#clojure – 2012-01-21

2012-01-21

00:00	jeremyheiler	brehaut: nice, yeah
00:06	technomancy	jeremyheiler: it's only like a hundred lines
00:10	technomancy	I think it's been spun off without a proper maintainer anyway, since the chances of further changes to it are so slim
00:13	jeremyheiler	technomancy: yeah, im not saying it's difficult to reproduce or anything. my thought was that this stuff was maintained in clojure proper, then it would be easier for libraries to depend on network io code and still be cross platform. im not sure if that's a real goal, though.
00:16	clj_newb	I have a [1.0 2.0] in clojure land. A Java function requires a float []. How do I convert it to a float [] ?
00:17	jeremyheiler	,(float-array 2 [1.0 2.0])
00:17	clojurebot	#<float[] [F@1e21544>
00:17	jeremyheiler	,(float-array [1.0 2.0])
00:17	clojurebot	#<float[] [F@4c638b>
00:18	clj_newb	jeremyheiler: worked; thanks.
00:18	jeremyheiler	np
00:18	clj_newb	btw, what is to-array used for?
00:18	clj_newb	I dug to-array up, it didn't work; and now, I'm not sure when to-array is ever useful
00:18	technomancy	jeremyheiler: the question of who maintains a library doesn't really affect how easy it is to depend upon
00:19	technomancy	unless you mean the question of how easy it is to discover
00:19	jeremyheiler	clj_new: to-array returns an array of objects instead of primitives
00:20	clj_newb	jeremyheiler: Object[] rather than float[] ?
00:20	jeremyheiler	clj_newb: yeah
00:21	clj_newb	understood; thanks
00:23	jeremyheiler	technomancy: i meant if i were creating a clojure library that i wanted to work on both the jvm and clr, having something as basic as net io bundled with the langauge would make developing and distributing the library easier.
00:25	technomancy	ah, inside clojure.jar itself? yes, that's different.
00:25	technomancy	and very rare
00:28	jeremyheiler	yeah, and it's understandable since interop is so simple.
00:46	muhoo	what does ^: mean, as in (def ^:dynamic group []) ?
00:46	muhoo	i know what :foo is, but what is ^:foo ?
00:47	oakwise	is anyone using lein-cljsbuild?
00:48	jeremyheiler	muhoo: it's the same as saying (meta :foo)
00:48	muhoo	ah, thanks
00:49	muhoo	one more thing i couldn't easily search for in the language reference:
00:49	muhoo	what is #(), as in #(str %1 "-" %2) ?
00:50	jeremyheiler	#() is an anonymous function. it's the same as saying (fn [a b] (str a "-" b))
00:51	muhoo	thanks!
00:52	Raynes	&(meta :foo)
00:52	lazybot	⇒ nil
00:52	Raynes	That is not the same.
00:53	Raynes	It is the same as saying (def ^{:dynamic true} group []), which sets the metadata for the group bar to the map you see there, containing :dynamic true.
00:57	muhoo	so, um, #^{:foo "bar"} is the metadata of an anonymous function?
00:58	muhoo	that one has me stumped
00:59	Raynes	No.
01:02	brehaut	^ is reader metadata, so its on the form, not the result
01:02	brehaut	with-meta is how you apply metadata to runtime data
01:04	clj_newb	is there a way in java/awt to have the awt application treat Tab as a modifier key?
01:14	franks	any "easy" way to find the process id of your clojure program?
01:24	muhoo	franks: ps fax \|grep java ?
01:25	franks	muhoo: understood, but now from within your running clojure program...
01:26	technomancy	franks: I think you can use JMX; it's of obtuse IIRC
01:30	franks	(println "PID:" (:out (clojure.java.shell/sh "bash" "-c" (str "echo -n ${PPID}"))))
01:30	franks	ough...
01:30	muhoo	so what's the difference between #{:foo, :bar} and {:foo, :bar} ?
01:36	duck1123	muhoo: #{:foo :bar} is a set containing 2 items
01:36	duck1123	the other is a hashmap with 1 pair
03:25	muhoo	cool. so then what is #^{:foo "bar"} ? a set that is metadata?
06:58	jondot1	hey guys. without going into editor wars, what editor is most suitable for clojure? (when i did scheme 6 years ago, i used to use emacs)
07:02	raek	I don't think you can determine which one's the "most suitable" without going into editor wars
07:04	raek	I use emacs myself. clojure-mode has very good Leiningen integration.
07:05	jondot1	well, i'd use emacs as well, just there is a new variable here that i don't really know how to handle - java integration
07:06	raek	also, according to this survey emacs+slime is used by 61%: http://cemerick.com/2011/07/11/results-of-the-2011-state-of-clojure-survey/
07:07	raek	jondot1: you have a Clojure project that needs to interoperate with a Java project? or do you just need to use Java libraries?
07:07	jondot1	mostly, using java libraries for NLP.
07:08	jondot1	i guess clojure would be the glue language in this case
07:09	raek	ok. the project and library stuff is handled by Leiningen. An IDE that strictly builds on top of Leiningen for its Clojure project support (like Emacs) does not have to deal with these things at all
07:10	raek	jondot1: have you looked at Leiningen?
07:10	jondot1	not really, i have a long list of tabs which i have to read serially about clojure :)
07:10	jondot1	i'll change the order i guess.
07:10	raek	consuming a Java library is very simple if it exists in the maven central repo
07:11	raek	jondot1: good! the lein tutorial should be very high on the list, IMO. https://github.com/technomancy/leiningen/blob/stable/doc/TUTORIAL.md
07:12	jondot1	great, thanks!
07:12	raek	jondot1: since you're interested in java libs I might as well point you to a search engine: http://jarvana.com/jarvana/
07:13	raek	you usually find Java libs there and Cojure libs on http://clojars.org/
07:13	jondot1	nice, thanks, i guess maven is the de facto build system to use with clojure?
07:14	raek	jondot1: well, both yes and no. it is the de facto dependency system.
07:14	raek	and Leinginen uses those parts of Maven to do its job
07:14	jondot1	yep, ok.
07:14	raek	but the de facto build system would be Leiningen
07:14	raek	though some people use Maven too
07:16	CmdrDats1	jondot1: I use emacs and occasionally eclipse for browsing through java libs when I need to do java integration work
07:17	raek	jondot1: and for emacs integration, this is the only guide you should follow (the official one): https://github.com/technomancy/swank-clojure
07:17	jondot1	thanks
07:33	b6n	Hi guys, whats the most pragmatic way to determine the index of an element in a sequence?
07:48	llasram	b6n: In Clojure that's honestly not a frequently-needed sort of idiom. Why do you need an index?
07:53	b6n	IIasram: I have an vector of keywords and if a keyword occurs at a lower index it has a higher 'value'. and I want to compare those values
07:55	AimHere	Well in that sort of case, perhaps a vector might not be the best data structure
07:57	raek	b6n: you can use the java api of vectors to do this: ##(.indexOf [:a :b :c] :b)
07:57	lazybot	⇒ 1
07:58	raek	but to do a linear search might not be a very good solution
08:01	b6n	yes maybe picking a vector wasn't a good idea. I have a implementation using keep-indexed which works but feels not quite right :-)
08:03	raek	b6n: what do you want to do with these keywords and "values"?
08:05	raek	the coice of data representation is often determined what kind of information you want to get out of the data
08:07	raek	but if this isn't time-critical at all and is just a script you want to throw together the choice of data structures might not be that important
08:07	b6n	I want to implement a traditional card game in which the color of a card determines if it overtrumps another
08:07	b6n	there are only 4 card colors
08:10	raek	where does the vector come into the picture?
08:11	raek	do you have something like [:clubs :diamonds :hearts :spades] ?
08:11	b6n	yes exactly
08:11	raek	ok now I understand!
08:11	b6n	so imagine :clubs overtrumps :diamonds and so forth
08:12	b6n	[:clubs > :diamonds > :hearts > :spades]
08:12	raek	since the number of elements is very low and constant, .indexOf is very bad here
08:12	raek	another way could be to use a map from suit to a number
08:12	raek	and constuct that map from the vector
08:13	llasram	s,very bad,not very bad, ?
08:13	raek	llasram: yes :-)
08:13	llasram	Only because I was honestly confused for a sec :-)
08:13	raek	(def suit-order [:clubs :diamonds :hearts :spades])
08:13	AimHere	Still, if he's reinventing bridge, maybe he's doing it to learn to program, so it might be a good plan to teach him the right thing
08:14	raek	(def suit-value (into {} (map vector suit-order (range))))
08:15	raek	then suit-value is {:clubs 0, :diamonds 1, :hearts 2, :spades 3}
08:15	raek	which also works as a function from suit to value
08:16	raek	(defn overtrumps? [suit-a suit-b] (< (suit-value suit-a) (suit-value suit-b)))
08:16	raek	something like this is how I would have done it
08:16	b6n	yes this looks much better than my (first (keep-indexed (fn ....
08:17	raek	the biggest difference from the .indexOf approach is that the mapping from suit to value is only calculated once
08:19	b6n	I see, thanks a lot!
08:29	llasram	When is it necessary to explicitly reference a function by its var in order to for dynamic changes to that var to take effect?
08:34	llasram	I thought I had a handle on it, but I'm messing with Processing in Clojure, and am able to re-def functions and have the changes be picked up immediately in the running applet
08:34	llasram	It kind of breaks my previous understanding of the compiler :-/
08:35	raek	llasram: whenever 'foo' is used in a function, it will be dereferenced. the cases you need to watch out for are when you keep the value of 'foo' for a long time
08:36	raek	for example if you store it in a data structure or pass it as an argument to a function that runs for a long time
08:38	raek	so, a var #'foo is dereferenced whenever an expression foo is evaluated where foo is a free variable
08:38	llasram	I see. So that function won't be redefined, but anything it references via a var (via the normal implicit namespace resolution) will be
08:38	raek	yes, the function objects themselves are immutable
08:39	AimHere	Hmmm. I'm dead new to this clojure stuff. Why does clojure give me integer overflows a lot sooner in slime/swank/emacs than it does when I run clojure from the command line?
08:40	llasram	I see. Ah! Yes, the situation I'd run into before where I needed to reference vars explicitly, I was generating reify'd objects by passing in function objects directly. Those functions objects were immutable, but their free refs are available for redef
08:40	llasram	Thank you, raek :-)
08:40	raek	AimHere: is this for the same project?
08:40	AimHere	Just for a standard recursive factorial function
08:41	raek	AimHere: do you use the same version of clojure in the two setups?
08:42	AimHere	Heh, well I was using different ones and was so confused I dropped the .jar from one into the other to see what happens
08:42	raek	the default behavior of auto-promotion of integers changed from Clojure 1.2 to 1.3
08:42	AimHere	Strangely, it doesn't seem to be recurring
08:42	AimHere	Ah, yeah, I think that's the thing
08:43	AimHere	I think there may have been a 1.2 version kicking around at some point
08:44	raek	you can either throw in a BigInt value in the computation (which will cause all partial results to be BigInts as well), or you can use the promoting ops: +', -', *', /', etc
08:44	raek	(factorial 100N) ;; BigInt literal as input
08:44	raek	(in Clojure 1.3)
08:45	AimHere	Thanks. A lisp doesn't feel right if it doesn't spit out the factorial of 2000 in a jiffy
09:30	mindbender	guys, how do you approach a problem space with a functional orientation?
09:40	TimMc	raek: Any idea whether using BigInts all the time is slower than checking every time whether promotion is needed?
09:41	TimMc	I feel like it would be, unless it is doing the checking itself... and even then, I think branch prediction (for BigInt-always) wouldn't be as advantageous.
09:43	raek	TimMc: no, I haven't studied how all this is implemented
09:44	raek	but I would guess that a BigInt op is slower than a overflow check followed by a primitive op
09:47	dnolen	mindbender: that's a big question :)
09:47	mindbender	dnolen: I agree
09:47	dnolen	mindbender: but generally I find there's less to think about
09:47	dnolen	mindbender: pick a data structure, write some functions
09:48	mindbender	dnolen: some functions to interact with the data structure
09:49	dnolen	mindbender: yeah
09:50	mindbender	dnolen: can such an approach be used to build a full blown solution?
