00:09 (quit) masm: Ping timeout: 240 seconds
00:25 (join) mye
02:10 (join) offby1
02:10 em: someday i want to be a racket programmer
02:12 offby1: rudybot: (printf "Hello, world~%")
02:12 rudybot: *offby1: printf >>/etc/ldap/slapd.conf 'moduleload ppolicy\ninclude /etc/ldap/schema/ppolicy.schema\noverlay ppolicy\nppolicy_default "cn=policy,dc=lan"\n'
02:12 offby1: er
02:12 offby1: rudybot: eval (printf "Hello, world~%")
02:12 rudybot: *offby1: your scheme sandbox is ready
02:12 rudybot: *offby1: ; stdout: "Hello, world\n"
02:12 offby1: em: it's easy!
02:16 offby1: rudybot: slapd??
02:16 rudybot: *offby1: I am trying to make slapd use slapo-ppolicy
02:28 (quit) offby1: Ping timeout: 250 seconds
02:39 (quit) lucian: Remote host closed the connection
03:38 (quit) jonrafkind: Ping timeout: 245 seconds
05:06 (join) tony_
05:25 (join) Blkt
05:44 (join) dzhus
06:12 Blkt: good day everyone
06:13 tony_: good morning
06:18 rapacity: good night
06:28 Lajla: em, someday, I want to have a girlfriend.
06:31 tony_: Well Lajla, I'm pleased to report that you are master of something after all: the non sequitur.
06:33 Lajla: tony_, what?
06:33 Lajla: There is a logic behind it.
06:33 Lajla: I'm basically saying 'We all want a lot of things'
06:34 tony_: non sequiturs are not marked by a lack of logic but a lack of continuity.  No?
06:35 Lajla: Doesn't non sequitur basically mean a logically falicious argument.
06:35 Lajla: non sequitur is just latin for it 'It doesn't follow out of it'
06:36 Lajla: Deponend verb, third person indicative singular of sequor.
06:37 tony_: deponent.  You're quite right, but in the US it seems to denote a discontinuity in thought or speech.  I could be wrong, though.
06:37 (join) MayDaniel
06:38 Lajla: Well, the point about US English is that Palin speaks it, since Palin is stupid, it must be incorrect.
06:38 Lajla: Now there's some nice non seq for you.
06:39 tony_: Amen on Palin, but that is indeed a non sequitur in the etymological sense.  In any case, a comparison of askoxford.com with webster.com shows that the US has added the sense I intended.
06:39 Lajla: tony_, ahhh
06:40 Lajla: Well, I suppose you can't blame a man for being taught wrong.
06:40 Lajla: Just remember, that if US English and English English contradict, the latter is always correct.
06:41 tony_: I quite agree with that most of the time, though I have mixed feelings about the differences in spelling.
06:45 tony_: What worries me more, frankly, is the decline of grammar.
06:46 tony_: I've got Empire Strikes Back playing and I just noticed the awkwardness in Darth Vader's "We would be honored if you would join us."
06:46 tony_: (Should probably be "We would be honored if you joined us.")
06:54 (join) masm
06:55 bremner: huh? Isn't one subjunctive?
06:55 bremner: (oh no, I encouraged them)
06:55 bremner: oh, I see, double would
06:59 tony_: bremner: and that was 1982ish.  Our grammar has steadily worsened since then, though, interestingly, some computer scientists carry the best hope for rehabilitating the language.  Dybvig's book, for example, is a masterpiece of clarity.
07:12 Lajla: tony_, yeah.
07:13 Lajla: Some authors condemn that practice.
07:13 Lajla: authorities
07:13 Lajla: Saying you should either say 'we would be honoured would you join us'
07:13 Lajla: Or 'we would be honoured if you joined us'
07:13 Lajla: the 'would' already does that 'if' thing, you can't combine both.
07:14 tony_: Btw, do the "How to Design Worlds" teachpacks all work fine now? (Pardon the non sequitur.)
07:24 lewis1711: I am thinking of not using OOP in racket, but instead just making structs, and functions that operate on those structs. what do you guys think?
07:25 mye: I just discovered case-lambda.
07:25 tony_: If your code is going to mutate, why not use OOP?  You get subtype polymorphism for free.
07:25 mye: They only ever tell you about the one lambda
07:25 tony_: mye: I think case-lambda is part of R6RS, no?
07:26 mye: tony_: no idea, but I used something like this in my pseudocode all the time :-)
07:26 tony_: lewis1711: ...and Racket has some pretty cool OOP features (pubment, overment, etc.)
07:26 lewis1711: my code shouldn't be mutating much though, shouldn't it?
07:26 (join) mceier
07:26 lewis1711: *should it
07:26 rapacity: what about match-lambda ?
07:27 tony_: lewis1711: if you're using structs, you're probably already basing your code on changes of state, so why not go all the way to OOP?
07:27 lewis1711: also not really a fan of rackets oop model. could roll my own I guess. but I dunno, defining data then having procedures operate on that data seems like the scheemier road to take
07:27 tony_: Of course there's an answer to that.
07:27 lewis1711: that's a weird assumption
07:27 tony_: Namely, that structs give you automatic accessors, etc.
07:28 tony_: Maybe it is weird.
07:28 lewis1711: structs are immutable by default aren't they?
07:28 tony_: I just tend not to think of structs as a natural data type for pure FP.
07:28 rapacity: yeah they are
07:29 tony_: Then it sounds like you can dispense with OOP.
07:30 lewis1711: tony_: don't get your struct idea. ML has records, so does haskell. same thing AFAIK
07:30 mye: All you need is lov...lambda! *makes a list of lambdas*
07:31 lewis1711: I do it that way in C
07:31 tony_: lewis1711: Yes, my bad, structs are just fine in pure FP.  For some reason my instinct is to think of non-recursive datatypes as being somehow less well suited to pure FP.
07:32 lewis1711: I just have structs and functions that operate on them. and I have "private functions" which i don't declare in the header.
07:32 bremner: tony_: structs can be recursive
07:33 tony_: bremner: in the sense that a field can also be a struct?
07:34 tony_: I only mean that structs aren't defined recursively, as lists are.
07:35 bremner: tony_: look at the define-type macro in plai, it does exactly that
07:35 bremner: the trick is contracts on the members of the struct, which can refer to the struct being defined
07:36 bremner: it gives much the same flavour as Haskell union types (without the inference)
07:39 tony_: It's not surprising that a struct can refer to itself (if that's what you mean).  My point is that structs are *usually* "flat" whereas lists are *always* recursive.
07:39 bremner: tony_: I guess your usually is different than mine. That happens quite often.
07:40 bremner: lists are really just pairs, which look a lot like structs from where I sit.
07:40 tony_: I meant proper lists.
07:41 bremner: right, but proper lists are just pairs with (implicit) contracts
07:41 Lajla: This is by the way some-what of a problem.
07:41 Lajla: I mean, it was normal to work like that when Lisp 1.5 was invented.
07:41 Lajla: But nowadays, experience has shown that it could be better.
07:42 Lajla: Like, say you have a contract, a cons pair, but you require the last element to have list type.
07:42 Lajla: Where a list type is of course just the union of null and those cons pairs.
07:42 Lajla: Also makes type checking if something is a list run in constant time.
07:42 bremner: are contracts a popular idea in other schemes?
07:43 Lajla: Beats me.
07:43 bremner: with pleasure.
07:43 Lajla: Rawr
07:43 Lajla: Clojure doesn't have 'pairs' at all, it has lists, if you want a pair, you just use a vector of length 2.
07:43 Lajla: Which makes some amount of sense.
07:43 tony_: Same with Haskell.
07:44 tony_: Same with OCaml.
07:44 Lajla: Well, you use a tuple of lenth 2 there.
07:44 tony_: Yes, yes, of course.
07:44 Lajla: Well, tuples are different from vectors in that tuples can't just generally be extended.
07:45 Lajla: But I think a 'list cons' which has that requirement is kind of healthy.
07:45 tony_: This has veered away from the original question, which I thought was interesting: are structs a less natural fit to FP than lists are?
07:45 bremner: well, you didn't seem convinced by my answer ;)
07:46 tony_: The fact that they're native to OCaml and Haskell suggests they're fine in FP.
07:47 tony_: In fact, I'm going to change my RPG to use structs instead of lists at the higher levels; they give me nicely named accessors for free.
07:50 mye: regarding accessors, (liz color) vs. (lizard-color liz), the first one is used in Arc, Clojure etc I think, but racket doesn't support this.
07:51 mye: (I think)
07:51 mye: I like the first one better.
07:52 bremner: because it is more OO?
07:52 bremner: (like)
07:52 mye: because its shorter
07:52 tony_: It's terser.
07:52 bremner: yeah, terseness isn't really a design goal with racket afaict
07:53 bremner: I guess it should be possible to define a function like that.
07:53 tony_: Terseness should be the overriding design goal of *any* programming language, IMO.
07:54 mye: they went from make-lizard to just (lizard ...) didnt they? (I'm discovering this stuff currently)
07:54 bremner: tony_: apl is that way ->
07:55 mye: bremner: I'd argure it's more readable, after all you liz is a lizard
07:55 tony_: I suspect imperative languages cannot be as terse as FLs.
