#clojure – 2011-06-09

2011-06-09

00:05	jtszzzzz	dnolen, Here is what I am going for https://gist.github.com/1016032
00:20	dnolen	jtszzzzz: does Submission have a fixed number of fields?
00:20	Scriptor_	Hi everyone
00:21	jtszzzzz	dnolen, I don't intend to use Submission anymore. Sorry for confusion. I just intend to return a map keyed by the provided key on the FieldSpec
00:22	dnolen	jtszzzzz: I think your gist is too much context, a smaller complete example of what you want to do would helpful.
00:22	dnolen	jtszzzzz: but so far I'm suspecting you don't need a macro.
00:29	jtszzzzz	dnolen, I just added a Generalized.clj to that gist. This should really boil it down to the guts of the problem
00:32	justinko	I just created a new project via lein and added a function. How can I call that function in "lein repl"?
00:33	dnolen	jtszzzzz: so the issue is you need to create those keys based on the # of lists?
00:33	jtszzzzz	yes, and i need the let bindings based on the number of lsits
00:33	jtszzzzz	and the args into the map based on the number of lists
00:38	jtszzzzz	justinko .. you need to load your namespace .. do something like (require 'myproject.core)
00:40	jtszzzzz	then you can navigate to your namespace like this (ns myproject.core)
00:41	justinko	jtszzzzz: that worked, thanks
00:41	jtszzzzz	np
00:41	dnolen	jtszzzzz: https://gist.github.com/1016073
00:43	justinko	jtszzzzz: just found out I can add ":main my-poject.core" to project.clj to automatically hop into that namespace
00:44	justinko	jtszzzzz: now is there a "reload" type function that reloads the clj source code?
00:44	justinko	for the repl
00:46	technomancy	justinko: (require 'foo.bar :reload)
00:46	jtszzzzz	dnolen, reading this now!
00:48	justinko	technomancy: thanks!
00:52	technomancy	np
00:53	jtszzzzz	dnolen, thanks a lot. That is really nice. I think this will work good for what I am doing!
00:54	dnolen	jtszzzzz: https://gist.github.com/1016088 another version that avoid metadata cleverness.
00:57	dnolen	jtszzzzz: update the last one for readability (IMO)
00:57	dnolen	s/update/updated
00:57	sexpbot	<dnolen> jtszzzzz: updated the last one for readability (IMO)
00:58	jtszzzzz	i like it!
01:10	justinko	omg you can define a :test metadata for inline tests? That is awesome.
01:10	justinko	do any of you guys actually use it?
01:10	justinko	or do you stick with the /test directory?
01:20	dnolen	justinko: I think usage was common in the early days when no tooling, now most people use maven or lein with /test dir.
01:20	dnolen	cake too
03:29	Dranik	hello all!
03:29	Dranik	I have a question
03:29	Dranik	I have the following variable: (def my (ref {:a 1, :b 2}))
03:30	Dranik	And I wanna a reference from inside the hash-map to my
03:30	justinko	nathanmarz: I just read your interview on geektalk
03:30	Dranik	(dosync (alter my assoc :c my)))
03:31	Dranik	but this leads to stack overflow
03:31	Dranik	is there a way to make a cyclic reference {:c my} inside 'my' ?
03:33	nathanmarz	justinko: cool
03:33	hoeck	Dranik: no, not with immutable maps
03:33	Dranik	hoeck: I need a mutable map
03:33	Dranik	that's why I use ref
03:34	hoeck	Dranik: right, but the map is still immutable
03:34	Dranik	hoeck: is there at least any way to do that? even not idiomatic?
03:34	hoeck	the :c key would have to be a ref too
03:34	Dranik	no, I've tried this also
03:35	Dranik	probably I should have used a common Java map...
03:37	hoeck	Dranik: having a ref in the map works, but can be kind of clumsy
03:38	hoeck	Dranik: when you use plain java maps, you loose the threadsafety and other guarantees of clojure maps
03:38	Fossi	probably you need to tell us more about your real problem
03:39	Fossi	then maybe somebody gcan tell you what would be idiomatic ;)
03:39	Fossi	what you are currently trying to do sounds like a huge mess to me tbh :>
03:41	hoeck	Fossi: circular datastructures are not that uncommon
03:41	Fossi	yes
03:41	hoeck	Fossi: well, maybe in the land of functional languages
03:41	Fossi	but this instance sounds messy
03:42	Fossi	esp with mutability and all
03:42	Dranik	Fossi: I'm impoementing a scheme interpreter in clojure using chapter 4 of SICP
03:42	Dranik	I need this cyclic reference when I create a procedure
03:43	Dranik	I define a binding for the body of the procedure and add the current environment to it
03:43	Dranik	hoeck: when I do (dosync (alter my assoc :c (ref my)))), it also throws the same exception
03:44	hoeck	Dranik: the printer throws that exception, because it fails to print your cyclic structure
03:45	Dranik	hoeck: looks like you're right!
03:45	Dranik	I'll try this...
03:45	hoeck	Dranik: I also have to correct myself, your first approach was correct, you don't have to use a nested ref
03:47	hoeck	,(let [my (ref {:a nil})] (dosync (alter my assoc :a my)) (-> @my :a deref :a deref :a deref type))
03:47	clojurebot	clojure.lang.PersistentArrayMap
03:48	Dranik	is there a way to disable the printer?
03:49	hoeck	implementing an interpreter using a circular datastructure does not feel 'right' in clojure
03:50	hoeck	Dranik: (set! print-level 10) on the repl
03:50	Dranik	trying this...
03:50	hoeck	limits printing of nested datastructures to n levels
03:54	Dranik	hoeck: you know what? just disabling the printer solved all the issues! :-)
03:54	Dranik	thanks a lot!
03:55	hoeck	Dranik: you're welcome
05:34	gilecham	Hmmm, anyone able to shed some light on why:
05:34	gilecham	(let (vector a 1) a)
05:34	gilecham	fails stating that "let requires a vector"
05:35	mrBliss`	gilecham: the syntax for let is: (let [a 1] a)
05:49	gilecham	mrBliss`: yes but (vector a 1) should return [a 1] ...
05:51	Chousuke	gilecham: when evaluated
05:51	Chousuke	gilecham: (vector a 1) is a list of two symbols and a number
05:51	Chousuke	evaluating it results in a vector of whatever a is and a number
05:52	Chousuke	s/a is/a evaluates to/
05:52	sexpbot	<Chousuke> evaluating it results in a vector of whatever a evaluates to and a number
05:53	gilecham	Chousuke: so the problem is that, because let is a macro, the call to vector is not evaluated before evaluationg the call to vector?
05:53	Chousuke	gilecham: to put it another way, all code you write must be "literal". If a syntax requires a vector, then you must give it a literal vector.
05:54	Chousuke	gilecham: But when you generate code you can of course generate the vector dynamically.
05:54	mrBliss`	gilecham: right (let is actually a special form)
05:55	Chousuke	hm
05:56	gilecham	Drat. I'm trying to write something which will allow me to automatically replace all arguments of a function with new ones generated by a function call on the original. I was hoping to just wrap a let... round the body of my function
05:56	Chousuke	You can do something like this though: &&(let [a 2] `(let ~(vector ['a a]) (do whatever)))
05:57	Chousuke	hm
05:57	mrBliss`	,`(let ~(vector 'a 1) ~'a))
05:57	Chousuke	how does that inline evaluation wrok?
