#clojure – 2010-10-14

2010-10-14

00:05	amalloy	kmc: hotspot's jit is pretty good, i think. but i'm not a performance expert by any means
00:12	rplevy	how far can one elaborate on something like this do you think? (defmacro with-turbo [turbo-on & body] `(let [~'map ~(if turbo-on `clojure.core/pmap `clojure.core/map)] ~@body))
00:12	rplevy	it's kind of the opposite of the Macintosh turbo button ;)
00:13	amalloy	rplevy: not sure what you mean by elaborate
00:13	rplevy	say if you don't know whether your code will be running on a quad core machine or a netbook and you want to provide an option
00:14	rplevy	amalloy: I mean, add more conditional stuff to take advantage of parallelism for free
00:14	amalloy	taking advantage of parallelism for free turns out to be a lot more difficult than it looks (cf kmc's efforts a few nights ago)
00:15	rplevy	amalloy: I'll take a look, sounds interesting
00:15	amalloy	rplevy: the main issue is that pmap needs to be given a smallish number of large tasks in order to be useful, because it spawns a thread for each task
00:15	amalloy	(it doesn't run all the threads at once, but it does create a new thread for each task)
00:16	rplevy	amalloy: what day's log is kmc's experiment in?
00:16	amalloy	uhhhhh
00:16	amalloy	sec
00:16	amalloy	mon oct 11
00:17	amalloy	on my computer, anyway, which is pacific time
00:17	amalloy	:P
00:17	amalloy	so in greenwich it's probably tuesday morning
00:18	amalloy	rplevy: search for pmap to find it quickly within the log file
00:18	rplevy	amalloy: yeah, so of course there is no free lunch, but still useful on occasion to wrapp small things in something like this, using judgment for specific cases
00:19	amalloy	rplevy: yes, clojure certainly provides a good groundwork for writing sensible parallel tasks
00:22	chouser	map has defined different semantics when the function you give it has side effects. pmap doesn't, I think.
00:22	chouser	clojure's got a lot more tools around concurrency at this point than it does around parallelism
00:23	_rata__	hi
00:23	amalloy	chouser: you don't think it's defined as "totally arbitrary and don't rely on it"?
00:23	amalloy	i mean, that's sorta the same as totally undefined, i guess
00:24	_rata__	records are persistent, right?
00:25	rplevy	chouser: interesting, it seems inevitable then for more high-level abstractions for parallelism to enter the language
00:27	amalloy	_rata__: yes, they are immutable, if that's what you mean
00:27	_rata__	amalloy, i mean, is it ok to use them with STM?
00:28	amalloy	_rata__: yes, because they are immutable. they implement java.util.map, which has a put() method, but if you actually call it they throw an exception
00:30	_rata__	i asked because of the optimization STM can do using persistent vectors/maps/lists
00:30	amalloy	what optimization is that?
00:30	_rata__	STM can do that optimization using any immutable data?
00:30	amalloy	(records implement IPersistentMap, so whatever they are it seems likely that STM can use them on records)
00:30	_rata__	not having to do copies
00:31	_rata__	ok
00:31	_rata__	*to make copies
00:31	_rata__	thanks
00:31	_rata__	I was unsure
00:31	amalloy	_rata__: i can't defrecord on clojurebot, but you can try (supers (class x)), where x is some defrecord'ed object
00:31	_rata__	ok
00:34	_rata__	good night
00:47	LauJensen	Good morning all
00:50	amalloy	morning lau. up early today?
00:51	amalloy	(note: i don't actually know what time it is where you are, but you're usually not around for two more hours or so)
00:52	LauJensen	amalloy: Yea I usually get here by 9AM but todays its 6:50AM now
00:54	notsonerdysunny	how can one create a soft-link in a cake-task?
00:55	LauJensen	soft link?
00:55	amalloy	as in the $ ln -s variety?
00:55	notsonerdysunny	LauJensen: yes soft link
00:55	LauJensen	amalloy: thats a symlink
00:55	KirinDave	Anyone here good with enlive?
00:55	notsonerdysunny	:) yes symlink
00:56	amalloy	LauJensen: i know, but it was likely to be what he meant
00:56	KirinDave	Having problems with nth-of-type.
00:56	LauJensen	(sh "ln" "-s" "src" "dest")
00:56	amalloy	i think you have to launch the shell; java doesn't have symlink feature
00:56	LauJensen	amalloy: yea you're just a lot smarter than me it seems :)
00:56	lancepantz	nah, you should use the ant task
00:56	amalloy	LauJensen: perhaps a lot dumber!
00:56	lancepantz	notsonerdysunny: http://ant.apache.org/manual/Tasks/symlink.html
00:56	lancepantz	so you would use cake's ant helper fn
00:57	LauJensen	lancepantz: any idea what that does on Windows?
00:58	amalloy	LauJensen: i wouldn't be surprised if it did what cygwin does
00:58	LauJensen	Which is?
00:58	notsonerdysunny	lancepantz: can you please give me the syntax for the cake's ant helper function to create a symlink
00:58	amalloy	which iirc is create a .lnk (windows shortcut file), with metadata attached to it that lets other cygwin processes treat it as a symlink
00:58	LauJensen	ah ok
00:58	notsonerdysunny	I think Windows 7 and vista support link
00:58	lancepantz	(ant Symlink {:link "foo" :resource "bar"})
00:59	notsonerdysunny	thanks lancepantz
00:59	lancepantz	LauJensen: there's a disclaimer at the end of the above link
01:00	amalloy	LauJensen: yeah, just checked my windows machine. ln -s creates a .lnk with some kind of special metadata
01:00	LauJensen	great.
01:01	amalloy	(on my 5-year-old XP system. sunny seems to think vista/7 are smarter)
01:11	notsonerdysunny	amalloy: I read about it some-where .. not very positive though
01:13	Bahman	Hi all!
01:14	LauJensen	Morning Bahman
01:14	Bahman	Hi there LauJensen!
01:21	chouser	NTFS has supported links forever, I think, but the rest of the OS and the applications all get very confused.
01:26	notsonerdysunny	chouser: so what is the command to create a symlink or hardlink in windows/NTFS then?
01:30	amalloy	chouser: this seems to be true. my ubuntu system can create a hardlink on the mounted ntfs drive that windows uses. whether or not windows can understand it i haven't tested
01:50	sandGorgon	is there any way for me to figure out what is the exact java command used by lein when I do "lein repl" in a project ? is there any verbose setting for example
01:52	LauJensen	sandGorgon: Well, there is an environment variable which you can set, to make it more verbose, it outputs the classpath at least. But I'd just run ps ax \|grep java if I wanted to know
01:54	kmc	you could use strace
01:54	sandGorgon	kmc, how do I use strace for that ?
01:55	amalloy	sandGorgon: or jps and then look at /proc/xxxxx/cmdline
01:55	kmc	strace -f -e execve -s 99999999 lein repl
01:57	sandGorgon	kmc, wow.. thanks that was great.
01:57	sandGorgon	LauJensen, thanks.. that worked as well ;)
01:59	LauJensen	np
02:05	sandGorgon	has anyone looked at the source code of cake ? it is kinda funky - for e.g. src/cake/tasks/jar.clj:53 why is "(ns cake)" a string, which is written into a file called build/jar/cake.clj when cake is invoked ?
02:09	notsonerdysunny	does the reader macro #' have a corresponding function
02:09	lancepantz	sandGorgon: yeah, so cake actually runs in two jvms
02:09	amalloy	,(get-var 'clojure.core/first)
02:09	clojurebot	java.lang.Exception: Unable to resolve symbol: get-var in this context
02:09	lancepantz	once for your project, and one that processes cake commands
02:09	amalloy	,(ns-resolve 'clojure.core 'first)
02:09	clojurebot	#'clojure.core/first
02:10	amalloy	notsonerdysunny: there's one that's more convenient than that, lemme go find it
02:10	lancepantz	then there is a ruby script that sits in front of those two, sending forms over the socket to get evaled in the jvms
02:10	LauJensen	,(= (var test) #'test)
02:10	clojurebot	true
02:10	notsonerdysunny	,(doc var)
02:10	clojurebot	notsonerdysunny: I don't understand.
02:10	notsonerdysunny	->(doc var)
02:10	sexpbot	⟹ "Special Form: Please see http://clojure.org/special_forms#var"
02:10	amalloy	,(var-get 'clojure.core/first)
02:10	clojurebot	java.lang.ClassCastException: clojure.lang.Symbol cannot be cast to clojure.lang.Var
02:11	kmc	what does sexpbot do?
02:11	notsonerdysunny	->(doc var-get)
02:11	sexpbot	java.lang.SecurityException: Code did not pass sandbox guidelines: (#'clojure.core/var-get)
02:11	LauJensen	,(-> #'var meta :doc println)
02:11	clojurebot	java.lang.Exception: Unable to resolve var: var in this context
02:11	amalloy	roughly the same as clojurebot, with a different codebase
02:12	notsonerdysunny	var is a special form ..
02:12	notsonerdysunny	so I guess we can't do #'var
02:12	sandGorgon	lancepantz, all right, but then why isnt there an inherent cake.clj in the project itself ? I dont think cake can be run from the commandline (using java commandline) because there is no cake.clj
02:12	notsonerdysunny	LauJensen: *
02:13	LauJensen	,(-> #'loop meta :doc println)
02:13	clojurebot	Evaluates the exprs in a lexical context in which the symbols in the binding-forms are bound to their respective init-exprs or parts therein. Acts as a recur target.
02:13	LauJensen	-> (doc loop)
02:13	sexpbot	⟹ "Special Form: Please see http://clojure.org/special_forms#loop"
02:13	amalloy	,(find-var 'clojure.core/first)
02:13	clojurebot	#'clojure.core/first
02:13	LauJensen	notsonerdysunny: ^^
02:14	amalloy	notsonerdysunny: the functions you probably want are intern, find-var, and ns-resolve
02:15	notsonerdysunny	thanks amalloy and LauJensen
02:25	amalloy	,(zipmap (map (juxt identity -) (range 4)))
02:25	clojurebot	java.lang.IllegalArgumentException: Wrong number of args (1) passed to: core$zipmap
02:25	amalloy	,(apply zipmap (map (juxt identity -) (range 4)))
02:25	clojurebot	java.lang.IllegalArgumentException: Wrong number of args (4) passed to: core$zipmap
02:25	amalloy	blahhhh
02:26	amalloy	,(apply hash-map (map (juxt identity -) (range 4)))
02:26	clojurebot	{[0 0] [1 -1], [2 -2] [3 -3]}
02:27	LauJensen	amalloy: let me guess, you used to do a lot of perl ? :)
02:27	kmc	,(doc juxt)
02:27	clojurebot	"([f] [f g] [f g h] [f g h & fs]); Alpha - name subject to change. Takes a set of functions and returns a fn that is the juxtaposition of those fns. The returned fn takes a variable number of args, and returns a vector containing the result of applying each fn to the args (left-to-right). ((juxt a b c) x) => [(a x) (b x) (c x)]"
02:27	amalloy	LauJensen: hah, not actually very much. why, am i trying to do something like perl?
