#clojure log - Mar 25 2009

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

0:01 slashus21: Anyone thought of adding some examples to clojure's wikipedia page?

0:05 durka42: clojurebot: examples?

0:05 clojurebot: examples is http://en.wikibooks.org/wiki/Clojure_Programming/Examples/API_Examples

0:06 durka42: those may be kind of old

0:06 slashus21: I made a syntax highlighting lang file for gedit.

0:07 It works pretty well. Still working on it.

0:07 durka42: nice

1:44 Gilbert987: does anyone know if there's a clojure interface for couchdb?

2:50 blbrown: Gilbert987, isn't couchdb a rest interface. probably wouldn't be too hard to cook one up using Java's URL libs

2:51 hiredman: there are java couchdb libs too

2:54 blbrown: hiredman, I have java code and I am invoking the clojure interpreter (legacy_script) is there a way to pass a Java object to that instance of the clojure script) Kind of as a global

2:55 hiredman: your java code is calling the clojure runtime?

2:56 blbrown: yea, just a copy of the clojure.main.java code

2:57 nevermind, I got it

3:04 OK, nevermind, dont have it. I thought I could do something with Var.invoke

3:06 Raynes: hiredman: You're still my hero.

3:06 hiredman: you want Var.intern I think

3:06 something like

3:06 Var.intern(Namespace.findOrCreate(Symbol.create("clojure")), Symbol.create("*console*"), console);

3:07 blbrown: nice

3:34 Raynes: c.c.duck-streams is the best thing evar.

3:41 All your clojure contrib are belong to us.

3:55 "Web browser in VB.NET" Failure failure, beep beep blue screen of death.

4:40 AWizzArd: ~seen cemerick

4:40 clojurebot: cemerick was last seen quiting IRC, 644 minutes ago

4:51 antifuchs: hmm, opening pages on the clojure wiki is a bit tedious with its cache control headers. who runs the server? (:

4:52 (opening /api takes 1.35 seconds. every time I load it. agh /-:)

4:54 AWizzArd: I think its hosted on a commercial wiki provider.

4:54 antifuchs: ah, wikispaces

4:55 AWizzArd: yup, that was the one

4:55 antifuchs: I'll ask them if they know their page load times are annoying (:

4:55 AWizzArd: *g*

5:05 antifuchs: right. hopefully they'll respond; this is getting slightly annoying /-:

5:21 hjlee: are there easy way to change radix? specifically decimal to binary.

5:38 AWizzArd: ,16rff

5:38 clojurebot: 255

5:39 AWizzArd: ,16r2

5:39 clojurebot: 2

5:39 AWizzArd: ,2r1611001

5:39 clojurebot: For input string: "1611001"

5:39 AWizzArd: nah

5:40 ,2r101011

5:40 clojurebot: 43

5:40 AWizzArd: hjlee: does that help?

5:40 hjlee: i want to get 101011 from 43 ...

5:40 antifuchs: hjlee: also, I see that cl-format has print radix control

5:41 AWizzArd: ,(Integer/toBinaryString 43)

5:41 clojurebot: "101011"

5:41 hjlee: that's it, thank you!

6:22 antifuchs: ok, so hm. is there a way to create a .class file (via compile of a .clj file containing a gen-class form) from swank-clojure without tearing my hair out over classpath issues?

6:29 Holcxjo: I get the impression that the classpath problems are one of the biggest hurdles for clojure -- why oh why can't that be adjusted at runtime?

6:31 antifuchs: suppose that would require a custom classloader thing and then other things would break

6:32 under the circumstances, it may be the best thing for now. it's just a pretty grating experience (:

6:34 ok, so how /do/ I add a directory (not a jar) to the classpath with swank-clojure?

6:34 hoeck: antifuchs: just set your classpath with [m-x] customize-variable [ret] swank-clojure-extra-classpaths

6:35 and restart emacs and slime

6:35 antifuchs: hoeck: but that will disable autoloading of the stuff in ~/.clojure - right?

6:35 or let's say auto-classpathing

6:36 hoeck: I don't know, who calls ~/clojure, swank-clojure or clojure.lang.main?

6:36 antifuchs: that's swank-clojure, IIANM

6:38 ok. I may have made an error setting the current classpath. is there a way to inspect the classpath that's currently set?

6:43 hoeck: there is (System/getProperty "java.class.path")

6:44 antifuchs: oh, thats cool

6:44 seems like the extra paths I'm pushing to swank-clojure-extra-classpaths isn't getting passed through to java.

6:45 so... is anybody using swank-clojure?

