00:02 (quit) swartzcr: *.net *.split
00:02 (quit) cataska: *.net *.split
00:02 (quit) sethalves: *.net *.split
00:02 (quit) jamessan: *.net *.split
00:06 (join) swartzcr
00:06 (join) cataska
00:06 (join) sethalves
00:06 (join) jamessan
00:49 (quit) jrslepak: Quit: What happened to Systems A through E?
01:02 (join) Fare
01:09 (quit) mithos28: Quit: mithos28
01:15 (join) nilyaK
01:42 (quit) jeapostrophe: Ping timeout: 252 seconds
01:44 (join) Skola
01:53 (quit) yoklov: Quit: computer sleeping
02:23 (join) mithos28
02:34 (join) fftb
02:43 (quit) fftb: Ping timeout: 250 seconds
02:44 (join) fftb
02:57 (quit) nilyaK: Quit: Leaving.
03:04 (quit) fftb: Ping timeout: 250 seconds
03:10 (quit) mithos28: Quit: mithos28
04:12 (join) vkz
04:45 (quit) dspt: Ping timeout: 260 seconds
04:45 (join) dspt
04:56 (quit) Skola: Quit: Lost terminal
05:02 (join) mye
05:05 (part) lewis1711
05:31 (join) neilv
05:31 (quit) neilv: Changing host
05:31 (join) neilv
05:46 (join) MayDaniel
06:10 (quit) MayDaniel: Read error: Connection reset by peer
06:27 (join) fftb
06:31 (join) malkomalko
06:32 (join) rn_
06:34 (join) rostayob
06:36 rn_: hello, community! I have a question. If we look at this http://groups.google.com/group/racket-users/browse_thread/thread/1fe13354bb641ae3/3389f195d637878e?lnk=gst&q=euler4e#3389f195d637878e we will se that it's recommanded to use (nested) define instead of let in this example. So, my question is: when to use nested define, and when to use let?
06:36 rudybot: http://tinyurl.com/7aggvek
06:38 neilv: rn_: it is a matter of personal preference. seems like more of the core racket developers are using internal define. seems like other people coming from scheme tend to use let
06:39 rn_: but felleisen said that define is faster?
06:42 (join) ambrosebs
06:42 (quit) ambrosebs: Client Quit
06:43 (join) realitygrill
06:43 neilv: he did?
06:43 rn_: yes, look at the last post
06:43 rn_: http://groups.google.com/group/racket-users/browse_thread/thread/1fe13354bb641ae3/8ac2c9d406cc10c5?lnk=gst&q=spirit+of+racket#8ac2c9d406cc10c5
06:43 rudybot: http://tinyurl.com/7os4dw6
06:44 neilv: looks to me like his change to from let to define was gratuitous, and that his performance improvement was to make palindromic use strings rather than lists
06:46 rn_: oh, so! i didn't notice the difference between the fuction palindromic in the first and last post
06:46 rn_: thanx
06:47 neilv: this is a new way in which internal define confuses people rather than helps them :)
06:47 rn_: you don't like internal define?
06:48 neilv: i don't have a strong argument against it. and it can make example code look more accessible to people who do not know racket at all yet
06:49 rn_: when is then recommandedto use the 'local' construct? it's unclear to me.
06:49 neilv: i don't use "local" either
06:52 neilv: you could try searching the email list for when to use internal define vs. local. i'm pretty sure it's been discussed recently
06:54 rn_: i've just found this: http://groups.google.com/group/racket-users/browse_thread/thread/3e934b513e018522/0505084b2327d39c?lnk=gst&q=local+define#0505084b2327d39c
06:54 rudybot: http://tinyurl.com/7ph7duf
07:01 neilv: http://www.ccs.neu.edu/home/matthias/Style/style/Choosing_the_Right_Construct.html#%28part._.Definitions%29
07:01 rudybot: http://tinyurl.com/7e4nqv9
07:02 neilv: that's all that i see matthias' style gude for core racket developers says about internal define
07:02 bremner: hmm. Not a super compelling reason ;)
07:03 rn_: i agree
07:08 (quit) djcb: Remote host closed the connection
07:08 (quit) vkz: Ping timeout: 276 seconds
07:09 neilv: i think there's been more arguments for it, but you'd have to search the email list
07:12 neilv: i don't like searching the email list because there are several different archives of it, so if i just go to google, i end up seeing the same threads over and over, for copies on different sites of varying ugliness
07:12 (join) vkz
07:13 bremner: is it on gmane?
