00:07 (quit) hanDerPeder: Quit: hanDerPeder
00:23 (join) jon5
00:25 (nick) jon5 -> jonrafkind
00:25 (quit) jonrafkind: Client Quit
00:25 (join) jonrafkind
00:56 (join) Naith
00:59 (quit) Naith: Client Quit
01:52 (join) Naith
02:27 (quit) jonrafkind: Ping timeout: 272 seconds
05:40 (join) masm
05:50 (join) arcfide
05:50 (part) arcfide
05:59 (join) MayDaniel
07:27 (quit) Naith: Quit: Naith
07:42 (part) MarcWeber
08:07 (quit) TheRealPygo: Ping timeout: 265 seconds
08:09 (join) pygospa
08:36 (join) hanDerPeder
08:37 (quit) hanDerPeder: Client Quit
09:08 (join) jeapostrophe_
09:14 (join) hanDerPeder
09:15 (quit) hanDerPeder: Client Quit
09:19 (quit) jeapostrophe_: Quit: jeapostrophe_
09:19 (join) jeapostrophe_
09:23 (quit) martinhex: Remote host closed the connection
09:33 (join) mceier
09:51 (quit) MayDaniel: 
10:09 (join) hanDerPeder
10:36 (join) danbrown
10:46 (quit) hanDerPeder: Quit: hanDerPeder
10:55 (join) b-man_
11:13 (join) MayDaniel
11:18 (quit) b-man_: Remote host closed the connection
11:38 (join) anRch
11:48 (join) hanDerPeder
11:52 (quit) jeapostrophe_: Quit: jeapostrophe_
12:07 (quit) mikeX: Ping timeout: 255 seconds
12:09 (join) mikeX
12:10 offby1: "We've thrown out about 200,000 lines of C++ code that built
12:10 offby1: on Xt, Win32, and Carbon. We've replaced it with about 30,000 lines of Racket code that builds on Gtk, Win32, Cocoa, Cairo, and Pango."
12:10 offby1: very cool
12:14 (join) emma_
12:15 (quit) emma: Read error: Connection reset by peer
12:24 (nick) emma_ -> emma
12:54 (quit) anRch: Quit: anRch
13:24 (join) Texx
13:25 (part) Texx
13:29 (join) cscheid
13:34 cscheid: Hey guys, I'm sure this is an idiotic question, but I'm having some trouble convincing Racket that one of my procedure calls respects the required contract, and I'm not quite sure how to solve it
13:35 cscheid: I only experience the problem when running the code with serve/servlet, so I have a hunch that some interaction with dynamic loading might be to blame here
13:36 cscheid: Specifically, there's a part? (from scribble/core) procedure which returns true when I run things from the interactive racket prompt, but returns false when I run the same code via a procedure passed to serve/servlet
13:37 cscheid: I end up getting a servlet error claiming "broke the contract (-> part? (or/c #f content?)) on part-title-content; expected <part?>, given: #<part>"
13:59 (join) jeapostrophe_
13:59 (quit) jeapostrophe_: Client Quit
14:02 offby1: doesn't sound idiotic; it sounds subtle
14:02 offby1: is your code simple enough that you can paste it somewhere, and let us try to repro it?
14:09 cscheid: It's not terribly complicated, but I can try to simplify it some more. Let me work on it for a bit.
14:15 offby1: and when I say "us", I mean "me"
14:16 cscheid: :) Thanks a lot, I appreciate the help. I think I can't make it smaller than two files, but they should be no bigger than ~50 lines each
14:17 offby1: probably OK.
