#clojure – 2011-04-12

2011-04-12

00:41	currentB	how to I get the first result of a strsplit
00:41	currentB	damn wrong room disregard
01:39	dnolen	hmm some interesting optimizations are possible with case, https://gist.github.com/914995
01:44	dnolen	predicate dispatch could be > 5x faster than multimethods in some cases and much more general...
01:45	amalloy	dnolen: multimethods have to support isa/hierarchies
01:45	amalloy	and i don't see anything "more general" about hardcoding a big long case statement?
01:46	dnolen	amalloy: yeah I'm not interested in supporting hierarchies with this at all, only thinking about defrecord/deftype
01:46	dnolen	amalloy: it would be "hardcoded" each method would modify the dag, and the optimized case statement be emitted.
01:46	dnolen	s/would/wouldn't
01:46	sexpbot	<dnolen> amalloy: it wouldn't be "hardcoded" each method wouldn't modify the dag, and the optimized case statement be emitted.
01:46	amalloy	haha good old sed
01:47	amalloy	dnolen: incidentally, if you're optimizing you might as well turn (case foo true a, false b) into (if foo a b)
01:47	amalloy	or do you want to specifically catch non-booleans and barf on them?
01:50	hiredman	case is faster than (if (= ...) ...)
01:51	hiredman	maybe faster than if period, given that if has to do nil checks, etc
01:52	dnolen	anyways food for thought, night.
02:12	devn	what /is/ the idiom in clojure for using and requiring? I've seen 100 seemingly different ways. Heavy :requires with :as, lots of :uses with :only, when do you choose which?
02:13	amalloy	hiredman: he was doing (case (= x y) ...); surely that's not faster than (if (= x y))?
02:13	devn	Not trolling, just curious at what people see the "standard" being in that instance?
02:13	amalloy	devn: i do some of both
02:13	devn	amalloy: what is the motivating factor behind each decision?
02:13	amalloy	can't easily explain which when
02:13	devn	Does it require a survey of the library?
02:14	amalloy	no, i don't think so. more about how related the library is to the core function of the file i'm in
02:14	hiredman	amalloy: if has to do a null check on the value
02:14	devn	an example?
02:14	amalloy	hiredman: really?
02:14	hiredman	so that is two branches vs. one for case
02:14	devn	^very interesting
02:14	hiredman	amalloy: nil and false are falsy
02:14	amalloy	oh, of course
02:15	amalloy	devn: (:use [somnium.congomongo :only [fetch update!]]) (:require [clojure.string :as str])
02:15	amalloy	in a file that mostly does database stuff with a little string manipulation
02:16	hiredman	I mean, I haven't read the part of the compiler that implements it, but it must call RT/booleanCast or similiar, which without inlining must be even slower then branching twice
02:16	amalloy	arguably it should be the other way around, devn, but this way feels more natural to me
02:16	amalloy	hiredman: yes, you're quite right
02:17	hiredman	I smell 1.3, jira is getting cleared out, the auto promotion of earmuffed vars to dynamic is gone
02:55	amalloy	hiredman: did someone decide on 1.3 vs 2.0?
05:46	matthias__	hrm, im trying to install penumbra with leiningen, but i only get a bunch of errors which i dont understand :(
05:50	matthias__	you guys know a good, up-to-date tutorial on how to set up clojure with emacs on ubuntu?
05:50	matthias__	the last one i used turned out to be outdated in the middle and im afraid i may have made a mess here
05:51	ejackson	matthias the best and easiest way is to use elpa as detailed in Technomancy's blog
05:52	ejackson	here: http://technomancy.us/126
05:52	ejackson	this looks like the same thing: http://davidhilton.wordpress.com/2009/07/08/clojure-getting-started-in-5-simple-steps/
06:32	matthias__	thanks, ejackson
06:33	ejackson	good luck, it's a bit of a fight the first time.
06:34	matthias__	looks like the stuff in technomancy's blog is also outdated :\|
06:41	matthias__	what are you using?
06:41	clgv	eclipse with counterclockwise
06:42	matthias__	gosh, the emacs-starter-kit removes the menu
06:42	matthias__	oh igot errors
06:47	matthias__	he says "you really don't need these; trust me" in the init.el
06:47	clgv	lol hihi, sounds funny
06:47	matthias__	i can sure do without scrollbars and the toolbar, but i think i might wanna keep the menu. but maybe he's right? hmm
08:12	raek	matthias__: if you use emacs-starter-kit, then the readme for swank-clojure explains the rest
08:20	markoman	how is internalization, common terms and text translations used to implement in clojure apps?
08:21	joodie	markoman: I use a map of language -> term -> string/function
08:21	clojurebot	map and the other sequence functions used to be lazy, but with the advent of chunked sequences, may or may not be lazy, consult your local ouija board
08:22	joodie	(def strings {:en {:bla "Bla text" :foo (fn [arg] (format...))}})
08:22	joodie	combined with a function to extract the translation given a keyword and language
08:23	markoman	somehow i would need to provide a path to the term because translation is often in some context
08:25	joodie	you can make the path part of the key, or nest more maps.
08:26	markoman	true, im thinking, if namespace could be part of the key
08:27	raek	,(ns-name ns)
08:27	clojurebot	sandbox
08:27	raek	you can use that to get the name of the current namespace as a symbol
08:28	markoman	how about getting name of the function I am in?
08:29	joodie	generally, you can't get the name of the function you're in
08:29	joodie	and I'm pretty sure you don't want to
08:30	clgv	markoman: you can write a frontend/wrapper to defn which enables you to get the name of the function you are in
08:30	markoman	in PHP i have __CLASS__ and __FUNCTION__ constants I can use for such tricks
08:30	markoman	or __FILE__
08:30	joodie	markoman: yeah, but in clojure you can move function definitions around
08:31	markoman	oh yes, true
08:31	raek	functions are just values, and vars can contain any value
08:33	clgv	the defn wrapper would do something like (binding [function-name fname] (do ~@body))
08:33	markoman	namespace could help a little but im afraid i need to be more specific with the path. the reduced problem is {:en {:kuusi "fir" :kuusi "six"}}
08:35	markoman	so I think if i can pass more precise path / context i can get right term
08:35	markoman	without too much har coding of course, im lazy by nature, lol
08:36	Chousuke	you could also do {:en {:kuusi ["fir" "six"]}} if you don't need the extra information about which is which
08:36	Chousuke	though hm
08:37	Chousuke	looks like you do :/
08:38	markoman	sure, but its not practical as is not my example. if hard coded its better: {:en {:tree {:kuusi "fir"} :number {:kuusi "six"}}}
08:38	Chousuke	I suppose you could mark localisable strings with a macro that replaces them with a function call that fetches the correct string at runtime
08:39	markoman	clgv: i need to play with that idea, binding
08:41	markoman	most often translatable terms are inside of hiccup forms...
08:42	markoman	so one idea would be to get class or id path of the term and use it for translation
10:11	sritchie	does anyone have any suggestions on a cleaner way to write this function ?
10:11	sritchie	https://gist.github.com/915542
10:12	sritchie	I'm looking to generate identity matrices, where each nested vector represents a row
10:15	mduerksen	sritchie: ##(vec (map #(assoc (vec (repeat 10 0)) % 1) (range 10)))
10:15	sexpbot	⟹ [[1 0 0 0 0 0 0 0 0 0] [0 1 0 0 0 0 0 0 0 0] [0 0 1 0 0 0 0 0 0 0] [0 0 0 1 0 0 0 0 0 0] [0 0 0 0 1 0 0 0 0 0] [0 0 0 0 0 1 0 0 0 0] [0 0 0 0 0 0 1 0 0 0] [0 0 0 0 0 0 0 1 0 0] [0 0 0 0 0 0 0 0 1 0] [0 0 0 0 0 0 0 0 0 1]]
10:17	sritchie	mduerksen: hey, that looks great, thanks
10:22	mduerksen	you're welcome :)
10:24	ejackson	mduerksen: slick !
