05:20 ambrosebs: in: (vector-ref a 1)
05:21 ambrosebs: My code:
05:21 ambrosebs: (define: a : Any #{1 2})
05:21 ambrosebs: (and (vector? a) (vector-ref a 1))
05:22 ambrosebs: I'm testing occurrence typing
05:22 noelw: You've given a type Any
05:23 ambrosebs: is there no way to recover (Vectorof Any) ?
05:28 ambrosebs: is there a function (Vector Index -> Any) ?
05:29 ambrosebs: it seems odd to me that vector? isn't a predicate of (Vectorof Any)
05:35 ambrosebs: I guess I can make my own predicate
05:36 (join) jesyspa
05:43 ambrosebs: I can't find a solution. Impossible?
05:45 (quit) nilyaK1: Quit: Leaving.
05:47 (join) masm
05:51 (join) rahul__
05:54 (join) antithesis
06:00 ambrosebs: oh, (define-predicate vector-of? (Vector Any)) fixed it
06:01 (join) bitonic
06:01 (quit) __rahul__: Quit: Leaving
06:10 (join) camilo_
06:11 (quit) rahul_: Remote host closed the connection
06:11 (quit) rahul__: Remote host closed the connection
06:12 jamessan: ambrosebs: your define isn't typing a as (Vectorof Any), but as Any.  your code works fine for me with a slight adjustment -- (define: a : (Vectorof Any) #{1, 2})
06:23 (quit) ivan\: Quit: ERC Version 5.3 (IRC client for Emacs)
06:27 ambrosebs: jamessan: right, I have a function that takes (Any Any -> Boolean), and both parameters could be vectors
06:28 ambrosebs: and I wanted to recover that they could be (Vectorof Any)
06:37 ambrosebs: I think it makes sense that I need to define a new predicate
06:59 ambrosebs: shouldn't the type for hash-ref allow the optional 3rd argument to be a value, not just a lambda?
07:00 ambrosebs: > hash-ref
07:00 ambrosebs: - : (All (a b c) (case-> ((HashTable a b) a -> b) ((HashTable a b) a (-> c) -> (U b c))))
07:00 ambrosebs: strange use of "value", but I think you get the idea :P
07:04 ambrosebs: (hash-ref some-hash key #f) doesn't fit the type sig, but seems valid to me
07:10 (join) tyson1
07:10 (part) tyson1
07:12 (join) tyson1
07:13 (part) tyson1
07:19 (quit) camilo_: Read error: Operation timed out
07:47 (join) camilo_
07:49 (join) ivan\
07:54 (part) camilo_: "Ex-Chat"
08:05 (join) kanak
08:06 ambrosebs: is a Function in typed racket kind of like an intersection type of arities?
08:06 ambrosebs: would there be any harm representing it as an intersection?
08:17 (join) jeapostrophe
08:19 (quit) ivan\: Ping timeout: 248 seconds
08:20 (quit) tfb: Ping timeout: 255 seconds
08:21 (join) ivan\
08:42 (join) netrino
08:46 (join) zyoung
08:48 (join) getpwnam
08:50 (quit) Demosthenes: Ping timeout: 245 seconds
08:57 (join) stis
09:31 (join) jasoegaard
09:38 (quit) getpwnam: Ping timeout: 260 seconds
10:00 (join) snearch
10:01 (quit) snearch: Read error: Connection reset by peer
10:01 (join) snearch
10:05 (join) dnolen
10:06 (join) djcb
10:17 (quit) jeapostrophe: Read error: Operation timed out
10:22 (join) flying_rhino
10:32 (nick) samth_ -> samth
10:32 samth: ambrosebs: yes, `Function' is a collection of arities, which are treated as an ordered intersection
10:33 samth: see our PADL 2012 paper for a few more details about this
10:37 (join) aalix
10:37 ambrosebs: samth: perfect! thanks :)
10:41 (join) RacketCommitBot
10:41 RacketCommitBot: [racket] plt pushed 2 new commits to master: http://git.io/xuk7pQ
10:41 RacketCommitBot: [racket/master] win64: glib DLL depends on libintl - Matthew Flatt
10:41 RacketCommitBot: [racket/master] fix non-portable test - Matthew Flatt
10:41 (part) RacketCommitBot
10:46 ambrosebs: samth: is the decidability of intersection types related to order?