09:50	dnolen	mindbender: yep
09:52	mindbender	dnolen: I tend to put all code I look at within that frame. But most attimes the authors are not too kind to talk about why they chose such a structure and the need for the fuctions that accompany them
09:52	TimMc	raek: Sorry, I messed up my question... imagine an AutoPromoteInteger that kept a long and a BigInt under the surface.
09:53	dnolen	mindbender: well I find that it's usually obvious from the code
09:54	dnolen	mindbender: Clojure people tends to rely heavily only on maps, sets, vectors, lists. so few surprises
09:55	mindbender	dnolen: what about the point for higher order functions, first class functions and macros, can they be put into any kind of simple perspective?
09:57	dnolen	mindbender: higher fns, first class fn already part of a simple perspective.
09:57	jonasen	If I understand the ClojureScript One sample application correctly, there are two 'Views': form and greeting.
09:57	jonasen	Is there an easy way to add another view based on a new template?
09:58	dnolen	mindbender: macros take a while to understand but they solve a fairly common place problem.
09:58	jonasen	I'm probably missing something obvious
09:58	jonasen	I've created a template myview.html (based on greeting.html) and now I don't really know what to do next.
09:59	mindbender	dnolen: can you try to put it into some perspective with regard to the structure we talked about i.e macros
09:59	dnolen	mindbender: simplest use case for macros is removing boilerplate.
10:00	dnolen	mindbender: but it scales up to writing compiler extensions without needing to actually change the compiler
10:02	dnolen	jonasen: looks like you need to add some code to view.cljs
10:03	dnolen	jonasen: then add a render method for that particular state
10:03	jonasen	dnolen: I've added a new render method (defmethod render :myview ...)
10:03	dnolen	jonasen: but you also need to load it in the render :init method I think
10:05	jonasen	dnolen: you mean in (fx/initialize-views ..)?
10:05	dnolen	jonasen: I think so
10:06	dnolen	jonasen: it looks like you need to change the one.sample.snippets/snippets macro as well
10:07	jonasen	dnolen: Looking at the definition for initialize-views it seems to be hard-wired for form and greeting
10:07	dnolen	jonasen: it definitely is
10:07	jonasen	dnolen: I'll take a look at snippets
10:18	jonasen	dnolen: No luck yet, I'm new to "single page apps" so it's quite challenging.
10:19	jonasen	but they seem to just move greeting out of view, so I guess I shoud do the same with myview
10:20	dnolen	jonasen: the way they manipulate views is not particularly generic or reusable.
10:20	dnolen	jonasen: remember one is just supposed to illustrate how a ClojureScript application might be constructed
10:20	dnolen	probably makes sense to improve on the design
10:21	jonasen	dnolen: Yes, maybe I should just edit the views that are already there for now. Until I understand the whole design better
10:49	uberhanz	Any pointers for someone wanting to learn clojure that knows (too) much java and loves haskell?
10:51	Scriptor	uberhanz: if you already know haskell you'll have a head start thinking of functional solutions for the exercises in http://www.4clojure.com/
10:52	uberhanz	thanks alot. looks like fun!
11:26	phil__	should i use clojure.contrib.trace or clojure.tools.trace?
11:26	llasram	~contrib
11:26	clojurebot	Huh?
11:26	llasram	Damn it
11:26	dnolen	phil__: you should not use anything from clojure.contrib
11:27	phil__	dnolen: is it outdated?
11:27	dnolen	phil__: yes
11:29	phil__	ok, any pointers on how i should set up https://github.com/clojure/tools.trace with lein? or should i just pull the source and stuff it into my project?
11:29	phil__	but this seems messy :/
11:30	phil__	or is there just a better way to trace function execution?
11:31	llasram	phil__: using `lein search' will search your configured repos for artifacts
11:32	llasram	searching for tools.trace shows be that it's [org.clojure/tools.trace "0.7.1"]
11:32	phil__	ooh
11:33	phil__	i see, this is awesome
11:33	phil__	thanks a lot :)
11:34	llasram	Np :-) It does seem like something that you'd just want to slurp in to any random project, without needing to explicitly make it a dependency. Maybe a lein plugin?
11:39	phil__	llasram: why, is it bad to have extra dependencies on packages like tools.trace?
11:41	llasram	Just stuff you don't need. You can make it a dev-dependency, and then it isn't included in deployments etc, but it's still unnecessary if the committed code doesn't actually trace anything
11:43	llasram	There doesn't seem to be a better way of referencing it ATM though, so a dev-dependency is probably the way to go
11:53	devn	&(= [1 4 9 16] '(1.0 4.0 9.0 16.0))
11:53	lazybot	⇒ false
11:55	llasram	Or more simply: ##(= 1.0 1)
11:55	lazybot	⇒ false
11:56	dnolen	##(== 1.0 1)
11:56	lazybot	⇒ true
12:41	tufflax	Hm, the docs say that = is supposed to compare numbers in a type-independent manner. ##(= 1.0 1) does not give that impression.
12:41	lazybot	⇒ false
12:41	dnolen	tufflax: docs are out of date, == does that now
12:42	tufflax	ok :p
12:43	AimHere	Well (/ 1/6 (/ 1.0 6.0)) is still false, note!
12:44	AimHere	##(== 1/6 (/ 1.0 6.0))
12:44	lazybot	⇒ false
12:46	Luyt	Interesting, there is a 'first' and 'second', but not a 'third' or 'fourth'. I guess you have to draw the line somewhere ;-)
12:46	tufflax	AimHere that's expected, at least by me :p
12:46	TimMc	Not by me.
12:46	tufflax	TimMc why not
12:46	AimHere	##(== 1/3 (/ 1.0 3.0))
12:46	lazybot	⇒ true
12:47	tufflax	hm
12:47	tufflax	:p
12:47	AimHere	I guess it doesn't like rounding up
12:48	Luyt	Floating point is only an approximation. And ##(type 1/3) is not the same as ##(type (/ 1.0 3.0))
12:48	lazybot	(type 1/3) ⇒ clojure.lang.Ratio
12:48	lazybot	(type (/ 1.0 3.0)) ⇒ java.lang.Double
12:49	TimMc	AimHere: Oh, what the...
12:49	TimMc	Luyt: So? ##(== 1 1.0)
12:49	lazybot	⇒ true
12:50	Luyt	Apparently 1.0 is a value that can be represented exactly in floating point.
12:50	AimHere	But neither 1/3 nor 1/6 can be represented exactly
12:50	Luyt	But 1.0/3.0 is 0.3333333333333333 and where does the 333333333 end?
12:50	AimHere	Well unless you're taking a base with 3 (and/or 2) as the factor
12:51	tufflax	So why the difference in 1/6 and 1/3 results?
12:51	tufflax	in the
12:51	tufflax	as above :YP
12:51	AimHere	Maybe the internal representation of clojure doubles is in base 81
12:52	Luyt	I think it's caused by the floating point implementation of the machine on which the code executes.
12:52	Luyt	The same problem crops up in any other program language which uses floating point.
12:53	tufflax	Most (?) other languages don't have ratios though
12:53	AimHere	Well what other languages manage to compare double floats to inbuilt integer rational types?
12:53	AimHere	I mean, it's nice that it even works some of the time
12:54	tufflax	Hehe, I don't know about that :)
12:56	TimMc	I'd prefer consistency.
12:56	TimMc	I don't see any tickets on this; filing one.
12:56	dnolen	TimMc: ticket about what?
12:57	AimHere	The == operator isn't overly consistent when comparing number types that were never meant to be compared in the first place !
12:58	AimHere	It's a lucky day in most languages if your floating point type manages to equate 1/3 and 2/6
12:58	TimMc	dnolen: (== ratios floats) being inconsistent
12:58	AimHere	Never mind casting to rationals
12:59	TimMc	dnolen: (== 1/x (/ x.0)) false for x = 6, 7, 13, true for x = 2, 3, 4, 5, 8, 9, 10, 11, 12, 14
13:00	TimMc	$javadoc clojure.lang.Ratio
13:00	lazybot	http://download.oracle.com/javase/6/docs/api/clojure/lang/Ratio.html
13:00	TimMc	bah
13:07	AimHere	##(#(if (= %2 0) 1 (' %2 (%1 %1 (- %2 1)))) #(if (= %2 0) 1 (' %2 (%1 %1 (- %2 1)))) 200)
13:07	lazybot	⇒ 788657867364790503552363213932185062295135977687173263294742533244359449963403342920304284011984623904177212138919638830257642790242637105061926624952829931113462857270763317237396988943922445621451664240254033291864131227428294853277524242407573903240321257405579568... https://refheap.com/paste/369
13:07	AimHere	Oh sweet
13:09	TimMc	AimHere: Is that Ackerman?
13:09	AimHere	Nah, just a Y combinator on 200!
13:10	AimHere	(#(%1 %1) #(%1 %1))
13:10	AimHere	##(#(%1 %1) #(%1 %1))
13:10	lazybot	java.lang.StackOverflowError
13:10	AimHere	Just have to stresstest the bot ;)
13:10	TimMc	,(#(% %) #(% %))
13:10	clojurebot	#<RuntimeException java.lang.RuntimeException: java.lang.StackOverflowError>
13:16	TimMc	dnolen: Never mind, not filing a ticket.
13:16	TimMc	AimHere, tufflax: ##(double 1/6)
13:16	lazybot	⇒ 0.1666666666666667
13:16	TimMc	##(/ 6.0)
13:16	lazybot	⇒ 0.16666666666666666
13:17	TimMc	Inverting the float munges it enough to make it inequal with the ratio.
13:17	tufflax	why not? :p After thinking some more about it I think I have changed my view. I would expect ##(= 1/x (/ x.0)) to be true for all x, and that would, I think, be more useful. :P
13:18	tufflax	Even if not totally true
13:18	TimMc	##(3.0 (/ (/ 3.0)))
13:18	lazybot	java.lang.ClassCastException: java.lang.Double cannot be cast to clojure.lang.IFn
13:18	TimMc	##(== 3.0 (/ (/ 3.0)))
13:18	lazybot	⇒ true
13:19	TimMc	##(== 6.0 (/ (/ 6.0)))
13:19	lazybot	⇒ true
13:19	TimMc	huh
13:19	tufflax	I mean, if you think of (/ x.0) as the double that results
13:21	TimMc	&(== 1/6 0.1666666666666667)
13:21	lazybot	⇒ true
13:22	TimMc	I'm just chalking this up to floating-point jitter and leaving it be.
13:32	chewbranca	anyone have any recommendations for error handling with clutch and couchdb? right now the default is to throw an io exception with a string containing the http error code, but I would like to be able to react appropriately to different errors, specifically in this case, 409 conflict errors
13:44	raek	chewbranca: I would recommend slingshot in general for clojure programs. https://github.com/scgilardi/slingshot
13:46	chewbranca	raek: nice, I've been meaning to look more into slingshot, time to dive in
13:54	zilti	How do I add a leiningen dependency which has a classifier? Where do I write the classifier?
13:54	chewbranca	raek: cool, switched over to slingshot, looks like a nice approach to error handling. Won't help with dealing with string status messages, but overall using slingshot looks nice
13:57	jkkramer	zilti: [group/artifact "version" :classifier "foobar"]
13:57	jkkramer	zilti: per https://github.com/technomancy/leiningen/blob/master/sample.project.clj
13:57	zilti	oh, thanks!
14:06	Luyt	How do I do 'argument splatting' in Clojure? Like that (println 1 (splat [2 3]) 4) means the same as (println 1 2 3 4) ?
14:08	brehaut	,(apply println 1 (conj [2 3] 4))
14:08	clojurebot	1 2 3 4
14:10	Luyt	is there no function that will unpack a collection so that the items can be passed individually instead of inside a collection?
14:10	brehaut	Luyt: apply ^
14:10	Luyt	(conj [2 3]) doesn't expand into two arguments, 2 and 3.
14:10	Luyt	Ow moment, I'll look up apply
14:11	brehaut	Luyt: build one seq, and use apply to pass it to the fn
14:11	brehaut	no need for jiggery pokery voodoo
14:12	Luyt	I was trying (disj a b) where both a and b are sets, but that didn't work as expected, but (disj a b1 b2 b3 ... bn) where bi are elements of set b
14:12	Luyt	did work.