10:24	sritchie	##(vec (for [row (range 5)] (assoc (vec (repeat 5 0)) row 1)))
10:24	sexpbot	⟹ [[1 0 0 0 0] [0 1 0 0 0] [0 0 1 0 0] [0 0 0 1 0] [0 0 0 0 1]]
10:25	sritchie	cool
11:10	hugod_	##(take 5 (partition 5 (cycle (concat [1] (repeat 5 0)))))
11:10	sexpbot	⟹ ((1 0 0 0 0) (0 1 0 0 0) (0 0 1 0 0) (0 0 0 1 0) (0 0 0 0 1))
11:27	dnolen	fogus`: did you see my snippet idea of DAG -> case for predicate dispatching? I'm curious as to how you approached the problem. I've been reading over Efficient Predicate Dispatching and papers on how they do it in Standard ML.
11:43	TimMc	hugod_: Nice.
12:16	amalloy	hugod_: ##(cons 1 (repeat 5 0)) seems cleaner than concat
12:16	sexpbot	⟹ (1 0 0 0 0 0)
12:18	hugod_	amalloy: indeed
12:47	cods	Is there a clojure mode for Emacs which is able to correctly indent case? (rather than aligning condition and expressions on the same column)
12:48	amalloy	cods: as far as i can tell that's the "right" way to indent case, bizarre as it seems to me
12:49	amalloy	if you get the latest version of clojure-mode.el, i've added a variable you can customize that would indent this right: M-x customize-variable clojure-defun-indents, then add case to the list
12:49	mec	how do you save once you've done that?
12:50	amalloy	mec: save what? the customization?
12:50	mec	ya
12:50	cods	amalloy: ok, trying it. Sound interesting :)
12:50	amalloy	C-x C-s works fine, mec
12:50	amalloy	the customization screen has Set and Save buttons if you prefer those
12:51	amalloy	cods: thanks for bringing it up. i haven't used case since adding that feature, so mine still indents "wrong". nice to have it look more natural
12:51	cods	hmm, actually I meant cond, but case is similar.
12:52	amalloy	cods: if you don't like cond's indentation you're doing it wrong :P
12:52	amalloy	should be (cond<RET>a b<RET>c d), which will indent fine
12:53	cods	sure if you don't break line between a and b :)
12:53	amalloy	mmm. that's harder. clojure's cond doesn't group pairs with parens, so emacs won't really be able to indent it (and there's no clearly "right" way)
12:54	amalloy	i see people put a blank line, or a comma, after b in that case
12:54	cods	I get lost because I'm used to Common Lisp, where (cond (a b) (c d)) is less confusing
12:54	amalloy	cods: yeah. it's rather more cluttered; it's a compromise
12:55	amalloy	i proposed at one point making (cond [a b] c d) equivalent to (cond a b c d) for just this reason
12:55	amalloy	since a literal vector is always truthy, this doesn't seem to break any existing code
12:56	mec	you could always (defmacro my-cond [& body] (cons 'cond (apply concat body)))
12:56	amalloy	mec: no good
12:56	amalloy	(cond a [b] c d)
12:57	amalloy	or even (cond a (inc b)) - the list goes on
12:57	cods	mec: that what I was thinking, but I prefer conforming to the language, rather than trying to import my habits from another language
12:58	amalloy	well, i'll slop up a quick cond* and see if i like it
12:59	mec	Do i need to change anything else if i upgrade from clojure-mode 1.7.1 to 1.8.1 or just drop-in?
12:59	amalloy	mec: drop-in afaik
12:59	amalloy	i did that when i was hacking it
13:09	amalloy	cods: https://github.com/amalloy/amalloy-utils/blob/master/src/amalloy/utils/core.clj if you're interested
13:11	cods	amalloy: thanks :)
13:12	amalloy	cods: i've carefully engineered that to depend on my amalloy-utils: if you want to use cond* you have to either depend on my repo or look up lazy-loop and rewrite without it
13:28	fogus`	dnolen: I have not had time to look at it yet. I will do so ASAP
13:30	dnolen	fogus`: seems like you get get a near 6X speed increase over multimethods and a with considerable amount of additional generality.
13:31	dnolen	supporting ad-hoc hierarchies might be tricky, but I'm not a big fan of those these days.
13:32	jcromartie	what general naming convention would you use for functions that return predicates?
13:33	jcromartie	make-<predicate>?
13:34	jcromartie	like (defn make-shorter-than? [n] (fn [x] (<= (count x) n)))
13:39	mec_	,(#('%))
13:39	clojurebot	java.lang.IllegalArgumentException: Wrong number of args (0) passed to: sandbox$eval71$fn
14:06	mduerksen	is there an elegant way to convert a record into a normal map, e.g. for sending it over wire? instead of (into {} rec)
14:08	amalloy	mduerksen: that's not elegant?
14:08	mduerksen	well, not elegant enough ^^ ideally it should work with nested records
14:09	mduerksen	thanks a lot i'm asking for, i know ;)
14:10	mduerksen	*thats
14:10	amalloy	mduerksen: (postwalk (partial into {}) nested-records)
14:10	amalloy	well, that's obviously rubbish
14:10	amalloy	but you get the idea
14:12	amalloy	if you want to copy transform-if from my utils library, you could write (postwalk (transform-if #(instance? clojure.lang.IPersistenMap %) (partial into {})) nested-records)
14:12	mduerksen	postwalk? that's not clojure.zip, where is that function in?
14:12	amalloy	clojure.walk
14:13	dnolen	warning; postwalk is horrendously slow.
14:13	mduerksen	ah, that's great, thanks. i think i know what transform-if means, i have an own version of that in my source (i called it change-when)
14:14	amalloy	heh
14:14	amalloy	mine returns a function rather than actually doing the change, which is handy for postwalk
14:15	jcromartie	how does "lein run" work? it says "No :main namespace specified in project.clj."
14:15	jcromartie	I have a key for :main with a symbol for my namespace with a -main fn
14:17	mduerksen	amalloy, dnolen: thanks. i take it as yet another nudge to look further than clojure.core
14:18	amalloy	dnolen: i'm always surprised by the number of people who want to use records, which seem to be primarily for performance, and then do all kinds of slow things with them
14:19	amalloy	using hash-maps all the way i usually wind up happier
14:20	mduerksen	in my case, i use records to add protocols to them, performance is but a nice side effect. where i don't use protocols, i usually just use hash-maps too
14:22	amalloy	mduerksen: protocols are mostly a high-performance, low-flexibility version of multimethods
14:23	technomancy	jcromartie: can you paste project.clj?
14:23	jcromartie	https://gist.github.com/d3e77832ace0e60462fa
14:23	llahna	hi, can partial be used to bind other args than the first one?
14:23	jcromartie	llahna: you want fnil
14:24	amalloy	jcromartie: i don't think that's true
14:24	technomancy	jcromartie: no need to quote symbols in project.clj
14:24	jcromartie	amalloy: ?
14:24	jcromartie	technomancy: ah, ok
14:24	amalloy	jcromartie: he wants to not pass the arg at all, not substitute something for nil
14:24	jcromartie	technomancy: you kids with your tricky macros
14:25	amalloy	llahna: anyway the answer is no. you can use #(some-fn %1 default-arg %2), for example
14:25	clojurebot	Roger.
14:25	amalloy	clojurebot: seriously? knock it off
14:25	clojurebot	Oh, I know the best knock-knock joke! It goes like this... OK, you go first:
14:25	technomancy	jcromartie: yeah, defproject would be a real pain to work with if you had to quote every symbol
14:25	jcromartie	Sorry I thought I had seen fnil used as a way to bind things other than the first arg
14:25	llahna	jcromartie: thanks
14:26	llahna	amalloy: oh ok
14:26	jcromartie	llahna: not so fast... amalloy says you want what he suggested
14:26	mduerksen	amalloy: true. still, a record provides me a type for dispatch. with a plain hashmap, i'd have to put an type identifier inside the map
14:26	jcromartie	technomancy: I un-quoted the namespace but still "No :main namespace specified in project.clj."
14:26	jcromartie	If I use -m findex.core it works
14:27	amalloy	llahna: i have a reorder function, which lets you reorder the arguments to a function and get out a new function. if you really wanted to, you could do like (partial (reorder subs) 2) to get a one-arg version of clojure.core/subs that has the second arg auto-filled in instead of the first
14:27	jcromartie	maybe I'll just update lein first :)
14:27	amalloy	but in practice i think that will lead to more confusion than enlightenment
14:27	dnolen	amalloy: personally I think multimethods are stop gap until we get something much better. the type dispatch function is a ball and chain.
14:28	dnolen	I mean the dispatch fn.