6:45 from the things I'm encountering, it looks like it's not really actively improved. /-:

6:46 hoeck: you need to restart emacs, swank-clojure-extra-classpaths is only evaluated when emacs loads the swank-clojure.el file

6:47 antifuchs: yes

6:48 I'm familiar with emacs. been using it for lisp development for 10 years (:

6:48 I also need to restart emacs after ,quit in the clojure swank repl. this is what was giving me that above impression (:

6:49 (need to restart because the next time I run clojure with slime, connecting to the swank port will hang)

6:52 hoeck: and you're right about the jars in ~/.clojure, they are ignored if you customize the extra classpath

6:53 antifuchs: what did was remove the 'clojure implementation from the swank impl alist, and add another one which works to my specifications

6:53 this isn't hyper-ideal, but for now it'll work well enough (:

6:55 hoeck: I have written a function for to do that for me, restarting emacs after changing the classpath for clojure is a bit odd

6:56 antifuchs: yeah. it's a bit annoying having to restart emacs after quitting clojure, too, let me tell you (:

6:57 hoeck: are the classpaths showing up in the string produced from the swank-clojure-cmd function?

6:58 antifuchs: now they are.

6:58 I'm thinking this may be easier if I just define one command to start clojure for each project

7:03 hoeck: I have always one clojure instance running, the classpath and clojure-package system works very well in separating all the projects

7:08 antifuchs: yeah, I was (naively) hoping I could avoid the whole initial setup cost.

7:08 mib_5zec882q: Anyone know if the ants demo still works with the latest version of clojure?

7:08 http://clojure.googlegroups.com/web/ants.clj?gda=AH2v2joAAAC2LrkjeC7f10uHiY7GOiyx4PWDLdmXu1edmOpq5JPfP-9OU0NQiFWgQuhmPR7veGf97daDQaep90o7AOpSKHW0

7:08 I just get a swing window with a blue square on it

7:08 but no ants

7:09 AWizzArd: did you look at the comments at the bottom?

7:09 (def ants (setup))

7:09 mib_5zec882q: no, I started at the top and didn't get to the bottom yet :)

7:09 AWizzArd: there you find commented out some calls that create the ants and start the simulation

7:09 mib_5zec882q: thanks, that should help

7:12 AWizzArd: Although that file could really need a little update. For example we find (. Thread (sleep evap-sleep-ms))

7:12 mib_5zec882q: Cool, I have ants now!

7:12 Thanks

7:12 AWizzArd: But today we would say: (Thread/sleep evap-sleep-ms)

7:13 Btw antifuchs, nice to have you here :)

7:14 mib_5zec882q: Also, a comment at the top saying "see the bottom of the file for how to run it" would have helped

7:14 antifuchs: AWizzArd: and you (:

7:14 I'm glad I finally have the time to work with clojure a bit (:

7:31 hm, how can I get the full path of a file whose (compile ) failed?

7:33 AWizzArd: What error message do you get when you try to compile?

7:36 It should tell you at least the filename + linenumber

7:46 clojurebot: max people

7:46 clojurebot: max people is 164

7:48 antifuchs: just the basename. I was hoping for a full path

7:52 leafw: antifuchs: grep is your friend then

7:56 antifuchs: leafw: like pathnames are never ambiguous

7:58 AWizzArd: but should there not be just one .clj file of that name in your classpath?

7:59 antifuchs: not sure

8:05 hoeck: AWizzArd: not necessarily, there could be hoeck/lib.clj and awizzard/lib.clj in the classpath, and the compiler just says error in "lib.clj"

8:17 AWizzArd: hoeck: but they will sit in different namespaces? (ns de.clojure.hoeck.lib ...) vs (ns de.clojure.awizzard.lib ...)

8:18 and when /home/clj/src/ is in the classpath, and when I know what I am compiling, I could dive into those dirs: emacs /home/clj/scr/de/clojure/hoeck/lib.clj

8:21 antifuchs: I think my problem was related to now knowing what I was compiling (:

8:21 anyway... that works now, thank goodness

8:21 now I'm getting odd ClassFormatErrors when I try to load the class files that compile emits

8:21 java.lang.ClassFormatError: Repetitive method name/signature in class file org/reprap/artofillusion/SwankTool (NO_SOURCE_FILE:0)

8:23 what is up with that? (-:

8:24 Holcxjo: Very Ruby-esque: Don't repeat yourself!

8:25 antifuchs: I wasn't meaning to (:

8:26 huh, seems like the :methods clause in ns does that. will debug further

8:26 Holcxjo: The error messages often have a hint of a Zen-like riddle.

8:26 How new-age is that?

8:31 antifuchs: it goes away once I remove the method declaration clause in ns. no idea if it's really needed. hm (:

8:34 hoeck: antifuchs: are you generating a class with additional methods?

8:53 antifuchs: yes, I am

8:56 oh wow, and look here. I have created my first .jar plugin to Art of Illusion that includes clojure (:

8:57 didn't have to write any java code at all; that wasn't too hard, let's see if I can get a Swank repl now (:

8:58 * hoeck congratulates

9:12 antifuchs: thanks - most of what kept me were misunderstandings that you all helped clear up (:

9:28 AWizzArd: antifuchs: grats to your .jar

10:08 cconstantine: Does anyone know a good resource for a high school kid who wants to get into programming (not clojure programming, programming in general)

10:11 leafw: cconstantine: a shell, gcc, an good text editor, and python/java/sbcl ... and their webpages with docs, that can give you most of what you need to tinker. Having a task in mind to accomplish will help a lot too.

