#clojure log - Jun 27 2011

The Joy of Clojure
Main Clojure site
Google Group
IRC
List of all logged dates

0:30 davekong: I have a function that is working fine in the REPL, but if the pcall version of it is called from a java class, it hangs after printing the result. Also, from java, whether calling the parallel version or not, it gives grossly inacurrate results, despite all of the number manipulation being done in clojure (the function is just passed arrays of strings). Any hints on how to fix these issues? Should I

0:30 paste my code?

0:34 amalloy: davekong: ##(doc shutdown-agents)

0:34 sexpbot: ⟹ "([]); Initiates a shutdown of the thread pools that back the agent system. Running actions will complete, but no new actions will be accepted"

0:35 amalloy: probably why it's "hanging". no ideas about the other thing

0:43 davekong: amalloy: thanks that seems to have fixed the first issue

0:45 amalloy: don't do that unless you want to shut down the system, though

0:45 davekong: oh

0:47 well I don't want to until much later except now while I am testing

0:48 I am doing some calculations in clojure but the main program is on java

0:50 Or you just mean shut down all the agents?

0:54 seems like half the time I get a bad result and half the time it is good

0:56 and it has been regular for several runs good, bad, good, bad

1:00 symbole: Why does (defn foo [&{:keys...}] work? Isn't everything after & collected into a list, and you can't destructure a list using {:keys...}?

1:06 davekong: Would java be less safe with memory or something like that and maybe memory is getting corrupted?

3:00 amalloy: ~source destructure

3:00 symbole: it works because the above makes it work

3:00 (i haven't looked actually, but where else can the logic be)

3:17 for the curious, it looks like line 3434 there says, "if you destructure a seq into a map, apply hash-map on it first"

3:20 &(let [{:keys [test]} '(:test 1)] test)

3:20 sexpbot: ⟹ 1

5:48 rfgpfeiffer: clojurebot: seen drcode

5:48 clojurebot: Pardon?

5:49 rfgpfeiffer: ~seen drcode

5:49 clojurebot: Excuse me?

7:32 tomoj: how does this make sense? https://gist.github.com/d24bf93fb851b9835305

7:32 I realize it's silly to create a bunch of new Randoms

7:33 but still, shouldn't I sometimes see more positive than negative?

7:33 every time I look for positives I get less positives than negatives

7:33 less positives than nonpositives I should say

7:34 every time I look for negatives I get fewer negatives than nonnegatives

7:34 yet every time I look for zeros I get zero...

7:37 if I instead use just one Random, the results make sense

7:37 hah

7:55 (repeateadly #(.nextLong (java.util.Random.))) is a strange thing

7:56 s/ead/ed/

7:57 raek: tomoj: I think that when a Random is instantiated the random seed is initialized with the current time

7:57 tomoj: so if you instantiate a lot of them in the same millisecond/microsecond, the will give the same values

7:57 *they will

7:59 tomoj: strange thing is, I can't ever see any dups

8:01 but since the behavior depends on how you consume the seq...

8:03 raek: ,(java.util.Random.)

8:03 clojurebot: #<Random java.util.Random@1746634>

8:03 tomoj: I think there's just some weird dependency where if you take longer to consume positive values, and so delay generation of Randoms a bit more for positive values than negative values, fewer positive values are produced in general

8:03 s/in general//

8:04 raek: ,(let [f #(.nextInt (java.util.Random.))] (loop [l (), n 10] (if (zero? n) l (recur (cons (f) l) (dec n)))))

8:04 clojurebot: (1135657559 -1653233976 1129285394 1624057155 -285451636 110824407 -1401623519 -906851758 1078467891 -246900125)

8:06 raek: ,(map #(.nextInt %) (loop [l (), n 10] (if (zero? n) (recur (cons (java.util.Random) l) (dec n)))))

8:06 clojurebot: ()

8:06 raek: ,(map #(.nextInt %) (loop [l (), n 10] (if (zero? n) l (recur (cons (java.util.Random) l) (dec n)))))

8:06 clojurebot: java.lang.ClassCastException: java.lang.Class cannot be cast to clojure.lang.IFn

8:06 raek: ,(map #(.nextInt %) (loop [l (), n 10] (if (zero? n) l (recur (cons (java.util.Random.) l) (dec n)))))

8:06 clojurebot: (1079184460 -905735116 -412502350 -735306677 -337491639 2058739835 1328101674 1724377717 608990656 386974692)

8:07 raek: ok, I can't seem to construct a example of what I'm thinking about... :-)

8:08 rpglover64: interop question

8:10 I (by which I mean my friend's students) have been given a jar, with the assignment that they should write program logic conforming to the api provided by the jar in java; I'd like to try to do this in clojure.

8:10 suggestions on how to start?

8:18 gfrlog: rpglover64: can you be more specific? like, are you already familiar with clojure's interop features, or do you need help with those?

8:19 (e.g., have you already read http://clojure.org/java_interop ?)

8:20 rpglover64: no, I'm not

8:20 in fact, I'm not familiar with java classloaders

8:20 I have read that

8:21 and my tripup right now seems to be getting access to the java api from within clojure

8:21 gfrlog: rpglover64: do you have the jar on the classpath? have you tried importing the relevant classes?

8:22 rpglover64: ah, the answer is simple. I'm an idiot who tries importing packages as if they were classes

8:22 gfrlog: :)

8:22 rpglover64: thank you

8:23 gfrlog: np

8:52 I wish (for) had an implicit do :(

8:53 rpglover64: can't you write a macro for that?

8:53 (I know I can't, but shouldn't it not be too difficult?)

9:00 terom: gfrlog`: what do you mean? doseq has similar bindings as for

9:00 gfrlog`: terom: does it return a seq?

9:00 ,(doseq [x (range 5)] (inc x))

9:01 clojurebot: Execution Timed Out

9:01 gfrlog`: ??

9:01 is it too early for writing code?

9:01 my repl says it returns nil

9:01 so doseq would not be a good substitute

9:01 terom: Maybe... doseq returns nil, it's for side effects.

9:02 gfrlog`: terom: right. I want to use for primarily, but adding debugging statements is a pain

9:02 I cannot say (for [x (range 10)] (prn x) (inc x))

9:04 terom: Well.. you could say (do (prn x) (inc x)), but the results can be a bit suprising.

9:04 gfrlog`: terom: that's what I do. I know that for is lazy.

9:04 When my lazy seq crashes in some completely other part of the code because of something in the body of the for, adding the (prn) there can usually help

9:09 rpglover64: looks like you've run into a problem common in haskell

9:31 tomoj: &(= (Long/toBinaryString -9223372036854775808) (Long/toBinaryString 27670116110564327424))

9:31 sexpbot: java.lang.Exception: EvalReader not allowed when *read-eval* is false.

9:32 tomoj: oh, unsigned..

9:32 wait..

