#clojure – 2009-10-27

2009-10-27

00:01	hiredman	http://pixel-shaker.fr/en/2009/09/representation-dimages-clojure-et-java-premiere-partie/
00:03	somnium	how do you specify multiple constructors? variable arity, variable type
00:03	somnium	or different, not variadic
00:04	hiredman	it turns out I missed redirecting a call to LispReader/read in LispReader/readDelimitedList so none of my incremental developed and tested code was ever really tested
00:04	somnium	but this is good, thans
00:04	hiredman	erm
00:04	hiredman	have you read the api docs?
00:04	somnium	yeah
00:05	hiredman	there is a :constructors keyword
00:05	somnium	yes
00:05	hiredman	and that didn't jump out as a way to implement constructors?
00:07	somnium	ostensibly yes, it did connect that :constructors and the vectors that followed were linked in a meaniful way
00:08	somnium	but it didn't jump out how to do [Object Object String] [] [Object String Interface] [] [Object ...]
00:08	somnium	http://en.wikibooks.org/wiki/Clojure_Programming/Examples/API_Examples/Java_Interaction#genclass
00:08	somnium	and this didn't help
00:09	somnium	though it did compile in slime, only to blow up in ant
00:09	hiredman	http://clojure.org/compilation
00:11	somnium	note the most complicated constructors invocation in the examples
00:15	hiredman	how much more do you want the constructor to do?
00:15	somnium	so much more man, so much more
00:16	hiredman	~laugh
00:16	clojurebot	ha ha
00:17	Licenser	somnium: does your driver support sorting?
00:17	somnium	I like clojure to be clojure and java to be the jvm, and the nasty 25 classes in one package interface didn't map well to the examples I found
00:17	somnium	Licenser: not yet, it seems it has be done with the cursor
00:17	Licenser	hmm there is a sort function at least in the console client
00:18	somnium	if the javadocs there's a skip and limit function defined for find
00:18	somnium	but its not found when I inspect the class
00:19	somnium	I'll wrap the cursor once I get this proxy business sorted, then skip, limit, sort etc. should be easy
00:19	somnium	try doing (clojure.contrib.repl-utils/show (.find some-collection))
00:19	hiredman	,(set (map #(.getName %) (.getMethods String)))
00:19	clojurebot	#{"compareTo" "equalsIgnoreCase" "subSequence" "trim" "codePointBefore" "compareToIgnoreCase" "getClass" "isEmpty" "length" "getChars" "toCharArray" "valueOf" "charAt" "indexOf" "notify" "toString" "codePointCount" "replaceAll" "substring" "intern" "matches" "copyValueOf" "lastIndexOf" "replace" "concat" "wait" "getBytes" "toLowerCase" "codePointAt" "startsWith" "toUpperCase" "format" "contentEquals" "notifyAll" "offsetBy
00:24	hiredman	my reader is dog slow
00:27	somnium	can't you just copy the one from SICP?
00:28	hiredman	I am copying the one from LispReader.java
00:28	Licenser	well I just self sort ^^
00:28	somnium	does he do reader macros?
00:28	hiredman	I don't type hint anything
00:28	hiredman	yeah, it's all reader macros
00:28	hiredman	just not user extendable
00:29	somnium	has there been some opposition to a heredoc or a """ ?
00:29	somnium	it seems to keep resurfacing but never gets anywhere
00:29	hiredman	there is a read table, if a character like '(' is in the read table control is passed to the matching reader macro
00:30	somnium	I got close to doing a string rearrangment of parentheses to make clojure like ruby
00:30	somnium	it always screwed up when the nesting got deep
00:30	hiredman	:(
00:30	hiredman	it?
00:31	somnium	it was more like, I wonder how hard this would be...
00:31	somnium	macros written with it even worked for simple things
00:32	somnium	it might attract some people, just to show flexible it can be without even reader macros, I was thinking
00:32	somnium	that was before clojure.walk, maybe I'll try again
00:33	somnium	I think I screwed myself with this proxy class, it won't let me override a superclass method
00:41	scottj	Is there a shortcut for doing something other than first argument with -> or is (-> "string" (#(str "prepend" %))) the way to do it?
00:43	hiredman	,(doc ->>)
00:43	clojurebot	"([x form] [x form & more]); Threads the expr through the forms. Inserts x as the last 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 last item in second form, etc."
00:43	hiredman	,(->> "foo" (format "prepend %s"))
00:43	clojurebot	"prepend foo"
00:48	scottj	hiredman: thanks
00:52	piccolino	Is it bad style to put a question mark at the end of a boolean variable? Is that for functions only?
00:52	hiredman	I do that
01:18	piccolino	Is there a way to write a linefeed character literal? I can't do \linefeed nor \lf, and gooling is coming up short for me...
01:29	piccolino	Nevermind, found char.
01:37	Licenser	(doc defn)
01:37	clojurebot	"([name doc-string? attr-map? [params] body] [name doc-string? attr-map? ([params] body) + attr-map?]); Same as (def name (fn [params* ] exprs)) or (def name (fn ([params ] exprs*)+)) with any doc-string or attrs added to the var metadata"
01:37	Licenser	,(doc defn-)
01:37	clojurebot	"([name & decls]); same as defn, yielding non-public def"
01:51	hiredman	,\newline
01:51	clojurebot	\newline
01:52	hiredman	,\return
01:52	clojurebot	\return
01:52	hiredman	,(str \newline)
01:52	clojurebot	"\n"
01:52	hiredman	piccolino:
01:52	Raynes	hiredman: Don't you use Vim for Clojure development? (or do I just need to sleep more often?)
01:53	hiredman	I do
01:53	Raynes	I've decided to move away from Emacs for a while to see what all the fuss is about Vim.
01:54	Raynes	I've been using it all day. I think I might keep using it.
01:54	Raynes	It's a little easier to remember individual commands than it is to remember key patterns. ._.
01:55	hiredman	a vim habit is hard to break, I often hit <esc>:wq while writing emails in gmail
01:56	Licenser	hiredman: I have a few word documents with :wq lines :P
01:56	Raynes	:>
01:56	Raynes	Once I figured out how to split panes in different directions and switch through buffers, I was a-go.
01:57	hiredman	I've noticed myself trying to use screen window switching commands in place of alt-tab
01:58	Licenser	yap that is bad
01:58	Licenser	evern worst when you switch between mac and PC
01:58	hiredman	and irssi's channel switching stuff in place of screen's window switching
01:58	Licenser	especially the german keyboard
01:58	Raynes	I'm using GVim until I get used to Vim.