07:13 neilv: it's everywhere
07:13 (join) RacketCommitBot
07:13 RacketCommitBot: [racket] plt pushed 4 new commits to master: http://git.io/HwcxdQ
07:13 RacketCommitBot: [racket/master] macro-debugger: internal debugging improvements - Ryan Culpepper
07:13 RacketCommitBot: [racket/master] macro-debugger: fix parser for local-actions in mod pass 2 - Ryan Culpepper
07:13 RacketCommitBot: [racket/master] db: fix mysql close-on-exec slowdown (TCP issue) - Ryan Culpepper
07:13 (part) RacketCommitBot
07:16 (quit) malkomalko: Remote host closed the connection
07:48 (join) malkomalko
07:51 (join) mceier
07:56 (quit) rostayob: Quit: WeeChat 0.3.5
07:57 (join) rostayob
08:01 (quit) rostayob: Client Quit
08:02 (join) rostayob
08:10 (quit) rn_: Ping timeout: 245 seconds
08:18 (join) samth
08:18 (quit) realitygrill: Quit: realitygrill
08:38 (join) masm
08:41 (join) jeapostrophe
08:44 (quit) masm: Ping timeout: 250 seconds
08:48 (quit) dous_: Remote host closed the connection
08:49 (join) masm
08:54 (quit) masm: Ping timeout: 260 seconds
08:57 (join) masm
08:58 (quit) jeapostrophe: Ping timeout: 240 seconds
09:03 (quit) masm: Ping timeout: 276 seconds
09:08 rostayob: not strictly racket but... is there any way to use paredit in a geiser prompt?
09:11 (join) Diarmid1
09:17 (join) dous_
09:17 (join) tgilray`
09:19 (quit) Diarmid: *.net *.split
09:19 (quit) tgilray: *.net *.split
09:22 (quit) dous_: Ping timeout: 260 seconds
09:23 (join) masm
09:24 (join) dous_
09:27 (quit) masm: Ping timeout: 252 seconds
09:35 (join) masm
09:39 (quit) masm: Ping timeout: 264 seconds
09:39 rostayob: also, why can't I match with '.' in racket?
09:40 (join) masm
09:41 rostayob: for example,  if I need to match (a b c . rest), what's the best way?
09:42 bremner: (a b c d ... ) maybe
09:42 rostayob: bremner: well but how do I refer to the "..."?
09:42 rostayob: also Ii think you meant (list a b c d ...)
09:42 bremner: everything gets shoved in d irrc
09:42 (quit) malkomalko: Remote host closed the connection
09:43 bremner: yeah, probably I did mean (list a b c d ...)
09:43 rostayob: oh, ok, I'll try
09:44 rostayob: still, wouldn't the dot notation be much more concise? I guess the way racke tdoes it is less ambiguous...
09:44 bremner: not sure. There is of course cons patterns if you just want the first.
09:44 bremner: I find cons is almost always better than list for my uses
09:58 (quit) mye: Ping timeout: 246 seconds
09:59 (quit) masm: Ping timeout: 246 seconds
10:00 (join) masm
10:04 rostayob: I want to create a macro that simply transforms an S-expression instead of a syntax-object, that would be an non-hygienic macro right?
10:05 rostayob: in my case, this is the transformation I want: http://vpaste.net/0DN9b
10:06 rostayob: so if I have an expression (parser . rest)
10:06 rostayob: I want 'rest' to be transformed according to that function
10:06 rostayob: how do I do that in racket?
10:06 (quit) masm: Ping timeout: 250 seconds
10:09 (join) masm
10:10 rostayob: and I'm pretty sure that I can't do that with an hygienic macro, since the point of that macro is to keep the provided identifiers
10:11 neilv: you can do anything in syntax-case or syntax-parse
10:12 rostayob: ok, so how would you do what I posted?
10:13 neilv: i'll have to look in a minute. running debugger
10:14 rostayob: ok, thanks :)
10:14 (quit) masm: Ping timeout: 250 seconds
10:18 (join) masm
10:31 (quit) masm: Ping timeout: 272 seconds
10:33 (join) realitygrill
10:35 (join) fftb_
10:35 (quit) fftb: Read error: Connection reset by peer
10:44 chandler: rostayob: It doesn't even look like you need more than syntax-rules here.
10:46 (quit) AlbireoX: Read error: Connection reset by peer
10:46 rostayob: chandler: so the names of the identifiers will be preserved through the transformation?
10:47 chandler: Uh, I'm not quite sure how to answer that.
10:48 chandler: I suspect you're thinking of syntax-rules et al as renaming identifiers under the hood.