05:57	clojurebot	(clojure.core/let [a 1] a)
05:57	mrBliss`	Chousuke: ##
05:57	Chousuke	oh right
05:57	Chousuke	damn
05:58	Chousuke	gilecham: note that that ` form only generates the code. to get it to do something, it must be the result of a macro (so that it will be evaluated at runtime) or you must eval it manually
05:58	mrBliss`	gilecham: can you give a small example because I don't really understand
05:58	Chousuke	gilecham: you can probably do that with a macro.
05:58	Chousuke	or you could use the templating facility in core
05:59	gilecham	Chousuke, mrBliss`: Thanks! I thought there was macro magic in here somewhere
05:59	Chousuke	gilecham: this might help http://richhickey.github.com/clojure/clojure.template-api.html
06:02	gilecham	mrBliss`: Intent was to be able to write:
06:02	gilecham	(defn foo [a]
06:02	gilecham	(let [a (transform a)]
06:02	gilecham	(original-body a)))
06:04	gilecham	Off to read up on templates and return with more questions, thanks again
06:06	terom	Why not just use a higher order function?
08:02	void_	hi there, what is this syntactic sugar please? ->
08:03	fliebel	void_: A thrush, almost at least.
08:03	manutter	Threading
08:03	manutter	Let me misuse it, as an illustration...
08:03	fliebel	&(1 (+ 1) (* 2))
08:03	sexpbot	java.lang.ClassCastException: java.lang.Integer cannot be cast to clojure.lang.IFn
08:03	fliebel	&(-> 1 (+ 1) (* 2))
08:03	sexpbot	⟹ 4
08:03	manutter	,(-> 1 (inc) (println) (inc) (println))
08:03	clojurebot	java.lang.NullPointerException
08:04	manutter	darn, too much abuse
08:04	manutter	so much for clarifying the issue
08:04	fliebel	&(macroexpand-1 '(-> 1 (+ 1) (* 2)))
08:04	sexpbot	⟹ (clojure.core/-> (clojure.core/-> 1 (+ 1)) (* 2))
08:04	fliebel	&(require 'clojure.walk)
08:04	sexpbot	⟹ nil
08:04	void_	&(doc ->)
08:04	sexpbot	⟹ "Macro ([x] [x form] [x form & more]); Threads the expr through the forms. Inserts x as the second item in the first form, making a list of it if it is not a list already. If there are more forms, inserts the first form as the second item in second form, etc."
08:05	fliebel	&(clojure.walk/macroexpand-all '(-> 1 (+ 1) (* 2)))
08:05	sexpbot	⟹ (* (+ 1 1) 2)
08:05	manutter	,(-> (identity 1) (+ 1) (* 2))
08:05	clojurebot	4
08:06	manutter	The basic idea is that it lets you take (foo (bar (baz x))) and turn it into (-> x (baz) (bar) (foo)), which can be more readable in certain circumstances
08:07	manutter	,(-> (identity 1) (inc) (println) (inc) (println) (inc))
08:07	clojurebot	java.lang.NullPointerException
08:07	void_	oh I see
08:07	manutter	still cant' get that one to work
08:07	manutter	o well
08:07	void_	yup, this is what google gave me, nice article: http://debasishg.blogspot.com/2010/04/thrush-in-clojure.html
08:07	fliebel	manutter: println return nil
08:07	void_	&(doc identity)
08:07	sexpbot	⟹ "([x]); Returns its argument."
08:08	manutter	doh, of course
08:08	fliebel	manutter: You can also skip the identity and enter 1 directly.
08:08	manutter	,(-> 1 (inc) (inc) (* 2))
08:08	clojurebot	6
08:08	void_	&(-> (1) (inc) (inc))
08:08	sexpbot	java.lang.ClassCastException: java.lang.Integer cannot be cast to clojure.lang.IFn
08:08	manutter	yeah, I was chasing a red herring there :)
08:09	void_	&(-> 1 (inc) (inc))
08:09	sexpbot	⟹ 3
08:09	fliebel	$(->> (range 10) (map #(* 2 %)) (reduce +))
08:09	fliebel	&(->> (range 10) (map #(* 2 %)) (reduce +))
08:09	sexpbot	⟹ 90
08:09	void_	what does it mean if you use ->> instead of -> ?
08:10	manutter	-> comes in handy for things like setting up swing components, where you create something and then do x to it, then do y to it, then do z to it
08:10	fliebel	void_: it inserts the argument at the end instead.
08:10	manutter	The difference between -> and ->> is where the "running" total goes
08:10	raek	void_: (-> x (f a b c)) means (f x a b c), but (->> x (f a b c)) means (f a b c x)
08:10	manutter	let me try again
08:10	fliebel	manutter: Actually, for swing, most actions don't return anything, so you'd use doto.
08:11	void_	this is cool
08:11	manutter	ok, I'm going to just shut up and drink my coffee, it's too early for me still!
08:11	manutter	:)
08:12	manutter	I don't use -> and ->> much yet, that's probably why I don't understand it as well as I think I do.
08:14	void_	it's amazing how whole source of compojure takes 4 files https://github.com/weavejester/compojure/tree/master/src/compojure
08:14	raek	it helps removing nesting: (into {} (map #(Integer/parseInt %) (remove empty? (line-seq rdr)))) -> (->> rdr (line-seq) (remove empty?) (map #(Integer/parseInt %)) (into {}))
08:15	raek	then you can put each step at its own line, and the code becomes beautiful again...
08:15	manutter	there, that's a much better example
08:17	manutter	Yeah, I really like compojure, though the real power behind it is in Ring
08:18	void_	guys, do you think it's a good idea to learn Clojure by building a web application?
08:18	manutter	I'd say so
08:18	manutter	especially if your goal is to learn how to write good web apps in clojure
08:19	void_	I still feel very lost in clojure code
08:19	kephale1	in the void so to speak
08:19	raek	one thing that can be fun to do is to build your own chat application and protocol
08:20	void_	oh yeah definitely
08:20	raek	something very simple
08:20	void_	use some sproutcore+websocket for clients
08:20	void_	or backbone.js or whatever
08:20	raek	then you have a little bit of state management too
08:20	void_	yeah I think that will be the next
08:20	void_	I did something like that in node.js
08:21	raek	I was thinking about basic I/O with ordinary sockets, but by all means use whatever knowledge you have to make it fun :-)
08:21	void_	but node.js seems so simple compared to clojure
08:21	void_	raek: yesterday I made hangman game for telnet
08:21	manutter	void_: have you done the 4clojure.org exercises?
08:21	void_	no sir
08:22	manutter	They're a good remedy for the "I feel lost in clojure" syndrome
08:22	void_	now that looks very nice
08:22	void_	I started koans
08:22	manutter	yeah, same idea
08:22	void_	it's amaing how many learnin resources are out there
08:23	manutter	indeed
08:29	void_	by the way "joy of clojure", anybody read it?
08:29	manutter	It's on my bookshelf, I'm partway thru it
08:30	manutter	It's not intended as an intro text, I like Clojure In Action for that
08:30	void_	yeah, same feelings here - not an intro text
08:31	manutter	though I just got the "Rough Cuts" version of Clojure Programming, and it's looking pretty good too
08:31	void_	rough cuts, like a beta version?
08:31	manutter	Yeah sort of
08:32	manutter	It's what O'Reilly calls their pre-release program
08:32	kephale1	is Raynes' book finished yet?
08:32	manutter	not sure
08:40	void_	hm, so I created a project with lein. And I would like to run REPL on tom of my existing code, is there any easy way to do it?
