#clojure – 2009-05-09

2009-05-09

00:42	replaca	danlarkin: are you still hanging out?
03:01	cads	is it hard to write a custom m-expression syntax for a lisp?
03:03	cads	eww gross, I've used a m-exp language before, mathematica
03:04	cads	though the array slicing was nice
03:04	cads	but eeew, I don't want to program in mathematica
03:05	p_l	cads: so far, any m-exp system died...
08:01	rhickey	http://www.javaworld.com/javaworld/jw-05-2009/jw-05-clojure.html
08:06	wlr	rhickey: congrats on the mainstream attention ^^
08:07	rhickey	wlr: thanks - hopefully Java One will help broaden awareness as well
08:09	wlr	rhickey: if folks put aside irrational bias against lispy things and listen to what you say, it will
08:29	nakyss	how do i register here??
08:30	eee	,(doc cond)
08:30	clojurebot	"([& clauses]); Takes a set of test/expr pairs. It evaluates each test one at a time. If a test returns logical true, cond evaluates and returns the value of the corresponding expr and doesn't evaluate any of the other tests or exprs. (cond) returns nil."
08:31	eee	are there any other special keys besides :else?
08:31	kensanata	nakyss: /msg nickserv help
08:31	eee	it appears cond is behaving more like "finally" for me maybe?
08:35	nakyss	+r
08:37	nakyss	Hello all
08:37	nakyss	I'm try to solve this prblem
08:37	eee	hi
08:38	eee	me too
08:38	eee	:(
08:38	eee	in my case cond seems to find a condition AND execute the else
08:38	nakyss	I can't get a hold of an interface
08:39	nakyss	(def sourceDataLine (AudioSystem/getSourceDataLine audioFormat))
08:39	nakyss	(. sourceDataLine open audioFormat)
08:39	eee	which is the object
08:39	nakyss	and i get : java.lang.IllegalArgumentException: No matching method found: open for class com.sun.media.sound.DirectAudioDevice$DirectSDL
08:39	eee	and which is the method
08:39	eee	and which is the param
08:41	eee	open the method you want?
08:42	nakyss	the sourceDataLine var is supposed to be an SourceDataLine interface, it has nithing to do with com.sun.media.sound.DirectAudioDevice$DirectSDL...
08:45	eee	did cond have a bug in some version that I have?
08:45	nakyss	what's your problem eee?
08:47	eee	when I do a (cond), one of the cases works
08:47	eee	but it still seems to do the else
08:47	eee	in addition
08:47	eee	i use :else
08:47	nakyss	paste your code maybe
08:48	eee	i'll group pertinent stuff together
08:50	lisppaste8	eee pasted "cond question" at http://paste.lisp.org/display/79928
08:50	eee	it's printing "got here" "got here2"
08:51	eee	also, it prints the empty vector even though I tried to cull it out
08:53	Lau_of_DK	Good afternoon gentlemen
08:53	eee	can you try (.open sourceDataLine audioFormat)?
08:53	eee	hi Lau_of_DK
08:53	nakyss	yes it does thtet same error
08:54	nakyss	your cond form seems to end before all the conditions
08:54	eee	well well
08:54	eee	you found it
08:55	eee	now I know where lisp got that acronym!
08:55	eee	"infernal parens"
08:55	eee	thanks
08:56	nakyss	yr welcom
08:56	eee	that's a huge lesson to remember to check
08:57	nakyss	use an editor with paran matching
08:57	eee	funny thing is, I do
08:57	eee	but i should have gone back line by line
08:57	eee	once you mentioned that
08:57	eee	i just ran the cursor up the right side
08:58	eee	now I can move on the the next thing that has had me perplexed for a day
08:58	eee	which is that when I use 100 for that special spot, everything works
08:58	eee	but when I use 99 or below it doesn't
09:01	nakyss	Does anyone have any idea?
09:01	nakyss	(def sourceDataLine (AudioSystem/getSourceDataLine audioFormat))
09:01	nakyss	(println "sourceDataLine :" sourceDataLine)
09:01	nakyss	prints java.lang.IllegalArgumentException: No matching method found: open for class com.sun.media.sound.DirectAudioDevice$DirectSDL
09:02	eee	you are saying that sun.media thing isn't what you are working with?
09:02	nakyss	even though getSourceDataLine return a SourceDataLine interface
09:02	nakyss	yeap
09:02	eee	so fully qualify
09:03	nakyss	it does the same :(
09:04	eee	seems like in some java file the ide imported the wrong thing
09:04	eee	go look in the java code for a bogus import
09:04	nakyss	i have a java version of the codde which work greats so...
09:05	eee	then your clojure class path has things in a different order than your java version?
09:08	nakyss	well it's possiblebut package name should be unique so that shouldn't interfere
09:08	eee	i agree
09:09	eee	there might be a way to unimport something in clojure
09:09	eee	i'm pretty sure
09:09	eee	seems like a hack
09:09	eee	but if it works
09:09	eee	it will tell you something
09:10	eee	unimport that bogus stuff
09:11	nakyss	whether it's a bug in clojure or i'm doing somthing wrong, the former is very unlikely
09:12	nakyss	should i use proxy to deal with interface?
09:13	eee	i wish I knew more to help
09:13	eee	i'm not too competent in clj
09:13	nakyss	thx anyway me neither as you can see
09:14	eee	i'd try unimporting that bugus package
09:14	nakyss	not very active irc channel here?
09:14	eee	sometimes you get 2 or three talkers
09:16	nakyss	i'll go post on clojure google group then
09:16	eee	that will work
09:19	lisppaste8	eee pasted "works with 100 fails with 99, why" at http://paste.lisp.org/display/79930
09:19	eee	i don't know why it matters what blank string I use
09:20	eee	for the stupid scorer, the values should be arbitrary
09:21	eee	it works with a letter, too
09:21	eee	just not a string that represents a number less than 100
09:21	eee	sooooo confusing
09:21	acieroid	hmm
09:22	acieroid	how can i eval code from a string ? something like (foo "(* 6 7)") -> 42
09:23	eee	i've seen that in the discussion posts
09:24	acieroid	where are those discussions posts ?
09:26	eee	http://groups.google.com/group/clojure
09:26	acieroid	'k
09:26	acieroid	thanks
09:27	acieroid	ok great
09:27	acieroid	(eval (read-string "(* 6 7)"))
09:28	acieroid	so simple :D
09:28	eee	how'd you find it so fast?