10:49 (join) asdfhjkl
10:49 chandler: That's not actually how it's implemented, though. It's better to think of it as not discarding scope information.
10:49 (join) masm
10:50 chandler: In this case, there's no reason to do that, nor is there anything I see which is introducing new identifiers.
10:50 rostayob: yes, that was my question.
10:50 rostayob: ok let me try... (i'm totally new to lisp macros)
10:53 rostayob: chandler: mhm... this http://vpaste.net/9jp4I always seems to match the last case, e.g. (parser a <- foo) doesn't throw an error
10:54 chandler: The way you match literals in `syntax-rules' is a bit different.
10:55 chandler: Instead of using '<- inside the rules, just use <- . You'll also need to put <- in the list of literals to the syntax-rules form, which you currently have as just ().
10:55 chandler: Here's a concrete example:
10:56 rostayob: chandler: oh well that works, thanks
10:56 chandler: rudybot: eval (define-syntax foo (syntax-rules (<-) ((foo <-) "got a <-") ((foo bar) "didn't get a <-")))
10:56 rudybot: chandler: Done.
10:56 chandler: rudybot: eval (foo <-)
10:56 rudybot: chandler: ; Value: "got a <-"
10:56 chandler: rudybot: eval (foo test)
10:56 rudybot: chandler: ; Value: "didn't get a <-"
10:57 chandler: rostayob: Also, the way you're using `error' there causes the error to be raised at run time, not at expansion time.
10:57 rostayob: chandler: ok, I should raise a syntax-error rigth?
10:58 chandler: This is something that's not particularly obvious with syntax-rules. I'm looking to see if there's a good way to do it in Racket.
11:00 chandler: Hm; I'm not seeing it. A lot of people do something like this:
11:00 rostayob: chandler: you mean the error?
11:01 chandler: How to raise an error at compile time in a syntax-rules macro.
11:01 rostayob: yeah
11:01 chandler: rudybot: eval (define-syntax syntax-error (syntax-rules () ((_) (syntax-error "syntax error"))))
11:01 rudybot: chandler: Done.
11:01 chandler: rudybot: eval (syntax-error "parser must end with non-binding form")
11:02 rudybot: chandler: error: #<undefined>:1:0: syntax-error: bad syntax in: (syntax-error "parser must end with non-binding form")
11:02 rostayob: yep that works
11:02 rostayob: thanks a lot
11:04 (quit) vkz: Quit: vkz
11:04 (join) yoklov
11:04 (quit) masm: Ping timeout: 255 seconds
11:06 chandler: Some kind of "magic" syntax-rule-error form would be nice to have, I think.
11:06 chandler: Magic in the sense that it would provide the right expression to `raise-syntax-error'.
11:19 (join) masm
11:27 (join) GeneralMaximus
11:36 (quit) cdidd: Ping timeout: 260 seconds
11:38 rostayob: ok, wrote my first racket program! a little parser combinator in the style of parsec
11:38 rostayob: https://gist.github.com/2075565
11:39 rostayob: if anyone could point out weirdness I'd be very grateful!
11:41 neilv: whew. http://lists.racket-lang.org/users/archive/2012-March/051066.html
11:41 neilv: rostayob: did you get your question answered, or should i look through the irc logs?
11:41 rostayob: neilv: yes, I did
11:42 rostayob: I had misunderstood how syntax objects work
11:47 rostayob: the only thing that I don't like is that pattern-matching is kind of verbose
11:49 (join) dnolen
11:49 (quit) masm: Ping timeout: 252 seconds
11:50 (join) masm
11:52 neilv: rostayob: congrats on your program.  you might like to try syntax-case next
11:53 rostayob: neilv: I'll read up on the difference :)
11:53 neilv: syntax-case lets you manipulate the syntax objects with arbitrary racket code
11:53 neilv: syntax-parse then adds a few things in addition to what syntax-case does
11:54 rostayob: ok
11:55 rostayob: now, markdown parser!
11:56 rostayob: btw, what ports does file-position work with?
11:56 rostayob: ah
11:56 rostayob: In position-setting mode, file-position raises the exn:fail:contract exception for port kinds other than file-stream and string ports.