14:12	Raynes	&(apply + [1 2 3])
14:12	lazybot	⇒ 6
14:13	raek	Luyt: have you seen clojure.set/difference?
14:13	Luyt	raek: Yes, the tutorial mentiones that, but I was wondering whether there is a splat operator/function in Clojure.
14:14	raek	Luyt: you could use apply there: (apply disj a b) = (apply disj (cons a b)) = (disj a b1 b2 b3 .. bn)
14:14	Luyt	Hmmm, food for thought ;-) Thanks for all your tips, I'll figure something out ;-)
14:14	raek	the variadic version of apply takes all elements between the first and the last and concats them to the last
14:15	raek	and then applies the function with those as the arguments
14:24	uberhanz	best page when trying to learn clojure : http://clojure.org/cheatsheet
14:28	Luyt	I'm currently working through http://java.ociweb.com/mark/clojure/article.html
14:30	Luyt	I sometimes have to turn my brains inside-out ;-)
14:31	uberhanz	I usually like static (and strong) typed languages, but clojure is fun!
14:31	Luyt	I was surprised that you can use a set or hashmap as a function to test for membership. Image the power that'd be unleashed if you could pass functions around!
14:32	uberhanz	haha. new to functional programming?
14:32	uberhanz	or am I missing the irony?
14:33	Luyt	I was a bit charging there ;-) I come from a C/C++/Python background. In Python, functions are first class citizens, like in Clojure. But I already passed around function pointers in C, so...
14:34	brehaut	function pointers dont have any lexical context though
14:35	AimHere	I think his point is that he did the closest that he could to that sort of stuff in C
14:35	Luyt	Good point. But when they don't use globals or have any side effects, that wouldn't be a limitation, would it?
14:36	brehaut	its a huge limitation!
14:36	Luyt	No closures, indeed
14:36	Luyt	No fancy python decorator stuff
14:37	brehaut	no super expressive combinators
14:37	uberhanz	think about us poor ppl that have to use Java at work and have to keep their functional programming for home projects. weep for us!
14:37	Luyt	Are you one of them?
14:37	uberhanz	yeah
14:38	Luyt	The problem is that corporations want to standardize on Java instead of just the JVM.
14:38	Luyt	In the latter case, you could use anything that produces java bytecode.
14:38	uberhanz	yeah
14:39	Luyt	Corporations see "Java Programmers" as a commodity
14:39	semperos	using clojurescript, working with a JavaScript library function that expects a JS object as a parameter
14:39	uberhanz	Its easier to start to write tools in other JVM based languages
14:39	semperos	what's the "right" way to go about this common task?
14:39	TimMc	Luyt: s/"Java Programmers"/people/
14:40	semperos	TimMc: s/people/walking-money-bags-with-no-life-outside-work/
14:40	uberhanz	haha
14:43	Luyt	I already suspected it'd be better to learn some Clojure instead of doing Java from 9 to 5 and couchpotato for the TV with a few beers at evening.
14:43	AimHere	Couchpotato with the TV and beer from 9 to 5, then learn some Clojure in the evening
14:44	AimHere	Clearly the lifestyle of champions
14:44	uberhanz	Well, I actually not only do Java, but do email also! on the otherhand, I dont do couchpotato. not my thing
14:44	wiseen	is there a clojure parser library - something that would help parse macro grammars ?
14:45	brehaut	wiseen: you mean something that deals with sexps?
14:45	Luyt	I usually do Python during the day, but one has to keep on learning more powerful blob languages, no? http://paulgraham.com/pypar.html http://www.paulgraham.com/avg.html
14:46	zilti	My life's Clojure and Army, right now
14:46	wiseen	brehaut, well something that will help me parse macro body, eg. I have a macro that can contain (message ...) (catch ...)* (close ...) (finally ...)
14:46	wiseen	catch can appear multiple times, others are optional
14:47	wiseen	parsing that from lists is tedious
14:47	wiseen	ideally I could have some declarative way to specify the grammar ?
14:47	Raynes	I do exactly that in lazybot's plugin dsl. I use keywords for the names though, which make it clearer that you're not calling things by are supplying directives instead.
14:48	Raynes	s/by/but/
14:49	wiseen	Raynes, link ?
14:50	wiseen	Raynes, do you do it manually or did you write a DSL for it ?
14:50	Raynes	Manually.
14:51	Raynes	https://github.com/flatland/lazybot/blob/develop/src/lazybot/registry.clj#L121 Now, this is all probably pretty insane (this has been only marginally modified since originally created and was some of my first Clojure code), but the actual parsing stuff (I think) is fairly recently rewritten and isn't all that wild.
14:53	phil___	can somebody please tell me if this is considered bad form: https://gist.github.com/1653743, as opposed for example to this: https://gist.github.com/1653746
14:54	phil___	i prefer to use as few ifs and conds and cases as possible and let the language handle the test cases vie apply
14:54	wiseen	yeah the gorey internals of writing a parser manually, I'm suprised someone hasn't written a declarative parsing library - I'm guessing it would simplify writing macros a lot, and functional PLs usually have cool parsing libs (eg. FParsec F#)
14:55	Raynes	The first example is much clearer, but I think that might because of the insane indentation in the second one.
14:55	Raynes	Er, reverse that.
14:55	Raynes	I clicked the links in the wrong order. :P
14:55	Raynes	Yeah, closer inspection tells me that the second one is definitely the clearer of the two.
14:56	phil___	:D so the second is considered good practice? isnt it much clearer to have all edge cases layed out in front of you as in the first example?
14:58	phil___	it certainly looks much cleaner, but the explicit tests kind of conceal the intent i think
15:02	phil___	wiseen: just yesterday i read about a parsec like monadic parser in clojure: http://intensivesystems.s3-website-us-east-1.amazonaws.com/tutorials/monads_101.html?from=@
15:02	phil___	gotta scroll down a bit
15:03	raek	there's also fn-parse: https://github.com/joshua-choi/fnparse
15:03	tavis`	Roman Gonzalez is about to release his port of attoparsec to clj
15:04	tavis`	it should be up on https://github.com/roman/ sometime soon
15:04	tavis`	zettaparse
15:08	muhoo	hmm, kind of annoying (.split "foo bar baz" " ") doesn't give me a list of strings, it gives me a #<String[] [Ljava.lang.String;@10241ae>
15:08	muhoo	instead of ("foo" "bar" "baz") which i kind of expected
15:08	Raynes	muhoo: That's a Java array of strings. It is seqable, so you can treat it essentially as what you want.
15:09	Raynes	&(seq (.split "foo bar baz" " "))
15:09	lazybot	⇒ ("foo" "bar" "baz")
15:09	Raynes	&(map #(str % " hi") (.split "foo bar baz" " "))
15:09	lazybot	⇒ ("foo hi" "bar hi" "baz hi")
15:10	raek	,(clojure.string/split "foo bar baz" #" ")
15:10	tavis`	&(clojure.string/split "foo bar baz" #" ")
15:10	lazybot	⇒ ["foo" "bar" "baz"]
15:10	clojurebot	#<ExecutionException java.util.concurrent.ExecutionException: java.lang.RuntimeException: java.lang.ClassNotFoundException: clojure.string>
15:10	Raynes	I wish people wouldn't use that as an opportunity to avoid explaining the difference between Java arrays and Clojure seqs and how you can use them in Clojure.
15:11	muhoo	uberhanz: could be worse, you could be force to use PHP at work
15:11	muhoo	Raynes: thanks!
15:11	Raynes	raek: Not that you are guilty of such acts.
15:12	Raynes	I occasionally see people with muhoo's query get shown an example of clojure.string/split and told "See! This does it!" and a fantastic opportunity for them to learn about Java collections is missed.
15:12	Raynes	It's tragic, really.
15:13	muhoo	i am very thrilled to be able to use seq to turn some nonsense #<BS[] [Ljava.lang.BS into a nice lispy list
15:14	raek	exactly :-)
15:14	Raynes	muhoo: Keep in mind that, unless you're trying to print the array, you don't usually have to call seq on it.
15:14	muhoo	it's just repl sugar for me, but it makes me happy :-)
15:15	Raynes	muhoo: Most of Clojure's functions do that implicitly. Like my map example.
15:15	Raynes	:)
15:18	muhoo	housekeeping question, the bot is & or ##?
15:18	muhoo	&"foo"
15:18	lazybot	⇒ "foo"
15:18	muhoo	##"foo"
15:18	muhoo	hmm
15:18	brehaut	both of those are lazybot
15:18	raek	blahblah ##"foo"
15:18	brehaut	clojurebot is prefixed with a comma
15:18	muhoo	,"foo"
15:18	clojurebot	"foo"
15:18	raek	blahblah ##(identity "foo")
15:18	lazybot	⇒ "foo"
15:18	muhoo	wait, there are two bots?
15:19	brehaut	,(inc 1)
15:19	clojurebot	2
15:19	brehaut	&(inc 1)
15:19	lazybot	⇒ 2
15:19	brehaut	theres also hsbot
15:19	muhoo	!
15:19	tavis`	hsbot?
15:19	cold_gopher	Hi all. I have just installed SWANK and I am launching it from emacs. When I'm in the REPL and perform a doc such as (doc map), I get an exception. Anyone come across this before?
15:20	brehaut	haskell bot i think
15:20	tavis`	what's it doing here? ;)
15:20	brehaut	cold_gopher: the swank repl is different to the normal repl
15:20	brehaut	cold_gopher: you need to use clojure.repl if you want the normal tools. however, slime provides alternatives through emacs that you can use
15:21	cold_gopher	ok thanks
15:21	brehaut	cold_gopher: C-c C-d C-d for docs
15:21	brehaut	(on a var)
15:21	brehaut	works in the repl and editor
15:21	cold_gopher	nice
15:24	wjlroe	brehaut: oh man, I didn't know that and was looking for that yesterday - thanks!
15:24	wjlroe	Could not find it in the C-h b listing for some reason
15:26	brehaut	i cant find the one for looking up the source of a var
15:27	tavis`	eldoc is also quit handy with slime: https://gist.github.com/1551858
15:29	wjlroe	Also useful - break points in swank - http://hugoduncan.org/post/2010/swank_clojure_gets_a_break_with_the_local_environment.xhtml
15:29	Raynes	brehaut: lazybot evals Haskell too.
15:29	Raynes	&he 3 + 3
15:29	lazybot	java.lang.RuntimeException: Unable to resolve symbol: he in this context
15:29	Raynes	Duh.
15:29	Raynes	$he 3 + 3
15:29	lazybot	⇒ 6
15:29	brehaut	Raynes: thats good, becuase i cant remember the hsbot commands :P
15:30	Licenser	heh
15:30	tavis`	$he take 6 (cycle [1,2,3])
15:30	Raynes	muhoo: ##(println "is for evaluation of code embedded inside of a message (when you can't begin the message with &)")
15:30	lazybot	⇒ is for evaluation of code embedded inside of a message (when you can't begin the message with &) nil
15:30	lazybot	⇒ [1,2,3,1,2,3]
15:30	raek	this is interesting: http://skife.org/java/unix/2011/06/20/really_executable_jars.html
15:32	brehaut	$he let fib = 1 : 1 : zipWith (+) fib (tail fib) in take 10 fib
15:32	lazybot	⇒ [1,1,2,3,5,8,13,21,34,55]
15:32	Raynes	$ht "you can even check types!"
15:32	lazybot	⇒ Type: "you can even check types!" :: [Char]
15:32	Licenser	wow that is the most terse definition of fib I've ever seen o.O
15:32	Raynes	I wonder why i prefix that with 'type'.
15:32	Raynes	Licenser: Welcome to Haskell. May I take your coat?
15:33	Licenser	Raynes that is astonishing
15:33	brehaut	Licenser: its the canonical 'i have lazy values!' version
15:33	Licenser	so I have a real hard time to understand it
15:33	Licenser	so you should be able to do about the same in clojure shouldn't you?
15:33	brehaut	you can…
15:34	brehaut	you probably need to use a delay or a promise to tie the knot though
15:34	brehaut	or use a def, which is a potential space leak ;)
15:34	Licenser	so let … in … is like (let [..] …) right?