09:28	acieroid	by searching in the group
09:28	acieroid	"eval string"
09:28	eee	cool
09:28	acieroid	a bit of luck \o/
09:29	eee	i have no approach for solving this prob
09:29	eee	perhaps I'll post the whole thing to the group soon
09:43	Cark	the article is not 100% acurate, but all in all it gives a good overview
10:55	acieroid	hmm
10:56	acieroid	how can I do smth like (re-seq #(str "foo" "\w+") foo)
10:56	acieroid	I mean, building regex at eval time, from concatened string or others functions
10:57	acieroid	hum, with re-pattern i think
11:26	replaca	acieroid: use re-pattern
11:26	replaca	oops, not yet enough coffee for reading comprehension!
11:28	danlarkin	replaca: you rang last night?
11:34	unlink	Does any .NET language have a decent REPL on Linux?
11:55	acieroid	replaca: :D
11:55	acieroid	thanks anyway
11:56	acieroid	:w
11:56	acieroid	oops
12:04	dreish	unlink: I think there are various projects to put Clojure on .NET, and there's rhickey's old project DotLisp. (That's the extent of my knowledge on the subject, though.)
12:06	unlink	I've tried fsi, ironpython, and booish, and they're all nearly unusable
12:20	dreish	What's wrong with them?
12:21	dreish	Without knowing much about it, I would have guessed ironpython would be handy, for example.
12:52	acieroid	does anyone have an exemple of using threads with clojure ?
12:53	dreish	,(.run (Thread. #(println "hello threads")))
12:53	clojurebot	hello threads
12:53	acieroid	oh, so simple
12:54	dreish	Make that .start to actually run in a different thread.
12:54	dreish	(I'm assuming that wouldn't work so well with clojurebot.)
12:54	dreish	,(.start (Thread. #(println "hello threads")))
12:54	clojurebot	nil
12:54	slashus2	dreish: You don't call .start?
12:54	dreish	Yeah, output goes off to nowhere.
12:54	slashus2	yeah
12:54	dreish	slashus2: You do. Just doesn't work with cjbot.
12:55	dreish	hiredman probably has "hello threads" somewhere at his repl now.
12:55	nakyss	can we import java class with the cjbot
12:56	durka42	,(import '[java.io File])
12:56	clojurebot	nil
12:56	dreish	acieroid: agents are maybe the more idiomatic way to do things in a different thread.
12:56	acieroid	agents ?
12:56	acieroid	I'll look at that
12:57	dreish	http://clojure.org/agents
12:57	nakyss	,(import '(javax.sound.sampled AudioFormat AudioInputStream AudioSystem LineUnavailableException SourceDataLine))
12:57	clojurebot	nil
12:57	durka42	,(binding [out out] (.start (Thread. #(println "hello threads"))))
12:57	clojurebot	nil
12:58	acieroid	but you know dreish, I just want that a call to a function don't block the repl
12:58	nakyss	(def sourceDataLine (AudioSystem/getSourceDataLine audioFormat))
12:58	nakyss	oops
12:59	nakyss	,(def sourceDataLine (AudioSystem/getSourceDataLine audioFormat))
12:59	clojurebot	DENIED
12:59	durka42	acieroid: then you want an agent
12:59	acieroid	ok
12:59	durka42	nakyss: no def
12:59	nakyss	no def :(
13:03	nakyss	sorry for the noise just testing some stuff
13:03	nakyss	,(.open (AudioSystem/getSourceDataLine (new AudioFormat 44100 16 1 true true)) (new AudioFormat 44100 16 1 true true))
13:03	clojurebot	java.lang.IllegalArgumentException: No line matching interface SourceDataLine supporting format PCM_SIGNED 44100.0 Hz, 16 bit, mono, 2 bytes/frame, big-endian is supported.
13:03	nakyss	does anyone know why this happened
13:06	nakyss	no just ignore that
13:44	slashus2	How do you access a static class inside of an interface like DataLine.Info in javax.sound.sampled?
13:47	liebke	slashus2: javax.sound.sampled.DataLine$Info
13:48	slashus2	Oh, I thought I tried that.
13:48	slashus2	That seems to work.
13:48	slashus2	I think
13:48	slashus2	alright, than kyou
14:28	cultureulterior	I'm new to clojure and lisp-like languages- what's the equivalent to python's star operator?
14:29	slashus2	multiply or in argument lists?
14:29	cultureulterior	argument lists
14:30	slashus2	(defn testfn [firstargument secondargument & containstherest])
14:33	cultureulterior	I must be asking wrong. I mean for when calling functions...
14:34	slashus2	cultureulterior: Can you show a python example of what you are trying to find the equivalence of?
14:35	cultureulterior	thefunction(*theargs)
14:36	cultureulterior	where theargs is a tuple or vector
14:38	slashus2	(apply function argumentlist)
14:40	cultureulterior	Thank you
14:46	stuhood	clojurebot: reduce?
14:46	clojurebot	I don't understand.
14:47	stuhood	clojurebot: reduce
14:47	clojurebot	No entiendo
14:47	stuhood	shoot... what is his quip about reduce not being lazy?
14:48	stuhood	~braindump
14:48	clojurebot	Titim gan �ir� ort.
14:50	durka42	~brain dump
14:50	clojurebot	brain dump is http://clj.thelastcitadel.com/clojurebot
15:28	nakyss	hi all
15:28	nakyss	how to get the string name of a symbol
15:29	nakyss	(def test nil)
15:29	nakyss	I want to print test
15:30	slashus2	,(str 'test)
15:30	clojurebot	"test"
15:30	Cark	(name 'test)
15:30	Cark	,(name 'test)
15:30	clojurebot	"test"
15:31	nakyss	thx
15:34	nakyss	it turns out it's not quite what I wanted
15:34	nakyss	I want to define such a fuction
15:34	nakyss	(defn debug [x]
15:34	nakyss	(println 'x ":" x))
15:34	nakyss	that would print the name of the var
15:34	Cark	mhh
15:34	nakyss	like (debug test) -> "test : nil"
15:35	slashus2	(var test)
15:36	nakyss	,(defn debug [x] (println (var x) ":" x))
15:36	clojurebot	DENIED
15:37	nakyss	it fails: Unable to resolve var: x in this context
15:37	slashus2	nakyss: You probably need to create a macro.
