#clojure log - Jul 02 2009

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

0:00 Chouser: http://java.sun.com/j2se/1.5.0/docs/api/java/util/regex/Pattern.html

0:01 Under "Special constructs (non-capturing)" that lists various pieces that can be included in the regex pattern (which is what #"" is)

0:01 for the flags you can see your options include each of idmsux

0:02 Under "Field Summary" you can see a list of english-named flags, including DOTALL

0:02 click on any of those to find out what flag letter to use, for example (?s)

0:03 newbie_: thanks you

0:04 *thank

0:06 Chouser: np

0:22 newbie_: if i have a struct like the following (defstruct atest :t1 :t2) and a vector [ 1 2 ] is there another way to convert this in to a struct then to call (nth vec 1) (nth vec 2)?

0:31 jackdempsey: newbie_: just learning clojure myself but if you're going to do that a lot you could create a function and use some destructuring

0:31 ,(defstruct atest :t1 :t2)

0:32 ,(def avector [1 2])

0:32 (defn make-atest [{t1 t2}] (struct atest t1 t2))

0:32 oops

0:32 ,(defn make-atest [{t1 t2}] (struct atest t1 t2))

0:32 ,(make-atest avector)

0:32 although i'm sure there's better ways, etc :-)

0:33 can use (first avector) and (second avector) too fwiw

0:33 Paul___: IF I understand what you are asking,

0:33 you can simply use apply.

0:33 In other words, if you have your struct as given,

0:33 you can do (apply atest [1 2])

0:33 woops

0:34 I meant, (apply struct atest [1 2])

0:34 Does that answer your question, newbie, or did I misunderstand ?

0:34 newbie_: thank you both that is what i need

0:34 jackdempsey: yea i think you got it.....basically looking for a way to unroll a sequence into individual args

0:35 nice, i have to remind myself of apply......used to *args from ruby

0:35 Paul___: Yeah, it's a similar use to *args ...

0:35 BRB.

0:36 pmooser: Ok, back.

0:53 jackdempsey: so a question of style

0:53 there're obviously a lot of java libraries out there

0:53 and working with java in clojure seems pretty simple and easy

0:53 do people find themselves writing wrappers to java libs often?

0:54 grrrt: not for simple stuff

0:54 jackdempsey: i'm looking to do some NLP stuff with clojure, and will probably make use of a bunch of java libs......but would prefer to code more like i'm in clojure

0:54 grrrt: but in my clojure SWT experiments, I can replace +/- 10 lines of java with 1 line of clojure

0:54 jackdempsey: nice

0:55 grrrt: I'm still doing things like (.open myWindow)

0:55 jackdempsey: so people wouldn't think it foolish to write some wrappers, or macros even i guess, to streamline what consumers of the lib might want ot do

0:55 gotcha

0:55 grrrt: that's what I'm doing - it might still be foolish though :)

0:56 Some things in java are truly hideous, like anonymous inner classes

0:57 I define functions for that, so I don't have to use (proxy ....) all the time

0:58 one drawback of wrapping could be that it might be difficult to still provide access to the "raw" java objects you're wrapping...

1:02 jackdempsey: yeah

1:03 i mean i'm still really early to this, so i think i'll just go with something, and see how it feels

1:10 Chouser: the general recommendation is to not wrap too early.

1:10 go ahead and use whatever java lib you need directly.

1:10 But if you start to see patterns or ways you could make your clojure code clearer with some wrappers, then go for it.

1:10 just no (defn parse-int [s] (Integer/parseInt s)) please

1:11 grrrt: no you don't really gain anything with that

1:12 Chouser: wait, am I still awake!?

1:12 * Chouser goes to fix that.

1:24 Knekk: anyone know offhand how to solve this for compiling vimclojure on OSX ? /Users/jorgeb/CLOJURE/vimclojure-2.1.1/build.xml:66: java.lang.ClassNotFoundException: clojure.contrib.pprint.PrettyWriter (pprint.clj:14)

2:07 netmage0606: Hello

2:09 If I understood it right then you can have access to all Java Libs with Clojure. So, there should be a way to program JavaFX with Clojure. Is there someone who tried this? Or are there some technical reasons which makes it for today impossible to program JavaFX in Clojure?

2:10 grrrt: as far as I know, there is no java API for javaFX

2:10 you have to program in JavaFX script (or whatever it's called)

2:11 there was a thread about this on the clojure google group not that long ago I think

2:13 netmage0606: grrrt: Thanks. I will look for this thread.

2:41 Lau_of_DK: Morning guys

4:26 duncanm: what's the right way to use the clojure.contrib.http.agent lib?

4:29 ah, i got it

8:45 AWizzArd: Compiling (via ant) a freshly cloned Clojure repo now results in 6 files: clojure, slim and sources, and those three again but with an "alpha" extension.

8:46 The ones without alpha: did they change since the last 1.0? Or are basically all the changes in the alphas?