10:46 (join) getpwnam
10:47 samth: ambrosebs: i think it would be decideable even without that
10:48 samth: the important restriction for decideability is that it's limited to function types
10:48 ambrosebs: ah
10:50 ambrosebs: I was hoping to use general intersection types in typed clojure, but I don't know where it gets hard
10:50 ambrosebs: wasn't aware it was undecideable either
10:51 samth: with general intersection types, you can type all of the lambda calculus terms that terminate
10:53 ambrosebs: it that bad?
10:58 (quit) jrslepak: Quit: This computer has gone to sleep
11:03 samth: ambrosebs: that solves the halting problem
11:06 ambrosebs: samth: ok, I'm not understanding some terminology. I was expecting to be bad if we can type `non' terminating terms, and good if we can type terminating terms. Where have I gone wrong?
11:29 (quit) djcb: Read error: Connection reset by peer
11:31 (join) djcb
11:38 (join) jao
11:38 (quit) jao: Changing host
11:38 (join) jao
11:42 (join) wah_java
11:42 (quit) wahjava: Read error: Connection reset by peer
11:43 (nick) wah_java -> wahjava
11:49 (quit) netrino: Ping timeout: 244 seconds
11:51 (join) netrino
11:55 asumu: ambrosebs: The problem is you can set your type system up so that the set of type-checkable terms are the halting terms. That means running your type-checker is a decision procedure for the halting problem.
11:55 asumu: Which is undecidable.
11:56 asumu: OTOH, if your type system could identify all terms *but* the halting terms, it's probably just as bad.
11:56 asumu: (but that doesn't actually happen)
11:58 (quit) djcb: Read error: Connection reset by peer
12:00 (join) djcb
12:02 ambrosebs: asumu: interesting, thanks
12:04 (join) mye
12:04 (quit) djcb: Read error: Connection reset by peer
12:06 (join) djcb
12:09 (join) jeapostrophe
12:16 (join) jonrafkind
12:16 (quit) jonrafkind: Changing host
12:16 (join) jonrafkind
12:30 (join) RacketCommitBot
12:30 RacketCommitBot: [racket] plt pushed 5 new commits to master: http://git.io/SYEdkg
12:30 RacketCommitBot: [racket/master] Start improving file system usage - Jay McCarthy
12:30 RacketCommitBot: [racket/master] New archives, smaller - Jay McCarthy
12:30 RacketCommitBot: [racket/master] Only repair the big archives - Jay McCarthy
12:30 (part) RacketCommitBot
12:33 (quit) jesyspa: Ping timeout: 244 seconds
12:35 (join) jesyspa
12:54 friscosam: asumu: re: Jazzscheme modules.  I don't remember if jazzscheme uses gambit's module system, but the last time I looked it was not totally documented and (I think) you could have identifier collisions
12:59 (join) RacketCommitBot
12:59 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/3Fiq_A
12:59 RacketCommitBot: [racket/master] add "alias" and "converted-arguments-variant" properties - Matthew Flatt
12:59 (part) RacketCommitBot
13:10 (join) karswell
13:15 (quit) snearch: Quit: Verlassend
13:25 (join) bluezenix1
13:26 (quit) bluezenix: Ping timeout: 272 seconds
13:49 (quit) cdidd: Ping timeout: 265 seconds
13:50 (quit) djcb: Read error: Connection reset by peer
13:52 (join) djcb
14:01 (join) jrslepak
14:05 (join) sstrickl
14:10 (quit) acarrico: *.net *.split
14:10 (quit) cataska_: *.net *.split
14:10 (quit) jrslepak_neu: *.net *.split
14:10 (quit) jschuster: *.net *.split
14:10 (join) jrslepak_neu
14:10 (join) acarrico
14:10 (join) cataska_
14:10 (join) jschuster
14:31 (quit) djcb: Read error: Connection reset by peer
14:31 (join) djcb
14:33 anonus: is it possible to make geiser to display docs about racket's functions ?