10:11 gnuvince: cconstantine: http://greenteapress.com/thinkpython/thinkCSpy/html/

10:12 cconstantine: leafw: I'm pretty sure this kid only has a windows machine so things like a decent shell and/or gcc could be hard to get working and just a frustration before they hit any kind of 'real' programming

10:12 Holcxjo: cconstantine: Squeak is meant for that sort of audience as well I believe

10:13 cconstantine: gnuvince: that looks like it might be useful

10:14 Holcxjo: thanks, looking into it

10:14 antifuchs: cconstantine: squeak is a really cool environment to get impressive results quickly

10:14 cconstantine: antifuchs: that'll be important :)

10:14 antifuchs: immensely. you get a really nice multimedia environment, and modifying it is really quite easy

10:17 Holcxjo: Danger is of course, that Smalltalk is much like Lisp in that it spoils you for life -- going back to other languages will be difficult

10:18 cconstantine: lol

10:18 hook 'em while they're young!

10:19 Holcxjo: You laugh -- but then they'll be made to program in Java or similar. Their peers will think it's cool to churn out mega bytes of interfaces, declarations etc but they are doomed to a life of misery.

10:20 gnuvince: Holcxjo: don't you know that IDE-oriented programming is the best new programming paradigm?

10:20 Why have a better language when you can click on a button and Eclipse spits out the 20 lines of code you need, every time you need them?

10:21 antifuchs: s/misery/productivity/ (-:

10:24 AWizzArd: Well, as they have no macros they need the ide to insert/maintain the code.

10:24 People want Lisp, they just don't know it.

10:24 gnuvince: AWizzArd: they don't even have closures

10:25 cconstantine: and to think there was a time that I thought having code blocks that looked almost identical was a good thing

10:26 gnuvince: that looked almost identical?

10:27 cconstantine: function after function, or class after class that were identical except for a single minor detail (like a different type)

10:27 Chouser_: sure, it indicates you've found a key pattern in your design

10:27 gnuvince: cconstantine: oh

10:27 Chouser_: ...and failed to abstract it. ;-)

10:27 cconstantine: right

10:28 AWizzArd: Although, I think Java had closures right from the beginning.

10:29 gnuvince: Java doesn't have closures

10:29 hoeck: but java has objects

10:29 AWizzArd: via anonymous inner classes

10:29 they are just an ugly syntax for closures

10:31 digash: Anonymous classes are not clojures.

10:31 cconstantine: would you call that a language feature though? I believe you can do that in C++ too

10:31 AWizzArd: http://kawagner.blogspot.com/2006/08/closures-for-java.html

10:32 Chouser_: wow, Java's had those much longer than I thought

10:32 AWizzArd: digash: In Java closures are called anonymous inner classes.

10:33 digash: As I said anonymous inner classes are not closures.

10:33 cconstantine: You can use closures in any language, even ASM... (a turing complete language is a turing complete language and all) but I wouldn't say ASM has closures

10:33 digash: They do no enclose environment.

10:33 gnuvince: http://video.google.com/videoplay?docid=4051253555018153503&hl=en

10:33 AWizzArd: digash: wrong, they do

10:33 gnuvince: That guy explains why anonymous inner classes fail at being closures.

10:34 AWizzArd: The guy in the blog explains why they are closures :)

10:34 gnuvince: AWizzArd: who's that guy?

10:34 AWizzArd: gnuvince: http://kawagner.blogspot.com/2006/08/closures-for-java.html

10:35 gnuvince: I linked you Neal Gafter

10:35 AWizzArd: and funny, Neal Gafter, who made your video, commented in the blog :)

10:35 gnuvince: http://gafter.blogspot.com/2006/08/closures-for-java.html

10:35 SethTisue: fwiw, inner classes weren't in Java "in the beginning" (Java 1.0)

10:36 they were added in Java 1.1. just a minor point

10:36 cconstantine: that's still pretty early

10:37 digash: AWizzArd: http://en.wikipedia.org/wiki/Closure_(computer_science)

10:42 AWizzArd: by that definition Java anonymous classes are not closure, but of cause you can call object closure and Java is a dialect of List because it has GC.

10:42 AWizzArd: Gafter seems to have no arguments for what his Closures can do and which is not possible yet with anon inner classes. He has good arguments why it is more practical.

10:43 But if we log at his blog entry, then we see Karsten Wagner responding there, explaining why Java already has full closures and why Gafters proposal is a bad idea.

10:43 sellout: I'm pretty sure Java doesn't have closures, since they recently decided not to add closures in Java 7.

10:44 AWizzArd: sellout: you can use anonymous inner classes to close over some environment.

10:45 digash: the main point is that closures are functions and not objects.

10:45 AWizzArd: Or even better: use Clojure :)

10:45 digash: yes, ugly to use

10:45 if the language had explicit support for closures it would make it much more usable.

10:46 Chouser_: but in clojure, functions *are* objects

10:46 AWizzArd: anyway, as we see, some people are still against explicit support, because that will also come with disadvantages.