14:28	amalloy	dnolen: like what?
14:28	dnolen	amalloy: matching multiple things means constructing a collection, downstream users of the multimethod are ball and chained to your dispatch fn.
14:29	technomancy	jcromartie: wfm on 1.5
14:29	jcromartie	technomancy: I am on 1.5.0
14:29	jcromartie	derp
14:29	jcromartie	derp derp derp
14:29	jcromartie	never mind
14:29	jcromartie	I had two project.clj in my Emacs
14:29	technomancy	it happens =)
14:30	jcromartie	now, next up: where did stdin go?
14:31	jcromartie	I'm piping the output of find to this program, and in -main I have (println (line-seq (BufferedReader. in))) which prints nil
14:32	technomancy	yeah, that's an ant bug
14:32	technomancy
14:32	jcromartie	hm
14:32	technomancy	they have no plans to fix it
14:32	jcromartie	so...?
14:33	jcromartie	I see
14:33	technomancy	have to use java -cp `lein classpath` clojure.main
14:34	dnolen	amalloy: efficient predicate dispatch and Haskell pattern match + guards - order/exhaustiveness issues are more interested direction than what we currently have.
14:34	jcromartie	that works
14:34	jcromartie	or do I need genclass in that case?
14:35	jcromartie	sorry, using clojure to actually run standalone programs is new to me
14:35	technomancy	jcromartie: you can use clojure.main to get a repl or clojure.main -e "(some-form)"
14:35	technomancy	or if you're on 1.3 you can do clojure.main -m my.ns
14:44	duncanm	amalloy: is choffstein someone who shows up here in #clojure?
14:44	duncanm	amalloy: he just wrote a blog post about hacking clojure
14:45	amalloy	yes
14:46	duncanm	amalloy: is that his irc nick?
14:46	amalloy	yep
14:46	duncanm	ahh, i 'll keep an eye out for him then, i want to find more clojure users in the boston area
14:46	amalloy	duncanm: you can $mail him
14:46	amalloy	$help mail
14:46	sexpbot	amalloy: Send somebody a message. Takes a nickname and a message to send. Will alert the person with a notice.
14:47	technomancy	duncanm: there should be a boston meetup announced soon for the first week of may
14:47	technomancy	or so I've heard...
14:48	duncanm	technomancy: oh cool, there was a meeting a few months ago that i went to, but it went silent afterward
14:48	technomancy	yeah, my boss is attempting to revive it
14:48	amalloy	duncanm: oh cool, i read the first half of that blog post but then got interrupted. neat to see that i'm at the bottom, and it explains why you asked me specifically
14:48	duncanm	technomancy: oh? who's your boss?
14:57	duncanm	dum de dum
14:57	mec_	is it better to make a simple macro that makes a bunch of nested ifs, or a more complex one that rips them all into one if
14:58	duncanm	magnet:
14:58	amalloy	mec_: better still to use cond?
14:58	duncanm	mec_: what about cond?
14:58	mec_	err sorry, not ifs, lets
14:58	amalloy	shrug let is free. do what you find most readable
14:59	duncanm	the reason why traditional Lisps have LET vs LET* is a performance trade-off, right?
14:59	amalloy	duncanm: i don't think so
14:59	duncanm	actually, space trade off?
14:59	duncanm	i don't remember anymore, i think we talked about it in my PL class when i was in school
15:00	duncanm	parallel bindings can be executed in parallel (but I don't know if any LISP systems actually do this)
15:00	duncanm	parallel bindings create a single new environment (scope) for all the bindings. Sequential bindings create a new nested environment for every single binding. Parallel bindings use less memory and have faster variable lookup.
15:00	duncanm	so i'm not totally wrong here....
15:00	mec_	so a big let vs a bunch of nested are more or less equal?
15:00	hiredman	if (let [x 1] ..) ~= ((fn [x] ..) 1), then (let [x 1 y x] ...) ~= ((fn [x y] ..) 1 x)
15:01	mec_	beyond the obvious that is
15:01	amalloy	duncanm: that's probably only relevant in interpreted code
15:01	technomancy	it may have to do with clojure's lack of reified environments
15:01	amalloy	certainly for "faster variable lookup" i can't see that being true in compiled code
15:01	matthias__	when i start swank with lein swank and then connect to it using M-x slime-connect, it always says "versions differ bla bla"
15:01	matthias__	is that bad?
15:01	amalloy	matthias__: nothing you need to worry about
15:02	duncanm	mec_: this is a good explanation: http://stackoverflow.com/questions/554949/let-versus-let-in-common-lisp/587837#587837
15:02	Bronsa	matthias__: put this in your .emacs => (setq slime-protocol-version 'ignore)
15:02	duncanm	but it's more geared towards Common Lisp
15:02	Bronsa	and stop worring about it
15:02	amalloy	i have in my .emacs: (eval-after-load "slime" '(setq slime-protocol-version 'ignore))
15:02	mec_	duncanm: is this valid for clojure tho?
15:03	matthias__	hmm, swank just got an exception. "invalid token: swank::"
15:03	amalloy	it's an artifact of the poor version-control practiced by slime devs, as i understand it
15:03	technomancy	amalloy: that's correct.
15:03	technomancy	well, poor release management.
15:03	duncanm	mec_: Clojure doesn't have LET, its 'let' has the same semantics as LET in older Lisps
15:03	technomancy	absent release management, rather
15:04	duncanm	mec_: so unless you really want to signify some difference in scope stylistically, (let [x 1] (let [y 2] ...)) is equivalent to (let [x 1, y 2] ...)
15:04	hiredman	well, clojure does have a let* but let and let* both behave like common lisp's let*
15:04	amalloy	technomancy: yes, i was being vague rather than trying to find the exactly-specific-enough term :P
15:04	duncanm	hiredman: oh really? let* is a synonym of let?
15:05	hiredman	no
15:05	hiredman	let* lacks destructuring support
15:05	duncanm	oh
15:05	mec_	let* is the builtin compiler function that let uses
15:05	matthias__	http://pastebin.com/YEVnzDb3 thats the exception i got from swank
15:05	matthias__	can you help me?
15:05	duncanm	i didn't know that
15:05	hiredman	let* is the real primitive, let is a macro on top
15:06	duncanm	oh, well, at least that doesn't quite invalidate what i said
15:06	amalloy	&(macro-expand '(let [[x :as all] (range)]))
15:06	sexpbot	java.lang.Exception: Unable to resolve symbol: macro-expand in this context
15:06	amalloy	&(macroexpand '(let [[x :as all] (range)]))
15:06	sexpbot	⟹ (let* [vec__18069 (range) x (clojure.core/nth vec__18069 0 nil) all vec__18069])
15:06	Chousuke	clojure doesn't officially have let* anyway
15:07	Chousuke	it's just an implementation thingy
15:07	duncanm	Chousuke: the naming makes it particularly confusing
15:07	Chousuke	I suppose, if you know CL
15:07	hiredman	it's like fn and fn*
15:07	duncanm	since LET and LET* have well-defined meaning in CL and Scheme
15:07	hiredman	~scheme
15:07	clojurebot	scheme is Scheme is like a ball of snow. You can add any amount of snow to it and it still looks like snow. Moreover, snow is cleaner than mud.
15:08	hiredman	hmm
15:08	duncanm	(i capitalize when i talk about scheme/lisp identifiers)
15:08	hiredman	~scheme
15:08	clojurebot	scheme is Scheme is like a ball of snow. You can add any amount of snow to it and it still looks like snow. Moreover, snow is cleaner than mud.
15:08	hiredman	bleh
15:08	kotarak	,map
15:08	clojurebot	#<core$map clojure.core$map@b36022>
15:08	kotarak	,MAP
15:08	clojurebot	java.lang.Exception: Unable to resolve symbol: MAP in this context
15:08	kotarak	hmmm
15:08	amalloy	kotarak: clojure doesn't SHOUT like common-lisp
15:08	kotarak	amalloy: I KNOW.
15:09	hiredman	~please tell us what the hyperspec says
15:09	clojurebot	hyperspec is not applicable
15:09	duncanm	i got used to seeing uppercase identifiers, it's kinda nice that it makes them standout more
15:09	TimMc	kotarak: Checking for case sensitivity?
15:09	hiredman	clojurebot: why doesn't clojure do like what I read in the hyperspec?