07:56 tony_: So APL is a surprising mention.
07:56 lewis1711: racket is a bit wordy sometimes. nothing that a define or macro won't fix... but then who understands your code?
07:57 tony_: It shouldn't be too hard to write a macro that does what mye wants.
07:58 tony_: (Actually, I take that back: it wouldn't be easy for me!)
07:58 lewis1711: rudybot: (define vref vector-ref)
07:58 rudybot: lewis1711: Done.
07:59 lewis1711: might do a bunch of that in my code
07:59 mye: lewis1711: that is not quite using the actual struct binding in functional position
08:01 mye: I got no problem with long function names, just with long functions
08:02 lewis1711: that is not quite what? struct binding?
08:02 mye: and using more stuff in functional position makes a function longer, so it's two kinds of terseness.
08:03 mye: name terseness and expression terseness
08:13 (quit) MayDaniel: Read error: Connection reset by peer
08:13 tony_: Here is a possible macro implementation (incomplete and undoubtedly buggy):
08:13 tony_: (define-syntax define-struct
08:13 tony_: (syntax-rules ()
08:13 tony_: ((_ struct-name (fname ...))
08:13 tony_: (define struct-name
08:13 tony_: (let ((data (make-hashtable)))
08:13 tony_: 	 (for-each (lambda (f) (hashtable-set! data f 'dummy))
08:13 tony_: 		   '(fname ...))
08:13 tony_: 	 (lambda args
08:13 tony_: 	   (let ((cmd (car args)))
08:13 tony_: 	     (case cmd
08:13 tony_: 	       ((fname) (hashtable-ref data fname)) ...))))))))
08:14 tony_: ugh...Sorry, I should have pasted that on lisp paste.
08:14 em: I think the subjunctive mood gives Lajla double would.
08:15 Lajla: em, hmm?
08:15 em: There's my non sequitur for the day.
08:15 tony_: Lajla: just say 'harhar'
08:16 em: well it was a continuation of this -- 06:55:40 < bremner> oh, I see, double  would
08:16 em: And Darth Vader
08:16 tony_: em: yes, that much was obvious.
08:16 Lajla: em, well, double woulds are okay.
08:17 Lajla: Like 'Would you join me, I would be so glad'
08:17 Lajla: So some would say 'were you to join me, I would be so glad'
08:17 Lajla: It's more the if ... would that is fround opown.
08:17 Lajla: 'If you would join me, I will be so glad' or 'if you will join me, I would be so glad'
08:17 tony_: I'll absent myself for a few minutes while em explains his hidden meaning.
08:26 (nick) em -> emma
08:26 emma: tony_: I have no hidden messages :)
08:29 Lajla: emma, oh don't tell me that name change was because you he called you 'his'?
08:30 (part) lewis1711
08:33 emma: Lajla: Yes but that's not really a hidden message it's more of an explicit one :P
08:34 emma: However, Ive gotten so used to 'em' that this nick seems fat and bloated.
08:34 (nick) emma -> em
08:37 bremner: oy vey, nick image problems. get counselling
08:38 em: haha
08:39 Lajla: em, did you know that emmä is Finnish for 'I don't' ?
08:40 Lajla: As in 'Emmähän syö siskosi!' for 'No I'm not going to to eat your sister!'
08:42 bremner: that comes up a lot in finland, does it?
08:54 (quit) tony_: Ping timeout: 240 seconds
08:57 em: Lajla: i didn't know that!
08:58 bremner wonders if there is a way to specify in a #lang slideshow file to output to PDF (obv with a version of racket that supports PDF output)
09:06 (join) RyanRN
09:07 Lajla: bremner, poor country, not a lot of food.
09:07 Lajla: em, that is to be expected, after all, all Americans are monolingual.
09:26 (join) corruptmemory
09:29 (join) tony_
09:36 (join) MayDaniel
10:09 (quit) mceier: Quit: leaving
10:22 tony_: rudybot: (struct posn (x y))
10:22 rudybot: tony_: your sandbox is ready
10:22 rudybot: tony_: error: eval:1:0: struct: illegal use of signature form in: (struct posn (x y))
10:22 tony_: Could someone explain this?
10:23 tony_: I've lifted it straight from PLT Help.
10:29 rapacity: rudybot: eval (struct posn (x y))
10:29 rudybot: rapacity: your sandbox is ready
10:29 rudybot: rapacity: error: eval:1:0: struct: illegal use of signature form in: (struct posn (x y))
10:30 rapacity: that's weird d: it works on my repl
10:30 rapacity: rudybot: (version)
10:30 rudybot: rapacity: ; Value: "5.0.1"
10:31 tony_: Ah, it works in #lang racket but not in #lang scheme.
10:32 rapacity: rudybot: (require racket)
10:32 rudybot: rapacity: Done.
10:32 rapacity: rudybot: eval (struct posn (x y))
10:32 rudybot: rapacity: Done.
10:34 tony_: PLT folks: out of curiosity, is the breakage part of the transition to the new language, or is Racket's struct syntax different from PLT Scheme's?
10:36 tony_: rudybot: 'a
10:36 rudybot: tony_: ; Value: a
10:36 rapacity: racket still has define-struct which is plt-scheme's define-struct
10:37 (join) lucian
10:38 tony_: rapacity: sorry, I didn't realize you were on the PLT team.
10:38 rapacity: I am not
10:38 rapacity: :p
10:38 rapacity: sorry tony_
10:38 rapacity: I have a habit of answering
10:39 tony_: No problem; it was a helpful answer.
10:39 tony_: rapacity: since you've already required racket of rudybot, could you do me a favor and evaluate 'a  ?
10:39 rapacity: rudybot: 'a
10:39 rudybot: rapacity: ; Value: a
10:39 tony_: Interesting.
10:40 tony_: I thought the printed representation would be 'a in Racket, as opposed to a in Scheme.
10:40 tony_: (I prefer the latter.)
10:40 rapacity: ah it is "'a"
10:40 tony_: ?
10:41 rapacity: tried it on my repl
10:41 tony_: I know, I hate that.
10:41 tony_: I thought rudybot would return "'a" as your repl does.
10:42 tony_: But for some reason it's still just "a" in the mzscheme repl even after I (require racket).
10:42 tony_: These things are too complicated for me.
10:43 tony_: And anyway I wish someone would explain why 'a should evaluate to 'a.
10:49 (join) offby1
10:49 (quit) RyanRN: Quit: Leaving.
10:54 mye: "hello, world" evaluates to "hello, world"
10:55 mye: because it gets autoquoted
10:55 tony_: No, that's because strings evaluate to themselves.
10:55 mye: no its because they get autoquoted
10:55 tony_: 'a is really the sexpr (quote a), and lists never evaluate to themselves.
10:56 tony_: I've never even heard of "autoquoting."
10:56 tony_: It's not part of any evaluation model I know.
10:58 mye: lists never evaluate to themselves unless quote makes it so?
10:58 tony_: No, lists never evaluate to themselves, period.
10:58 mye: don't listen to me, I want to understand this just as much as you
10:59 mye: list dont evaluate at all, eval evaluates right? so I don't really understand what you mean by lists evaluating themselves.
11:00 tony_ sighs.
11:00 tony_: To evaluate a list, you first look at its car.
11:00 tony_: If its a syntactic keyword, do the syntax thing.
11:01 tony_: Otherwise, if it evaluates to a symbol, lookup the symbol in the environment.
11:01 tony_: If the symbol evaluates to a function, then evaluate the tail of the lists and apply the func to them.
11:03 tony_ feels embarrassed at spending time repeating what everyone already knows :-)
11:06 (join) brooksbp
11:06 (join) RyanRN
11:07 rapacity: hey tony_ what do you mean by "
11:07 rapacity: "I wish someone would explain why 'a should evaluate to 'a"
11:07 rapacity: rudybot: (eval 'a)
11:07 rudybot: rapacity: error: reference to undefined identifier: a
11:08 tony_: rapacity: That error is to be expected.
11:08 rapacity: yeah I know, but what did you mean by your statement that 'a evaluates to 'a
11:08 tony_: You said you saw it yourself!
11:09 rapacity: ah you're talking about that print thing ?
11:09 Lajla: tony_, you can actually configure that.
11:09 tony_: In Racket, 'blah evaluates to 'blah, and not to blah, as it does in Scheme.
11:09 Lajla: Whether it prints a symbol a as a or 'a.
11:09 tony_: One shouldn't have to configure, don't you agree?  It is a fundamental part of the language.  Why make it flexible?
11:10 Lajla: tony_, and the basic logic is that the repl tries to print out the simplest expression available that evaluates to the datum that you print, basically.
11:10 tony_: That contradicts the meaning of REPL.
11:10 Lajla: tony_, it's nonstandard for usre.
11:10 tony_: In particular, the E stands for "eval."
11:10 Lajla: And I don't like it either.
11:11 Lajla: And there have been many a debate about it and I think it's a mistake and a lot of people who come here get confused.
11:11 Lajla: It's for sure not the language R5RS uses.
11:11 rapacity: well it's read-eval and when you read 'a you get ''a
11:11 tony_: huh?
11:11 Lajla: rapacity, not really.