02:27	kmc	cool function
02:28	amalloy	kmc: yes, it's one to keep in mind. it turns out to be useful in a lot of interesting ways
02:28	LauJensen	amalloy: I once did something like what you just did, ie. tweaking until something working came out. Someone rebuked me saying "You need to understand what you're doing, not just tweak until it works, thats perl development"
02:28	amalloy	,(map (juxt inc dec) (range 5))
02:28	clojurebot	([1 -1] [2 0] [3 1] [4 2] [5 3])
02:28	kmc	amalloy, yeah, i've used something similar in Haskell
02:29	amalloy	LauJensen: in perl, tweaking is your only option. clojure gives you the choice to understand what you're doing...or just keep pounding on the keyboard like i was doing
02:29	kmc	heh
02:29	LauJensen	amalloy: well, im just gla you're not in charge of marketing
02:29	kmc	them's fightin' words
02:29	amalloy	i always have trouble creating a hash-map out of a map
02:29	LauJensen	kmc: naah, all the fighting words are on my blog :)
02:29	amalloy	,(apply hash-map (mapcat (juxt identity -) (range 4)))
02:29	clojurebot	{0 0, 1 -1, 2 -2, 3 -3}
02:30	amalloy	there we go
02:30	amalloy	kmc, what time zone are you in, if you don't mind?
02:30	kmc	EDT
02:30	LauJensen	,(map (juxt (comp inc rand-int) +) (range 10))
02:30	kmc	GMT-4
02:30	clojurebot	([1 0] [1 1] [1 2] [1 3] [1 4] [5 5] [2 6] [5 7] [7 8] [8 9])
02:31	kmc	comp is function composition?
02:31	amalloy	yes
02:31	LauJensen	yes
02:31	amalloy	blurgh. i used to stay up that late, but now that i have a job i can't stay up past like 1 without killing myself (pacific time)
02:31	LauJensen	(comp f g) = (f (g x))
02:31	LauJensen	amalloy: you're getting older :)
02:31	amalloy	LauJensen: not so. ((comp f g) x) = (f (g x))
02:32	LauJensen	tsk
02:32	amalloy	LauJensen: and yeah, getting older, but mostly getting up earlier
02:33	amalloy	(or at least that's what i tell myself)
02:34	LauJensen	I really would like to know what they eat, the guys who run on 4 hours of sleep every night, that must be great
02:34	kmc	,((comp inc *) 5 4)
02:34	clojurebot	21
02:35	kmc	sweet
02:35	amalloy	LauJensen: future years of their lives
02:36	bobo_	and efficency of the ours they are awake
02:37	bobo_	they probably get less done in more time
02:37	LauJensen	bobo_: Tell that to Obama, Denmarks former Primeminister and Dana White, who all seem to get a lot done
02:37	amalloy	i get about six hours a night, and ten hours once a week when i realize i'm exhausted :P
02:38	kmc	((comp + *) 5 4)
02:38	kmc	,((comp + *) 5 4)
02:38	clojurebot	20
02:38	LauJensen	amalloy: yea I used to do that, if was just 5 and 10
02:38	kmc	,((comp (partial + 1) *) 5 4)
02:38	clojurebot	21
02:38	amalloy	,((juxt + *) 5 4)
02:38	clojurebot	[9 20]
02:44	notsonerdysunny	can anybody help me figure out why #(-> % second :name) goes through fine while #(-> % second (fn [x] (:name x))) does not .. both seem identical to me...
02:45	amalloy	,(macroexpand '(-> % (fn [x] (:name x))))
02:45	clojurebot	(fn* % ([x] (:name x)))
02:45	amalloy	,(macroexpand '(-> % ((fn [x] (:name x)))))
02:45	clojurebot	((fn [x] (:name x)) %)
02:45	amalloy	notsonerdysunny: you want the second of these
02:47	amalloy	(though what you actually want is (comp :name second))
02:47	notsonerdysunny	amalloy: aha .. I think I see what is happening .. thanks .. :)
02:48	notsonerdysunny	yea you are right comp is a better way of doing it
02:48	amalloy	,((comp :name second) [{:name 10} {:name 15}])
02:48	clojurebot	15
02:51	LauJensen	,15
02:51	clojurebot	15
02:51	LauJensen	much easier
02:51	amalloy	LauJensen: you're right. i don't know why i bother writing any programs; i should just figure things out by hand and then write the answer in a clj file
02:51	notsonerdysunny	LauJensen: :)
02:51	amalloy	,(identical? 15 15)
02:51	clojurebot	true
02:52	LauJensen	,(identical? 130 130)
02:52	clojurebot	false
02:52	amalloy	that's the result i would have expected
02:52	amalloy	why did i get true?
02:52	kmc	haha
02:52	kmc	interning of small integers?
02:52	kmc	,(doc identical?)
02:52	LauJensen	amalloy: The first 128 integers are hardcoded in the jvm
02:52	clojurebot	"([x y]); Tests if 2 arguments are the same object"
02:52	amalloy	,(identical? -5 -5)
02:52	clojurebot	true
02:53	kmc	,(let [x 130] (identical? x x))
02:53	clojurebot	true
02:53	amalloy	LauJensen: the first 256 :P
02:53	kmc	the first 127 and the last 128
02:53	LauJensen	sure
02:53	LauJensen	punks
02:53	LauJensen	:)
02:53	amalloy	kmc: that adds up to 255. you have to include 0 in one of the two sets
02:54	LauJensen	,(true? (Boolean. "true"))
02:54	clojurebot	false
02:54	kmc	that constructs a Java Boolean from the string "true"?
02:54	kmc	is there a better way to write #(identical? % %) ?
02:54	LauJensen	,(Boolean. "true")
02:54	clojurebot	true
02:54	kmc	,(Boolean.)
02:54	clojurebot	java.lang.IllegalArgumentException: No matching ctor found for class java.lang.Boolean
02:54	amalloy	kmc: identical?
02:54	amalloy	oh
02:54	kmc	,(new Boolean)
02:54	clojurebot	java.lang.IllegalArgumentException: No matching ctor found for class java.lang.Boolean
02:55	kmc	,(new Boolean "true")
02:55	clojurebot	true
02:55	kmc	in general, is there a better way to write #(f % %)
02:55	lypanov	,(new Boolean TRUE)
02:55	clojurebot	java.lang.Exception: Unable to resolve symbol: TRUE in this context
02:55	LauJensen	(new Boolean "true") == (Boolean. "true")
02:55	LauJensen	lypanov: You're thinking of Boolean/TRUE
02:55	lypanov	ah, thats the one. forgot the syntax.
02:55	kmc	,Boolean/TRUE
02:55	clojurebot	true
02:55	kmc	and that's a static field of the Boolean Java class?
02:55	kmc	,(Integer. 3)
02:55	clojurebot	3
02:56	amalloy	kmc: i'm not sure, about (f x x). there must be an easy way
02:56	LauJensen	,(set (range 32))
02:56	clojurebot	#{0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31}
02:56	LauJensen	,(set (range 33))
02:56	clojurebot	#{0 32 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31}
02:56	kmc	heh
02:56	lypanov	cute
02:56	kmc	,(class 5)
02:56	clojurebot	java.lang.Integer
02:56	kmc	,(class 5000000000000000000000000)
02:56	clojurebot	java.lang.ExceptionInInitializerError
02:57	amalloy	!
02:57	kmc	;P
02:57	lypanov	huh
02:57	LauJensen	This is almost better than PHP :)
02:57	amalloy	oh
02:57	kmc	,(Bignum. 500000 500000)
02:57	clojurebot	java.lang.IllegalArgumentException: Unable to resolve classname: Bignum
02:57	kmc	,(java.lang/Bignum. 500000 500000)
02:57	clojurebot	java.lang.ClassNotFoundException: java.lang
02:57	kmc	hrm
02:57	LauJensen	kmc: in 1.3 you can to (class 5000000000000000000N)
02:57	amalloy	,(class 500000000000000000L)
02:57	clojurebot	Invalid number: 500000000000000000L
02:57	amalloy	,(class 500000000000000000M)
02:57	clojurebot	java.math.BigDecimal
02:58	kmc	cool
02:58	LauJensen	,(class (bigint 5000000000000000000000000))
02:58	clojurebot	java.lang.ExceptionInInitializerError
02:58	LauJensen	eh?
02:58	LauJensen	,(clojure-version)
02:58	clojurebot	"1.2.0-master-SNAPSHOT"
02:58	LauJensen	-> (class (bigint 5000000000000000000000000))
02:58	amalloy	,(doc bigint)
02:58	sexpbot	java.lang.Exception: EvalReader not allowed when read-eval is false.
02:58	clojurebot	"([x]); Coerce to BigInteger"
02:58	LauJensen	,read-eval
02:58	clojurebot	false
02:58	amalloy	LauJensen: (bigint) doesn't get called until 50000000000000 is evaluated, and it's an illegal integer literal
02:59	LauJensen	anyway, I should really do some work
02:59	amalloy	,100N
02:59	lypanov	hehe
02:59	clojurebot	Invalid number: 100N
02:59	LauJensen	amalloy: works in the repl here though
02:59	amalloy	hm
02:59	LauJensen	same clojure version
03:00	kryft	Does this just destructure the value of the function neighbors? ([size yx] (neighbors [[-1 0] [1 0] [0 -1] [0 1]] size yx))
03:00	kryft	(Looking at listing 5.1 in JoC)
03:01	amalloy	kryft: no, that code is doing no destructuring
03:01	amalloy	it is treating the vector [size yx] as a function, and calling it with the result of (neighbors ...)
03:01	amalloy	when vectors are called as functions, they treat their argument as an index
03:02	amalloy	,([\a '- "qqq"] 1)
03:02	clojurebot	-
03:02	Chousuke	that looks more like a variadic function definition for the [size yx] parameter vector
03:02	Chousuke	taken out of context :P
03:02	amalloy	however, if there's a (fn) lurking around near the code that you didn't show us...