15:37	nakyss	any suggestions? plz
15:38	slashus2	nakyss: search for clojure debug macro into google and go to the first result.
15:38	eee	hiyas
15:40	Cark	(defmacro d [name] `(println '~name "=" ~name))
15:40	Cark	that should do it
15:40	Cark	usage : (d my-var)
15:41	Cark	hello eee
15:41	Cark	then you can do fun stuff like : (d (+ 1 3))
15:42	Cark	and refactor the macro to return the result
15:42	eee	i;m still stuck i've been really stuck for days
15:42	nakyss	ok great thx
15:43	eee	nakyss you got that problem fixed now?
15:43	eee	what was it?
15:43	AWizzArd	Cark: maybe you can embed the println in the macro inside a do
15:43	rhickey	eee: stuck on what?
15:43	nakyss	eee do you earlier today or just now
15:44	nakyss	> eee do you mean earlier today or just now
15:44	AWizzArd	Cark: or in a let, if you don't want to compute the result twice. Example: (defmacro d [name] `(let [result# ~name] (println '~name "=" result#) result#))
15:44	eee	now I gotta figure out how to put it back the way it was
15:44	eee	i have a 15-puzzle
15:44	eee	where it shouldn;t matter what strings I use
15:44	eee	i define one to be the blank
15:44	eee	that moves around
15:45	Cark	AWizzArd : sure
15:45	eee	when I make it a letter
15:45	eee	it works fine
15:45	eee	when I make it a number string less than 100
15:45	eee	it doesn't work
15:45	eee	but my heuristic doesn't care about numbers
15:45	eee	just equality
15:45	Cark	AWizzArd : though you might want to define a prog1 macro first for easier reading
15:45	Cark	or a returning function =)
15:45	eee	cause I didn't switch to the manhatten heuristic yet that does care about numbers
15:46	eee	i've pasted the code before
15:46	eee	but let me get it back to that state if I can
15:46	eee	made a generic a-star search with the heap
15:46	eee	trying to show some examples in clojure
15:48	nakyss	keep it concise eee. here isn't really a place to post go to google group :)
15:49	eee	this might have ben one prob:
15:49	eee	,(= [] nil)
15:49	clojurebot	false
15:50	eee	which I seemed to remember from lisp
15:55	lisppaste8	eee pasted "works with 100 fails with 99, why" at http://paste.lisp.org/display/79939
15:55	eee	probably pretty wacky code ... i'd built it up from smaller things
15:59	eee	it runs forever if you use "99" for the blank
16:00	eee	or "98"
16:00	eee	must be something low-level I don't get
16:01	rhickey	eee: right, (= [] nil) is bad, what are the possible values of goal-s ?
16:01	rhickey	are you using [] as a 'done' sentinel?
16:02	eee	no, trying to throw those out
16:02	eee	because they are bad transitions
16:02	rhickey	if so, then (if (seq goal-s) ...
16:02	eee	if the blank is in the cornet or on the side
16:02	eee	corner
16:03	eee	ok
16:03	eee	so i thought with "next" that worked the way it is
16:03	eee	with rest I need seq
16:03	eee	as tested before
16:03	eee	will make the change tho
16:03	eee	no change
16:04	eee	runs forwever with "99" solves with "100" or "B" for the blank
16:04	eee	but what I use for my string should be arbitrary in my mind ... for score-state-stupid
16:06	lisppaste8	eee annotated #79939 "made emptyVec? predicate" at http://paste.lisp.org/display/79939#1
16:07	rhickey	eee: are you comparing strings somewhere?
16:07	eee	in case I didn't answer the question, '(["1" "2" "3" "4" "5" "6" "7" "8" "9" "10" "11" "12" "13" "14" "15" "99"])
16:07	eee	that is the goal
16:07	eee	yes
16:08	eee	comparing strings in 'score-state-stupid
16:08	eee	well comparing list of strings
16:08	eee	element by element
16:09	rhickey	eee: you are saying it behaves differently if you replace the 2 "99"s in the call with "100"s?
16:09	eee	yes
16:10	eee	it completes right away
16:10	eee	well no
16:10	eee	one other place too
16:10	rhickey	3 places to change
16:10	eee	yes
16:14	eee	rhickey, I really appreciate you looking ... I feel bad now that I have to take off in a minute or two
16:15	rhickey	if the strings are treated completely like symbols, this is hard to believe - you never do math with them ?
16:15	eee	not that I know of
16:15	eee	I can check it all in like this
16:16	eee	and it's in my heap project
16:16	eee	i was about to do math in the new heuristic
16:16	eee	which is why I found the problem
16:17	eee	http://code.google.com/p/jc-pheap/source/checkout ... if this is interesting
16:17	eee	i hope it's not something too dumb
16:18	eee	gotta go, will leave my notebook open in case there's more on the subject. Thanks again
16:20	cultureulterior	What version of java should I be running for an an optimal clojure experience? I'm referring here to GC and such
16:20	AWizzArd	,(* 0.4 150 1.5)
16:20	clojurebot	90.0
16:21	AWizzArd	,(* 1.5 0.4 150)
16:21	clojurebot	90.00000000000001
16:21	AWizzArd	rhickey: do you see that?
16:21	gnuvince_	AWizzArd: in Python:
16:21	gnuvince_	>>> 0.4 * 150 * 1.5
16:21	gnuvince_	90.0
16:21	gnuvince_	>>> 1.5 * 0.4 * 150
16:21	gnuvince_	90.000000000000014
16:22	rhickey	AWizzArd: I see floating point doing what it always does
16:23	rhickey	,(* 4/10 150 15/10)
16:23	clojurebot	90
16:23	rhickey	,(* 15/10 4/10 150)
16:23	clojurebot	90
16:26	AWizzArd	I tried clisp. For both forms it returned 90.0
16:27	gnuvince_	AWizzArd: try C, C++, Java, Python, Ruby, Haskell, etc. they'll all give the same answer as Clojure I bet.
16:28	AWizzArd	I just tried Ruby. Got 2x the 90.0
16:28	svdm	,(apply * (map bigdec [1.5 0.4 150]))
16:28	clojurebot	90.00M
16:28	AWizzArd	In C, C++, Java I won't be surprised
16:29	AWizzArd	there I expect that floats do what every computer scientist knows
16:29	gnuvince_	AWizzArd: Haskell, Python and OCaml gave me the same answer as Clojure.