8:47 Chouser: I would assume the alpha and non-alpha are identical

8:48 yeah, they're the same

8:48 it's just that one set has the version number so you'll know, the other left it out -- just to be shorter?

8:49 frodef: when a java method calls for "byte[]", how do I make a buffer like that?

8:49 Chouser: frodef: (into-array Byte/TYPE (map byte [1 2 3]))

8:50 or make-array

8:50 frodef: Chouser: thanks.. how is that different from (make-array java.lang.Byte 100) which doesn't seem to work?

8:51 Chouser: frodef: it's the difference between an array of Byte and an array of byte

8:52 frodef: fascinating..

8:52 rhickey: anyone want to take a crack at chunked 'for'?

8:52 Chouser: Byte is an instance of a subclass of Object

8:53 frodef: byte is a primitive type, handled specially by the JVM. It's not an Object at all, doesn't have any methods of its own, take up exactly one real byte in memory, etc.

8:54 frodef: Chouser: right, but it looks to me that the into-array thing returns an array of Byte? I get #<Byte[] [Ljava.lang.Byte;@5a71c3>

8:54 Chouser: (make-array Byte/TYPE 0) ==> #<byte[] [B@359ecd80>

8:55 (make-array Byte 0) ==> #<Byte[] [Ljava.lang.Byte;@319c0bd6>

8:56 frodef: Chouser: oh right, I missed the Byte/TYPE part, sorry.. it works fine. thanks!

9:26 AWizzArd: Chouser: thanks for the alpha info

9:27 jweiss_: can someone tell me a good resource to learn how to *read* clojure? I can look up function names, but I have no idea what some symbols mean, like #, #^, etc.

9:27 dnolen: http://clojure.org

9:28 jweiss_: dnolen: uh, yeah

9:28 dnolen: look at the part on reader

9:28 the reader

9:28 AWizzArd: I think that code is written in Java.

9:30 jweiss_: dnolen: ah ok that's what i was looking for thanks

9:31 dnolen: np

9:40 Jetien: hi! is there a more elegant way to check if a vector v contains a given element x than this?: (some #(== %1 x) v)

9:42 Chouser: (some #{x} v)

9:44 Jetien: thanks chouser

10:14 i can' figure out how to solve this: for a given vector and element x i need the subvector of v starting with the first occurrence of x and all remaining elements

10:15 my idea was using subvec but there doesn't seem to exist a function that yields the index of the first occurrence of an element in an vector

10:16 Chouser: hm. Well, there's .indexOf

10:17 (defn subvec-starting-with [v x] (subvec v (.indexOf v x)))

10:17 Jetien: oh :)

10:17 thanks

10:18 i forgot that there's a java part to all collections

10:18 Chouser: :-)

10:18 you rarely need it.

10:18 that is, most of the useful operations are pulled out into clojure.core functions

10:19 but some of the linear-search ones are not, I guess because they're discouraged?

10:19 your vector doesn't happen to be in sorted order, does it?

10:19 any sorted order?

10:19 Jetien: no, it's an arbitrary sequence of integers

10:20 and the order is important

10:20 Chouser: yeah, this seems the most sensible way I can think of.

10:21 similarly .contains is not exposed via a function, because vectors are a poor fit if you're going to be doing that kind of thing -- sets are better.

10:22 Jetien: so a clojure vector implements java.util.Vector ?

10:22 Chouser: but in this case a vector seems really right -- you care about order and you want to use subvec.

10:22 (instance? java.util.Vector [1 2 3]) ==> false

10:22 duck1123: is there a public javadoc for the java part of clojure hosted anywhere?

10:23 Jetien: chouser: so how did you know you could invoke indexOf ?

10:23 Chouser: duck1123: not that I know of. Writing javadocs is on the todo, but I suppose when c-in-c is done that becomes a bit moot, or at least different...

10:24 Jetien: You can use (supers (class [1 2 3])) to see what it implements, but I just used clojure.contrib.repl-utils/show

10:24 (show [1 2 3]) lists all the methods you can get at.

10:24 Jetien: thanks

10:48 jackdempsey: can't seem to run that show

10:48 am i missing something obvious?

10:48 java.lang.Exception: Unable to resolve symbol: show in this context (NO_SOURCE_FILE:42)

10:48 user=> (show [1 2 3])

10:49 cgrand: jackdempsey: show is in clojure.contrib.repl-utils

10:49 jackdempsey: hm yeah, i guess its not loading anymore

10:50 k brb

10:50 Chouser: (use '[clojure.contrib.repl-utils :only [show javadoc source]])

10:55 jackdempsey: thx Chouser somethings not right with jars/paths/etc

10:55 ah the fun

11:03 yea i think when i updated to 1.0.0- contrib needed updating as well......and now i just have to figure out where to get what from to build. good experience tho

11:04 clojure-contrib is still in svn right?

11:04 just the main clojure on github?

