00:20 (quit) bertorob: Quit: Ex-Chat
01:24 (join) carleastlund
02:31 (quit) carleastlund: Quit: carleastlund
02:31 (join) jonrafkind
03:10 (quit) hanDerPeder: Quit: hanDerPeder
03:48 (quit) jonrafkind: Ping timeout: 265 seconds
03:51 (join) spidermario
04:04 (quit) rbarraud: Read error: Connection reset by peer
04:07 (join) rbarraud
05:22 (join) masm
05:40 (join) maoo
06:10 (join) MayDaniel
06:41 (quit) rudybot: Ping timeout: 276 seconds
06:53 (join) rudybot
06:57 (quit) chturne: Quit: Leaving
06:59 (join) mceier
07:15 (join) rbarraud_
07:15 (quit) MayDaniel: 
07:15 (quit) rbarraud: Ping timeout: 252 seconds
07:44 (join) martinhex
08:11 (join) Karpenter
08:12 Karpenter: hey im a beguiner with scheme i just wanted to ask some questions, i know its not a big deal, but i cant seam to be able to code this.
08:12 Karpenter: i wanna make a procedure to count the number of digits
08:12 Karpenter: in a number
08:13 Karpenter: i know i can do ln(specific number)/ln(10) + 1
08:13 Karpenter: but i wanted to do it recursevly cant seam to know how
08:13 Karpenter: ex: (define (count-digit n)
08:13 Karpenter: (cond (= n 0)
08:14 Karpenter: after this i cant get the algorithm rigth
08:14 Karpenter: im not asking to make the code for me
08:14 Karpenter: but just give a tip or something
08:44 (join) MayDaniel
08:44 (quit) MayDaniel: Changing host
08:44 (join) MayDaniel
08:49 (join) jeapostrophe
08:51 Karpenter: or at least how the math is done
08:51 Karpenter: how to get the number of digits out of a number
08:51 Karpenter: mathematecaly speaking
08:52 (join) chturne
09:01 bremner: Karpenter: well, generally this is closely related to log base 10, right?
09:01 Karpenter: y
09:01 bremner: oh, you want to do it the hard way ;), I see.
09:01 Karpenter: precisely
09:01 bremner: well, how many times can you divide by 10?
09:02 Karpenter: :)
09:02 Karpenter: the point was to divide by 10
09:02 Karpenter: and assign
09:02 bremner: make a recursive formula for that first
09:02 Karpenter: 1
09:02 Karpenter: how to i assign (/ n 10) .-----> 1
09:02 Karpenter: an expretion to a number
09:02 bremner: doesn't make sense to me.
09:03 Karpenter: well take the number 560
09:03 Karpenter: 56,0
09:03 Karpenter: you divide by 10
09:03 Karpenter: you count 1 digit
09:03 Karpenter: you divide again
09:03 Karpenter: you count another
09:03 Karpenter: you divide again and again to reach 0
09:03 Karpenter: so the result would be 3 digits
09:04 bremner: so, think about how to write digits(560) in terms of digits(560/10)
09:04 Karpenter: i see
09:13 Karpenter: how do i tell it to stop
09:13 Karpenter: when it reaches 0
09:13 Karpenter: hmm
09:14 Karpenter: that well
09:14 Karpenter: pity scheme doesnt do loops
09:14 Karpenter: pff
09:17 bremner: Karpenter: you need a base case in your recursion
09:17 Karpenter: n=0
09:17 Karpenter: ?
09:17 bremner: Karpenter: this is not a scheme issue, it is a writing a recursive function issue. Have you done that before?
09:18 Karpenter: well yeah
09:18 Karpenter: its just i cant figure, this specific recursion
09:18 Karpenter: tbh
09:18 Karpenter: i know how to define factorials
09:18 Karpenter: and similar stuff
09:18 Karpenter: but to count
09:19 Karpenter: digits in a number
09:19 Karpenter: hmm
09:19 bremner: hint: what numbers have only one digit?
09:19 Karpenter: 0-9
09:19 bremner: (assuming non-negative integers)
09:19 bremner: and how do you check for 0-9?
09:20 Karpenter: 10000
09:20 Karpenter: 10 100 1000
09:20 Karpenter: hmm
09:20 Karpenter: first ten numbers are assigned
09:20 Karpenter: digit 1
09:20 Karpenter: next multiplyed by ten
09:20 Karpenter: 2 digits
09:21 Karpenter: i may try it this way
09:21 Karpenter: tx
09:29 Lajla: Karpenter: http://codepad.org/ObdBNn02
09:30 Lajla: The definiotn of 'helper' is the important one
09:30 Lajla: The rest around it is just to make you be able to specify the base. But assume base 10 with no base specified.