01:58	Licenser	the @ sign when typed wrongly will do horrible things
01:59	Raynes	GVim makes a Emacsist feel at home.
01:59	Licenser	if I use the widows key combo for @ on os X it will close a window, if I use the os x key combo on windows it will log me out ...
01:59	hiredman	Raynes: I have to warn you, vimscript is horrible
01:59	Raynes	hiredman: Huh?
01:59	hiredman	vim's scripting language
02:00	Raynes	I never did any custom Emacs scripting anyways.
02:00	Raynes	I just googled what I needed if I needed something.
02:01	Licenser	night Raynes
02:20	Licenser	what is the reason that there is no !=?
02:20	Licenser	so everyone can has fun with cons?
02:23	arsatiki	huh?
02:23	arsatiki	Are you looking for not= perhaps?
02:24	Licenser	,(doc not=)
02:24	clojurebot	"([x] [x y] [x y & more]); Same as (not (= obj1 obj2))"
02:24	Licenser	yap I'm looking for that :P
02:28	bradfordstephens	Whoa, there's people in here!
02:30	Licenser	bradfordstephens: and most of them even are helpfull is my experience :) clojure has a great community
02:30	bradfordstephens	I'm working through Paul Graham's book on Lisp now...my friend is one of "Those People" who kept insisting. So I gave in ;)
02:31	Licenser	heh
02:31	Licenser	for me Clojure was the reason to bother with LISP
02:37	kylesmith	Are there any good genetic algorithm libraries in clojure? The first few hits on google don't look very appealing to me.
02:39	seangrove	require 'rubygems'
02:39	seangrove	require 'sqlite3'
02:39	seangrove	require 'activerecord'
02:39	seangrove	# connect to database. This will create one if it doesn't exist
02:40	seangrove	MY_DB_NAME = ".my.db"
02:40	seangrove	MY_DB = SQLite3::Database.new(MY_DB_NAME)
02:40	seangrove	# get active record set up
02:40	Licenser	seangrove: ?
02:40	seangrove	ActiveRecord::Base.establish_connection(:adapter => 'sqlite3', :database => MY_DB_NAME)
02:40	seangrove	# create your AR class
02:40	seangrove	class Update < ActiveRecord::Base
02:40	seangrove	end
02:40	seangrove	# do a quick pseudo migration. This should only get executed on the first run
02:40	seangrove	if !Update.table_exists?
02:40	seangrove	ActiveRecord::Base.connection.create_table(:updates) do \|t\|
02:40	seangrove	t.column :account_name, :string
02:40	seangrove	t.column :last_update_time, :timestamp
02:40	seangrove	t.column :last_update_id, :integer
02:40	seangrove	end
02:40	seangrove	Sorry guys, emacs is going a bit wacky
02:40	seangrove	I meant to paste that into a different buffer, but emacs reconnected to irc and started popping up irc channels
02:41	Licenser	:P
02:41	Licenser	def my_companies_secret_code ...
02:41	seangrove	Hahaha
02:42	Licenser	hrm I hate the backtrace if it leaves out some function :/
02:44	Licenser	uh
03:23	wooby	anyone a vimclojure user?
03:34	vy	What's the prefered way to swap two elements of a seq, particularly a vec?
03:38	arsatiki	Any two items or e.g. first two?
04:02	Fossi	hi
04:13	vy	arsatiki: Any two items.
04:22	vy	Why Clojure complains that "clojure.lang.LazySeq cannot be cast to clojure.lang.IFn" for (let [{v :v} {:v (take 20 (iterate inc 0))}] (v 10))?
04:26	jdz	vy: because LazySeq does not implement IFn?
04:29	vy	jdz: What should I do to make (v 10) work?
04:29	jdz	vy: either use a vector or nth
04:29	piccolino	Is there some function I can call to allow multiple forms to be evaluated and return the last one?
04:30	piccolino	Like progn?
04:32	jdz	piccolino: do
04:33	piccolino	Ah, there we go. Thank you.
04:38	wavis_	is there a way to do something like this:
04:38	wavis_	(let [oldns ns throw-away (clojure.lang.Namespace.)] (in-ns throw-away) (foo) (in-ns oldns))
04:38	wavis_	IE, avoid global state by sandboxing functions
04:39	jdz	wavis_: like (binding [ns (new clojure.lang.Namespace)] (foo))?
04:39	wavis_	binding requires declaring vars in the namespace first
04:39	wavis_	ohh sorry...
04:40	wavis_	i didn't know I could do that
04:41	wavis_	right... so constructing a namespace is still a problem
04:41	jdz	wavis_: how about create-ns
04:41	jdz	,(doc create-ns)
04:41	clojurebot	"([sym]); Create a new namespace named by the symbol if one doesn't already exist, returns it or the already-existing namespace of the same name."
04:42	wavis_	doesn't that leave the namespace lying around though, still resolvable?
04:43	jdz	wavis_: well, you have try/catch and remove-ns
04:43	jdz	wavis_: you'd actually want try/finally
04:45	wavis_	sounds kinda do-able, but the namespace would be global until removed, possibly creating a concurrency problem
04:45	jdz	,(doc gensym)
04:45	clojurebot	"([] [prefix-string]); Returns a new symbol with a unique name. If a prefix string is supplied, the name is prefix# where # is some unique number. If prefix is not supplied, the prefix is 'G__'."
04:46	wavis_	good call
04:46	jdz	wavis_: or go and look at clojurebot source
04:46	jdz	it definitely has a sandboxing in there
04:46	wavis_	where is that?
04:47	jdz	like, google clojurebot?
04:47	wavis_	ok jdz. thanks
04:48	jdz	,(+ 2 2)
04:48	clojurebot	4
04:48	jdz	(+ 2 2)
04:48	clojurebot	4
04:48	jdz	hmm
04:48	jdz	nvm
05:22	adityo	hey nipra :)
05:22	nipra	adityo, Hello!
05:23	nipra	adityo, wass up :-)
05:24	G0SUB	adityo, nipra
05:25	nipra	G0SUB, Hello!
05:25	AWizzArd	omg
05:25	AWizzArd	;)
05:26	adityo	~max people
05:26	clojurebot	max people is 185
05:27	AWizzArd	wtf, it was 191 yesterday
05:28	AWizzArd	clojurebot: max people is 191
05:28	clojurebot	Ack. Ack.
05:28	jdz	clojurebot: max people is 255
05:28	clojurebot	c'est bon!