16:29	gnuvince_	Don't have Ruby installed.
16:29	hiredman	well, clojure floats are java floats, so ...?
16:30	gnuvince_	I don't see the problem either since Clojure can just use BigDecimals or Rationals.
16:30	AWizzArd	Should it not default to correct results? We should ask for speed if correctness is not preferred.
16:30	hiredman	it is correct
16:31	hiredman	it follows the spec for floating point math
16:31	hiredman	if you don't want floating point math, don't use floats
16:31	AWizzArd	it is clear that it must give these outputs of the FPU is used, but the results are mathmatically not correct
16:31	AWizzArd	of the FPU ==> if the FPU
16:32	hiredman	they are correct for floating point math
16:32	hiredman	which is what happens when you use floats
16:33	AWizzArd	Why are floats used as default and not only when one explicitly askes for them?
16:33	hiredman	,(class 1)
16:33	clojurebot	java.lang.Integer
16:33	AWizzArd	that would match the behaviour for integers
16:33	hiredman	,(class 0.1)
16:33	clojurebot	java.lang.Double
16:33	AWizzArd	(+ 999999999999999999999999999999999999999999 99999999999999999999999999999999999999999999999999999999)
16:33	AWizzArd	,(+ 999999999999999999999999999999999999999999 99999999999999999999999999999999999999999999999999999999)
16:33	clojurebot	100000000000000999999999999999999999999999999999999999998
16:33	AWizzArd	no overflow
16:33	AWizzArd	why correct behaviour here but not for floats?
16:34	slashus2	I guess that is the behavior that most people expect.
16:34	hiredman	because the bahviour is correct for floats, by definition
16:34	rhickey	floating point is an inexact representation, as you perform successive operations you accumulate error - there are some numbers it simply can't represent
16:35	AWizzArd	right, it can represent nearly none of the real numbers
16:36	AWizzArd	I did not expect 90.000000000000014 when I tried 1.5 * 0.4 * 150 in Ruby or clisp
16:36	AWizzArd	uhm, that last one was for slashus2 :)
16:37	gnuvince_	AWizzArd: and should Clojure do the right thing when (+ 1 "2") obviously meant (+ 1 2)?
16:38	AWizzArd	I can see a difference between (+ 1 "2") and (* 1.5 0.4 150)
16:38	AWizzArd	Why are those numbers floats or doubles and not BigDecimals?
16:38	hiredman	well, obviously, I meant I wanted to concat the toString of 1 to "2"
16:39	gnuvince_	AWizzArd: just write a patch
16:39	gnuvince_	But I don't think you'll get much traction.
16:39	AWizzArd	Can that be written in Clojure or does it require Java knowledge?
16:39	hiredman	I don't think java does the auto-promotion stuff with bd
16:40	hiredman	so you would have to cast every single time you pass a number to a java method
16:40	AWizzArd	,(+ 1.5M 0.4M 150)
16:40	clojurebot	151.9M
16:40	AWizzArd	,(* 1.5M 0.4M 150)
16:40	clojurebot	90.00M
16:40	hiredman	,(.charAt "foo" 1M)
16:40	clojurebot	\o
16:40	AWizzArd	Maybe the M should be implicit, while floats or doubles require a F or a D
16:40	hiredman	huh
16:40	hiredman	go figure
16:41	gnuvince_	And break backwards compatibility?
16:41	gnuvince_	Rich and Stu are gonna love you :)
16:41	AWizzArd	Would it break backwards compatibility if the M will still be supported?
16:41	gnuvince_	I think defaulting to Double is a fine choice
16:42	eee	your questions helped me figure it hour rhickey
16:42	gnuvince_	It's what the Java people expect.
16:43	AWizzArd	gnuvince_: and would getting overflows for multiplying big numbers also be a good choice then?
16:43	eee	drove back to tell you not to waste the cycles:
16:43	eee	:)
16:43	AWizzArd	I think a good choice would be: correctness first, speed when people ask for it
16:44	AWizzArd	I think the typical Java programmer is intelligent enough to live with mathematically correct results. It won't reduce their productivity very much I guess.
16:44	chrizel	for most applications floating point math is enough
16:44	AWizzArd	I am not against the use of the FPU
16:45	AWizzArd	one could just simply ask for floats
16:45	AWizzArd	for most applications correct math would also be enough
16:45	chrizel	no because its slower
16:45	eee	rhickey: the answer is that the heap breaks ties using the hash code of the object. so 99 has a different hash than 100 ... thus in the course of A-star . . . the order of pop can be different if different states have the same score
16:46	AWizzArd	chrizel: Does that mean you don't like Clojures behaviour in not letting integer operations overflow?
16:46	AWizzArd	What Clojure is now doing is slower. It could simply use ints and longs under the hood und overflow for more speed.
16:46	eee	ok see yah again
16:47	AWizzArd	And if someone cares for speed: why not specify it? (float 1.5) or 1.5F
16:47	AWizzArd	In nearly all aspects Clojure has the motto: let us do the correct thing by default.
16:47	slashus2	AWizzArd: It would be a unique direction to go. Would be pretty neat, but would be at odds with the behavior of most programming languages.
16:48	slashus2	Clojure is pretty unique anyway.
16:48	AWizzArd	I just wanted to say that ;)
16:48	AWizzArd	people who don't come from the Lisp world will see 80% of new stuff
16:48	AWizzArd	imo it's a minor issue to live with mathematically correct results
16:49	gnuvince_	Not if every time you want to send x to a Java method, you need to cast it back to a double
16:50	gnuvince_	The thing with the Integer/BigInteger promotion is that it happens only when needed. Otherwise, it stays an Integer.
16:50	gnuvince_	So you can say (.foo bar (* x y)) without having to cast that down to Integer
16:51	AWizzArd	What if (* x y) results in a number of 600 digits?
16:52	gnuvince_	AWizzArd: then you couldn't pass it to that method anyway
16:52	gnuvince_	As it would expect an Integer parameter.
16:52	AWizzArd	Can't Clojure do this casting back to float/double under the hood
16:52	AWizzArd	?
16:53	chrizel	How would you determine whether a float is "too big" or would be incorrect? (are there any languages doing this?)
16:53	AWizzArd	You mean a language implementation?
16:54	gnuvince_	AWizzArd: it could, but if you have a BigDecimal holding an unrepresentable Double value, the result would be changed during the conversion, so the result would be off.