11:08 achim: jackdempsey: both are on github

11:08 jackdempsey: oh ok, i guess i missed contrib

11:08 jeez

11:08 duh

11:08 hadn't shown up in a search, strange

11:09 achim: the current clojure contrib won't work with clojure 1.0, i believe

11:09 here's the last tree before test reorganization: http://github.com/richhickey/clojure-contrib/tree/3073f0dc0614cb8c95f2debd0b7e6a75c1736ece

11:11 jackdempsey: oh

11:11 hehe

11:11 yeah i'm going in circles

11:11 downloaded both sources, built clojure then clojure-contrib specifying the location of the clojure.jar

11:11 load them up in classpath, startup the repl, and its still not finding show

11:11 will check out that tree, thanks achim

11:13 Chouser: jackdempsey: are you getting an error when you run that 'use' line?

11:13 jackdempsey: yeah, its not finding it, so something simple but bad is happening :-)

11:13 java.io.FileNotFoundException: Could not locate clojure/contrib/repl_utils__init.class or clojure/contrib/repl_utils.clj on classpath: (NO_SOURCE_FILE:0)

11:14 let me start fresh one more time, and will paste my commands.....still new to ant and very possible i've completely missed something basic here, but i did have this working at one point. brb

11:15 Chouser: if you're not getting errors when you build, that's probably not the issue.

11:15 jackdempsey: hmm, yeah the builds seem ok

11:15 slim isn't necssary right?

11:16 thats only if you don't want the compiled clojure code

11:16 Chouser: then it's more likely your classpath when you start java

11:17 achim: jackdempsey: how do you start the REPL?

11:17 jackdempsey: pasting

11:18 Knekk: any idea about this error while compiling the contrib jar?

11:18 [java] java.io.FileNotFoundException: Could not locate clojure/walk__init.class or clojure/walk.clj on classpath: (dataflow.clj:17)

11:19 jackdempsey: achim: http://gist.github.com/139520

11:19 achim: Knekk: compiling with clojure 1.0?

11:19 jackdempsey: i'm wondering if the clojure.lang.Repl is old huh

11:19 Knekk: compiling against clojure from May04, using -Dclojure.jar

11:19 achim: yes, from here http://code.google.com/p/clojure/downloads/list

11:19 on OSX

11:19 achim: Knekk: the current contrib doesn't work with 1.0 anymore :-(

11:20 Knekk: achim: I got http://github.com/richhickey/clojure-contrib/tree/3073f0dc0614cb8c95f2debd0b7e6a75c1736ece

11:20 achim: isn't that the last working version to 1.0?

11:20 jackdempsey: some tags would be helpful

11:21 achim: i thought it was ...

11:21 Knekk: a nice bundled release with contrib would be nice too :)

11:22 jackdempsey: yep, so much to do, so little time :-)

11:22 achim: jackdempsey: does it work if you just call "java -cp path/to/clojure.jar:path/to/clojure-contrib.jar clojure.main" ?

11:23 jackdempsey: thats a negative ghostrider

11:23 Clojure 1.1.0-alpha-SNAPSHOT

11:23 user=> (show [1 2 3])

11:23 java.lang.Exception: Unable to resolve symbol: show in this context (NO_SOURCE_FILE:1)

11:23 hmm

11:24 (use '[clojure.contrib.repl-utils :only [show javadoc source]]) returns a nil tho

11:24 Chouser: that's good!

11:24 jackdempsey: ah

11:24 yep

11:24 and now show works

11:24 so, hm, excuse me but wtf :-D a bit confused

11:24 so its just not loading repl-utils at the start?

11:25 i guess that would make sense, huh

11:25 man, that is so useful btw

11:29 k, so i was putting clojure-contrib in the path, but not actually loading it......now its loading, so i can just type the use line and things are good

11:29 but, i'm guessing you all don't type out that use everytime you want to use show....so where should things like that go to load it up for you automaticaly?

11:31 lisppaste8: Chouser pasted "my current clj launch script" at http://paste.lisp.org/display/82905

11:31 Chouser: wait, are you on linux?

11:32 jackdempsey: nope os x

11:32 Chouser: ah, well ... not sure how well that will do then.

11:33 jackdempsey: ok well, its a similar beast though

11:33 i think my load script is ok for now

11:33 unless

11:33 does this loadup the repl-utils for you

11:33 Chouser: anyway, I have a $HOME/.clojure directory where I put a repl-init.clj where I set *print-level* and (use repl-utils) etc.

11:33 jackdempsey: ahh

11:33 i see

11:33 Knekk: where do I get a contrib version that will compile against 1.0?

11:33 jackdempsey: any chance you can pastie that?

11:34 Knekk: did you try that http://github.com/richhickey/clojure-contrib/tree/3073f0dc0614cb8c95f2debd0b7e6a75c1736ece

11:34 Chouser: that part is just the -i repl-init.clj part, followed by optional -e or -i's

11:34 jackdempsey: i believe i have that working against edge of clojure

11:34 Chouser: er, -r's

11:34 jackdempsey: yeah, didn't realize thats where you set those things up

11:34 i think thats all i need now

11:35 kotarak: Knekk: everything before the walk, template, test-is move should work against 1.0

11:36 achim: re: rlwrap: is there a way to have multiline history?