15:09	clojurebot	hyperspec is not applicable
15:10	duncanm	otherwise, you kinda resort to some quoting convention
15:10	kotarak	TimMc: not really
15:10	amalloy	duncanm: blech
15:10	amalloy	clojure supports unicode symbol names. upcasing them would be a disaster for letters you don't recognize
15:11	kotarak	amalloy: clojure supports ascii symbols. Everything else is an implementation detail.
15:11	technomancy	elisp does fine with quoting in docstrings like so: "Pivots down to the `bargle' position."
15:11	duncanm	i'm not saying it's great, but it's easy to follow that uppercase means it's some bounded identifier
15:11	duncanm	technomancy: that's the elisp convention? `foo' like that?
15:11	fibeu	I have a quick qeustion: which IDE is best for doing GUI development in clojure?
15:11	technomancy	duncanm: yeah, erc even knows to highlight it
15:12	amalloy	duncanm: but it doesn't mean that! even in CL, if something returns the symbol A, you don't know if it's bound
15:12	hiredman	http://dev.clojure.org/display/design/Library+Coding+Standards
15:12	duncanm	fibeu: there's not a lot of support for GUI development in clojure - i do quite a bit of it, but i just use emacs, and i don't use any form designer
15:12	TimMc	fibeu: I don't think GUI dev is going to be much different from ordinary dev.
15:13	amalloy	&((juxt identity #(Character/toUpperCase %)) \u039b)
15:13	sexpbot	⟹ [\Λ \Λ]
15:13	jweiss	is there a simpler way to write #(not (nil? %))
15:13	amalloy	(complement nil?)
15:13	amalloy	longer, but simpler :)
15:13	fibeu	So I guess best approach is using java GUI builder to build classes and just call them from clojure?
15:13	kotarak	jweiss: % if you know it's not a boolean
15:14	amalloy	kotarak: ##(not (nil? 4))
15:14	sexpbot	⟹ true
15:14	amalloy	he is effectively casting to a boolean, which yours doesn't
15:14	hiredman	,(if 4 :x :y)
15:14	clojurebot	:x
15:14	TimMc	&(Character/toUpperCase \u131)
15:14	sexpbot	java.lang.IllegalArgumentException: Invalid unicode character: \u131
15:14	kotarak	,(if 4 "o.O" "O.o")
15:14	clojurebot	"o.O"
15:14	hiredman	,(if (not (nil? 4)) :x :y)
15:14	clojurebot	:x
15:14	jweiss	i just want to take-while from an infinite seq while it's not nil
15:15	TimMc	&(Character/toUpperCase \u0131)
15:15	sexpbot	⟹ \I
15:15	jweiss	what pred to use
15:15	duncanm	fibeu: there's something nice about typing it all by hand, if you organize things
15:15	amalloy	&((juxt identity #(Character/toUpperCase %)) \u03bb)
15:15	sexpbot	⟹ [\λ \Λ]
15:15	duncanm	fibeu: if you organize things just the right way, you can change the app while it's running
15:15	hiredman	if does an implicit boolean cast
15:15	kotarak	jweiss: (take-while identity ...) or (take-while (complement nil?) ...)
15:15	kotarak	jweiss: the latter is more secure
15:15	jweiss	ah identity. that's it :)
15:15	kotarak	Ah
15:15	fibeu	duncam: yes, in the REPL, I tried some basic stuff with Swing, it is cool
15:15	amalloy	duncanm: ^ is the example i was looking for about why it would be evil to uppercase symbols
15:16	jweiss	it's iether going to be a string or nil, so i like identity
15:16	fibeu	reminds me of doing things in Smalltalk
15:16	TimMc	&(= (Character/toLowerCase (Character/toUpperCase \u0131)) \u0131)
15:16	sexpbot	⟹ false
15:16	amalloy	&\0131
15:16	sexpbot	java.lang.Exception: Unsupported character: \0131
15:16	duncanm	amalloy: sure
15:16	amalloy	&\u0131
15:16	sexpbot	⟹ \ı
15:16	TimMc	^ This is why we don't do case-folding.
15:16	amalloy	TimMc: what dang character is that
15:16	TimMc	amalloy: Turkish dotless i.
15:16	duncanm	fibeu: yeah, but it takes a lot of care to do it right, often times, i just restart the app
15:16	amalloy	fun
15:17	duncanm	(well, kill the frame and start again)
15:17	amalloy	i guess i knew there were characters that didn't transition nicely, but i never would have found one. thanks
15:17	mec_	wow apparently there are limits on method length ;p
15:17	duncanm	fibeu: also, it's easy to accumulate too many dangling references in the REPL, and then you either can clear it out with (System/gc) or you have to restart your slime session (or you figure out some slime way to clean up....)
15:18	duncanm	mec_: is that a JVM limitation?
15:18	amalloy	mec_: yeah, they're crazy large though. but you can't write your whole app in one big macro :P
15:18	mec_	I would assume so
15:18	amalloy	duncanm: yes
15:18	TimMc	amalloy: http://gizmodo.com/#!382026/a-cellphones-missing-dot-kills-two-people-puts-three-more-in-jail < Interesting case where lack of support for dotless i led to someone being killed. (Sorry for Gizmodo link.)
15:19	duncanm	dotless i? it's used in turkish or something?
15:19	fibeu	duncanm: hmm, never tried slime, I use IDEA
15:19	mec_	looks like the limit is 65536
15:19	duncanm	fibeu: oh, slime is what you'd use with emacs
15:19	duncanm	i tried to like IDEA, but it just doesn't look very nice on the machines that I ran it on (mac and windows)
15:20	TimMc	duncanm: Turkish has both dotted and undotted upper and lowercase Is.
15:20	amalloy	mec_: 65536 of anything in particular? methods can't have more than 64k rubber ducks?
15:20	mec_	on the plus site nested let vs 1 big let doesnt seem to run any slower
15:20	TimMc	English has a dotted lower and an undotted upper. Localization hilarity ensues.
15:20	hiredman	k
15:21	mec_	i think it's characters but I cant be sure
15:21	hiredman	method size it bytes is limited to 64k I believe
15:22	fibeu	I think I will use netbeans to write some GUI classes in java, do the functional part in clojure and glui it together
15:22	duncanm	mec_: in Scheme, i think (let ((a 1)) (let ((b 2)) ....) is the same as (let* ((a 1) (b 2)) ....)
15:22	amalloy	TimMc: that's a pretty amazing localization story
15:22	duncanm	fibeu: i'd be interested to hear how that works out for you
15:23	duncanm	fibeu: i write java code in NBs, but i neverg got around to using their form designer
15:23	duncanm	fibeu: i wrote all the layout code by hand ;p
15:23	fibeu	oh, netbeans GUI designer is great I think
15:23	matthias__	hmm, according to something i found, my problem might be that i got the wrong version of slime, but i did install it with package.el
15:24	duncanm	fibeu: i just don't really like having generated code
15:24	duncanm	but yeah, i should embrace it, it's a much easier/faster way to work
15:24	fibeu	Iyes, lot"s of boilerplate
15:26	amalloy	duncanm: write macros you understand to generate all the code, instead of letting a tool fill your program with code you don't understand
15:26	mec_	anyone know where let* is defined? all I can find is a reference in compiler of defining the symbol LET
15:27	mec_	wow dejavu
15:27	amalloy	mec_: it's in the compiler. the symbol LET is bound to Symbol.intern("let*")
15:27	mec_	but where from there?
15:27	duncanm	amalloy: yeah, that's why i want to write a define-action macro for Clojure+Swing
15:27	matthias__	swank keeps crashign with that unreadable message error :\|
15:28	amalloy	LetExprParser, iirc
15:28	amalloy	LetExpr.Parser
15:28	duncanm	amalloy: the idea is to kinda do SAM conversion, by making define-action return an AFn that also implements the given SAM interface
15:28	TimMc	It's in the Compiler.java file.
15:28	raek	matthias__: which slime version are you using? the one from emacs-starter-kit?
15:29	matthias__	yes
15:29	matthias__	how can i check which version i have though?
15:30	raek	and you connect with M-x slime-connect?
15:30	matthias__	slime-20100404.1
15:30	matthias__	yes
15:30	matthias__	and then swank crashes
15:31	raek	I'm looking in my .emacs.d/ and I have that version too
15:31	matthias__	"unreadable message: (:emacs-rex (swank:autodoc (quote ("+" swank::%cursor-marker%)) :print-right-margin 80) "user" :repl-thread 4)"
15:31	raek	which version of swank-clojure?