16:54	gnuvince_	Let's say there's a trivial method called addOne. (.addOne myBigDecimal)
16:54	AWizzArd	yes
16:55	gnuvince_	You'd expect an answer that's exactly one larger than myBigDecimal.
16:55	gnuvince_	However, because of the inherent imprecision of doubles, that could not be case.
16:55	AWizzArd	does addOne expect a float, a double or a BigDecimal object?
16:55	AWizzArd	If it expects a float or double I would not expect a correct result
16:56	gnuvince_	I assume here it's expecting a Double, sicne we're talking about implicit conversion from BigDecimal to Double
16:56	gnuvince_	since*
16:56	AWizzArd	then I would not expect that the number is exactly one more than my input
16:57	AWizzArd	It could mean that the program must stop or do something else.
16:57	gnuvince_	Why not
16:57	gnuvince_	"Clojure has correct floating point mathematics"
16:57	AWizzArd	The potential problem is only data that comes into the program during execution
16:58	AWizzArd	for all data that is a consequence of my program I can find out if there are problems by just running my code
16:58	AWizzArd	If new data enters a program each good programmer will let it only go deeper if it meets the requirements
16:59	AWizzArd	If you read a big double and add one it could overflow and become a very big negative number
16:59	AWizzArd	or not?
16:59	AWizzArd	maybe it can become infinite instead, not sure
17:00	AWizzArd	so, I must check if my input to which I later plan to add one will meet the requirements at the time the data flows into my program
17:00	AWizzArd	If it can pass my tests then the conversion from BigDecimal to double would not cause problems and I get my number+1
17:01	slashus2	,(Math/pow 50000 50000)
17:01	clojurebot	Infinity
17:01	AWizzArd	exactly
17:01	acieroid	hmm, i'd like to do something like (take 100 foo), with a string foo, but it returns me a list of char. How can I have a string ?
17:02	AWizzArd	Each good programmer will check input data. If the input is so big that adding one will cause that Infinity problem, the program must not accept the data.
17:02	gnuvince_	AWizzArd: how long do you think it would take to compute the real answer?
17:02	chrizel	acieroid, (apply str (take 100 foo))
17:03	AWizzArd	does it have to do with why (* 1.5 0.4 150) should result in 90.000000000000014?
17:03	acieroid	thanks :o
17:03	slashus2	Has anyone completed a pow and sqrt clojure implementation that does correct math?
17:03	gnuvince_	Whatever
17:04	gnuvince_	This discussion is going nowhere, I'm walking out. If you truly care about it, put your code where you mouth is and submit a patch that we can actually play with instead of mentally jerk off over.
17:05	AWizzArd	slashus2: for only a very small set of real numbers there exists a result that can be algorithmically expressed
17:05	AWizzArd	for calculating roots and so on
17:06	AWizzArd	gnuvince_: do you see it is legitimate to report bugs and discuss about things that one percieves as bugs without providing a patch?
17:08	slashus2	The BigDecimal pow function only takes in integer, which isn't the best way to go. I don't think.
17:08	slashus2	integers*
17:09	AWizzArd	agreed
17:09	slashus2	It is crippled compared to Python's Decimal package
17:10	AWizzArd	yes, seems that way
17:10	AWizzArd	maybe this will become better when Sun .. hmm.. Oracle wants to push Fortress a bit more?
17:10	AWizzArd	(if they want to do so at all)
17:12	gnuvince_	,(let [v [1 2 3 4]] (time (dotimes [_ 1e6] (first v))))
17:12	clojurebot	"Elapsed time: 239.412 msecs"
17:12	slashus2	I think Python's decimal suffers from the same thing, in that you can't raise a Decimal to a non-integer power. It does support sqrt though.
17:12	gnuvince_	,(let [v [1 2 3 4]] (time (dotimes [_ 1e6] (v 0))))
17:12	clojurebot	"Elapsed time: 102.065 msecs"
17:13	AWizzArd	There is this gnu calculator program, don't remember its name right now.. can that do 1.2345^6.757 ?
17:16	AWizzArd	clisp can :)
17:16	gnuvince_	bc
17:16	AWizzArd	ah right, bc, thx
17:16	slashus2	What is special about 1.2345 to the 6.757?
17:17	AWizzArd	uh nothing, just typed some digits
17:17	slashus2	,(Math/pow 1.2345 6.757)
17:17	clojurebot	4.151502692911114
17:18	AWizzArd	,(Math/pow 1.2345M 6.757)
17:18	clojurebot	4.151502692911114
17:18	AWizzArd	so here it worked for a BigDecimal?
17:18	AWizzArd	(.pow 1.2345M 6.757)
17:18	slashus2	I don't think so
17:19	AWizzArd	,(.pow 1.2345M 6.757)
17:19	clojurebot	3.539537889086624823140625M
17:19	slashus2	That didn't work like you think it did. It changed 6.757 to 6
17:19	AWizzArd	,(.pow 1.2345M 6)
17:19	clojurebot	3.539537889086624823140625M
17:19	AWizzArd	this is as you said right? The BigDecimal class has no pow method for doing what I wanted
17:19	slashus2	right
17:20	slashus2	It doesn't even have a sqrt
17:20	slashus2	:-(
17:20	AWizzArd	What surprised me is that after nearly 15 years of Java there still is no class for complex numbers
17:21	AWizzArd	,(* 1.5 0.4 150)
17:21	clojurebot	90.00000000000001
17:21	AWizzArd	,(* 1.5 0.4 150.0)
17:21	clojurebot	90.00000000000001
17:22	AWizzArd	gnuvince_: can you try in Haskell: 1.5 * 0.4 * 150 vs 1.5 * 0.4 * 150.0 ?
17:22	gnuvince_	Same as in Clojure, already told you.
17:23	AWizzArd	now I changed something: the 150 became 150.0 (point zero added)
17:23	gnuvince_	ghci> 1.5 * 0.4 * 150
17:23	gnuvince_	90.00000000000001
17:23	gnuvince_	ghci> 1.5 * 0.4 * 150.0
17:23	gnuvince_	90.00000000000001
17:23	AWizzArd	k
17:26	gnuvince_	It wouldn't change anything
17:26	gnuvince_	because there are Double in the expression, the 150 is infered to be a Double as well.