9:33 ,(long 27670116110564327424)

9:33 clojurebot: java.lang.ExceptionInInitializerError

9:34 gfrlog`: tomoj: I've noticed the bots don't like reading big numbers

9:34 ,2823589289358923847289348923523423423

9:34 clojurebot: 2823589289358923847289348923523423423

9:34 gfrlog`: okay nm I don't know what I'm talking about

9:35 ,(first (range 8923895829348929835982389498234923))

9:35 clojurebot: java.lang.ExceptionInInitializerError

9:44 mattmitchell: anyone know of a good validation lib? I want to do things like, make sure input is of a certain type, provide defaults, provide useful error messages etc.. This will be used in a library full of search finder functions

9:52 jcromartie: -> and update-in are just about the best things ever

9:52 next to Nutella

9:56 gfrlog`: mattmitchell: I know clojure provides function preconditions built in

9:56 mattmitchell: but that's probably not as extensive as what you want

9:58 mattmitchell: gfrlog`: yeah I want to play with the preconditions a bit. The simpler the better for sure.

10:00 gfrlog`: mattmitchell: the defaults you could probably handle with a macro pretty simply

10:01 I think clojure.core had a private method that it used for declaring constraints for inputs and reporting errors

10:01 wastrel: hi i stopped learning clojure because of busy

10:02 mattmitchell: gfrlog`: interesting, that'd be a good thing to have

10:02 gfrlog`: could you show me an example of using a macro for defaults?

10:02 gfrlog`: sure, one sec

10:02 the clojure.core macro is here: https://github.com/richhickey/clojure/blob/master/src/clj/clojure/core.clj#L1383

10:03 I can't guarantee that one's a good idea though; some of the code in core is primitive

10:04 mattmitchell: what kind of default behavior do you want? do you just want nil args to be converted to something else, or do you want the function to be able to accept any number of args?

10:06 mattmitchell: gfrlog`: well for example, if I have a param called :sort, I want to make sure it is an array, and contains valid sort fields: [:score.dec :price.asc]. If the values aren't valid, then fallback to a default like [:score.desc]. But it'd be great to throw an exception if needed too.

10:09 gfrlog`: mattmitchell: so for each arg you want to define what valid means and you want to provide an alternative for invalid inputs?

10:09 or you want to throw an exception? :/

10:10 mattmitchell: gfrlog`: yes, the former. The exception stuff isn't as important, thanks :)

10:15 edw: I'm consistently getting a "No value supplied for key: <PORTNUM>" exception when using M-x clojure-jack-in. Anyone else run into this? This is new to me; I've been using jack-in for a while now.

10:16 gfrlog`: mattmitchell: try this and tell me how it breaks :) https://gist.github.com/1048931

10:17 edw: Hmm, seems related to incanter-1.2.3... Hmm.

10:19 pyr: whenever timers are brought up

10:19 concurrency issues are mentionned

10:20 raek: edw: iirc, incanter has an old version of swank-clojure as a dev-dep

10:25 mattmitchell: gfrlog`: thanks!

10:25 gfrlog`: will check it now

10:25 gfrlog`: yepps

10:31 mattmitchell: gfrlog`: OK running your example I get "Can't let qualified name: user/arg-nam"

10:31 sorry, that last part was "arg-name"

10:31 gfrlog`: mattmitchell: that sounds plausible. Kay I'll put it in my repl and beat out the problems :)

10:31 mattmitchell: gfrlog`: cool thanks!

10:32 gfrlog`: mattmitchell: you see what it's trying to do though right?

10:32 mattmitchell: gfrlog`: yes i think i understand the idea there and how it works. I'm not real experienced with macros though, so the quoting throws me off a bit

10:32 gfrlog`: yeah it's a bit weird

10:35 mattmitchell: okay, I updated it. Seems to work now.

10:35 mattmitchell: awesome i'll check it

10:36 gfrlog`: mattmitchell: using macroexpand probably makes it more explicit what it's doing

10:36 mattmitchell: gfrlog`: good idea

10:37 gfrlog`: mattmitchell: maybe renaming it let-defaults would make more sense :/ dunno

10:37 or....whatever you want

10:37 mattmitchell: ok sure yeah. i'm going to play around with this. thanks for taking the time to write this up :)

10:38 gfrlog`: no problem

10:55 mattmitchell: is there a predicate for testing if something is an array-like type?

10:56 Or is it just (or (vector? x) (list? x) ?

10:57 pcavs_: mattmitchell: (seq? my-list) I believe

10:57 mattmitchell: that's what i thought too, but then:

10:57 terom: seq? returns false for vectors

10:58 mattmitchell: ,(seq? [1])

10:58 clojurebot: false

10:58 mattmitchell: yeah

10:58 pcavs_: well it depends what you're testing

10:58 ,(seq? (list 1 2 3))

10:58 clojurebot: true

10:58 pcavs_: is because it implements (head my-list) and (rest my-list)

10:58 ,(head (list 1 2 3))

10:58 clojurebot: java.lang.Exception: Unable to resolve symbol: head in this context

10:59 pcavs_: doh

10:59 terom: first

10:59 pcavs_: w/e it's called in clojure

10:59 there we go

10:59 ,(first (list 1 2 3))

10:59 clojurebot: 1

10:59 pcavs_: ,(rest (list 1 2 3))

10:59 clojurebot: (2 3)

10:59 pcavs_: ,(type [1])

10:59 clojurebot: clojure.lang.PersistentVector

11:00 mattmitchell: ,(first [1])

11:00 clojurebot: 1

11:00 mattmitchell: ,(rest [1 2 3])

11:00 clojurebot: (2 3)

11:00 pcavs_: mattmitchell: now I'm confused myself... =\

11:01 mattmitchell: ok :). yeah I was totally expecting seq? to return true for vectors

11:01 terom: first and rest probably call seq on the vector

11:03 bsteuber: what you want is sequential?

11:03 ,(sequential? [1 2 3])

11:03 clojurebot: true

11:04 bsteuber: ,(sequential? (list 1 2 3))

11:04 clojurebot: true

11:04 bsteuber: ,(sequential? {:a 4})

11:04 clojurebot: false

11:07 rfgpfeiffer: ,(doc sequential?)

11:07 clojurebot: "([coll]); Returns true if coll implements Sequential"

11:08 mattmitchell: ,(sequential? "test")

11:08 clojurebot: false

11:08 mattmitchell: good

11:10 rfgpfeiffer: ,(seqable? ())

11:10 clojurebot: java.lang.Exception: Unable to resolve symbol: seqable? in this context

11:10 raek: re. seq vs seqable, I recommend this post: http://kotka.de/blog/2011/06/On_types.html

11:10 pcavs_: raek: thanks

11:11 bsteuber: ,(clojure.contrib.core/seqble? "foo")

11:11 clojurebot: java.lang.ClassNotFoundException: clojure.contrib.core

11:12 raek: when the ISeq is eventually protocolized, checking for "seqable" would be more intuitive

11:12 rfgpfeiffer: is there a way to refer to array types in multimethod signatures?

11:13 raek: rfgpfeiffer: what do you mean by "multimethod signatures"? dispatch value? type hints?