15:31	matthias__	i got swank-clojure 1.3.0-SNAPSHOT
15:32	raek	same as the one I use too
15:32	matthias__	:(
15:32	raek	I'l try to update it
15:33	matthias__	swank crashed once i type more than a ( into slime
15:33	raek	...and you don't have any old installation of slime lying around?
15:33	matthias__	*crashes
15:34	matthias__	erm
15:34	raek	ok, this is strange...
15:34	matthias__	i dont know
15:34	matthias__	i might have made a mess, i tried multiple different clojurei nstallation tutorials :D
15:34	matthias__	but i removed the .emacs and .emacs.d and then installed the starter ki
15:34	matthias__	t
15:34	raek	ok
15:34	raek	that's good
15:35	raek	matthias__: there's a slime package in ubuntu (assuming you are using some GNU/Linux variant). you don't have it installed?
15:36	matthias__	who knows B) ill look
15:36	matthias__	i do
15:36	matthias__	it actually seems to be a newer version
15:36	raek	try removing it
15:36	matthias__	on it
15:37	raek	newer versions are known to break compability with swank-clojure...
15:37	matthias__	yeah i heard
15:39	matthias__	still doesnt work, but its different now
15:39	raek	did you install both slime and slime-repl with package.el?
15:39	matthias__	"exception in read loop
15:39	matthias__	java.lang.Exception: Error reading swank message
15:39	matthias__	at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
15:40	matthias__	yes
15:40	matthias__	but who knows what else i had lieing around D:
15:41	raek	what does "java -version" say?
15:41	matthias__	java version "1.6.0_20"
15:41	matthias__	OpenJDK Runtime Environment (IcedTea6 1.9.7) (6b20-1.9.7-0ubuntu1)
15:41	matthias__	OpenJDK 64-Bit Server VM (build 19.0-b09, mixed mode)
15:42	raek	matthias__: could you post the stacktrace somewhere?
15:43	matthias__	http://pastebin.com/53knb18c
15:45	raek	you could try switching the Java implementation to the Sun variant
15:46	matthias__	huh, i dont even get a repl anymore when i run slime-connect
15:46	matthias__	is that normal?
15:46	matthias__	used to see that fancy anymation right away
15:46	amalloy	no
15:47	matthias__	can i open one manually?
15:47	amalloy	that's a broken, there
15:47	matthias__	i got a slime-events buffer
15:47	matthias__	so is openjdk known to cause issues or was that just a wild guess?
15:48	raek	I'm not sure, but getting an exception from sun.reflect.NativeConstructorAccessorImpl.newInstance feels suspicious :/
15:49	matthias__	hasnt crashed this time, but i cant try running code because i dont see a repl
15:49	matthias__	i uninstalled cl-swank and something else. dont even know why i had cl stuff, never used it
15:50	hiredman	raek: that is just a reflective call to a constructor
15:53	raek	matthias__: anyway, here's some instructions on how to install the sun jvm: http://www.webupd8.org/2010/09/how-to-install-java-jre-and-java-plugin.html
15:54	raek	matthias__: does the problem persist if you use swank-clojure 1.2.1?
15:55	matthias__	uh
15:56	matthias__	how i do that?
15:56	raek	how do you start swank-clojure currently?
15:56	matthias__	lein swank
15:56	matthias__	do i have to uninstall 1.3.0 and install 1.2.1?
15:57	raek	matthias__: did you add swank-clojure to you project.clj or did you do a lein plugin install?
15:57	raek	in the latter case, yes
15:58	mec_	How could I do this? `(foo ~(rest (bar-macro ~@body))) I want to replace the head of the result from (bar-macro ~@body) with foo
15:58	matthias__	yeah, the latter
15:58	raek	lein plugin uninstall swank-clojure "1.3.0-SNAPSHOT" && lein plugin install swank-clojure "1.2.1"
15:59	matthias__	"Plugin "swank-clojure 1.3.0-SNAPSHOT" doesn't appear to be installed.
15:59	matthias__	" what the hell :\|
15:59	raek	mec_: you can't, since that subsitution will hapen after the foo macro is done
15:59	Chousuke	mec_: hmmh, that's a bit tricky
16:00	raek	or can you?
16:00	mec_	(cons `foo (rest (macroexpand-1 `(bar-macro ~@body)))) works but seems iffy
16:00	Chousuke	I was about to type something like that
16:01	Chousuke	it's not that bad :/ the macroexpansion still happens at macroexpansion time so it's not like you're using eval or something :)
16:02	tsdh	Why are there no nightly clojure builds since december last year?
16:02	amalloy	do you control bar-macro? it sounds like it would really rather be a function than a macro
16:02	amalloy	tsdh: sun hasn't gone down yet, in antarctica?
16:02	mec_	amalloy: ya I was just thinking about pulling bar-macro out into a helper function that is referenced by both
16:02	matthias__	oh, there's lisp connection closed unexpectedly: connection broken by remote peer in messages
16:02	matthias__	but swank doesnt seem to have crashed
16:03	tsdh	amalloy: ;-)
16:03	Chousuke	you could also use the macro function instead of the macro so you don't need macroexpand-1.
16:03	Chousuke	but that's maybe overkill :P
16:03	mec_	do what now?
16:04	Chousuke	,(@(var defn) 'foo '[args] 'body 'here)
16:04	clojurebot	java.lang.RuntimeException: java.lang.RuntimeException: java.lang.IllegalArgumentException: Don't know how to create ISeq from: clojure.lang.Symbol
16:04	Chousuke	hmm
16:04	Chousuke	oh right macros take the weird extra args too :/
16:04	amalloy	Chousuke: it wants some nils for &env and &body
16:04	Chousuke	never mind, best not try thtat then :D
16:05	matthias__	erm, i had something about 1.2.0 in my project.clj :\|
16:05	amalloy	,(#'and nil nil 'a)
16:05	clojurebot	a
16:05	matthias__	i dont even remember where i got that one
16:05	matthias__	still not getting a repl in emacs
16:05	mec_	are &env and &body supposed to be anything besides nil?
16:05	matthias__	is there really no way to manually start a slime repl?
16:05	amalloy	mec_: the compiler sets them
16:05	amalloy	matthias__: yes, you manually start it with slime-connect
16:05	amalloy	that's it
16:06	matthias__	:(
16:06	matthias__	how i find out why it doesnt work?
16:06	raek	matthias__: so you have put a more recent one in your project.clj and done lein deps?
16:06	amalloy	it can't start a repl with a swank server it can't connect to
16:06	matthias__	i didnt do line deps
16:07	matthias__	got a bunch of artifact missing errors
16:07	matthias__	it did say it connected
16:07	matthias__	Connected. Lemonodor-fame is but a hack away!
16:08	matthias__	perhaps i need to add some kind of repository for lein
16:08	matthias__	hmm
16:08	raek	you have the latest version of lein?
16:09	raek	you could remove lib/, add [swank-clojure "1.3.0-SNAPSHOT"] to project.clj and a lein deps again
16:09	raek	swank-clojure is in clojars, so you shouldn't have to add any repo
16:10	raek	what artifact is missing?
16:10	raek	also, do you still have the global swank-clojure install (form lein plugin install)? if so, you don't need any in project.clj
16:11	matthias__	i cant tell which is missing
16:11	matthias__	now it worked (after i removed it from project.clj)
16:12	matthias__	the lein deps that is
16:12	raek	sounds like a typo
16:13	matthias__	"[leiningen/lein-swank "1.3.0-SNAPSHOT"]" iswhat is said before
16:13	raek	ooh, that's also old... :)
16:13	matthias__	slime still wont show me a repl, but is connected
16:13	raek	"lein swank" is in swank-clojure now
16:13	matthias__	as i said i tried a bunch of outdated tutorials on how to set up clojure and made a mess :(
16:13	raek	matthias__: replace that with [swank-clojure "1.3.0-SNAPSHOT"]
16:14	raek	then you should be good to go :-)
16:15	matthias__	still getting no repl
16:16	raek	first, did the swank server start?