05:28	jdz	make AWizzArd happy
05:28	AWizzArd	no, I'm going to cry, I want the real numbers
05:29	jdz	no, you're cheating
05:29	jdz	:)
05:29	Chousuke	clojurebot: forget max people
05:29	clojurebot	I forgot max people
05:29	Chousuke	~max people
05:29	clojurebot	max people is 194
05:29	Chousuke	I think it's a special function that just looks like a factoid. :)
05:29	Chousuke	and the definitions were overriding it
05:30	AWizzArd	I need to get my own bot in here.
05:32	AWizzArd	I want to access my server S via https. On S a Jetty is running with a certificate that I created myself. Any ideas how I can make my jvm not throwing a SSLHandshakeException?
06:19	Jetien	hi! i have a general question. when i want to maintain a 2d state array, what is the better choice: 1. a vector of refs to vectors or 2. a vector of vectors of refs ?
06:21	Jetien	or should i use some java data type?
06:21	Chousuke	are the states all independent of each other?
06:21	Chousuke	because the easiest option would be just to have a ref to a vector of vectors
06:25	Chousuke	a ref containing refs makes it difficult to have a consistent snapshot
06:26	Jetien	yeah they are very independent of each other
06:27	Chousuke	the vector of refs approach can be fine if the refs or their values don't depend on each other.
06:28	Jetien	hm what do you mean by depend? in my case there are just integers
06:28	Jetien	s/there/they're
06:28	Chousuke	as in, to determine the value of one ref you need to read the others as well
06:28	Chousuke	in general, you should keep the granularity low at first and increase it if it becomes a bottleneck.
06:28	Jetien	ah i see, no not the case
06:29	Chousuke	coordinating many refs at once is difficult, and a vector of vectors of refs is really little better than a mutable 2d vector :/
06:29	Jetien	how does this approach perform anyway? is it not better just to use some java data type?
06:30	Chousuke	depends on what you need, I guess.
06:30	Jetien	hm huge in size like 1000x1000 with a lot of changes
06:30	Chousuke	with java data types Clojure can offer no assistance
06:31	Chousuke	hmm.
06:31	Chousuke	and it'll contain integers?
06:31	Jetien	but only one thread
06:31	Jetien	yes
06:31	Chousuke	then I suppose having a million refs would be unacceptable overhead, memory wise.
06:32	Jetien	and time wise?
06:32	Chousuke	probably.
06:32	Jetien	hm okay
06:33	Chousuke	and if it's single-threaded there won't be much concurrent access to protect against.
06:34	Chousuke	so if you just use an array you'll only have to be careful not to depend on any previous value of the matrix
06:34	Jetien	i always wondered how a ref to vector performs if there are a lot of changes.. does the garbage collector clean up immediately?
06:34	Chousuke	it should be fine.
06:34	Chousuke	if your changes are done as a batch you can also use transients
06:35	Jetien	actually i never used concorrent programming ;) i'm all in for the lisp + java thing
06:35	Jetien	let me google "transients"...
06:36	Chousuke	the array will likely perform best, but you will also not get all of the advantages immutability
06:36	Chousuke	~transients
06:36	clojurebot	http://clojure.org/transients
06:37	Chousuke	basically a local mutable version of a persistent collection for making a lot of changes, with a fast operation for transforming it back into an immutable one again.
06:40	Jetien	ah that's nice...
06:40	Jetien	i think that's what i'm looking for
06:41	wooby	Jetien: http://clj-me.cgrand.net/2009/10/15/multidim-arrays/
06:41	Jetien	thanks
06:41	wooby	no problem, i was trying to figure this out myself the other day
08:27	vy	What's the best way to model a matrix in a functional style in Clojure? Using a 1D vectors and mapping 2D coordinates to 1D coordinates via abstraction functions?
08:29	wooby	vy: i've done it that way, here's another: http://clj-me.cgrand.net/2009/10/15/multidim-arrays/
08:29	wooby	as to what's best, i don't know enough to know ;)
08:30	yason	vy: how large matrices? for what purpose?
08:32	yason	vy: if you're doing simple 2D/3D stuff with your matrices, you could just use a vec of suitable length (like 6 entries for 2 x 3 matrix in 2D space) and use destructuring in defning the basic matrix ops since you pretty much know what you're going to add/multiply
08:33	vy	yason: I'll issue quite much col/row swap operations on a matrix collection of size ~100 where matrices are of size ~1000x1000.
08:33	yason	vy: that's a bit too much for destructuring :)
08:36	yason	vy: [Disclaimer: I'm a beginner] You probably want speed too so you could consider make-array even though it's mutable. The functional approach would probably be to make the high-level functions pure -- even though you did mutations inside some matrix operation
08:37	octe	when running emacs slime with clojure i get "Lisp connection closed unexpectedly: connection broken by remote peer
08:37	octe	"
08:37	octe	used to work..
08:37	octe	anyone have any clue what it could be?
08:38	yason	octe: slime/swank crashed?
08:38	yason	octe: sync to the latest sources and see if it reproes
08:39	octe	yason: i sycned slime and swank-clojure to the latest
08:39	octe	should i sync clojure too?
08:52	vy	In a `struct-map' call, Clojure complains that "clojure.lang.PersistentStructMap cannot be cast to clojure.lang.PersistentStructMap$Def", what could be causing this error?
08:56	arsatiki	vy: please post the code to a pastebin
08:59	chouser	vy: you need (struct-map foo ...) where foo is the name you gave in a (defstruct foo ...)
09:05	vy	arsatiki: chouser: Thanks. I think I figured out the problem.
09:25	yason	octe: IIRC there's a different swank/slime for Clojure 1.0 and trunk
09:25	yason	octe: but if you're using the latest Clojure, then you probably should have latest slime/swank
09:53	chouser	huh. Proposal: Moratorium on Python language changes: http://lwn.net/Articles/357996/rss
09:54	chouser	"frequent changes to the language cause pain for implementors of alternate implementations"
10:03	cemerick	yeah, I saw that. The writing was on the wall years ago, tho.
10:03	chouser	90% of the comments on lwn are negative.
10:05	cemerick	the ISO standardization notion at the end is hilarious
10:05	spuz	I don't understand that article. If you're changing python every month it doesn't imply that you need to change Jython every month does it?
10:06	chouser	I think it's interesting to consider Clojure's related position in light of cinc.
10:07	rhickey	chouser: in what way?
10:07	cemerick	spuz: I think that's a red herring. The language has been largely frozen for years, it's just been window dressing for some time.
10:07	spuz	chouser: why, how much is cinc going to affect users?
10:07	chouser	once cinc is done, the amount of work to create a Clojure host-port will be pretty small compared to the amount of work to create Jython
10:07	spuz	I see
10:08	chouser	and with that small creation work comes smaller maintenance work. It should be rather less effort to keep the ports in sync.