08:40	void_	oh
08:40	void_	lein repl
08:41	manutter	and if you're using emacs, you can use lein swank and then use M-x slime-connect to run a repl right in emacs
08:41	void_	yeah I'm not
08:41	void_	:)
08:41	void_	textmate
08:41	manutter	Ah
08:41	manutter	I've never used that one
08:41	manutter	what OS is that/
08:41	manutter	?
08:41	void_	Mac
08:42	manutter	Ah, ok
08:42	manutter	I should play with that some time
08:45	kephale1	there was also a recently revitalized lisp GUI for OS X that now uses Clojure and can play fairly nicely with lein… unfortunately the name of it has completely slipped my mind
08:46	manutter	doh!
08:46	kephale1	there is*
08:47	manutter	Hmm, first google result is Emacs...
08:47	manutter	Lemonodor?
08:47	manutter	MCLIDE?
08:48	void_	kephale1: I would like to try that
08:48	kephale1	MCLIDE!
08:49	kephale1	right because it was developed for MCL
08:49	void_	oh ok
08:49	manutter	Might have to give that a whirl
08:49	manutter	First I'm going to stop by and submit the web site to web-sites-that-suck.com tho :P
08:50	void_	If I wanted to play with my code without reloading jvm each time (I use lein run), what is the best way?
08:50	void_	except emacs
08:50	void_	because I'm not in the mood for learning new editor
08:50	void_	manutter: that link is dead :P
08:50	manutter	eh, I probably spelled it wrong
08:50	kephale1	manutter: lol
08:50	manutter	but still
08:50	void_	http://www.webpagesthatsuck.com/
08:51	manutter	That were it! :)
08:53	mrBliss`	void_: are you using https://github.com/swannodette/textmate-clojure ?
08:53	void_	mrBliss`: yes
08:53	void_	but that still doesn't give me repl
08:54	void_	I mean I can execute one line
08:54	void_	which is the same ...
08:54	manutter	omg even the MCLIDE icon is that horrible plaid
08:54	mrBliss`	Command-Shift-L loads a file
08:55	kephale1	manutter: at the least I think the project is pretty recently revived… lets hope they're putting all their efforts into the backend ; )
08:55	kephale1	i concur, emacs FTW
08:55	kephale1	especially when it comes to lisps
08:56	manutter	+1 on both counts
09:02	Vinzent	Is there function combinators somewhere in core or contrib? like (map (fn/and string? empty?) ["" [] ()] => (true false false)
09:03	void_	anybody uses mongo with clojure?
09:03	manutter	Vinzent: there's comp and partial, though from your example I'm not sure that's what you're looking for
09:05	Vinzent	manutter, (fn/and string? empty?) should return the predicate that returns true for things that are strings and empty
09:05	kephale1	Vinzent: i think you'd have an issue with the and since its a macro
09:05	kephale1	but 1 sec i may have an idea
09:05	Vinzent	same as #(and (string? %) (empty? %)), but shorten
09:06	raek	Vinzent: I think fogus suggested such functions on the clojure dev mailing list a while ago
09:06	hoeck	Vinzent: I'm looking for that too :)
09:06	mrBliss`	found it: http://dev.clojure.org/jira/browse/CLJ-729
09:08	raek	and the discussion: http://groups.google.com/group/clojure-dev/browse_thread/thread/899349c6a9b526e0/10233af5f5f13831
09:08	kephale1	Vinzent: I think you'd have to write a function like the one fogus is suggesting
09:08	kephale1	but you might be able to use juxt in it
09:09	Vinzent	found the lib: https://github.com/zahardzhan/fn
09:10	Vinzent	kephale1, right, juxt suites me in my case
09:11	Vinzent	but it's more verbose
09:12	kephale1	yeah, but you can use it to write your function which might be something like a comp of a partial and of a juxt
09:12	kephale1	err that last and is the macro, not the conjunction
09:12	kephale1	trying to get it to work now
09:22	kephale1	though i'm avoiding using that fn library...
09:22	kephale1	,(every? #(apply (fn [x y] (and x y)) ((juxt string? empty?) %)) (list "" ""))
09:22	clojurebot	true
09:22	kephale1	,(every? #(apply (fn [x y] (and x y)) ((juxt string? empty?) %)) (list "" '(a b)))
09:22	clojurebot	false
09:23	kephale1	kinda ugly, but you could wrap something like that into a funcion
09:23	kephale1	function*
09:23	kephale1	hooray for a reason to use juxt : P
09:26	kephale1	again there is weirdness in this because i'm not using fn/and but:
09:26	kephale1	(def predicator (fn [& predicates] #(apply (fn [x y] (and x y)) ((apply juxt predicates) %))))
09:28	kephale1	someone might have a better idea than the nested application there though
09:30	mrBliss`	(defn every-pred [& preds] (fn [& args] (every? #(apply % args) preds)))
09:31	kephale1	thats the one fogus presented right?
09:32	kephale1	hrm, yeah mine is far from solving the 0 arg issue
09:32	raek	,(every? odd? [])
09:32	clojurebot	true
09:35	kephale1	,(and)
09:35	clojurebot	true
09:35	__name__	Interestin
09:35	__name__	*Interesting even
10:25	Dranik	Is there a way to add a library to leiningen project's classpath and not import it into the local maven repository?
10:25	clgv	Dranik: yes, just copy it to your lib directory
10:26	clgv	Dranik: and use :disable-implicit-clean true in your project.clj
10:26	Dranik	when does leiningen clean the lib directory?
10:27	raek	Dranik: when you run lein deps, I think
10:27	Dranik	ah, yeah...
10:27	Dranik	thanks
10:27	raek	Dranik: are you sure the lib is not in Clojars or Maven Central?
10:28	raek	if you make this project publically available, it's much simpler for others if there is an artifact for the dependency
10:28	Dranik	its j2ee5-api.jar...
10:29	fliebel	http://blogs.oracle.com/henrik/entry/moving_to_openjdk_as_the
10:30	kephale1	Dranik: or you can just run your own repository with something like archiva
10:30	raek	Dranik: http://jarvana.com/jarvana/search?search_type=project&project=javaee-api
10:31	Dranik	raek: what's that?
10:31	raek	jarvana?
10:32	Dranik	aha
10:33	raek	jarvana is a search engine for maven artifacts
10:33	Dranik	wow! thanks!
10:33	raek	...the place to look at for non-clojure libraries
10:34	clgv	raek: great. didnt know that before
10:43	dnolen	fliebel: https://gist.github.com/1016858
10:43	dnolen	tabling fixed
10:43	fliebel	dnolen: yea, got a notice from jira, great!
10:44	dnolen	fliebel: caveat at the moment is you can't table defns only def + fn because of lexical scoping issue.
10:44	dnolen	bbl
10:44	fliebel	dnolen: Does htis also mean run is no longer lazy?
10:49	devn	must...find...book: http://www.amazon.com/Art-Prolog-Second-Programming-Techniques/dp/0262193388/ref=reg_hu-rd_add_1_dp
10:49	devn	(for less than 105.00, preferably)
10:52	joly	Am I correct that trampolined functions can't return a vector or map since they implement IFn?
10:53	joly	,(trampoline (fn [a] [1 a 3]) 2)
10:53	clojurebot	[1 2 3]
10:53	joly	... I guess not
10:54	joly	ah, parentheses error locally
10:54	mrBliss`	,(map fn? [(fn []) [] {} #{}])
10:54	clojurebot	(true false false false)
10:56	joly	thanks, I was doing a 4clojure trampoline problem and was using (instance? clojure.lang.IFn r). Totally missed fn?.