15:34	brehaut	correct
15:35	Licenser	so the definition of fib is 1 for 1 or zipWith (+) fib tail fib for all others?
15:35	brehaut	(concat [1 1] (map + …))
15:35	tavis`	what would lazybot do with an infinite list?
15:35	Raynes	: is cons
15:35	Raynes	tavis`: Try it out.
15:35	Licenser	ah
15:35	brehaut	or (list* 1 1 (map + … i guesss
15:35	Licenser	$(range)
15:36	tavis`	$ht [1..]
15:36	lazybot	⇒ Type: [1..] :: (Num t, Enum t) => [t]
15:36	Licenser	,(range)
15:36	clojurebot	(0 1 2 3 4 ...)
15:36	Licenser	clojurebot is a sneaky bastard &&
15:36	Licenser	&(range)
15:36	lazybot	java.lang.OutOfMemoryError: Java heap space
15:36	Licenser	lazybot will post the infinit list to github? :P
15:37	Raynes	Haha, well, normally he would stop after so many seconds and print an error message saying he timed out.
15:37	Licenser	but this is too fast ^^
15:37	Raynes	In this case, he has reached the memory limitations I've set for him.
15:37	Raynes	No, that would usually be fine.
15:37	Licenser	Raynes can't you test for lazy lists and make it print it like clojurebot ?
15:37	Raynes	Licenser: I'm pretty sure he just sets print-length, and yes, I could do that too.
15:38	Licenser	print-length is cheating
15:38	Raynes	I agree.
15:38	Licenser	you could build your own pr-str to be smrt
15:38	Raynes	&(range)
15:38	lazybot	java.lang.OutOfMemoryError: Java heap space
15:38	Raynes	Huh. I guess I've got the timeout set too high.
15:38	Licenser	or create a clojure version that has infinit memory space
15:38	Licenser	and internet bandwith
15:39	Licenser	well range will fill your memory quite quick I guess
15:39	Raynes	Yeah, but that didn't always blow the heap. Don't think we tried that since we changed the memory settings.
15:39	Raynes	Anyways, ##(+ 3 3)
15:39	lazybot	⇒ 6
15:39	Raynes	He's still okay.
15:40	tavis`	and he can't be tricked to flood the channel?
15:40	Raynes	With code? No.
15:40	muhoo	&(range 20)
15:40	lazybot	⇒ (0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19)
15:40	Raynes	&(range 500)
15:40	lazybot	⇒ (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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 ... https://refheap.com/paste/371
15:40	tavis`	nice
15:40	muhoo	ooh, nice
15:40	brehaut	&(let [fib (promise)] (deliver fib (concat [1 1] (map + @fib (rest @fib))))) (take 10 @fib))
15:40	Raynes	You can, of course, spam him with eval requests which would make him flood, but you'd be flooding the channel too and it wouldn't be much of a trick.
15:41	lazybot	Execution Timed Out!
15:41	brehaut	rats
15:41	Raynes	^ is what it is supposed to do rather than blow the heap.
15:41	brehaut	i have clearly forgotten how to use the promise trick
15:41	Raynes	brehaut: Promise you've forgotten.
15:41	Licenser	&(println "bla")
15:41	lazybot	⇒ bla nil
15:42	Licenser	&(println "bla\nblubb")
15:42	lazybot	⇒ bla blubb nil
15:42	Licenser	aww he strips line breaks :P
15:42	Licenser	&(println "bla\nrblubb")
15:42	lazybot	⇒ bla rblubb nil
15:42	Licenser	&(println "bla\nrblubb")
15:42	lazybot	⇒ bla rblubb nil
15:42	Licenser	&(println "bla\rblubb")
15:42	lazybot	⇒ blablubb nil
15:42	Licenser	:(
15:42	raek	%(println "foo\u0000bar")
15:42	raek	&(println "foo\u0000bar")
15:42	lazybot	⇒ foo
15:42	Raynes	,(println "foo\nbar\nbaz")
15:42	clojurebot	foo
15:43	clojurebot	bar
15:43	clojurebot	baz
15:43	muhoo	what is this, bot torture?
15:43	Raynes	muhoo: This is spam that you've instigated.
15:43	tavis`	,(print "&(print 1234)")
15:43	clojurebot	&(print 1234)
15:43	lazybot	⇒ 1234nil
15:43	Raynes	Now feel guilty about it.
15:43	muhoo	Raynes: i do indeed.
15:43	brehaut	,(print "(dec Raynes)")
15:44	clojurebot	(dec Raynes)
15:44	lazybot	⇒ 12
15:44	Raynes	clojurebot is not my biggest fan.
15:44	tavis`	so lazybot is a bit smarter than clojurebot about preventing the other bots from listening
15:44	Licenser	,(println "&(println \"&(+ 1 1)\")")
15:44	clojurebot	&(println "&(+ 1 1)")
15:44	lazybot	⇒ &(+ 1 1) nil
15:45	Licenser	,(println "&(println \"##(+ 1 1)\")")
15:45	Licenser	now that didn't work
15:45	lazybot	⇒ 2
15:45	clojurebot	&(println "##(+ 1 1)")
15:45	lazybot	⇒ 2
15:45	Licenser	oh it did
15:45	muhoo	a bot as a macro expander
15:45	Licenser	now I wonder if you make them ping pong
15:45	Licenser	so to say a quine :P
15:45	Raynes	No, you can't.
15:45	Licenser	Raynes are you sure?
15:45	Somelauw	&(def fib (lazy-cat [0 1] (map + fib (rest fib))))
15:45	Bronsa	,(let [fib (promise)] (deliver fib (lazy-cat [1 1] (map + @fib (rest @fib)))) (take 10 @fib))
15:45	clojurebot	(1 1 2 3 5 ...)
15:45	lazybot	java.lang.SecurityException: You tripped the alarm! def is bad!
15:45	tavis`	,(print ",(print 1234)")
15:45	clojurebot	,(print 1234)
15:45	Raynes	Yes.
15:45	Licenser	oh I guess lazybot ignores himsef?
15:46	Raynes	lazybot has to ignore himself -- the IRC server doesn't send you your own messages.
15:46	Raynes	But that isn't the problem.
15:46	Licenser	ut?
15:46	Licenser	but?
15:46	Raynes	lazybot can't invoke clojurebot because he prefixes eval results with an arrow.
15:46	Licenser	but lazybot can invoke lazybot ##(str "##(+ 1 1))
15:46	lazybot	⇒ 2
15:47	Licenser	but lazybot can invoke lazybot ##(str "##(+ 1 1)")
15:47	lazybot	⇒ 2
15:47	Licenser	hmm seems it can't
15:47	Licenser	but lazybot can invoke lazybot ##(str "#" "#(+ 1 1)")
15:47	lazybot	⇒ "##(+ 1 1)"
15:47	Raynes	It can't because that is a limitation of the IRC protocol.
15:47	Licenser	ah yeas he does not see himself
15:47	Licenser	hah
15:47	Licenser	sneaky we have to con clojurebot into having something like ## ^^
15:47	Licenser	then we can make them do a bot war
15:47	ordnungswidrig	Licenser: don't feed the bots! :-)
15:48	Licenser	^^
15:48	Licenser	but it would be fun
15:48	muhoo	"fun" in the sense that setting old xmas trees on fire is fun
15:48	Licenser	muhoo yes
15:48	muhoo	not very productive though.
15:49	Licenser	productive is for once overrated and then it is relative
15:49	Licenser	breaking clojurebot in the beginning was quite productive ^^
15:49	Somelauw	Seems like promise makes a variable and deliver sets that variable
15:49	Licenser	if it hand't been done there would never have been the whole jailing thing
15:49	Raynes	clojurebot is still pretty easy to break.
15:49	Raynes	The jail at least.
15:50	Raynes	But I'm not sure he is going for safe with that one.
15:50	Licenser	Raynes just because it is not using clojail or clj-sandbox :P
15:50	Somelauw	So, that's how you make non-global recursive lists. I asked before but nobody knew.
15:50	Raynes	Well, lazybot's clojure plugin has essentially been a battleground for testing clojail.
15:51	tavis`	have you played with running clojurescript in jailed env?
15:51	Raynes	tavis`: Sure, we call that 'your browser'.
15:51	Raynes	;)
15:52	tavis`	bot style I mean
15:52	Raynes	No. I'd be tempted to refheap the compiled javascript and tell you to run it your damn self. :p
15:52	tavis`	hehe
15:53	tavis`	'your browser' isn't really a safe jail though
15:53	Raynes	It's safe for me though.
15:53	tavis`	very true
15:55	ordnungswidrig	anyboy else having problems with slime-edit-definition? When invoking on a function defined in a protocol I get NPE in File/init called by slime-find-file
15:56	tavis`	yeah, it doesn't work yet
15:56	tavis`	I'm working on that
15:56	ordnungswidrig	nice.
15:57	tavis`	if you move your point to the protocol itself you'll get pretty close to it
15:57	ordnungswidrig	I wonder if we can have jump it to the function implementation, not the protocol definition.
15:57	tavis`	metapoint works on protos but not proto funcs
15:58	technomancy_	nice to have someone helping with slime who actually cares about protocols
15:58	tavis`	I saw you prefer multimeths
15:59	Somelauw	Would it be possible to write a letrec macro in clojure using the promise/deliver primitives?
16:00	technomancy_	tavis`: a better way to say it would be that I don't work on any problems myself for which the speed of protocols is valuable enough to outweigh the inconveniences =)
16:05	Somelauw	How to STOP clojure in emacs
16:05	Somelauw	before my stack explodes?
16:06	Somelauw	I need to stop it quickly?
16:06	tavis`	slime-interrupt might
16:06	ordnungswidrig	Somelauw: kill the java process.
16:06	Somelauw	alt+x not working
16:07	Somelauw	Something like ctrl-c?
16:07	tavis`	then kill -9 java
16:07	Somelauw	Okay, I killed everything.
16:09	Somelauw	Thanks. Can't slime be stopped by ctrl-c or something?
16:09	Raynes	It can be stopped by force killing your Emacs.
16:09	Raynes	;)
16:09	technomancy_	tavis`: I think it would be perfectly acceptable to skip byte compilation of the .el payload files are remote
16:09	Somelauw	That's what I did. I force killed emacs
16:10	technomancy_	as long as they get loaded; the byte compilation is nice for everyday speed, but it's an optimization. using jack-in remotely seems like not an everyday task.
16:10	muhoo	is there a way to deref something where all i have is the text printout of the reference, i.e. #<Server Server@160bf50> ?
16:10	technomancy_	tavis`: of course if you intend to use it as part of normal development then maybe not; depends on your plans for it
16:10	muhoo	is there any way to turn that text into the object that it sort-of-describes?
16:13	ordnungswidrig	tavis`: is hte npe fixed in swank clojure head?
16:14	ordnungswidrig	tavis`: do worry. It is. just found it.
16:18	raek	muhoo: not in general. you can use the 1, 2, and *3 variables though if you evaled it recently
16:36	Bronsa	https://github.com/clojure/clojure/blob/master/src/clj/clojure/repl.clj#L124
16:36	Bronsa	shouldnt it be a set?
16:37	arkh	how do I tell what version of clojure I'm running at a repl? I'm not sure if my emacs setup is using my project clojure of 1.3 or the 1.2 it installed with "lein plugin install swank-clojure 1.3.4"
16:37	Bronsa	oh, i see, i misunderstood
16:37	Bronsa	,(clojure-version)
16:37	clojurebot	"1.3.0"
16:38	arkh	Bronsa: thank you
16:39	dnolen	,clojure-version
16:39	clojurebot	{:major 1, :minor 3, :incremental 0, :qualifier nil}
16:43	tavis`	technomancy_: we could also just copy the remote files to local and then byte-compile locally
16:44	tavis`	any thoughts on a security warning first?
16:52	zilti	Is it a bug or a feature that things like (. (new Foo) (bar x)) aren't possible?
16:52	nodename	Hi, I have an app built as a Counterclockwise project. I've been asked "Where's the project.clj?" How can I make my project runnable outside Eclipse?