10:08	rhickey	the biggest difference is that a Lisp has so little in the language/compiler proper - most is libraries
10:09	chouser	many changes that would count as "grammar or language semantics" won't require any porting work at all
10:09	notallama	shouldn't pypy do that, too? you only need to implement the part for pypy requires.
10:09	chouser	right, because they're implemented in the core "library"
10:10	chouser	but even changes on the scale of "defclass" which are likely to be much rarer will still be largely discreet pieces and comparatively little effort, I think. Well, I hope anyway.
10:10	djork	indeed, it's pretty hard to make a "breaking change" to the syntax of a Lisp
10:10	notallama	i thought that was the entire point of pypy, actually: to make porting python easier.
10:11	cemerick	I thought it was an optimization and 'alternative feature' (e.g. continuations) playground.
10:11	djork	although Clojure's reader is more complex than any other Lisp I've seen
10:12	notallama	complex meaning it has 3 different kinds of brackets?
10:12	djork	well yeah, that and a lot of the reader dispatch macros
10:14	djork	() [] {} #{} ' ^ @ # #"" #' #_ #() #^ ^# #^{} ` ~ ~@
10:14	djork	but the read table is a nice implementation
10:15	djork	I wish I could extend it
10:16	djork	but maybe not :)
10:16	chouser	djork: you're not the only one to express interest. :-)
10:17	djork	I guess I shouldn't trust myself with the ability to add reader macros if I can't trust other people
10:21	arsatiki	djork: what does ^# do?
10:21	djork	oops maybe not that one
10:22	chouser	#^ is pretty much the same as #^{}
10:22	tmountain	is there a summary table online somewhere with all the reader macros in it?
10:23	chouser	tmountain: http://clojure.org/reader
10:24	notallama	how would user reader macros work with namespaces? presumably you'd want a way to use a namespace minus a reader macro.
10:25	chouser	notallama: that's similar to rhickey's main objection
10:26	chouser	I had a proposal to solve the namespace issue and address some percentage of reader macro use-cases, but I think it was seen as sort of a "worst of each" compromise.
10:28	djork	I think it could cause maintainability and readability issues to let people add custom reader macros. Regular macros are powerful enough.
10:28	djork	it starts to look like Perl with too many of them
10:31	ambient	i wouldn't necessarily have a problem with encapsulated reader macroes that implemented a DSL
10:31	ambient	meaning, I would find that thing probably very useful
10:31	djork	probably
10:31	chouser	think xpath or a custom regex implementation
10:31	djork	granted pretty much anybody that is working with Clojure code is working on it by themseves
10:32	djork	or how about XML-in-Clojure
10:32	djork	like, literally XML
10:32	cemerick	oh, god, no
10:32	djork	:)
10:32	djork	like Scala
10:32	cemerick	seriously, no.
10:32	djork	I rest my case
10:32	vy	What's wrong with (defn distinct-rand-int-seq [n m] (first (remove #(apply = %) (repeat #(repeat m (rand-int n)))))) that Clojure complains that "Don't know how to create ISeq from: foo$rand_int_seq__2939$fn__2944"?
10:32	ambient	array slicing operations would be nice with reader macroes afaict, as little as I know about them
10:33	chouser	cemerick: but weren't you just calling for literal Java? you could do that with a reader hook
10:33	djork	inline Java?
10:33	djork	interesting idea
10:33	chouser	ambient: I don't think a reader macro helps much there -- maybe slightly more terse than (slice x 4 5)
10:34	cemerick	chouser: Big difference between simplifying integration with the host language and choosing to elevate a particular data representation to a first-class entity.
10:34	chouser	cemerick: right, I'm just saying a reader hook allows both
10:35	cemerick	java-in-parens would surely require a reader mod, but I don't think that's an argument for a generalized hook.
10:35	chouser	vy: I'm not sure repeat does what you think it does.
10:35	djork	yeah
10:36	ambient	chouser i should look more into lisp implementation of Matlab or Numpy style array processing, but I don't believe that it would be so readable nor concise
10:36	djork	,(repeat 10 (rand-int 10))
10:36	clojurebot	(2 2 2 2 2 2 2 2 2 2)
10:36	djork	,(repeat 10 "hi")
10:36	clojurebot	("hi" "hi" "hi" "hi" "hi" "hi" "hi" "hi" "hi" "hi")
10:36	vy	hrm...
10:36	chouser	,(repeat 3 #())
10:36	clojurebot	(#<sandbox$eval__3820$fn__3822 sandbox$eval__3820$fn__3822@1558d62> #<sandbox$eval__3820$fn__3822 sandbox$eval__3820$fn__3822@1558d62> #<sandbox$eval__3820$fn__3822 sandbox$eval__3820$fn__3822@1558d62>)
10:36	djork	hah
10:36	djork	,(take 10 (repeatedly #(rand-int 10)))
10:36	clojurebot	(2 5 9 0 4 0 3 3 9 6)
10:39	octe	ugh.. my slime setup still gives me "Lisp connection closed unexpectedly: connection broken by remote peer"
10:39	octe	run latest of.. everything
10:39	octe	anyone have any idea how to debug it?
10:42	vy	what does inferior lisp buffer tell?
10:43	octe	http://pastebin.ca/1645194
10:44	octe	that's all
10:44	patricius_	would Clojure ever be able to be as 'efficient' as Erlang in terms of "process spawning"?
10:44	octe	http://pastebin.ca/1645195 <- slime-events
10:44	patricius_	I'm asking since Clojure uses 'real threads' as opposed to Erlangs 'green threads'
10:45	vy	octe: M-x slime-abort-connection M-- M-x slime clojure (when asks for an extra inferior lisp, say no.)
10:45	djork	patricius_: I think Clojure and Erlang have different priorities
10:46	djork	patricius_: Clojure is better at balancing computation on multiple cores, Erlang is better at distributing computation across multiple servers
10:46	patricius_	djork: clojure is for parallelism and erlang for massive concurrency?
10:46	ambient	patricius_ the way I understood it, agents use 'green threads', but I haven't really studied or used them in depth
10:46	octe	patricius_: that seems to work
10:47	octe	err
10:47	octe	vy: that seems to work
10:47	octe	doesn'
10:47	octe	t hnmm
10:47	octe	says Connected. Hack and be merry!