03:03	amalloy	(and i agree with Chousuke, it looks like it is)
03:06	kryft	amalloy: http://pastebin.com/YTPJ8iUq (Listing 5.1 in Joy of Clojure)
03:06	amalloy	kryft: Chousuke is correct
03:06	Chousuke	yeah. it calls the 3-arg version of neighbor from the 2-arg version
03:06	amalloy	it's saying that neighbors can take 2 or 3 arguments. the 2-argument version calls the 3-arg version
03:06	Chousuke	the first arg is just a vector of vectors.
03:07	kryft	Ah, ok, I don't think that was covered in the book yet. :)
03:08	kryft	I forgot about that syntax for indexing vectors even though I just read about it. :P
03:10	amalloy	kryft: see 2.3.2
03:10	amalloy	discussing variadic args
03:14	tomoj	anyone using octobot in clojure yet?
03:22	amalloy	ah, kryft, he does mention using vectors as functions in chapter 5, but it's hidden so well i had to reread the whole damn chapter to find it
03:22	amalloy	it's in 5.2.2
03:30	bartj	,([1 2 3] 0)
03:30	clojurebot	1
03:33	tomoj	ztellman is my hero
03:33	amalloy	,([1 2 3] -1)
03:33	clojurebot	java.lang.IndexOutOfBoundsException
03:33	amalloy	:(
03:33	LauJensen	tomoj: why?
03:35	tomoj	http://github.com/ztellman/gloss
03:35	tomoj	not to mention aleph
03:36	zkim	and penumbra
03:38	LauJensen	Yea he's a cool and crafty guy, no doubt
03:42	kmc	cool
03:42	kmc	embedded gpgpu programming for clojure
03:42	kmc	i like that a lot
03:42	LauJensen	kmc: Yea he's done 2 amazing ports of my blogposts to Penumbra
03:43	LauJensen	First Brians Brain, when in optimized Clojure it does a 160x160 board in 1.6ms, Zach kept it at 1.6ms per iteration, but on a 2000x2000 board. My Fluid sim ran about 35 fps (I think) on a 80x80 board, Zach had it on 55 fps on a 800x600 board.
03:43	kmc	nice
03:43	kmc	i love GPU programming
03:43	kmc	need to get back into it
03:44	kmc	i was doing fractal video feedback loops at 4096 x 4096 x 60fps, in 2007, without particularly optimizing my code
03:44	kmc	there's just so much power there
03:44	kmc	that card is probably in every midrange machine now
03:44	LauJensen	I read that the largest supercomputer in 1998 had the same power as a 500$ graphics card today
03:44	kmc	hehe
03:45	hiredman	thats a hell of a screensaver
03:45	kmc	well, i'd bet that supercomputer had more ability to do different things at once
03:45	LauJensen	no
03:45	LauJensen	it would only solve solitaire puzzles
03:45	kmc	haha
03:45	LauJensen	j/k ofc :)
03:46	kmc	http://en.wikipedia.org/wiki/Semi_Automatic_Ground_Environment ← the largest computer, ever
04:37	jjido	no clojurebot today
04:37	LauJensen	jjido: ?
04:38	jjido	,true
04:38	LauJensen	clojurebot: where are you ?
04:38	LauJensen	-> true
04:38	sexpbot	⟹ true
04:38	LauJensen	Well, there are alternatives
04:46	jjido	-> (:foo nil)
04:46	sexpbot	⟹ nil
04:52	jjido	killall -s java \| wc -l --> 6
04:52	jjido	killall java
04:52	jjido	how many stray processes do I have?
04:57	jjido	Is there a string representation of Clojure objects which is resilient to structural loops (with an atom)?
05:01	notsonerdysunny	thanks raek
05:01	notsonerdysunny	that actually highlights the use of atom
05:01	notsonerdysunny	:)
05:02	raek	sorry for being offensive about the def thing... I should have mentioned what "the right way" was, rather than just saying "you're doing it wrong"
05:04	bmh	dear #clojure -- I love clojure so much. The Java interop makes me so happy
05:04	jjido	-> (take 10 (.toString first))
05:04	sexpbot	java.lang.SecurityException: Tried to call: toString on clojure.core$first@137e670 which is not allowed.
05:06	notsonerdysunny	raek: no problem ... being critical is what makes this fun.. :)
05:06	esj	bmh: absolutely, playing with libraries like JFreeChart from a repl makes me uncommonly happy
05:07	bmh	esj: I found a C++ library that did the things I wanted it to do, but bugger, it's written in C++. I found a Java library that did some of the things I want to do, so bam. I can use the Java bits to port the C++ code I need!
05:07	bmh	(sorry if that was incoherent)
05:07	esj	bmh: not at all
05:07	raek	notsonerdysunny: actually, thing the counter example would be a good example to have on clojuredocs. it very clearly shows how the function is applied to the data structure
05:09	jjido	-> (take 10 (str first))
05:09	sexpbot	⟹ (\c \l \o \j \u \r \e \. \c \o)
05:10	jjido	but I guess that won't prevent a stack overflow.
05:12	raek	jjido: what are you trying to do? :)
05:13	jjido	raek: I want to find out what I did wrong (error), so I want to print out objects. But that goes in infinite loop because of an atom.
05:14	raek	jjido: does the data structure in the atom have a reference to the atom itself?
05:14	jjido	yes
05:15	raek	then it might be a good thing to wrap the printing code in a (binding [print-level 10] ...)
05:16	raek	when an atom is printed/stringified, it will turn its contents into a string first
05:16	raek	at the repl, you can also do a (set! print-level 10)
05:16	jjido	is that the depth or the string length?
05:17	raek	form depth
05:17	raek	there is a print-length too
05:20	notsonerdysunny	can somebody help me with an example for apply-macro from clojure.contrib.apply-macro
05:24	notsonerdysunny	(defmacro hello [x]
05:24	notsonerdysunny	`(sayhello ~x))
05:24	notsonerdysunny	(macroexpand-1 (apply-macro hello '(10)))
05:25	notsonerdysunny	http://gist.github.com/625927 is my example .. but I keep getting a stack overflow ..
05:34	jjido	raek: ok, I inadvertently shared an atom :(
05:52	jjido	I see no effect of using the print-length binding. print-level works great though.
06:20	raek	,(binding [print-length 3] (println ["one" "two" "three" "four"]))
06:20	clojurebot	[one two three ...]
06:21	raek	it inly restricts the number of element printed from collections
06:21	raek	,(binding [print-level 3] (println ["one" ["two" ["three" ["four"]]]]))
06:21	clojurebot	[one [two [three #]]]
06:23	raek	print-length is useful when dealing with infinite or very long sequences and print-level when dealing with circular structures or deep trees
06:26	jjido	raek: thanks
06:34	rfgpfeiffer	how can i give a function a fixed time to return or kill it?
06:36	raek	rfgpfeiffer: I know that clj-sandbox (http://github.com/Licenser/clj-sandbox) does something like that
06:37	raek	http://github.com/Licenser/clj-sandbox/blob/master/src/net/licenser/sandbox.clj#L35
06:38	LauJensen	rfgpfeiffer: http://gist.github.com/626002
06:38	rfgpfeiffer	thanks
06:38	LauJensen	np
06:41	raek	,(class (future 1))
06:41	clojurebot	clojure.core$future_call$reify__5492
06:43	raek	hrm, what happens with the running code if a timeout occurs?
06:43	LauJensen	it gets barged
07:56	tobiasraeder	anyone ever done anything with wings and clojure?
08:06	carkh	wew that wing thing is ugly
08:09	LauJensen	carkh: ?
08:11	carkh	i was reacting to tobiasraeder's query
08:11	bmh	I'm trying to use penumbra and the universe is exploding on me when I run 'lein test'
08:13	bmh	hurumph, looks like a case of a broken test
08:16	LauJensen	carkh: Yea, just wondering what you were looking at
08:16	LauJensen	ie. link
08:17	carkh	http://wingsframework.org/wingset/WingSet/
08:17	LauJensen	bmh: just start a project vm and run the examples
08:17	bmh	LauJensen: I just spotted it on the bug tracker
08:17	bmh	thanks
08:23	kumarshantanu	I have a question on Java interop
08:23	kumarshantanu	,(. "Hello" :toLowerCase) ; this works
08:23	clojurebot	"hello"
08:24	kumarshantanu	,(. "Hello" :substring 3 4) ; but this doesn't
08:24	clojurebot	java.lang.IllegalArgumentException: Malformed member expression
08:24	kumarshantanu	can anyone help me understand this?
08:24	carkh	,(. "hello" substring 3 4)
08:24	clojurebot	"l"
08:25	carkh	don't know why using a keyword should work on the first example
08:25	carkh	you're supposed to use simple symbols
08:25	kumarshantanu	carkh: I want to be able to compose the method name at runtime (I don't know them in advance)
08:25	LauJensen	kumarshantanu: The first example shouldnt work, thats just a result of how keywords are implemented
08:25	carkh	kumarshantanu: what's your use case ?
08:26	kumarshantanu	LauJensen: I see
08:26	kumarshantanu	carkh: I have a bunch of setter methods to be called, so want to create a helper function to call them
08:26	kumarshantanu	but I don't know them in advance
08:27	carkh	you might have to use reflection
08:27	carkh	there is a helper function somewhere in clojure (java side)
08:27	carkh	clojrue.lang.Reflector has it i think
08:28	kumarshantanu	example: (setter! some-obj {:method1 [m1-value1] :method2 [m2-value1 m2-value2]}) --> (doto some-obj ((.method1 m1-value1) (.method2 m2-value1 m2-value2))
08:28	carkh	(clojure.lang.Reflector/invokeInstanceMethod instance method-name-as-string array-of-args)
08:28	kumarshantanu	okay looking
08:29	carkh	note that args are an array, not a vector
08:29	carkh	http://github.com/richhickey/clojure/blob/master/src/jvm/clojure/lang/Reflector.java
08:31	kumarshantanu	carkh: there seems to be a c.c.reflect too -- http://richhickey.github.com/clojure-contrib/reflect-api.html
08:31	carkh	ahyes indeed !