11:36 i'd ditch emacs right away

11:37 Chouser: achim: I dunno -- I use a keyboard shortcut to merge pastes into a single line before dropping them into rlwrap

11:37 achim: Chouser: that's clever!

11:38 Knekk: Thanks, 3073f0dc0614cb8c95f2debd0b7e6a75c1736ece compiles

11:38 jackdempsey: cool

11:39 Knekk: now vimclojure compiles as well. Awesome

11:39 Chouser: achim: it actually does a bunch of other stuff too: http://agriffis.n01se.net/skel.hg/index.cgi/file/d59d4656cbec/bin/eatpaste

11:39 kotarak: Knekk: VC is in sync with 1.0 at the moment. Have to incorporate the latest changes in HEAD in the bleeding-edge branch, eg. the stacktrace move.

11:40 Knekk: kotarak: do updates get announced in here?

11:41 jackdempsey: Chouser: nice. out of curiousity, would you mind pasting your repl-init.clj file?

11:41 kotarak: Knekk: no. Releases are announced via the group and twitter

11:41 Knekk: But commits go unnoticed (unless you follow it on bitbucket)

11:43 Knekk: thanks

11:43 back from the office in a few.

11:44 achim: Chouser: that's useful, thanks!

11:54 kotarak: How can I get continuous updates of a JTextField

11:54 ?

11:54 ie. w/o the user hitting enter?

11:59 jackdempsey: Chouser: amazing, your clj script works for me. thx!

11:59 Chouser: jackdempsey: oh, great. you figured out the .clojure/cp-1.0 stuff?

12:00 jackdempsey: yeah i had a couple things to change, but now the use i have in repl-init correctly loads up show and friends

12:00 are there any other useful things to loadup/set that i may not know of? you mentioned *print-level* i think?

12:00 achim: ah, rlwrap has some multi-line support: you can specify a line break substitute, like rlwrap -m " #_| " ...

12:00 Chouser: well, good. I haven't put it up anywhere because it keeps changing and/or break. :-/

12:01 achim: hitting ctrl-^ then lets you edit a multi-line expansion of the current expression

12:01 Chouser: jackdempsey: I'll paste my .repl-init in a minute -- found a bug in 'add-break-thread!'

12:01 jackdempsey: oh ok cool, thx

12:01 Chouser: achim: oh! cool.

12:03 achim: that's good enough for me

12:03 * achim moves Aquamacs to trash can

12:09 jackdempsey: heh

12:11 lisppaste8: Chouser pasted "repl-init.clj" at http://paste.lisp.org/display/82906

12:15 Chouser: (e) is experimental for me... not yet sure if I like it.

12:30 jackdempsey: ok cool, thanks man

12:31 stuartsierra: Chouser: yes, it's experimental. :)

12:33 Chouser: stuartsierra: fwiw, it seems to me that most of the times I find it insufficient it's because I need more than the one cause that's printed.

12:34 I've been wondering lately if it would help to print the whole trace in order rather than having each cause in the reverse order of the frames

12:34 that ordering caused me confusion for a long time at the beginning, and still slows me down at times now.

12:39 stuartsierra: So the "root" cause should come first?

12:49 * stuartsierra is creating a contrib branch for Clojure 1.0

12:51 * stuartsierra created a contrib branch for Clojure 1.0: http://tinyurl.com/nh52r4

12:53 dysinger_: http://groups.google.com/group/clojure/t/55e55941bd05c5b5 :)

12:55 jackdempsey: hehe, nice

12:55 nice stuart

12:56 Chouser: thx again for the help, i tried to document all of it so others don't have to bug you the same way :-) http://jackndempsey.blogspot.com/2009/07/opening-your-mind-with-clojure.html

12:56 Knekk: so the recommended approach to multidimensional arrays of mutable data in Clojure is to instantiate as Java arrays?

12:58 texodus: Anyone in here know their licensing well enough to answer a few simple questions?

12:59 drewr: texodus:

12:59 Just ask; it'll start the conversation.

12:59 texodus: fair enough

12:59 clojure is epl

12:59 does that mean I can't distribute if my clojure project relies on gpl libs?

13:00 drewr: Perhaps, but that wouldn't be a restriction of Clojure's license.

13:01 You can bundle Clojure programs without dependencies and tell people to go get them if that's a problem.

13:01 texodus: true, but that's a pain for none-technical users

13:02 drewr: What non-technical user cares about Clojure? :-)

13:02 I would think programming in Lisp on the JVM automatically elevates you to technical status. At least that's why I'm here.

13:03 stuartsierra: texodus: you can distribute Clojure however you want.

13:04 drewr: texodus: Are you referring to the compatibility section of http://en.wikipedia.org/wiki/Eclipse_Public_License?