10:47	octe	but doesn't give me a repl buffer
10:47	octe	and sending soething to it gives an error
10:48	octe	error in process filter: slime-dispatch-event: Elisp destructure-case failed: (:write-string "hi
10:48	octe	")
10:48	octe	error in process filter: Elisp destructure-case failed: (:write-string "hi
10:48	octe	")
10:48	octe	nil
10:49	lisppaste8	rhickey pasted "playing with defclass" at http://paste.lisp.org/display/89364
10:51	rhickey	still todo: Seqable, Counted, toString, Associative...
10:52	vy	octe: See http://paste.lisp.org/display/89365
11:03	djork	rhickey: how much work for defclass is "under the hood?"
11:03	djork	looks like IObj and ILookup?
11:04	chouser	those are just interfaces
11:04	djork	oh right
11:05	rhickey	djork: not sure about your question, defclass and deftype build on defclass* which is a generic exposure of code generation capability that had been locked inside fn
11:05	djork	oh ok
11:05	octe	vy: still doesnt work, but now i get some errors when doing the connection with M- prefix
11:05	octe	http://paste.lisp.org/display/89366
11:06	chouser	defclass* doesn't do closures though, right? so fn will still need compiler help?
11:07	ambient	to implement a repl in a swing window would i just have to read, eval, loop? :p
11:07	rhickey	chouser: there is a minor partitioning around closure, defclass* roots the environment at the field set, fn and newnew don't
11:07	rhickey	newnew is still there and shares all of the benefits, like bridge generation
11:07	chouser	rhickey: ok. can they share gen code?
11:07	rhickey	oh yes
11:07	chouser	ah, fantastic.
11:08	chouser	ambient: yep. it's been done a couple times.
11:08	ambient	cool, thanks. i just thought that I was being blindsighted by something obvious :)
11:08	rhickey	when/if I expose volatile again it will likely be for defclass* only
11:09	chouser	this will be such a clean bytecode-gen api, I wonder if other languages will use it.
11:09	chouser	implement my-new-toy-language in clojure instead of java+asm
11:09	ambient	i'd think Forth was better for that ;)
11:27	gerry_	hello
11:28	Licenser	aloa
11:31	gerry_	i'm looking at hickey's paste code, (:d x 42) used for what?
11:33	djork	what's the best way to reload a lib from a repl?
11:34	djork	oh :reload
11:37	chouser	gerry_: that's a ILookup by keyword :d, with 42 used as the "not found" value.
11:37	gerry_	it seems not work
11:37	chouser	I haven't tried it. what does it return?
11:38	gerry_	nil
11:38	chouser	hm.
11:39	gerry_	(:d x 42) return 42, but (:d x) return nil
11:40	chouser	oh, that's correct
11:40	chouser	:d doesn't exist in x, so (:d x) should return nil and (:d x foo) should return foo
11:41	gerry_	(defclass one [#^int a #^String b])
11:41	gerry_	i got " Can't type hint a primitive local " error msg
11:43	gerry_	chouser: you are right
11:43	headius	that's the defclass syntax?
11:43	headius	what is #^ doing there?
11:44	chouser	headius: defclass barely exists yet. http://paste.lisp.org/display/89364
11:45	chouser	headius: the #^int a attaches metadata {:tag 'int} to the symbol a, which defclass then interprets as the type it should make field a to be.
11:46	headius	I see
11:46	headius	is that a clojurism or a lisp thing (the #^ syntax)?
11:47	chouser	headius: I believe #^ is not in other lisps
11:50	gerry_	rhikey's paste code works here, good job, congr richkey!
11:51	stuartsierra	#^ is a Clojure reader macro to attach metadata to the following symbol or data structure
11:51	headius	ok
11:54	Licenser	muhaha I forked my first github project!
11:56	djork	congrasturbations
11:56	Licenser	thank you djork and it's even in clojure :P
11:57	djork	did you fork your own project or someone else's? :)
11:57	Licenser	I forked somnium mongodb driver
11:57	Licenser	because I spend the day in airplains and took the time to fix some problems I had with it
11:57	Licenser	or started to fix :P
11:57	Licenser	so I figured why not try out how it works
12:53	yuras	i would like to change behavior of function accept-fn of contrib/server-socket. what is the idiomatic way to make it if i dont want to copy&paste whole server-socket server? is there a way?
12:55	Chousuke	~def server-socket
12:55	clojurebot	excusez-moi
12:55	Chousuke	;(
12:55	yuras	~def create-server
12:57	Chousuke	hmmh
12:58	yuras	basically i want to make it passing socket object into function, i know i can rewrite it completely, but thinking about the better way. the thing that i need to know ip address of remote socket
12:59	Chousuke	I don't think there's a better way
12:59	fabe	hi beginners question: is their a sum function like in python ?
12:59	Chousuke	fabe: it's + :)
12:59	Chousuke	,(apply + [1 2 3 4])
12:59	clojurebot	10
12:59	fabe	(+ coll) ?
12:59	ambient	,(reduce + [1 2 3])
12:59	clojurebot	6
12:59	Chousuke	or that
12:59	fabe	k thx
13:00	arsatiki	is there a significant performance difference either way?
13:00	arsatiki	oh wait, I have a repl. I can check for myself.
13:00	Chousuke	well, + uses reduce internally
13:00	ambient	such micro-optimazion likely has no value
13:00	Chousuke	yeah
13:01	Chousuke	often, apply is better though.
13:01	Chousuke	for example, (apply str '[foo bar whatever]) is faster than (reduce str '[foo bar whatever])
13:03	arsatiki	yeah. no value whatsover =)
13:03	Chousuke	str uses a stringbuilder internally in the apply case, but using reduce will force it to create temporary strings
13:03	tmountain	on my box, reduce is faster in this case (not that it matters much)
13:03	tmountain	(time (reduce + (range 1 10000000))) "Elapsed time: 804.801736 msecs"
13:03	tmountain	(time (apply + (range 1 10000000))) "Elapsed time: 3275.64601 msecs"
13:04	hiredman	that is +
13:04	Chousuke	hmm, that might actually be because of chunked seqs
13:04	hiredman	Chousuke is talking about str
13:04	Chousuke	perhaps using apply will lose the chunkiness information of range, so the reduce in + can't take advantage of it
13:05	tmountain	I need to read up on chunked sequences
13:05	tmountain	so many new concepts coming down the pipe that it's hard to keep up ;-)
13:06	ambient	i thought the chunkiness was mostly automatic, at least I haven't had to burden my brain with it
13:06	ambient	at least I shouldn't ought to
13:06	Chousuke	(letfn [(foo [a b & therest]] (chunked-seq? therest))] [(apply foo (range 1000)) (chunked-seq? (range 1000)])
13:06	Chousuke	.,(letfn [(foo [a b & therest]] (chunked-seq? therest))] [(apply foo (range 1000)) (chunked-seq? (range 1000)])
13:06	Chousuke	,(letfn [(foo [a b & therest]] (chunked-seq? therest))] [(apply foo (range 1000)) (chunked-seq? (range 1000)])
13:06	clojurebot	Unmatched delimiter: ]
13:06	arsatiki	tmountain: fascinating. I can't reproduce that at all (I get 800 +- 20 msec with both)
13:06	Chousuke	:P
13:06	Chousuke	hm
13:06	ambient	haha :D
13:06	Chousuke	gah
13:06	Chousuke	,(letfn [(foo [a b & therest] (chunked-seq? therest))] [(apply foo (range 1000)) (chunked-seq? (range 1000)])
13:06	clojurebot	Unmatched delimiter: ]
13:07	arsatiki	tmountain: do you have a cutting edge clj perhaps?