11:56 rostayob: which is more or less what I need
11:58 neilv: i'm not sure why you'd need to do random access i/o. you have recursion. (if you needed really fancy stuff for speculative search-like parsing, you could use first-class continuations, but i doubt you need that)
12:03 (join) jldavis76
12:06 (quit) realitygrill: Quit: realitygrill
12:09 (join) realitygrill
12:10 (quit) masm: Ping timeout: 246 seconds
12:12 (join) masm
12:14 (join) Shvillr_
12:14 (join) vkz
12:15 (quit) Shviller: Disconnected by services
12:15 (nick) Shvillr_ -> Shviller
12:16 (quit) vkz: Client Quit
12:17 (quit) realitygrill: Quit: realitygrill
12:17 (quit) masm: Ping timeout: 276 seconds
12:19 (join) masm
12:19 (quit) fftb_: Remote host closed the connection
12:21 rostayob: neilv: ? you mean I should just carry around a string instead of manipulating a port?
12:25 (quit) masm: Ping timeout: 244 seconds
12:25 neilv: i'm talking about writing a recursive descent parser, recursively
12:25 rostayob: well, I will write it recursively using those combinators
12:26 rostayob: actually now that I'm thinking about it it might be a bit annoying without lazyness...
12:26 neilv: so the frames (or whatever) for the recursion can hold onto characters that you need for backtracking
12:27 (join) fftb
12:29 rostayob: mhm... what I'd do in Haskell (and what parsec does) is to consume a string and implement 'try' saving the string, trying to parse and putting the initial string back if we fail
12:29 rostayob: I'm not sure if this is what you suggest
12:29 (join) masm
12:29 rostayob: I wanted to try to do it with ports because I want interoperability with the html parser, that works with ports
12:31 neilv: ok, i see. so you might want to look at input and later decide to put it back for a different parser you don't control to look at
12:32 rostayob: yes
12:32 neilv: if you only need a little lookahead, you can use peek-char instead of read-char
12:32 rostayob: yeah I know but I don't want to worry about those things, I just want a combinator that tries to parse and restore the previous situation if it fails.
12:32 neilv: or you could make a port abstraction that lets you unread or push characters onto it
12:33 neilv: you can define your own ports in racket
12:33 rostayob: yes that would work, but why bother?
12:33 rostayob: I mean what's the disadvantage of how it works currently, apart from the fact that it works only with strings and files?
12:34 (quit) masm: Ping timeout: 260 seconds
12:34 neilv: probably no practical difference in this case
12:34 neilv: the difference is whether you can stream
12:35 rostayob: "stream"?
12:35 rostayob: as in using custom ports?
12:35 neilv: i just mean that, with ports, you can read some input before all the input is available, and you don't necessarily have to keep all the input in memory at once
12:36 rostayob: well but that would work anyways, assuming that the port implementation for files reads incrementally
12:36 (part) jldavis76
12:36 rostayob: anyway it's not a big deal for me, all I need to parse is strings and files that fit in memory :P
12:38 (join) masm
12:42 (quit) Diarmid1: Quit: Leaving.
12:42 (join) Diarmid
12:44 (join) jonrafkind
12:45 (quit) masm: Ping timeout: 252 seconds
12:47 (join) masm
12:53 (quit) fftb: Remote host closed the connection
13:01 (quit) masm: Ping timeout: 252 seconds
13:04 (join) masm
13:04 (join) MayDaniel
13:10 (quit) masm: Ping timeout: 265 seconds
13:11 (quit) neilv: Quit: Leaving
13:13 (join) masm
13:15 (quit) dnolen: Ping timeout: 245 seconds
13:17 (join) realitygrill
13:20 (quit) masm: Ping timeout: 240 seconds
13:22 rostayob: mh. If I import a module with prefix-in and then I use a macro from that module, if that macro then expands using functions from that module it seems that it can't find it, possibly because the macro doesn't qualify it with the prefix
13:22 rostayob: is there a way around this?
13:23 (join) mithos28
13:24 (join) masm
13:24 (join) djcb
13:28 rostayob: no, my bad
13:29 (quit) masm: Ping timeout: 264 seconds
13:32 (join) masm
13:32 (quit) MayDaniel: Read error: Connection reset by peer
13:36 (quit) masm: Ping timeout: 244 seconds
13:42 (join) masm
13:47 (join) dnolen
13:48 (quit) yoklov: Quit: computer sleeping
13:49 (quit) masm: Ping timeout: 264 seconds
13:49 (quit) Fare: Ping timeout: 246 seconds
13:52 (quit) dous_: Remote host closed the connection
13:53 (join) Skola
13:53 (join) masm
13:59 (quit) realitygrill: Quit: realitygrill
14:00 (join) bdirks
14:01 (join) jrslepak
14:02 (join) dous_
14:02 (quit) dous_: Changing host
14:02 (join) dous_
14:05 (quit) masm: Ping timeout: 246 seconds
14:10 (quit) bdirks: Quit: Colloquy for iPad - http://colloquy.mobi
14:10 (join) masm
14:14 (quit) masm: Ping timeout: 250 seconds
14:16 (join) masm
14:25 (quit) dous_: Remote host closed the connection
14:40 offby1: So ... my racket program is using a surprisingly large amount of memory.  How does one go about investigating?