14:37 (join) anRch
14:38 (join) bluezenix
14:41 (quit) bluezenix1: Ping timeout: 245 seconds
14:45 (part) sstrickl
14:45 (join) sstrickl_away
14:46 (nick) sstrickl_away -> sstrickl
14:47 (part) sstrickl
14:47 (join) sstrickl
14:50 asumu: anonus: I don't know if geiser does, but I know some people here have emacs bindings to do that.
14:50 asumu: stamourv: ^
14:50 (quit) acarrico: *.net *.split
14:50 (quit) cataska_: *.net *.split
14:50 (quit) jschuster: *.net *.split
14:51 (join) acarrico
14:51 (join) cataska_
14:51 (join) jschuster
14:55 (join) hash_table
14:55 (quit) bluezenix: Quit: Leaving.
14:56 (nick) hash_table -> brogrammer
14:58 (quit) jrslepak: Quit: This computer has gone to sleep
15:01 (join) ynniv
15:03 (join) bro_down
15:29 (quit) anRch: Quit: anRch
15:33 (join) noam
15:37 (join) bluezenix
15:44 (quit) jonrafkind: Quit: Ex-Chat
15:45 (join) jonrafkind
16:02 (join) jrslepak
16:17 (quit) Kaylin: Quit: Leaving.
16:18 stamourv: friscosam: Gambit doesn't have a module system. There's a 3rd party one (Black Hole), though.
16:19 stamourv: anonus: I have something lying around.
16:19 anonus: aand?
16:20 stamourv: It doesn't display docs inline, but it makes it possible to look up identifiers in the browser.
16:20 anonus: is it for geiser?
16:20 anonus: browser
16:20 anonus: it is a bit difficult
16:20 stamourv: foof's scheme-complete + eldoc does a bit of inline docs, but mostly for R5 stuff.
16:20 anonus: and geiser cannot into docs for racket?
16:21 stamourv: anonus: It works fine with geiser. It's just a function that finds the identifier under the cursor, launches the browser and searches for the identifier in the docs.
16:21 stamourv: anonus: I haven't looked into geiser much, so I don't know if it does it.
16:21 stamourv: I should post what I have somewhere.
16:21 stamourv tries to figure out how github gists work.
16:21 anonus: ok, so i just need to invent method to run local browser for emacs-over-ssh =/
16:22 stamourv: Oh, yeah, that could be an issue.
16:22 stamourv: The Racket docs should look fine in w3m, which runs fine through ssh.
16:22 stamourv: So maybe something like that could work.
16:22 anonus: btw i don't believe than no one but me got this problem before
16:23 anonus: i mean running local browser from remote server
16:24 stamourv: anonus: https://gist.github.com/2721375
16:24 stamourv: It's a hack, but it works.
16:26 anonus: interesting
16:32 (join) igibson
16:36 (quit) ynniv: Remote host closed the connection
16:43 (join) lem_e_tweakit
16:53 (quit) dnolen: Ping timeout: 265 seconds
16:56 (quit) kanak: Ping timeout: 252 seconds
17:04 (part) stis
17:05 (quit) dented42: Quit: Computer has gone to sleep.
17:06 (quit) getpwnam: Ping timeout: 244 seconds
17:06 (quit) lem_e_tweakit: Ping timeout: 265 seconds
17:09 shawnps: is there an easy way to collapse a list of lists into one list?  like '("foo" '("bar")) -> '("foo" "bar")?
17:09 shawnps: i guess that's not a list of lists necessarily
17:10 (join) yoklov
17:13 shawnps: flatten
17:13 (quit) antithesis: Quit: yes leaving
17:14 shawnps: ah but that doesn't work with pairs
17:15 shawnps: would like to do someting like: '((a . "foo") '((b . "bar") (c . "baz"))) -> '((a . "foo") (b . "bar") (c . "baz"))
17:21 jrslepak: so you want to leave improper lists alone, or just the ends of improper lists, or what?