17:27	AWizzArd	makes sense
17:32	AWizzArd	btw, does .net have complex numbers and an equivalent to BigDecimal?
17:37	gnuvince_	~seen Cark
17:37	clojurebot	Cark was last seen in #clojure, 111 minutes ago saying: or a returning function =)
17:42	gnuvince_	Cark: if you look at my clj-starcraft project, I've made a change that gave me ~12% speed improvement over before; I replaced some destructuring bindings with explicit vector indexing.
17:43	slashus2	I wouldn't that that destructuring would be that inefficient.
17:43	slashus2	think*
17:45	AWizzArd	gnuvince_: can you please say a bit more about it?
17:45	gnuvince_	"it"?
17:46	AWizzArd	the speed improvement
17:47	gnuvince_	I had a program that took 80s to parse 1,050 files. In some deeply nested loops, I replaced things like (let [[a b c] v] ...) with (let [a (v 0) b (v 1) c (v 2)] ...) and now the execution time is down to ~70s
17:47	gnuvince_	My goal is to get within 50% of Java's performance without sacrificing too much of the Clojure feel.
17:47	gnuvince_	(Java takes 12 seconds for the same 1,050 files)
17:47	AWizzArd	surprising
17:50	gnuvince_	Not unexpected though; the vector -> seq conversion can really start to make a difference when it's done roughly 3.5M times.
17:51	gnuvince_	The two largest bottlenecks in my code at the moment are calls to first and calls to aset-int
17:54	AWizzArd	what profiler do you use?
18:05	AWizzArd	Btw, an interesting video about self-improving AI which also deals with nano technology. It's the official Stanford University youtube channel, and they also mention that in about 10-15 years we may have processors with a billion cores :)
18:05	AWizzArd	http://www.youtube.com/watch?v=omsuTsOmvsc&fmt=18
18:55	chrizel	Should the function docstring be before or after the parameter list? I saw different styles, there seem to be no strict convention.
18:55	cmvkk	the docstring has to come before the parameter list.
18:57	hiredman	after the param list is a bug
18:58	cmvkk	technically it wouldn't cause an error, I guess
18:58	cmvkk	but it wouldn't be a docstring
18:58	hiredman	it would not, but it would not give you a docstring either
19:00	chrizel	most clojure code seems to do this -- after a grep through clojure-contrib there are at least in monads.clj some funktions with the other style...
19:01	hiredman	I cannot speak for contrib, core definetly has the docstrings before the params list
19:01	chrizel	ok i see, the other style is not a real docstring because it's not saved in the meta data
19:01	chrizel	thank you, now it makes sense
19:06	chrizel	ironically this docstring inconsistency is even in the ants.clj example by rich himself - that's what confused me too
19:08	cmvkk	haha, you're right.
19:09	cmvkk	to be honest, it's not very intuitive that the docstring is in that particular location. but that's how it has to be.
19:11	liebke	cmvkk: it makes more sense when you have a function with multiple arities
19:11	cmvkk	yep. that's why it's necessary.
19:13	cads	hey, I'd like to see if a set a is a subset of set b, should I just (defn subset? [a b] (= (clojure.set/union a b) b)) somewhere?
19:20	edgar	hello
19:20	edgar	i have see the web site of clojure
19:20	edgar	and i see is a language for jvm
19:21	edgar	is there plans to make it a language such as ruby or python? independent of the JVM
19:21	liebke	cads: that should work, you can also do (empty? (difference a b))
19:22	ataggart	edgar: one of the points of being on the JVM is to benefit from the many libraries available in java
19:22	ataggart	though I think there is a project to try and get clojure running on dot net
19:25	edgar	well, on dot net may be great
19:28	ataggart	why?
19:30	ataggart	what deficiency do you see by running on the JVM?
19:30	unlink	Running on the JVM is supposed to be a strength -- you can leverage a huge body of existing libraries.
19:31	ataggart	plus the JVM itself is badass
19:31	unlink	Though I'm not sure I'm a fan of creating a separate class for every function.
19:31	ataggart	I think (though may be completely off the mark) that that might be addressed with the addition of invokedynamic
19:33	unlink	What is that? A planned JVM feature?
19:33	dnolen	has anybody messed around with protecting java primitive arrays with the Clojure reference types? Is this a good approach?
19:33	ataggart	new opcode for the jvm in JSE 7
19:34	unlink	OK.
19:34	unlink	Interesting.
19:37	unlink	In my mind the only big thing really lacking from Clojure is static typing.
19:38	ataggart	funny, I thought the lack thereof was supposed to be a feature ;)
19:38	unlink	It seems to have been omitted by design, but once you've used ML, it seems like there's no excuse to leave it out.
19:39	unlink	Algebraic types and type inferencing make static typing useful, usable and powerful.
19:39	ataggart	example?
19:40	unlink	type tree = Null \| Leaf of int \| Node of tree * tree
19:41	unlink	Haskell has an even more powerful type system, but some find it overwhelming.
19:42	ataggart	I should say, what are you not able to do as easily, usefully, and powerfully without the static typing?
19:42	dnolen	in Haskell, is it possible to have a List of things that are not of the same type? Just curious... I've been meaning to investigate it.
19:43	unlink	You can't in SML, but I find that a feature. I have never wanted a heterogeneous list. Every time I have created one, it was in error. (except in Lisp, of course, I'm referring to using data as data, not as code).
19:44	unlink	got to run...
19:44	hiredman	that is an interesting point
19:44	ataggart	which brings its own philosophical issue
19:44	hiredman	the issue of homoiconicity
19:44	unlink	Is static typing irreconcilable with homoiconicity? discuss.
19:44	unlink	:)
19:45	ataggart	lol
19:46	ataggart	I do like strong typing when it comes to what my IDE can do for me.
19:46	ataggart	not having all those nifty features was one of the things that made me hesitant to start with dyanamic languages
19:47	ataggart	I recall with my brief foray into ruby, asking a fellow developer how I can find where a function was defined and who was using it. His reply: search the code base.
19:48	ataggart	I went back to java the next day
19:48	AWizzArd	in principle it could be possible that one can put an optional type system on top of dynamic languages
19:48	ataggart	and clojure sort of has that
19:48	ataggart	with meta data
19:49	AWizzArd	meta data could be a place where to store those type information
19:49	ataggart	yup, and multimethods can trigger off of it
19:49	AWizzArd	it can go even deeper
19:50	AWizzArd	http://groups.google.com/group/clojure/browse_frm/thread/7a48f48e4b197a15/47389c2a78e82b38
19:50	hiredman	well, metadata cannot be attached to everything
19:51	ataggart	java strings and primitives iirc, right?