14:50 (join) yoklov
15:07 (join) realitygrill
15:12 (quit) realitygrill: Client Quit
15:23 (join) jeapostrophe
15:37 rostayob: is it normal that something that compiles with #lang racket doesn't with #lang lazy?
15:37 rostayob: with "compiles" I mean that I can load it
15:37 (quit) jonrafkind: Ping timeout: 246 seconds
15:39 offby1: I don't use lazy, but it's not surprising.
15:39 offby1: they are, after all, different languages
15:41 (quit) GeneralMaximus: Quit: Leaving
15:42 (join) realitygrill
15:46 rostayob: uhm, ok
15:47 rostayob: I'm really missing lazyness, I'm defining a parser using some parser combinators and for recursive definitions it's a big win
16:23 Diarmid: rostayob: Why don't you use the lazyness normally available (not in #lang lazy)?
16:23 rostayob: Diarmid: you mean create thunks manually?
16:23 rostayob: or is there some lazy keyword?
16:23 chandler: rostayob: `delay' and `force' implement explict laziness.
16:24 rostayob: actually I'm thinking of converting the "choice" operator in a macro, which would solve problems.
16:24 rostayob: ah. cool.
16:25 Diarmid: rostayob: http://docs.racket-lang.org/reference/Delayed_Evaluation.html
16:26 rostayob: Diarmid: that's great. thanks.
16:26 rostayob: I still think I'll create a macro
16:26 rostayob: to explore macros more ehe
16:26 rostayob: Diarmid: btw, I'm quite satisfied with my parsec clone apart from this problem :)
16:27 rostayob: I like this language!
16:29 (join) nilyaK
16:31 Diarmid: rostayob: I like parseq and fmt.
16:32 (join) jonrafkind
16:32 rostayob: I wonder h ow they solve that problem
16:38 rostayob: also, is there a top-level letrec?
16:39 chandler: Hm. What are you trying to do?
16:39 chandler: Multiple `define' forms at the top level are roughly equivalent to `letrec'.
16:40 chandler: rudybot: eval (begin (define f (lambda (x) (if (zero? x) #t (g (sub1 x))))) (define g (lambda (x) (if (zero? x) #f (f (sub1 x))))))
16:40 rudybot: chandler: Done.
16:40 chandler: rudybot: eval (f 4)
16:40 rudybot: chandler: ; Value: #t
16:40 rostayob: no, what I meant is something of the form (define foo ...some expression using foo ...)
16:40 chandler: That's fine too.
16:41 rostayob: oh right
16:41 rostayob facepalms
16:41 chandler: As long as you don't need to take the value of `foo' in the process of evaluating that expression.
16:43 rostayob: chandler: I asked this because this: http://vpaste.net/TBNLE doesn't work, while defining it as a 0-ary function - (pexpr) - works
16:43 rostayob: but the problem is probably elsewhere
16:45 chandler: Hm. I wouldn't expect to see multiple forms in the expression part of a `define' form like that. I know there was some talk of allowing that recently on the Racket list, but I don't see any reference to that being allowed in the documentation.
16:45 chandler: rudybot: eval (define x (define y 1) y)
16:45 rudybot: chandler: error: #<undefined>:1:0: define: bad syntax (multiple expressions after identifier) in: (define x (define y 1) y)
16:46 chandler: rudybot: eval (define x (begin (define y 1) y))
16:46 rudybot: chandler: error: #<undefined>:1:17: define: not allowed in an expression context in: (define y 1)
16:46 rostayob: yeah, that is the problem... but then why does it work if it's a function?
16:46 chandler: rudybot: eval (define x (local (define y 1) y))
16:46 rudybot: chandler: error: #<undefined>:1:18: define: bad syntax in: define
16:46 rostayob: rudybot: (define (x) (write 'foo) (write 'bar))
16:46 rudybot: rostayob: your sandbox is ready
16:46 rudybot: rostayob: Done.
16:46 rostayob: rudybot: eval (define (x) (write 'foo) (write 'bar))
16:46 rudybot: rostayob: Done.
16:47 chandler: rostayob: Because the body of a procedure is like the body of a `let' or other such binding form, and allows internal definitions.
16:47 chandler: rudybot: eval (define x (let () (define y 1) y))
16:47 rudybot: chandler: Done.