08:33	carkh	yes that would help tremendously
08:38	LauJensen	You guys should use emacs, fuzzy completions expands c.c to clojure.contrib
08:40	carkh	i do but fuzzy completion is not perfect for namespaces
08:40	carkh	maybe some setup problem
08:40	carkh	i walways end up with clojure.contrib.sql/<space>
08:43	kumarshantanu	,(clojure.contrib.reflect/call-method java.lang.String :substring [] "hello" 3 4)
08:43	clojurebot	java.lang.ClassNotFoundException: clojure.contrib.reflect
08:45	kumarshantanu	I am still getting this error on my local REPL: java.lang.NoSuchMethodException: java.lang.String.substring()
08:46	carkh	the vector should have classes inside it
08:47	carkh	[java.lang.String java.lang.Integer java.lang.Integer] i guess
08:47	carkh	i think it's easier to use (clojure.lang.Reflector/invokeInstanceMethod instance method-name-as-string array-of-args)
08:48	jk_	are there any old-time common lispers here who resent the mixture of messy, stateful java into lisp? i've been really interested in the controversy and when most old-timers go with clojure, it seems like it's with a shrug like it's the lesser of 2 evils (using only java, for example)
08:48	jk_	probably a bad question in this forum :)
08:49	kumarshantanu	carkh: I am passing [Integer Integer] too, but the signature accepts primitives
08:49	kumarshantanu	not working....I guess that might be the reason
08:50	carkh	jk_: well common lisp is very imperative as well
08:51	carkh	kumarshantanu: really you should try c.l.reflector/invokeInstanceMethod, i had success using it
08:52	kumarshantanu	carkh: I will use Reflector, I was just wondering the clojure.contrib version may be better supported
08:52	carkh	kumarshantanu: clojure itself uses reflector so i guess it's here to stay =P
08:53	jk_	carkh: what about exceptions? does that seriously detract?
08:54	carkh	jk_: conditions sure are way better, but exceptions are good enough in most cases
08:54	carkh	and we have a library for condition-like behaviour
08:54	carkh	though i never used it
08:54	kumarshantanu	,(clojure.lang.Reflector/invokeInstanceMethod "Hello World" "substring" (into-array [3 4]))
08:54	clojurebot	"l"
08:55	carkh	kumarshantanu: yay !
08:56	kumarshantanu	carkh: invokeInstanceMethod should be promoted to c.c (!)
08:56	jk_	carkh: ok, one last question... do you think seq is a big win over the predominance on cons cells? as a newcomer, that shift in focus looks like an advantage
08:56	carkh	yes i beleive so
08:57	carkh	most of the time you'll be working with vectors instead of lists
08:57	carkh	it's very handy to be able to use seq abstraction on these
08:57	carkh	and vector are _fast_
08:57	carkh	vectors*
08:58	jk_	thanks :)
08:58	carkh	also they grow at the end, so you can avoid most of this reverse non-sense
09:00	jk_	carkh: i was watching some online lectures where they are teaching with scheme and i was wondering if lispers thought something was lost or radically different with clojure because they always had such a strong link to the cons cell data structure before
09:00	carkh	well we still have cons cells
09:00	carkh	kinda
09:01	carkh	we have single linked lists anyways, for a cons cell we use a simple vector
09:01	carkh	but really i think it's a plus to have the right data structure for the right job
09:02	carkh	sure the jvm has its drawbacks, but it has many good features
09:03	carkh	and clojure is way above CL for functional purity (though not to the haskell level)
09:04	carkh	take CL's hash tables for instance
09:04	carkh	very imperative
09:05	carkh	we have fully immutable and persistent maps in clojure for that
09:11	pdk`	i recall some article once whining about clojure wasnt lisp all the way down and how evil this was cause it was on the jvm
09:11	pdk`	though on the other hand
09:11	pdk`	the same person who wrote that still thought his efforts to make a multi ghz lisp machine for today's world were remotely practical
09:11	carkh	there is strong opposition to clojure in the CL community
09:12	_fogus_	jk_: Common Lisp is not necessarily centered around the cons cell. It has a powerful sequence abstraction.
09:13	carkh	_fogus_: though the sequence abstraction doesn't go as far as in clojrue
09:13	_fogus_	carkh: I haven't noticed a very strong opposition at all.
09:13	jk_	pdk`: http://www.loper-os.org/?p=46 probably this guy :)
09:13	carkh	i guess it withered down with time, they're mostly ignoring clojure
09:13	pdk`	oh yeah i was just about to look for that jk
09:13	_fogus_	carkh: It's not as pervasive no. But the point is that the Common Lisp world does not revolve around the cons cell.
09:14	pdk`	but the tl;dr of it is it boiled down to whining from an ivory tower
09:14	carkh	_fogus_: anyways i can see their point, CL has been there for decades, clojure is but a blip on their radar
09:15	Bahman	Hi all!
09:16	jk_	pdk`: well i was reading through a lot of this blogs and i don't think it's really ivory tower whining. he's claiming that something truly valuable was lost when lisp machines went away because the same abstractions when all the way down giving crystal clarity to the whole structure (that's how i read it anyway, in a nutshell)
09:16	_fogus_	carkh: If you're a long-time Common Lisp programmer then you're probably happy with what you have. Plus, you've seen "other Lisps" come and go numerous times. ;-)
09:16	djpowell	where should i report a clojure-clr issue?
09:17	carkh	_fogus_: i never was fully happy with CL, the multi-processing issues (under windows) and the impertive nature of it were always a drawback
09:17	jk_	_fogus_: thanks for the input. i guess i was getting that feeling (re cons cell) from some of R.H.'s comments and from the SICP lectures i'm viewing
09:18	pdk`	are these lectures recent
09:18	_fogus_	jk_: Those SICP lectures are like 20 years old no?
09:18	jk_	pdk`: mid-80's i think
09:18	jk_	and it's scheme, not CL
09:20	_fogus_	That's not to say that cons cells are not used in CL, they are an incredibly robust structure. Only, that they are not used exclusively.
09:21	jk_	gotta go to work. thanks for the insightful comments, guys.
09:21	djpowell	,(with-out-str (pprint {:a {:b "c"}}))
09:21	clojurebot	java.lang.Exception: Unable to resolve symbol: pprint in this context
09:21	djpowell	,(with-out-str (clojure-version))
09:21	clojurebot	""
09:21	djpowell	,(clojure-version)
09:21	clojurebot	"1.2.0-master-SNAPSHOT"
09:22	carkh	,(with-out-str (clojure.contrib.pprint/pprint {:a {:b "c"}}))
09:22	clojurebot	java.lang.RuntimeException: java.lang.RuntimeException: java.lang.RuntimeException: java.lang.ClassNotFoundException: clojure.contrib.pprint
09:22	carkh	=/
09:22	carkh	,(require 'clojure.contrib.pprint)
09:22	clojurebot	java.io.FileNotFoundException: Could not locate clojure/contrib/pprint__init.class or clojure/contrib/pprint.clj on classpath:
09:22	djpowell	hmm, anyway pprint in clojure-clr totally mangles that expression
09:23	djpowell	as: {:a "c"{:b }}
09:23	_fogus_	,(with-out-str (clojure.pprint/pprint {:a {:b "c"}}))
09:23	clojurebot	"{:a {:b \"c\"}}\n"
09:26	carkh	ah i'm still with good old 1.1 =P
10:08	bmh	well this seems to be an energetic channel. Any penumbra users about, I'm trying to make sense of ztellman's marble vertex shader
10:22	cemerick	bmh: quiet just means work is getting done :-)
10:26	hircus	has anyone tried this snippet? http://gist.github.com/624638
10:26	hircus	I tried it with Clojure 1.2.0 on Linux x64 -- both JDK 1.6.0u22 from Sun and IcedTea from Fedora
10:27	hircus	and the Clojure REPL freezes at (defprotocol...) -- it returns nil but never prompts for the next command
10:36	_fogus_	hircus: Hrm. That'
10:36	_fogus_	s odd
10:37	_fogus_	hircus: Does it do that for every defprotocol ever?
10:39	cemerick	hircus: funky; works fine in both direct and ccw repls.
10:40	hircus	_fogus_:
10:41	hircus	_fogus_: I updated my comment in the gist. as it turns out it's just that the prompt is not printed after defprotocol. If I enter a simple expression (e.g. a number) then it's evaluated and the prompt is then printed
10:41	hircus	and it does not always happened -- I sometimes get the prompt back after defprotocol. odd eh
10:45	kryft	#(< -1 % size) <- Hmm, what does the % do?
10:46	MayDaniel_	,(#(- 10 %) 15)
10:46	clojurebot	-5
10:47	kryft	MayDaniel_: Duh, of course :P
10:58	raek	kryft: #(< -1 % size) is the same as (fn [%] (< -1 % size))
11:00	kryft	raek: Yeah, thanks. :) I did remember that #() was literal syntax for anonymous functions, but for some reason I didn't remember (and couldn't figure out) that naturally % must be an argument for the function.
11:03	kryft	There are no stupid questions, but some days I manage to get pretty close!
11:05	bhenry	is there any utility that will tell you all the namespaces on the classpath in which a given function is called?
11:09	raek	bhenry: with emacs and swank-clojure, there is "C-c C-w c: List all callers of a given function"
11:10	bhenry	awesome!
11:12	bhenry	raek i get error in process filter: Wrong type argument: char-or-string-p, nil
11:12	bhenry	i'm leaving the argument as is after the command
11:15	_rata_	hi
11:16	dpritchett	does the newest vimclojure integrate with lein-nailgun?
11:17	dpritchett	I really want to like vimclojure but I haven't gotten used to classpaths... it's so nice to let leiningen deal with them for me.
11:19	_rata_	is there a function like tree-seq but for graphs? (i.e., the children function can give nodes already visited by the function)
11:21	jkkramer	_rata_: http://github.com/jkk/loom or clojure.contrib.graph
11:21	_rata_	clojure.contrib.graph doesn't have anything like that... I've already checked
11:22	jkkramer	lazy-walk does if you use ccg's graph structure
11:23	jkkramer	loom has loom.alg-generic/bf-traverse and others for general-purpose traversal
11:24	_rata_	to use a ccg's graph structure I'd need a function like the one I'm asking for
11:25	_rata_	I have my own data structure and I want to walk over it
11:26	jkkramer	then loom would be a better fit. it's alpha though so the API isn't concrete
11:35	fliebel	Hey
11:35	jkkramer	_rata_: if you do end up using loom, i'm interested in any feedback
11:36	fliebel	chouser or _fogus_: You probably know this, but the index of Joy of Clojure(which I just bought) says 'firehos', missing the 'e'.
11:37	_rata_	jkkramer, I think my team wouldn't be happy to include another library dep, but I can see bf-traverse is self-contained, so I'd like to copy it right into my code... don't you mind?