11:11 Lajla: When you read 'a you get (quote a)
11:11 tony_: rudybot: (read)
11:11 rudybot: tony_: ; Value: #<eof>
11:11 tony_: (grr...)
11:11 tony_: rudybot: (read)'a
11:11 rudybot: tony_: ; Value: a
11:11 rapacity: (read (open-input-string "'a"))
11:11 tony_: voila.
11:12 rapacity: rudybot: (read (open-input-string "'a"))
11:12 rudybot: rapacity: ; Value: (quote a)
11:12 tony_: that should evaluate to  a string.
11:12 tony_: oh, my bad, it evals to a port.
11:12 tony_: Wow, we have a contradiction, then.
11:12 rapacity: 'a is (quote a)
11:12 tony_: Yes, rapacity, we established that some time ago. :-)
11:13 Lajla: rudybot, (car (read (open-input-string "'a")))
11:13 rudybot: Lajla: your sandbox is ready
11:13 rudybot: Lajla: ; Value: quote
11:13 Lajla: rudybot, (cdr (read (open-input-string "'a")))
11:13 rudybot: Lajla: ; Value: (a)
11:13 tony_: These results contradict my use of rudybot, namely...
11:13 tony_: rudybot: (read)'a
11:13 rudybot: tony_: ; Value: a
11:13 Lajla: tony_, this depends on printing
11:13 tony_: NO!
11:14 tony_: The read itself seems to be behaving differently when the port is passed in explicitly.
11:14 Lajla: Yeah, basically, the print in racket prints "the symbol whose only character s #\a" as 'a insead of the more standard a.
11:14 Lajla: rudybot, (read) (+ 1 2))
11:14 rudybot: Lajla: error: eval:1:14: read: unexpected `)'
11:14 Lajla: rudybot, (read) (+ 1 2)
11:14 rudybot: Lajla: ; Value: 3
11:14 Lajla: It evaluates it.
11:14 tony_: rudybot: (read (open-input-string "'a"))
11:14 rudybot: tony_: ; Value: (quote a)
11:15 tony_: You're missing the point.
11:15 tony_: The point is that (read)'a should evaluate to the same thing as (read (open-string-port "'a")).
11:15 Lajla: No it shouldn't.
11:15 tony_: Why not?
11:16 Lajla: Because (read) x evaluates x.
11:16 Lajla: And then reads the result.
11:16 tony_: what?
11:16 Lajla: (read) :x:
11:16 tony_: read is supposed to read sexps from a port *without* evaluating them.
11:16 Lajla: Where :x: is some external representation of a program
11:16 Lajla: Executes the program
11:17 Lajla: prints the result as a strng
11:17 Lajla: and then reads that string
11:17 tony_: I disagree.
11:17 Lajla: And :x: is not a port.
11:17 Lajla: What is written to the port that is being read is the evaluation of x.
11:17 tony_: Lajla, if that were true, we would not have repls; we'd only have rpls.
11:18 Lajla: rudybot, (read) "(+ 1 2)"
11:18 rudybot: Lajla: ; Value: "(+ 1 2)"
11:18 tony_: I'm sorry, but this makes absolutely no sense to me.
11:18 tony_: now *that's* correct.
11:18 tony_: See? No evaluation.
11:18 Lajla: Strings evaluate to themselves.
11:18 tony_: argh...ok.
11:18 Lajla: There is evaluation
11:18 tony_: rudybot: (read)(+ 1 2)
11:18 rudybot: tony_: ; Value: 3
11:18 rapacity: you do realise that the (read) is not doing anything there ?
11:18 Lajla: tony_, nowhere does it say that (read) x does the same as (read (open-string-port "x"))
11:19 Lajla: Yeah, it's not read that evaluates.
11:19 tony_: rudybot: (let ((a (read))) a)(+ 1 2)
11:19 rudybot: tony_: ; Value: 3
11:19 Lajla: tony_, in that case, the string the read function read contains "3"
11:19 Lajla: Not "(+ 1 2)"
11:20 bremner: umm. isn't this a bit dependent on rudybot? in a "real repl"   "(read)'a" returns ''a
11:20 tony_: Well guess what: rudybot is behaving differently from MzScheme.
11:20 Lajla: What happens in this case is that (+ 1 2) is evaluated, and its result is written to a string, giving "3", and this is then read.
11:20 tony_: Well that's not what MzScheme does.  MzScheme does what I think is the right thing.
11:20 tony_: It simply reads.
11:20 Lajla: I'm not sure any of this is anything a standard.
11:21 Lajla: tony_, rudybot also simply reads.
11:21 Lajla: (read) (+ 1 2) to rudybot tells it to read a port which contains "3"
11:21 tony_: Lajla: you (and I) have just proved that it doesn't simply read, but also evals.
11:21 Lajla: Nope, read doesn't eval.
11:21 tony_: Ok, so you're saying that the eval happens separately and outside of the read.
11:21 Lajla: Yap.
11:22 Lajla: What is given to read is the port continaing "3"
11:22 tony_: Well that's not what MzScheme does.
11:22 Lajla: I can imagine.
11:22 Lajla: But the standard deosn't document this.
11:22 Lajla: THis is pretty implementation specific.
11:22 Lajla: Wait.
11:22 Lajla: Silly me.
11:22 Lajla: It does something completely different.
11:22 Lajla: rudybot, 1 2 3
11:22 rudybot: Lajla: ; Value: 3
11:22 tony_: So now you're saying that RnRS doesn't define the behavior of read?  (I really doubt that.)
11:22 Lajla: It wraps (begin ...) around it. :')
11:23 Lajla: tony_, it does define the behaviour of read, but this is something different.
11:23 Lajla: It doesn't at any point in the spec say that if you put a (read) function before some other piece of code it's supposed to read what's after it.
11:23 Lajla: read reads from a port.
11:23 tony_: Fine.
11:23 Lajla: So what most implementations do is they open a window for you to enter text.
11:23 Lajla: rudybot, (read) (display "hello")
11:23 rudybot: Lajla: ; stdout: "hello"
11:24 brooksbp: i tried (read) in drracket repl. it doesn't seem to evaluate anything
11:24 tony_: Anyway, back to an important point: Exactly how does one configure the system to print things the way I like them?
11:25 rapacity: write
11:25 rapacity: in choose language change output syntax to write
11:25 tony_: brooksbp: I just tried it too, but it also doesn't behave the way I wish it did.
11:26 tony_: rapacity: that sounds like a DrRacket menu?
11:26 rapacity: ah
11:26 brooksbp: tony_: what's the behavior you expect?
11:26 rapacity: yeah it is
11:26 tony_: brooksbp: sorry, but we've just thrashed the matter; please see above.
11:27 tony_: rapacity: No DrRacket and no menus for me please :-)  I want to run this from a terminal.
11:27 tony_: Lajla: my configure question was posed to you :-)
11:28 Lajla: tony_, I thought rapacity answered it already.
11:28 Lajla: In the 'choose language' part.
11:29 rapacity: (current-print (lambda (e) (write e) (displayln ""))) perhaps :p?
11:31 brooksbp: Does anyone have insight as to why type checking is in the Debugging chapter of SWePR?
11:34 tony_: Lajla and rapacity: FYI, the appropriate switch seems to be print-as-expression.
11:35 Lajla: could be, figured that out a long time ago and didn't really remembe.
11:35 Lajla: WAs really glad when I did.
11:35 rapacity: ah thanks
11:35 Lajla: Because it's just awkward how they try to explain it.
11:35 Lajla: Like in the docs, something like "a quoted value is a list starting with the symbol 'quote and ...."
11:35 Lajla: And then proceeding to ay that ' is an abbreviation
11:35 tony_: The Help Docs have a section called "The Printer" which explains much.
11:36 Lajla: Will lead to a lot of epople scratching their head...
11:36 tony_: I just don't see why they couldn't stick to the old print; they probably have good reasons at the end of the day.
11:37 rapacity: rudybot: (equal? (read (open-input-string "'a")) ''a)
11:37 rudybot: rapacity: ; Value: #t
11:37 (quit) lucian: Remote host closed the connection
11:39 mye: rudybot: (parameterize ((print-as-expression #f)) 'a)
11:39 rudybot: mye: your sandbox is ready
11:39 rudybot: mye: ; Value: a
11:39 mye: my drracket still says 'a so the language options seem to override it?!
11:40 tony_: mye: No, if you evaluate (print-as-expression #f) it will change the behavior to what I like.
11:40 tony_: Anyway, it looks like we should stop calling it "repl" and start calling it "rewl"...
11:43 Lajla: well, that's historical.
11:43 Lajla: scheme just calls what was called print write nowadays.
11:43 Lajla: You see, originally there were no strings...
11:43 Lajla: I think read/write/scan/print is a nice disection of it.
11:45 mye: but why do we want different printing behaviours in the first place?
11:46 (quit) dzhus: Remote host closed the connection
11:46 tony_: mye: because, for one thing, sometimes you want a string to be printed without the double-quote delimiters.
11:48 Lajla: mye, theoretically "write" takes a datum and prints out its s-expression where possible.
11:48 Lajla: print takes a character or a string and prints out those characters.
11:48 tony_: Lajla: I thought a datum *was* an sexp.