10:59	zenoli	devn: AbeBooks has a bunch of copies listed ...2nd ed for ~$40, 1st ed for ~$10.
11:10	anonymouse89	is there a way to exclude a function/macro from being "used" by another ns without explicitly excluding?
11:10	anonymouse89	or to rephrase, something like defn- for defmacro?
11:11	fliebel	anonymouse89: If you can add metadata to a macro, you can say ^:private I guess.
11:12	raek	anonymouse89: (defn- foo ...) is the same as (defn ^{:private true} foo ...) and you can use the latter way for macros
11:12	fliebel	raek: Sure? ##(doc defmacro)
11:12	sexpbot	⟹ "Macro ([name doc-string? attr-map? [params] body] [name doc-string? attr-map? ([params] body) + attr-map?]); Like defn, but the resulting function name is declared as a macro and will be used as a macro by the compiler when it is called."
11:13	anonymouse89	fliebel, raek: yes, this works. Thanks!
11:15	raek	I think that basically (defmacro foo [x y] ...) is equvalent to (defn ^{:macro true} foo [x y &env &form] ...)
11:16	raek	hrm, yes. using an attribute map looks prettier: (defmacro foo {:private true} ...)
11:16	anonymouse89	raek: for whatever reason (defn- mymacro ^{:macro true} ... ) gives me a StackOverflow when I call it
11:20	mrBliss`	,(macroexpand '(defmacro m [args] 1))
11:20	clojurebot	DENIED
11:20	mrBliss`	(do (clojure.core/defn m [args] 1) (. (var m) (setMacro)) (var m))
11:20	mrBliss`	so it's not ^{:macro true}, but (.setMacro ..)
11:21	devn	zenoli: thanks!
11:43	dnolen	fliebel: so I think I have an approach that can work for Rel, will be more involved to implement but will address the generality that I want.
11:43	fliebel	dnolen: Coll, I'm all ears :)
11:44	dnolen	fliebel: write a macro that generates the Rel type for all IFn methods, 1-21 args.
11:45	dnolen	Rel will take 23 params, name meta and 21 fns, possibly initial nil, these are unsynchronized mutable, setters will generate for all o these.
11:45	dnolen	Clojure will all automatically pack all args past 20 into the rest parameter.
11:45	dnolen	the 21st arg I mean.
11:48	dnolen	will need to think about dealing with indexing, but I think this approach can work. I'm interested in a syntax for indexing any arg, as well compound indexes.
11:50	fliebel	dnolen: Okay, so we're now writing a NoSQL db for relations? :)
11:50	dnolen	fliebel: no, some Prologs support this from what I understand.
11:51	dnolen	fliebel: but yeah, large knowledge bases could be queried efficiently.
11:51	fliebel	So Prolog contains an ad hoc, informally-specified, bug-ridden, slow implementation of half of MongoDB.
11:51	dnolen	fliebel: other way around.
11:52	fliebel	So MongoDB contains an ad hoc, informally-specified, bug-ridden, slow implementation of half of Prolog.
11:54	fliebel	dnolen: But, sure sounds cool and all. Only... 20 args! I assume Clojure does this for efficiency?
11:55	clgv	fliebel: "Nobody will need more than 20 args!"
11:55	dnolen	fliebel: no Clojure supports arbitrary number of args, so will Rel. it just that anything more than 20 gets packed into a rest arg, so you need to unpack them yourself.
11:56	dnolen	so only fast path for 1-20.
11:56	fliebel	clgv: Nobody will need more then 5 args 90% of the time, so 20 + varargs seems a weak optimization with a bad sideeffect.
11:56	dnolen	fns that take > 10 args are gross.
11:57	fliebel	dnolen: On the other hand, think about functions like defn and vector...
11:57	clgv	my juxt-map cannot handle varargs yet, since it was too much work and I was content with 8 pairs = 16 args...
11:58	dnolen	fliebel: ?
11:58	fliebel	dnolen: (vector 1 2 3 4) goes over 20 args real quick.
11:58	clgv	fliebel: how often do you specify it manually like this? ;)
11:58	fliebel	clgv: Juxt-map?
11:59	dnolen	fliebel: vector does take arbitrary args, it takes 5 at most.
11:59	fliebel	clgv: Not at all, but I do define functions with more than 20 statements.... maybe.... sometimes
11:59	dnolen	s/doesn/doesn't
11:59	dnolen	s/does/doesn't
11:59	sexpbot	<dnolen> s/doesn'tn/doesn'tn't
11:59	clgv	fliebel: like juxt but instead of buildin a seq/vec it does build a map with specified keys
11:59	dnolen	ergs
12:17	Vinzent	In midje, how can I say that fn returns approximately some number?
12:18	manutter	Round off the result and say it equals exactly some number
12:19	manutter	Hmm, I'm trying to think of a way to make that more specific...
12:20	manutter	Ok, let me rephrase that:
12:20	mprentice	how can i get the direct children of a type in a hierarchy created with make-hierarchy?
12:20	manutter	The value x is approximately equal to y if the absolute value of (x - y) is less than some "close enough" factor
12:22	manutter	so write your midje test to calculate the absolute value of target number minus actual number, and assert that it is < e (where "e" is the margin of error)
12:23	Vinzent	Yes, I can do it, but the right way is to use roughly:
12:23	Vinzent	(fact (/ 666666.666 13) => (roughly 51282))
12:24	manutter	roughly?
12:24	manutter	Oh, is that a midje function?
12:24	Vinzent	Yep
12:25	Vinzent	or, if I want to specify "close enough" number: (fact (/ 666 13.0) => (roughly 51 0.5))
12:25	manutter	well there you go, now I learned something
12:26	Vinzent	Midje is really awesome, isn't it? :)
12:26	manutter	mprentice: I don't know how to get the direct descendants of a type in a hierarchy, but if I had to try, I'd get (descendants h type) and then compare them to each other with isa? to discard the ones that are not direct descendants
12:26	manutter	Yeah, I love midje
12:29	mprentice	manutter: hmm, i guess i'm trying to use hierarchy in a way it's not designed for. i want to pre-walk the hierarchy
12:31	dnolen	mprentice: yeah there's parents, but no children.
12:32	manutter	mprentice: can you build your tree structure first, and then use that to build your hierarchy?
12:33	manutter	That way they'd be parallel structures, but you could walk the one, and use the other for isa? testing and so on.
12:33	mprentice	i don't need the hierarchy, i was just hoping to use it as an easy way to build a tree
12:34	mprentice	but i don't think i gained anything by doing that
12:35	manutter	yeah hierarchies are more a tool for describing relationships I think
13:33	void_	hi. I asked this before but: Is there any way to start REPL from inside my program. Something like read-line -> evaluate it in an infinite loop.
13:33	void_	I was just wondering if there is anything that does this.
13:36	manutter	hmm, there's a clojure.repl namespace
13:37	void_	nah there's only things like dir, source, apropos
13:37	manutter	Ah, no I found it
13:37	technomancy	clojure.main/repl
13:37	void_	oh that
13:37	manutter	omg technomancy out-typed me again
13:37	void_	dvorak
13:37	void_	right technomancy? ;)
13:37	manutter	lol
13:37	technomancy	absolutely
13:38	technomancy	also terseness
13:39	void_	I love how anybody can add examples to http://clojuredocs.org/clojure_core/clojure.main/repl
13:39	void_	or clojuredocs.org
13:42	void_	awesome!