11:39	dpritchett	firehoss
11:39	jkkramer	_rata_: sure, go for it. just pass along changes if you make any
11:40	_rata_	ok... I think that lazy-seq should go at the beginning of the step function
11:41	jkkramer	my reasoning for not is that the first node is always available in the queue
11:41	jkkramer	and subsequent nodes require some work to find
11:43	jkkramer	but i'm not an expert in the subtleties of lazy-seq
11:44	_rata_	I think it's preferable to have a complete lazy seq than a seq that's mixed
11:45	jkkramer	opinions from the #clojure sages? is the form (cons x (lazy-seq ...)) discouraged?
11:46	Chousuke	why would it be?
11:46	_rata_	also I've always seen (lazy-seq (cons x ...)) in the functions returning lazy seqs
11:46	Chousuke	well the latter is lazier
11:48	Chousuke	If you do (defn foo [] (cons 1 (lazy-seq (foo)))) and call it it realizes the first element immediately.
11:49	jkkramer	right. when you have the first element in-hand that's not bad, is it?
11:49	Chousuke	Well, the recommendation is to make sequences as lazy as possible
11:50	jkkramer	if you did (member:defn foo [] (lazy-seq (cons 1 (foo)))) and (foo) takes some time to process, the first element would take longer to return
11:50	KirinDave	Because of chunking, right? It's already going to do sequences in blocks.
11:50	jkkramer	oops, scratch "member:"...colloquey weirdness
11:51	Chousuke	KirinDave: Chunking doesn't really matter here.
11:51	KirinDave	Chousuke: Right, but realizing the first several elements probably doesn't offer efficiency gains unless they're terribly expensive.
11:52	jkkramer	in the case of a graph traversal, finding neighbors not yet visited could be a less-than-fast task
11:54	jkkramer	and you'd be doing work for a step one farther than the consumer asked for
11:55	_rata_	I don't think so
11:57	_rata_	every function that returns a lazy-seq I've seen use the lazy-seq macro at the beginning of the function and they don't realize more elements than the ones asked for (except for chunking)
11:57	Chousuke	That's what we're saying
11:57	jkkramer	taking pre-traverse as an example: http://github.com/jkk/loom/blob/master/src/loom/alg_generic.clj#L30 if you request the one item, no work is done -- the start node is returned. if lazy-seq was moved before cons, it would find the neighbors of the starting node but still only return the starting node. right? i fully acknowledge i mgiht be misunderstanding lazy-seq
11:59	_rata_	what's the difference between pre-traverse and bf-traverse? just DFS vs BFS?
11:59	jkkramer	yes. pre-traverse is preorder
11:59	jkkramer	and bfs right now has more options for filtering neighbors, depth, etc
12:00	jkkramer	still messing with the API
12:00	_rata_	ok
12:00	_rata_	why preorder = DFS?
12:01	_rata_	it'd be more consistent to call it df-traverse
12:01	jkkramer	because dfs can be preorder or postorder
12:03	_rata_	ah ok
12:08	dpritchett	Does anyone know how to use the vimclojure/server jar on clojars? Meikel suggests there should be a `lein run` task to start it up but that doesn't seem to be the case.
12:08	dpritchett	(http://groups.google.com/group/vimclojure/browse_thread/thread/69abeea9f57cb73b)
12:11	_rata_	jkkramer, I've tried both versions of pre-traverse, your and the one with lazy-seq at the beginning, and the later is faster when I (take 1 ...)
12:12	jkkramer	_rata_: how are you benchmarking?
12:12	_rata_	nothing serious, just (time ...) a couple of times
12:13	_rata_	maybe the graph I'm giving to the function is too small
12:14	_rata_	I'm calling it like (take 1 (pre-traverse {:a [:b :c], :b [:c], :c []} :a))
12:21	jkkramer	(dotimes [_ 10] (time (doall (take 1 (pre-traverse {:a [:b :c], :b [:c], :c []} :a))))) is consistently faster for me with my version -- note the doall to realize the first element
12:23	jkkramer	also, (dotimes [_ 10] (time (doall (take 1 (pre-traverse #(do (Thread/sleep 100) ({:a [:b :c], :b [:c], :c []} %)) :a))))) demonstrates when the 'neighbors' function takes a long time to execute. with lazy-seq before cons, requesting the first element does the work for the second element
12:25	jkkramer	_rata_: ^
12:25	dpritchett	woo, i figured it out!
12:25	@rhickey	-XX:MaxInlineSize=100 seems to help Clojure perf a good bit
12:25	@rhickey	interested in feedback from others if it helps you
12:25	dpritchett	add the vimclojure/server jar to lein's dev dependencies
12:26	dpritchett	then start the server with a quick alias including lein classpath
12:26	dpritchett	alias nailgun='java -cp "`lein classpath`" vimclojure.nailgun.NGServer 127.0.0.1'
12:27	_rata_	jkkramer, ok
12:28	_rata_	but normally no one will ask for just the first element... they already have that element
12:29	jkkramer	_rata_: it applies to any node. with lazy-seq before cons, work is done for nodes subsequent to the one requested
12:30	_rata_	that's weird... it should be lazier, but it's the other way around then?
12:32	jkkramer	basically it's lazy for the nodes that require work to find
12:33	jkkramer	if there's a better way to do arbitrary-sized lazy graph traversal pre/post/bf, i'm very interested to know
12:35	jkkramer	lazy-walk from clojure.contrib.graph, which uses concat, dies on 100,000+ node graphs
12:35	_rata_	I've tried it and you're completely right... putting lazy-seq before cons realizes always one more than needed
14:00	fliebel	I'm reading Joy of Clojure, and in some of the examples some collections from java.util are used instead of clojure.lang. Why is this and what are the differences?
14:00	d0m_	fliebel: The goal is to be able to use both depending of when you need them
14:01	nishant	test message
14:02	fliebel	d0m_: I see… But what are the differences? Why might you prefer one over the other?
14:03	d0m_	Well suppose you need to call a method on an existing Java object, but what you have available right now is a clojure list
14:03	d0m_	You can call directly the java method with the clojure list.. without conversion
14:04	nishant	hey everyone, I'm new to clojure, I've a question about java interop. I am trying to create an object of a java class using the (new Class args) form, but I get an InvocationTargetException, even after fully qualifying the name of the class, any ideas? I also verified that the constructor is visible using the show function available in repl-utils. Please help!
14:05	d0m_	fliebel: The list in clojure and in java are different implementations, even if they share some interfaces. So, if you're coding only in clojure, you can use only the clojure one if you like.. but then, if you have to use some java methods, it's nice to be able to convert java list to clojure list.
14:06	fliebel	Right :)
14:06	d0m_	so because of the interface they share, you can use java implementation of list in your high level function in clojure
14:06	d0m_	that's awesome :)
14:07	hiredman	ping?
14:07	clojurebot	PONG!
14:10	raek	nishant: can you paste the full stack trace somewhere?
14:11	raek	also, which class do you try to instanciate
14:11	nishant	@raek should I paste it here?
14:12	nishant	@raek it's a class that I wrote and I'm importing it through lein
14:12	raek	nishant: put it in gist.github.com and paste the link here
14:12	raek	nishant: you can check whether clojure can find the class by just trying to evaluate the class name
14:12	raek	,java.util.ArrayList
14:12	clojurebot	java.util.ArrayList
14:13	nishant	yeah it can find the class
14:13	raek	,java.util.Something
14:13	clojurebot	java.lang.ClassNotFoundException: java.util.Something
14:13	raek	great
14:13	nishant	okay here's the stack trace git://gist.github.com/626696.git
14:14	raek	ok, I tried http://gist.github.com/626696
14:14	raek	(you pasted the git url)
14:14	nishant	oops
14:14	raek	hrm, looks like it can't figure out the constructor in some way
14:15	raek	,(java.util.ArrayList.)
14:15	clojurebot	#<ArrayList []>
14:15	raek	,(java.util.ArrayList. "foo" "bar")
14:15	clojurebot	java.lang.IllegalArgumentException: No matching ctor found for class java.util.ArrayList
14:15	raek	hrm, but it's not that exception..
14:16	raek	nishant: if you press 1 in the exception buffer (goes to the cause) what do you get?
14:16	raek	my current theory is that the constructor throws an exception
14:16	nishant	I get No message [Thrown class.java.NullPointerException]
14:17	nishant	umm, let me try to see if I can use the constructor in Java
14:17	raek	nishant: does the class have multiple constructors that take the same number of arguments
14:17	raek	?
14:17	nishant	@raek yeah the same number, but different types
14:18	raek	what types?
14:19	nishant	well there are two that take 1 argument, one of them takes a string, and the other one takes a JSON object (it's a java JSON implementation)
14:19	_rata_	why can't I use ^ instead of (meta ...) in a place that meta has no problems? like in ^(:a b)
14:19	_rata_	isn't ^ just a reader macro?
14:20	raek	_rata_: ^ as a shortcut for 'meta' is deprecated
14:20	raek	it worked up to 1.1
14:20	_rata_	raek, really??
14:20	_rata_	and what's ^ for now?
14:20	raek	in 1.2 ^ is the new #^
14:20	raek	so, use the more verbose (meta ...) for getting the metadata
14:21	raek	and since 1.2, you can use ^ to attach metadata and type hints
14:21	_rata_	mmmm... I don't get it yet
14:21	raek	,(meta ^{:a 1} :foo)
14:21	clojurebot	Metadata can only be applied to IMetas
14:22	_rata_	is there no shortcut for (meta ...) then?
14:22	raek	_rata_: not anymore...
14:22	_rata_	:(
14:22	raek	attaching meta is probably much more common
14:22	nishant	@raek the ctor works from Java
14:23	raek	nishant: can you post the stack trace of the cause?
14:23	_rata_	than retrieving it?? how can that be? everytime you attach metadata there will be one or more times you get it
14:24	nishant	@raek http://gist.github.com/626720
14:25	raek	_rata_: a very common usage is type hints for the compiler
14:25	raek	which only the compiler looks at
14:26	raek	_rata_: anyway, ^ (or #^ in 1.1) only works for literals
14:26	raek	if you have some data in your program that you want to attach metadata to, you must use with-meta
14:26	_rata_	mmmmm... so metadata isn't used that much.... that's a pity
14:27	_rata_	yeah, I'm using with-meta
14:27	_rata_	but the (meta ...) thing is annoying... with ^ the code would be much clearer
14:27	raek	when interfacing with clojure.core, you more often declare it than use it yes...