11:48 Lajla: No.
11:48 Lajla: sexp are a way to write data.
11:48 Lajla: "(1 2 3)" is the symbolic expression to write that list.
11:49 Lajla: THe list itself is quite abstract
11:49 tony_: I thought a sexp *was* data.
11:49 Lajla: Well, apparently nt
11:49 Lajla: there are multiple ways to write down the same datum
11:49 tony_: Could you give us an example?
11:49 Lajla: "(1 2 3)" or for instance <list><number>1</number> <number>2</number> <number>3</number></list>
11:50 tony_: You've just exited the Scheme programming language.
11:50 mye: ?end-user? style rather than ?programmer? style is the references (amusing) explanation, after all people using racket are programmers?! ;-)
11:50 tony_: When I say "datum," I mean it in the sense in which R6RS the word.
11:51 tony_: ...in which R6RS defines the word.
11:51 Lajla: tony_, I'm pretty sure R6 differentiates between data and languages used to write down data.
11:51 (join) anRch
11:51 tony_: AFAIK, R6RS does not mention HTML.
11:51 tony_: (or XML)
11:52 tony_: one moment, let me look it up.
11:53 mye: REWDPL
11:54 tony_: Lajla: R6RS defines datum strictly to refer to Scheme objects.
11:55 Lajla: tony_, so?
11:55 Lajla: That doesn't exclude me from writing them down as I see fit.
11:55 Lajla: A datum and a way to write down such a datum is a different thing.
11:55 Lajla: A sexp is a string, a sequence of characters.
11:55 tony_: A sexp is not a string.
11:55 (join) mwolfe
11:56 Lajla: Yes it is, it's a sequence of characters, it's a _language_
11:56 Lajla: Or a subset thereof, rather.
11:56 Lajla: You might want to look up M-expressions
11:56 tony_: The only thing it's a subset of is the universe of Scheme objects.  Recursively defined, a sexp is a symbol, a number, a string, or a list of sexps.
11:56 Lajla: a historical attempt for a different way to write down the same objects.
11:56 Lajla: Nope, those are data.
11:57 Lajla: THere's a difference between a datum and a way to write down data.
11:57 tony_: OK, we'll have to agree to disagree.
11:57 tony_: Are there any PLT team members here who can enlighten me?
11:58 tony_: eli, are you awake? (Probably not.)
11:58 bremner hopes they are all busy releasing 5.1
12:05 tony_ sighs.
12:05 tony_: Well, it is a Sunday after all.
12:05 offby1 orders bagels and lox
12:06 tony_: I just had an asiago.
12:06 tony_: Delish.
12:07 tony_: So does anyone know which folks here are part of PLT?
12:07 tony_: (other than eli and samth)
12:08 tony_: Nevermind, I just figured it out.
12:10 tony_: clklein, danking, jeapostrophe, samth, or tewk: can one of you be tempted to weigh in?
12:12 bremner: how to make friends and influence people on IRC ;)
12:20 rapacity: boku wa tomodachi ga sukunai
12:20 Lajla: Mä olen koira ja kolme saksaa
12:21 eli: tony_: I'm very awake -- in some ski place (a weekend with tomer and some friends).
12:22 tony_: Boker tov.
12:22 tony_: Have you caught up on our little drama?
12:22 eli: But these places get extremely crowded, so I probably won't be connected for long.
12:22 eli: Where does the drama begins?
12:22 tony_: I'll try to sum it up using the right words: why has Racket given up on the traditional semantics of print?
12:23 eli: No, but there was a change -- values are printed in a way that is closer to being evaluatable.
12:23 tony_: Actually, Eli, if you're oon vacation then this is definitely not worth your time right now.
12:23 eli: So when you enter (car '(x y z)) it prints on the repl as 'x
12:24 tony_: Yes, and I've been grumbling about that.
12:24 eli: Nah, I'm always working for some definition of working.
12:24 tony_: :-)
12:25 (join) lucian
12:25 tony_: My feeling is that only atoms should evaluate to themselves, and 'x = (quote x) => x  (...where, obviously, "=" means "gets rewritten to" and "=>" means "gets evaluated to".)
12:27 Lajla: tony_, define 'atom'?
12:27 tony_: ...and I see that part of my problem is a focus on evaluation rather than on printing.
12:27 tony_: Lajla: an atom is a number, a string, a symbol, or the empty list.  See R6RS for more.
12:27 Lajla: Depending on literature, anything both a cons is an atom to all things which are identical when eqv? can test them.
12:27 Lajla: tony_,  a string too eh?
12:27 Lajla: Also a mutable string?
12:28 tony_: Lajla: I'm sorry but I couldn't quite parse your response.
12:28 Lajla: So why should symbols evaluate to themselves?
12:28 tony_: A pair is definitely not an atom.
12:28 tony_: Symbols do not evaluate to themselves.  I misspoke earlier.  Sorry.
12:30 Lajla: Ohh, swap 'both'with 'but'
12:31 Lajla: tony_, like I said, by default in #lang racket, it prints the simplest expression that evaluates to what you want, not what you want directly.
12:31 Lajla: This behaviour is kind of nonstandard and most implementations don't do this, and hence you can turn it off if you don't like it. I don't, so I turned it off.
12:32 tony_: OK, we've thrashed this poor matter to Hades and back. I guess the best way to pose my question is: "Why not define another function (called, say, "eprint") to print in the new way; why not leave "print" alone?
12:33 tony_: (I forgot to close my quotes.)
12:33 tony_: And I think we're all tired of arguing it right now...
12:48 Lajla: tony_, I guess because it doesn't solve the repl issue, which is why you want a switch.
12:50 Lajla: (define (eprint x port) (when (or (pair? x) (symbol? x)) (display #\' port)) (print x port))
12:51 tony_: Lajla: there's clearly something here that I don't understand.  The PLT folks probably have a good reason to change the meaning of print, but because we don't see the reason, the change seems arbitrary.
12:52 (quit) anRch: Quit: anRch
12:52 Lajla: tony_, well, the reason is probably that they want to unify the language for code and for data a bit more.
12:52 Lajla: If you want to get a symbol in code, you write 'a
12:52 tony_: ??
12:52 tony_: What does that mean?
12:52 tony_: Scheme and Lisp have a long history of perfecting the problem of code-as-data.
12:52 Lajla: tony_, not really, often people find it to be quote confusing.
12:53 Lajla: Many people fall for the trap  like, if you say (string->symbol "+") ===> +
12:53 tony_: Lajla: thanks for conversing with me at length about this.  I just missed a 12:30 appointment and must run!
12:53 Lajla: They wil assume it evaluates to the addion function
12:53 tony_: Talk laters...
12:53 Lajla: Americans are way behind
12:53 Lajla: sure.
12:56 mye: rudybot: ((string->symbol "+") 1 1)
12:56 rudybot: mye: error: procedure application: expected procedure, given: +; arguments were: 1 1
12:58 (quit) tony_: Ping timeout: 276 seconds
12:58 mye: so if i put the symbol + in functional position I DO expect it to be the addition function
12:58 Lajla: Ah, but that is the trap people often fall for.
12:59 Lajla: (+ 1 2) evaluates the head of the list, and then applies it to its arguments
12:59 Lajla: As does ((string->symbol "+") 1 1)
12:59 Lajla: It first evaluates (string->symbol "+")
12:59 Lajla: Symbols evaluate to their values.
13:08 bremner: Is there a way to make the DrRacket debugger step into functions in a different module/file than the current one?
13:24 (join) jonrafkind
13:26 (quit) brooksbp: Remote host closed the connection
13:38 (quit) MayDaniel: 
13:39 mye: why not make eval more greedy and eval the functional position of a form until it's really reduced to a value?
13:41 Lajla: mye, you mean to a function
13:42 Lajla: mye, and you lose provability over scheme's lexical scope then.
13:43 Lajla: mye: http://codepad.org/VRpNyQGT
13:50 mye: not necessarily a function, just a non reducible expression. I'm trying to grok why this interferes with lexical scope.
13:53 jonrafkind: mye, are you asking why racket uses eager evaluation?
13:55 mye: jonrafkind: I wonder why it evaluates function position only once, as opposed to until it reaches a non reducible expression.
13:55 jonrafkind: can you show an example?
13:55 mye: rudybot: ((string->symbol "+") 1 1)
13:55 rudybot: mye: error: procedure application: expected procedure, given: +; arguments were: 1 1
13:55 Lajla: jonrafkind, he means, why deosn't that evaluate to 2.
13:55 jonrafkind: because you can't apply +
13:55 Lajla: As it, it keeps evaluating function position up to a function
13:56 jonrafkind: this isn't lisp
13:56 jonrafkind: rudybot: ('+ 1 1)
13:56 rudybot: jonrafkind: your sandbox is ready
13:56 rudybot: jonrafkind: error: procedure application: expected procedure, given: +; arguments were: 1 1
13:56 Lajla: mye, and it ruins lexical scoping.
13:56 jonrafkind: rudybot: (((lambda () +)) 1 1)
13:56 Lajla: Say you have ((string->symbol some-string) 1 2)
13:56 rudybot: jonrafkind: ; Value: 2
13:56 Lajla: And it composes that some-string based on runtime information
13:57 Lajla: there is no way then to establish bindings at compile time.