13:07	Chousuke	...
13:07	tmountain	Clojure 1.1.0-alpha-SNAPSHOT
13:07	Chousuke	,(letfn [(foo [a b & therest] (chunked-seq? therest))] [(apply foo (range 1000)) (chunked-seq? (range 1000))])
13:07	clojurebot	[true false]
13:07	Chousuke	huh
13:07	Chousuke	interesting.
13:08	Chousuke	I would have expected the opposite
13:08	arsatiki	tmountain: right. I have 1.0.0
13:09	chouser	,(chunked-seq? (seq (range 1000)))
13:09	Chousuke	,(time (reduce + (range 1000000)))
13:09	clojurebot	499999500000
13:09	clojurebot	"Elapsed time: 771.331 msecs"
13:09	clojurebot	true
13:09	Chousuke	,(time (apply + (range 1000000)))
13:09	clojurebot	499999500000
13:09	clojurebot	"Elapsed time: 1472.001 msecs"
13:09	arsatiki	huh
13:09	tmountain	I'm guessing clojurebot is bleeding edge?
13:09	Chousuke	,(time (reduce str (range 10000)))
13:09	clojurebot	"0123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177
13:10	clojurebot	"Elapsed time: 1562.735 msecs"
13:10	Chousuke	,(time (apply str (range 10000)))
13:10	clojurebot	"0123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177
13:10	clojurebot	"Elapsed time: 30.616 msecs"
13:10	Chousuke	funky, huh :P
13:10	tmountain	wow
13:10	arsatiki	yeah
13:10	ambient	the basics of string concatenation
13:10	Chousuke	in the str case it's obvious why apply is so much faster, but the + case is surprising
13:11	hiredman	clojurebot: clojurebot is not a benchmarking platform
13:11	clojurebot	c'est bon!
13:11	ambient	reduce is (f (f (f (f str)))) apply is (join str-a str-b str-c)
13:12	Chousuke	ambient: the (f (f (f ...))) one is iterate though. :) reduce is (f str3 (f str2 (f str1 init)))
13:13	Chousuke	though the parameter order is the other way around :P
13:13	hiredman	,(time (reduce #(.append %1 %2) (StringBuilder.) (range 10000)))
13:13	clojurebot	#<StringBuilder 0123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172
13:13	clojurebot	"Elapsed time: 1415.392 msecs"
13:14	Chousuke	,(time (reduce #(.append #^StringBuilder %1 %2) (StringBuilder.) (range 10000)))
13:14	clojurebot	#<StringBuilder 0123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172
13:14	clojurebot	"Elapsed time: 9.241 msecs"
13:14	hiredman	zounds
13:14	Chousuke	avoiding reflection = important
13:14	ambient	Chousuke yes, but i was trying to illuminate the fact that reduce is recursive and apply iterative, at least in this case, AFAIK
13:14	ambient	hmm, or is iterative the word im looking for
13:14	hiredman	erm
13:15	hiredman	reduce is iteriation implimented via recur
13:15	ambient	the point is that reduce does string=string+stuff at each step and apply does string=a+b+c+d+e+f
13:16	hiredman	ambient: but str has to use a loop that is just like reduce internally to process all the args
13:17	ambient	str can do (let [new-string-len (+ (length a) (length b) (length c))] ...)
13:18	fabe	wow getting your head around functional programming is harder then i thought took me an hour to to write my first small programm :)
13:18	ambient	whereas string=string+stuff doesn't know the final result and thus has to do O(n) operation at each step of iteration
13:18	hiredman	ambient: no, it can't
13:18	hiredman	strings are immutable
13:19	ambient	well then, create a new string which is the old string plus some other stuff. same diff
13:19	hiredman	so you can't just declare a string of length n and fill it in, and you would still have to use a reduce-ish looking loop
13:19	ambient	i see
13:20	ambient	but clojure strings are java strings, i read that somewhere
13:20	hiredman	and java strings are immutable
13:20	ambient	so perhaps the innards are still mutable
13:20	hiredman	nope
13:20	arsatiki	I tried to do a faster integer sum with type hits but ran into a snag. How do I tag primitive types?
13:20	chouser	The difference is StringBuilder is mutable and so can cheat
13:21	arsatiki	Using #^int says "Unable to resolve classname"
13:21	chouser	(apply str ...) uses a StringBuilder internally where (reduce str ...) can't
13:21	hiredman	~performance
13:21	clojurebot	http://clojure.org/java_interop#toc46
13:22	hiredman	arsatiki: that's for you
13:23	chouser	hiredman: have you considered a way to tell clojurebot to direct an answer at someone? don't know what the syntax would be.
13:23	rhickey	deftype is alive!
13:23	arsatiki	hired: tnx. Turns out that was a bad idea :)
13:24	hiredman	clojurebot: tell so and so about something
13:24	clojurebot	Gabh mo leithscéal?
13:25	chouser	rhickey: so cool. I won't have a chance to port finger trees for a couple days, but I'm looking forward to it. Should be a great fit.
13:26	chouser	rhickey: ought to be able to use 'new' to implement defclass interfaces until protocols are ready? or am I jumping the gun?
13:27	rhickey	chouser: I'm not sure what about what you;ve already done would have to change
13:28	chouser	gen-interface stays for now?
13:28	rhickey	but generally, yes, you can always wrap an interface in a protocol
13:28	rhickey	I'd like a definterface that matches deftype/defclass syntactically
13:28	rhickey	but same plumbing as gen-interface
13:28	chouser	yeah, ok.
13:29	chouser	so right now each node and digit data is stored in a newnew closure
13:29	chouser	with hand-written accessors as needed
13:29	chouser	moving that to deftype would be a win I think
13:30	rhickey	accessors used outside the node?