14:28	raek	,(meta (with-meta [1 2 3] {:a 1}))
14:28	clojurebot	{:a 1}
14:28	_rata_	the old shortcut could have been replaced by another one
14:30	_rata_	and one that could be used everywhere, not just with literals
14:31	raek	meta is the "opposite" of with-meta
14:31	raek	nishant: is com.complabs.Annotators.EntityAnnotator the class you try to make an instance of?
14:31	nishant	yeah
14:32	_rata_	so (:changed? (meta (:volume c))) could be written as (:changed? &(:volume c)) or something similar... the later is much more readable I think, no boilerplate
14:35	nishant	ping
14:35	KirinDave	Okay, something that's been bugging me...
14:36	KirinDave	Due ot profound mental deficiency, I don't know how to (require ...) a single namespace with an as without nesting. Is it possible?
14:36	KirinDave	I want to say (require '([clothesline.service :as service]))
14:36	KirinDave	Obviously that doesn't work.
14:36	KirinDave	and I know I could say (require '([clothesline [service :as service]]))
14:37	raek	nishant: is com.complabs.Annotators.EntityAnnotator the class you try to make an instance of?
14:37	nishant	@raek yes
14:37	raek	KirinDave: (require '[clothesline.service :as service])
14:38	KirinDave	raek: Oh weird.
14:38	raek	the extra parens introduce a "prefix list"
14:38	KirinDave	raek: I thought require just wanted seqs.
14:38	KirinDave	But I guess lists vs vectors actually matter?
14:38	raek	KirinDave: it is recommended to use vectors and lists as they are used at clojure.org/libs
14:39	raek	other variants might work, but those are not documented
14:39	raek	and could be an implementation detail
14:39	raek	the general syntax is:
14:39	raek	foo.a
14:39	raek	[foo.a ... options ...]
14:40	raek	foo.a foo.b foo.c --> (foo a b c)
14:40	raek	or combinded:
14:40	KirinDave	raek: Thanks. I didn't realize the type was significant.
14:40	raek	[foo.a ..opts..] [foo.b ..opts..] [foo.c ..opts..] --> (foo [a ..opts..] [b ..opts..] [c ..opts..])
14:40	KirinDave	raek: (btw notice I was saying (require ...) not (:require ...)
14:41	KirinDave	raek: I write the way recommended by libs in ns statements because it looks good and some things parse it
14:41	KirinDave	But it felt strange to do it in an actual require statement. I should get over it.
14:41	raek	the only difference between require and :require in a ns, is that the former requires you to quote the form
14:41	raek	(foo) is actually the degenerate case of the prefix list
14:41	raek	meaning, require nothing with the prefix foo
14:42	raek	how ever, a vector in that position simply means "no options"
14:42	nishant	raek: Yes I am trying to instantiate the EntityAnnotator class
14:43	raek	I think that is about the only case where vectors and lists are treated differently in the current implementation
14:43	raek	nishant: what do you pass to the constructor?
14:43	jarpiain	nishant: do you use the internal class name com.complabs.Annotators$EntityAnnotator ?
14:43	raek	is there any docs for the clas online?
14:43	nishant	raek: a string
14:44	nishant	raek: no there are no docs
14:44	raek	what jarpiain said looks interesting
14:44	nishant	raek: the constructor I'm trying to invoke takes a string, I tried creating an object through java code using the same ctor, and it works
14:45	nishant	jarplain: no I haven't tried that
14:45	nishant	:jarplain I get a ClassNotFound when I try that
14:46	raek	I think I am out of ideas...
14:47	raek	but maybe one could type hint the param
14:47	raek	(fn [^String s] (com.complabs.Annotators$EntityAnnotator. s))
14:47	raek	nishant: have you tried passing a literal string or something?
14:48	raek	what value does the calling code pass?
14:50	nishant	I am getting a slightly different behavior when I pass a literal string (the string is supposed to be a file path)
14:50	raek	the same exception?
14:51	nishant	the same exception, but when I try to debug the cause I get a filenotfound exception
14:52	raek	nishant: can you double-check that the code you used previously used to generate the string doesn't give nil?
14:53	nishant	raek : I've hardcoded the string for testing
14:55	raek	so there is a difference whether the string is a literal or not?
14:56	raek	if so, could you try it with the type hint code I wrote before?
14:57	nishant	raek: yeah let me try that
14:59	nishant	raek: the behavior is same for literals and variables
15:00	raek	and if the file whose path is in the string exists, you get NullPointerException, otherwise FileNotFoundException?
15:01	nishant	raek: yes, it's weird
15:01	raek	should the file contain some data in a specific format?
15:02	nishant	raek: the file contains a json formatted string
15:03	raek	I think clojure finds the correct contsructor, since you get a FileNotFoundException
15:04	raek	it seems unlikely that it tries to open a file using a JSON object (or whatever the other type was)
15:05	nishant	yeah, I think it's calling the right ctor
15:05	LauJensen	/whois nishant
15:06	LauJensen	nishant: You're not the guy who used to write a ton of C# tutorials on codeproject are you?
15:06	nishant	LauJensen: no :)
15:06	LauJensen	ok :(
15:07	hiredman	,(* 16 3)
15:07	clojurebot	48
15:08	nishant	:raek I just noticed that there might be multiple jars which contain this class, you think that could result in this behavior?
15:09	raek	possibly
15:09	raek	I don't know, but I guess that could be worth investigating
15:09	raek	it can cause weird things, for sure
15:10	nishant	okay let me try that, thanks for your help!
15:18	nishant	raek: the jar was old :)
15:18	nishant	sorry for wasting your time
15:24	raek	nishant: no worries :)
15:25	nishant	thanks
15:49	_rata_	if I have a map from objects to their weights, how do I randomly choose one of those objects based on the given distribution? is make-distribution from clojure.contrib.probabilities.finite-distributions the best way?
15:56	_rata_	why are there two math libs in clojure.contrib?
16:15	amalloy	_rata_: finite-distributions, as far as i can tell, defines a way to create pdfs, but not a way to select from them, so there must be some other lib that uses them
16:16	amalloy	if you
16:16	amalloy	re interested, http://rosettacode.org/wiki/Probabilistic_choice#Clojure is an implementation of what you're doing from scratch
16:17	_rata_	can't I use the monad that make-distribution returns directly? (I don't know anything about monads... any reference to understand them would be appreciated :) )
16:17	amalloy	i don't either :P
16:18	raek	_rata_: rhickey's ants.clj example contains the 'wrand' function, which seems to do something similat to what you are trying to obtain: http://raek.se/ants.clj
16:21	_rata_	raek, oh thanks. that's something very similar to what I'm looking for
16:22	_rata_	"=" for refs is the same as "identical?"?
16:25	amalloy	,(let [[a b :as c] (repeatedly (atom 0))] (map (juxt = identical?) c))
16:25	clojurebot	java.lang.RuntimeException: java.lang.ClassCastException: clojure.lang.Atom cannot be cast to clojure.lang.IFn
16:25	amalloy	,(let [[a b :as c] (repeatedly #(atom 0))] (map (juxt = identical?) c))
16:25	clojurebot	java.lang.IllegalArgumentException: Wrong number of args (1) passed to: core$identical-QMARK-
16:25	amalloy	bleh
16:26	amalloy	,(let [[a b] (repeatedly #(atom 0))] ((juxt = identical?) a b))
16:26	clojurebot	[false false]
16:27	_rata_	,(for [a (repeatedly 5 #(atom 0))] (map (juxt = identical?) a))
16:27	clojurebot	java.lang.IllegalArgumentException: Don't know how to create ISeq from: clojure.lang.Atom
16:27	amalloy	_rata_: looks like yes
16:27	_rata_	,(for [a (repeatedly 5 #(atom 0))] ((juxt = identical?) a))
16:27	clojurebot	java.lang.IllegalArgumentException: Wrong number of args (1) passed to: core$identical-QMARK-
16:28	amalloy	_rata_: you can't possibly do it with just a for. only one a is in scope at a time
16:28	_rata_	mmm ok
16:29	amalloy	but since 0 is identical to 0, and two refs to it are neither equal nor identical, it looks like equal and identical are the same for refs
16:29	_rata_	but anybody knows if that result is an implementation detail or I can relay on that behaviour?
16:30	raek	_rata_: for clojure types, = means value equality for immutables and indentity equality for mutable things (atoms/refs/agents/vars, etcs)
16:30	_rata_	ah ok :)
16:30	raek	,(= [1 2 3] '(1 2 3))
16:30	clojurebot	true
16:30	_rata_	thank you amalloy and raek
16:40	ztellman	bmh: still have questions about penumbra?
16:41	BufferUnderpants	Hi
16:42	BufferUnderpants	I'm trying to use the ZK web framework from within Clojure and I'm having trouble with a proxy class.
16:43	nishant	does repl-utils/source not work on functions I define in the user namespace?
16:44	BufferUnderpants	The method I'm invoking receives an instance of an instance of an interface, very much like Swing.
16:44	raek	nishant: no, but their definitions needs to be loaded through a require
16:44	raek	or C-c C-k in emacs
16:45	raek	if you load then in the repl or with C-x C-e or C-M-x file and line metadata is not stored for them
16:45	amalloy	oh neat, raek, i didn't realize source worked for C-c C-k. i just knew it didn't work if you put them into the repl so i never tried
16:47	nishant	raek:thanks
16:48	colinsteele	hey folks how do I turn a string into a java.io.Reader or a java.io.InputStream??
16:48	colinsteele	<---- java noob, pardon my cluelessness :)
16:50	lpetit	,(java.io.StringReader. "a string")
16:50	clojurebot	#<StringReader java.io.StringReader@6d9e14>
16:50	lpetit	colinstee: ^^
16:50	colinsteele	ahhhh
16:50	colinsteele	voila, thanks
17:07	hsuh	hello folks... i wrote this (http://pastie.org/1221883) and thought it was too slow, what am i doing wrong ?
17:09	Chousuke	hsuh: 7ms per database connection & query doesn't sound too bad.
17:09	Chousuke	well, 8ms
17:10	Chousuke	hsuh: it opens a new connection to the database for every query, you hopefully won't be doing that in production code.
17:11	arkh	how do I :use namespaces from multiple files (of my own making)?
17:11	hsuh	Chousuke: not anymore... so what should i be doing ? (RTFM is a valid answer)
17:12	Chousuke	hsuh: I think you'll want to open a connection and use that for all your queries. (You'll have to read the docs to find out how, though)
17:13	raek	arkh: other than (ns my.ns.one (:use the-ns)) (ns my.ns.two (:use the-ns)) etc, for each file that wants to use it?