16:47 chandler: rudybot: eval x
16:47 rudybot: chandler: ; Value: 1
16:47 rostayob: uh?
16:47 rostayob: rudybot: eval (x)
16:47 rostayob: chandler: well anyways. if it's a function it works, which confused me
16:47 rudybot: rostayob: ; stdout: "foobar"
16:47 chandler: Right, hence the recent discussion on the list about allowing this, but I don't remember what became of it.
16:48 rostayob: why would it be different for procedures?
16:49 chandler: Because (define (f a ...) e ...) is just syntactic sugar for (define f (lambda (a ...) e ...)) .
16:49 chandler: And in the body of a `lambda', you can use local definitions.
16:50 chandler: Personally, I've always preferred to just write (define f (lambda ...)), which I've been informed is an IU tic. (Guilty as charged.)
16:50 rostayob: IU?
16:50 chandler: Indiana University
16:50 rostayob: ah :)
16:52 chandler: I'm a little confused as to why (local (define y 1) y) doesn't work, actually.
16:53 chandler: Er, in a top-level context that is.
17:06 (quit) realitygrill: Quit: realitygrill
17:12 (join) stis
17:13 stis: Hey! Why does racket have one local-syntax lookup function that lookup syntax-parameters and one that don't?
17:13 jeapostrophe: chandler: you are missing a paren
17:14 jeapostrophe: rudybot: (local [(define y 1)] y)
17:14 rudybot: jeapostrophe: your sandbox is ready
17:14 rudybot: jeapostrophe: ; Value: 1
17:14 chandler: Oh!
17:14 chandler: Sorry. I must not be able to read today.
17:15 jeapostrophe: stis: syntax-local-value looks up the phase-1 binding associated with a piece of syntax, syntax-parameter-value is different because there is no phase-1 binding associated with the stx parameter , or rather there is, but it's not the value the parameter is currently bound to
17:16 stis: thanks!
17:18 jeapostrophe: stis: syntax-parameter-value is much less useful than syntax-local-value
17:18 jeapostrophe: stis: and, syntax-local-value is pretty advanced macrology, so you probably don't need it for what you're doing... i'm guessing :)
17:19 jeapostrophe: stis: and finally, if you DID need it, it would be better to use 'static' from 'syntax/parse'
17:19 jeapostrophe: than calling it directly
17:19 chandler: Oh, it's `block' I was trying to think of, not `local'.
17:20 chandler: Though apparently `block' can't be used in an expression context...
17:21 jeapostrophe: chandler: ya, that's different/nicer, but actually unnecessary because 'begin' already does that now
17:22 chandler: Is there a form equivalent to (let () body ...) to let internal definitions be used in an expression context? It's obviously semantically unnecessary, but it's nice to not have an empty binding form where it's not needed.
17:26 stis: jeapostrophe: I've ported syntax parse over to guile and needed something like
17:27 jeapostrophe: chandler: begin does that
17:27 chandler: rudybot: eval (define x (begin (define y 1) y))
17:27 rudybot: chandler: your sandbox is ready
17:27 rudybot: chandler: error: #<undefined>:1:17: define: not allowed in an expression context in: (define y 1)
17:27 stis: syntax-local-value, We have the option there to let syntax local value return the binding of a syntax-parametrized syntax parameter as well, can that be a bad strategy?
17:28 chandler: What am I missing?
17:29 jeapostrophe: rudybot: (version)
17:29 rudybot: jeapostrophe: ; Value: "5.2.1"
17:31 jeapostrophe: chandler: hmm, i guess the splicing behavior of begin causes a problem. block works there though
17:31 jeapostrophe: stis: that's not what syntax-local-value does... it's not really about syntax-parameters
17:31 chandler: rudybot: eval (define x (block (define y 1) y))
17:31 rudybot: chandler: error: #<undefined>:1:17: define: not allowed in an expression context in: (define y 1)
17:33 jeapostrophe: rudybot: (require racket/block) (define x (block (define y 1) y))
17:33 rudybot: jeapostrophe: hmm. i guess it depends on the kind of people that you're trying to ensure understand your code. i might be wrong, but i'd think any seasoned lisp programmer would automatically translate car, cdr to what those functions represent (i.e., first and rest). i agree it's a stumbling block for new programmers (since they have to understand new notation) and that the fewer such requirements are made for people to mem
17:33 (join) realitygrill
17:33 chandler: Heh.
17:33 chandler: offby1: That's really confusing.
17:33 stis: ok! I need to read melt things for know, thanks again!