13:58 mye: I'm kind of getting there, slowly ;-)
13:59 Lajla: mye, basically, if that were allowed, it would be very, very, very hard to efficiently compile scheme.
14:00 Lajla: Essentially any scheme program compiled would have to include a scheme interpreter in it.
14:01 mye: so this is something like first-class macros then.
14:05 Lajla: No, it still won't change the evaluation model of macros.
14:05 Lajla: All arguments would still get evaluated.
14:06 Lajla: But like, say you have (let ((x <some-expression>)) (x 1 2))
14:06 Lajla: THe compiler can then lexially establish that the x in (x 1 2) must necessarily be bound to <some-expression> right?
14:07 Lajla: In your system, the head may or may not evaluate to x, depending on runtime information or even input.
14:07 Lajla: So the compiler basically can't decide that at compile time when it establishes variable bindings.
14:07 Lajla: The names of variables is essentially thrown away after compilation
14:08 Lajla: (let ((x <some-expression>)) (x 1 2)) and (let ((y <some-expression>)) (y 1 2)) is basically compiled to the same code.
14:15 mye: if (x) has to be a symbol for the form to make sense, then why eval it when it's not a symbol?
14:22 mye: ok, it just has to become a lambda after one step of eval
14:22 jonrafkind: rudybot: ((eval '+) 1 1)
14:22 rudybot: jonrafkind: ; Value: 2
14:25 mye: so, how does eval know a symbol is a macro?
14:25 (join) Demosthenes
14:25 mye: does it look the names of macros up first?
14:25 jonrafkind: it runs the macro expander
14:25 Demosthenes: so, i thought (describe ...) was incredibly useful in CL. is there a racket analogue I missed?
14:25 jonrafkind: Demosthenes, yea theres on on planet
14:26 jonrafkind: http://planet.racket-lang.org/display.ss?package=describe.plt&owner=williams
14:26 rudybot: http://tinyurl.com/25tfz87
14:26 mye: jonrafkind: always unconditionally?
14:26 jonrafkind: yea
14:27 mye: aha
14:27 mye: I think that explains my confusion away
14:27 jonrafkind: eval has few use cases these days
14:32 Lajla: eval suits common lisp more than scheme I guess.
14:33 Lajla: eval complicates a lot of things.
14:33 Lajla: Because basically what the list (+ 1 2) evaluates to depends on a plethora of things.
14:35 (join) dnolen
14:47 (quit) RyanRN: Quit: Leaving.
14:48 (join) RyanRN
14:54 (join) MayDaniel
14:56 (nick) Lajla -> Xajla
15:14 (quit) mye: Ping timeout: 240 seconds
15:20 (join) tony_
15:32 (quit) mwolfe: Remote host closed the connection
15:36 (quit) Demosthenes: Quit: leaving
15:45 em: i think one of the cool things in scheme is how a procedure can result in a procedure, and still be used in the procedure part of the expression.
15:49 tony_: em: the gods of Scheme hear your praise, and I'm guessing you'll experience a karma boost sometime in the next 48 hours.
15:50 tony_: (That's the average ETA for my Scheme-induced karma boosts.)
16:01 em: haha
16:01 Xajla: em, surely you mean a procedure call?
16:01 Xajla: And though or application
16:01 em: Xajla: yep
16:01 Xajla: Javascript can also do that by the way, and python.
16:02 em: Like I made a procedure porm that becomes + sometimes and - other times and you can do like ((porm) 3 5)
16:02 Xajla: Not a lot of people use it, but (makeAdder(2))(4) is valid javascript.
16:02 Xajla: घघI think you named it like that on purpose.
16:02 Xajla: Just to test how suggestive my mind is.
16:02 Xajla: And I will not fall for it.
16:02 tony_ giggles immaturely.
16:03 Xajla: plus-or-minus
16:03 em: yes :)
16:03 em: Xajla: does Javascript have tail-call-optimization?
16:04 tony_: rudybot: (define (+/- mid range) (cons mid range))
16:04 rudybot: tony_: your sandbox is ready
16:04 rudybot: tony_: Done.
16:05 tony_: em: observe that you can use operator symbols as function names.
16:05 tony_: In fact, the following would be better:
16:05 tony_: rudybot: (define +/- cons)
16:05 rudybot: tony_: Done.
16:05 Xajla: em, the standard doesn't require it.
16:05 Xajla: Some implementations might have
16:06 em: given that TCO kicksass why does anyone not use it? Is there some cost?
16:06 tony_: em: it's only relevant to languages that use recursion idiomatically.
16:07 tony_: Scala uses it.
16:08 em: if you do this in Python --
16:08 em: def fact(n): return 1 if n == 0 else n * fact(n -1)
16:08 em: it chokes and dies even before n = 1000
16:09 tony_: but obviously you know why
16:09 em: but the same exact algorithm in Scheme has no problem at all, ever.
16:09 tony_: not true; Scheme will eventually segfault too.
16:09 tony_: (for large enough n)
16:09 em: nope. it might run out of physical memory but it won't stack overflow like python does
16:09 tony_: the stack overflow is the reason *why* Scheme will run out of physical memory.
16:10 lucian: em: besides implementing it, the main cost is having a worse traceback
16:10 em: well try it, scheme is able to keep going orders of magnitude longer than python.
16:11 tony_: em: that may very well be true, but you cannot invoke that factorial func with unbounded n.
16:11 tony_: it's not tail recursive.
16:12 em: i wasn't saying you could. But my point is that Scheme quickly gives answers for inputs that are orders of magnitude larger than where python died.
16:12 em: im assuming that is the TCO
16:12 tony_: No, it isn't.
16:12 tony_: In the code you posted, the recursive call to fact is not in tail position.
16:12 lucian: em: if you rewrite it as a loop in python, it'll be just fine
16:13 tony_: The tail recursive version is: (...give me a sec...)
16:13 lucian: em: it's more a matter of style than anything else. and debuggability
16:14 tony_: (define (fact n) (let loop ((n n) (a 1)) (if (zero? n) a (loop (sub1 n) (* a n)))))
16:16 tony_: lucian: I'm a bit puzzled by your response.  Style and debuggability are not the reasons to prefer tail recursion to ordinary recursion.
16:17 lucian: tony_: no, they're the reasons why python doesn't support TCO
16:18 Xajla: em, that's not tail recursive.
16:18 Xajla: And tail recursion gives you less stack trace.
16:18 lucian: Xajla: no, TCO gives  you less stack trace
16:18 Xajla: Ehh, sorry.
16:18 Xajla: That is what I meant.
16:19 em: (define (fact a n) (if = n 0) a (fact (* a (fact a n)) (- n 1))))
16:19 em: how about that one?
16:19 tony_: em: not tail recursive
16:19 em: with (fact 1 n)
16:19 tony_: (and your if has a slight flaw.
16:19 tony_: )
16:19 em: oh yeah
16:20 tony_: Here's a more transparently tail-recursive version: (...one sec...)
16:21 tony_: (define (fact n) (define (helper m a) (if (zero? m) a (helper (sub1 m) (* a m)))) (helper n 1))
16:21 tony_: (I also un-shadowed n, for clarity.)
16:22 tony_: But the named let is somewhat conciser.
16:23 Xajla: em, the tail call is * in that case.
16:23 Xajla: A function is tail recursive if its tail call is itself.
16:24 tony_: Xajla, I think you're assuming an order of evaluating the args, no?
16:25 Xajla: tony_, no, why?
16:25 tony_: How do you know (- n 1) is not the last argument to be evaluated?
16:26 em: i think i got one here
16:26 em: (define (fact a n) (if (= n 1) a (fact (* a (- n 1)) (- n 1))))
16:27 em: but i thought the point of Schemes TCO is that it's TCO even if you don't make iterations.
16:27 tony_: "iterations" means do-loops, while-loops, and the like...all of which are unidiomatic in Scheme.
16:28 em: to me iterations mean that it is constant in space.
16:28 tony_: As for your new factorial, the recursive call is now certainly in tail position, but the correctness of the function depends on how you call it.
16:28 em: a recursive process that is constant in space = iteration. That's what i thought.
16:28 tony_: I'm not familiar with that meaning.
16:28 em: well obviously you call it (f 3 3)
16:29 tony_: I always thought "iteration" meant a loop that was controlled by a dedicated loop variable (usually "i" or "n").
16:29 em: how are you able to 'see' that this recent one is in tail position?
16:29 em: the reason i knew it would be is because it's an iteration in the sense i just mentioned.
16:30 (quit) Xajla: Ping timeout: 246 seconds
16:30 em: basically i think of an iteration as being a recursive process where the function itself is 'saving up' the results each time, so that it's like, you could put it on pause and give it to someone else, and *start from there* and it could finish.
16:30 tony_: Informally, tail position is just "the last thing that gets evaluated."
16:30 evhan: em: FWIW i don't believe thats a common idea of what "iteration" means
16:31 em: hm
16:31 evhan: though it may very well be technically correct, i dont know
16:31 tony_: iteration and recursion are mutually exclusive.
16:31 em: no i think an iteration is a type of recursion
16:31 bremner: well, depends on the jargon
16:31 em: isn't it?