10:46 http://gafter.blogspot.com/2006/08/closures-for-java.html?showComment=1156164900000#c115616492091390744

10:47 digash: Chouser_: that is an implementation detail.

10:47 AWizzArd: Chouser_: hidden from the Clojure user

10:48 here another nice example: http://kawagner.blogspot.com/2006/08/some-more-remarks-on-closures-in-java_22.html

10:48 digash: I will ask Guy Steel once the talk is over if he thinks that anonymous classes in Java are closures.

10:49 leafw: digash: don't waste his time, it's semantics.

10:49 jsankey: at the end of the day, it doesn't matter whether they are closures or not, really

10:49 they still suck for various reasons :)

10:50 Chouser_: right, java anon classes do what they do, Clojure fns do what they do. Doesn't matter too much what they're called.

10:50 AWizzArd: I really suggest to write the application in Clojure instead of Java.

10:51 jsankey: arguing against the closures for java proposals on the basis that anonymous inners are closures misses the point

10:52 AWizzArd: jsankey: he mentioned other arguments

10:52 Chouser_: a better argument might be: if you want to write that kind of code why not use Clojure?

10:52 jsankey: Awizzard: sure, there are valid arguments against them to weigh

10:53 but saying java already has closures is meaningless

10:53 Chouser_: jsankey: how about: java already has anonymous inner classes that can be used in a lot of cases where you might otherwise use a closure?

10:54 jsankey: because the fact is that using anonymous inners to implement higher-order functions and custom control structures ranges between painful and impossible

10:54 Chouser_: hm.

10:54 digash: if you have closures you do not need Visitor pattern and some other ones.

10:55 jsankey: higher order functions are painful because functions aren't first class

10:55 AWizzArd: jsankey: I can agree on that humans don't like to do it this way, as there is too much syntax. But they still can technically be used for what one wants to do with closures.

10:55 Chouser_: Java doesn't have functions, does it?

10:55 jsankey: so you need to define interfaces describing your function types

10:56 (not to mention that almost none of these have a standard defintion, so lots of peopledefine their own)

10:56 Chouser_: it has methods instead

10:56 AWizzArd: digash: But the visitor pattern is mostly about method dispatch. What does this have to do with closures?

10:56 jsankey: AWizzArd: technically is right. but i care more about practically :)

10:57 Chouser_: yes, there are no free functions, everything is a method on some type

10:57 AWizzArd: I think nearly 100% of all people care about this. That is probably the reason why anon inner clases are very rarely used in place of closures.

10:57 So yeah, forget Java, write it in Clojure :)

10:57 * AWizzArd cheers

10:57 jsankey: :)

10:58 brianh2: it is possible (if incredibly painful) to do functional programming in raw java code

10:58 jsankey: i guess i would say i understand the desire to add more support for closures to Java, as is would allow higher-level abstractions

10:58 AWizzArd: brianh2: yes, although this is now coming closer to the turing argument :)

10:58 cconstantine: brianh2: first step; write a clojure compiler in java.

10:58 jsankey: but in reality, yeah, just use clojure!

10:59 Chouser_: second step: write a clojure compiler in clojure

10:59 brianh2: cconstantine: even without a compile step ;)

11:00 cconstantine: hehe

11:02 brianh2: the main problem is the poor sucker who has to come behind me and re-write all my code 'cuz they don't have a clue about what the code is really doing.

11:03 AWizzArd: apropos closures: very often, it seems, people think that anon functions are closures. Things like #(...) or (fn [x y] ...). In general this is not the case.

11:04 Chouser_: more semantics? when it closes over one binding it becomes a closure?

11:04 AWizzArd: Well, at least I personally I would find this to be a better definition.

11:04 Because otherwise really every function is a closure, even if it just captures the empty environment.

11:05 Chouser_: but that's how it actually is, at least in clojure -- every FnExpr has a 'closes' map, which is frequently empty.

11:06 AWizzArd: Yeah, in that case I would call it a function.

11:06 Closures are a subset of functions, and not the set of functions.

11:06 A true subset I mean

12:52 Cark: ~def str

12:54 are ou there hirdeman ?

12:56 ~seen hiredman

12:56 clojurebot: hiredman was last seen in #clojure, 589 minutes ago saying: Var.intern(Namespace.findOrCreate(Symbol.create("clojure")), Symbol.create("*console*"), console);

13:19 st3fan: ola .. is there a downloadable copy of the documentation available?

13:19 i'm traveling and i would like to read things while not online

13:19 leafw: st3fan: there's a pdf somewhere in the website

13:19 otherwise, wget spider is your friend :)

13:19 st3fan: hmyeah

13:24 what is the official clojure-contrib project? there is one on github, sf and google code

13:26 Cark: google code

13:28 st3fan: got it

13:34 dhea: has anyone had any luck getting clojure + processing working?

13:36 hoeck: dhea: yes

13:37 dhea: http://github.com/rosado/clj-processing/tree/master

13:37 dhea: using it, im having a problem getting the core.jar and processing.jar loaded

13:37 added it to classpath, cant find it though

13:40 hoeck: is it in (System/getProperty "java.class.path")

13:42 are you using clojure with emacs-slime or another ide or just from the shell?