13:43	void_	this (clojure.main/repl)
13:57	timvisher	hey all
13:57	timvisher	anyone know where I can read some good docs on compojure sessions?
13:58	manutter	compojure sessions or ring sessions?
13:58	timvisher	i suppose that compojure would just be using ring sessions
13:58	timvisher	i'm not sure
13:58	timvisher	is there such a thing as compojure sessions?
13:59	timvisher	better yet, all I'm trying to do is stuff a string into a session
13:59	timvisher	i found some references to session-assoc
13:59	timvisher	but i can't find that in the source anywhere
14:01	manutter	all I can find is this: http://clojuredocs.org/ring/ring.middleware.session/wrap-session
14:01	dnolen	timvisher: compojure is really just a routing DSL these days, all session stuff comes from Ring
14:02	manutter	looks like the wrap-session handler just adds a hash map to the :session key on the request object, and you store stuff by putting it in a hash-map your attach to the :session key on the response object
14:02	manutter	(but don't quote me)
14:05	timvisher	so my site is defined using the handler/site function
14:05	timvisher	which auto-wraps sessions around all the routes
14:05	timvisher	I'm trying to figure out how to alter the session now
14:05	timvisher	that's where I'm falling down
14:06	manutter	so if I'm understanding the docs correctly, you should be getting a request map that contains a :session key
14:06	Vinzent	timvisher, just get :session key from request map and assoc what you need
14:07	manutter	The :session key on the request is your session data coming in, and then to modify it, create a new map (just use ordinary assoc on the map you get from session) and attach it to a :session key in your response
14:08	timvisher	gotcha
14:08	timvisher	i think i was expecting things to be more complex. :)
14:08	timvisher	shame on me
14:08	manutter	well, I'm passing on what I think I've heard, so grain of salt, etc
14:08	manutter	but I think that's all there is to it
14:09	manutter	If you use Sandbar there's some more complex session handling available
14:09	manutter	but Ring is pretty vanilla
14:09	timvisher	gotcha
14:09	Vinzent	iirc you should write your middleware to save session, if your handler doen't modify it
14:10	manutter	I think that the standard wrap-session middleware updates the session data
14:10	manutter	The docs say that if you add a :session key to the response, "the session is updated with the new value"
14:11	manutter	If I'm understanding that correctly, that should preserve any session values that you did not change
14:13	Vinzent	manutter, may be, I just remember I've read somewhere about it
14:14	manutter	Yeah, I played with it a while ago, I'm trying to remember what I discovered at the time
14:14	manutter	but I think I'm somewhere in the ballpark anyway
14:14	timvisher	is the referrer sent when a form posts to a url?
14:15	amalloy	timvisher: should be, but you also shouldn't really rely on it
14:15	timvisher	here's the big picture of what i'm trying to do.
14:16	timvisher	I've got a long page with a bunch of small elements, I'd like to
14:16	timvisher	send a post to a route, and then focus the result of the function
14:16	timvisher	called there when I return to the previous page
14:16	timvisher	i'd like to return to whatever url they were on previously, as well.
14:16	timvisher	although at the moment i just redirect to root
14:17	amalloy	timvisher: a popular way to do the latter is to post to /myform?backto=somepage
14:18	timvisher	amalloy: brilliant. :) don't know why i didn't think of storing the window.location in the form
14:18	timvisher	ok, so that solves that problem
14:19	timvisher	then, i can store the title in the session on the way back and focus it there
14:19	timvisher	awesome
14:53	lawfulfalafel	how do you find out the size of a list?
14:53	amalloy	count
14:54	amalloy	&(map count ["string" '(some list) [1 2 3 4]])
14:54	sexpbot	⟹ (6 2 4)
14:54	amalloy	lawfulfalafel: ^
14:59	lawfulfalafel	thanks you amalloy
15:05	pdk	does anyone know if clojure bit-shift-right is equivalent to java >> or >>>
15:05	pdk	iirc >>> sets the high order bits to 0
15:05	TimMc	,(bit-shift-right -1)
15:05	clojurebot	java.lang.IllegalArgumentException: Wrong number of args (1) passed to: core$bit-shift-right
15:05	TimMc	,(bit-shift-right -1 5)
15:05	clojurebot	-1
15:07	TimMc	Looks like sign-extend to me.
15:09	pdk	,(type (bit-shift right 255 1))
15:09	clojurebot	java.lang.Exception: Unable to resolve symbol: bit-shift in this context
15:09	pdk	,(type (bit-shift-right 255 1))
15:09	clojurebot	java.lang.Integer
15:10	zakwilson	Am I understanding correctly that I can't do variable capture in a Clojure macro?
15:11	pdk	you should be able to
15:12	pdk	it has a reader macro for gensym after all
15:12	dnolen	zakwilson: you can, ~'foo
15:12	dnolen	,`(foo)
15:12	clojurebot	(sandbox/foo)
15:12	dnolen	,`(~'foo)
15:12	clojurebot	(foo)
15:13	zakwilson	dnolen: thanks
15:14	zakwilson	pdk: not using gensyms where they would ordinarly be called for usually throws an exception.
15:14	pdk	,~'foo
15:14	clojurebot	java.lang.IllegalStateException: Var clojure.core/unquote is unbound.
15:15	pdk	wait durr
15:28	pdk	is there a changelog for 1.2.1
15:34	miwillhite	@technomancy, one time I was in here asking how to get environment variables…when you told me you prefixed the function call with a comma (it was something like ",(System/getenv)")…was that a typo, or does the comma mean something when used that way?
15:35	Chousuke	miwillhite: on this channel , invokes clojurebot
15:35	raek	miwillhite: comma is whitespace in clojure. it is also the prefix clojurebot listens to for code evaluation
15:35	miwillhite	ah okay, thanks :P
15:36	miwillhite	,(+ 1 1)
15:36	clojurebot	2
15:36	miwillhite	sweet
15:36	ampleyfly	,_
15:36	clojurebot	java.lang.Exception: Unable to resolve symbol: _ in this context
15:36	TimMc	,,,,,,
15:36	clojurebot	EOF while reading
15:39	amalloy	is there a good/standard way to implement IFn by hand rather than with (fn …)? I've written (reify IFn (invoke [this] …)), and that works for 0 args. if i call it with any args, it throws an AbstractMethodError. in some ways that's fine since i don't want to allow any args, but it would be nice if it threw an arity exception instead
15:39	amalloy	i can proxy AFn, but it's my understanding that you should stay away from proxy if you can help it
15:40	amalloy	to avoid an XY issue: i'm doing this because i want to implement IDeref as well as IFn
15:43	hiredman	A nasty way to do it would be to use a Var
15:44	hiredman	actually, well, the fullfils what you said in the most pointless way
15:46	amalloy	well. the more i poke at it the less certain i am that implementing IFn is really a good idea anyway
15:49	lawfulfalafel	can someone help me with this code? http://pastebin.com/AytnsLps
15:49	dnolen	amalloy: I would write a helper macro to throw ArityException on the other arities. I'm planning on implementing IFn and I see there's a lot of boilerplate that could be abstracted away.
15:49	lawfulfalafel	I don't really get how I am supposed to keep the numbers I am generating
15:49	bhenry	lawfulfalafel: you're not thinking functionally
15:50	hiredman	yeah, a deftype+helper macro for replacing AFn is just waiting to be written
15:50	lawfulfalafel	bhenry: so where is the error in my thinking?