19:51	hiredman	java objects in general
19:51	ataggart	mm ya
19:51	AWizzArd	those already carry type information I guess.. don't know if that information still exists in the bytecode
19:51	hiredman	,^(with-meta (Object.) {:foo 1})
19:51	clojurebot	java.lang.ClassCastException: java.lang.Object cannot be cast to clojure.lang.IObj
19:51	hiredman	,^(with-meta [] {:foo 1})
19:51	clojurebot	{:foo 1}
19:52	ataggart	a not altogether unreasonable cost for switching to using java stuff directly
19:52	hiredman	AWizzArd: if that was sufficient, why are type hints needed to prevent reflection?
19:52	ataggart	though I can imagine it hurts when common clojure value types (e.g., strings) have that limitation
19:52	AWizzArd	I think rhickey once mentioned that type information still exists in the byte code. So for optional static typing support this should not be a problem
19:53	AWizzArd	hiredman: I don't know enough about the internal. I just think that reflection is needed when Clojure comes into play
19:54	AWizzArd	(defn foo [a b] (a b)) <-- here it is not clear (from that code snippet alone) what a and b will be at run time
19:54	AWizzArd	With gradual typing however this can be found out before compilation
19:54	hiredman	yet (defn a [x] (.toString x)) seems very clear
19:55	hiredman	but unless you type hint x, you will get a reflection waring
19:55	AWizzArd	yes, for you and me it is clear
19:55	AWizzArd	the clojure compiler however does not make use of that information
19:55	AWizzArd	that is the dynamism about the language
19:56	AWizzArd	only at runtime it will be clear what type the object x has
19:56	hiredman	AWizzArd: but you are calling a java method, if the java objects (which all clojure things are anyway) are carryng type information, what do we need reflection for?
19:56	AWizzArd	because Clojure is dynamically typed. Can you tell what class x will have?
19:57	AWizzArd	there are thousands of .toString methods. Which should be called?
19:57	AWizzArd	currently Clojure uses type information only for speedups. That is great. I think ataggarts point is that this information could also (and I step in and add "optionally") be used to make compile time checks
19:58	AWizzArd	In a java program variables have a type. In Clojure those are untyped, but objects (at runtime) have a type.
19:58	ataggart	actually, I'm not sure I have a point. :)
19:58	ataggart	I'm on my 4th beer
19:58	AWizzArd	A java program can always look up when you say x.toString() what type x must have
20:00	AWizzArd	uh, it is late.. a java program can of course not look up anything, as that is just a string. I mean: javac can extract the information what type x has out of the java program file(s)
20:01	ataggart	yes, which carries over into the IDE issue, namely, with the strong/static typing everything that exists is known to exist ex ante
20:01	ataggart	with dynamic languages you can't know ex ante what's what
20:01	AWizzArd	in principle you can.. it just is not very easy
20:01	ataggart	well, yeah, you just have to run the code
20:02	AWizzArd	you can, if you add a static type system on top of the dynamic language
20:02	ataggart	well yes
20:02	AWizzArd	that type system could even sit completely in your editor
20:02	ataggart	I think I'm okay with allow us to make ad hoc type systems
20:03	AWizzArd	so the compiler won't know anything about the types
20:03	AWizzArd	the compiler could be the same. Only your Clojure editor knows about the types.
20:03	AWizzArd	When you say (def x 10) it can tell you that it thinks that x should be of type "byte".
20:03	ataggart	it'd be nice to get an editor to do that
20:04	AWizzArd	when you later do a (count x) it looks up that count expects a collection, not a byte
20:05	AWizzArd	all type hints could be put into metadata or comments, and the editor can use code folding to hide them by default
20:07	AWizzArd	ataggart: if you have some time click the link I gave you 15 minutes ago. In that posting there is a video linked (I think in Meikels post)
20:07	ataggart	k
20:07	ataggart	I'm reading the thread
20:09	ataggart	watching...
20:09	AWizzArd	hiredman: btw, how do you turn on the reflection warnings? Via a binding before you compile?
20:22	slashus2	(set! warn-on-reflection true)
20:23	gnuvince	AWizzArd: both -Xprof and -Xrunhprof:cpu=samples report roughly the same numbers.
20:25	slashus2	If clojure ever uses involkdynamic in its implementation, will it improve performance? I guess this will be a while since involkdynamic would be incompatible with JVM 5 and 6.
20:34	AWizzArd	slashus2: ah good thank. I tried alter-var-root ^^
20:34	AWizzArd	gnuvince: didn't know about those options, thanks
20:40	slashus2	No one has any idea about the implications of invokedynamic?
20:53	slashus2	I guess that it could only use the invokedynamic technique when invokedynamic exists, and the current way when not.
20:54	cmvkk	yeah, a separate version of the compiler for a new java version
20:54	cmvkk	there are a few good articles out there about exactly what invokedynamic will do for languages like clojure
20:54	cmvkk	but i don't know what they are offhand
20:56	eee	i thjought this thing would keep the full history
20:56	eee	dern it
20:56	eee	now i don't know what rhickey said
20:57	hiredman	~logs
20:57	clojurebot	logs is http://clojure-log.n01se.net/
20:57	eee	is the log saved some
20:57	eee	er
20:57	eee	thanks
20:57	eee	:)
21:02	eee	i iguess not much of today in the logs yet
21:02	clojurebot	everyday is Rich Hickey Appreciation Day
21:03	slashus2	I think he said that invokedynamic wouldn't do much for clojure.
21:04	hiredman	~google dynamic languages and invokedynamic
21:04	clojurebot	First, out of 1840 results is:
21:04	clojurebot	Invoke dynamic languages dynamically, Part 1: Introducing the Java ...
21:04	clojurebot	http://www.ibm.com/developerworks/java/library/j-javascripting1/
21:04	hiredman	:(
21:11	eee	is the way to go from "3" to '3 to use javas Integer/parseInt
21:11	eee	?
21:12	eee	or is there something in clojure?
21:14	hiredman	,(symbol "3")
21:14	clojurebot	3
21:14	hiredman	,'3
21:14	clojurebot	3
21:15	cmvkk	interesting.