17:22 jrslepak: also, is that second ' supposed to be there?
17:22 shawnps: from a list that contains pairs and lists of pairs, i'd like to have a list of pairs
17:23 shawnps: i thought i almost had it with this:
17:23 shawnps: (define (foolatten lst)
17:23 shawnps: (cond
17:23 shawnps: ((empty? lst) '())
17:23 shawnps: ((list? (car lst)) (cons (foolatten (car (car lst))) (foolatten (rest lst))))
17:23 shawnps: (else (cons (car lst) (foolatten (rest lst))))))
17:23 (join) nilyaK
17:23 shawnps: actually i can see a couple of mistakes with that
17:24 (join) duncan
17:24 duncan: does anyone know the enitre code for creating an animation where an image moves left right up or down when you hit the corresponding arrow keys
17:24 (nick) duncan -> Guest63637
17:26 (quit) djcb: Remote host closed the connection
17:33 bremner: Guest63637: you should stick around a bit this time
17:34 jrslepak: shawnps: e.g. (foolatten (car (car lst))) ?
17:35 shawnps: jrslepak: yeah, that's wrong isn't it
17:37 jrslepak: well, I have to go now, but you are probably also going to want an append in there rather than cons
17:37 shawnps: hmm, i figured changing that to just (car lst) would work but no
17:37 shawnps: jrslepak: cool, thank you
17:37 jrslepak: remember, flatting a list still leaves you with a list
17:37 jrslepak: *flattening
17:38 (quit) jrslepak: Quit: This computer has gone to sleep
17:39 (quit) jonrafkind: Ping timeout: 265 seconds
17:52 (quit) jeapostrophe: Ping timeout: 240 seconds
17:53 (join) tfb
17:53 (join) jonrafkind
17:56 (join) lem_e_tweakit
17:56 (join) getpwnam
17:57 (join) hash_table
17:57 (join) Quetzalcoatl_
18:07 (join) peeeep
18:13 friscosam: stamourv: I think blackhole was just starting up around the last time I looked at Gambit seriously.  At the time I thought they were trying to do a Planet like system.
18:14 (quit) flying_rhino: Quit: KVIrc 4.0.4 Insomnia http://www.kvirc.net/
18:14 danking: shawnps: Shawn Smith? What're you doing in the UK?
18:14 stamourv: friscosam: Could be. Been a while since I looked at it.
18:16 friscosam: yeah, there are so many nice things in Racket that everything else has a high bar to reach ;)
18:16 friscosam: so I don't look at much anymore
18:20 (join) jyc
18:22 (join) RacketCommitBot
18:22 RacketCommitBot: [racket] plt pushed 2 new commits to master: http://git.io/hhig_Q
18:22 RacketCommitBot: [racket/master] Pango kerning for Mac OS X - Matthew Flatt
18:22 RacketCommitBot: [racket/master] slideshow/pict: for `text' in caps mode, kern captials with following - Matthew Flatt
18:22 (part) RacketCommitBot
18:25 (quit) tfb: Quit: sleeping
18:32 (join) dnolen
18:43 (quit) nilyaK: Quit: Leaving.
18:46 (quit) mceier: Quit: leaving
18:47 (join) tfb
18:49 (join) jeapostrophe
18:51 (quit) jasoegaard: Ping timeout: 245 seconds
18:55 (quit) jeapostrophe: Ping timeout: 244 seconds
18:55 (join) jeapostrophe
19:07 (quit) yoklov: Quit: computer sleeping
19:08 (quit) jesyspa: Quit: leaving
19:22 (quit) bluezenix: Quit: Leaving.
19:26 (join) dented42
19:29 (join) jacius
19:51 (quit) tfb: Quit: sleeping
19:55 (quit) hash_table: Ping timeout: 245 seconds
19:55 (quit) getpwnam: Ping timeout: 245 seconds
19:56 (quit) lem_e_tweakit: Ping timeout: 272 seconds
20:11 (join) zyoung_
20:13 (quit) zyoung_: Remote host closed the connection
20:14 (quit) zyoung: Ping timeout: 245 seconds
20:25 (join) zyoung
20:28 (quit) zyoung: Remote host closed the connection
20:29 (join) kvda
20:35 (join) mye_
20:35 (quit) dnolen: Remote host closed the connection
20:38 (quit) mye: Ping timeout: 272 seconds
20:39 (join) dnolen
20:40 (quit) masm: Quit: Leaving.