11:13 bsteuber: rfgpfeiffer: clojure.lang.Sequential

11:13 rfgpfeiffer: dispatch value

11:14 I can write (def bytes (type (byte-array 0)))

11:14 and use that as a dispatch value

11:14 raek: I think the easiest way is to create an instance of an array with the correct type and then store the (class the-array) in a var

11:15 I've seen this in clojure core code, I think

11:15 rfgpfeiffer: ,(type (byte-array 0))

11:15 clojurebot: [B

11:16 raek: ,(resolve '[B)

11:16 clojurebot: Unmatched delimiter: )

11:16 raek: ,(resolve (symbol "[B"))

11:16 clojurebot: [B

11:16 raek: ,(Class/forName "[B")

11:16 clojurebot: [B

11:16 raek: ,(Class/forName "[Invalid")

11:16 clojurebot: java.lang.ClassNotFoundException: [Invalid

11:17 raek: hrm, this might work too

11:17 rfgpfeiffer: ,(= (Class/forName "[B") (type (byte-array 0)))

11:17 clojurebot: true

11:17 rfgpfeiffer: interesting

11:19 raek: mattmitchell: for macros, one idea is to define cases for seq? (lists), vector?, map?, set?, and everything else.

11:20 if you want to define new meaning for clojure source code data structures

11:26 sritchie: hey all -- is there some way to get metadata at read-time, for a function passed in to another function as an argument?

11:27 (defn get-meta [x] (retrieve-meta x))

11:27 which would just be retrieve-meta, of course

11:28 oh, actually, better question -- can function values hold metadata?

11:28 pdk: the return values or the functions themselves

11:29 also (meta x) returns a map with the metadata of x

11:29 sritchie: the functions themselves

11:29 gfrlog`: ,(vector (meta cons) (meta (var cons)))

11:29 clojurebot: [{:line 28} {:ns #<Namespace clojure.core>, :name cons, :file "clojure/core.clj", :line 22, :arglists ([x seq]), :doc "Returns a new seq where x is the first element and seq is\n the rest.", :added "1.0"}]

11:29 pdk: functions can carry metadata

11:30 sritchie: pdk: cool, that's perfect

11:30 thanks

11:30 gfrlog`: I'm curious why one is :line 28 and the other is :line 22

11:30 the function seems to be about six lines long...

11:31 rpglover64: how is proxy expected to be used (there seems to be a dearth of examples); I'm slightly confused as to what would be the equivalent of (in java) writing a class which implements an interface.

11:32 gfrlog`: rpglover64: I think you can implement java interfaces with defrecord

11:36 raek: rpglover64: a typical usage is to interface with swing, e.g. you might need to implement ActionListener

11:37 http://www.dreamincode.net/code/snippet3252.htm

11:37 (example by Raynes)

11:38 rpglover64: ah; illuminating

13:10 derp__: I want to try and make a simple project using lein, but I don't really know where to go after creating my main function in core.clj

13:10 would anyone recommend a small example project using lein for me to look at?

13:10 technomancy: clojurebot: google mire

13:10 clojurebot: First, out of 4660000 results is:

13:10 Mire | Define Mire at Dictionary.com

13:10 http://dictionary.reference.com/browse/mire

13:10 technomancy: oops

13:10 clojurebot: google mire clojure

13:10 clojurebot: First, out of 103 results is:

13:10 in which slime and mire and other sludges come into play - Technomancy

13:10 http://technomancy.us/122

13:12 technomancy: derp__: lame link, but whatever; what I meant to say is that I built mire as a sample type app

13:12 derp__: technomancy: thank you, this looks like a great place to start

13:12 wastrel: is clojurebot yours ?

13:12 technomancy: ^^

13:13 technomancy: wastrel: I convinced hiredman to implement it.

13:14 gfrlog`: clojurebot belongs to whoever feeds him the most

13:15 wastrel: botsnack?

13:15 !botsnack

13:15 gfrlog`: ~botsnack

13:15 clojurebot does not belong to me

13:15 raek: clojurebot: have some botsnack

13:15 ejackson: $botsnack

13:15 sexpbot: ejackson: Thanks! Om nom nom!!

13:15 justinlilly: ,botsnack

13:15 ejackson: haha !

13:16 gfrlog`: ,(prn "botsnack")

13:17 pjstadig`: ~ping

13:17 manutter: clojurebot: say something foreign...

13:17 gfrlog`: this is what happens when you forget to love your bot

13:17 manutter: sigh, he died

13:18 gfrlog`: manutter: is that the one phrase that he will absolutely always respond to unless he is completely dead?

13:18 hiredman: I suspended the screen it runs it

13:19 manutter: gfrlog`: I dunno, whenever I try to have a conversation clojurebot always says somthing foreign to me

13:19 gfrlog`: oh right

13:20 derp__: I'm trying to start using slime/swank, but for some reason this tutorial isn't helping me: http://data-sorcery.org/2009/12/20/getting-started/

13:20 specifically launching "lein swank" in the mire directory isn't working, I get "That's not a task. Use "lein help" to list all tasks."

13:20 manutter: clojurebot: welcome back.

13:20 technomancy: aquamacs? =\

13:20 clojurebot: Titim gan éirí ort.

13:20 derp__: nope

13:21 emacs23 on 11.04

13:21 gfrlog`: ~botsnack

13:21 clojurebot: Thanks! Can I have chocolate next time

13:21 technomancy: derp__: https://github.com/technomancy/swank-clojure#readme

13:21 raek: derp__: try "lein plugin install swank-clojure 1.3.1"

13:25 derp__: okay, I tried that command and lein swank, but I got this error: https://gist.github.com/1049322

13:25 technomancy: derp__: see "known issues" in the readme

13:27 derp__: technomancy: okay, I removed the swank requirement in mire, but now I get this error

13:27 https://gist.github.com/1049325

13:27 woops

13:30 technomancy: I assume you were referring to the :dev dependencies bug, just to be clear

13:30 technomancy: right

13:31 derp__: but this new error, is that related to my own mistake or an actual bug?

13:33 technomancy: your last gist is also caused by having more than one copy of swank installed

13:35 derp__: uh, I guess I had it through elpa (totally forgot about that). when I do M-x package-list-packages, what is the command to delete/remove a highlighted package?

13:36 technomancy: not sure; rm -rf ~/.emacs.d/elpa/foo works for me

13:37 but that problem is caused by the swank-clojure lein plugin, not the swank-clojure elisp package

13:37 derp__: that okay to do? emacs isn't going to throw a fit about a package being removed unceremoniously or whatever?

13:38 technomancy: you should definitely get rid of the swank-clojure elisp package

13:38 it's been deprecated for over a year

13:39 if I could delete it off elpa I would

13:52 derp__: now I am getting "symbols value as variable is void: slime-clj"

13:52 but lein swank works, so I'm halfway there

13:53 technomancy: never heard of that one before

13:56 derp__: I have clojure-mode (1.9.1), slime (20100404.1), slime-clj (0.1.6), slime-fuzzy (20100404), slime-repl (20100404) installed

13:56 technomancy: I have never heard of slime-clj before

13:57 hiredman: looks like it comes from swank-clj

13:59 hugod: slime-clj is from swank-clj

14:00 derp__: ooh, now I think I got it almost working

14:00 it says Host: 127.0.0.1

14:00 but the mode-bar thingy hasn't changed

14:01 now it just has (Lisp Interaction ElDoc Paredit AC yas Fill)

14:02 raek: derp__: note that swank-clojure and swank-clj are different projects

14:03 davekong: If I have a function that takes an argument of type {[""] <double>} what would be the type I would pass from java to call it?