14:24 cscheid: ok - http://dl.dropbox.com/u/2785146/repro.tar.gz
14:24 cscheid: untar that anywhere, then trigger it like this
14:25 cscheid: cd repro; racket
14:25 cscheid: $ racket Welcome to Racket v5.0.1. > (require "posts.rkt") > (define posts (load-posts "posts")) > (part->xexpr (first posts)) '(div   (div ((class "part-title")) (h1 "This is a new title"))   (p "There would be some blocks here")) >
14:25 cscheid: (this is when it works)
14:26 cscheid: however, if you instead say $ racket serve.rkt
14:26 cscheid: you'll get the exception and stack trace
14:28 cscheid: In case it is of any use, I get this on a user-level install of Racket v5.0.1 (that is, racket lives inside /home/cscheid/...), from source, on x86_64 Ubuntu 10.04, which I downloaded from racket-lang.org maybe a month ago
14:38 cscheid: I just posted this to the mailing list too. Thanks for prodding me into making a reproducible case :)
14:43 (join) evhan
14:45 cscheid: (I gotta run, be back in 30-45min)
14:54 (join) jeapostrophe_
15:01 (quit) jeapostrophe_: Quit: jeapostrophe_
15:09 offby1: cscheid: other than 'cannot open input file: "/tmp/repro/./conf/not-found.html"', I get no errors via "racket serve.rkt"
15:09 offby1: This is "v5.0.1.900", which I guess I built from source
15:09 offby1: Ubuntu 10.10
15:10 offby1: it opens my browser and shows me a web page that says "This is a new title / There would be some blocks here", which I assume means it's working
15:10 offby1: so: I can't repro your problem :-(
15:13 offby1: maybe you have some content in your posts that I don't, that's triggering the problem?
15:22 (join) Naith
15:39 cscheid: Hey, I'm back.
15:39 cscheid: This is exactly the content that triggers the issue.
15:41 cscheid: Hm - Jay McCarthy on the mailing list told me I should use the #:make-servlet-namespace and #:servlet-namespace keyword args
15:43 cscheid: Which makes perfect sense in hindsight, but I guess doesn't explain the discrepancy between your output and mine. You're using v5.0.1.900, and I'm using v5.0.1. Perhaps that's it
15:48 offby1: I wonder if you've found a racket bug
15:49 cscheid: unlikely
15:50 offby1: can't say I understood Jay's response, but as long as _you_ understand it, all is well :)
15:50 cscheid: at least in my experience :)
15:51 cscheid: I'm not sure I understood it either, really
15:52 cscheid: because the module is only being requested at one place in serve.rkt. That could make sense if the server somehow sent data across servlet boundaries
15:57 cscheid: However, the code _does_ work when I add the keyword arg #:servlet-namespace (list "posts.rkt")
16:05 eli: It's not that difficult.
16:05 eli: rudybot: (define a (delay (+ 1 2)))
16:05 rudybot: eli: your sandbox is ready
16:05 rudybot: eli: Done.
16:05 eli: rudybot: give offby1 a
16:05 rudybot: offby1: eli has given you a value, say "rudybot: eval (GRAB)" to get it (case sensitive)
16:05 eli: offby1: Try to force that.
16:07 eli: rudybot: give cscheid a
16:07 rudybot: cscheid: eli has given you a value, say "rudybot: eval (GRAB)"
16:07 eli: cscheid: In case you want to try it.
16:08 cscheid: rudybot: eval (GRAB)
16:08 rudybot: cscheid: your sandbox is ready
16:08 rudybot: cscheid: ; Value: #<promise:a>
16:08 cscheid: rudybot: eval (force a)
16:08 rudybot: cscheid: error: reference to undefined identifier: a
16:08 cscheid: I guess I lose at this game :)
16:09 eli: cscheid: try (force (GRAB))
16:09 cscheid: ah, right
16:09 cscheid: rudybot: eval (force (GRAB))
16:09 rudybot: cscheid: ; Value: #<promise:a>
16:09 eli: cscheid: now try (promise? (GRAB))
16:09 cscheid: rudybot: eval (promise? (GRAB))
16:09 rudybot: cscheid: ; Value: #f
16:10 cscheid: fun
16:10 cscheid: but I guess my point is that in this case you generated the data in your namespace (or whatever it is the right name for this)
16:10 eli: cscheid: `struct' in racket (formerly known as `define-struct') is a *generative* tool for creating new structs.
16:11 eli: Since we have separate sandboxes on rudybot, each sandbox has its own idea of what is a promise.
16:12 cscheid: while it seemed to me that in the case inside that repro.tar.gz, the same namespace (sandbox?) which called load-posts used part-title-content
16:15 (quit) evhan: Ping timeout: 255 seconds
16:16 eli: cscheid: That I don't know much about -- it depends on how the web server runs code -- and it certainly does create a new namespace, hence Jay's answer.