20:45 (quit) jonrafkind: Read error: Operation timed out
21:05 shawnps: danking: hi :) i'm not sure why i'm on a UK freenode server, i'm in san francisco as usual
21:06 shawnps: danking: if anything i'd like to be in japan
21:08 (quit) netrino: Quit: Ave!
21:11 (join) cdidd
21:13 (join) mye__
21:17 (quit) mye_: Ping timeout: 245 seconds
21:17 (quit) jeapostrophe: Ping timeout: 245 seconds
21:20 (join) mye_
21:23 (quit) mye__: Ping timeout: 240 seconds
21:25 (join) mye__
21:27 Shambles_: I'm having trouble with part of the Racket Guide.  It doesn't discuss contracts with multiple arguments, but no optional or keyword arguments.  I'd expect this is the most common case.  From what I can gather from the documentation (-> foo? bar/c baz/c) gets read as "this function expects a foo and bar type argument, and returns a baz type one".
21:29 Shambles_: That's all well and good if I only return one argument, which is probably the commonl case, but what happens if you use 'values'?  The documentation (Reference, not Guide) mentions the contract has to specify results for each value, but not how this is done.  It seems rather difficult since there's no grouping to depict what's input and what's output in the contracts.
21:29 (quit) mye_: Ping timeout: 260 seconds
21:32 offby1: I assume you use ->*
21:32 offby1: let's see.
21:33 offby1: yeah, use (values this/c that/c) in the "range" part.
21:33 offby1: http://docs.racket-lang.org/reference/function-contracts.html?q=-%3E*#(form._((lib._racket/contract/base..rkt)._-~3e*))
21:33 asumu: rudybot: (contract (-> integer? (values integer? integer?)) (lambda (x) (values x x)) 'pos 'neg)
21:33 rudybot: asumu: your sandbox is ready
21:33 rudybot: asumu: ; Value: #<procedure>
21:33 asumu: rudybot: (define f (contract (-> integer? (values integer? integer?)) (lambda (x) (values x x)) 'pos 'neg))
21:33 rudybot: asumu: Done.
21:33 asumu: rudybot: (f 5)
21:33 rudybot: asumu: ; Value: 5
21:33 rudybot: asumu: ; Value#2: 5
21:33 asumu: Like so
21:33 offby1: oh, you don't need ->*
21:33 offby1 makes a mental note
21:33 offby1: on a mental Post-It
21:34 Shambles_: What bothers me is: The ->* contract combinator produces contracts for functions that accept optional arguments (either keyword or positional) and/or arbitrarily many arguments.
21:35 offby1: why does that bother you?
21:35 Shambles_: If I write a function like I usually would, to accept a fixed number of arguments and return a fixed number of values (but more than one!), it sounds like I shouldn't use that.
21:35 offby1: What bothers _me_ is: the refrigerator running loudly at night.  Wakes me up
21:35 Shambles_: As for why I care about the multiple values thing... well, Python got me hooked on using it.  It's actually pretty darn handy to not have to unpack the things every time.
21:36 offby1: Python doesn't return multiple values, if I recall correctly; but it _does_ let you deconstruct easily.
21:37 Shambles_: Essentially, yes.  You can    x, y = someFunction(blah)
21:37 Shambles_: Which can be really nice.
21:37 Shambles_: You usually return a tuple.
21:39 Shambles_: So I don't really need ->* ?  The documentation is right, it's only for optional or keyword arguments?  I can use '(values whatever? here?)' as the end of the contract?
21:39 offby1: *shrug*
21:39 offby1: it's after 5:00 so I don't answer any hard questions.
21:42 (quit) dented42: Quit: Computer has gone to sleep.