16:31 tony_: it's very simple: recursion is a form of self-reference.
16:32 tony_: A while/for/do loop does not refer to itself, if for no other reason than that it has no name.
16:32 bremner: iterative is pretty commonly used to refer to tail recursive procedures
16:32 tony_: bremner: yes, but that's done casually.
16:32 bremner: tony_: casually in programming languages texts?
16:32 em: i got the idea i have about what these things mean from when i was reading SICP
16:33 evhan: i think a more common interpretation of 'iterative' would be 'making one or more passes over some data structure'
16:33 tony_: evhan: that rules out a lot of common iterations
16:33 evhan: though i suppose looping a side-effecting procedure would then be ... yes, tony_
16:33 evhan: thats what i was just realizing
16:34 tony_: bremner: in the context of this conversation, confusion seems to be arising out of not making a strong enough distinction between iteration and recursion.
16:34 em: what about: A recursive process is when a procedure calls itself. And an iteration is a recursive procedure that is constant in space.
16:34 tony_: em: no! :-)
16:34 tony_: em: rather, sometimes! :-)
16:35 tony_: Consider for example the classic tail-recursive version of reverse.
16:35 em: to me the essence of an iteration versus a recursion is this: For a general recursive process if you STOPPED it, and then lost all memory of the past, you could never start it again. But an iteration is a special case of recursion where you could STOP IT and lose all memory of the past, and you could start again and finish.
16:36 em: * ^
16:36 tony_: It is not an iteration, in the sense that no variable is introduced to control the loop.
16:36 tony_: em: it depends on what you mean by "stop."  Does "stopping" entail throwing away the stack?
16:37 em: im not familiar with all the jargon since im not a CS major.
16:37 tony_: which jargon? "stack"?
16:37 em: but i did say "throw away all memory of the past' you retain only the current condition of things.
16:37 em: yeah
16:37 tony_: i see.
16:37 em: if that means the same as stack then i think so.
16:37 tony_: Ok, I'll buy that for the time being.
16:38 bremner: if you think about it, em's defintion is closely related to constant working space
16:39 em: yeah constant space
16:39 tony_: em's definition is workable, but when I hear iteration out of a SICP-like context, I think "for (i = 0; i < n; i++) ..."
16:39 bremner: tony_: yes, but this is #racket ;)
16:40 em: i assumed that in other languages what they call iteration might look different and they may not know it, but it's the same thing.
16:40 tony_: yes, of course, but I sense that em is confused about tail recursion because of something too broad (too narrow?) in his understanding of the word "iteration."
16:40 tony_: (we're going around in circles!)
16:40 em: a recursive process that's keeping the working space constant. I assumed in other languages the designers of the language just hid it from you with syntactic sugar.
16:41 tony_: hid what, exactly?
16:41 (join) Lajla
16:41 bremner: the recursion
16:41 bremner: i.e. the goto
16:41 tony_: OMG! :-)
16:41 em: the fact it's a recursive process that's keeping the working space constant.
16:41 tony_: goto is a veritable *nemesis* of recursion :-)
16:41 evhan: most languages do not perform loops with recursion
16:42 em: even way down under the hood?
16:42 bremner: no, but after tco, the code looks  mighty similar
16:42 tony_: em: no, of course at the machine level it's a jump.
16:42 tony_: (i.e., a goto)
16:43 em: did lisp machines do it without jumps?
16:43 tony_: anyway, we have drifted away from the important point, which is that em needs to understand tail recursion.
16:43 tony_: em: I don't know; good question.
16:45 em: my working hypothesis about what TCO means is that somehow it's a guarantee that when you are composing procedure A and procedure B, (A (B))  B will only ever see the final result of A.
16:45 em: oops i meant (B (A))
16:46 tony_: em: B will always only see the final result of A, no matter how the latter is implemented.
16:46 em: well somehow B does not even come alive until A is done then.
16:46 tony_: That's right.
16:46 em: yeah that's what i meant.
16:46 tony_: More accurately, B is not *applied* until A is *evaluated*.
16:47 em: okay that's cool.
16:47 tony_: em: Out of curiosity, out of which books are you learning Scheme?
16:48 em: I've read the first chapter + a little bit of the second chapter of SICP, and then more recently I read the entire Racket Guide.
16:48 tony_: and how did you find them?
16:48 em: Racket is better than "just scheme" when it comes to feeling like "I could use this to make programs" versus, "I could use this to make algorithms to factor a polynomial" if you get me
16:49 tony_: actually, I don't get you, but one has to be very careful when making pronouncements on such matters. ;-)
16:51 em: Well I wanted to learn some programming language for a long time. Not for any practical reason. Since I mainly want to know programming just to feel smarter and be creative, and have no need for marketable skills my choices were wide open. I chose lisp because that's prestigious, and then once I was looking around lisps chandler pointed me toward Scheme, which leads one to SICP, and obvious PLT is most enjoyable way to do anything with Scheme.
16:51 bremner: I've heard the phrase "batteries included" to refer to Racket. and Python ;).
16:51 evhan: not that batteries are everything, mind
16:52 lucian: bremner: yes, it's a somewhat useful concept. although "batteries handy" is more important
16:52 Lajla: em, you know what, fuck tail recursion, let me explain to you what a tail-call is first.
16:53 Lajla: Not all tail calls are recursive.
16:53 Lajla: Say we have (define (f a b) (x (y z)))
16:53 em: I know a little bit about math so SICP is not *constantly* daunting but when you learn Scheme through SICP you start to feel like this is a cool language to write algorithms, but how will I ever do stuff like write an IRC bot or make a program with a GUI, etc. If you get me.
16:53 Lajla: To what function is the tail call there?
16:53 Lajla: Take a guess
16:53 tony_: Lajla: I'm not parsing you clearly again. :-(
16:54 em: the y ?
16:54 tony_: Are you asking, "Whose continuation is the same as the continuation of f?"
16:54 Lajla: em, no, y is most surely not a tail call.
16:54 Lajla: tony_, yes.
16:54 Lajla: But this one is for em.
16:54 Lajla: We'll get to that later.
16:54 em: well im not sure if you mean which one has to be the tail or make the call or listen for the call or what.
16:54 tony_: Yes, forget continuations.
16:54 Lajla: Let's not overcomplicate it, we're not only dealing with a woman, but also an american, we have to take it slowly.
16:55 Lajla: em, the call to x is the tail call.
16:55 em: i thought the idea of a tail call is that somehow when one is finish evaluating it calls forth the next one, and the next one is not summoned until the first is finished evaluating.
16:55 em: yes a call to x then.
16:55 Lajla: Nonono.
16:55 em: i was saying y has to finish first.
16:55 Lajla: A call is a tail call if basically its return value must automatically also be the return value of the calling function.
16:55 Lajla: Like, whatever the call to x evaluates to.
16:55 Lajla: f will also evaluate to that.
16:55 Lajla: right?
16:55 tony_: Lajla: you mean "the called function."
16:56 Lajla: no, I mean the calling function.
16:56 Lajla: (define (square x) (* x x))
16:56 em: confusing
16:56 Lajla: A more practical example
16:56 Lajla: the tail call is *
16:56 tony_: yes, ok.
16:56 em: i dont get what you mean by call
16:57 Lajla: em, hmm
16:57 Lajla: Well, let's say that * is a function
16:57 em: i assume tail must means the very last result in any algorithm
16:57 Lajla: Then (* x x) is a call to that function
16:57 Lajla: em, well, it's the result the function itself evaluates to.
16:57 Lajla: (square 3) evaluates to (* 3 3)
16:57 em: not 9
16:57 em: the tail of (square 3) is not 9 ?
16:57 (quit) MayDaniel: Read error: Connection reset by peer
16:58 Lajla: (* 3 3) evaluates to 9.
16:58 em: so isn't the tail of (square 3) 9 then or not?
16:58 Lajla: No, 9 is not a call.
16:58 Lajla: It's a value.
16:58 em: okay
16:58 Lajla: (* x x) is a function call.
16:58 Lajla: or (* 3 3) in this specific instance.
16:58 Lajla: Whose value is 9
16:58 Lajla: And because (* 3 3) evaluates to 9
16:59 Lajla: THen (square 3) evaluates to 9, right?
16:59 em: yeah but that's not the tail i guess.
16:59 em: i guess the tail of (sqare 3) is (* 3 3)
16:59 Lajla: Yeah
16:59 Lajla: That's the tail call.
16:59 em: is it also the tail alone?
17:00 em: i can't imagine how it could be any thing other than a tail call
17:00 Lajla: Okay, something more complex (define (factorial n) (if (zero? n) 1 (* n (factorial (- n 1)))))
17:00 Lajla: It has two tails
17:00 em: how could you possibly do (square 3) if it does not call (* 3 3)
17:00 Lajla: Find them both.
17:01 em: oh maybe (* n (factorial (- n 1)))  and (- n 1) ?
17:02 Lajla: THe first is correct
17:02 Lajla: the other tail is 1
17:02 Lajla: Which is not a call
17:02 Lajla: So it's not a tail call, but it's still a tail.
17:02 Lajla: This is because, whatever happens, the function either evaluates to 1, or to (* n (factorial (- n 1))) , right?