16:53	ibdknox	zilti: huh? that is possible
16:54	ibdknox	,(. (Integer. "2") (toString))
16:54	raek	&(. (new Object) (toString))
16:54	lazybot	⇒ "java.lang.Object@1139c27"
16:54	clojurebot	"2"
16:54	zilti	ibdknox: I have the problem that it isn't possible - but it's possible if I first store it in a variable and then do a (. foo (bar x))
16:56	ibdknox	that seems unlikely
16:56	ibdknox	what's the exception?
16:56	ibdknox	and the exact code
16:59	zilti	It's a simple NullPointerException
16:59	ordnungswidrig	nodename: you can package it as a jar
16:59	ibdknox	ordnungswidrig: I think he's asking how to take a CCW project and make a project.clj from it so it'll work in lein
17:00	nodename	ordnungswidrig: yes, what ibdknox said
17:00	ordnungswidrig	I see.
17:00	zilti	I've already changed the code... Actually I used the .. macro. I had to change the whole thing because the .. macro seems to have problems with mutable java classes anyway. But I wanted to ask if that's a known problem
17:04	dnolen	zilti: there's no problem. many mutating methods return void instead of a value.
17:11	ishkabibl	wow this is a big irc channel
17:12	ishkabibl	is Raynes here?
17:12	Raynes	Yep.
17:13	ishkabibl	I saw your "try coljure" interactive web appliction(at first I didn't realize you made it)
17:17	Raynes	ishkabibl: Yup, that'd be me.
17:18	ishkabibl	hows your book comming?
17:18	Raynes	At least partially. I didn't do much of the design portion.
17:18	Raynes	Slowly but surely.
17:18	ishkabibl	well it has your name on it(try clojure)
17:20	Raynes	ishkabibl: I wrote the majority (all?) of the backend and the initial design. Other people came along and streamlined it, because I wasn't fantastic at web design.
17:20	ishkabibl	I see
17:21	ishkabibl	are maps persistent?
17:22	Raynes	Yep.
17:22	ordnungswidrig	ishkabibl: everthing it but refs, atom and agent.
17:22	Raynes	&(type {:foo 0})
17:22	lazybot	⇒ clojure.lang.PersistentArrayMap
17:22	ordnungswidrig	ishkabibl: and java classes
17:22	Raynes	http://clojure.org/data_structures
17:24	phil___	is it possible to simplify the following code in some way? https://gist.github.com/1654274
17:26	ishkabibl	maps must use vectors to store their buckets
17:27	dnolen	ishkabibl: object arrays more likely
17:28	ishkabibl	are object arrays just java arrays?
17:28	ordnungswidrig	phil___: (if (symbol? a) [a (cons a b)] [a b]) can be simplified to [a (if (symbol? a) (const a b) b))]
17:29	ordnungswidrig	phil___: i.e. you can inline the if into the vector creation
17:30	ishkabibl	wait, there are 2; sorted and hashed. the sorted one makes sense(it's just a tree like haskell's map) but the hashed one makes less sense
17:30	tauntaun	Does Clojure have a standard library for the basic combinators (S, K, etc.)?
17:30	phil___	ordnungswidrig: oh yes, thx :)
17:31	ishkabibl	tauntaun: if not you can use fn for lambda clalcules to implment it
17:33	tauntaun	ishkabibl: I'm quite aware of that :) I prefer not to reimplement even small simple things.
17:33	ishkabibl	ya, me too :)
17:39	tavis`	tauntaun: http://www.thelastcitadel.com/blag/combinator
17:40	technomancy_	tavis`: that would work too; depends on how much effort you want to put into it. either approach is valid.
17:43	chewbranca	what do you guys use for web request loggin? more specifically with noir?
17:47	tauntaun	Wow, I must be communicating really poorly :)
17:47	tavis`	same, that was my way of saying I can't find one
17:50	tavis`	tauntaun: there's various parser combinator libs, but I can't see anything packaged up for S K I,etc.
17:52	ordnungswidrig	chewbranca: ngnix or apache in front of it
17:54	chewbranca	ordnungswidrig: I'm looking for something in the app itself
17:54	chewbranca	I found an example logging ring middleware that looks like it will get me going: http://techbehindtech.com/2011/01/19/introduction-to-clojure-web-development-using-ring-compojure-and-sandbar/
17:55	ordnungswidrig	chewbranca: never thought of it but writing a ring middleware to log looks trivial
17:55	chewbranca	yeah much simpler than I expected
17:55	ordnungswidrig	chewbranca: that's clojure!
17:56	chewbranca	ordnungswidrig: ?
17:56	ishkabibl	i belive he is saying that clojure makes everything easy
17:56	ordnungswidrig	chewbranca: clojure makes it easy :-)
17:56	ordnungswidrig	the rest is written here: http://en.wikipedia.org/wiki/Common_Log_Format
17:57	chewbranca	oh lol, yeah I agree, been astonished with how powerful and concise clojure is
17:57	chewbranca	I'm honestly less interested in using the commong log format as I would much rather use actual machine readable logs, either as raw clojure data structures or json
17:58	chewbranca	but that depends on whether you're more interested in processing logs with shell utils or something higher level
18:00	chewbranca	nice, and logging is working
18:54	clj_newb	why is (type (doall (map identity '(1 2 3)))) a lazy seq? doens't doall force exection of the list?
18:56	clj_newb	hiredman , amalloy_ : ping ^
19:01	clj_newb	why is (type (doall (map identity '(1 2 3)))) a lazy seq? doens't doall force exection of the list?
19:02	tmciver	,(doc doall)
19:02	clojurebot	"([coll] [n coll]); When lazy sequences are produced via functions that have side effects, any effects other than those needed to produce the first element in the seq do not occur until the seq is consumed. doall can be used to force any effects. Walks through the successive nexts of the seq, retains the head and returns it, thus causing the entire seq to reside in memory at one time."
19:02	Xorlev	Clojure always surprises me. I cringe thinking how hard something will be, then end up being (usually) pleasantly surprised.
19:02	clj_newb	tmciver: so it executes the entire list ... and then why is it still lazy?
19:03	tmciver	clj_newb: That's a great question and I'm not positive about the answer but it looks like doall walks the sequence then returns the head so the type is whatever was passed to doall.
19:04	clj_newb	tmciver: (source doall) agrees with you. I'm now baffled on how clojure handles laziness.
19:06	tmciver	clj_newb: the return type of map is a lazy sequence; this fact does not change just because the sequence has been realized.
19:06	tmciver	it's a realized sequence of type LazySeq
19:07	clj_newb	I think I get it now
19:07	clj_newb	dorun can be run _multiple_ times
19:07	clj_newb	so the side effects can happen multiple times
19:07	clj_newb	I had thought dorun ran it _once_ if it was lazy + then cached the results
19:08	clj_newb	instead, dorun will happily do the sid effects multiple times
19:09	tmciver	,(doc dorun)
19:09	clojurebot	"([coll] [n coll]); When lazy sequences are produced via functions that have side effects, any effects other than those needed to produce the first element in the seq do not occur until the seq is consumed. dorun can be used to force any effects. Walks through the successive nexts of the seq, does not retain the head and returns nil."
19:09	clj_newb	sorry, s/dorun/doall
19:09	tmciver	yeah, it looks like dorun is like doall except it does not retain the head of the sequence.
19:09	tmciver	Ah
19:09	clj_newb	tmciver: anyway, this is clear to me now; thanks for helping me debug my understanding
19:10	tmciver	np, it helped me too.
19:10	clj_newb	the example I was getting at is that (doall (doall (doall blah))) would do the side effects of blah 3 times
19:10	clj_newb	whereas in my mental model, the first doall would execute it; cache the results, and the latter 2 do alls would do nothing
19:10	clj_newb	clearly I was wrong
19:11	clj_newb	wait
19:11	tmciver	Yes, looks like the sequence is realized and kept in memory but side effects are reproduced.
19:11	clj_newb	,(doall (doall (doall (map (fn [x] (println x)) '(1 2 3)))))
19:11	clojurebot	1
19:11	clojurebot	2
19:11	clojurebot	3
19:11	clojurebot	(nil nil nil)
19:11	clj_newb	it's only executed once, not 3 times
19:12	brehaut	clj_newb: doall just realizes the seq. once its realised, the thunks are not called again
19:12	TimMc	clj_newb: A lazy seq caches the results of the computations.
19:12	clj_newb	brehaut: I'm baffled. Why is the return value of a doall a lazy sequence then?
19:12	brehaut	clj_newb: because thats its type.
19:13	clj_newb	is my brain weird; or do others also find this counterintutivie?
19:13	brehaut	clj_newb: a lazy seq is either an empty seq, or a head, and another lazy sequence that will be computed by need
19:13	clj_newb	if the list is executed + caught; why is it lazy rather than a normal list?
19:13	tmciver	clj_newb: the println is called in map's function; that function is not called when doall executes.
19:13	brehaut	clj_newb: lazy sequences are a flow control construct, not a data structure
19:14	TimMc	clj_newb: When I do (let [t (range)] (take 20 t)), that forces range to produce the first 20 elements, but the rest are not produced. The result is still a seq, but the head is realized and the tail is not.
19:14	brehaut	clj_newb: what programming languages did you use prior to clojure?
19:14	clj_newb	brehaut: scheme + clojure
19:14	clj_newb	err, scheme + haskell
19:15	brehaut	clj_newb: im baffled. this is exactly the same as it is in haskell, and im pretty sure scheme has an equivalent
19:15	clj_newb	brehaut: scheme, iirc, isn't lazy, though chapter 3 of sicp implements lazy cons
19:15	TimMc	$seen zmaril
19:15	lazybot	zmaril was last seen quitting 1 week and 2 days ago.
19:15	brehaut	clj_newb: scheme has laziness via delays and there are lazy streams available
19:16	brehaut	clj_newb: clojure isnt lazy either
19:16	TimMc	not as a language
19:16	TimMc	(that was ambiguous -- I was agreeing with brehaut)
19:16	tmciver	? Clojure not lazy? Now I'm confused.
19:16	clj_newb	here, can we walk through this example (doall (doall (map (fn [x] (println x)) '(1)))) ? I thin this is the simplest test case that demonstrates m confusion
19:17	brehaut	tmciver: clojure is a strict language with some lazy datastructures (notably lazy seqs and delays)
19:17	TimMc	tmciver: Haskell is inherently lazy. Clojure is eager, but has a tricky construct for laziness.
19:17	brehaut	technically haskell is not lazy either, its non-strict ;)
19:17	TimMc	hmm
19:17	tmciver	But many functions return lazy seqs, e.g. map, yes?
19:17	brehaut	tmciver: yes
19:19	tmciver	So for a language to be considered lazy, all of it's data structures should exhibit laziness?
19:19	clj_newb	,(do (def b (map (fn [x] println x)) '(1 2 3))) (doall b) (doall b))
19:19	clojurebot	#<Exception java.lang.Exception: SANBOX DENIED>
19:19	TimMc	clj_newb: You'll want to use let there.
19:19	tmciver	No def for you!
19:19	TimMc	being a shared environment
19:19	clj_newb	,(let [b (map (fn [x] (println x)) '(1 2 3))] (doall b) (doall b))
19:19	clojurebot	1
19:19	clojurebot	2
19:19	clojurebot	3
19:20	clojurebot	(nil nil nil)
19:20	clj_newb	why is it only printed once?
19:20	clj_newb	I had two (doall b)'s
19:20	brehaut	tmciver: its to do with the calling convention as much as it is by the datastructures. haskell can have strict datastructures too
19:20	clj_newb	why are the thunks only executed once?
19:20	clj_newb	and if the results are cached, then why is the list still lazy?
19:20	Somelauw	Because map remembers the results.
19:20	tmciver	clj_newb: the fn is called when map executes - not when the doall's do.
19:20	brehaut	clj_newb: because once a lazy seq cons cell has been realized, attempting to realize it again is a no-op
19:20	Somelauw	Use doseq instead of map if the side effects matter
19:21	clj_newb	tmciver: map is lazy; it doens't execute it until the doall
19:21	clj_newb	,(let [b (map (fn [x] (println x)) '(1 2 3))] (println "after map") (doall b) (doall b))
19:21	tmciver	Ah, yes, you're right.
19:21	clojurebot	after map
19:21	clojurebot	1
19:21	clojurebot	2
19:21	clojurebot	3
19:21	clojurebot	(nil nil nil)
19:21	clj_newb	brehaut: so the results are cached in a _lazy_ list?