15:51	raek	lawfulfalafel: use let instead of def inside functions
15:51	raek	def is not for local variables
15:51	bhenry	,(doc let)
15:51	clojurebot	"([bindings & body]); Evaluates the exprs in a lexical context in which the symbols in the binding-forms are bound to their respective init-exprs or parts therein."
15:52	raek	lawfulfalafel: have you learned about loop and recur?
15:54	raek	lawfulfalafel: also, in a functional solution, you will typically not have any calls to 'do', 'doseq' or any other construct that is made for side-effects
15:55	raek	lawfulfalafel: example of a loop in Clojure to sum all numbers between 0 and x: (loop [acc 0, i x] (if (zero? i) acc (recur (+ acc i) (dec i))))
16:05	amalloy	lawfulfalafel: also line 15 you need whitespace after the +
16:14	naeu	is there a simple way of listing a given object's fields/methods?
16:14	raek	naeu: show
16:15	naeu	raek: thanks
16:15	naeu	now i need to figure out how to unfreeze emacs
16:15	naeu	i tried this in a slime repl: (println (vec (.getMethods (class f))))
16:15	naeu	and it totally froze up
16:15	raek	clojure.contrib.repl-utils/show
16:16	naeu	killing emacs seems cruel
16:16	raek	naeu: I think you can press C-c C-c in the repl
16:16	naeu	raek: yeah, i've tried that
16:17	raek	so it froze emacs itself, and not the swank server?
16:18	naeu	raek: yup
16:18	naeu	raek: i even killed the swank server and emacs is still frozen
17:08	raek	has anyone successfully managed to use autodoc lately?
17:09	raek	if I use the standalone jar with java -jar, it chokes on my usage of clojure.java.io (seems like autodoc evals the code in clojure 1.1)
17:10	raek	if I add it as a dev-dependency, it fails because clojure-contrib 1.1.0-master-SNAPSHOT does not exist anymore
17:11	void_	so I was thinking .. Android is running on JVM. Clojure produces JVM bytecode. Can you create Android apps with Clojure?
17:11	void_	I mean, yes you can (I googled that), but does anybody do it?
17:12	raek	void_: it is possible, but quite slow currently. I hear they are working on it: http://dev.clojure.org/display/design/Android+Support
17:15	offby1	void_: actually, Android doesn't run on the JVM. Google wrote their own, incompatible VM called "Dalvik".
17:15	offby1	In theory, however, it is probably possible to add a Dalvik backend to the Clojure compiler
17:15	void_	oh yeah I now recall that they have their own
17:16	void_	so Google have their own Java compiler too?
17:20	void_	ohhh ->/->> is cool
17:20	void_	(->>
17:20	void_	(mongo/fetch :expenses)
17:20	void_	(map (fn [e] (:title e)))
17:20	void_	)
17:20	void_	this is such beautiful code
17:21	raek	void_: if you tend to use 'map' with anonynous functions a lot, consider using 'for' instead
17:21	Chousuke	void_: why not just (map :title)?
17:21	raek	(for [e (mongo/fetch :expenses)] (:title e))
17:21	void_	thanks guys thanks ;)
17:22	raek	eh, what Chousuke said ;-)
17:22	void_	Chousuke: didn't know you can do that
17:22	Chousuke	it's one of the little awesome features that Clojure has :P
17:22	raek	but seriously. the autodoc deps seems to have been broken for half a year now.
17:23	raek	is the conclusion that nobody document their code?
17:23	raek	or have everyone switched to Marginalia?
17:23	raek	(Marginalia is very neat, but not what I want in this case)
17:25	ohpauleez	raek: I use Raynes' autodoc AND marginalia in my stuff
17:25	ohpauleez	and generate both
17:25	Raynes	ohpauleez: I'm fine with just using marginalia, fyi.
17:26	raek	ah, now Raynes' version works for me. wonder what I were doing wrong.
17:26	solussd	,(->> 3 (* 3) (/ 4) (* 5)) ;it can make ugly code too!
17:27	clojurebot	20/9
17:27	Raynes	I have a version of autodoc?
17:27	Raynes	Cool.
17:27	raek	Raynes: thanks for fixing it
17:27	raek	[org.clojars.rayne/autodoc "0.8.0-SNAPSHOT"]
17:27	Raynes	You're welcome. I'd tell you how I did it if I could remember actually doing it.
17:27	Raynes	I seem to recall doing something minor to it for my cake autodoc plugin.
17:41	void_	hmm why doesn't "\n" make a newline?
17:42	void_	\newline neither
17:42	void_	oh I got it
17:42	void_	I wouldn't pass it to println
18:00	lawfulfalafel	just out of curiosity, why does the empty list evaluate to true?
18:01	lawfulfalafel	It seems like it would be useful for recursive functions that are being passed a list
18:01	lawfulfalafel	or maybe that's just my n00b opinion
18:01	dnolen	lawfulfalafel: you have (seq foo) for that. only nil and false are false-y thank heavens.
18:02	raek	lawfulfalafel: the rules for truthiness is that nil and false are falsey, everything else is truthy
18:02	raek	that something empty can be non-nil allows for more laziness
18:03	raek	otherwise a lazy-seq producing function would have to force the first element to check whether it should return nil or not
18:51	zakwilson	Is there a built-in inverse of coll? (like CL's atom)
18:52	hiredman	coll? doesn't just test for collections, it tests for IPersistentCollection
18:53	Raynes	(comp not coll?)
18:53	hiredman	so java collections will all return false
18:53	hiredman	etc
18:54	amalloy	zakwilson: no, mainly because clojure has a fuzzier dividing line. it's not "lists vs everything else"
18:56	zakwilson	I'd be safe ignoring Java collections in this case, but is there a generic test for single values versus collections that works on everything?
18:57	hiredman	it really depends on what you are doing with it, there is no generic always safe test
18:58	hiredman	if you are writing a macro and dealing with clojure forms coll? and (comp not coll?) are good enough
19:01	zakwilson	That's exactly what I'm doing.
19:01	hiredman	I figured as much
19:02	hiredman	I imagine it being part of some elaborate system of macros
19:02	hiredman	and then I sigh and try to think of something else
19:03	choffstein	Hey all -- I have a question. I have a situation where I am building html content using hiccup with different functions. In on area, I have something like "[:li {:id "myTab"} (fn-to-render-content arg1 arg2)]". My issue is that I need fn-to-render-content to return two vectors, not a vector of two vectors. I was thinking about constructing some sort of macro to "unwrap" a vector and return the individual elements ... like "(u
19:03	choffstein	[1 2 3 [4 5]]))" becomes "1 2 3 [4 5]"
19:03	choffstein	Is that possible?
19:04	hiredman	sounds horrid, but zakwilson is a macro writer, so he can help you
19:05	zakwilson	No elaborate system of macros, but the macro does call a multimethod to mangle its arguments.
19:07	zakwilson	choffstein: you could (apply u [1 2 3 [4 5]])
19:07	zakwilson	Well, not u(nwrap), but whatever function it is you want to call on 1 2 3 [4 5]
19:09	choffstein	zakwilson: I don't want to apply a function though
19:10	choffstein	something more like "[a b c (unwrap [d e f])]" would return "[a b c d e f]"
19:10	hiredman	not possible
19:10	amalloy	choffstein: that's not possible
19:11	choffstein	boo
19:11	amalloy	and for good reason
19:11	choffstein	alright. can't know unless you ask, right? :)
19:11	zakwilson	,(flatten [1 2 3 [4 5]])
19:11	clojurebot	(1 2 3 4 5)
19:11	zakwilson	that's about as close as it gets
19:12	choffstein	Yeah ... but not quite what I am looking for unfortunately. It is alright.