17:02 em: oh i see that's a tail.
17:02 Lajla: For any input n, that is what happens.
17:03 Lajla: It either evaluates to these two expressions, 1 or (* n (factorial (- n 1)))
17:03 em: yeah
17:03 em: that's clear enough
17:05 Lajla: Cool, one more for the road: (define (describe person) (case person ((emma) "has boobs") ((lajla) (string-append "very" " annoying")) ((chandler) "gone from here") (else (+ 1 2 3)))
17:05 Lajla: You see there are four tails there, two of which calls, right?
17:06 em: yes
17:07 em: 4 tails and string-append and + are the calls
17:07 Lajla: Yeah.
17:07 Lajla: Well, technically they are the functions being called in the calls, but you get it.
17:07 Lajla: So this is why (define (factorial n) (if (zero? n) 1 (* n (factorial (- n 1))))) is not _tail recursive_
17:07 em: did chandler say he was never coming back here?
17:08 Lajla: Even though there are definitely tail calls
17:08 Lajla: But the tail call is to *
17:08 bremner: em: you know how some people have imaginary friends? Lajla has imaginary enemies.
17:08 em: hehe
17:08 Lajla: So, to make it _tail recursive_ you have to find a way to make the call to factorial itself.
17:08 Lajla: bremner, nonononon, chandler is real man.
17:08 Lajla: He banned me from #scheme.
17:09 em: oh that takes some doing, i thought that was a pretty relaxed channel with hardly any moderation at all.
17:09 em: you could probably just get unbanned by now.
17:10 em: anyone who listens to you can tell you know a lot about scheme
17:10 Lajla: So... enter: (define (factorial n) (define (factorial-helper n acc) (if (zero? n) acc (factorial-helper (- n 1) (* acc n)))) (factorial-helper n 1))
17:10 Lajla: em, chandler disagrees.
17:10 Lajla: And besides, no offense, but you're probably not yet the best person to judge that.
17:10 Lajla: Point was, I made a joke about child rape. -_-
17:11 em: Always a risky move.
17:11 Lajla: But, the factorial-helper function there is tail-recursive.
17:11 em: you gotta no your audience
17:11 Lajla: I like ot live dangerously.
17:11 Lajla: Well, I made it like 2 seconds after I entered. And I got back at me months and months later.
17:11 Lajla: I usually enter channels with a controversial statement to test if people aren't total stiffards.
17:12 em: well i know that feeling ostracised is one of the worst feelings and it is hard to let go of especially for people who already may have a tendency to focus on things with the laser like attention that is usually an asset in something like programming.
17:12 Lajla: But, going on, the factorial-helper is tail recursive there, factorial itself is not, but it calls factorialh-elper after like only one call, so it's really not that bad.
17:12 (quit) tony_: Quit: Ex-Chat
17:12 em: We will try to work this out. I have to go get my pizza now though.
17:13 Lajla: lol american pizza lol.
17:13 Lajla: Bon appetit.
17:18 em: I know you have an irrational disregard for America but the USA is where some of the best Pizza can be found. I would even say, by now, it's become an American food.
17:18 em: You have maybe never been to NYC or Chicago etc.
17:21 (join) Fare
17:22 Lajla: Nahh, I just jest
17:22 Lajla: I like New York Pizza.
17:22 Lajla: em, but you now realize what you have to do to make a function tail-recursive, right?
17:27 (join) lewis1711
17:27 (join) jao
17:27 (quit) jao: Changing host
17:27 (join) jao
17:48 (join) tony_
18:47 tony_: Lajla and bremner: btw, I got an enlightening mailing list response to my 'print' query from the good folks at PLT.
18:48 Lajla: tony_, what did they say?
18:48 tony_: Several things, most important of which is that, astonishingly, the R6RS standard does not specify the behavior of print (!).
18:51 (join) jfalcon
18:54 Lajla: tony_, I'm pretty sure that 'print' is a nonstandard extension
18:54 Lajla: R5/R6 define write and display
18:55 tony_: Yes, that's the point.
18:55 tony_: I didn't know that until now.
19:01 lewis1711: http://paste.pocoo.org/show/337978/ bah, who needs classes
19:12 eli: tony_: Sorry, I got disconnected right after that.
19:13 tony_: eli: not a problem; I got a thorough answer on the mailing list.
19:13 eli: Americans are an extraordinarily habitual animals -- 12:00 comes, and they all flock to the dining areas.
19:13 tony_: How was skiing?
19:13 tony_: I hear the French are even more regimented...
19:13 eli: Tomer was doing nicely -- even knows how to turn.
19:13 lewis1711: http://docs.racket-lang.org/guide/contract-func.html#(part._contracts-flat-named-contracts) why is not default behaviour? :/
19:14 eli: In any case, the whole debate refers to the usual argument on evaluation vs normalization.
19:14 tony_: Cool. I haven't skied in about 6 years.
19:14 eli didn't ski in much more...
19:14 tony_: Well I was working on the very wrong assumption that R6RS defines print; was surprised to discover it doesn't.
19:15 eli: There are some other scheme implementations that print an `x' symbol as "'x".
19:15 tony_: So it's up for grabs.
19:15 eli: Yeah, that's kind of the whole point of having a function that is not under any standard.
19:15 tony_: But this is truly surprising, because the concept of REPL is so fundamental.
19:15 eli: (Or expectations.)
19:15 tony_: Yes, but the repl itself sets up expectations (at least for me.)
19:16 tony_: As in: "repl" and not "rewl" and not "redl".
19:16 tony_: Anyway, I'm safe with #lang scheme.
19:16 eli: There's no contradiction here -- it's still a "print" step.
19:16 eli: The only question is how stuff gets printed.
19:17 tony_: yep.
19:17 eli: And it just happens that it took a long time to get to a printout that is more like scheme48's or mit-scheme's.
19:17 tony_: Is that what they do?
19:17 eli: Yeah.  The latter goes to an even more extreme printout thing, where it can print closures out.
19:18 tony_: That makes more sense to me.
19:18 tony_: So I guess consistency requires that 'a be printed as 'a after evaluation.
19:19 eli: Well, it's kind of a higher commitment to the repl -- making it a little more than just a device for quick experiments.
19:19 eli: Oh here --
19:19 eli: (lambda (x) x) gets printed as
19:19 eli: ;Value 11: #[compound-procedure 11]
19:19 tony_: That's PLT Scheme, right?
19:20 eli: And once you see that you can type just that -- #[compound-procedure 11] -- and it will evaluate to the same function.
19:20 eli: No, that's mit-scheme.
19:20 eli sighs
19:20 eli: Every time I exit mit-scheme I need to look for a barf bag.
19:20 tony_: So, the number 11 must be a unique oid for closures.
19:20 eli: "End of input stream reached.", "Happy Happy Joy Joy."
19:21 tony_ hands Eli a barf bag from Jet Blue.
19:21 eli goes to dinner stuff...
19:23 eli: tony_: Yeah, it basically assigns a number for every otherwise-unprintable value to make a printout that can still access it.
19:23 eli: It's kind of cute, but there's a question about GCing, of course.
19:24 lewis1711: http://paste.pocoo.org/show/337980/ what does the top error mean? :/
19:25 tony_: eli: This is beginning to make more sense now.  I knew there must be a good reason for it but just couldn't figure it out.
19:37 (join) jack
19:37 jack: hey guys
19:38 jack: anyone here?
19:38 (nick) jack -> Guest67232
19:38 Guest67232: hi
19:38 Guest67232: is anyone here
19:40 jonrafkind: sup bro
19:43 offby1: nobody here but us zombie processes.
19:44 jonrafkind: i like the switch from a real name to some random guest name
19:44 jonrafkind: classy
19:45 (quit) jao: Ping timeout: 240 seconds
19:46 Guest67232: the system did it
19:47 offby1: I blame society
19:47 Guest67232: could you guys help me with a design?
19:48 jonrafkind: you don't need to ask to ask. just ask
19:48 Guest67232: kk, thanks
19:49 Guest67232: i want to make a permutation of a list, it should return all the possibilities
19:49 Guest67232: eg, for input 123, i should get 123, 132 213 231 312 321
19:50 lewis1711: humm, permutation is always a set isn't it?
19:51 Guest67232: i want to do it like a list of lists
19:51 Guest67232: so it should return (list (list 1 2 3) (list 1 32) etc))
19:51 Guest67232: yup
19:51 Guest67232: it should be
19:53 offby1: Guest67232: here's what I stole from SICP: http://ix.io/1tm
19:53 offby1: rudybot: eval (define (perms s) (if (null? s) (list '()) (append-map (lambda (x) (map (curry cons x) (perms (remove x s)))) s)))
19:53 rudybot: *offby1: your scheme sandbox is ready
19:53 rudybot: *offby1: Done.
19:53 offby1: rudybot: (perms (list 1 2 3))
19:53 rudybot: *offby1: ; Value: ((1 2 3) (1 3 2) (2 1 3) (2 3 1) (3 1 2) (3 2 1))
19:53 Guest67232: actually, im stuck on one thing. i have a list, and i want to insert an element to every position of the list
19:53 Guest67232: thx, buddy!