17:33 jeapostrophe: rudybot: (define x (block (define y 1) y))
17:33 rudybot: jeapostrophe: error: #<undefined>:1:17: define: not allowed in an expression context in: (define y 1)
17:34 chandler: jeapostrophe: Nevermind rudyboy; I see now.
17:34 jeapostrophe: doesn't work because it is parsed as (#%app block (define y 1) y)
17:34 chandler: "rudybot"
17:34 chandler: Yes, I figured that out.
17:34 jeapostrophe: because block isn't in racket or racket/base, but racket/block
17:35 offby1: chandler: it only evals if it can "read" what you say into exactly one expression.
17:36 chandler: offby1: Not to be an ass about this, but what exactly is the purpose of the old-message-quoting thing? I hardly ever see it triggered deliberately, and I don't think it's ever been amusing.
17:38 offby1: chandler: I think it's very funny.
17:38 offby1: And that is its purpose.
17:38 offby1: in fact I'm working hard at "improving" it now.
17:38 jamessan: offby1: does it build the response with markov chains?
17:38 chandler: OK. Personally I think it'd be most useful if anything that didn't look like a recognized command was treated as a sequence of expressions.
17:39 offby1: chandler: perhaps I could have it do the funny-response only if it couldn't parse a sequence of expressions.
17:40 offby1: jamessan: no.  It's pretty simple: it breaks what you say up into words, then looks through its "corpus" of previous utterances, finds the word that you said that appears the fewest times (and hence is most "interesting"), then chooses a random utterance that contains that word.
17:40 offby1: that's all
17:40 offby1: right now I'm trying to make the corpus live in a sqlite db instead of just hanging out in RAM
17:47 (quit) acarrico: Ping timeout: 260 seconds
17:49 (part) stis
17:50 (join) acarrico
17:55 (join) tiagosr
17:58 (quit) jonrafkind: Ping timeout: 248 seconds
18:07 (quit) nilyaK: Quit: Leaving.
18:17 (quit) Skola: Quit: Lost terminal
18:18 (quit) djcb: Remote host closed the connection
18:19 (join) djcb
18:51 (join) rostayob1
18:52 (quit) yoklov: Quit: computer sleeping
18:54 (quit) rostayob: Ping timeout: 276 seconds
18:56 (join) yoklov
19:09 (join) vkz
19:20 rostayob1: uhm. is there anything wrong with this macro: http://vpaste.net/f0UHH ?
19:20 offby1: yeah, it's got weird colors :)
19:21 rostayob1: :) apart from that?
19:21 offby1: seriously, I dunno; my syntax-fu is but weak
19:23 rostayob1: because I'm unable to use it. however I call it it says "bind: illegal use of syntax "
19:23 (quit) jeapostrophe: Read error: Operation timed out
19:24 chandler: I'm not sure how you're intending this macro to be used, but one thing to note is that `syntax-case' patterns look just like `syntax-rules' patterns.
19:24 chandler: Also, that `define-syntax' form doesn't really look right. Low-level macros receive one argument, which is a "syntax object".
19:24 chandler: Here's an example of a `syntax-rules' macro and its `syntax-case' equivalent:
19:25 chandler: rudybot: eval (define-syntax foo (syntax-rules () ((_ a) 1) ((_ a b) 2)))
19:25 rudybot: chandler: Done.
19:25 rostayob1: chandler: I've already redefined the old macro with syntax-case, and it works
19:25 chandler: rudybot: eval (define-syntax foo^ (lambda (stx) (syntax-case stx () ((_ a) #'1) ((_ a b) #'2))))
19:25 rudybot: chandler: Done.
19:26 rostayob1: chandler: ok, so it always gets one argument that then you pattern match
19:26 chandler: Yup.
19:26 rostayob1: the thing is that
19:26 rostayob1: I want to execute something
19:26 rostayob1: and then spit out some syntax depending on the result of the computation
19:27 rostayob1: can I do that inside define-syntax?
19:27 chandler: Right, well, that's what this is here for. :-)
19:27 chandler: But there are two big things you'll need to be aware of. One is the difference between syntax and an ordinary datum. The other is phases.
19:28 rostayob1: chandler: so syntax = syntax object, datum = scheme list, right?
19:29 chandler: Yes - though a datum can be other things, too.
19:29 rostayob1: like?
19:29 chandler: Like a symbol, or a literal.
19:29 rostayob1: oh, right.
19:30 rostayob1: just a sec...