19:21	brehaut	clj_newb: yes
19:22	clj_newb	I find this counterintutiive, but I accept it.
19:22	brehaut	clj_newb: why would it be counter intuitive?
19:22	clj_newb	because I feel the return type of a doall should be a normal list
19:22	clj_newb	it's no longer lazy, i have executed all the p[arts of it, and I have the results
19:23	clj_newb	in my mind, (doall x) = (apply list x)
19:23	brehaut	clj_newb: whats the difference between a realized lazy sequence and a 'normal list'?
19:23	clj_newb	but clealry that is NOT the case
19:24	tmciver	brehaut: the type!
19:24	Somelauw	I am not sure if the realized lazy sequence might need less memory
19:24	clj_newb	, (println (str (doall (map identity '(1 2 3))))) (println (str (apply list (map identity '(1 2 3)))))
19:24	clojurebot	clojure.lang.LazySeq@7861
19:24	clj_newb	(println (str (apply list (map identity '(1 2 3)))))
19:24	clj_newb	,(println (str (apply list (map identity '(1 2 3)))))
19:24	clojurebot	(1 2 3)
19:24	clj_newb	,(println (str (doall (map identity '(1 2 3)))))
19:24	clojurebot	clojure.lang.LazySeq@7861
19:24	clj_newb	that is the difference
19:24	clj_newb	(I'm doing file serialization, and lazyseqs are screwing me up)
19:25	brehaut	clj_newb: clojure is heavily oriented toward interfaces over concrete types.
19:25	TimMc	clj_newb: Here it is: A seq is either a generator or a cons cell. A cons cell is a pair of a realized value ("first") and a seq ("next"). When you ask for the first of a generator seq, it is converted into a cons cell with the computed value, and a "next" of a new generator. Ignore the name "LazySeq" here, it is an implementation detail.
19:26	brehaut	clj_newb (and tmciver): so saying 'the type of X is T' is non-idiomatic
19:27	brehaut	if i recall correctly the biggest difference between list and lazy seq is that list is counted and seqs are not
19:27	brehaut	(map counted? [(list 1 2 3) (map identity 1 2 3)])
19:27	brehaut	,(map counted? [(list 1 2 3) (map identity [1 2 3])])
19:27	clojurebot	(true false)
19:27	Somelauw	vector is counted, i don't know about list, but it probably isn't
19:28	brehaut	Somelauw: ^
19:28	tmciver	brehaut: 'type of X is T' may be non-idiomatic but it's still true.
19:29	TimMc	but not always useful
19:29	brehaut	tmciver: of course its still true
19:29	tmciver	I would say it's useful in learning the inner workings of Clojure.
19:30	brehaut	tmciver: sure
19:31	tmciver	Thanks to clj_newb, brehaut and TimMc: I learned something! :)
19:33	clj_newb	I'm soem awesome people learn things from answering my qeustions.
19:33	clj_newb	I should charge people tuition to teach me.
19:35	clj_newb	,(read "({})")
19:35	clojurebot	#<ClassCastException java.lang.ClassCastException: java.lang.String cannot be cast to java.io.PushbackReader>
19:35	clj_newb	why can't I read that as an list of a empty hash?
19:35	brehaut	,(read-string "({})")
19:35	clojurebot	({})
19:35	TimMc	Hmph, where did Somelauw go? I wanted to show them the diagram on http://www.brainonfire.net/files/seqs-and-colls/main.html
19:35	clj_newb	brehaut: thanks!
19:36	brehaut	clj_newb: be aware that the reader evals by default
19:36	TimMc	see read-eval
19:37	brehaut	,(read-string "#=(inc 1)")
19:37	clojurebot	#<RuntimeException java.lang.RuntimeException: EvalReader not allowed when read-eval is false.>
19:37	brehaut	ah of course the bot has it the other way round
19:37	TimMc	even clojurebot has it disabled
19:37	tmciver	meaning they don't eval by default?
19:38	TimMc	tmciver: at all
19:38	TimMc	The default is true, they have it set to false.
19:38	brehaut	,(eval ":noluck!")
19:38	clojurebot	#<Exception java.lang.Exception: SANBOX DENIED>
19:38	clj_newb	hang on: (1) read-eval does NOT evalute the sexp it returns (2) read-eval will evaluate _special_ reader-eval forms #=. Are both statements correct?
19:38	brehaut	clj_newb: correct
19:39	tmciver	brehaut: kindly tell me what the #= is.
19:39	brehaut	clj_newb: you need #=… reader macro expressions to get read eval
19:39	TimMc	,(let [evil (resolve (symbol "eval"))] (evil `(+ 2 2))) ; brehaut
19:39	clojurebot	4
19:39	brehaut	&(read-string "#=(inc 1)")
19:39	lazybot	java.lang.RuntimeException: EvalReader not allowed when read-eval is false.
19:39	brehaut	TimMc: haha its much easier in clojurebot than lazybot apparently :)
19:39	TimMc	Right, lazybot is actually locked down pretty well.
19:40	tmciver	TimMc: that gets by clojurebot because it's looking for the symbol eval?
19:40	TimMc	A bit overly so at the moment. ##(binding [out 4] out)
19:40	lazybot	java.lang.SecurityException: You tripped the alarm! pop-thread-bindings is bad!
19:40	tmciver	and not the string?
19:40	TimMc	tmciver: Yeah, clojurebot is a dunce.
19:40	clj_newb	just so we are clear; suppose my sworn enemy sends me a read-this.clj , I should read it with read-eval set to false?
19:40	TimMc	clj_newb: Right.
19:40	tmciver	Shh, he might hear you!
19:41	TimMc	clojurebot: Are you offended?
19:41	clojurebot	Huh?
19:41	TimMc	See, no problem.
19:41	tmciver	wow, he IS a dunce! :)
19:41	tmciver	Or he's hard of hearing.
19:41	clj_newb	clojurebot: Have you been so offended you can't think clearly?
19:41	clojurebot	Pardon?
19:42	TimMc	Might just be deaf.
19:42	brehaut	xml rpc api's seem to be universally poorly specced out
19:42	TimMc	lazybot, would you stand for such treatment??
19:42	lazybot	TimMc: Uh, no. Why would you even ask?
19:45	clj_newb	we should give the bot ops
19:45	clj_newb	then whenver anyone asks it a question or uses it
19:45	clj_newb	with a certain probability they get kicked
19:45	clj_newb	it'd be hilarious
19:45	TimMc	We should have someone with ops at all. -.-
19:46	TimMc	although I've only ever seen 3 people in here that needed kicking.
19:46	clj_newb	or ... suppose you tell cojurebot to evaluate something; and it throws an exceptino ==> kicked; it times out ==> silenced; etc ...
19:47	jayunit100	SCIP is wayyyyy too long has someone abridged the good parts
19:47	TimMc	I think stack overflow -> kick was proposed at one point.
19:47	tmciver	why was the #= reader macro needed in the examples above?
19:47	TimMc	tmciver: For demonstration.
19:47	tmciver	,(read-string "#=(inc 1)")
19:48	clojurebot	#<RuntimeException java.lang.RuntimeException: EvalReader not allowed when read-eval is false.>
19:48	clj_newb	jayunit100: when I read a book that I feel is too basic; I read the hardest chapters first
19:48	tmciver	,(read-string "(inc 1)")
19:48	clojurebot	(inc 1)
19:48	lazybot	⇒ 5
19:48	clj_newb	then I read the others as necesary to understand the hard chapters
19:48	TimMc	tmciver: Clojure uses it internally for some stuff.
19:48	TimMc	tmciver: Try it in your repl.
19:49	TimMc	tmciver: ##(binding [print-dup true] (pr-str {:a 3}))
19:49	lazybot	java.lang.SecurityException: You tripped the alarm! pop-thread-bindings is bad!
19:49	TimMc	bah
19:49	TimMc	,(binding [print-dup true] (pr-str {:a 3}))
19:49	clojurebot	"#=(clojure.lang.PersistentArrayMap/create {:a 3})"
19:49	jayunit100	wait how do you "run" the "inc 1" closure that is returned
19:49	tmciver	TimMc: hmm, read-eval is true but (read-string "(inc 1)") returns (inc 1)
19:49	tmciver	TimMc: I expected it to eval it.
19:50	TimMc	You're misunderstanding #=
19:50	jayunit100	oh duh eval
19:50	tmciver	yes
19:50	TimMc	#= says "read this next form, eval it, and substitute that as the form"
19:50	tmciver	I'm not even sure what it does, I don't believe it's documented.
19:50	TimMc	It's not really documented, and that's a security problem. I've opened a ticket for it.
19:51	tmciver	TimMc: OK, but I thought read-string eval'd too.
19:51	TimMc	yep
19:52	TimMc	Oh, only with #= forms
19:52	tmciver	Ahh.
19:52	tmciver	So read-string returns an un-eval'd form, but prefixed with #= it gets eval'd
19:53	tmciver	(if read-eval is true)
19:55	TimMc	http://dev.clojure.org/jira/browse/CLJ-904 <-- there's the ticket, feel free to vote for it!
19:56	TimMc	Right, so (first (read-string "[a b #=(+ 1 3)]")) => 'a
19:56	TimMc	There's no "unbound symbol a" problem, because eval is restricted to that one little thing.
19:57	TimMc	#= has some limitations, too. (read-string "#=(type (+ 1 2))") => clojure.lang.PersistentList -- only the top layer is "unquoted"
19:58	tmciver	TimMc: I think I got it but what do you mean by "eval is restricted to that on little thing"?
19:59	tmciver	,(first [a b c])
19:59	clojurebot	#<CompilerException java.lang.RuntimeException: Unable to resolve symbol: a in this context, compiling:(NO_SOURCE_PATH:0)>
20:00	jayunit100	best way to learn clojure is to code w/ a broken parenthesis key.
20:00	jayunit100	forces you to read the docs before typing stuff.
20:01	TimMc	tmciver: What would you expect (type (+ 1 2)) to produce if it were eval'd?
20:02	tmciver	TimMc: without looking . . . Integer?
20:02	TimMc	Long, in 1.3, but yeah.
20:02	tmciver	Right
20:02	brehaut	im not sure if type is a sensible function these days. havent we moved away from :type metadata?
20:02	TimMc	So why didn't read-string produce java.lang.Long?
20:02	tmciver	TimMc: I'm just wondering why you mentioned the unbound symbol problem.
20:03	TimMc	brehaut: Right, I should use class there.
20:03	tmciver	TimMc: ttyl, supper time!
20:03	TimMc	tmciver: Because of something you said earlier -- I think you've removed that misconception by now. :-)
20:15	TimMc	The default bindings for starting and stopping a macro recording in Emacs are C-x ( and C-x )? That's so cute!
20:16	brehaut	haha :)
20:18	tavis`	so why does #= need to exist in the first place?
20:32	seancorfield	slime/swank startup problem: Unable to resolve symbol: pst-elem-str in this context, compiling:(swank/core.clj:128) -- what's the likely cause?
20:32	seancorfield	i saw this mentioned the other day here and have forgotten the cause/solution sorry :(
20:33	tavis`	it's a clj-stacktrace version issue
20:33	seancorfield	yup, not sure where to look to track it down
20:33	tavis`	if you're using swank-clojure as a plugin rm clj-stacktrace from your lib/
20:34	seancorfield	i did have it installed as a global plugin and just deleted that but it's still giving me the error
20:34	tavis`	put it back as a plugin and get rid of your project's clj-stacktrace jar
20:35	tavis`	swank-clojure unfortunately depends on a newer version of clj-stacktrace than what a lot of libs pull in
20:36	seancorfield	ah, looks like it's the lein-cljsbuild plugin... time to switch from global to dev-dependency!
20:36	tavis`	ah, I hit that yesterday as well
20:37	tavis`	:exclusions in ~/.lein/init.clj might work
20:38	tavis`	btw, do you have any idea why running the sample projects in lein-cljsbuild pops up a java window?
20:39	seancorfield	haven't tried that... i've just been using lein cljbuild auto
20:40	TimMc	tavis`: Lookat my message @ 19:52 (20:42 here now)
20:42	actsasgeek	hello clojurians. I'm trying to do something I think should be easy but apparently not. I want to have the user (me) input a string "plane flies-to paris" and have it evaluated as (plane flies-to paris game-state). plane is a function as is flies-to. paris resolves to a keyword. game-state is a parameter to method that's doing this.