16:16	matthias__	all i get is some stuff in Messages and a slime-events buffer
16:16	matthias__	oh i crashed again
16:16	matthias__	didnt do that before
16:16	raek	i.e. do you get the #<ServerSocket ServerSocket[addr=localhost/127.0.0.1,port=0,localport=4005]> line?
16:16	matthias__	i did, but then it crashed
16:17	matthias__	same as the last error i put on pastebin
16:17	raek	what does "ls ~/.lein/plugins/" say?
16:18	raek	and could you paste your project.clj?
16:18	matthias__	swank-clojure-1.3.0.jar swank-clojure-1.3.0-SNAPSHOT.jar
16:18	matthias__	http://pastebin.com/3S64NFRW
16:19	matthias__	thanks a lot for helping me out, btw
16:19	matthias__	now i'm wondering why i have two swanks
16:19	raek	ok, do a rm ~/.lein/plugins/* ~/.lein/bin/*
16:20	raek	and then remove lib/ in your project and do a lein deps and lein swank
16:21	raek	...and before that, change version to "1.3.0" in project.clj
16:21	raek	(didn't realize that one was released)
16:21	raek	[swank-clojure "1.3.0"]
16:22	matthias__	gosh
16:22	matthias__	can i also do that after? :p
16:22	raek	yes, just do a lein deps
16:23	raek	it should remove the old stuff from lib/ too, I was just a bit paranoid before...
16:23	matthias__	i have a feeling the problem is actually with slime right now. maybe
16:23	matthias__	i mean it says it connects
16:23	matthias__	but doesnt give me a repl
16:23	matthias__	the newer version of slime always did, but then crashed after
16:23	raek	is there any slime-repl clojure buffer anywhere?
16:23	matthias__	or made swank crash rather
16:24	matthias__	no
16:24	matthias__	only slime-events
16:25	raek	so you get "Connecting to Swank on port 4005.."
16:25	raek	and then "Connected. Take this REPL, brother, and may it serve you well."? (or other random quote)
16:25	matthias__	yes
16:25	matthias__	actually is said [2 times] after the connected
16:26	raek	but no repl? that's odd..
16:26	matthias__	and then it says byte-code: beginning of bugger [6 times]
16:26	matthias__	*buffer
16:26	matthias__	after the random quote
16:26	raek	and the "lein swank" terminal hasn't changed?
16:26	matthias__	nope
16:27	matthias__	not this time anyways :\|
16:28	raek	the only thing I can think of now is to remove .emacs.d and replace it with a fresh install of emacs-starter-kit, and then install clojure-mode, slime and slim-repl again
16:28	matthias__	hmm, i kinda only just installed emacs-starter-kit
16:28	raek	I'm not sure what happens if you try to install those while you have the system slime package installed
16:28	matthias__	but i guess i should try that :(
16:28	matthias__	ah
16:28	matthias__	maybe
16:29	matthias__	huh, swank crashed after i closed emacs
16:29	raek	I get that at my setup too, if I exit emacs while the slime<->swank conenction is running
16:30	raek	hrm, and then I get the same exception as you did...
16:30	matthias__	i think this is looking like slime is the problem
16:30	raek	probably means that swank-clojure got an EOF while reading or something
16:32	matthias__	reinstalled the starter kit
16:34	matthias__	ihuh, slime-repl failed to install
16:34	matthias__	Error: No setf-method known for slime-connection-output-buffer
16:35	raek	try installing one at a time, and with slime before slime-repl
16:35	matthias__	did have slime before slime repl
16:37	edw	Does ztellman of aleph fame hang here?
16:37	raek	matthias__: and you're not trying to install the "swank-clojure" package, right?
16:37	matthias__	no, i installed clojure-mode and slime and then tried slime-repl
16:38	matthias__	i got the exact same error as the commenter on that: http://webcache.googleusercontent.com/search?q=cache:http://blog.clojurls.com/setting-up-clojure-with-emacs-and-leiningen
16:38	matthias__	i'm too dumb to copy stuff out of emacs :D
16:39	raek	matthias__: M-w :-)
16:40	matthias__	slime-repl.el:122:39:Error: No setf-method known for slime-connection-output-buffer
16:40	matthias__	haha
16:40	matthias__	thanks
16:41	matthias__	seems like slime-repl might actually be installed, but just not byte-compiled
16:41	matthias__	according to 3. here http://groups.google.com/group/swank-clojure/tree/browse_frm/thread/20141a1e19c60339?_done=%2Fgroup%2Fswank-clojure%2Fbrowse_frm%2Fthread%2F20141a1e19c60339%3F&
16:42	matthias__	i was actually going to install swank-clojure after slime-repl :>
16:42	matthias__	i dont even know what it is, but some random part of the internet told me to
16:42	raek	heh, that package is from the time before lein
16:42	matthias__	technomancy, you're phil hagelberg, right?
16:42	technomancy	yeah
16:42	technomancy	why slime-repl ever got separated out of slime.el I will never know. bunch of nonsense.
16:42	matthias__	i dont know how i knew that
16:43	raek	"When you perform the installation, you will see warnings related to the byte-compilation of the packages. This is normal; the packages will work just fine even if there are problems byte-compiling it upon installation."
16:43	matthias__	well this was an error, no warning
16:44	matthias__	huh, got a repl and it seems to work 8)
16:44	raek	matthias__: is slime-connect available anyway?
16:44	matthias__	yeah
16:44	matthias__	thanks again, raek
16:44	matthias__	btw, what happens when i run just slime?
16:44	raek	matthias__: I can tell you, it's downhill from here ;-)
16:44	matthias__	better be :D
16:45	technomancy	what was the root cause? mostly just confusion due to the apt-get version of emacs?
16:45	matthias__	oh man
16:45	matthias__	i didnt quite understand it all. but that was one of the problems
16:45	technomancy	I'm open for suggestions to make it clearer
16:45	matthias__	i didnt get a cool animationthis time though
16:46	technomancy	unfortunately I can't fix blogs with misleading info =\
16:46	raek	a combo of the slime debian package, [leiningen/lein-swank "1.3.0"] for [swank-clojure "1.3.0"], among other things
16:46	matthias__	your newest explanation is outdated. perhaps someone should make a tutorial on how to set it all up and actuallt keep is up to date
16:46	raek	...coming from outdated blogs, I think
16:46	technomancy	matthias__: which one are you refering to as the newest?
16:46	matthias__	some random one i found
16:46	matthias__	:D
16:47	raek	matthias__: this one? http://technomancy.us/126
16:48	matthias__	yeah
16:48	matthias__	its not one of the problematic ones since its clear that its outdated
16:48	technomancy	that points to the official readme at least
16:48	matthias__	the link is also outdated and the official wiki thing explains some stuff about inferior lisp mode
16:48	matthias__	which sounded wrong to me
16:49	raek	I should really make a screecast about how to do this...
16:49	technomancy	it's an alterante approach
16:49	technomancy	not invalid, just more simplistic
16:49	matthias__	yeah, but it seems there was no current one about the swank-clojure + slime approach
16:50	matthias__	that might have made it clearer 8)
16:50	raek	time to write that mail re. write access to the wiki
17:04	duncanm	la la la
17:07	amalloy	edw: i've seen him once or twice in here, but seems very rare
17:08	duncanm	i can never remember the arguments to the methods for JOptionPane
17:10	jcromartie	am I completely misunderstanding redis-clojure?
17:11	jcromartie	or is sadd broken... because doing (redis/sadd "key" "val") and then (redis/smembers "key") returns #{"3"}
17:17	jcromartie	it appears to be storing the length of the value instead of the value itself... must be a protocol mismatch
17:21	edw	amalloy: Thanks. I rooted through the source and found the answer to one of my questions.
17:22	amalloy	edw: i met him a month ago, though. i think there are lamina and/or aleph usergroups that he does follow if you want to get his attention
17:23	edw	amalloy: Thanks again. I'm curious about what sort of scalability benefits using aleph makes possible, as I've recently worked on a small but non-trivial Node.js project, and it got me getting very sick of the C-like syntax required for this very Lisp-like job.
17:28	lucian	edw: if you ever have to write JS, give CoffeeScript a try
17:28	raek	technomancy: do you think it would be resonable to present slime + swank-clojure as the primary / most common approach at Getting Started with Emacs in confluence?
17:29	edw	lucian: I feel like I'm at Hacker News right now! ;) I tried it, it's not compressing my code much, it requires too much thinking (do I NEED those parens?!) and it adds complexity to the development process.