21:43 (quit) cdidd: Remote host closed the connection
21:43 (join) jeapostrophe
21:44 (join) nilyaK
21:52 (quit) ambrosebs: Ping timeout: 244 seconds
21:53 asumu: Shambles_: I'm confused. What are you trying to do?
21:53 asumu: ->* is only for optional/keyword arguments.
21:53 Shambles_: How would you write a contract for a function that takes no arguments?  Say one you just call to update the screen after you've drawn to a backbuffer.
21:53 asumu: (-> result/c)
21:54 asumu: rudybot: ((contract (-> void?) void 'pos 'neg))
21:54 rudybot: asumu: Done.
21:54 asumu: rudybot: ((contract (-> integer?) void 'pos 'neg))
21:54 rudybot: asumu: error: #<undefined>:1:1: self-contract violation, expected: integer?, given: #<void> <NEWLINE>   contract from: pos, blaming: pos <NEWLINE>   contract: (-> integer?) <NEWLINE>
21:54 (quit) mye__: Ping timeout: 252 seconds
21:54 asumu: (sorry bad example, but void was the first nullary function to come to mind)
21:55 Shambles_: asumu: I'm trying to understand how to write contracts, working through the guide.  There's nothing in the guide for the common case of wanting to accept multiple values, but all values are required.  It also doesn't explain contracts for returning multiple values, but offby1 seems to have answered how you do that.
21:55 Shambles_: So I'm guessing the simplest contract would be (-> void?) for a function that takes no arguments and returns nothing?
21:56 asumu: Oh, that's an odd omission.
21:57 asumu: Shambles_: That depends on if you care about the return type.
21:57 asumu: That contract will really enforce that it "does nothing" as opposed to return a string.
21:57 Shambles_: asumu: It's probably easier to notice when you don't already know the material.
21:57 asumu: (-> any) is the simplest contract for a no-argument function, I think.
21:57 asumu: (if you don't care that it's truly void)
21:58 asumu: Also, there are, it turns out, some examples with multiple arguments in 7.4. But yeah, probably better to show one earlier.
21:58 Shambles_: asumu: Yeah, I read about the performance implications of checking void versus using just "any/c".  I figure if you're using contracts you've decided it's worth giving up some performance to have reasonable debugging.  I like DbC (I'm actually surprised Racket comes with it), and am not one to worry about the overhead for them.
21:59 Shambles_: I'd rather it shout at me for screwing up and returning something.
21:59 asumu: Maybe it's surprising, but Racket had the first higher-order contract system.
22:00 asumu: Yeah, I generally agree with you.
22:00 asumu: I am in a habit of writing (-> any) though I've been writing contracts that go in the core codebase, so I try to make them efficient.
22:00 Shambles_: That is surprising.  I figured it'd be one of the ML or Haskell variants.  I'm not a big fan of 'fancy' type systems, but I do like contracts.  I don't have trouble passing the right type.  I do have trouble passing nonsense values at times (and yes, I know you can bludgeon values into types, but you know that's not easy).
22:00 asumu: *though since
22:01 (join) cdidd
22:01 asumu: Some people have come up with contract systems for Haskell, I think. But after Racket.
22:02 Shambles_: It's super awesome to have unit tests blow up your code right where the bug starts, rather than having to stare at it in the debugger, and if you miss something in the unit tests, have it still start the debugger exactly where the problem started, rather than playing detective.
22:02 asumu: Also contracts are not so great with laziness.
22:02 asumu: (you either don't check violations or you break the laziness)
22:03 Shambles_: I wonder if that's true in general?  I've used DbC addons  in Python, and it works fine with generators, though granted that's a limited example of lazy evaluation.  It's a mighty useful example though.
22:05 (join) jrslepak
22:05 asumu: So if you have a lazy sequence (like with a generator) and you attach a contract like (sequence/c integer?) then either 1) you check that every element is an integer? and thus break the laziness
22:05 asumu: or 2) you don't check it until someone requests an element, in which case you may never signal a violation.
22:06 asumu: Unless there's some other case I forgot about. There's a paper reviewing various approaches to this that Haskell people have done.