09:30 Karpenter: i see
09:30 Karpenter: thanks
09:31 Lajla: Basically you have a base case. If the number is smaller than the base, it is 1. And a recursion case if it's as large as the base or larger.  Where it's 1+ the amount of digits in the numnber divided by the basd.e
09:32 Lajla: This only works on natural numbers though
09:32 Karpenter: i see
09:35 Lajla: bremner, no fault me for making people dependent on shitty R5RS when I should explain to them the wondrous optional arguments
09:36 (quit) martinhex: Remote host closed the connection
09:52 Karpenter: anyways im making an optional argument
09:52 Karpenter: now
10:00 Lajla: Using (define (count-digits number (base 10))
10:00 Lajla: ?
10:19 Karpenter: something similar
10:19 Karpenter: but no
10:19 Karpenter: wasnt thinking going that way
10:19 Karpenter: using quotient
10:19 Karpenter: and remainer
10:20 Karpenter: that wont be recursive
10:20 Karpenter: if i go that way
10:20 Karpenter: i could just make (ln(number) /ln 10 ) 1
10:20 Karpenter: i could just make (ln(number) /ln 10 ) +1
10:21 Karpenter: and the integer part
10:21 Karpenter: being the number of digits
10:21 Karpenter: not recursive
10:21 Karpenter: :(
10:21 Karpenter: since im studying recursive cases
10:21 Karpenter: i need to do it this way
10:21 Karpenter: its not the easyest  i know
10:21 Karpenter: or fastest
10:21 Karpenter: or the most comon
10:22 Karpenter: but it got to be this way
10:27 (join) b-man_
10:29 (join) mije
10:48 (quit) b-man_: Remote host closed the connection
10:59 mije: "This is no Scheme programming sweatshop, such as you'll find in the north-eastern United States, where innumerable young men are forced to work twenty hours a week, or more, to produce code and dissertations. " <- why is this written on http://schematics.sourceforge.net/ ?
11:02 Lajla: Karpenter, I found my code to be surprisingly small easy.
11:02 Lajla: Karpenter: http://codepad.org/Pa0vaFYE
11:03 Lajla: This is the version using racket's extensions
11:06 Lajla: Karpenter: http://codepad.org/wbF14nQ3 here we have the same algorithm in PHP
11:10 Karpenter: you just crashed my head
11:11 Lajla: Why?
11:11 Karpenter: i was over this for so mutch time
11:11 Karpenter: i eventualy got tired
11:11 Lajla: Yes, but I am brilliant.
11:11 Karpenter: and made
11:11 Karpenter: this
11:11 Karpenter: (define (cd n)   (if (< n 10) 1)   (- (quotient (round (log n)) (round (log 10))) 1))
11:11 Lajla: log
11:11 Karpenter: it works
11:11 Karpenter: xD
11:11 Karpenter: epicness to the last
11:11 Karpenter: and since it rounds 3.9 to 4.0
11:12 Karpenter: i have to subtract
11:12 Karpenter: 1
11:12 Lajla: 1+ the amount of digits in the thing divided by 10 works fine too.
11:12 Karpenter: in order for the maths to be good
11:12 Lajla: It only counts the digits before the 0 you see 172.178127134 counts a three
11:13 Karpenter: you know
11:13 Lajla: Some might argue that to do it superly clean you would need if you test 1234 1 + the amount of digits in 234 or 123
11:13 Karpenter: it seams so
11:13 Lajla: But it works anyway in this way.
11:14 Karpenter: well dont think ill need that
11:14 Karpenter: :D
11:14 Karpenter: yeah true
11:14 Lajla: (The latter case is just dividing by the base anr rounding)
11:14 Karpenter: so
11:14 Karpenter: in that code
11:14 Karpenter: you add +1
11:14 Lajla: which showcases my divine brilliance.
11:14 Lajla: Yes.
11:14 Karpenter: to the amount of digits
11:14 Lajla: to the amount of digits that the number divided by the base has.
11:15 Karpenter: oh i see
11:15 Lajla: You could use (floor (/ number base))
11:15 Karpenter: witch is
11:15 Lajla: If you want to be supar clean
11:15 Karpenter: 1
11:15 Karpenter: briliant
11:15 Karpenter: xD
11:16 Lajla: Yes, it expans to (count-digits 1234) = 1 + (count-digits 123) = 1 + 1 + (count-digits 12) ...