17:30	technomancy	at least you don't have to think "did I get all my return statements in?"
17:30	lucian	edw: meh. i find the tradeoff worth it
17:30	technomancy	that's like a 50% bug reduction right there when it comes to the JS I write.
17:30	lucian	i get indentation-syntax, more expressions and nice function literals
17:30	edw	technomancy: Yeah, I hear you. And lucian, I'm not begruding anyone their coffeescript; it's just not for me.
17:31	lucian	edw: sure, never blamed you of anything
17:31	amalloy	explicit return statements are so sad
17:31	technomancy	raek: yes, but only if it's simple enough. I keep thinking it is, and people keep proving that false.
17:31	lucian	amalloy: to each his own. i think it's not the best of reasons for disliking JS :)
17:31	edw	I hate syntax. I love lisp.
17:31	raek	fair point.
17:32	amalloy	lucian: it's actually a reason for disliking php, for me :P. make a function to be an alias of another, and spend ages debugging to find i forgot the return
17:33	lucian	amalloy: i suppose that's down to familiarity. as a pythonista, i pretty much never do that
17:33	edw	And aleph, which someone on HN pointed me to, looks like it could let me write async threadless server apps in Clojure.
17:33	lucian	but there are plenty of things that are clearly wrong & stupid in both JS and especially PHP
17:34	edw	lucian: You spend much time in a lisp, where everything's an expression, you quickly forget the `return' ever existed. I stared at some of that Node.js code for ten minutes until I realized I needed a `return' in there.
17:34	lucian	edw: yeah, i realise that. i get the reverse when i write lisp, but not with returns
17:35	lucian	edw: btw, there's also ParenScript. it's a bit odd, you might hate it (i do)
17:35	edw	With pattern matching (destructuring assignment) and implicit returns and a banishment of assignment, Javascript would be so much nicer.
17:36	edw	Anything that's a prepocessor is going to be a drag.
17:36	lucian	well, still has assignments, but they're done right and less necessary
17:36	edw	lucian: CS has destructuring assignment?
17:36	lucian	edw: doesn't it? i might be wrong, but i thought it did
17:37	lucian	i treat CS as a compiler
17:37	lucian	or you can get it to compile in the browser if it's really a problem
17:38	lucian	edw: yeah, it has destr. assi.
17:38	lucian	edw: http://jashkenas.github.com/coffee-script/
17:38	lucian	works like in python or js1.7
17:39	edw	lucian: Thanks, I hadn't gotten that far down in the docs yet.
17:39	lucian	it might depend for your particular case, but i'm willing to jump some hoops to replace JS with CS
17:39	lucian	it's just that much nicer
17:39	edw	I'm using Joyent's Node.js Cloud, so the CS would interfere with the git push goodness.
17:40	lucian	well, you could run coffee -c -w and commit .coffee files
17:40	lucian	s/.coffee/.js/
17:40	sexpbot	<lucian> well, you could run.js -c -w and commit .js files
17:41	lucian	sexpbot: bad bot
17:41	brehaut	lucian: if you dont want sexpbot to do that, there is an exclude list
17:41	lucian	well, my regex is wrong
17:42	lucian	brehaut: i should get my regexen right instead :)
18:45	raek	technomancy: what is the difference between the slime, slime-repl and clojure-mode packages when comparing ELPA and Marmalade
18:45	raek	?
18:46	technomancy	raek: clojure-mode is newer at marmalade. slime and slime-repl are probably the same.
18:48	raek	is it mentioned in the swank clojure readme beacuse it gets newer version more quickly?
18:49	technomancy	raek: elpa has basically stopped accepting new versions in hope that people will migrate to marmalade
18:51	raek	technomancy: ah, so maybe it's a good idea to mention it instead of elpa in Getting Started with Emacs? (since the packages mentioned at that page are all in marmalade anyway)
18:51	technomancy	raek: oh, for sure
18:52	raek	ok, thanks!
18:54	TimMc	marmalade... never heard of it
18:55	raek	just discovered it myself :-)
18:55	technomancy	it's spreadable elisp
19:54	raek	there. today's work is done: http://dev.clojure.org/display/doc/Getting+Started+with+Emacs
19:54	raek	good night.
19:57	technomancy	raek: thanks!
19:59	technomancy	does lein repl actually work for inf-lisp?
19:59	clojurebot	the repl is holding onto the head of sequences when printing them.
20:00	technomancy	clojurebot: nonsense
20:00	clojurebot	Pardon?
20:02	znutar_	oh nice! I just started with clojure and emacs on saturday and the nebulous state of the documentation on what to actually use was a real drag
20:03	technomancy	well the old wiki docs were already halfway decent; the problem is more that people don't read them
20:04	technomancy	and that everyone put lots of google juice into the assembla pages and now they're on confluence
20:06	znutar_	Yeah, it's a pain figuring out what docs are good out of context. Otoh it's usually a sign that people are actually doing something and not just blogging.
20:07	technomancy	more like people who don't know what they're doing are the ones doing the blogging
20:10	znutar_	That's almost always going to be the case, though. If you have everything running smoothly it's rarely interesting enough to warrant telling people about.
20:10	technomancy	quite true.
20:11	technomancy	"Hey guys, check out the readme! It tells you how to use it. Shocking, I know." <= lemme go retweet that!
20:11	technomancy	</rant>
20:24	znutar_	I wonder if the whole game console-like achievements angle for something like labrepl would do a good job of illustrating the most painless way to get yourself set up in some known state for newbies
21:10	sritchie	does anyone here know how to force stdout to print to the current emacs repl buffer, with SLIME?
21:10	sritchie	cake currently directs it to .cake/cake.log, but I've seen a few screencasts where stdout shows up directly in emacs
21:59	MiggyX	Hi guys, if you need to write a program that goes through a CSV file to sequentially match rows (i.e. order is important) is there a standard way to maintain this 'state' in clojure ?
22:03	amalloy	sritchie: that's weird. it prints to the repl for me
22:04	amalloy	(and i use cake)
22:04	sritchie	amalloy: hmm, odd! I've been running cake swank at the command line, then slime-connect in emacs
22:04	dnolen	sometimes things are so easy in Clojure it seems criminal, https://gist.github.com/916831
22:05	sritchie	the solution floating around the web is to put (setq SWANK:GLOBALLY-REDIRECT-IO t) in ~/.swank.lisp, but no luck there
22:05	amalloy	sritchie: same. maybe you have an old version of one of those?
22:06	sritchie	amalloy: I just updated everything off of ELPA, so it's not that --
22:06	amalloy	errr, cake isn't in elpa, is it?
22:07	sritchie	sorry, I meant slime, etc
22:07	sritchie	cake is maybe a month old
22:07	amalloy	cake --version?
22:08	sritchie	just ran cake upgrade, so now it reads cake 0.6.2-20110221.082826-17-g4d002b1 -- let me test it now
22:11	brehaut	MiggyX: without knowing more details you probably want to look at reduce
22:12	sritchie	amalloy: no dice. normal side effects show up, and the results of function calls, just not anything that's meant to be piped to stdout
22:12	sritchie	cascalog output, or pallet output, for instance
22:13	amalloy	oh. piped? i guess i don't understand the problem
22:14	amalloy	but if i did understand it i still wouldn't have a solution, so probly not worth your effort to clarify
22:14	MiggyX	brehaut: Basically processing a huge csv file (30GB+) and matching rows together etc
22:14	MiggyX	brehaut: brb
22:16	sritchie	amalloy: no problem, thanks for taking a look. the clarification's short -- non-repl threads currently print to .cake/cake.log, so I was just looking for a way to get all threads, repl or non-repl, to send output to the repl buffer
22:18	brehaut	MiggyX: a) http://mywiki.wooledge.org/XyProblem b) thats still an extremely general description of a problem
22:21	MiggyX	brehaut: Well basically I have a CSV file that contains two feeds of data. We're talking more than 9 million rows combined. From the first feed (let's call it feed A) there should be a matching entry - or at least a similar entry - in feed B. The problem is that there are no global identifiers and so the file needs to be processed sequentially. Basically I want to match the two.
22:23	brehaut	MiggyX: how are they 'matched' ?