13:32	chouser	used by other instances of the same class, yes.
13:33	chouser	specifically, the pre/mid/suf triplet that makes up a deep-node.
13:33	rhickey	ok, so with defclass you can get .field access to other instances of same type, with deftype only :field access
13:33	rhickey	in either case no manual accessors needed
13:34	rhickey	(defclass gives you :field as well)
13:34	chouser	right. so deftype's dynamism would have been lovely during development, but now I'd probably go straight to defclass
13:35	rhickey	that's the idea, although for many use cases you;d never need to go to defclass, swapping defclass for deftype and . for : is a breeze
13:35	chouser	right, I've already crossed an ugly line or two trying to wring performance out of finger trees, so defclass for me.
13:36	chouser	the one area I haven't worked on performance is the measure/reduce stuff, but I have a vague notion that deftype/defclass may be great there too.
13:37	chouser	right now, users pass in a map of vectors of fns: {:size [(constantly 1) + 0] :str [str str ""]}
13:38	chouser	well, two fns and a seed value
13:38	chouser	but it probably ought to be either a map or a plain list of instances that implement IMeasureReduce or something
13:39	chouser	hm, except I don't know the types involved, and primitives would be really nice is I could find a good way to store them...
13:49	leafw	is there any way to pretty-print maps of maps with indentation?
13:49	chouser	leafw: clojure.contrib.pprint
13:50	Chousuke	contrib has a pretty-printer. maybe you can use that?
13:50	Raynes	How would I filter through a sequence, removing every other element of the sequence?
13:50	chouser	,(partition 2 1 (range 10))
13:50	clojurebot	((0 1) (1 2) (2 3) (3 4) (4 5) (5 6) (6 7) (7 8) (8 9))
13:50	leafw	chouser, Chousuke: I called (pprint {:a 1 :b 2 :c {:f 11 :g 44} :r 34}) and printed all in one line.
13:50	chouser	,(partition 1 2 (range 10))
13:50	clojurebot	((0) (2) (4) (6) (8))
13:50	chouser	,(map first )(partition 1 2 (range 10))
13:50	clojurebot	java.lang.IllegalArgumentException: Wrong number of args passed to: core$map
13:50	leafw	I am trying to figure out where are the docs
13:50	chouser	,(map first (partition 1 2 (range 10)))
13:50	leafw	there may be options.
13:50	clojurebot	(0 2 4 6 8)
13:51	Raynes	Thanks.
13:51	Chousuke	leafw: contrib documentation is on github
13:51	Chousuke	linked in the wiki
13:52	Chousuke	http://richhickey.github.com/clojure-contrib/pprint-api.html here. :)
13:53	leafw	thanks.
13:55	leafw	woah, pprint doesn't indent by default and the docs are just ... way too low level
13:57	leafw	so, despite all the effort put by the author, I find pprint useless. Dead-brain examples would be nice.
13:59	hiredman	,(doc pretty-print)
13:59	clojurebot	No entiendo
13:59	chouser	try pprinting a bigger map.
13:59	hiredman	uselesss
13:59	chouser	I'm pretty sure it's done a nice job intedenting without any special options for me.
13:59	hiredman	~def pretty-print
13:59	clojurebot	I don't understand.
13:59	leafw	hiredman: (doc pprint)
14:00	hiredman	leafw: pretty-print is one of pprint's tuning vars
14:00	hiredman	the docs say bind to true if you want pretty printing
14:00	hiredman	but the source reveals that it defaults to true
14:01	hiredman	maybe play around with miser-width
14:02	leafw	hiredman: I read the pprint_base.clj code and I I saw that too: all one can do is limit the width.
14:04	hiredman	,(take-nth (range 10) 2)
14:04	clojurebot	java.lang.IllegalArgumentException: Don't know how to create ISeq from: Integer
14:04	hiredman	bah
14:04	hiredman	,(take-nth 2 (range 10))
14:04	clojurebot	(0 2 4 6 8)
14:07	leafw	to concat two vectors? I keep forgetting this simple things
14:07	leafw	it's not conj, neither conc or concat
14:07	Chousuke	(into vec1 vec2) works
14:08	leafw	into ... reads like misusing it, but thnaks
14:08	Chousuke	not really :)
14:51	yuras	how to make hash map from sequence? like (a 2 b 4) -> {a 2, b 4}
14:53	leafw	yuras: a combination of zipmap and interleave ... it can be done.
14:55	chouser	,(apply hash-map
14:55	clojurebot	EOF while reading
14:55	chouser	,(apply hash-map '(a 2 b 4))
14:55	clojurebot	{a 2, b 4}
14:56	leafw	there, eev easier.
14:56	yuras	oh, blame me
14:59	yuras	speaking of shortest ways. i made transformation of ((1 2) (3 4)) -> {1 2, 3 4}
14:59	yuras	this way
14:59	yuras	(apply hash-map (reduce concat '((1 2) (3 4))))
14:59	yuras	is that the shortest one? :)
15:00	yuras	,(apply hash-map (reduce concat '((1 2) (3 4))))
15:00	clojurebot	{1 2, 3 4}
15:01	chouser	,(into {} '((1 2) (3 4)))
15:01	clojurebot	java.lang.ClassCastException: java.lang.Integer cannot be cast to java.util.Map$Entry
15:02	chouser	,(into {} '([1 2] [3 4]))
15:02	clojurebot	{1 2, 3 4}
15:02	chouser	,(into {} [[1 2] [3 4]])
15:02	clojurebot	{1 2, 3 4}
15:02	chouser	yuras: the inner things must be lists and not vectors?
15:02	yuras	that what re-split returns
15:03	chouser	,(into {} (map vector '((1 2) (3 4))))
15:03	clojurebot	java.lang.IllegalArgumentException: Vector arg to map conj must be a pair
15:03	chouser	,(into {} (map vec '((1 2) (3 4))))
15:03	clojurebot	{1 2, 3 4}
15:03	yuras	oh man, you are wizard
15:04	chouser	AWizzArd: <-- no he's the wizard apparently. :-)
15:06	djork	into is magic
15:09	yuras	yeah, digging the sources
15:10	yuras	is there a way to peek sources of function from repl?
15:11	djork	I dunno but it's a nice idea.
15:12	yuras	cheat sheet contains source as doc fun, but its not resolved in 1.1 snapshot
15:13	djork	,^#'into
15:13	clojurebot	{:ns #<Namespace clojure.core>, :name into, :file "clojure/core.clj", :line 4561, :arglists ([to from]), :doc "Returns a new coll consisting of to-coll with all of the items of\n from-coll conjoined."}
15:13	djork	oh god it's perl
15:13	djork	so there you get the file and line
15:13	djork	and you can go from there probably
15:13	leafw	suppose I have an input stream with clojure code to evaluate. What is the best way to do it? Perhaps load-reader ?