13:04 That does make it sound ominous.

13:06 stuartsierra: License terms don't matter until somebody sues you. If you're worried, distribute your app in two packages -- one for each license.

13:08 technomancy: stuartsierra: spoken like someone who works with lawyers. =)

13:10 stuartsierra: technomancy: yes.

13:20 Chouser: jackdempsey: thanks!

13:20 jackdempsey: no, ty! :-)

13:20 Chouser: jackdempsey: so you'll be handling support requests related to the clj script. ;-)

13:20 jackdempsey: hehe, yep, i'll be happy to help where i can :-D

13:21 technomancy: what's this about a clj script?

13:25 Chouser: technomancy: http://jackndempsey.blogspot.com/2009/07/opening-your-mind-with-clojure.html

13:27 technomancy: what's the rationale for not having a shell script in clojure itself? just that nobody has gotten around to it?

13:27 or to get people comfortable with the fact that they need editor integration? =)

13:28 Chouser: not sure. I suppose there's some question as to what exactly such a script ought to do.

13:28 there were several such scripts shared in a g.group thread a while ago, and I was surprised by the variety of functionality they provided.

13:28 not to mention the various platforms

13:31 technomancy: I guess the OS-specific ext_dirs etc are tricky.

13:36 Lau_of_DK: Good evening gents

13:37 technomancy: wow, I stick my head out on the mailing list, and I get a flood of patches.

13:37 that's awesome.

13:40 Lau_of_DK: patches for...?

13:41 Chousuke: ll

13:42 technomancy: Lau_of_DK: for clojure-mode and swank

13:42 Lau_of_DK: Ok - I had some NASTY trouble with Slime the other day - Will you consider taking that on ?

13:43 jackdempsey: hehe

13:43 technomancy: no, today I only accept patches

13:43 nah; just ask. =)

13:44 Lau_of_DK: :(

14:01 texodus: ah, two licenses, that does sound like a good idea

14:01 I dont anticipate being sued, but potentially harassed

14:02 and I'd like to distribute in a binary form so people without programming experience can use it

14:02 'binary'

14:02 thanks for the advice, all

15:03 Chouser: pop quiz, Q1: what does (assoc [] 0 1) return?

15:03 pop quiz, Q2: what does (assoc nil 0 1) return?

15:12 dysinger: I'd rather make my own special purpose shell scripts than have "standard" one - it's just java

15:12 but - for example - the macports version of clojure comes with bin/clj

15:13 (referring to 10:30am conversation)

15:15 kotarak: Chouser: ad 1: hmm... throws? ad 2: nil? geez, where do you get these questions from?

15:16 ok

15:16 Chouser: kotarak: :-)

15:17 kotarak: thanks for playing, we have a nice parting gift for you.

15:17 kencausey: So those answers were corect?

15:17 kotarak: The "Trostpreis"

15:17 no

15:17 drewr: [1] & {0 1} respectively

15:17 Interesting.

15:17 kencausey: oh, I guess 'parting' should imply that

15:18 kotarak: A nice flower. :)

15:19 The first is not so far fetched, but somehow ugly (special case). I don't like the second.

15:20 Chouser: right, so you're allowed to assoc into a vector at any existing index or at one index past the end which acts like conj

15:20 unlink: What's the best way to say (zero? (truncating-division n k))

15:21 kotarak: I would probably avoid both cases, although the first might have some nice applications. (less special treatment)

15:22 Chouser: (assoc nil 0 1) is perhaps like the corollary to (conj nil 1)

15:22 kotarak: hmm...

15:22 I don't like either....

15:22 Chouser: in each case 'nil' is a bit too vague, so Clojure could either throw something or pick a "reasonable" meaning for nil

15:24 kotarak: I think there is a difference between eg. nil and {}. So (assoc nil some thing) and (assoc {} some thing) means also something different to me.

15:50 slurms: hello

15:50 Chouser: hi

15:51 kotarak: do you'd prefer an exception over the current default behavior? or you have something else in mind?

15:52 kotarak: I would probably prefer an exception. How can assoc or conj decide, what to do on nil? assoc and conj don't act on sequences, but on collections. I think passing nil is an error.

15:52 unlink: Is there a way I can do this with reduce? http://paste.pocoo.org/show/126386/

15:54 Chouser: unlink: *whew*, an easy one: yes you can.

15:54 unlink: yeah

15:54 I just noticed

15:55 although the condescending attitude is not appreciated

15:55 Chouser: oh. well, "how" may not be so easy to answer. :-)

15:55 unlink: I'm sorry, I mean it to be funny, no condescending. I'll stop talking now.

15:55 unlink: oh, my apologies

15:57 Chouser: unlink: I certainly did not mean to suggest the conversion to reduce would be easy. I'd be happy to help if you need any.