22:24	MiggyX	brehaut: There is a sequence number and various fields that should match up - it's unlikely that the same combination would appear within a few minutes of each other (though theoretically possible). However trying to load the file and match in one go won't work as you will have great trouble trying to isolate the correct one
22:25	brehaut	MiggyX: is this tallyman / TCRM data?
22:26	MiggyX	brehaut: forex price feed / trade
22:26	brehaut	close then. those formats are always miserable
22:27	MiggyX	brehaut: Tell me about it :/
22:27	amalloy	brehaut: i'm going to have to start prefacing all my questions with "I'm on Gentoo so I know what I'm doing"
22:27	brehaut	amalloy: golden eh :)
22:27	brehaut	via raganwald's tweeter
22:28	MiggyX	Thing is, I can't see how I can do it without maintaining some sort of state - or sticking most of the code into a single function...
22:28	brehaut	MiggyX: are you fairly new to the functional programming thing?
22:29	MiggyX	brehaut: yeah - read a lot about it, but have yet to actual do anything useful with it.
22:30	brehaut	MiggyX: ok. theres a function called reduce (sometimes called 'fold' in other languages) that takes a function, some data and a sequence. The function it takes as an argument is given items from the sequence one at a time along with some data (sometimes called the accumulator/accumulation) and it returns a new accumulation
22:31	brehaut	MiggyX: a simple example would be ##(reduce + 0 [1 2 3])
22:31	sexpbot	⟹ 6
22:31	brehaut	MiggyX: the ## just tells sexpbot to evaluate it
22:32	amalloy	btw, you can see what the reduce is doing "under the hood" with ##(reduce (partial list '+) 0 [1 2 3])
22:32	sexpbot	⟹ (+ (+ (+ 0 1) 2) 3)
22:32	brehaut	amalloy: ah yes good example
22:32	MiggyX	okay, with you so far :)
22:32	brehaut	MiggyX: so the data passed to reduce is the seed value for the accumulator, if its not present clojure will use the first argument in the sequence
22:33	amalloy	brehaut: you missed my stroke of genius to explain the difference between apply and reduce: ##((juxt apply reduce) (partial list '+) (range 4))
22:33	sexpbot	⟹ [(+ 0 1 2 3) (+ (+ (+ 0 1) 2) 3)]
22:33	TimMc	We don't have foldl and foldr, do we?
22:33	brehaut	amalloy: nice :)
22:34	brehaut	TimMc: nope
22:34	amalloy	TimMc: no
22:34	brehaut	MiggyX: so in your case you might have something like (reduce match-csv (match-acc) csv-data)
22:35	brehaut	where match-csv is a function that operates record at a time on csv-data and (match-acc) is a function that returns an empty accumulator
22:36	MiggyX	can the csv-data see the contents of match-acc ?
22:36	brehaut	no
22:36	brehaut	csv-data needs to be a sequence of csv records from your csv file
22:37	MiggyX	if I understand correctly, each row (assuming I set it up that way) will be passed to the match-csv function which will be called uniquely (i.e no side effects) for each row.
22:37	brehaut	also because im a muppet, it should probably just be empty-match-acc or something; no need for it to be a function call
22:37	MiggyX	From that I'm not sure how I could match saw row one with row ten
22:38	brehaut	MiggyX: match-csv can return anything it likes. whatever it returns is used as the accumulator argument for the next call to match-csv
22:38	brehaut	MiggyX: think of it as simple recursive over a list but without you needing write the recursive logic and tests
22:39	MiggyX	okay, so match-csv is called with the result of the accumulator and the next row from the file?
22:39	brehaut	the result _is_ the accumulator
22:40	amalloy	brehaut: haskell teaches us that everything should be a function call
22:40	TimMc	Oh, that's right... nullary functions for constants.
22:40	brehaut	amalloy: this is indeed what tripped me up
22:40	amalloy	haha
22:41	amalloy	i didn't even know that, TimMc
22:41	amalloy	i just figure, if brehaut says something crazy, blame haskell
22:41	TimMc	Well, don't take it as gospel.
22:42	brehaut	amalloy: if you cant have any side effects due to the type system, and the funcall syntax doesnt require parens, then there is no difference between a function of no argumnts and a constant value
22:42	TimMc	brehaut: Can you confirm my assertion?
22:42	TimMc	heh
22:42	brehaut	TimMc: yes. although its probably elided back to just constant values by the compiler
22:43	TimMc	brehaut: Also, where do you live? I'm wondering about your use of the word "muppet".
22:43	brehaut	TimMc: New Zealand
22:43	TimMc	It's a Brtish thing, yeah?
22:43	mec	amalloy: do you have an example of using your lazy-recur in a non tail position?
22:43	brehaut	TimMc: probably
22:44	dnolen	hmm are strings reliable unique identifiers for classes on JVM? I wondering the what's the most stable thing considering redef'ing types/records is common.
22:44	amalloy	mec: https://github.com/amalloy/amalloy-utils/blob/master/src/amalloy/utils/core.clj, but pretty much every use of lazy-recur isn't in tail position
22:44	amalloy	dnolen: no
22:45	amalloy	two classes with the same package and name can be loaded by different classloaders and not conflict
22:45	mec	amalloy: err right
22:45	TimMc	amalloy: "not conflict" is an interesting phrase
22:46	TimMc	It can be problematic if you accidentally try to cast one to the other.
22:46	amalloy	TimMc: feel free to propose more-accurate statement
22:46	TimMc	nuh-uh
22:46	MiggyX	stupid question - I don't suppose reduce runs in paralllel :)
22:46	brehaut	MiggyX: nope
22:46	TimMc	MiggyX: preduce
22:46	dnolen	amalloy: huh ... how does the JVM differentiate them? If I load some other jar that references one of those classes, which will get instantiated ?
22:46	TimMc	Wait, how would that even work?
22:46	MiggyX	TimMC: I assume that would screw up the accumulator though
22:47	amalloy	dnolen: suppose class A1 and A2 are loaded by classloader L1 and L2
22:47	amalloy	if someone in class B refers to class A, they get the class named A which was loaded by the same classloader as B
22:47	amalloy	iirc
22:49	TimMc	MiggyX: There's something called preduce, but I think it parallelizes operations on different key-value pairs of maps.
22:49	dnolen	amalloy: if classes can't cross class loader boundaries then how can using string as an identifier be problem?
22:50	brehaut	MiggyX: thres no such thing as magic parallization; you need to know your domain well enough to device reasonable partitioning, operations on those partitions, and reconstitutions for instance
22:50	MiggyX	TimMc: I always seem to pick hard tasks :)
22:50	amalloy	dnolen: you can find references to other classloaders, and ask them to load a class for you, or return it if it already exists
22:50	brehaut	MiggyX: and your strong ordering requirement may cause you a lot of grief
22:50	amalloy	MiggyX: you pick vaguely-defined tasks
22:50	amalloy	all such tasks are hard :)
22:51	amalloy	http://download.oracle.com/javase/1.4.2/docs/api/java/lang/ClassLoader.html#getSystemClassLoader()
22:51	MiggyX	brehaut: If I was doing it in say Python, I would split the file into say 1GB chunks and then start the matching process. any rows that aren't match are passed back, put in an ordered list and then processed as a final job. I think that should work
22:51	MiggyX	but I was hoping to have a good excuse to spend a few days doing it in clojure :D
22:51	brehaut	MiggyX: you would do the same in clojure then
22:52	phenom_	hey, in my code I create a lot of vectors which i pass around to single argument functions which in turn deconstruct the vector ... YourKit is showing me that alot of my space is taken up by clojure.lang.PersistentVectors ... any idea why?
22:52	phenom_	shouldn't they be garbage collected? I dont hold on to them at all
22:53	brehaut	MiggyX: but you will have a potentially very large post-parallel bottleneck which is going to seriously impede your performance gains
22:53	brehaut	phenom_: the collector runs in its own time
22:53	amalloy	phenom_: they get gced eventually, but in the meantime you're creating a lot of them. what else can you expect but that they take up space meanwhile?
22:54	MiggyX	brehaut: I wouldn't expect it to be that bad. If I slice the CSV file, only a few rows would be affected either side of it... there shouldn't be much left to process after the main chunks I'd have thought
22:54	phenom_	amalloy, brehaut: but i eventually get an OOM exception :S with -Xmx1G
22:54	amalloy	then something is wrong; fix it :P
22:55	TimMc	phenom_: Are you holding onto the head of seq you shouldn't be?