16:16 eli: (And I think that it's something that has changed in some way in 5.0.2)
16:26 cscheid: eli and offby1: alright, then; thanks for the help!
16:37 offby1: rudybot: eval (force (GRAB))
16:37 rudybot: *offby1: your scheme sandbox is ready
16:37 rudybot: *offby1: ; Value: #<promise:a>
16:37 offby1: promises, promises
16:39 eli: offby1: The point is that it doesn't force it, since for you, it's not a promise...
16:39 offby1: huh
16:39 offby1: rudybot: eval (promise? (GRAB))
16:39 rudybot: *offby1: ; Value: #f
16:39 offby1 strokes beard thoughtfully
16:40 eli: That's the point of having a module instantiated or not...  Since rudybot's sandboxes don't share anything, all code is instantiated from scratch there,
16:40 eli: so all structs are disjoint from structs in other sandboxes.
16:46 (quit) MayDaniel: 
16:49 (quit) cscheid: 
17:01 (quit) hanDerPeder: Quit: hanDerPeder
17:40 offby1: now, that makes sense.
17:40 offby1: But I didn't know promises were structs.
18:00 (join) MayDaniel
18:08 His_Shadow: rudybot: eval ("What do you do when you are asked to call a string with no arguments?")
18:08 rudybot: His_Shadow: your sandbox is ready
18:08 rudybot: His_Shadow: error: procedure application: expected procedure, given: "What do you do when you are asked to call a string with no arguments?" (no arguments)
18:12 offby1: rudybot: what do you call a procedure with no arguments?
18:12 rudybot: *offby1: the formal parameters of a procedure are bound within the body of an expression yes
18:14 (join) jonrafkind
18:22 (quit) Naith: Quit: Naith
18:24 (quit) emma: Remote host closed the connection
18:42 His_Shadow: offby1, insightfull albeit nonsensical response.
18:42 His_Shadow: I like this idea though ("string" :ref 3) instead of (string-ref "string" 3)
18:43 jonrafkind: "string.3
18:44 (join) Axsuul
18:44 (join) Naith
18:50 (join) jeapostrophe_
18:50 His_Shadow: jonrafkind, yes, that is the idea.
18:51 His_Shadow: But we all know that message pasing is faaaar more elitist than method calling
18:51 His_Shadow: And we all know that being elitist is faaaaar more elitist than not being elitist.
18:51 jonrafkind: "string".3 can be a message, just like in ruby
18:51 His_Shadow: That is true.
18:52 His_Shadow: But I don't know, unifying function cals and message sends through the same interface has its charms.
18:52 His_Shadow: Kind of like French people who talk about aeternal love.
18:53 jonrafkind: the only thing i know about french culture is what I read in the 'tropic of cancer'
18:54 (quit) jeapostrophe_: Client Quit
18:54 His_Shadow: Oh wow, you can even string substitute in ruby by string = "jonrafkind"; string['raf'] = 'turbo';
18:54 His_Shadow: Now string holds 'jonturbokind'
18:54 His_Shadow: That's a bit too much
18:54 jonrafkind: yea
18:55 jonrafkind: well you could imagine a similar immutable function in racket, (substitute "blah" 2 2 "foo") => "bfooh"
18:56 His_Shadow: Yeah, sure.
18:56 His_Shadow: But I think it's cute that you can 'index' strings like that.
18:56 His_Shadow: THough it's kind of 'adding feature above feature'
18:57 jonrafkind: its nice to have convienences
19:09 His_Shadow: jonrafkind, calvinism f0 lyyfe.
19:09 His_Shadow: Thou shalt work
19:23 (join) b-man_
19:26 (quit) MayDaniel: 
19:35 (join) emma
19:36 (quit) emma: Client Quit
19:37 (join) emma
19:41 (quit) mikeX: Ping timeout: 276 seconds
19:41 (quit) Naith: Quit: Naith
19:42 (quit) emma: Read error: Connection reset by peer
19:42 (join) mikeX
19:43 (join) emma
19:50 offby1: His_Shadow: Paul Graham's "arc" language, if I recall correctly, does something like ("string" :ref 3) -- maybe just ("string" 3)
19:51 offby1: sequences are really functions on indices, I think it is.
19:51 His_Shadow: offby1, that was my original idea yes.