22:08 asumu: Oh right, there's worse: you can end up blaming the other party due to laziness.
22:09 Shambles_: Python does 2.
22:09 Shambles_: Which I'm okay with really.
22:09 asumu: Since in a fully lazy language, the lazy value might flow somewhere else before its forced. And then might blame some code elsewhere.
22:11 Shambles_: Yeah.  I think the 'right place' for contracts is trapping bugs at their origin, rather than as some abstract pointless formalism, or as a full solution to detecting bugs.  There won't ever be a full solution to detecting them.  The machine can only know what you said, not what you meant.  And even for internal-consistency errors there are problems (Godel's incompleteness theorem).
22:12 Shambles_: 2 works pretty well in practice.  If the code never blows up during unit testing or during use, it's worth questioning whether the bug 'matters'.
22:13 asumu: Well, I think it'll work fine for a lazy data structure in a strict language.
22:13 asumu: But if the language is lazy too, it might not be forced until way later.
22:13 asumu: In which case, you have lost track of the origin of the bug.
22:14 Shambles_: Assuming the contract implementation works right it should still tell you what the state was, what contract was violated, and what the relevant code was, even if the value doesn't get 'pulled through' when you expect it.
22:15 asumu: Yes, that's true. It just might give you a misleading location.
22:15 Shambles_: Besides, Racket appears to be eager.  At least for Racket this stuff isn't a problem, right?
22:16 (join) zyoung
22:16 asumu: Might be an issue for Lazy Racket. But yeah, I was just pondering about Haskell w/ contracts.
22:18 Shambles_: I'm not a big fan of 'default' lazy evaluation, but it is admittedly very handy at times.  I wonder if the world will eventually catch on to (hopefully non-default) speculative evaluation.  I think it's been around since something like SISAL.
22:29 (join) Demosthenes
22:39 (join) RacketCommitBot
22:39 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/HiTW0Q
22:39 RacketCommitBot: [racket/master] Fix a bug with class/c & interface ctc interaction - Asumu Takikawa
22:39 (part) RacketCommitBot
22:42 (quit) stamourv: Ping timeout: 244 seconds
22:43 (join) stamourv
22:43 (quit) stamourv: Changing host
22:43 (join) stamourv
22:48 (quit) jao: Ping timeout: 240 seconds
23:00 shawnps: so, if i want to use alist->form-urlencoded, i can only have strings as the second element in a pair?  something like (alist->formurlencoded '((api_key . (get-pubkey user)) doesn't work :(
23:01 shawnps: er, alist->formurlencoded*
23:01 (join) jonrafkind
23:13 (quit) noelw: Ping timeout: 244 seconds
23:16 asumu: Try `((api_key . ,(get-pubkey user))
23:16 asumu: (assuming get-pubkey produces a string)
23:16 shawnps: thanks
23:17 shawnps: what does that do?
23:17 asumu: shawnps: it lets you escape from a quotation back into Racket land for a bit and sticks the result into the quoted expression.
23:17 asumu: See http://pre.racket-lang.org/docs/html/guide/qq.html
23:18 (quit) jeapostrophe: Ping timeout: 244 seconds
23:18 asumu: Where ` starts a "quasi-quote" and , lets you escape into Racket.
23:18 shawnps: hmm, that didn't work unfortunately
23:18 asumu: Did it at least give a different error?
23:18 shawnps: string->bytes/utf-8: expects argument of type <string>; given ',(get-pubkey user)
23:19 asumu: You want ` (backtick) and not ' (quote)
23:19 shawnps: ah!
23:19 shawnps: that worked
23:19 shawnps: thank you :D
23:20 asumu: NP. :)
23:20 shawnps: sorry, i saw you mentioned quasi-quote but hadn't seen you use it before
23:21 (join) getpwnam
23:21 (join) lem_e_tweakit
23:22 (join) hash_table
23:28 (quit) dnolen: Ping timeout: 244 seconds
23:42 (join) nilyaK1
23:45 (quit) nilyaK: Ping timeout: 255 seconds
23:57 (join) dented42
23:59 (quit) stamourv: Ping timeout: 244 seconds