11:16 Karpenter: ah now i get it
11:16 Karpenter: the thing was
11:16 Karpenter: start
11:16 Karpenter: i didnt had a base
11:16 Lajla: That I am brilliant, hence I solved this hands done.
11:16 Karpenter: that i understood
11:16 Karpenter: rofl
11:17 Karpenter: :)
11:17 Lajla: Base case is 1 digit. Which is the number being < 9
11:17 Karpenter: i shall name you Lajla the briliant one
11:17 Lajla: Good idea.
11:17 Lajla: I will eat you last of my minions.
11:18 Lajla: Actualyl most people in this channel would though.
11:23 (quit) jeapostrophe: Quit: jeapostrophe
11:25 (join) jeapostrophe
11:29 Karpenter: minions hmmm delicious
11:30 Lajla: Yes, let's take a bite of bremner
11:34 (quit) mije: Ping timeout: 265 seconds
11:38 (join) anRch
11:39 (join) rmitt
11:41 (quit) rmitt: Client Quit
12:01 (join) rmitt
12:02 rmitt: general question: is it wrong of me to dislike Java?
12:02 (join) jonrafkind
12:02 Karpenter: no
12:02 Karpenter: imo
12:03 Karpenter: some people think its a mess
12:03 Karpenter: others think its average
12:03 Karpenter: but is isnt bad to dislike java
12:03 Karpenter: im a python freak
12:03 Karpenter: so ...
12:03 Karpenter: :p
12:04 rmitt: I just worry about the consequences of my taste on my employability ;-)
12:04 Karpenter: lol
12:04 Karpenter: :D
12:04 Karpenter: a sugestion dont say your tastes
12:04 Karpenter: just go there
12:04 Karpenter: ill do anything
12:05 rmitt: I'd much rather have a job writing functional code
12:05 rmitt: and I like static typing as much as dynamic, but I prefer that the former be inferred.
12:06 Karpenter: work and likings to bond together
12:06 Lajla: rmitt, not at all, you'll fit right in our elitist circles.
12:06 Karpenter: one must dislike work in order to do it good
12:06 Karpenter: imo
12:07 Lajla: rmitt, the lack of inference isn't as much the piss as the lack of type variables.
12:07 rmitt: Karpenter: interesting perspective...and I wonder if you just might be right
12:07 Karpenter: yeah you know
12:07 rmitt: Lajla: I'd LOVE to fit into your elitist circles... if only I could afford it :-)
12:08 Karpenter: hahahah
12:09 Lajla: rmitt, just say all the time that racket is the most superior language for any purpose except human resource management and you should be fine.
12:09 Karpenter: even 3 dimentional representation
12:09 Karpenter: dont think so
12:10 Karpenter: it can be done
12:10 Karpenter: but i doubt anyone did a 3d environment out of racket
12:10 Karpenter: not saying its imposible
12:10 Karpenter: but is it easyer
12:10 Karpenter: or superior
12:10 Karpenter: ?
12:16 Lajla: You mean a Raquet 3D engine/
12:16 Karpenter: y
12:17 Karpenter: or maybe
12:17 Karpenter: not go that far
12:17 Karpenter: just make a small 3d rendering
12:17 Karpenter: program
12:17 Karpenter: that acepts ligth sources
12:17 Karpenter: and a small .obj file
12:17 Karpenter: dunno how to do that
12:17 Karpenter: from what i know of scheme or racket
12:18 Karpenter: just curious if its possible to do that
12:19 Karpenter: i doubt it would be easy from my ligth experience with C and C++
12:19 Karpenter: since there are some things c++ does better
12:19 Karpenter: from my perpective
12:19 Karpenter: than racket
12:21 (quit) jonrafkind: Ping timeout: 240 seconds
12:26 Karpenter: still got to say Lajla racket in a curriculum could be good :D
12:26 (quit) chandler: Ping timeout: 240 seconds
12:27 (join) chandler
12:27 (nick) chandler -> Guest24533
12:36 (quit) Guest24533: Changing host
12:36 (join) Guest24533
12:36 (nick) Guest24533 -> chandler
12:39 (quit) anRch: Quit: anRch
12:46 (join) Scifi
12:47 (quit) Karpenter: Quit: Page closed
12:47 Scifi: can someone help me
12:47 Scifi: oops
12:47 Scifi: http://racket-lang.org/irc-chat.html
12:47 Scifi: (define number 234)  (define (existe-digito number x)   (cond      ((= number x) #t)     ((= (* (remainder number 10) 10) x) x)     (else (existe-digito (= (* (remainder number 10) 10) x) x))))