13:42 dhea: emacs-slime

13:42 its not in the java.class.path, but ive added the jars with add-classpath

13:49 st3fan: is clojure-mode good to start? i'm a little afraid of slime and for now i just want to type and eval things

13:49 Cark: that's what i'm using

13:50 Chouser_: st3fan: that's what rhickey uses.

13:52 dhea: I get the error: java.lang.RuntimeException: You need to use "Import Library" to add processing.core.PGraphicsJava2D to your sketch. (NO_SOURCE_FILE:0)

13:54 hoeck: dhea: are you trying the example from clj-processing?

13:54 dhea: i've add a (:import (processing.core PGraphicsJava2D))

13:54 yes, simple.clj

13:55 the swing frame is up and running, its only when the (.init p5-applet) line is called

13:55 hoeck: mhh, maybe better adding the jars to the clojure-swank-extra-classpaths or into ~/.clojure/

13:56 dhea: jars in ~/.clojure are auto added to classpath?

13:57 hoeck: yes, by default, swank-clojure builts a list of all jars from ~/.clojure/

13:58 or you do [m-x] customize-variable [ret] swank-clojure-extra-classpaths

13:59 but you must to restart emacs after resetting the classpath

13:59 dhea: yup, that did it

14:00 excellent

14:00 thank you sven hoeck

14:02 lisppaste8: hoeck pasted "swank-clojure-cmd-update-classpath" at http://paste.lisp.org/display/77591

14:02 hoeck: np, or use ^^^ that function instead of restarting emacs

14:03 btw, for processing, the P3D renderer is much faster (for me) than the default P2D

14:13 antifuchs: haha, the wikispaces.com guys deny that clojure.com is hosted by them... should I be worried? (-:

14:13 Chouser_: clojure.org?

14:14 antifuchs: erm, yeah

14:14 mistypo'd (:

14:14 Chouser_: well, I guess it redirects

14:16 cads: may I load a jar into the classpath from inside the repl?

14:16 antifuchs: it's probably just a little misunderstanding, but I'm amused that their helpdesk guy would point me to a page that has their email prominently displayed (-:

14:17 hiredman: cads: there is add-classpath but it is not reliable

14:18 antifuchs: is there a rule for when add-classpath will work or not?

14:22 cads: thanks hiredman, I'll try it out

14:24 cemerick: antifuchs: short of having control over your classloader (moreso than you get from java's boot classloader), I think it's generally best to assume that add-classpath will never work.

14:24 antifuchs: good to know. thanks

14:24 cemerick: if you have stuff in a jar, there's no reason to not simply include it in your classpath.

14:24 antifuchs: right

14:25 hiredman: cconstantine has the highest non bot url count

14:25 cemerick: if you want to be able to hot-deploy clojure code that depends on a new jar, then set up your deployment environment so that the classpath points at a directory, and unzip your jar(s) into that directory before attempting to load them from your (completely dynamically loadable) clojure.

14:26 that's likely something that could be made very clean with a library, but I don't think anyone's bothered with such a mechanism yet.

14:27 cconstantine: hiredman: hu?

14:29 hiredman: cconstantine: for the channel, the only nick that has pasted more urls then you is lisppaste8

14:29 (this is only in the last week or so)

14:30 cconstantine: ah

15:25 lisppaste8: slashus2 pasted "with-out-file" at http://paste.lisp.org/display/77597

15:28 slashus2: Accidentally hit enter when trying to format it :-(

15:28 Would it be good practice to make the boolean-append to be optional?

15:31 Chouser_: maybe an optional :append flag? maybe put it before the filename?

15:39 pop quiz: what does the following return? (no fair trying it out): (binding [* '+] (* '{1 / - * + 6}))

15:39 gnuvince: probably an error?

15:39 You can't call a symbol

15:40 kotarak: I would think so, too.

15:40 hiredman: fools

15:40 +

15:40 Raynes: I'll go with Error.

15:40 gnuvince: oh

15:40 cemerick: hrm

15:40 hiredman: you can call a symbol

15:40 kotarak: Symbols are functions of maps?

15:40 gnuvince: ,(ifn? 'foo)

15:40 clojurebot: true

15:40 hiredman: just like a keyword

15:40 Raynes: It returns 6.

15:40 gnuvince: 6 then?

15:40 Raynes: >_>

15:41 hiredman: fine

15:41 Chouser_: 6 it is.

15:41 now wasn't that fun?

15:41 cemerick: Chouser_: that could be a regular blog series for you :-P

15:41 gnuvince: Pop quiz: what does this do: zip.ap fmap.(id &&& wtf)

15:42 Raynes: gnuvince: Eats my kittehs.

15:44 * agriffis_ figures out Chouser_'s quiz

15:45 kotarak: The depths of undocumented features...

15:47 gnuvince: What's undocumented?

15:47 Raynes: Why are virtually all twitter clients built on AIR?