14:03 hugod: I think I should rename swank-clj to something completely different…

14:03 raek: clojure-mode, slime and slime-repl are the elisp packages you would use with swank-clojure

14:04 derp__: http://dev.clojure.org/display/doc/Getting+Started+with+Emacs

14:05 amalloy: davekong: huh? what type is that?

14:05 a map from vector of string to double?

14:07 just pass it an IPersistentMap, or a j.u.Map if your java code is lazy

14:08 davekong: amalloy: yea

14:23 mattmitchell: is it possible to do nested destructuring for function args?

14:25 raek: mattmitchell: yes

14:26 each parameter can be destructured with the usual rules

14:27 but the whole "paramter vector" is not a destructuring form, so you can't do something like (defn foo [a b c :as all] ...)

14:27 amalloy: &((fn [{[x] :values}] x) {:values [1 2 3 4]}) ;; mattmitchell

14:27 sexpbot: ⟹ 1

14:27 fmw`: I'm trying to add a test to https://github.com/clojure/tools.logging, but it doesn't seem to be run when I execute mvm install or mvn test. I can't find where the test functions are called either. any suggestions?

14:28 mattmitchell: ok thanks. I'll play around with that for a bit

14:42 wastrel: do you use emacs GUI or cli?

14:44 sritchie: gui

14:47 amalloy: even the cli is pretty gooey

14:48 gfrlog`: :-|

14:49 * gfrlog` waits for amalloy to apologize

14:52 sritchie: on OS X, I like using command as the meta key

14:52 and terminal likes escape

14:52 (though this sounds like a fixable issue :))

14:59 hugod: fmw`: you could try mvn clojure:test if it is using the clojure-maven-plugin

15:53 rando: I'm building a project using leiningen and I'm trying to add the S3 wagon following the steps here: http://blog.anzix.net/2010/12/07/using-amazon-s3-as-a-maven-repository/ How can I add the extension to my pom.xml without lein clobbering it?

16:13 symbole: rando: You might be able to include that in your settings files.

16:14 rando: symbole: Yeah, i was trying that, but there's no place to put build/extensions in settings.xml that it won't yell at me

16:16 symbole: rando: It probably goes inside a profile.

16:17 rando: symbole: any idea where? http://maven.apache.org/ref/2.2.1/maven-settings/settings.html#class_profile

16:18 symbole: rando: http://maven.apache.org/guides/mini/guide-building-for-different-environments.html

16:18 rando: oh, look at that

16:20 symbole: nope, still yells: Caused by: org.codehaus.plexus.util.xml.pull.XmlPullParserException: Unrecognised tag: 'build'

16:21 thats annoying that the docs say its possible

16:21 derp__: so in emacs, what's the key combo to go to a function definition?

16:21 amalloy: M-.

16:21 in slime, anyway

16:22 rando: symbole: yeah, that example is for the pom.xml

16:23 derp__: okay when I type M-. I get "Find tag (default server)" and I have slime/swank talking to each other I think

16:23 but I don't know where this tag file is or how to generate it

16:24 amalloy: then slime and swank aren't talking to each other

16:24 if slime is going in the current buffer then M-. doesn't try to do tags stuff

16:26 scgilardi: for OS X, I use https://github.com/aim-stuff/cmd-key-happy as part of a setup to allow my cmd key to act as a meta key with terminal-based emacs.

16:26 derp__: okay, now it's working

16:26 so how do I do the opposite of M-. and go backwards?

16:27 scgilardi: M-, (or meta *)

16:33 derp__: when would one use defn-?

16:35 amalloy: mad cow disease. (an overstatement, but generally don't use it)

16:35 dnolen: derp__: I see it as a way to control which definitions get imported w/ use. Better I think to lift private stuff into a separate ns I think.

16:36 Somelauw: defn is private so that it won't conflict with methods of other modules and namespaces

16:36 and reduces complexity

16:36 symbole: rando: Maybe this will help http://maven.apache.org/guides/mini/guide-wagon-providers.html

16:37 Somelauw: I think

16:43 rando: symbole: I'll give it a shot

17:21 sjl: Ugh, this "no unsigned bytes datatype" is really annoying.

17:28 derp__: what does the # sign do?

17:28 amalloy: like a hundred things

17:28 derp__: I tried (doc #) but got a weird error from slime

17:28 well what's it called?

17:28 if I want to google it

17:29 sritchie: the dispatch macro

17:29 amalloy: derp__: it's a "reader macro": based on the next character in the file, it reads forms specially

17:29 sritchie: search for that here: http://clojure.org/readera

17:29 http://clojure.org/reader

17:29 amalloy: eg, {a b} is a map, but #{a b} is a set

17:31 derp__: okay, thank you

17:35 Cozey: hello. what's the best way to return json under compojure?

17:36 middleware or overriding render ?

17:36 or .. ?

17:38 amalloy: uhhhh?

17:39 just return a string, possibly by using a middleware that converts your map to json, and add a middleware that adds a content-type header

17:39 sjl: What's the best way to say: "split this byte into (int top-four-bits) and (int bottom-four-bits)"? I must be missing something simple...

17:41 sritchie: (juxt #(bit-and % 0x1100) #(bit-and % 0x0011) )

17:42 sorry, missing some two ones and zeros from each

17:42 s/some//

17:42 sexpbot: <sritchie> sorry, missing two ones and zeros from each

17:42 sjl: hmm

17:43 sritchie: don't trust that guess, actually, I think something's wrong there

17:43 amalloy: sritchie: the problem that hex isn't binary, i think :P

17:44 &(let [b 137] (map #(bit-and b %) [0xf0 0x0f]))

17:44 sexpbot: ⟹ (128 9)

17:45 sritchie: amalloy: haha, yes, there we go

17:45 sjl: sritchie: the tricky part is that (byte 137) doesn't work :\

17:45 amalloy: &((juxt quot rem) 137 0x10)

17:45 sexpbot: ⟹ [8 9]

17:46 sritchie: sjl: amalloy's got you covered -- bit-masking the number will wipe out the first two and last two bits, giving you back the ints you wanted

17:46 amalloy: sritchie: only if he has an int, though - it won't work with a byte

17:47 sritchie: true. back to observation mode for me :)

17:48 sjl: Grrrr (.toString (BigInteger. (byte-array [(byte -122)])) 2) -> "-1111010"

17:48 amalloy: sjl: working with unsigned numbers in java is terrible, yes

17:48 sjl: Is there a "just show me the damn bits you're using to represent this" function somewhere?

17:49 amalloy: &(.toString (BigInteger. (byte-array [(byte 0) (byte -122)])) 2)

17:49 sexpbot: ⟹ "10000110"

17:49 amalloy: sjl: not really. a number of libraries have implemented it

17:49 sjl: Hmm, ok, that's a start. Thanks

17:50 Stupid Minecraft protocol and its using nibbles for data.

17:50 amalloy: *chuckle*

17:50 justinlilly: wait what? They have a protocol which uses the nibbles game?