15:59 unlink: thanks

16:08 dysinger: So everybody in here has full time clojure work ? :)

16:08 Chouser: hm, I appear to be headed toward (last ... (iterate ...)) rather than reduce

16:09 unlink: Chouser: or just (/ log(n) log(1024))

16:09 technomancy: dysinger: /me raises his hand

16:09 Chouser: unlink: heh. oh. :-)

16:10 dysinger: technomancy: also raising hand

16:10 kotarak: not full time, but I snuck some Clojure, hehe.

16:10 Chouser: I've checked in a little bit of clojure at my full time job...

16:10 kotarak: snuck in some Clojure, I meant..

16:10 * Chouser gives kotarak a high-five.

16:10 Chouser: :-)

16:10 kotarak: :)

16:11 pure self-defense against corporate IT

16:11 Chouser: and then hosted a 3-hour con call walking my team though it.

16:11 unlink: :(

16:11 well good for clojure

16:11 bad for business

16:11 Chouser: unlink: so you have a solution you like, or you want me to keep poking at it.

16:12 unlink: you can if you like

16:12 I'm actually just doing it for fun anyway, my final solution will be in python

16:12 I can't afford to give that 3 hour conference call.

16:12 Chouser: usually 3 hours of training is much more expensive.

16:13 kotarak: luckily my boss looks in the other direction... "As long as the job gets done..."

16:13 Waiting for corporate IT means doing steps backwards. :(

16:14 unlink: What's the right way to go from java.lang.Double -> java.lang.Integer?

16:14 specifically, the ceiling of the double

16:16 Chouser: probably (int (Math/ceil x))

16:16 unlink: Isn't that unchecked?

16:17 Chouser: (int (Math/ceil 1e100)) ==> 2147483647 so I think you're safe

16:17 unless you want an exception, I suppose.

16:18 unlink: I want a BigInteger

16:18 (am I asking for too much?)

16:20 Chouser: I don't know Java math very well. I don't see any BigInteger methods that take Doubles

16:20 or any Double methods that return BigIntegers

16:21 I imagine you could route it through a String, but that smells bad.

16:25 (.toBigIntegerExact (BigDecimal. (Math/ceil 1e23))) ==> 99999999999999991611392

16:25 :-/

16:25 but I guess that's the Double's fault

16:26 unlink: lawl

16:55 jkantz: ,(if (Boolean. "false") 1 2)

16:56 no clojurebot?

16:57 kotarak: on vacation

16:57 jkantz: is the above really supposed to return 2?

16:57 that's what clj 1.0 is doing

16:57 unlink: supposed to? I don't know. I do know what it does.

16:58 jkantz: gah, I mean it returns 1

16:58 in clojure 1.0

16:58 kotarak: (Boolean. "true") returns true, (Boolean. "something else") false => hence the above is correct.

16:58 Oh. 1 is wrong, no?

16:59 unlink: *some other string

16:59 jkantz: yeah, it seems there is only one false value

16:59 unlink: Not surprisingly, (Boolean. true) returns true

16:59 jkantz: (false? (Boolean. "False"))

16:59 => false

16:59 technomancy: triple negative?

17:00 jkantz: (true? (Boolean. "true") => false

17:00 unlink: hold your horses

17:00 new Boolean is almost always the wrong thing to use

17:00 user=> (false? (Boolean/valueOf "false"))

17:00 true

17:02 rtfm

17:03 Note: It is rarely appropriate to use this constructor. Unless a new instance is required, the static factory valueOf(boolean) is generally a better choice. It is likely to yield significantly better space and time performance.

17:03 i.e. it will return one of the interned values true and false

17:04 like, the same ones that the java compiler emits when it sees the literals "true" and "false"

17:05 jkantz: ok. yeah, reading through docs. but it what situation is the new constructor used? why is it even there?

17:05 unlink: If you need a new Boolean instance.

17:05 I can't think of any reason other than completeness' sake. I've never used it.

17:11 slurms: I have some example java code that looks something like new StyledText( shell, SWT/V_SCROLL | SWT/H_SCROLL) My problem is, I don't even know what that pipe is doing there. In the SWT docs I have some constructors have two or three of those. How do I emulate that in Clojure?

17:12 grrrt: that's a bitwise or I think

17:13 slurms: (new StyledText *shell* SWT/V_SCROLL) will work, when you only use one. BUt I'm looking to use two scroll bars here.

17:13 kotarak: (bit-or SWT/V_SCROLL SWT/H_SCROLL) or something like that, or simply (+ SWT/V_SCROLL SWT/H_SCROLL), not tested, though.

17:13 slurms: hmm

17:13 grrrt: I've used + which works with SWT

17:13 * slurms tries

17:13 kotarak: bitwise or is in this case equivalent to +, I think

17:14 grrrt: but bit-or is closer (well identical) to the | operator

17:14 slurms: oh, cause SWT/H_SCROLL is returning a number value (binary) ?

17:14 grrrt: yes

17:14 slurms: ah, ok

17:14 kotarak: For Boolean. issue: I think similar considerations apply to Integer. ...