15:48 :|

15:48 I don't like AIR.

15:48 It's too AIRy.

15:48 slashus2: plus is assigned 6, but since the symbol plus is bound to * retrieving * is actually retrieving '+ from the map.

15:48 kotarak: gnuvince: I think that Symbols are callable. I only knew for keywords.

15:48 gnuvince: or did I miss that on the api page?

15:49 hiredman: http://clojure.org/data_structures <-- under Symbols

15:49 gnuvince: Yeah they are

15:50 I was confusing with the fact that lists are not callable

15:50 ,(ifn? ())

15:50 clojurebot: false

15:50 kotarak: Hmm... I should reread that from time to time.

15:50 hiredman: clojurebot: symbols?

15:50 clojurebot: It's greek to me.

15:50 hiredman: clojurebot: symbols is <reply>http://clojure.org/data_structures#toc10

15:50 clojurebot: Alles klar

15:51 slashus2: ,(ifn? '+)

15:51 clojurebot: true

15:51 hiredman: ,(ifn? +)

15:51 clojurebot: true

15:51 slashus2: ,(ifn? 'a)

15:51 clojurebot: true

15:52 Chouser_: ,(ifn? #'+)

15:52 clojurebot: true

15:52 Chouser_: ,(#'+ 2 4)

15:52 clojurebot: 6

15:52 Chouser_: ,(@#'+ 2 4)

15:52 clojurebot: 6

15:52 hiredman: ,(class +)

15:52 clojurebot: clojure.core$_PLUS___3313

15:53 hiredman: bah

15:53 ,(class #'+)

15:53 clojurebot: clojure.lang.Var

15:53 Chouser_: ,(class @#'+)

15:53 clojurebot: clojure.core$_PLUS___3313

15:53 hiredman: ,(map #(.getName %) (.getMethods clojure.lang.Var))

15:53 clojurebot: ("applyTo" "call" "alter" "bindRoot" "pushThreadBindings" "popThreadBindings" "setTag" "setMeta" "setMacro" "isMacro" "fn" "internPrivate" "setValidator" "doSet" "hasRoot" "unbindRoot" "commuteRoot" "alterRoot" "releaseThreadBindings" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "invoke" "

15:56 slashus2: ,@#'*ns*

15:56 clojurebot: #<Namespace sandbox>

17:44 lisppaste8: rich_holygoat pasted "Confusing sort behavior" at http://paste.lisp.org/display/77605

17:45 rich_holygoat: could someone more experienced than I comment on this apparently-odd behavior of sort on a map?

17:45 the default comparator sorts just fine, but explicitly specifying a comparator does not.

17:45 this is with r1337.

17:46 stuhood: rich_holygoat: the comparator would need to accept Map.Entry

17:46 i think

17:46 kotarak: The map is turned into sequence of map entries. Obviously the Comparator expects a String => Failure. (at least the is my understanding of the error messages)

17:46 Maybe you want a sorted-map?

17:47 rich_holygoat: ah yes, and the ordinary sort actually calls compare on the map entries

17:47 kotarak: apparently I do :)

17:48 is there literal syntax for a sorted map, or a function to turn an associate thing into one without having to pass the contents as args to sorted-map?

17:48 s/associate/associative

17:49 hiredman: ,(class (into (sorted-map) {:a 1 :b 2}))

17:49 clojurebot: clojure.lang.PersistentTreeMap

17:49 kotarak: (into (sorted-map) a-normal-map)

17:49 rich_holygoat: ah, thank you folks

18:41 stuhood: ~def into

18:42 ~def c.l.PersistentTreeMap

19:34 cconstantine: ~def union

19:36 So, how is into different from union... from a caller's perspective?

19:38 hiredman: ,(doc clojure.set/union)

19:38 clojurebot: ping?

19:38 cmvkk: sudden death

19:38 clojurebot: "([] [s1] [s1 s2] [s1 s2 & sets]); Return a set that is the union of the input sets"

19:38 PONG!

19:39 sohail: the JVM was collecting garbage

19:39 antifuchs: there can be a lot of that on irc channels (:

19:47 * sohail squints at antifuchs

19:51 antifuchs: yes, hello sohail (:

19:51 * sohail squints at antifuchs

19:52 cconstantine: Gotta love seeing java take 200+% CPU

21:11 ~def time

22:06 I'm making a macro, and I"m using '"`" to quote a block, how do I get a sub-expression to unquote and execute at macro-expand time?

22:07 Chouser: ~

22:07 clojurebot: rest never returns a unicorn

22:07 Chouser: ha!

22:10 cconstantine: thanks

22:11 grr, it appears to work in macroexpand.. and doesn't when I make the call

22:13 ,(defmacro foo [expr] `([ ~expr ~(str expr)])) (macroexpand '(foo asdf))

22:13 clojurebot: DENIED

22:14 durka42: when you do (foo asdf) it says unable to resolve asdf, which is true

22:14 ([asdf "asdf"]) isn't going to work

22:14 cmvkk: you're beter of doing `([ ~expr (str ~expr)])

22:17 cconstantine: bah, getting the right "layer" of macro execution is hard

22:17 cmvkk: it is slightly annoying because you can't do macros that require the value of the arguments passed to them;

22:18 since those values aren't available at compile-time. in common lisp you could do it that way i'm pretty sure

22:18 because in CL there's no distinction between compile time and run time, etc..