12:47 Scifi: ill paste bin it
12:48 Scifi: http://pastebin.com/Zn0Wdchs
12:48 Scifi: dont understand
12:48 Scifi: =: expects type <number> as 1st argument, given: #f; other arguments were: 4
12:49 Scifi: i was trying to
12:49 Scifi: make
12:49 Scifi: a identifier , identify if a digits is on the number
12:49 (quit) jeapostrophe: Quit: jeapostrophe
12:49 Scifi: if anyone can help id apretiate
12:57 (quit) chturne: Quit: Leaving
13:00 (join) karpenter
13:00 Scifi: wtf
13:00 Scifi: how come you have the same ip as mine
13:00 Scifi: are you my neighbour or something
13:01 karpenter: ´heh
13:01 karpenter: depends u«if we are using the same network
13:01 Lajla: Scifi, your else clause is fault.y
13:01 Lajla: You give existe-digito the result of (= (* (remainder number 10) 10)
13:01 Lajla: Which is always #t or #f
13:01 Lajla: And it expects a number
13:01 Scifi: hm i see
13:02 Lajla: You might also want to generalize it for bases.
13:03 Scifi: hm
13:03 Scifi: why does it expects a number
13:03 Scifi: the base
13:03 Scifi: is wrong
13:03 Scifi: it seams
13:04 Scifi: wow
13:04 Scifi: im doing it wrong
13:04 Scifi: y
13:05 (quit) Scifi: Quit: Page closed
13:16 (quit) spidermario: Remote host closed the connection
13:18 (quit) karpenter: Ping timeout: 265 seconds
13:28 (join) hanDerPeder
13:31 (join) jonrafkind
13:47 (quit) rmitt: Quit: Page closed
14:10 (join) chturne
14:24 (quit) MayDaniel: 
14:26 (join) b-man_
15:12 (join) MayDaniel
15:34 (join) rmitt
15:36 rmitt: ok, so why does DrRacket print '(a) when I enter '(a)?  The repl should output (a) instead.
15:36 rmitt: (btw, this is with #lang racket)
15:41 eli: rmitt: Why should the repl print "(a)" instead?
15:41 eli: Why not "a list with a single element which is the symbol whose name is `a'"?
15:42 Lajla: rmitt, it prints the simplest expression it can find which evaluates to what your expression evaluates to I guess.
15:42 Lajla: Let me rephrase.
15:43 Lajla: It outputs a stream of characters which when read by the reader and the result of that evaluated in some environment whose properties I shan't be paedantic enough to fully specify evaluate to the result of the expresson you input.
15:46 Lajla: rmitt, I agree by the way that this outform form is 'counter intuitive' and only work insofar you consider the data as a program.
15:47 eli: Lajla: Speaking about majority of people etc, printing "'(a)" is by far much more intuitive than "(a)".
15:47 Lajla: It leaves me to wonder why (+ 1 2) outputs "3" and not "'3" or "(-10 7)" or "(string->number "11" 2)"
15:48 Lajla: Plus the slashes
15:48 Lajla: eli, I guess it doesn, that's the annoying part about 'intuitive', it differs for people.
15:48 Lajla: The fact that most REPL's of lisp do not, or historically haven't has led to some confusion I guess.
15:49 Lajla: The difference of a datum and its evaluation
15:49 eli: Yes, but that's a small minority.
15:49 eli: Why would you expect 3 to print as "'3"?
15:49 Lajla: rmitt, and I are the rebels man.
15:50 eli: It's pretty simple -- if you enter a number, you get to see a number.
15:50 eli: If you enter a string -- you get to see a string *in* double quotes.
15:50 eli: You don't want to see the string without the double quotes, since that's going to be confusing.
15:50 Lajla: Well, "3" is the external repraesentation of an expression which evaluates to the exact number 3. As is "'3" or "(+ 1 2)" and so forth.
15:50 eli: In exactly the same way, displaying the symbol `a' as just "a" is confusing.
15:51 Lajla: Well, I just quote all things now to indicate that output is still a sequence of characters.
15:51 Lajla: I guess but "a" is reader syntax for a symbol.
15:51 Lajla: And "\"a\"" for a string
15:52 eli: Actually, "a" is reader syntax for an *identifier*, not for a symbol.