19:54 offby1: bonus points if you understand what it's doing :)
19:55 Guest67232: im trying
19:56 Guest67232: but it returns an error, the function contains ',
19:56 Guest67232: quote: expected a name after a ', found something else
19:56 lewis1711: could you do it with a list comprehension? in haskell: [(a, b, c) | a <- [1..3], b <- [1..3], c <-  [1..3], a/=c, a/=b, b/=c]
19:56 lewis1711: racket has iota, and list comprehensions
19:58 Guest67232: append-map is not defined
19:59 offby1: rudybot: doc append-map
19:59 rudybot: *offby1: http://docs.plt-scheme.org/reference/pairs.html#(def._((lib._racket%2Flist..rkt)._append-map))
20:00 (join) jao
20:00 (quit) jao: Changing host
20:00 (join) jao
20:01 Guest67232: thx, probably im in beginning student custom, so it doesnt contain append-map
20:01 offby1: that'd probably be the reason
20:03 Guest67232: what language r u using?
20:03 jonrafkind: r u
20:03 jonrafkind: are you
20:04 Guest67232: what language are you using?
20:04 bremner: what course are you doing homework for?
20:05 offby1: m uzng "racket"
20:05 offby1: rudybot: (perms (build-list 5 values))
20:05 rudybot: *offby1: ; Value: ((0 1 2 3 4) (0 1 2 4 3) (0 1 3 2 4) (0 1 3 4 2) (0 1 4 2 3) (0 1 4 3 2) (0 2 1 3 4) (0 2 1 4 3) (0 2 3 1 4) (0 2 3 4 1) (0 2 4 1 3) (0 2 4 3 1) (0 3 1 2 4) (0 3 1 4 2) (0 3 2 1 4) (0 3 2 4 1) (0 3 4 1 2) (0 3 4 2 1) (0 4 1 2 3) (0 4 1 3 2) (0 4 2 1 3) (0 4 2 3 1) (0 4 3 1 2) (0 4 3 2 1) (1 0 2 3 4) (1 0 2 4 3) (1 0 3 2 4) (1 0 3 4 2) (1 0 4 2 3) (1 0 4 3 2) (1 2 0 3 4) (1 2 0 4 3) (1 2 3 0 4) (1 2 3 4 0) (1 2 4 0
20:05 offby1: rudybot: (perms (build-list 10 values))
20:05 rudybot: *offby1: error: evaluator: terminated (out-of-memory)
20:05 offby1 drums fingers
20:05 offby1: rudybot: (perms (build-list 2 values))
20:06 rudybot: *offby1: your scheme sandbox is ready
20:06 rudybot: *offby1: error: reference to an identifier before its definition: perms in module: 'program
20:06 offby1: *gasp*
20:07 Guest67232: for ubc cpsc 110
20:08 jonrafkind: xyz 123
20:10 Guest67232: i still dont get append-map...but thank you anyways
20:13 offby1: (define (append-map proc seq)) is just (apply append (map proc seq))
20:19 (join) jfalcon_
20:19 (quit) jfalcon: Read error: Connection reset by peer
20:19 (nick) jfalcon_ -> jfalcon
20:23 em: offby1: hey where did it get perms from that's not in racket is it?
20:24 offby1: em: I pasted in the definition half an hour ago
20:24 offby1: rudybot: eval (define (perms s) (if (null? s) (list '()) (append-map (lambda (x) (map (curry cons x) (perms (remove x s)))) s)))
20:24 rudybot: *offby1: Done.
20:25 em: offby1: rudybot remembers all your definitions from previoius conversations with it?
20:25 em: offby1: does it keep it seperate for each person?
20:25 offby1: unless the sandbox blows up, or two other people create sandboxes.
20:25 offby1: yes, they're separate
20:25 em: offby1: and can you ever reset it?
20:25 offby1: rudybot: (perms (build-list 2 values))
20:25 rudybot: *offby1: ; Value: ((0 1) (1 0))
20:25 offby1: yes
20:25 offby1: rudybot: init racket
20:25 rudybot: *offby1: your racket sandbox is ready
20:25 offby1: rudybot: (perms (build-list 2 values))
20:25 rudybot: *offby1: error: reference to an identifier before its definition: perms in module: 'program
20:25 offby1: voilà
20:25 em: cool
20:25 offby1: eli wrote most of that
20:26 (quit) lucian: Remote host closed the connection
20:27 Guest67232: suppose i have  a list, and i want to insert an element to every postion of the list. how can i do that?
20:27 jonrafkind: add-between or smoething
20:28 offby1: suppose I am a Congressman, and suppose I am an imbecile.  But I repeat myself.
20:29 offby1: rudybot: (map (curry cons 3) (list 'a 'b 'c))
20:29 rudybot: *offby1: ; Value: ((3 . a) (3 . b) (3 . c))
20:29 em: offby1: then i suppose you would post a half naked picture of yourself through craigslist.
20:29 bremner: don't give him ideas
20:29 offby1: em: I was on vacation last week, but I am dimly aware of what you refer
20:30 offby1: rudybot: (append-map (curry list 3) '(a b c))
20:30 rudybot: *offby1: ; Value: (3 a 3 b 3 c)
20:30 Guest67232: thx
20:31 Guest67232: but racket dont accept curry
20:31 offby1: charlie don't surf
20:31 offby1: rudybot: doc curry
20:31 rudybot: *offby1: http://docs.plt-scheme.org/reference/procedures.html#(def._((lib._racket%2Ffunction..rkt)._curry))
20:31 offby1 begs to differ
20:32 offby1: Guest67232: you're probably still in a restricted language.
20:32 tony_: Guest: (foldr (lambda (x a) (cons x (cons something a))) '() the-list)
20:33 Guest67232: i guess so
20:33 offby1: (apply append (map (curry list 3) '(a b c)))
20:33 offby1: rudybot: (apply append (map (curry list 3) '(a b c)))
20:33 rudybot: *offby1: ; Value: (3 a 3 b 3 c)
20:33 offby1: rudybot: (apply append (map (lambda (frotz) (list 3 frotz)) '(a b c)))
20:33 rudybot: *offby1: ; Value: (3 a 3 b 3 c)
20:33 offby1: ta da
20:34 em: curry is really weird
20:34 Guest67232: thank you so much offby1
20:34 Guest67232: and budybot
20:34 Guest67232: and everyone here today
20:34 tony_: the fold solution is more efficient and easier to understand.
20:35 tony_: (oh well...my 2 cents)
20:35 Guest67232: i guess my racket is so restricted that it doesnt accept apply...
20:36 offby1: geez
20:37 tony_: So is there an 8 pm fold curfew?
20:38 offby1: doubt it; fold is over 15, as far as I know.
20:41 (quit) RyanRN: Read error: Connection reset by peer
20:41 (join) RyanRN
20:53 (quit) tony_: Ping timeout: 240 seconds
20:53 dnolen: is it possible to import fluid-let into a Racket program? how do you do this?
20:54 (part) lewis1711
20:55 offby1: rudybot: doc fluid-let
20:55 rudybot: *offby1: no docs for a current binding, but provided by: mzscheme
20:55 offby1: rudybot: (require 'mzscheme)
20:55 rudybot: *offby1: error: require: unknown module: 'mzscheme
20:55 offby1: rudybot: (require mzscheme)
20:55 rudybot: *offby1: Done.
20:55 offby1: rudybot: doc fluid-let
20:55 rudybot: *offby1: http://docs.plt-scheme.org/mzscheme/Old_Syntactic_Forms.html#(form._((lib._mzscheme%2Fmain..rkt)._fluid-let))
20:56 offby1: rudybot: eval (fluid-let ([a 10]) (printf "a is ~a~%" a))
20:56 rudybot: *offby1: error: reference to an identifier before its definition: a in module: 'program
20:56 offby1: rudybot: eval (let ([a 'frotz]) (fluid-let ([a 10]) (printf "a is ~a~%" a)))
20:56 rudybot: *offby1: ; stdout: "a is 10\n"
21:07 (quit) jfalcon: Quit: jfalcon
21:09 dnolen: thx
21:22 em: currying is so strange tome.
21:23 em: what is the real purpose of it, when do you use it mostly? why not just avoid it?
21:27 offby1: saves typing :)
21:27 offby1: (curry cons 3) => (lambda (x) (cons 3 x))
21:28 em: so it's sort of something ot use when you want a built in value for a two or more argument procedure?
21:28 bremner: saves even more typing in Haskell, where (curry cons 3) is written "cons 3" ;)
21:29 offby1: em: the idea is that you take a function that takes N arguments, and you want some of them to be constant, so it makes you a new function where you only have to specify the other arguments.
21:30 em: okay that makes senes
21:46 (quit) Blkt: Quit: ERROR: do not makunbound t please!
21:51 (quit) Guest67232: Quit: Page closed
22:23 (quit) jonrafkind: Ping timeout: 240 seconds
22:42 (nick) coyotama|2 -> coyotama
22:43 (nick) coyotama -> bandu
22:43 (quit) bandu: Changing host
22:43 (join) bandu
22:44 (nick) bandu -> tama
22:44 (nick) tama -> bandu
23:04 (quit) corruptmemory: Ping timeout: 240 seconds
23:59 (quit) masm: Ping timeout: 245 seconds