22:19 cconstantine: cmvkk: there is... I just want the expression as a string, and the expression and is confusing

22:20 cmvkk: oh i see. you don't want the value, you just want the string of the expression itself.

22:20 cconstantine: right

22:21 I'm writing a macro that records runtimes for expressions to a global map of accumulcated time-to-execute

22:23 cmvkk: (def(defmacro foo [expr] vector ~expr ~(str expr)))

22:23 err, ignore the first 4 characters

22:23 and it's missing `(

22:23 darn you 'screen'

22:24 cconstantine: hehe

22:24 cmvkk: that should do it though, the problem i was getting was that you were expanding to a vector inside a list...meaning it was trying to call the vector

22:25 ....and i now see that durka42 already explained this.

22:32 cconstantine: so I'm doing my first sync, and ref-set inside my first macro... yay fun tim!

22:32 time

22:33 Chouser: :-)

22:34 cconstantine: i wanted ~(str expr)

22:35 cmvkk: ~(str expr) works, right?

22:36 cconstantine: yeah

22:36 I did (str ~expr) at first and got a string of the result

22:36 which for :a was the same ;)

22:38 now I have a macro (with-timing [expr]) that will return the result of expr, and accumulate time-to-execute and store it in a map keyed off the expr as a string

22:39 cmvkk: out of curiosity, how do you store the time to execute? i know (time ...) prints it out but...

22:39 lisppaste8: cconstantine pasted "with-timing" at http://paste.lisp.org/display/77615

22:40 cconstantine annotated #77615 "untitled" at http://paste.lisp.org/display/77615#1

22:40 cconstantine: that's stolen from the implementation of the 'time' macro

22:41 durka42: did you also write time-val

22:41 also, (sync nil ==> (dosync

22:41 cmvkk: is there any different currently between sync and dosync--

22:41 cconstantine: oh right!

22:41 cmvkk: you read my mind

22:42 durka42: cmvkk: what i said is the definition of dosync

22:42 lisppaste8: cconstantine annotated #77615 "time-val" at http://paste.lisp.org/display/77615#2

22:43 cconstantine: yeah... I'm always suspicious when an API doc says "just pass null/nil for this argument"... or when str-tok's docs say "don't use this"

22:43 cmvkk: one day, though, there might be flags for sync

22:43 cconstantine: I know, but in the meantime it raises flags for me ;)

22:44 durka42: "reserved for future use" means i can use that field for my own app-specific settings, right? ;)

22:44 * durka42 glares at the windows api

22:44 cconstantine: lol

22:45 so, I would love for this to make it into clojure... is it reasonable to have that global reference to a map?

22:45 or would it be better form to make some kind of structure that needs to be passed in?

22:46 durka42: global variables, two thubs down

22:46 thumbs

22:47 i dunno, carrying the structure around sounds unwieldy though

22:47 Chouser: you mean like *out*, *err*, *warn-on-reflection*, etc?

22:47 cconstantine: yeah

22:47 durka42: this macro smells like the beginning of a profiler library

22:47 cconstantine: part of hte problem is that timings are kind of intrinsically a side-effect

22:47 durka42: yes

22:48 durka42: with that raw data (or better data) you could get hings like average runtime, number of execs, max, min, yada yada

22:48 durka42: that would be cool

22:49 cconstantine: on any form... not just functions (which I'm assuming all the java-level profilers limit you too)

22:49 * durka42 wonders if a contrib library with with-timings and a few of those sorts of functions would be welcome

22:49 durka42: then you'd have clojure.contrib.timings/*all-timings* which sounds more reasonable

22:49 cconstantine: I'm a programmer by trade... so I understand how hurtful bad code can be to a code-base.

22:50 hell, I suffered through a 3 hour code review of stuff that was forced to be terrible by the terrible APIs it had to work with....

22:50 today

22:50 cmvkk: do most of the people in here program professionally, I wonder

22:50 durka42: i don't

22:50 not yet anyway

22:51 cconstantine: hehe

22:51 cmvkk: me neither

22:51 cconstantine: I'm threatening to rewrite our testing framework in clojure :)

22:51 No one there knows it, but the current java framework is pretty bad

22:51 *it == clojure

22:52 ah... clojure.contrib.timings... that would mean I'd need to learn namespaces :/

22:53 durka42: they're like java packages

22:53 correspond to folders

22:53 * cconstantine kinda sorta never learned those either.

22:53 cconstantine: packages... I know what folders are *bush*

22:53 *blush*

22:53 * cconstantine fail tping

22:53 cmvkk: I get namespaces but I totally don't understand all this classpath stuff

22:54 what annoys me is if I load a file with load-file, and that file defines a namespace, then i try to (use ...) that namespace

22:54 it doesn't work if the file isn't in the classpath.