15:52 Lajla: At least, what it seems to do to me is print the external repraesentation of the 'simplest' expressoin it can find which evaluates to our result.
15:52 (join) jao
15:53 eli: In any case, that's something that is more common in Scheme than in Lisps -- because in Scheme (most of them, at least) we have more than just plain "symbols".
15:53 Lajla: (symbol? (read (open-input-string "a"))) this evaluates to #t though
15:54 eli: This is because `read' discards some information -- and it happens that when you discard that information from an identifier you're left with a symbol.
15:54 Lajla: At least, I haven't bothered to exhaust this or read this at any point. But as it looks to me now. What it outputs is some thing that if read by the reader and then evaluated, it produces the result of the original expression in the REPL, unless you have a counter-example.
15:55 eli: Cointinuing the above -- see for example scsh/scheme48, which does exactly that quoting thing.
15:55 eli: Also see mit-scheme that does neither, and instead prints ";Value: a".
15:56 Lajla: Well, I have nothing against it per se. I'm just trying to explain to rmitt what it prints out.
15:56 eli: It's roughly that, doesn't work in all cases, obviously.
15:56 Lajla: Which makes me wonder though, why not print (lambda ....) instead of #<procedure>
15:58 eli: Because you still can't print something evaluatable in all cases.
15:59 eli: This is *after* you'll solve the problem of displaying ((lambda (x) (lambda (y) (+ x y))) 3) as (lambda (y) (+ 3 y)).
16:00 Lajla: I guess. This is why I personally would not opt for the quote syntax for symbols, it seems some-what arbitrary.
16:00 Lajla: But I guess that is the nature of human intuition.
16:01 eli: I don't see how what I said is related to displaying quoted symbols.
16:01 Lajla: I mean to in one case (the symbols) do display an expression which evaluates to the result, and not in another case (the functions)
16:02 eli: So how is not printing the quote making things better?
16:03 eli: I'll save you the trouble --
16:03 eli: Yes, the quote-less printout is supposed to be the external representation blah blah blah --
16:03 eli: but even that is an approximation only, since functions *don't* have an external representation.  Ever.
16:04 eli: Same goes for a whole bunch of stuff -- like ports and more,
16:04 eli: so the fact that there is some data without an external representation is still there -- whether you try to make it evaluatable or not.
16:04 Lajla: So you see this simply as 'some string of characters meant to give a human reader an idea of what it is'?
16:05 eli: It would have been arguably cleaner for `write' to just throw an error whenever it runs into a value that cannot be serialized in some way -- but that just didn't happen.
16:05 eli: Yes, exactly.
16:05 Lajla: I guess that would be a nice solution yes. GHC does that with show.
16:06 eli: And it happens that "giving a human reader an idea" is overall less confusing if it's written in the same langugae the human is using -- racket -- rather in some obscure semi-language of "external representations".
16:06 Lajla: Or make things like #<procedure:+> actually readable by the reader depending on the definitions.
16:06 Lajla: I gues you're right.
16:07 eli: That would be impossible.  It inherentrly refers to a name, rather than a function.
16:07 Lajla: I found it confusing at first though, but it isn't and won't be a formal thing I guess.
16:07 eli: But in DrRacket's student languages (at least in the beginner one) this is exactly what happens -- and there, the addition function is printed as ... "+".
16:07 eli: Without quote.
16:08 Lajla: Well, (eval '+) in that definition evaluates to that procedure, the reader could use that, right?
16:08 Lajla: Ahhh
16:08 Lajla: I did not know that.
16:09 eli: Oh, yes, an equivalent replacement for a printout of "+" is "(eval '+)" -- except that those languages will never have `eval' bound.
16:10 Lajla: No, I mean, to enable the reader in a program where + is defined or even lajla to read #<procedure:lajla> as what-ever function the symbol lajla is bound to.
16:14 rmitt: well I never thought I'd provoke a debate :-)
16:14 rmitt: basically I'm asking why we can't stick to R5RS
16:14 rmitt: on this point
16:15 rmitt: More particularly, I once heard a nice challenge to devise a s-expression that "evaluates to itself"
16:15 Lajla: I'm pretty sure that R5RS says very little if any about repls and interfaces of implementations
16:15 rmitt: With the current #lang racket this is no longer a challenge
16:15 rmitt: which is a pity...<sigh>
16:15 Lajla: You mean to the string?
16:15 Lajla: Or a quine?