17:15 slurms: hah.. worked

17:15 thank you so much.

17:16 FOO/MY_BAR can return whatever though right? I have to look up the docs or introspect to find out, right?

17:17 ozzilee: kotarak: I've got another issue with ClojureQL. Every query I attempt says "java.sql.SQLException: Operation not allowed after ResultSet closed".

17:17 kotarak: slurms: To be on save side, yes. Whenever the docs say FOO | BAR, it's numbers, which should work with +.

17:17 slurms: great. Thanks :)

17:18 ozzilee: The user I'm using has only SELECT privileges on the db, but that should be enough for (raw "SELECT * FROM foo"), yes?

17:18 kotarak: ozzilee, you have to realise the resultsetset-seq with doall before leaving the run form.

17:18 Yes. should work. I think laziness gets in the way here.

17:19 grrrt: kotarak, slurms: there is a small gotcha with using + though

17:19 kotarak: Try: (run *conn-info* (doall (query * foo)))

17:19 grrrt: binary 00000111 | 00000001 = 00000111

17:19 but 00000111 + 00000001 = 00001000

17:19 kotarak: ozzilee: or (with-connection [conn *conn-info*] (doall (execute-sql (query * foo) conn)))

17:20 grrrt: this will not happen for such constants.

17:20 grrrt: true, because they're all 2, 4, 8 etc etc

17:20 kotarak: grrrt: such constants have the form 00001 | 00010 ....

17:20 grrrt: that's why I said: "in this case"

17:24 ozzilee: (run *conn-info* (doall (query * foo))) gives me the same error. with-connection works.

17:25 kotarak: The first thing I posted was directly from Lao's post on the group, so either the post is wrong or something got broken.

17:25 kotarak: -.- run needs an overhaul...

17:26 ozzilee: you mean the compile error?

17:27 Or the raw?

17:28 ozzilee: kotarak: The raw. Lao's post doesn't use raw, but query has the same effect.

17:29 kotarak: ozzilee: you can try the same doall with with-connection and the raw. Or (run [*conn-info* results] (raw "SELECT * FROM foo") (doall results)).

17:29 Lau BTW, not Lao ;)

17:33 ozzilee: kotarak: Ah, ok. That works. Thanks. Not sure where I got Lao from either...

17:34 kotarak: ozzilee: beware changes in JOIN and run syntax... But the possibilities are interesting.

17:37 ozzilee: kotarak: Indeed, it looks like it's going to be pretty powerful.

17:38 kotarak: ozzilee: although, it's probably not an option if you need to go for the last tweak... It provides some abstraction. That comes with a cost. But FULL JOIN with Derby (which doesn't support it) is nice! :)

17:43 ozzilee: kotarak: I imagine I'm going to try pushing it pretty hard, so I'll let you know what I hit my head against :-)

17:44 kotarak: ozzilee: hehe, maybe the list were you didn't cut your fingers is shorter. ;) But that's good. I'd like to see some (more or less) real world examples, which we can test.

17:44 The code is not very well tested at the moment.

17:45 ozzilee: cut my fingers?

17:45 kotarak: hit your head against something, loads of rough edges.

17:47 ozzilee: kotarak: Ah, okay, gotcha :-) Yeah, I suppose we'll find out.

17:47 Is there a NOT NULL predicate?

17:48 kotarak: geez... NOT is missing.... Does foo <> NULL also work? (I'm not a SQL guru...)

17:50 ozzilee: kotarak: No, foo <> NULL doesn't work (on mysql, anyway).

17:50 kotarak: Ok. Will look into NOT.

17:51 ozzilee: "foo = NULL" doesn't work either. Needs to be "foo IS NULL".

17:52 * kotarak hates SQL.

17:53 ozzilee: Hahahaha. It is a bit... different.

17:53 technomancy: I think the word you're looking for is horrible

17:55 ozzilee: Nah, it's not so bad. It gets a bit difficult once you get up to a certain amount of complexity, but most things do.

17:56 I mostly run into problems with the implementations, as opposed to the language. Performance, mostly.

17:57 * ozzilee gives mysql a dirty look.

18:01 kotarak: I hate the manyfold differences in syntax of the different implementations. I'm not sure on which basis they claim to something called "SQL"...

18:01 ozzilee: Well, I've got a query working at least. Time to head out for the day, thanks for the help kotarak.

18:01 kotarak: ozzilee: np

18:01 be sure to file bug reports on github, thanks

18:01 ozzilee: kotarak: Will do.

18:02 btw, have you check out SQLAlchemy (Python)? It has some of the same ideas, namely the queries as higher-order functions.

18:02 *checked

18:02 kotarak: ozzilee: no, but will do.

18:03 ozzilee: kotarak: Cool. Later.

18:04 kotarak: cu

19:44 Anniepoo: I just got bit by trying to assemble a Swing GUI using map to go through the form. The map, of course, didn't actually execute because it's lazy. Is there a non lazy version of map?

19:49 emacsen: Is there a simple way to turn a vector of chars into a string?