19:12	choffstein	I'll figure it out
19:12	amalloy	wait, choffstein, you want to use this for hiccup?
19:12	zakwilson	It can be done in Common Lisp, but Clojure doesn't have multiple return values.
19:12	choffstein	yes
19:12	amalloy	just return (list [ret1…] [ret2…])
19:13	amalloy	hiccup renders seqs by expanding them into the parent element, roughly
19:13	choffstein	amalloy: hiccup seems to be running into issue having a parent element with no tag
19:13	amalloy	choffstein: with a list parent element? vectors are not lists
19:13	choffstein	doh.
19:14	amalloy	choffstein: try (html [:ul '([:li 1] [:li 2])])
19:14	amalloy	i think that ought to return a :ul with two children
19:15	choffstein	list worked ... vector didn't. doh.
19:15	choffstein	feeling stupid.
19:15	choffstein	thanks :)
19:15	amalloy	choffstein: that's because vectors are for introducing html elements, and you don't want to introduce an html element ;)
19:15	choffstein	that makes a completely reasonable amount of sense :)
19:40	gigamonkey	So everybody seen Fogus's interview with Hickey? http://www.codequarterly.com/2011/rich-hickey/
19:41	brehaut	gigamonkey: and the companion interview with abelson? ;)
19:42	hiredman	I sure hope everyone has seen it, it's been online for hours
19:44	gigamonkey	brehaut: yeah, you can read than one too. ;-)
19:44	gigamonkey	hiredman: come on, we're on Internet time here.
19:44	hiredman	sorry
19:45	hiredman	it's been online for beats
19:45	gigamonkey	heh.
19:46	brehaut	hah
19:46	devn	gigamonkey: i have now
19:48	offby1	what devn said :)
19:53	offby1	"optional type hints", eh? Cool
20:17	offby1	"Hammock-Driven Development". I'd sign up for that.
20:19	Moominpapa	I imagine I'll get a "doc" answer for this one, but is there a function in the standard library equivalent to (fn [coll f] (reduce #(assoc %1 %2 (f %2)) {} coll))?
20:21	brehaut	(into {} (map (juxt identity f) coll))
20:22	hiredman	wrong
20:22	brehaut	figures
20:22	hiredman	Moominpapa: not really
20:22	hiredman	,(let [a {:a 1}] (zipmap (keys a) (map inc (vals a))))
20:22	clojurebot	{:a 2}
20:23	Moominpapa	I guess group-by does the opposite...
20:23	hiredman	brehaut: pardon me, I think I misread that
20:24	brehaut	hiredman: no problem
20:27	Moominpapa	Thanks.
20:32	technomancy	clojurebot: uh oh
20:32	clojurebot	Excuse me?
20:33	Moominpapa	It uses leiningen... :)
20:52	__name__	Nice nick, btw :)
20:53	offby1	there's an offbyone out there, too; we run into each other occasionally and duel
20:54	__name__	Heh
21:36	mattb[]	howdy
21:37	mattb[]	brand new to clojure, lisp, and functional programming in general (raised on C)
21:37	mattb[]	I was looking at different fibonacci implementations and found one using data recursion rather than function recursion:
21:38	mattb[]	(def fib-seq (lazy-cat [0 1] (map + (rest fib-seq) fib-seq)))
21:38	mattb[]	two questions about this:
21:38	amalloy	mattb[]: yeah, brehaut likes that one
21:39	mattb[]	1. does evaluating (rest fib-seq) and fib-seq in the map reevaluate the sequence for each, or once the lazy sequence is evaluated does it cache the values?
21:39	amalloy	lazy seqs are cached
21:39	mattb[]	2. how does the functional community look at problems like this; prefer data recursion, or prefer function recursion?
21:39	mattb[]	alright cool, I figured
21:39	amalloy	mattb[]: the definition there is one that's popular in haskell
21:39	mattb[]	something about data recursion seems ridiculously elegant to me
21:39	brehaut	data recursion like that is good in haskell, less good in clojure
21:40	mattb[]	ah
21:40	brehaut	because its hard to define that in a non-global way
21:40	brehaut	s/hard/awkard/
21:40	sexpbot	<brehaut> because its awkard to define that in a non-global way
21:40	mattb[]	gotcha
21:40	dnolen	brehaut: I disagree.
21:40	dnolen	mattb[]: that style is natural in Clojure as it has neither TCO nor Haskell laziness.
21:40	amalloy	in clojure i like ##(take 15 (map first (iterate (fn [[a b]] [b (+ a b)]) [0 1])))
21:40	sexpbot	⟹ (0 1 1 2 3 5 8 13 21 34 55 89 144 233 377)
21:42	mattb[]	I've come to appreciate s-expressions pretty quickly, but I still have a crazy hard time wrapping my head around code
21:42	amalloy	$google clojure rosetta-code fibonacci
21:42	sexpbot	First out of 61 results is: Fibonacci sequence - Rosetta Code
21:42	sexpbot	http://rosettacode.org/wiki/Fibonacci_sequence
21:42	brehaut	dnolen: im weak on binding-fu and vars, but i dont know how to create a definition of fib like the recurvsive one mattb[] showed that doesnt have a global var (and the head-retention that goes along with it)
21:42	amalloy	mattb[]: an explanation of my code is there
21:43	mattb[]	awesome thanks amalloy
21:43	brehaut	dnolen: what have i got wrong ?
21:43	dnolen	brehaut: you can name anon fns.
21:43	brehaut	dnolen: im not sure if we are talking about the same thing?
21:43	brehaut	oh
21:44	brehaut	actually we are and i misread the definition
21:44	dnolen	mattb[]: lazy-seqs are particularly powerful as they can be mutually recursive.
21:45	mattb[]	so this wikipedia article claims the haskell-inspired fib to be "more elegant"; what makes you prefer the iterate version amalloy?
21:45	amalloy	mattb[]: well, if someone on rosettacode says it's more elegant, it must be true...
21:46	mattb[]	oh lol this isn't wikipedia.. duh
21:47	amalloy	mattb[]: the way the haskell version is implemented on rosetta code involves holding the head of the lazy sequence, so would eventually run out of memory
21:47	mattb[]	"map first ;; throw away the "metadata" (see below) to view just the fib numbers" it doesn't explain this; what's the deal with metadata?
21:47	mattb[]	ahh ok
21:47	mattb[]	that's what was meant by head retention
21:47	amalloy	dnolen is probably right in that it could be implemented better
21:48	amalloy	mattb[]: when i wrote that, i meant something like "turn each [prev, next] pair into just one integer"
21:48	amalloy	feel free to edit in a way as to make that clearer
21:48	mattb[]	ah k
21:50	mattb[]	this does seem a lot more sane than the "more elegant" version
21:51	amalloy	it's an acquired taste, i'm sure
21:53	mattb[]	thanks for the help guys
21:55	mattb[]	by the way, in the doc for iterate: "f must be free of side-effects" how is this explicitly enforced? are builtins flagged for whether or not they allow side effects?
21:55	brehaut	mattb[]: its not
21:55	danlarkin	it should read "f _should_ be free of side-effects"
21:56	mattb[]	k
21:57	brehaut	mattb[]: an effects tracking system is requires a lot of complexity (see also: haskell) akin to (or part of) a type system
21:57	mattb[]	aye
21:58	mattb[]	am I right in seeing haskell as more of a research language than for practical use?