16:16 rmitt: no, I mean an sexp that evalutes to itself
16:16 (quit) MayDaniel: 
16:16 Lajla: As in, a program whose output is its own source.
16:16 rmitt: yes
16:16 Lajla: And sexp always does that, seeing that they are strings.
16:16 Lajla: Ahh
16:16 rmitt: ??
16:16 Lajla: That's called a quine.
16:16 rmitt: one moment while I look up quine....
16:16 (quit) b-man_: Remote host closed the connection
16:17 rmitt: (I assume it's named after the philosopher)
16:17 Lajla: A quine is simply a computer programme which outputs its own source code.
16:17 Lajla: Yes.
16:18 rmitt: yes, so quines are no longer challenges, which is a pity
16:18 (join) Tyr42
16:19 Lajla: rmitt, well, it's not really output.
16:19 Lajla: It's just the termination value of the program in a way.
16:19 Lajla: Hence racket colours them differently from stuff that came from display and print and write et cetera.
16:21 rmitt: what's weird is that mzscheme does NOT do this funky quote thing, whereas DrRacket does
16:21 rmitt: any reason for that?
16:21 rmitt: (and I certainly hope that this comment won't prod the PLT team to change mzscheme too)
16:22 Lajla: DrRacket is aimed at noobs such at I who are too lazy to use a commandline properly. =)
16:22 rmitt: I don't think that's the real reason, do you?
16:23 rmitt: eli's response was earnest, and not really blase about noobs
16:23 Lajla: I like to think they give me special treatment because of my incompetence.
16:26 rmitt: speaking of which (not), has anyone built a Scheme that runs on the JVM?
16:27 rapacity: kawa scheme
16:27 offby1: yep
16:27 offby1: there's another one too, I think
16:27 offby1: SISC maybe
16:30 eli: Yes, and also jscheme.
16:30 eli: rmitt: It should print the same in r5rs, except for the mutable thing.
16:30 Lajla: rmitt, they have a tendency to not give TCO and some other cool stuff, unless you really really beg them though.
16:30 eli: rmitt: As for `mzscheme' -- that uses the `scheme' language by default, instead of `racket'.
16:30 eli: IIRC, SISC does have tail calls.
16:31 eli: rmitt: And what you're talking about are quines -- and they're stil; the same kind of challenge.
16:32 eli: "Write an expression that evaluates to its own representation as an s-expression" is still the same problem.
16:32 eli: And 'foo is not doing that -- it evaluates to 'foo but the question asks for something that would make it evaluate to ''foo .
16:32 eli: There's a ton of these kind of problems.
16:32 Lajla: I think the idea of a quine is 'whose evaluation has a side-effect of printing that s-expression' though.
16:33 eli: No, there's nothing explicit about printouts.
16:33 eli: Though I can see mainstreamers butchering it into that -- since they have no good representation for syntax.
16:35 rmitt: no quines should not be about side effects
16:35 rmitt: "mainstreamers"?  Is that the official pejorative for folks who hack C, Java, etc?
16:36 Lajla: Blokes without long hair obviously.
16:36 Lajla: Unlike Richard Stallman.
16:36 Lajla: He's underground.
16:36 Lajla: At least under all the dirt in his hair.
16:36 rmitt: ew
16:36 rmitt: btw, is he in Boston?
16:36 rmitt: (I'm full of questions until I finish my bagel)
16:37 eli: rmitt: Yeah.  "People who are high on side-effect drugs."
16:37 eli: For them, making it *return* an expression is meaningless unless they happen to be compiler hackers.
16:38 rmitt: So are mainstreamers the kind to think that quines are impossible (or "undecidable", which is not the right word)? ;-)
16:38 eli: No, just the kind that think that it must be something related to some printout.
16:38 Lajla: Well, it is generally undecidable if a program is a quine.
16:39 eli: Otherwise they can have almost the same kind of fun.
16:39 rmitt: I guess I've forgotten my theory: why is quine-detection considered undecidable when a computer could just simulate the computation?
16:41 Lajla: rmitt, you can't know if it will terminate in advance.
16:41 Lajla: MAybe it will enter an infinite loop.
16:41 Lajla points at Rice's theorem
16:47 offby1 points at Lajla's behind
16:47 Lajla: To be fair, it is where the money's at.
16:50 rmitt: i'm point at my own behind.  It's pleasing, but the folks at Starbucks are looking at me funny.
16:53 Lajla: The people at Starbucks will look at you funny if you drink coffee there.
16:53 Lajla: They are mainstream
16:53 Lajla: we are the underground man.