21:15	hiredman	,(= (symbol "3") '3)
21:15	clojurebot	false
21:15	Raynes	,(read-string "3")
21:15	clojurebot	3
21:15	hiredman	,(class '3)
21:15	clojurebot	java.lang.Integer
21:15	Raynes	Mines better.
21:15	Raynes	:)
21:15	eee	ok
21:16	cmvkk	your actually works.
21:16	cmvkk	,(+ (symbol "3") 4)
21:16	clojurebot	java.lang.ClassCastException: clojure.lang.Symbol cannot be cast to java.lang.Number
21:16	eee	(read-string ("println \"Hello\""))
21:16	eee	,(read-string ("println \"Hello\""))
21:16	clojurebot	java.lang.ClassCastException: java.lang.String cannot be cast to clojure.lang.IFn
21:16	Raynes	,(read-string "2.3")
21:16	clojurebot	2.3
21:16	hiredman	anyway, it depends what you mean by '3
21:17	eee	,(read-string "(println \"Hello\""))
21:17	clojurebot	java.lang.RuntimeException: java.lang.Exception: EOF while reading
21:17	eee	,(read-string "(println \"Hello\")")
21:17	hiredman	'foo general means the symbol foo
21:17	clojurebot	(println "Hello")
21:17	eee	cool
21:17	Raynes	I remember read-string because I was pissed off after writing my own function to do that for decimals and finding it.
21:17	eee	read-string is god
21:17	hiredman	if '3 means the symbol 3
21:17	hiredman	(symbol "3")
21:17	eee	,(read-string "(println \"Look at it go\")")
21:17	clojurebot	(println "Look at it go")
21:17	cmvkk	hmm...it hadn't occurred to me before this that '3 could be both a symbol and a number.
21:17	eee	clever colution!
21:18	cmvkk	i don't think '3 should be a valid symbol.
21:18	cmvkk	i can't think of a situation in which that would be useful...
21:19	eee	well, it's just a matter of not making an exception
21:19	eee	stuff can start with numbers
21:19	eee	start with means "followed by n >= 0 stuff"
21:20	replaca	is stuart sierra around these parts anywhere?
21:20	cmvkk	i guess it would mostly never come up, since (symbol "3") is the only way to make that symbol.
21:20	cmvkk	since '3 just becomes the number 3.
21:23	slashus2	agony?
21:24	Raynes	slashus2: I have very little experience in languages that aren't totally wrapped in parentheses. After using Clojure, looking at anything else hurts my eyes. I suppose this will pass soon.
21:36	hiredman	snort
21:36	hiredman	eee: are you in #scala?
21:36	eee	no
21:36	eee	are you?
21:36	hiredman	yeah
21:36	hiredman	everyday someone bitching about compiler bugs
21:36	eee	i'm just aying the thought crossed my mind
21:36	eee	that's all
21:36	eee	(ducking)
21:37	hiredman	#scala does not sell scala at all
21:37	eee	do you use that lang, too?
21:37	hiredman	nope
21:37	hiredman	just curious
21:38	eee	i understand they inheritted clojures cool persistent datastructures
21:38	eee	or some of them
21:38	hiredman	they definetly have their share of problems
21:38	unlink1	ataggart: no, clojure absolutely does not have an optional type system.
21:38	unlink1	(static) type system
21:38	hiredman	eee: apparently the guy who wrote the research papers rhickey based them on is in the scala camp
21:39	hiredman	bagwell?
21:39	eee	i see
21:39	eee	i'm writing one
21:39	hiredman	but clojure got them first :P
21:39	eee	you may know
21:40	eee	here's a little math riddle. probably easy for ya'll
21:41	eee	maybe i need a simple if-check that say's if, zero than 4. But what's a more elegant way to make mod 4 return values, 1 2 3 . . .. like normal, but 4 instead of zero
21:42	hiredman	,(- (mod 10 5) 1)
21:42	clojurebot	-1
21:42	hiredman	hmm
21:42	eee	it's funny what I can just believe in my head . . .. vice what I can program sucincly
21:42	hiredman	,(- (mod 10 4) 1)
21:42	clojurebot	1
21:42	hiredman	,(+ (- (mod 10 5) 1) 5)
21:42	clojurebot	4
21:42	eee	,(mod 10 4)
21:42	clojurebot	2
21:43	eee	still needs to be 2
21:43	eee	i thought I had it .. .. but doesn't work for 1 2 or 3
21:43	eee	here's 15 mod 4
21:43	hiredman	,(* (mod 10 8) 2)
21:43	clojurebot	4
21:43	eee	,(int (mod 15/4 8))
21:43	clojurebot	3
21:43	hiredman	,(mod 10 8)
21:43	clojurebot	2
21:44	eee	correct
21:44	eee	(int (mod 10/4 8))
21:44	eee	,(int (mod 10/4 8))
21:44	clojurebot	2
21:44	eee	,(int (mod 12/4 8))
21:44	clojurebot	3
21:44	eee	ay
21:44	eee	should have been 4
21:44	eee	,(int (mod 12/4 16))
21:45	clojurebot	3
21:45	eee	i'm dumb
21:45	eee	i'll just do the if
22:02	eee	yahoooo clojure is the coolest!
22:02	eee	i.e. i found my problem
22:02	eee	:)
22:03	eee	dynamic typing is the way to go again
22:09	unlink1	nah nah
22:22	gcv	is it just me, or is slime-compile-defun flaky with resolving classes imported earlier in (ns ... (:import ...)) forms? slime-compile-and-load-file seems to work fine, which suggests some kind of namespace matching problem. I used slime-repl-set-package, though... and it seemed to work fine until I restarted the repl a minute ago
22:37	defn	user=> user=> java.lang.Exception: Unable to resolve symbol: lazy-cons in this context (core.clj:70)
22:38	defn	I get this when starting slime/clojure repl using swank
22:39	defn	nevermind it fixed itself when i updated the repo
22:40	cmvkk	yeah lazy-cons is no more
23:07	gcv	slime-compile-defun definitely has trouble with namespace resolution. I just saw a function compile fail in a file, then succeed when I saved that file under a different name and tried again
23:55	gcv	definite flakiness with current package tracking in swank-clojure
23:56	gcv	it picks up a nil package sometimes, which the code translates to the user namespace, which wouldn't have necessary imports for a particular file except by accident
23:56	gcv	still trying to figure out where the nil namespace comes from