17:51 seems really interesting to transport information as 100x100 grids of a nibbles game in progress :)

17:53 fmw`: hugod: Sorry, got interrupted by a long skype conversation with the girlfriend. I've added a test function, but it doesn't seem to be run, however the other 21 functions are run, so the tests themselves are running fine

17:57 hugod: but it seems the test is within a comment function, so the current tests aren't all being run either

17:57 hugod: so sorry for bothering you ;)

18:01 hugod: fmw`: no bother :) mvn test should work too, looking at the parent pom

18:29 DerGuteMoritz: hello everyone, I am trying to install http://search.maven.org/#artifactdetails%7Cjavax.ws.rs%7Cjsr311-api%7C1.1.1%7Cjar via leiningen like this: lein install javax.ws.rs/jsr311-api 1.1.1

18:29 however, I get this error: Exception in thread "main" java.lang.NullPointerException (NO_SOURCE_FILE:0)

18:29 am I doing something wrong?

18:31 dnolen: sjl: https://gist.github.com/1050007

18:31 DerGuteMoritz: hm putting it into a project.clj and running lein deps works though

18:32 dnolen: sjl: not sure if you need something perf conscious, but if all you want to do is introspect this should be fine.

18:32 sjl: dnolen: Ohh, nice. But is that extremely slow compared to magical bit shifting?

18:32 dnolen: sjl: if you want something fast you need to bit-and then shift.

18:33 (bit-and n 0xf0) and (bit-and n 0x0f)

18:34 I'm sure there are faster ways, I'm not a bit-twiddling expert

18:35 sjl: hmm, (defn top [b] (bit-shift-right (bit-and b 0xf0) 4))

18:38 dnolen: sjl: yeah that's about 2 orders of magnitude faster.

18:40 sjl: yeah, that's much more workable

18:41 Raynes: DerGuteMoritz: lein and cake's install features are meant to install a jar created from a Clojure project's project.clj file.

18:41 sjl: thanks!

18:41 dnolen: sjl: np

18:43 DerGuteMoritz: Raynes: I see - that's a little odd as at least on my installation the maven.org repository is included by default and it does indeed work through lein deps. cljr install javax.ws.rs/jsr311-api 1.1.1 also works though.

18:44 Raynes: DerGuteMoritz: I was wrong anyway. The lein install task works a little differently than the cake one now.

18:44 So, you have reason to complain, methinks. :p

18:45 seancorfield: anyone using clojure.contrib.sql with SQL server or PostgreSQL?

18:45 DerGuteMoritz: Raynes: I was just surprised by that, not really complaining :-)

18:45 seancorfield: i just pushed a fix for clojure.java.jdbc for generated keys which needs testing against those DBs... 0.0.3-SNAPSHOT

18:45 DerGuteMoritz: the reason why I wanted that package in the first place is more complaint-worthy

18:47 dnolen: sjl: yeah the issue is that a bit-shift-right on a negative number fills w/ 1

18:47 sjl: yep

18:47 technomancy: DerGuteMoritz: thanks for the bug report; just fixed it

18:47 I don't think anyone's ever tried to use that task on a non-clojure project

18:47 DerGuteMoritz: technomancy: oh, it was an actual bug? you're welcome then! :-)

18:48 ok looks like I am way off common territory here so I might as well ask for advice, maybe I am doing something wrong

18:48 Raynes: technomancy: Man, you should really let me know when you add functionality like that so that I can do the same for cake. Stop being competitive! <3

18:48 DerGuteMoritz: I'd like to construct an URI (well, URL to be more precise) from a given one

18:49 turns out that java.net.URI and URL are not quite up to that

18:49 technomancy: Raynes: changelog's that way =)

18:49 Raynes: I'm allergic to changelogs.

18:49 DerGuteMoritz: so I found out about javax.ws.rs.core.UriBuilder

18:50 which, judging from the API, looks quite promising

18:50 first question!

18:51 how do I know from that reference page (which looks kind of official to me) that it is indeed not part of the standard Java distribution? http://download.oracle.com/javaee/6/api/javax/ws/rs/core/package-tree.html

18:51 or is my JDK version too old or something?

18:51 technomancy: DerGuteMoritz: IIUC "javax" means "not part of the JDK"

18:51 kind of like clojure.contrib

18:51 DerGuteMoritz: a-ha!

18:51 * DerGuteMoritz makes mental note

18:51 seancorfield: btw, if anyone does want to test that, they'd need to pull from sonatype snapshots :)

18:52 DerGuteMoritz: thanks technomancy

18:52 so has anyone ever used UriBuilder?

18:53 carllerche: technomancy: Any idea how I might be able to have lein deploy use an s3 wagon? It seems that it requires setting build extensions in the pom file

18:54 or is that something that would require writing a plugin for?

18:54 DerGuteMoritz: I am new to Clojure but the reference page has this example at the beginning: UriBuilder.fromPath("{arg1}").build("foo#bar"), which I converted into this hopefully idiomatic Clojure call: (. (UriBuilder/fromPath "{arg1}") build "foo#bar")

18:54 technomancy: carllerche: hmm... it could probably go in a plugin and/or future lein version

18:54 DerGuteMoritz: this results in this interesting exception: java.lang.ClassNotFoundException: com.sun.ws.rs.ext.RuntimeDelegateImpl

18:55 technomancy: carllerche: pom entries usually correspond to pretty straightforward calls on the Model object, so often it's easy to translate to the Java API

18:55 rando: technomancy, carllerche: specifically, the pom entry in step 1 of this: http://blog.anzix.net/2010/12/07/using-amazon-s3-as-a-maven-repository/

18:56 DerGuteMoritz: does that mean that the jsr311-api jar is missing a dependency perhaps?

18:56 * technomancy luls at <version>3.0.0.RELEASE</version>

18:57 amalloy: technomancy: that's not accurate. javax.swing.*, for example, is part of the jdk

18:57 DerGuteMoritz: ; just in case

18:57 hehe

18:57 !

18:57 technomancy: amalloy: huh; didn't know that.

18:57 DerGuteMoritz: then what is javax? hm

18:57 Java + Chemical X

18:57 Raynes: It means the developers got sick of typing 'java' and wanted to spice things up a bit.

18:58 amalloy: i think it's like..."stuff that not every deployment platform is required to have"? or "less stable"? i dunno

18:58 DerGuteMoritz: searching the tubes for com.sun.ws.rs.ext.RuntimeDelegateImpl yields all kinds of results referencing something named "Jersey"

18:59 Bronsa: http://mindprod.com/jgloss/javax.html

18:59 carllerche: technomancy: Alright, I will take a stab at writing a plugi

19:00 technomancy: carllerche: start a mailing list thread if you get stuck

19:00 carllerche: cool, will do, thanks

19:00 technomancy: will probably have 1.6 released in a few days, otherwise I'd consider putting it in leiningen itself

19:00 DerGuteMoritz: Bronsa: thanks! why does that page contain a link to file:///J:/mindprod/jgloss/javax.html?

19:00 Bronsa: DerGuteMoritz: i have no idea.

19:00 just found it on google

19:01 DerGuteMoritz: oh it seems to be some kind of offline cache

19:01 carllerche: ha, well, i'll see if i can get it done tonight :P I have yet to dig into maven though... so we'll see

19:01 technomancy: if you want to do it in a fork of lein rather than a plugin that could work too; less messing with hooks that way

19:02 because that really should be built-in

19:02 but then it won't work with older versions.

19:07 DerGuteMoritz: ok, apparently I need an implementation of JAX-RS

19:07 any recommondations?