17:14	Chousuke	it's a good habit to avoid :use though
17:14	arete	hsuh: you'll want to use a connection pooling ds and the :ds option for clojure.contrib.sql
17:14	Chousuke	at least without :only
17:14	arete	http://jdbc.postgresql.org/documentation/publicapi/org/postgresql/ds/PGConnectionPoolDataSource.html
17:14	arkh	raek: how do namespaces work with the class path? i.e., how does clojure know where to find namespaces in the files I make / want to load
17:14	raek	(I interpreted your question as that you wanted to use "the-ns" from "my.ns.one" and "my.ns.to"
17:14	hsuh	arete: tks
17:15	raek	arkh: yes. (ns foo.bar.some-thing ...) corresponds to foo/bar/some_thing.clj
17:15	arkh	raek: cool - that's what I didn't know. Thank you
17:15	raek	where the foo/ directory should be in a directory on the class path (presumably src/)
17:15	arete	hsuh: actually there's an even better example here: http://en.wikibooks.org/wiki/Clojure_Programming/Examples/JDBC_Examples
17:15	dpritchett	how long has leiningen had the "lein classpath" task anyway?
17:15	raek	note that hyphens in namespace names correspond to underscores in file names
17:16	arkh	k
17:16	hsuh	arete: otoh the datasource - postgresql example there says "Note that this is not recommended for production. Use c3p0 or similar instead."
17:17	bmh	ztellman: missed you message an hour back. Can pixel shaders be written in java/clojure or am I tied into using C/C++?
17:17	ztellman	bmh: pixel shaders have to be written in GLSL, which is a C-99 derivative
17:18	arete	yeah dunno why it says that, perhaps there are/were issues with the connection pooling
17:18	ztellman	but there's a Clojure DSL that offers some of the same functionality
17:18	ztellman	under the covers, it just translates into GLSL
17:18	bmh	ztellman: hrm. Do you know a good simplex noise/fBM implementation?
17:19	bmh	I'm playing around with some algorithmic terrain stuff and figured I could try generating everything in the pixel shader
17:19	ztellman	bmh: you were looking at the marble program, right?
17:19	ztellman	that uses perlin nosie
17:19	bmh	ztellman: I saw. Perlin fBM. Perlin has dimensionality issues, though
17:19	hsuh	arete: there is also http://github.com/briancarper/postgres-pool
17:20	ztellman	bmh: well, it's generally difficult to use pre-existing GLSL code
17:21	ztellman	perlin noise is pretty friendly to the hardware
17:21	ztellman	I'm not sure about simplex noise, but it shouldn't be too hard to create your own version
17:22	bmh	ztellman: Have you read any of the papers analyzing Perlin? It actually has some terrible characteristics
17:23	ztellman	bmh: I haven't, but I'm really not an expert
17:23	bmh	For example, the hash function that Perlin used is highly correlated
17:24	bmh	If I was an expert I wouldn't be lurking on irc picking the brains of people more in the know than I ;-)
17:24	ztellman	ha, well I don't think the difference between terrains generated by perlin noise and simplex noise are all that obvious
17:25	ztellman	but if you want to pursue another fractal generator and need any pointers on using Penumbra, I'm available
17:26	bmh	Thanks. Have you ever looked at turbulence? It works by perturbing you x,y,z coordinates with... wait for it... coherent noise
17:26	edoloughlin	Hi, I'm trying to remove nil values from a map (I'm new to Clojure). I'm using
17:26	edoloughlin	(def record {:a 1 :b 2 :c nil})
17:26	edoloughlin	(merge (for [[k v] record :when (not (nil? v))] {k v})
17:26	edoloughlin	but this gives:
17:26	edoloughlin	({:a 1} {:b 2})
17:26	edoloughlin	Whereas I would like:
17:26	ztellman	bmh: I haven't but that seems very GPU friendly
17:26	edoloughlin	{:a 1, :b 2}
17:27	bmh	edoloughlin: why don't you use remove
17:27	edoloughlin	Wow. Thanks. Will look it up!
17:27	bmh	(remove (fn [k v] (nil? v)) collection)
17:27	bmh	http://clojure.github.com/clojure/clojure.core-api.html#clojure.core/remove
17:27	edoloughlin	Thanks.
17:28	bmh	rich & co. did a great job of getting all the functional stuff right.
17:28	dpritchett	what's the latest in clojure on AppEngine hand-holding libs?
17:28	raek	,(apply merge (for [[k v] {:a 1, :b 2, :c nil} :when (not (nil? v))] {k v}))
17:28	clojurebot	{:b 2, :a 1}
17:28	raek	(into {} (for [[k v] {:a 1, :b 2, :c nil} :when (not (nil? v))] [k v]))
17:29	dnolen	,(into {} (remove (fn [[k v]] (nil? v)) {:a 1 :b 2 :c nil}))
17:29	clojurebot	{:a 1, :b 2}
17:29	bmh	ztellman: Is there any reading you could suggest for GPU programming? I'm a bioinformatician by trade, so graphics is a big, magical world to me
17:29	ztellman	bmh: I haven't found a good conceptual book yet
17:30	ztellman	bmh: but the orange GLSL guide isn't all that bad as a reference
17:30	edoloughlin	Thanks everyone.
17:30	ordnungswidrig	is there something like zip for multiple seqs? [[1 2 3] [:a :b :c] ["!" "§" "$"]] -> [[1 :a "!"] [2 :b "§"] [3 :c "$"]]
17:31	raek	,(map vector [1 2 3] [:a :b :c] ["!" "§" "$"])
17:31	clojurebot	([1 :a "!"] [2 :b "§"] [3 :c "$"])
17:32	raek	map hase the useful property of "taking one thing from each seq"
17:32	bmh	raek: I had an interview earlier this week. I solved one of the problems they gave me with a few applications of map, filter, and reduce.
17:33	bmh	We spent the remainder of the interview discussing the finer points of roguelikes...
17:34	ordnungswidrig	ah, cool, I didn't know this one.
17:34	ordnungswidrig	,(doc map)
17:34	clojurebot	"([f coll] [f c1 c2] [f c1 c2 c3] [f c1 c2 c3 & colls]); Returns a lazy sequence consisting of the result of applying f to the set of first items of each coll, followed by applying f to the set of second items in each coll, until any one of the colls is exhausted. Any remaining items in other colls are ignored. Function f should accept number-of-colls arguments."
17:35	ordnungswidrig	,(map vector [1 2] [:a :b :c])
17:35	clojurebot	([1 :a] [2 :b])
17:36	hsuh	arete: 10x improvement with postgres-pool
17:37	ztellman	bmh: I'm heading out, but hit me up on the penumbra mailing list with any questions you have
17:38	bmh	ztellman: thanks. I'll subscribe.
17:39	arete	hsuh: sounds speedy
17:44	_rata_	has anybody problems with the debugger of swank-clojure and clojure 1.3?
17:45	lancepantz	so i was just playing around with tuning my jvm a little
17:45	lancepantz	i took 100m off of my Xmx, which dropped my virtual memory usage by 160m
17:46	lancepantz	anyone have any ideas where the extra 60 is coming from?
17:50	amalloy	lancepantz: what was your Xmx to begin with?
17:51	lancepantz	took it from 500 down to 400 in this case
17:52	amalloy	my guess, at any rate, would be some magic mumbo-jumbo to do with powers of two
17:52	_rata_	has anybody experienced problems with the debugger of swank-clojure and clojure 1.3?
17:58	nishant	does clojure have something like for-each?
17:58	nishant	bold question
17:59	AWizzArd	map, doseq, for
17:59	AWizzArd	,(map inc [10 20 30])
17:59	clojurebot	(11 21 31)
17:59	nishant	okay, I was looking for something that only side effects
17:59	AWizzArd	,(doseq [n [10 20]] (println n))
17:59	clojurebot	10 20
18:01	kmc	heh, clojurebot captures output?
18:05	nishant	thanks!
18:08	gfrlog	is there a faster way to create a map from a list of pairs aside from (apply hash-map (reduce concat pairs))?
18:08	dnolen	gfrlog: into is your friend
18:09	dnolen	,(into {} [[1 2] [3 4]])
18:09	clojurebot	{1 2, 3 4}
18:09	gfrlog	,(into {} [[1 2] [3 4]])
18:09	clojurebot	{1 2, 3 4}
18:09	gfrlog	awesome, thanks
18:11	kmc	,(doc into)
18:11	clojurebot	"([to from]); Returns a new coll consisting of to-coll with all of the items of from-coll conjoined."
18:11	kmc	,(into [1 2] [3 4])
18:11	clojurebot	[1 2 3 4]
18:12	kmc	,(conj [1 2] 3)
18:12	clojurebot	[1 2 3]
18:13	gfrlog	(into #{} {:a 2 :c 5})
18:13	gfrlog	,(into #{} {:a 2 :c 5})
18:13	clojurebot	#{[:a 2] [:c 5]}
18:19	gfrlog	,(defrecord Tom [jill tim])
18:19	clojurebot	DENIED
18:23	gfrlog	hmm. I do a (defrecord Graph [order edges]) in a certain namespace, (use) that namespace from the repl, but then Graph doesn't resolve
18:25	KirinDave	So
18:25	Chousuke	gfrlog: you need to import records
18:25	Chousuke	gfrlog: they become classes
18:25	gfrlog	oooh
18:26	gfrlog	very good
18:26	KirinDave	It's not possible to use the extend+merge technique to simulate inheritance with (reify ...), unless you want to use a macro and promise compile-time support. Right?
18:27	KirinDave	With (extend ...) you can merge two maps together
18:28	KirinDave	But extend requires a specific type to extend.
18:28	KirinDave	If you want anonymous types to fit a protocol, you have to have the entire proposition be static, right?
18:28	KirinDave	(unless, of course, you want to eval a form)
18:28	KirinDave	There's no way to provide default/inherited behavior to reified objects.
18:29	Chousuke	I suppose you could use reflection but it'd be fragile
18:30	KirinDave	Chousuke: I mean, a macro is easy enough to turn {:name (fn...) ...} into a '(name (fn...))
18:30	KirinDave	Chousuke: But...
18:30	KirinDave	That's purely static.
18:37	AWizzArd	How can I declare a deftype field as :unsynchronized-mutable and ^int at the same time?