19:30 (quit) rostayob1: Quit: WeeChat 0.3.5
19:30 (join) rostayob
19:30 (join) jonrafkind
19:31 rostayob: chandler: ok. now, back to my question... how can I execute some computation inside define-syntax while mantaining all the info in the syntax object? it looks like I shouldn't be able to
19:31 chandler: You definitely can do that. What are you trying to compute?
19:32 rostayob: well this is the original definition of >>=: http://vpaste.net/DOHwz
19:32 rostayob: I'd like to define a macro that does the same. this is to avoid the execution of "f" if >>= is a normal function
19:33 (join) dzhus
19:33 rostayob: so the macro would generate a function (lambda (port) ...), where (p port) gets executed, and if it fails it just stops. if it succeeds it executes 'f result'.
19:34 rostayob: so if I have something like (>>= a (b c d)) if a fails (b c d) won't be executed
19:34 rostayob: (if I understand correctly how macros work).
19:34 chandler: OK. I don't think you actually need to compute anything at compile-time here; I think you just need to create a small macro that expands into the right definition.
19:35 chandler: rudybot: eval (begin (define (my-big-hairy-expensive-function) (printf "I'm running!\n") #t) (define (my-test-function) #f))
19:35 rudybot: chandler: Done.
19:36 chandler: rudybot: eval (define (my-proc test-fn val) (if (test-fn) val 42))
19:36 rudybot: chandler: Done.
19:36 chandler: rudybot: eval (my-proc my-test-function (my-big-hairy-expensive-function))
19:36 rudybot: chandler: ; Value: 42
19:36 rudybot: chandler: ; stdout: "I'm running!\n"
19:36 chandler: So, that's the problem, right?
19:37 rostayob: chandler: yes.
19:37 chandler: rudybot: eval (define (my-proc^ test-fn compute-val) (if (test-fn) (compute-val) 42))
19:37 rudybot: chandler: Done.
19:38 chandler: rudybot: eval (define-syntax my-macro (syntax-rules () ((_ test-fn val) (my-proc^ test-fn (lambda () val)))))
19:38 rudybot: chandler: Done.
19:38 rostayob: oh right, I can create a thunk manually, I suppose.
19:38 chandler: rudybot: eval (my-macro my-test-function (my-big-hairy-expensive-function))
19:38 rudybot: chandler: ; Value: 42
19:38 rostayob: and then unthunk it
19:38 rostayob: right. thanks.
19:39 chandler: You can also use `lazy' and `force' there if it's useful to be able to pass around the promise internally.
19:40 rostayob: yes, Diarmid suggested that, but I wanted to explore macros more :)
19:41 chandler: Sure. You might still want to use macros to hide the laziness even if you do end up using promises.
19:43 offby1: my-big-fat-greek-function
19:48 rostayob: chandler: whooo, it works. thanks a lot.
20:03 (join) dous_
20:11 offby1: it's cold here in Seattle, but I'm keeping warm because I'm running "make install" on my laptop! :-P
20:11 offby1: even better than a cat.
20:18 (join) jeapostrophe
20:23 (quit) mceier: Quit: leaving
20:30 rostayob: chandler: the problem is that that doesn't really solve my problem, because all the other combinators that I use will still evaluate their arguments
20:30 rostayob: in general I can't rely on them... I think I'll just use `lazy'
20:32 rostayob: is there something like "delay" and "lazy" but without the need to force them explicitely?
20:32 asumu: offby1: Funny enough, it's very warm here in Boston.
20:33 asumu: Weird weather this year.
20:33 offby1 has a Kramer moment
20:33 offby1: if only there were a way to ship our weather East, and yours West ...
20:34 (quit) jeapostrophe: Ping timeout: 245 seconds
20:39 (quit) gf3: Excess Flood
20:40 (quit) yoklov: Quit: computer sleeping
20:42 (join) gf3
21:07 (quit) tiagosr: Ping timeout: 245 seconds
21:13 (join) cdidd
21:17 (quit) dzhus: Remote host closed the connection
21:31 (quit) rostayob: Ping timeout: 276 seconds
21:35 (quit) dous_: Remote host closed the connection
21:45 (quit) masm: Quit: Leaving.
21:48 (quit) asdfhjkl: Ping timeout: 252 seconds
21:52 (join) AlbireoX
22:18 (quit) samth: Ping timeout: 276 seconds
22:21 (quit) mithos28: Quit: mithos28
22:47 (join) nilyaK
23:02 (quit) offby1: Read error: Connection reset by peer
23:02 (quit) rudybot: Read error: Connection reset by peer
23:03 (join) offby1
23:03 (quit) offby1: Changing host
23:03 (join) offby1
23:58 (join) jeapostrophe