19:10 ok, Jersey it is

19:56 XPherior: Can anyone point me to a good place to learn lambdas? I couldn't find anything good on Google. Maybe I'm just blind.

19:56 amalloy: XPherior: new to functional programming? what other languages are you used to?

19:57 XPherior: I've been trying out Lisp for a few weeks. I sort of understand functional programming. Used to Ruby, Python, Java, C#

19:57 For the record, Clojure is the coolest thing I've seen in a while. It just gives me shivers looking at how idiomatic it can be.

19:58 amalloy: XPherior: don't python and ruby have lambdas? are you just looking for clojure's syntax?

19:59 XPherior: Er, yeah. Looking for how it's done in Clojure.

19:59 Sorry for not being clear.

19:59 amalloy: ,((fn [x y z] (+ x (* y z))) 1 2 3)

19:59 clojurebot: 7

19:59 DerGuteMoritz: XPherior: it's called `fn' in Clojure

20:00 XPherior: Woah, evaluator bot. Cool.

20:00 Why's a function definition called defn and a lambda fn, then?

20:00 gfrlog`: ,(let [f (fn [x y] (- x y))] (+ (f 12 15) 9))

20:00 clojurebot: 6

20:00 XPherior: Ha! That's brilliant.

20:00 gfrlog`: XPherior: "defn" sorta means "def fn"

20:01 * gfrlog` waits for someone to correct him

20:01 DerGuteMoritz: XPherior: (defn foo ...) is short for (def foo (fn ...))

20:01 XPherior: Does that imply that all functions return functions?

20:01 DerGuteMoritz: Clojure is a weird mix of Common Lisp and Scheme in that regard

20:01 gfrlog`: XPherior: no

20:02 DerGuteMoritz: XPherior: depends on what you mean by that :-=

20:02 :-)

20:02 XPherior: DerGuteMoritz: Do all functions return a (fn ...) ?

20:02 gfrlog`: XPherior: do you understand the expression (def foo "bar")?

20:02 DerGuteMoritz: XPherior: no, `fn' constitutes a function, (fn ...) *is* a function

20:02 XPherior: I think. It's a function named foo that's given a string "bar" parameter?

20:03 DerGuteMoritz: note the difference between def and defn!

20:03 XPherior: Mind boggling!

20:03 I don't think I quite got it.

20:03 gfrlog`: XPherior: no; it's a lot closer to setting a variable to a value like foo = "bar"

20:03 * gfrlog` waits for someone to scold him

20:03 amalloy: DerGuteMoritz: i don't think your distinction between "constitutes" and "is" makes any sense

20:03 XPherior: Oh! Right. foo would be like a global, correct?

20:03 amalloy: right

20:03 gfrlog`: XPherior: sort of -- it's still qualified by whatever namespace you're in

20:03 DerGuteMoritz: amalloy: that's possible, I made that up :-)

20:04 gfrlog`: so not like a $global in ruby

20:04 XPherior: Okay, wait. So if fn is lamba, what's the hash do?

20:04 gfrlog`: XPherior: syntactic sugar

20:04 #(+ % 6) <=> (fn [%] (+ % 6))

20:04 amalloy: ,'#(inc %)

20:04 clojurebot: (fn* [p1__72#] (inc p1__72#))

20:04 XPherior: ,(reduce (fn + %1 %2) (range 100))

20:04 clojurebot: java.lang.RuntimeException: java.lang.RuntimeException: java.lang.RuntimeException: java.lang.UnsupportedOperationException: nth not supported on this type: Symbol

20:04 XPherior: Ah D:

20:05 Hm.. That's interesting

20:05 amalloy: demonstrations in #clojure: if you can still read it, there isn't enough punctuation

20:05 DerGuteMoritz: amalloy: wait, it makes a little sense. "constitutes" emphasizes the function of fn while "is" emphasizes the result of the expression, or something like that!

20:05 XPherior: Haha

20:06 gfrlog`: ,(#(% %) #(% %))

20:06 clojurebot: java.lang.StackOverflowError

20:06 amalloy: DerGuteMoritz: whatever helps you sleep at night. if you said "constructs" instead of "constitutes" i might buy it

20:06 XPherior: Is it correct to say the body of defn is always a fn, and the fn's result is returned?

20:06 DerGuteMoritz: ("function of" not to be taken literally there)

20:07 amalloy: ok, let's say that instead!

20:07 XPherior: defn doesn't return anything, it binds a function to a name

20:07 gfrlog`: XPherior: (defn foo [x] (bar x)) <==> (def foo (fn [x] (bar x))) <~~> foo = lambda{|x|bar(x)}

20:07 DerGuteMoritz: hm wait, I haven't actually verified that

20:07 gfrlog`: DerGuteMoritz: I think it returns the var

20:08 DerGuteMoritz: I see

20:08 well, but that's secondary anyway

20:10 XPherior: Okay. So what's with all the wacky %s?

20:10 amalloy: XPherior: gfrlog` is just being a smart-ass

20:10 XPherior: Ah

20:10 amalloy: but when you use the #(...) syntax for a lambda, the paramters are implicitly named %1, %2, %3...

20:10 and % is short for %1

20:10 XPherior: But even (fn [%] (+ % 6) uses it

20:10 Ahh, cool

20:10 So I can still do (fn[name age] (...)) if I wanted?

20:10 amalloy: right

20:10 XPherior: That's nice that I can still name arguments in anonymous function

20:10 Functions* Keeps the code more understandable sometimes

20:10 amalloy: XPherior: in *that* context, % is just a symbol like any other. you could replace % with a, or with lkdfahfakf, or with -!$+

20:10 XPherior: That makes sense.

20:10 amalloy: only in #(..) does it have special meaning

20:10 XPherior: Ironically, I really didn't want any documentation of lambas in Clojure.

20:10 amalloy: haha

20:11 XPherior: Hmm

20:11 What's the purpose of the body being a fn?

20:11 I mean, why would we want that?

20:11 amalloy: XPherior: "the body being a fn" is something you keep saying that i don't understand

20:11 XPherior: (def foo (fn [x] (bar x)) ... (fn [x] (bar x)) <- I'm calling that the body

20:12 What would one call that?

20:12 amalloy: don't. call it the value

20:12 because def just assigns a value to a var

20:12 DerGuteMoritz: XPherior: in your example, (bar x) would be the body of the function actually

20:13 amalloy: defn does the work of turning the body into a function and then calling def

20:13 XPherior: Okay. So def assigns a variable, fn is a lambda, and defn assigns a fn to a def'ed var?

20:13 amalloy: full marks

20:13 hiredman: except for the use of the word variable

20:14 * gfrlog` :-Ds

20:14 XPherior: What do Clojure people call it?

20:14 amalloy: we call it "oh god, please no, never let anything vary"

20:14 DerGuteMoritz: heh

20:14 XPherior: Hahaha

20:14 gfrlog`: unless you call def again

20:14 * DerGuteMoritz is curious too

20:14 hiredman: XPherior: def creates a var

20:14 technomancy: varies-but-not-at-runtime-able

20:14 DerGuteMoritz: and var is not short for variable?

20:14 hiredman: no

20:15 DerGuteMoritz: aha!

20:15 XPherior: I really like this language. I hope I go far with it.