22:54 cconstantine: hehe

22:54 durka42: right because use goes straight for the file

22:54 not that that's logical...

22:54 cconstantine: (use ...) tries to find the compiled clojure/java stuff?

22:55 durka42: if the files defines a namespace isn't it already used

22:55 cmvkk: I guess.

22:55 well it would be loading a file with a different namespace than the one i'm in already.

22:55 then what i really want to do is refer everything.

22:57 Chouser: if you loaded it with load-file, you can bring the symbols into your namespace with refer

22:57 (doc refer)

22:57 clojurebot: refers to all public vars of ns, subject to filters. filters can include at most one each of: :exclude list-of-symbols :only list-of-symbols :rename map-of-fromsymbol-tosymbol For each public interned var in the namespace named by the symbol, adds a mapping from the name of the var to the var to the current namespace. Throws an exception if name is already mapped to something else in the current namespace. Filters can be

22:57 cmvkk: so something like (load-file "bar-namespace.clj") (ns foo) (refer bar)

22:58 Chouser: I suppose.

22:59 you've got essentially different levels of api

22:59 cmvkk: i know it's the wrong way to do things, i don't actually do it that way.

22:59 but the fact of the matter is, i often know where my other files are much better than i know what the classpath is.

23:00 i like a system where all I need to know is where the other code is, rather than knowing what the classpath is.

23:00 Chouser: load-file works on files directly and in-ns and refer work on namespaces directly

23:01 ns, use, and require work on "libs" which are an abstraction that combine files and namespaces.

23:03 duderdo: What's the equivalent to 'assoc' from cl in clojure?

23:03 cmvkk: assoc

23:04 duderdo: really? Hm

23:04 cmvkk: ,(doc assoc)

23:04 clojurebot: "([map key val] [map key val & kvs]); assoc[iate]. When applied to a map, returns a new map of the same (hashed/sorted) type, that contains the mapping of key(s) to val(s). When applied to a vector, returns a new vector that contains val at index. Note - index must be <= (count vector)."

23:04 cmvkk: (assoc [1 2 3] 1 4)

23:04 durka42: no, CL's assoc is "get"

23:05 http://www.ai.mit.edu/projects/iiip/doc/CommonLISP/HyperSpec/Body/fun_assoccm_a_assoc-if-not.html

23:05 cmvkk: ohh yeah.

23:05 duderdo: ah that's what I thought

23:05 durka42: maps are callable too

23:05 duderdo: thanks.

23:05 durka42: ,({:a 1 :b 2} :a)

23:05 clojurebot: 1

23:05 durka42: ,(:b {:a 1 :b 2})

23:05 clojurebot: 2

23:05 cmvkk: i can't believe i've forgotten that much about CL already

23:06 * durka42 had to look it up

23:06 duderdo: I'm trying forget, I keep getting mixed up

23:07 Chouser: you already forgot "to", so you're well on your way.

23:09 duderdo: heh

23:11 Chouser: :-)

23:14 duderdo: How can I make a keyword from a string? Is this possible?

23:14 durka42: ,(keyword "boo")

23:14 clojurebot: :boo

23:14 duderdo: Sweet

23:15 Thanks guys. night.

23:15 cconstantine: ooo with a rich programmable timing infrastructure you could make something that takes a set of expressions that implement the same functionality and do automatic detection of fastest algorithm/source (like, an expression for each server to issue requests to, and it could choose the server that's tending to respond faster)

23:16 durka42: that would be really cool

23:16 cconstantine: or the parallel/non-parallel algorithms

23:19 durka42: (defmacro halts? [expr] `(= Double/POSITIVE_INFINITY (time-val ~expr))

23:19 i mean

23:19 (defmacro halts? [expr] `(< Double/POSITIVE_INFINITY (time-val ~expr))

23:19 no still backwards

23:19 failed joke :(

23:20 cconstantine: (defmacro halts? [expr] `(> Double/POSITIVE_INFINITY (first (rest (time-val ~expr)))))

23:20 there ya go :)

23:21 durka42: frest

23:21 also that macro might need some optimization :)

23:21 Chouser: fnext

23:21 durka42: oh yeah that

23:21 cconstantine: and should probably return the value of ~expr

23:22 but I get it... if it's runtime is less that positive infinity it halted

23:22 oh right, returns a bool

23:23 didn't halt... bah

23:25 ahh, the halting problem

23:25 someone at work today claimed they had a turing complete language that had solved it.

23:25 shockingly they were mistaken

23:26 durka42: (defmacro does-halt [expr] `(if (halts? ~expr) (loop [] (recur)) "Logical fallacy complete."))

23:27 * cconstantine golf claps.

23:27 durka42: http://www.reddit.com/r/programming/comments/6wamg/the_terminator_project_taking_on_the_halting/

23:27 cconstantine: that reminds me of using exec() in C... exec( ...); exit(1);

23:29 ie, if exec() returns, you have a problem.

Logging service provided by n01se.net