16:53 Lajla: Like the people in London's subway network.
16:54 Lajla: ain't that right, Eli my brother from another mother?
16:56 Tyr42: Hello
16:56 Lajla: Yo Blair
16:58 Tyr42: I'm having some trouble running amb
16:58 Tyr42: http://planet.racket-lang.org/display.ss?package=amb.plt&owner=murphy
16:58 Tyr42: I'm getting
16:58 Tyr42: abort-current-continuation: continuation includes no prompt with the given tag: #<continuation-prompt-tag:amb>
16:58 Tyr42: when I run the example from the doc
17:01 offby1: you gotta tell us what you're typing
17:01 jonrafkind: he means this http://planet.racket-lang.org/package-source/murphy/amb.plt/1/1/planet-docs/amb/index.html
17:01 (notice) rudybot: http://tinyurl.com/2eb3lca
17:02 jonrafkind: maybe.. Tyr42 that example worked for me
17:02 Tyr42: Oh
17:02 Tyr42: I see what I forgot
17:03 offby1: *shrug* works for me
17:03 Tyr42: (amb-collect (solve-kalotan-puzzle))
17:03 (quit) lisppaste: Quit: Want lisppaste in your channel? Email lisppaste-requests AT common-lisp.net.
17:03 Tyr42: not just (solve-kalotan-puzzle)
17:03 Tyr42: thanks
17:04 (join) lisppaste
17:12 eli: jonrafkind: ping
17:12 jonrafkind: pong
17:13 eli: You don't have any opinions?
17:13 jonrafkind: i briefly read the email but didnt have any strong emotions
17:13 jonrafkind: something about replacing => with more specific arrows?
17:13 eli: There were two important points.
17:14 eli: One of them was the confusion of doing (test (+ 1 3) 5) instead of (test (+ 1 3) => 5).
17:14 eli: This mistake means two passing tests instead of one failure.
17:14 eli: That's the thing about `tests' etc.
17:14 eli: The other issue is that `=>' catches all errors,
17:15 eli: which means that if you do, for example, (define (fact n) (if (< n 1) n (* n (fact (- n 1)))))
17:15 eli: and then (test (fact 5) => (* 5 (fact 4))), you'll get a bogus success.
17:16 eli: Um, wrong error.
17:16 jonrafkind: i think => should awlays be present
17:16 eli: (define (fact n) (if (< n 1) 1 (* n (fact (- n "one")))))
17:16 eli: You'd get a bogus success since both sides are evaluated with catching errors, and the error will be the same.
17:16 jonrafkind: right, if the LHS wasn't supposed to be an error then there shouldnt be an equivalence between the error on the RHS
17:17 eli: Re the first point -- that was Robby's conclusion,
17:17 eli: but it gets in the way of conveniently specifying just expressions to test.
17:17 eli: Eg, (test (fact 5) => 120 (even? (fact 120)))
17:17 jonrafkind: (even? (fact 120) => #t
17:17 jonrafkind: I always use that style
17:18 jonrafkind: I think its better since it makes it extremely explicit whats going on
17:18 eli: What about (memq ...)
17:18 eli: What Robby suggested for that was some pred thing
17:18 jonrafkind: how is memq different
17:18 eli: It doesn't return #t/#f.
17:18 jonrafkind: neither does (+ 1 2)
17:19 eli: So what Robby suggested was (test (even? (fact 120)) =pred> values)
17:19 eli: Either way, that becomes very inconvenient when you're dealing with calls to functions that have more tests.
17:19 eli: As in (test ... (run-some-tests) ...)
17:19 jonrafkind: you mean (test (f)) where f has a (test) expression inside it?
17:19 eli: Yes.
17:20 jonrafkind: oh because it doesnt return anything (or void maybe) ?
17:20 eli: Yes.
17:20 jonrafkind: hm, ok i never use that style
17:20 eli: I (and Jay) use it very extensively.
17:20 eli: "tests/lazy/main.rkt"
17:20 eli: "tests/scribble/main.rkt"
17:20 eli: etc.
17:21 jonrafkind: well im reasonablly happy with the way eli-tester works right now, if it changes drastically ill probably fork it
17:21 jonrafkind: but if you are just adding things that I will never use, thats ok
17:21 eli: Forking will be bad -- it's (roughly) the way testing has been done for ages.
17:21 jonrafkind: i cant remember the last time i got stuck by having a syntax error on both sides and inadvertently had a test pass that should have failed
17:21 eli: Ie, each one writes their own little macro.