20:15 hiredman: a var is a specific thing with precises semantics, a variable is not

20:15 amalloy: presumably var was short for variable when it was named, but they are now two distinct concepts

20:16 DerGuteMoritz: as opposed to atoms and refs and whatnot?

20:16 * DerGuteMoritz still has to read up on those

20:16 amalloy: DerGuteMoritz: those aren't variables either

20:16 DerGuteMoritz: yeah

20:16 gfrlog`: well how do I get my hands on a damn variable then??

20:16 DerGuteMoritz: I mean err

20:16 amalloy: clojure doesn't have variables in the classic sense; just locals, which don't vary

20:17 and reference types

20:17 DerGuteMoritz: you have things like vars, bindings, refs and atoms which are lumped together as "variables" in most other languages

20:17 XPherior: So it's like Haskell in its emphasis on immutability?

20:17 gfrlog`: guess it's time to "lein new variables-clj"

20:17 amalloy: XPherior: a little less so

20:17 but a lot more than scala, say

20:17 DerGuteMoritz: gfrlog`: I hear there are transients which are mutable

20:17 amalloy: or, god forbid, ruby

20:17 XPherior: Think reading Programming Clojure is the quickest way to really learn the details of the language?

20:18 technomancy: gfrlog`: there's with-local-vars, but you didn't hear that from me

20:18 amalloy: Joy of Clojure

20:18 gfrlog`: DerGuteMoritz: ah yes. So my library will just figure out how to make transients less restrictive

20:18 XPherior: I kind of want to make an IRC bot in Clojure. Seems like an alright first project.

20:18 amalloy: XPherior: fork sexpbot or clojurebot

20:18 technomancy: gfrlog`: *whimper*

20:18 amalloy: $whatis source

20:18 sexpbot: source is http://github.com/cognitivedissonance/sexpbot

20:18 gfrlog`: :)

20:18 amalloy: clojurebot: source?

20:18 clojurebot: source is http://github.com/hiredman/clojurebot/tree/master

20:18 XPherior: Thanks!

20:18 DerGuteMoritz: gfrlog`: why not use Scheme then ;-)

20:18 gfrlog`: DerGuteMoritz: clojure is the hipster language right now

20:19 XPherior: I notice Clojure's author didn't write any books on it. Wonder why

20:19 DerGuteMoritz: gfrlog`: ah okay, I didn't know that was a requirement

20:19 gfrlog`: XPherior: he didn't have to?

20:19 XPherior: It's just kind of common, I guess.

20:19 amalloy: busy writing it, i imagine

20:19 DerGuteMoritz: by the way, on the clojars site there are no links to the project sites or upstream repositories, or am I mistaken?

20:19 amalloy: he's done several good lectures, which are online

20:20 gfrlog`: "The Richness of Clojure"

20:20 technomancy: DerGuteMoritz: clojars needs help

20:20 XPherior: I'll check those out.

20:20 technomancy: there's a list of "somebody needs to implement this please" features

20:20 XPherior: Are there any good IDES for Windows other than Enclojure?

20:20 technomancy: nobody's bothered yet

20:20 gfrlog`: VirtualBox?

20:20 halfprogrammer: But he has released a bunch of wonderful videos :-)

20:20 DerGuteMoritz: technomancy: cool, where is that list kept?

20:21 halfprogrammer: I mean Rich ^

20:21 technomancy: DerGuteMoritz: http://groups.google.com/group/clojars-maintainers/browse_thread/thread/d4149ec96316d5b1

20:21 XPherior: Does Rich ever come in here?

20:21 gfrlog`: yes

20:22 it makes all but the most confident of us shut up for 30 minutes

20:22 XPherior: Hahah

20:22 * gfrlog` doesn't really know if that's true or not

20:22 XPherior: I think Clojure is going to take my skills to the next level.

20:22 technomancy: if anyone is interested in hacking clojars, I've got an automated VM setup for hacking clojars that I'm working on

20:22 will publish it soon

20:23 it's a bit of a bear to config by hand

20:23 gfrlog`: clojars isn't a 30-line compojure app?

20:24 cemerick: XPherior: check out Counterclockwise for eclipse

20:24 technomancy: gfrlog`: there's a bunch of scp-handing stuff necessary to handle deploys

20:24 gfrlog`: ~900 LOC

20:24 gfrlog`: technomancy: in clojure?

20:24 technomancy: aye

20:25 gfrlog`: that's respectable

20:25 c.c.shell-out ftw

20:25 technomancy: gfrlog`: no, scp *server* handling

20:26 can't shell out for that

20:26 gfrlog`: it'd be fun to have a repl plugin where any command that throws an exception gets retried as a shell command

20:26 s/command/expression

20:26 sexpbot: botsmack

20:26 worthless

20:27 technomancy: I see.

20:27 amalloy: gfrlog`: i remember someone trying to do that as a macro

20:28 basically impossible for about ten reasons

20:28 gfrlog`: haha

20:28 wait really?

20:28 it sounds too easy

20:28 technomancy: gfrlog`: eshell accepts shell commands and lisp in the same prompt

20:28 DerGuteMoritz: does anyone know of a maven repository that has Jersey 1.8?

20:28 technomancy: it's awesome

20:28 gfrlog`: technomancy: I bet

20:28 technomancy: apart from the lack of lexical scope, I mean. =(

20:29 gfrlog`: I once tried to write a macro that would allow any [sub-]expression to be in reverse order; THAT was impossible

20:29 DerGuteMoritz: lexical-let ftw :-)

20:29 gfrlog`: you mean a reverse polish notation macro?

20:29 gfrlog`: DerGuteMoritz: (defmacro with-dyslexia ...)

20:30 scgilardi: technomancy: any way to compare the awesomeness of eshell with, say, the awesomeness of the piranhamoose?

20:30 DerGuteMoritz: hm

20:30 technomancy: scgilardi: that's hard! the piranhamoose is _pure_ awesome, whereas eshell is extremely awesome but not 100%.

20:31 XPherior: cemerick: I'll check it out now.

20:31 scgilardi: got it, thanks.

20:31 amalloy: gfrlog`: interesting. i would have guessed the "optional reversing" would be totally viable. what problems did you have?

20:31 i mean, aside from compile times in the 2^n range

20:31 gfrlog`: amalloy: well the strategy would be to wrap every [sub-]expression in a try-catch, where the catch tries in reverse

20:31 DerGuteMoritz: that's what I expected

20:32 technomancy: clojurebot: piranhamoose is unadulterated awesome: http://blog.howfastareyou.net/wp-content/uploads/2009/08/piranhamoose.jpg

20:32 clojurebot: Ok.