19:50 texodus: (apply str [\a \b \c])

19:50 emacsen: texodus, wow... that was easy :)

19:50 texodus: ,(apply str [\a \b \c])

19:51 hrm, not good with clojurebot apparently

19:51 emacsen: works for me tho

19:53 cp2: clojurebot is mia

20:04 texodus: poor guy - I hope he's ok :)

20:06 Anniepoo: anybody in the bay area, the Clojure meetup starts at 5:45 at Runa

20:13 emacsen: Okay, one more dumb question. remove is great, but is there a way to do remove only on the first match?

20:13 so you don't remove all the instances of the match, just the first

20:22 texodus: not aware of one

20:23 (defn remove-first [col i] (if (= (first col) i) (rest col) (cons (first col) (remove-first (rest col) i))))

20:24 though there's prolly a better way

20:26 emacsen: yeah... what we need is a filter-first

20:26 but thanks

20:26 (I just read how remove works, it uses filter)

20:33 Knekk_: What are the normal uses for (constantly x) ?

20:36 It always returns the first-ever value it was invoked with?

20:36 user=> (def f (constantly x))

20:36 #'user/f

20:36 user=> (f 1 2 34 5 6)

20:36 1

20:36 user=> (f 2 )

20:36 1

20:37 oh I did (def x 1)

20:38 ok, I get it

21:27 texodus: emacsen. if you look at the code for take-while, it works the same way, so I guess that is pretyt much the best way to do it

21:31 emacsen: texodus, thanks

22:02 Chouser: has anyone compiled contrib's clojure-1.0-compatible branch? It's just completely failing for me

22:02 java.lang.NoSuchMethodError: clojure.lang.Namespace.importClass(Ljava/lang/Class;)Ljava/lang/Class;

22:04 jackdempsey: hmm

22:04 seems to work for me

22:05 git checkout origin/clojure-1.0-compatible

22:05 ant -Dclojure.jar=../clojure/clojure.jar

22:05 compiles fine

22:05 (and the clojure dir is up to date)

22:05 duck1123: does rich have the 1.0 branch, or is it somewhere else

22:05 I just fetched and i'm not seeing it

22:06 jackdempsey: not as a remote?

22:06 jack@localhost ~/git/clojure-contrib comp $ gb -a

22:06 ...

22:06 remotes/origin/clojure-1.0-compatible

22:06 and thats from rich

22:06 ( finally something i have half a clue about...)

22:08 Chouser: jackdempsey: hmph, I think that's exactly what I'm doing.

22:08 jackdempsey: hmmm

22:09 duck1123: Chouser: did you clean?

22:14 Chouser: ah, my 1.0 branch wasn't up to date.

22:26 oh, nope, still fails :-P

22:33 ok, git's just toying with me. I've got it now

23:04 gstamp: I saw this on a blog... (def fibs (lazy-cat [0 1] (map + fibs (rest fibs)))) ... and i'm not sure how it works. How can fibs refer to itself in that way?

23:05 Chouser: gstamp: sorry, I've got a reply started, but not done yet

23:07 maybe the best way to understand how it works is to realize that a lazy seq mutates internally

23:07 gstamp: lol. was that you're blog? I guess I can wait :-)

23:07 Chouser: :-)

23:08 each step of a lazy seq starts as a closure that knows how to compute a cons cell (first and rest)

23:09 when you ask for either, that closure is called

23:10 but then the closure is thrown away and its return value is cached instead

23:14 so as long as you refer only to parts of the seq that have realized, nothing has to recurse at that point.

23:16 and that's what happens with 'fibs' -- each element is computed only from earlier elements -- all the way back to the [0 1] constants.

23:16 oh

23:16 gone

23:16 :-P

23:17 liebke: add-break-thread works again, woot!!! :)

23:18 Chouser: liebke: nobody told me it was broken!

23:18 though I apparently I didn't notice either. :-P

23:18 liebke: my bad, I should have told you :)

23:18 I missed it

23:21 menSan: nn all

23:31 gstamp: Blasted http://webchat.freenode.net/ hung on me.

23:31 Chouser: gstamp: http://blog.n01se.net/?p=33#comment-1230

23:32 gstamp: does that make any sense?

23:32 gstamp: Thank you for that. I think I understand

23:33 I'll just keep looking at it for another hour to make sure. It's a beautiful piece of code.

23:34 Chouser: it is, though it does not perform very well

23:36 gstamp: Hopefully future versions of clojure can optimize that inconvience truth away.

23:37 unlink: speaking of optimizations, #^Object should be implicit for all parameters w/o type annotations

23:38 Chouser: unlink: would that help something?

23:39 unlink: .toString & friends by default use reflection

23:39 e.g. (fn [x] (.toString x))

23:41 getClass, hashCode too

23:46 Chouser: unlink: hm!

23:48 unlink: It's not often a function gets passed a non-Object in clojure ;)

Logging service provided by n01se.net