17:22 eli: So you did have that happen?
17:22 jonrafkind: I agree, but I like the current interface and if you change it to something I dont like I will probably do something about it
17:22 eli: (That's the other problem, btw.)
17:22 jonrafkind: no ive never had inadvertent errors
17:22 jonrafkind: that i can remember
17:23 eli: In any case, re the first problem -- you won't need to fork it more than (require (rename-in tests/eli-tester [test* test]))
17:23 eli: But the question is what should the default `test' be, in face of that problem.
17:23 jonrafkind: so the one test vs multiple tests?
17:23 eli: Yes.
17:23 eli: Robby's last suggestion was:
17:24 eli: - (tests ...) is like the current `test', except that you *must* write arrows.
17:24 eli: (Making this roughly like what you want)
17:24 eli: - and if you want a single non-arrow test, then you wrap it in a `test'.
17:24 jonrafkind: ok but what about your use-case of (tests (f))
17:24 eli: So: (tests (fact 5) => 120 (test (even? (fact 10))) (test (> (fact 10) (fact 9))))
17:25 jonrafkind: oh ok
17:25 eli: Yes, that's exactly the problem.
17:25 eli: It becomes very inconvenient:
17:25 jonrafkind: so (test (f)) ?
17:25 eli: (tests (test (foo)) (test (bar)))
17:25 jonrafkind: well.. i would say you (eli and jay) should write some macro to make that case more convient
17:25 eli: Or (tests (test (begin (foo) (bar))))
17:25 jonrafkind: (sub-tests (foo) (bar))
17:26 eli: There's no need for that -- since that would be `test*'.
17:26 eli: Again, the question is what's the better default.
17:26 jonrafkind: i vote for robby's behavior
17:26 eli: Considering that people *will* want to write "test suites".
17:27 eli: OK, and what about the second problem then?
17:27 eli: The one where `=>' catches all errors on both sides.
17:27 jonrafkind: i think the default behavior should treat errors specially
17:27 jonrafkind: and if you want to test equivalence between errors then use some other form
17:27 jonrafkind: =error> or whatever
17:27 eli: What do you mean "treat errors specially"?
17:28 jonrafkind: if the LHS produces an error then make `test' signal some error instead of doing (eq? LHS RHS)
17:28 eli: Robby's suggestion there is to *not* catch them, so if there are errors, you will get the errors.
17:28 jonrafkind: or whatever it does
17:28 jonrafkind: uh yea ok that too
17:28 jonrafkind: just pass them on
17:28 jonrafkind: ok i guess the current behavior has a with-handlers, so just get rid of that
17:28 eli: Then there would be `=error>' (there already is that) which will catch runtime errors.
17:29 eli: But even that won't catch syntax errors -- that will be a new `=stxerror>' arrow.
17:29 eli: (Possibly different name.)
17:29 jonrafkind: yea ok
17:29 (part) Tyr42
17:30 eli: "Yea" == "I like it", or == "whatever"?
17:30 jonrafkind: "i like it -- but ill probably never use it so whatever"
17:30 eli: You never test errors??
17:30 jonrafkind: no
17:31 eli: Weird.
17:31 jonrafkind: errors dont exist, what are you a bad programmer or something?
17:31 jonrafkind: we dont need no steenking type system!
17:31 eli: Anyway, do you mind sharing your opinion on both of these on the thread?
17:31 jonrafkind: ok
17:32 jonrafkind: btw about the ppa thing. i had some insane problems setting it up, but long story short ill fix it soon
17:32 jonrafkind: basically i tried to test the build in a virtual machine and got a ton of impossible to debug errors, so ill switch virtual machine technologies to something that actually works
17:32 eli: ppa?
17:32 jonrafkind: ubuntu
17:33 eli: Oh, that.
17:45 (quit) jonrafkind: Ping timeout: 252 seconds
18:29 (quit) rmitt: Ping timeout: 265 seconds
18:47 (join) b-man_
19:12 (quit) mceier: Quit: leaving
19:18 (join) kingless
19:23 (quit) kingless: Remote host closed the connection
20:05 (quit) masm: Quit: Leaving.
20:12 (join) rmitt
20:32 (join) danbrown
21:21 (quit) rmitt: Quit: Page closed
21:43 (quit) b-man_: Remote host closed the connection
21:56 (quit) danbrown: Quit: danbrown
21:56 (join) danbrown
21:58 (join) jeapostrophe
22:07 (quit) danbrown: Quit: danbrown