19:51 His_Shadow: But I changed it to this, because I can now also ("string" :slice 2 5) et cetera
19:51 His_Shadow: ("string" :len)
19:52 His_Shadow: Clojure by the way treats vectors and strings et cetera as functions of naturals to their contents
19:53 His_Shadow: In Clojure I think (random-set bla) also returns bla if bla is a member of that set, end nil otherwise.
19:53 His_Shadow: Not sure if sets can contain nil.
19:59 offby1 keeps meaning to play more with Clojure, but never gets around to it
20:01 His_Shadow: offby1, to be honest, while clojure seemed great at the start, the more I learn about it, the less I like it.
20:01 His_Shadow: A lot of things about it are devoid of aeternal logic. Like (set 1 2 3 4) is a set of those four numbers, makes sense, right?
20:02 His_Shadow: But (set "abcd" "efg" "bla") is a set of all the characters in those strings, not a set of those strings.
20:02 His_Shadow: internal logic*
20:03 offby1: that is indeed surprising
20:03 His_Shadow: offby1, well, it follows 'the principle of least surprise'.
20:03 offby1: presumably there's _some_ way to make a set of those three strings.
20:03 His_Shadow: But I find that that principle surprises a lot.
20:03 His_Shadow: Well, I suppose that #{"string" "other-string"} wouldwork.
20:04 His_Shadow: What I do like though, [a b c d] is a vector of four symbols.
20:04 His_Shadow: But like, a vector that is evaluated evaluates to a vector which contains the sequential evaluations of the old vector.
20:05 His_Shadow: So typing [(+ 1 2) (+ 3 5)] in code, which is a vector of two lists in data, becomes [3 8]
20:11 offby1: *eeeeerppp
20:11 offby1: wait.
20:11 offby1: in [a b c d], the symbols aren't evaluated as variable names ...
20:11 offby1: but in [(+ 1 2) (+ 3 5)], the + _is_ evaluated as a variable naming a procedure?!
20:12 offby1: where's the logic in that?
20:12 His_Shadow: offby1, they are.
20:12 His_Shadow: [a b c d] is a vector of four symbols, but it is evaluated to for instance [1 2 3 4]
20:13 offby1: so if i define a as 1, b as 2, etc; _then_ [a b c d] evaluates to 1 2 3 4.
20:13 His_Shadow: Like (let [a 0 b 1 c 2 d 4] [a b c d]) evaluates to [0 1 2 3]
20:13 offby1: fair enough.
20:13 His_Shadow: But I would not be surprised by clojure if they broke that rule.
20:13 offby1: oh, I thought [a b c d] => [a b c d].
20:13 His_Shadow: To be honest, I; m not even sure if (set "abc "efg") displays the behaviour I described above.
20:13 His_Shadow: It could be that it again does that once you make a set of multiple sequences
20:20 offby1: clojure is so hard to install that I don't even want to bother in order to find out
20:21 His_Shadow: But it runs on the JVM!
20:21 His_Shadow: Easy to use.
20:21 His_Shadow: And it interfaces with Java!
20:22 His_Shadow: You just don't have TCO.
20:22 His_Shadow: offby1, if I were to call you as the last think I would do before I die, you would take my place and not take up an additional stack frame, would you?
20:48 (quit) mceier: Quit: leaving
20:52 (quit) b-man_: Remote host closed the connection
21:17 (join) Naith
21:29 (quit) lisppaste: Ping timeout: 255 seconds
21:47 (quit) masm: Quit: Leaving.
21:56 (quit) Naith: Quit: Naith
22:03 (quit) jonrafkind: Ping timeout: 260 seconds
22:22 offby1 fidgets uncomfortably
22:22 offby1: that's an awful big responsibility
22:22 offby1: you could just pass your continuation on to me; I'd promise to use it faithfully
22:41 (quit) danbrown: Remote host closed the connection
22:55 His_Shadow: offby1, we don't use a call stack model? =(
22:55 His_Shadow: The Java programmers would hate us.
22:55 His_Shadow: offby1, racket doesn't use CPS, right?
23:24 offby1: I have no idea
23:50 (quit) mikeX: Ping timeout: 240 seconds
23:52 (join) mikeX
23:57 (join) danbrown