21:58	brehaut	mattb[]: http://www.codequarterly.com/2011/rich-hickey/ search for "I think Haskell is a fantastic" and read the next 3 paragraphs
21:58	dnolen	mattb[]: not true these days, people using haskell for pratical things.
21:58	brehaut	mattb[]: not really.
21:58	brehaut	mattb[]: its practical once you master it, its just not easy to master
21:59	mattb[]	ah
21:59	dnolen	mattb[]: also possible to get verification for Clojure via Datalog/Prolog, I'm working on that.
21:59	mattb[]	nice link brehaut
22:03	mattb[]	something I was having a hard time finding is in situ algorithms in functional languages; the quicksort example I saw for clojure specifically mentioned that "this isn't really quicksort since it's not in situ"
22:03	mattb[]	what would something like that look like?
22:04	brehaut	it would look like the imperative example that had been mangled a little, and it would operate on a transient
22:04	brehaut	in the future it might operate inside a pod, and in haskell it would exist inside an effectful type context.
22:04	mattb[]	is this a case where such an algorithm would just be a poor choice for a functional implementation?
22:05	brehaut	it depends?
22:06	mattb[]	I mean, do you reap the memory benefits of an in situ algorithm if it's not in situ?
22:06	dnolen	mattb[]: remember Clojure's not didactic about purity, you can implement quicksort on a mutable array if you like.
22:06	mattb[]	aye, I'm thinking more from a general functional context though
22:06	dnolen	functional languages like OCaml and Standard ML have similar escape hatches.
22:06	mattb[]	still trying to wrap my head around how you guys get anything done without state :p
22:06	brehaut	mattb[]: we use state when we need to
22:07	brehaut	clojure excels at handling state
22:07	mattb[]	I see
22:07	dnolen	mattb[]: Clojure probably has more state management facilities than pretty much any language out there except Haskell.
22:07	brehaut	as dnolen said, theres mutable arrays, we have access to javas mutable collections, transient versions of clojures collections, and a grab bag of great reference types
22:10	brehaut	mattb[]: imperitive state is also trivially modelable as a bunch of functions that take a 'world' and return a new 'world, and through the power of higher order functions, you can abstract that away
22:12	brehaut	you might have encountered these in C# or F# as Linq or Workflows if you've lived in dotnet land
22:13	mattb[]	yep, coming from C# actually
22:13	brehaut	C# is actually a pretty decent halfway point to functional languages
22:13	brehaut	its just the average stdlib that lets it down
22:13	brehaut	(ie, all the old stuff)
22:14	mattb[]	the biggest hurdle so far is remembering what data structures look like in clojure; I keep trying to pass an integer from a vector to map lol
22:14	mattb[]	the whole lisp thing is eating my brain
22:14	brehaut	i dont follow
22:15	mattb[]	just misinterpreting what I'm actually passing around
22:16	mattb[]	is that "ants" demo from rich hickey still a good example to learn from? I know a few minor things changed syntax-wise
22:16	mattb[]	(e.g. vec in lieu of apply vector)
22:17	brehaut	theres also a few fairly significant library additions that make the ode a bunch cleaner
22:17	brehaut	and defstruct has been deprecated
22:17	mattb[]	aha
22:17	mattb[]	that's good to know
22:17	brehaut	it uses (dorun (for …)) or something (could be doall, or have a map or something) where doseq would be better
22:18	brehaut	and it uses send-off in place of (for example) future from memory
22:19	mattb[]	dorun retains the head like doall?
22:19	brehaut	nah one of them doesnt
22:20	mattb[]	k
22:20	brehaut	but doseq should generally be your first choice for side-effects on sequences
22:20	brehaut	not least of all because it has full comprehension notation supported
22:21	brehaut	doall retains the head btw, dorun discards
22:21	mattb[]	yeah
22:22	mattb[]	found a couple list messages about them
22:23	mattb[]	by the way, can fn [_] be anonymized? I found I couldn't get it to infer an argument in an anonymous function without using it
22:23	mattb[]	(by anonymous I mean #, maybe that's just sugar?)
22:23	brehaut	by fn [_] you mean a functio nthat ignores its argument?
22:24	mattb[]	yea
22:24	brehaut	you cant use the #() reader form to abbreviate it no
22:24	mattb[]	k
22:24	brehaut	but constantly might be what you are after
22:24	brehaut	&(map (constantly :a) (range 10))
22:24	sexpbot	⟹ (:a :a :a :a :a :a :a :a :a :a)
22:25	amalloy	&(map #(do % :a) (range 10))
22:25	sexpbot	⟹ (:a :a :a :a :a :a :a :a :a :a)
22:25	brehaut	amalloy: yuk :P
22:25	mattb[]	haha
22:25	amalloy	not recommended, but fulfills the stated requirements
22:26	mattb[]	what does evaluating the arguments even do?
22:26	amalloy	?
22:27	mattb[]	if you did that to a list with a function in it, would it call the function?
22:27	mattb[]	(the do %)
22:27	amalloy	no
22:27	amalloy	it's a no-op, although it might not get entirely compiled away
22:27	mattb[]	k
22:28	mattb[]	when was defstruct deprecated? it's not documented as such in the api ref
22:28	brehaut	it was deprecated when defrecord was introduced
22:30	brehaut	http://clojure.org/data_structures#Data%20Structures-StructMaps
22:30	brehaut	the word 'depreciated' isnt used
22:30	mattb[]	indeed
22:30	mattb[]	I disregarded that note before
22:30	brehaut	"Note: Most uses of StructMaps would now be better served by http://clojure.org/datatypes."
22:30	mattb[]	so defstruct was a hashmap and defrecord is an actual java class with final fields?
22:31	brehaut	defrecord is still a map
22:31	brehaut	but yes, its backed by a custom class
22:32	brehaut	consider them an intermediate feature of the language though, you dont need them
22:33	mattb[]	so in hickey's example, (defstruct cell :food :pher) would be better served by defrecord or a different structure entirely?
22:34	brehaut	'better' how?
22:34	brehaut	for performance, yeah a record
22:34	brehaut	but a map would suffice
22:34	mattb[]	in terms of deprecation/more "functional"
22:34	mattb[]	k
22:34	brehaut	neither is 'more functional'
22:35	brehaut	structs were just maps with certain keys given a performance optimisation
22:35	mattb[]	aye
22:36	brehaut	the clojure idiom is to write your code in terms of the abstraction, in this case maps (with keyword keys)
22:36	brehaut	and maps are the general case of that
22:37	brehaut	with structs and records as implementation details
22:37	mattb[]	gotcha
22:37	brehaut	thats good, cause i made a meal of the typing ;)
23:08	dnolen	man I love macros
23:10	brehaut	dnolen: any particular reason today?
23:11	dnolen	brehaut: implementing IFn deftype in pure Clojure, lots of crazy boilerplate eliminated.
23:11	brehaut	oh true cool
23:12	brehaut	ifn has a million arg variations right?
23:12	dnolen	yeah
23:47	amalloy	psh. i'm sure careful consideration went into each line of AFn.java
23:48	brehaut	im sure it was either a hand manging emacs chord or an arcane jumble of vim
23:50	dnolen	AFn.java would drive anyone screaming to macros.
23:51	amalloy	dnolen: so you might think, but look at MouseAdapter in awt
23:52	mattb[]	hahaha
23:52	mattb[]	just looked up afn.java
23:53	mattb[]	I'd like to believe someone did this by hand