18:43	_rata_	I don't get what make-distribution does
19:03	gfrlog	does (first #{2 3 4}) require the axiom of choice?
19:04	amalloy	AWizzArd: i think it looks something like this: (deftype a [^{:volatile-mutable true :tag int} h])
19:05	amalloy	that creates for me an A class, with a 1-arg int constructor. i'm not sure if it's mutable since i don't do deftypes very often, but it seems like if the tag works the mutability probly does too
19:07	AWizzArd	I will try that, thanks for the idea
19:15	gfrlog	,#{4 4 5}
19:15	clojurebot	Duplicate key: 4
19:19	jashmenn	hey, im wondering if someone can help me figure out a macro problem: https://gist.github.com/fdd8b7b902b1ab8170f7
19:20	jashmenn	i realize that macros don't evaluate their operands
19:20	jashmenn	but i'm wondering how i can evaluate them if i want to
19:21	gfrlog	jashmenn: I think you can only set them up to be evaluated
19:22	gfrlog	it looks like you're trying to call the function twice?
19:22	jashmenn	i want to use the var x as an argument to defining the output of the macro
19:22	jweiss	anyone know how to switch slime from clojure back to lisp? when i run slime it tries to load clojure
19:23	jweiss	or better yet, be able run either one :)
19:23	technomancy	jweiss: getting rid of swank-clojure should help
19:23	gfrlog	jashmenn: could you make foo a function? maybe one that calls another macro?
19:24	technomancy	I don't use CL myself though, I could use some help making swank more Cl-frinedly
19:24	jweiss	technomancy: so once swank-clojure is installed, it assumes you're only using clojure as inferior-lisp?
19:24	dnolen	jashmeen: there are two hidden params handed to macros &env and &form
19:24	jweiss	i don't know CL much myself. i learned a bit briefly before finding clojure.
19:24	technomancy	jweiss: maybe? getting rid of swank-clojure.el is a good idea even if you don't use CL. it's just got some heinous hacks that were necessary when lein didn't exist, but don't make sense any more.
19:25	technomancy	it just complicates things
19:25	jashmenn	dnolen: this sounds interesting. let me look that up
19:25	jweiss	technomancy: so it is not necessary to run lein swank, and then slime-connect?
19:25	jweiss	it meaning swank-clojure.el
19:25	technomancy	correct
19:26	jweiss	excellent, that's most likely the problem then, i set up clojure dev here before lein came along (or maybe just before i discovered it)
19:27	jweiss	technomancy: i have swank-clojure and clojure-mode elpa packages - do i need those?
19:28	technomancy	just clojure-mode and slime-repl
19:28	jashmenn	dnolen: okay, so i've got &env and it shows the LocalBinding. now im not exactly sure how to safely get the value of `g` out of it
19:29	dnolen	jashmenn: ah sorry you led you astray, &env is just so you can the surrounding context from w/in a macro
19:29	jashmenn	dnolen: ah okay
19:30	dnolen	jashmenn: the only way to get at the value of something is eval, but what exactly are you trying to do?
19:31	jashmenn	dnolen: i want to pass an argument into a macro and use the value of it (unquoted) to help build the sexp within the macro
19:33	jashmenn	but, like gfrlog was saying, i think i need to just set it up to be evaluated, not evaluate it early
19:35	gfrlog	jashmenn: can you use eval?
19:35	dnolen	jashmenn: curious as to the end goal, why do you need the actual value of a literal to influence the macro expansion?
19:36	dnolen	jashmenn: sometimes it is indeed useful to eval in a macro, but usually unnecessary.
19:36	gfrlog	I've never done it
19:37	dnolen	I've only needed once, because I was using the contents of files on disk to generate code.
19:41	amalloy	jashmenn: why is the value being passed in a variable instead of a literal? (not just asking to tell you you're wrong; i think you can probably make this happen)
19:49	rplevy	using clojure contrib 1.1.0, I am trying out contrib.condition. it doesn't do quite what I would expect. it certainly responds to the raised condition as specified but the exception goes uncaught delivering the message in tomcat. what I need it to do and what I think the semantics of it is, is for handler-case to be like try and for handle to be like catch in the sense that I can do something with the result. Am I doing it wrong or was it jus
19:50	technomancy	rplevy: support for catching regular exceptions in handler-case just got committed last week IIRC
19:50	technomancy	might be able to backport it to 1.1, but it might be tricky; easier to just upgrade to 1.3 =)
19:53	jashmenn	dnolen: well when i try to eval it says "can't eval a local"
19:53	dnolen	jashmenn: heh, yeah I forgot about that.
19:53	jashmenn	amalloy: right, if i pass in the value as a literal it works fine
19:53	jashmenn	but, obviously this is just extracted from a much larger macro
19:53	jashmenn	and `x` in this case "has to" be a variable
19:54	jashmenn	i don't mind passing it in as a literal in the `let` if you have advice on the best way to do that
19:55	amalloy	is this a macro embedded within another macro?
19:55	amalloy	or called from another, whatever
19:55	jashmenn	amalloy: yes
19:55	amalloy	okay. usually you don't need to do that
19:56	amalloy	write your inner macro as a function, not a macro
19:56	jashmenn	but what if i need both inner and outer to do list destructuring like ~@body ?
19:57	amalloy	eg (defn foo [x] `(prn ~x))
19:57	amalloy	backtick is legal in functions too; the only special thing about macros is that their arguments aren't evaluated
19:57	jashmenn	i didnt think that was possible with just a plain defn but i guess im wrong
19:57	jashmenn	hmmm
19:58	jashmenn	ah ha
19:58	jashmenn	okay
19:58	jashmenn	thanks
19:58	amalloy	,(let [x (fn [] `(prn 1))] (x))
19:58	clojurebot	(clojure.core/prn 1)
19:59	jashmenn	okay, this is very helpful. thanks amalloy (and everyone else)
19:59	amalloy	welcome jashmenn
19:59	amalloy	this is a macro "trick" that is widely applicable: don't use macros :)
20:06	tomoj	haha
20:08	kmc	i see a few libraries for continuations in Clojure, implemented by performing a continuation-passing-style transform
20:08	kmc	but it seems to me that CPS is fatally limited without TCO
20:08	kmc	is this a problem for using continuations in Clojure, or do they manage to insert trampolines cleverly to get TCO?
20:11	tomoj	kmc: http://intensivesystems.net/tutorials/cont_m.html
20:12	tomoj	does it look like that approach is limited by lack of TCO?
20:15	kmc	yes
20:15	kmc	most of those functions end in a tail call
20:15	kmc	in CPS you replace "return to parent, which does something else" with "tail-call a continuation"
20:15	kmc	this is going to use more stack, unless you have TCO
20:16	kmc	however i bet that some clever CPS macros for Clojure could replace calls with returns and use trampoline
20:16	kmc	which is what i was asking about
20:17	amalloy	You know we have trampoline, right?
20:18	kmc	yeah i mentioned it twice
20:18	amalloy	Haha, just checking that you didn't mean to implement it first
20:19	tomoj	would you just have to add a trampoline at every tail call site?
20:19	amalloy	On my mobile atm, apologies if I miss some context
20:20	kmc	tomoj, i'm not really sure
20:20	tomoj	I don't understand CPS yet :(
20:20	kmc	other languages implemented on systems without TCO sometimes wrap the whole program in one trampoline
20:20	kmc	and my understanding is that Clojure could have done this to get transparent TCO
20:20	kmc	but it would slow down everything
20:20	tomoj	ah, right
20:22	kmc	tomoj, in CPS every function takes an extra argument (the continuation) and, instead of returning, calls that
20:22	kmc	so (fn [x] (+ x 2)) becomes (fn [x k] (k (+ x 2)))
20:22	kmc	except that in a full CPS transform, we'd CPS-transform (+) too, and eliminate every non-tail call
20:23	kmc	(fn [x k] (+ x 2 k))
20:24	kmc	and (fn [x] (* (+ x 2) 3)) becomes (fn [x k] (+ x 2 #(* % 3 k))) or some such
20:27	kmc	and if you carry this to its conclusion, your big nested expressions get unfolded to a linear sequence of small steps
20:27	kmc	which is why CPS transform is useful as a stage in a compiler
20:28	kmc	and indeed is close to the SSA form used in lots of compilers for languages not traditionally called "functional"
20:29	tomoj	kmc: actually, doesn't the article there explain it?
20:29	tomoj	search for "But why?"
20:29	kmc	yeah
20:29	kmc	that's kind of orthogonal to what i said above
20:30	tomoj	oh
20:30	kmc	that's answering why you as the programmer should care about CPS
20:30	kmc	what extra power it gives you
20:30	kmc	not why a compiler should use it internally, or mechanically how it's carried out
20:31	tomoj	well, doesn't that article show a way to do the trampolining to get around lack of TCO?
20:31	kmc	yeah i see it now
20:32	tomoj	by placing bounces you can control which parts of the chain avoid function call overhead at the expense of stack space, I guess?
20:32	kmc	yeah
20:33	kmc	and if you baked the bounce into your monadic bind, you'd have full TCO at the expense of slower function calls
20:34	tomoj	unrelatedly, have you seen aleph?
20:34	kmc	no
20:34	tomoj	http://github.com/ztellman/aleph/wiki
20:34	tomoj	bells are going off in my head but I'm too stupid to tell whether it's a false alarm
20:35	kmc	ding
20:44	kmc	so how expensive are threads in Clojure?
20:45	kmc	are they 1:1 with Java threads or something more lightweight?
21:20	amalloy	i don't remember if it was tomoj or kmc who linked the intensivesystems monad tutorial, but thanks. so far i'm finding it a lot more understandable than the others i've looked at
21:21	amalloy	anyway, have to run, but just wanted to mention that
21:30	gfrlog	does anyone else think that c.c.sql could use an ensure-connection macro?
21:30	lancepantz	cc.sql could use alot of things
21:31	gfrlog	,(defn alot-of-things [& args])
21:31	clojurebot	DENIED
21:31	ninjudd	like a way to load huge datasets without bringing the whole thing into memory!!!
21:31	ninjudd	;)
21:33	lancepantz	gfrlog: there is a clj-sql project that addresses alot of the shortcoming, but i don't actually use it
21:34	lancepantz	i usually end up calling the jdbc methods directly
21:36	gfrlog	lancepantz: clj-sql doesn't have it -- are you suggesting that project might be easier to contribute to than cc?
21:36	lancepantz	gfrlog: yep
23:57	yayitswei	any pointers for setting up log4j with clojure.contrib.logging? just want a simple way to append-to-file