20:32 technomancy: attaboy

20:32 gfrlog`: you could do that, but higher order functions might run in reverse-form as well

20:32 maybe something about special forms as well? :/ I dunno, maybe it'd be halfway workable

20:33 DerGuteMoritz: ooh, like that, I see. that's ambiguous then

20:33 gfrlog`: I guess you could do better than try-catch some of the time by resolving the symbol and seeing what it means

20:35 ...if it's even a symbol

20:37 DerGuteMoritz: just make it a one character macro that reverses the current expression

20:37 gfrlog`: DerGuteMoritz: well the idea was that you could accidentally input things in reverse, so you wouldn't be indicating it explicitely

20:38 DerGuteMoritz: right but that's not possible

20:38 because of ambiguities

20:38 gfrlog`: so yes, redefining the goal does make it easier :)

20:39 DerGuteMoritz: or ... unintended reverse evaluation

20:39 amalloy: DerGuteMoritz: a risk you have to take when you're dyslexic. not a failing of the library

20:39 gfrlog`: (defmacro <> [& forms] (map reverse forms))

20:39 * gfrlog` knows that doesn't work

20:39 amalloy: things often work, but then sometimes for reasons you don't understand you have to manually reverse the order to make it work

20:40 sounds like dyslexia to me

20:40 gfrlog`: that is like 17 steps away from working

20:40 gfrlog`: ((forms reverse map) [forms &] <> defmacro)

20:40 maybe it'll work that way

20:41 amalloy: hah

20:41 gfrlog`: it can't be seventeen, it's more like 3

20:41 amalloy: yes, first step when defining a new macro: see if it's already included in the language

20:41 gfrlog`: I think it's two, but I said three just to be safe

20:41 amalloy: i'm choosing to count each irc message you send as a step in adjusting your thinking the right way

20:42 gfrlog`: I don't accept the rules of your game

20:42 okay let's try my two fixes

20:43 (defmacro <> [& forms] (list `do (map #(if (list? %) (reverse %) %) forms)))

20:43 aw crap

20:43 s/list/list*

20:43 or s/list/cons

20:43 amalloy: s/list*/cons

20:44 gfrlog`: list* in case I decide to put anything else in there :)

20:44 amalloy: gfrlog`: you could do a good job of making it totally unreadable with postwalk-replace

20:44 gfrlog`: amalloy: but then I'd have to learn how to do that

20:45 amalloy: just wrap the top-level form with (<> ...) and then you'll have a terrible time making anything work

20:45 hiredman: ,(pl (↕map (replicate 3 (↕apply vector $ (↕map range $ 10 inc · inc · inc) call · ⌽* $ 10 · call · (⌽+ -2) map)) shuffle)))

20:45 clojurebot: ([80 50 100 60 10 40 70 30 20 90] [80 90 30 60 20 100 10 50 70 40] [70 50 20 100 30 90 40 80 60 10])

20:45 gfrlog`: &(pl (↕map (replicate 3 (↕apply vector $ (↕map range $ 10 inc · inc

20:45 sexpbot: java.lang.Exception: Unable to resolve symbol: pl in this context

20:46 gfrlog`: it's clojurebot voodoo

20:46 amalloy: gfrlog`: (use 'clojure.walk) (defmacro <> [& args] (postwalk-replace reverse args))

20:46 XPherior: Thanks for your wonderful help everyone! You've been excellent. I shall be back!

20:46 Good bye.

20:46 gfrlog`: XPherior: bye

20:46 amalloy: ahh -- that's different from mine too isn't it?

20:47 amalloy: quite

20:47 gfrlog`: reversing at all levels instead of top

20:47 amalloy: yeah

20:47 gfrlog`: I will submit a talk to clojure conj offering to compare and contrast the two approaches

20:48 hiredman: ,(pl inc $ inc $ 1)

20:48 clojurebot: 3

20:48 gfrlog`: hiredman: what's going on?

20:48 hiredman: ,(macroexpand '(pl inc $ inc $ 1))

20:48 clojurebot: (do (inc (inc 1)))

20:49 hiredman: I like to trot pl when people talk about crazy macros

20:49 pl out

20:49 gfrlog`: ah ha

20:49 amalloy: haha

20:49 gfrlog`: very good

20:49 amalloy: oh, i read that as "pl out"

20:49 like "it's done its job and it's leaving"

20:52 gfrlog`: alright fine I'll go mudge around in hiredman's github account looking for pl

20:53 hiredman: ~functional

20:54 clojurebot: I don't understand.

20:54 hiredman: ~pl

20:54 clojurebot: examples is http://en.wikibooks.org/wiki/Clojure_Programming/Examples/API_Examples

20:54 hiredman: ~transform

20:54 clojurebot: transform is http://github.com/hiredman/odds-and-ends/blob/8a84e6ddbad9d71f714ba16c3e1239633228a7eb/functional.clj

20:56 gfrlog`: that kevin.clj is pretty weird

20:56 maybe using emacs results in files like this...

21:03 tomoj: gfrlog`: like what?

21:03 gfrlog`: tomoj: lots of functions mixed with expressions trying them out

21:04 I think emacs lets you highlight stuff and eval it, so it makes sense of that

21:04 I only use stone-age tools, so I don't know what it's like to have that kind of power.

21:04 tomoj: not exactly good to stick them in toplevel

21:04 gfrlog`: tomoj: I don't think it's serious code at all

21:04 tomoj: because e.g. they might run on uberjar and stuff

21:04 sure

21:05 but if one expression takes a long time to run and you C-c C-k (loads the entire file)

21:05 ...

21:05 better to stick those in (comment ) blocks

21:05 where they still work just as well for emacs

21:05 s/blocks/forms/

22:42 symbole: Is there a lazy http library for Clojure? I looked at clj-http, but it seems like it downloads entire files before returning.

22:51 clojure.contrib.http.agent seems to be what I want.

23:06 mrnex2010: are there functions to modify a fn, such as to add an arity overload to a fn?

23:08 (im building an util to add arity overload to all the functions that use a fn that has been overloaded, all the way up to the functions that use it

23:08 )

23:14 symbole: mrnex2010: I don't quite follow.

23:15 Clojure has arity overloading, which is specified during function definition.

23:21 mrnex2010: symbole: sorry for the vaguity, what i meant was a function that returns a function with another arity

23:22 overload

23:23 i can do it inserting it into the function as a string, but i got curious if we had higher order fns that did that kind of stuff

23:23 symbole: You can use macros to do that for you.

23:24 Make a macro that creates a function with however many params you want.

23:24 amalloy: this sounds like the sort of thing that doesn't need a macro, but it's hard to understand what the goal is. are you just reinventing partial?

23:25 eg, maybe you could gist an example usage of your function and the desired result

23:25 mrnex2010: no, supose you have a fn with 2 params, another fn uses it, and yet another one uses the intermediate fn

23:26 then you make an overload on the first

23:26 but your fingers are too delicate to rewrite the other fns with overloads that reflect the overload on the first fn

23:27 amalloy: mrnex2010: you're looking for like a refactoring tool that will add a default argument?

23:27 or pass along an argument, i guess

23:28 mrnex2010: amalloy: yes a refactoring tool but not necesarilly a default arg but the arg that is added on a fn yes to pass it along

23:28 sorry for the boraty writing on the prev line lol

23:29 amalloy: i see. on the one hand, i can see why that would be convenient. on the other, if you have N functions calling each other, and a change to function f requires a change in the api to all N functions, you're probably doing something wrong

23:30 mrnex2010: yeah i can imagine that stuff

23:30 i just meant it for abstractions like the ones on SICP first chaps

23:31 and hacking libs

23:36 amalloy: i don't think there's any nice way to do it. you could write a macro which wraps a defn, adds an extra arg to it, and is told what function to internally pass along that function to

23:36 but i think it will be less readable and more work

23:37 * pass along that arg to

Logging service provided by n01se.net