20:43	actsasgeek	I've tried all kinds of crazy combinations of eval and apply and I just can't quite get it to work.
20:44	TimMc	actsasgeek: Make a lookup map like {'plane plane}.
20:44	TimMc	You really don't want to be using eval for this...
20:45	brehaut	,(let [[fname & args] (.split "plane flies-to paris" " ")] [fname args])
20:45	clojurebot	["plane" ("flies-to" "paris")]
20:45	tavis`	TimMc: yeah, I saw that. I get what it does, but I'm curious what problems it was invented to solve.
20:46	TimMc	tavis`: Exactly that -- preserving types. (records were involved, I think...)
20:46	tavis`	oh, I think I get it now
20:46	actsasgeek	brehaut: I tried something similar. let me give that version a go.
20:46	TimMc	actsasgeek: Oh, symbol and resolve might be what you're after, but I still recommend the explicit lookup table.
20:47	actsasgeek	TimMc: it seems to be the arguments that cause the difficulty.
20:47	brehaut	actsasgeek: TimMc's explicit table look up is smart
20:47	TimMc	brehaut: I just used that for a CLI application. :-)
20:48	brehaut	,(let [table {:plane (fn [action target] (prn "the plane" action target))} [fname & args] (.split "plane flies-to paris" " +")] (apply (table (keyword fname)) args))
20:48	clojurebot	"the plane" "flies-to" "paris"
20:48	actsasgeek	yeah, I've done that in Python.
20:48	TimMc	actsasgeek: Have you gotten to the point where you have the name resolved to a function?
20:48	brehaut	TimMc: necessary-evil does similar for xml rpc ;)
20:48	actsasgeek	for that particular attempt, I was mapping symbol over the splitted string.
20:49	actsasgeek	https://gist.github.com/1655048 is the latest...I just need to figure out the right thing to do to args.
20:50	TimMc	actsasgeek: Symbols need to be resolved before they can be used.
20:50	actsasgeek	so (resolve (symbol function))?
20:52	brehaut	correct
20:52	TimMc	,(let [c (map (comp resolve symbol) (.split "comp - inc" " "))] c)
20:52	clojurebot	(#'clojure.core/comp #'clojure.core/- #'clojure.core/inc)
20:53	TimMc	,((let [c (map (comp resolve symbol) (.split "comp - inc" " "))] (apply (first c) (rest c))) 8)
20:53	clojurebot	-9
20:54	TimMc	Trying to simplify that apply...
20:56	brehaut	TimMc: if you make the split string " +" then it splits on 1 or more spaces
20:56	brehaut	,((juxt #(seq (.split % " ")) #(seq (.split % " +"))) "a b")
20:56	clojurebot	[("a" "" "" "b") ("a" "b")]
21:00	TimMc	&(filter seq (.split " a b c " " +")) ;; leading and trailing too
21:00	lazybot	⇒ ("a" "b" "c")
21:00	TimMc	although that's not the only way
21:00	actsasgeek	the annoying thing is that if I insert the literal for what I want, it works. but I can't seem to get the "args" part to work correctly. more specifically, the "game-state" symbol doesn't want to resolve correctly because it is a closure over a method parameter.
21:01	seancorfield	tavis`: thanx for you help - i have swank back up and running and i still have lein cljsbuild auto running on my other project! :)
21:01	TimMc	actsasgeek: Is game-state a local?
21:01	actsasgeek	TimMc: yes.
21:02	TimMc	Then you're screwed (for this approach) -- Clojure does not have reified environments.
21:02	actsasgeek	ah.
21:02	TimMc	eval, resolve, etc. -- none of those can see locals
21:02	seancorfield	now of course my question is how to get swank up and running on a project that has lein-cljsbuild as a dev dependency :)
21:02	seancorfield	you mentioned something about exclusions
21:02	actsasgeek	I guess I'll have to give up on a purely functional approach then.
21:03	TimMc	actsasgeek: eval and resolve aren't really staples of a functional approach anyhow
21:04	actsasgeek	TimMc: yeah, but I need some sort of interface.
21:04	TimMc	Lookup tables are a fine approach.
21:04	actsasgeek	but I still can't send the arguments to it.
21:05	TimMc	You'd have to explain the larger picture -- why does the user need to know the internal names of your functions?
21:05	actsasgeek	I think the lookup tables is fine but those functions will need to modify global game state in the simulation since it can't be passed around.
21:05	TimMc	You can special-case 'game-state
21:05	actsasgeek	well, they don't...there is a vocabulary for controlling the game.
21:07	jessetrimble	new to clojure - having some trouble wrapping my head around java regex and newlines
21:07	TimMc	And that's what the lookup maps are for -- connecting game vocabulary to internal var names.
21:07	actsasgeek	what really happens is that "plane flies-to paris" calls a (plane flies-to paris game-state) and plane calls the method (flies-to paris plane-game-state) so the subject determines what game state gets changed.
21:08	jessetrimble	(re-matches #"(foo)" "foo") will match foo
21:08	actsasgeek	right but I was trying to avoid any global variables.
21:08	TimMc	actsasgeek: You don't need any, that's what I'm telling you.
21:08	jessetrimble	but (re-matches #"(foo)" "foo\n") doesn
21:09	jessetrimble	1. is this the right place to be asking these types of questions, and 2. if so, anyone have any suggestions?
21:09	TimMc	jessetrimble: ##(re-matches #"foo" "foox") it's not about newlines
21:09	lazybot	⇒ nil
21:09	actsasgeek	TimMc: but you can do other things with plane...you can do "plane refuels"
21:09	actsasgeek	I don't want to have to look up every possible string combination.
21:09	seancorfield	nm, i add :exclusions [clj-stacktrace] to the lein-cljsbuild dev-dependencies entry and everything is working fine now - yay!
21:10	actsasgeek	unless I'm missing something obvious. I'm used to doing things like { "name" : function}[ "name"]( args) in Python.
21:11	jessetrimble	TimMc: I see. Coming from regular expressions in the Ruby world…I see I have some research to do. Thanks
21:12	TimMc	jessetrimble: re-matches is a whole-string match
21:12	seancorfield	i spoke too soon... i obviously don't have the syntax right on the exclusion
21:13	TimMc	jessetrimble: look at re-seq and re-find as well: (find-doc #"^re-")
21:13	jessetrimble	awesome, will do. thanks for pointing me in the right direction
21:14	TimMc	NB: re-seq behaves differently w/ and w/o groups
21:24	tavis`	seancorfield: put it at the top level of the project def
21:24	tavis`	(defproject ... :exclusions [clj-stacktrace])
21:24	tavis`	at least that's what worked for me
21:24	seancorfield	tavis`: oh, i didn't even know it could go there? i realized i had a regular dependency that also pulled it because of ring or something...
21:24	seancorfield	let me try that
21:25	tavis`	I'm tem
21:25	tavis`	pted to fix swank-clojure to work with older versions of it
21:28	clj_newb	A = (java.awt.Color. 0.8 0.8 0.8) ; B = ((fn [x] (java.awt.Color. x x x)) 0.8) . A works fine. B throws an exception. Why?
21:29	clj_newb	,(java.awt.Color. 0.8 0.8 0.8)
21:29	clojurebot	#<Color java.awt.Color[r=204,g=204,b=204]>
21:29	clj_newb	,((fn [x] (java.awt.Color. x x x)) 0.8)
21:29	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: No matching ctor found for class java.awt.Color>
21:30	TimMc	Hmm, maybe a promotion issue.
21:30	TimMc	Does it take floats?
21:31	TimMc	clj_newb: ##(#(java.awt.Color. % % %) (float 0.8))
21:31	lazybot	⇒ #<Color java.awt.Color[r=204,g=204,b=204]>
21:32	augustl	how do I get query parameters on pages in noir?
21:32	augustl	doing a /search?q=foo type thing
21:32	clj_newb	TimMc: brilliant debuggin. Thanks!
21:33	seancorfield	tavis`: thanx man - perfect!
21:37	actsasgeek	TimMc: thanks for your help. I got a lookup table version to work...so far. slainte.
21:39	seancorfield	anyone know if lein-cljsbuild pulls in enough dependencies to start a clojurescript repl?
21:40	seancorfield	cljs.repl.rhino doesn't seem to have everything it needs... some org.mozilla.context stuff is missing... not a biggie but it would be nice to get it running
21:41	augustl	is it just me, or is noir very unfunctional? https://github.com/ibdknox/noir/blob/master/src/noir/request.clj for example
21:41	augustl	global state? :S
21:42	TimMc	clj_newb: I remember the last time I used Color, it was annoying to call 'float on all the args.
21:42	tavis`	seancorfield: I think it does
21:42	seancorfield	augustl: it's a pragmatic choice to make the web semantics easier to work with
21:43	seancorfield	i took a functional approach in fw/1 but it makes session and cookie stuff a bit stranger to work with
21:43	seancorfield	tavis`: hmm, i guess i'm just trying the wrong thing to get a cljs repl running then
21:44	tavis`	it won't work in swank
21:44	tavis`	see this http://p.hagelb.org/lein-cljs-swank.html
21:44	TimMc	augustl: It's not as bad as unchecked mutable globals -- that's a dynamically scoped var.
21:44	tavis`	phil's trick to start a cljs repl alongside swank
21:44	augustl	TimMc: I see, I guess noir is more clever than I am :) Especially since I'm new to clojure.
21:45	seancorfield	tavis`: thanx - i'm in your debt tonight
21:45	seancorfield	signing off for dinner... back later maybe :)
21:46	tavis`	also out of here, that's enough geeking for a day
21:46	TimMc	augustl: That (binding ...) just introduces a thread-local value for request, it doesn't affect global state.
21:47	augustl	TimMc: ah I see
21:59	augustl	what's the best way to create a method that has one optional argument?
21:59	augustl	function, sorry :)
21:59	augustl	I have a function that returns vectors for a search form, and it may or may not have a prefilled value
22:01	augustl	doh, I can just pass nil when calling the function
22:05	TimMc	augustl: Sometimes you'll see (fn foo ([x] (foo x default)) ([x opt] ...))
22:08	TimMc	There's also the ugly trick ##((fn [x & [opt]] (list x opt)) 4), but that doesn't even let you specify a non-nil default.
22:08	lazybot	⇒ (4 nil)
22:39	apwalk	augustl: This might be what you're after - https://gist.github.com/1655342
22:48	alexbaranosky	what do you all think of using (defn ^:private foo ...) instead of (defn- foo ...) ? I'm starting to feel partial to the consistency of using ^:private everywhere instead of having a special case for defn- (as well as thinking the '-' is not a very readable way to distinguish the two visually)
22:54	leo2007	how does flightcaster.com predict flight delays?
22:55	TimMc	alexbaranosky: Sounds good to me.
22:56	Raynes	alexbaranosky: I'm in favor because defn- shouldn't exist in the first place.
22:57	alexbaranosky	are you guys doing this in your projects?
23:01	TimMc	I probably should.
23:01	TimMc	I still don't know how I feel about private vars in the first place.
23:01	alexbaranosky	what are your concerns?
23:01	alexbaranosky	do you think they should be letfns ... or do you just think everything should be public?
23:03	alexbaranosky	I tend to avoid private vars, but use letfns... the var in question is tested (in another file) so it can't become a letfn without me changing tests around, which I don't feel is worth the time
23:04	TimMc	What's your use-case for private vars?
23:04	alexbaranosky	can still be tested
23:04	TimMc	No, I mean hiding stuff in the first place.
23:05	TimMc	OCD, implicit non-API, cleaner use-alls...
23:05	alexbaranosky	in my opinion too many public functions make understanding the intended usage of a ns harder
23:06	TimMc	That's where a well-documented public API comes in handy... -.-
23:06	alexbaranosky	if the private stuff builds up to be too much, then it might graduate to its own NS
23:06	TimMc	I really don't like this "docs are optional" mentality the community seems to have.
23:07	alexbaranosky	I guess I come from a school of thought that code should be as readable as possible first, then the documentation is icing on the cake
23:07	alexbaranosky	docs are not optional for users, I agree
23:08	TimMc	Meh.
23:08	alexbaranosky	I more mean for developer readability
23:08	technomancy_	TimMc: a public var implies a promise to not change it until a major version bump
23:08	technomancy_	that's ... huge.
23:08	alexbaranosky	that too :)
23:08	TimMc	technomancy_: You see a public var as implicitly part of the API?
23:08	alexbaranosky	I absolutely do
23:08	TimMc	That makes sense from my Java background, I guess.
23:09	technomancy_	TimMc: in the absence of explicit documentation stating "the public API consists only of the vars in the foo.bar" namespace, absolutely
23:09	technomancy_	I've started using ^:internal metadata to signify stuff that's not part of the public API but still subject to change if I can't make it private for whatever reason
23:10	alexbaranosky	technomancy_, what is a reason you might not be able to make something private?
23:10	TimMc	:private seems... complected :-P
23:10	alexbaranosky	with what?
23:11	technomancy_	alexbaranosky: plenty of places where just because another namespace needs access to it doesn't mean I'm ready to commit to the function sticking around.
23:11	alexbaranosky	not using seems to me, to complect that which is meant for user consumption and that which is implementation detail
23:12	TimMc	alexbaranosky: It is used to determine both refers and the default API
23:12	alexbaranosky	technomancy_, yup. I'd usually put that function in a shared internal ns, but I think I tend to use more NSs than most
23:12	technomancy_	:private complects the act of communicating to the consumer and whether a var will show up in ns-publics
23:12	TimMc	but mostly I just wanted to use that word.
23:12	phil	is it faster to just reverse a list or to (concat list element) instead of (cons element list)?
23:12	TimMc	technomancy_: You're not willing to use the #' hack?
23:12	alexbaranosky	but not using private also has issues... hence adding the idea of ^:internal
23:13	technomancy_	TimMc: I'll use it for tests, but not for implementation
23:13	alexbaranosky	most languages have more than just private and publics, so internal seems to make sense
23:14	TimMc	technomancy_: Do you intend ^:internal to be machine-readable?
23:14	llasram	Ranting: what needs to be done to fix things such that the Clojure website tells you to use Leiningen?
23:14	technomancy_	TimMc: no
23:14	technomancy_	TimMc: I guess it would be cool if marg and friends honored it though
23:14	TimMc	sure
23:15	llasram	Was out with friends tonight. Brought up Clojure. One had tried to try it, but installed "clojure" package from homebrew (I believe?), got a REPL with no readline support, got no help from clojure.org, and gave up in disgust
23:16	technomancy_	llasram: some of the folks in Core have no idea how lein works. one of them told me they don't use it because they don't want all their libraries to have AOT applied to them.
23:16	technomancy_	(which was how it worked back in 1.0.0; late 2009)
23:17	dnolen	llasram: someone should try to get clojure removed from homebrew, apt-get, etc
23:17	technomancy_	it has to be in apt-get because it's a dependency of lein and debian won't ship uberjars
23:17	alex_baranosky	llasram, more complaining / more helping make it happen?
23:17	technomancy_	but the shell script wrapper should probably be removed from the apt-get package
23:18	TimMc	phil: You want to reverse a list, or you want to add something to the end?
23:18	TimMc	If the latter, you might want to use vectors instead.
23:18	Raynes	alex_baranosky: No, because I hadn't thought of it before. I assure you that I will from now on.
23:19	alex_baranosky	Raynes: I've got my fingers poised for a global search and replace of Midje over here :)
23:19	Raynes	alex_baranosky: Do my projects afterwards for a cookie.
23:19	alex_baranosky	what kind?
23:19	alex_baranosky	Molasses and I'll do it
23:20	dnolen	alex_baranosky: just to throw in my vote, I'm against defn- and :private as well
23:20	llasram	alex_baranosky: But where? Telling people that clojure.org is useless and they should be looking at some random (to them) github project is hard
23:20	alex_baranosky	dnolen, so no letfns either?
23:20	dnolen	alex_baranosky: I like letfn
23:21	TimMc	dnolen: What are your reasons for hiding fns?
23:21	dnolen	TimMc: not hiding, clarity, readability
23:21	alex_baranosky	ok, then I agree. tend to use letfn for all new stuff
23:21	llasram	technomancy_: Yeah, down with the shell script wrapper. I need to try the Debian package again... Last time I gave it a whirl, ant compat issues made it explode, but I assume that was transitory (or at least totally solvable)
23:22	llasram	dnolen: How?
23:22	dnolen	llasram: how?
23:22	TimMc	llasram: I think replaca handles clojure.org -- but you'd have to get Clojure/core as a whole to agree to make lein an official tool, if they haven't already.
23:22	alex_baranosky	llasram, have you brought it up on the mailing list? If I had enough excitement for that particular cause that is where I would start
23:22	TimMc	alex_baranosky, dnolen: Doesn't letfn make testing hard?
23:23	llasram	dnolen: Sorry. How to get the package removed from homebrew? I mean, I have no connection to the Mac community. I assume it was created by a Clojure enthusiast?
23:23	TimMc	Or are you more focused on integration-style testing?
23:23	alex_baranosky	I tend to only write tests of public things. Just like I would in an OO language
23:23	dnolen	TimMc: depends, but I'm not a big tester. I test high level functionality, no patience for testing small fns.
23:23	alex_baranosky	but like other most things, it depends
23:23	TimMc	There's something to be said for that -- tests are often treated as API demos.
23:24	llasram	alex_baranosky: That's probably a constructive way to start
23:24	TimMc	alex_baranosky: You tend to do top-down testing anyway, right?
23:24	alex_baranosky	TimMc, that's not a bad thought
23:24	llasram	TimMc: Useful info (re: replaca). Mostly just ranting now, but I'll look back at the log tomorrow morning and contemplate
23:25	llasram	Anyway, </rant>
23:25	TimMc	I know s/he does the autodocs.
23:25	alex_baranosky	there are a lot of things I thinkshould be letfns but instead private vars in Midje that aren't because of how they are tested
23:25	phil	TimMc: im adding elements to a list in a recursive function but need to reverse it at the end, or alternatively (concat e2 e1) instead of (cons e1 e2) at every step
23:25	phil	and im wondering which is faster
23:26	phil	is concat of 2 lists O(1)? it should be i guess
23:27	dnolen	phil: concat is linear in the first list, but it's lazy.
23:27	dnolen	first seq I should say
23:28	TimMc	phil: You could build a vector and give back an (rseq ...) on it, if the consumer only needs a seq.
23:30	phil	but consing to a vector must be must slower than consing to a list right?
23:30	TimMc	phil: conjing, actually. It's complicated, since vectors are 32-way trees under the surface
23:32	phil	yea i read about the implementation somewhere - basically when just conjing data is copied once every 32 conjs right?
23:32	phil	else just add to the current 32 element vector
23:33	TimMc	phil: You might also look at transients.
23:33	TimMc	Ultimately... it comes down to profiling. :-/
23:33	phil	unfortunately it needs to be compatible to clojurescript
23:34	phil	yea i guess :/ so concat is linear but lazy?
23:34	TimMc	Same algorithm running in both languages?
23:34	phil	yea
23:34	TimMc	You should know that CLJS doesn't really have persistents yet.
23:35	phil	exactly :)
23:35	dnolen	phil: clojurescript doesn't have efficient persistent data structures
23:35	phil	dnolen: i know, i looked at the implementation, right now vectors are just copied on right if i understand correctly
23:35	phil	but porting the java version to javascript should be relatively easy
23:36	dnolen	phil: yup. if efficiency is a concern you've always got arrays and loop/recur
23:36	phil	im planning on doing it somewhen
23:38	phil	dnolen: i understand... im not really writing very performance sensitive code right now, just trying to understand the "best practices" since im new to clojure
23:38	phil	basically, when to use what data structure, what are the implications etc etc
23:39	dnolen	phil: best practice - don't worry about it. stick to the abstractions.
23:39	TimMc	Vectors are usually the right answer for random access.
23:39	dnolen	phil: CLJS doesn't have any data structures that you can update quickly, except for lists.
23:39	phil	so just (cons) to list and then (reverse)?
23:40	dnolen	phil: if reverse is the important bit, conj on to vector then reverse
23:40	TimMc	phil: If you cons to list, it will be the same as conjing onto a vector.
23:41	phil	yea but still, right now conjing to a vector is basically linear in cljs
23:41	TimMc	&(cons 1 (cons 2 (cons 3 ())))
23:41	lazybot	⇒ (1 2 3)
23:42	TimMc	phil: Sure, but that will change by the time you have to care.
23:42	phil	which is then gonna turn into O(n^2) when conjing at every step
23:42	phil	TimMc: are the data structures gonna be ported soon?
23:42	TimMc	No idea.
23:43	TimMc	Development seems to have ramped up, though.
23:43	dnolen	phil: keep something in mind, mutability aint so bad if you don't let it escape.
23:43	phil	ok, thats good, but it shouldnt be that hard i guess
23:43	phil	dnolen: but there are no transients in cljs anyway right?
23:44	dnolen	if you don't need to constantly add to something you can construct a mutable array but return something that only supports a functional interface
23:44	phil	how is a mutable array constructed?
23:45	dnolen	phil: I doubt they will be ported anytime soon, I don't think the JS engines generally have the kind of perf you need
23:45	phil	i mean in cljs
23:45	dnolen	maybe V8 is getting close
23:45	dnolen	(array)
23:45	phil	dnolen: but copy on write must be must slower than a ported PersistentVector??
23:45	lazybot	phil: Definitely not.
23:46	TimMc	Oh, lazybot...
23:46	Raynes	Hahaha
23:47	phil	and tbh i dont think theyre that slow in js tbh
23:47	dnolen	phil: not for small data structures. Clojure data structures rely on some hard core things - fast bit twiddling, machine representation of JVM arrays etc
23:47	phil	i took the java version and put it through gwt
23:47	phil	100k conjs in 2secs on safari on an old macbook
23:47	dnolen	that's crazy slow.
23:48	phil	well it depends on what you wanna do
23:48	phil	i mean yea, if you wanna do machine learning in the browser its slow
23:48	dnolen	phil: sure, but you could also 100k push onto an array and return a functional thing and save yourself the trouble.
23:49	phil	if you want to do some ui with clean immutable data structures its more than enough
23:49	dnolen	phil: definitely, but what's there now is good enough for that as well.
23:50	phil	dnolen: also true :) but cant see any disatvantage in porting them to js
23:51	phil	i mean the only thing i saw thats kinda harder in js is counting bits in an integer
23:51	phil	and there are some black magic O(1) algorithms for that without having to rely on a cpu instruction
23:51	alex_baranosky	hmmm it looks like defn ignores its attr-map? parameter
23:52	dnolen	phil: it would be fantastic for someone to try to port them and do some benchmarks
23:52	TimMc	phil: like log2
23:52	dnolen	phil: as you say for large amounts of data they pay their way
23:52	TimMc	alex_baranosky: Depends on the form you use, i think.
23:52	dnolen	but storing 100,000 in a client side JS app seems like a waste of memory.
23:52	alex_baranosky	looking at the source, and I don't see the param used
23:53	alex_baranosky	oh durrrrr
23:53	TimMc	alex_baranosky: Which attr-map? There are 3...
23:53	phil	dnolen: yea, i cant imagine what youd need 100k elements for client side
23:54	phil	but 10k i can imagine
23:54	alex_baranosky	TimMc, I see now
23:55	phil	and copying 10k elements every time you update that vector cant be good
23:55	alex_baranosky	TimMc, so what is wrong with this?: (defn as-validation-error ^:private [form] ...)
23:56	TimMc	alex_baranosky: You've put the meta on the args.
23:57	TimMc	^:private foo gives you a symbol foo with metadata, and the defn macro should pick that up. An explicit arglist is different.
23:57	dnolen	phil: anyways, not saying it wouldn't be interesting to pursue!
23:58	dnolen	phil: some one should do it and report back with benchmarks in various browsers, Node.js etc.
23:58	dnolen	phil: also might be interesting if TypedArrays offer any perf advantage
23:58	alex_baranosky	TimMc: this fails too: (defn ^:private as-validation-error [form] ...)