15:14	yuras	oh. (source x) is in (use 'clojure.contrib.repl-utils) !
15:14	djork	yay, I thought it might be
15:14	yuras	,(source into)
15:14	clojurebot	Source not found
15:15	djork	aw
15:15	djork	clojure src must not be in the classpath for the bot
15:16	djork	maybe it could be if we ask nicely?
15:17	yuras	~def into
15:18	yuras	i guess its like that and no more
15:19	djork	is it just me or is into not tail-recursive?
15:19	djork	could it blow the stack with a big enough from?
15:19	chouser	'recur' is illegal outside a tail position
15:19	djork	oh ok
15:19	djork	I must not understand tail position
15:19	chouser	yet another reason to love it
15:20	djork	if you have (if foo bar), foo is in tail position?
15:21	chouser	yes
15:21	chouser	no
15:21	yuras	its not
15:21	chouser	(if test then else)
15:21	chouser	test is not in tail position, but then and else are
15:23	yuras	even after python, clojure operations on data look like huge boost, so many and so powerfull
15:23	djork	err yeah that's what I meant
15:23	djork	(if test foo bar) foo is in tail position
15:23	chouser	djork: yes
15:24	djork	(if test (bar (foo)) bat) foo is not in tail position
15:24	chouser	correct
15:25	chouser	(if test bar (bat (foo))) foo is still not in tail position, even though it's "last"
15:26	djork	right
15:26	yuras	b/c foo is applied earlier than bat?
15:26	djork	because it's the returning call that is in tail position
15:26	djork	and there can be many tail positions
15:26	chouser	yes
15:26	djork	OK that clicks now!
15:27	djork	I never made the connection between tail call and returning
15:28	chouser	(do x y z tail) (if x tail tail) (let [a b c d] x y z tail)
15:28	hiredman	well, let is do + bindings
15:28	chouser	(fn [a c] x y z tail) (loop [a b c d] x y z tail)
15:29	hiredman	,(lett [x 1 y 2] (+ x y))
15:29	clojurebot	java.lang.Exception: Unable to resolve symbol: lett in this context
15:29	chouser	(cond a tail b tail c tail)
15:32	hiredman	,(lett [x 1 y 2] (+ x y))
15:32	clojurebot	3
15:32	hiredman	,(macroexpand-1 '(lett [x 1 y 2] (+ x y)))
15:32	clojurebot	((clojure.core/fn [x] (sandbox/lett [y 2] (+ x y))) 1)
15:33	hiredman	explodes if you try to recur across it, which the real let doesn't
15:46	tomoj	how do you name your structmaps?
15:46	chouser	what's the best possible stack trace behavior for macro expansions?
15:46	tomoj	I find I want to call my local variables the same thing I used as the name of the structmap
15:47	hiredman	I like <foo> with foo being the name of the factory function
15:47	tomoj	wow, I hadn't realized <foo> was valid, nice
15:48	hiredman	it might not be technically, but it works
15:48	chouser	I suppose a sort of nested stack? Where you could see the filename and line number of each level of the macro expansion?
15:48	chouser	i think that would be desirable. is it possible?
15:49	hiredman	hmmm
15:49	tomoj	with my structs named <foo> I don't have to use silly local variables like my-foo or the-foo :D
15:50	hiredman	I think according to http://clojure.org/reader <foo> is not technically legal
15:50	somnium	tomoj: what's wrong with f? :)
15:50	tomoj	huh
15:50	hiredman	,'!foo!
15:50	clojurebot	!foo!
15:50	tomoj	I wonder why it works, then
15:50	hiredman	,'\|foo\|
15:50	tomoj	,'<foo>
15:50	clojurebot	\|foo\|
15:50	clojurebot	<foo>
15:50	chouser	much is permitted that is not supported
15:51	hiredman	tomoj: the actually implementation is very lax in what it allows
15:51	hiredman	:/
15:51	tomoj	well I like <foo>. easy to search/replace if it stops working someday.
15:51	hiredman	hmmm, \|foo\| is not legal either
15:51	hiredman	-foo-
15:51	chouser	you could use Foo
15:51	hiredman	,-foo-
15:51	clojurebot	java.lang.Exception: Unable to resolve symbol: -foo- in this context
15:52	somnium	,(let [&$!# '<foo>] &$!#)
15:52	clojurebot	<foo>
15:52	hiredman	:P
15:52	hiredman	clojurebot: much is permittted
15:52	clojurebot	Roger.
15:52	somnium	really, I don't get why the perl community isn't all over clojure
15:53	arbscht	somnium: &$!# that.
15:53	somnium	:)
15:54	tomoj	I hate uppercase
15:56	yuras	,(into {} (for [[k v] '((1 2) (3 4))] [k v]))
15:56	clojurebot	{1 2, 3 4}
15:56	yuras	,(into {} (for [[k v] '((1 2) (3 4))] [(keyword k) v]))
15:56	clojurebot	java.lang.RuntimeException: java.lang.IllegalArgumentException: No matching method found: intern
15:56	hiredman	keyword takes a string
15:57	yuras	arhg
15:57	yuras	thanks
16:06	somnium	does this do copy-on-write: PersistentHashMap.create(new java.util.HashMap()); ?
16:06	Chousuke	probably not.
16:06	somnium	hmm
16:06	chouser	how could it?
16:07	somnium	presumably there's something in the map
16:07	chouser	java.util.HashMap could change and PHM wouldn't know
16:09	chouser	but anyway, no. it uses temporary transients to build a PHM copy of the juHM at the moment the PHM is created.
16:09	Chousuke	temporary transients :)
16:10	somnium	what about: PersistentHashMap.create(some-mutable-java-map-from-the-db-with-mutable-stuff-in-it);
16:10	chouser	Chousuke: yeah, maybe that was a bit unnecessary. :-)
16:11	chouser	If you've got a non-temporary transient you're already in trouble
16:11	chouser	somnium: PersistentHashMap.create doesn't do anything deep, so you'd get a PHM of mutable things.
16:12	somnium	ok, thanks
16:12	chouser	,(let [m {:a (into-array [1])}] (aset (:a m) 0 99) (-> m :a first))
16:13	clojurebot	99
16:16	somnium	is there anything like an into-with (a la merge-with) in core or contrib?
16:19	chouser	what would it do?
16:20	hiredman	sounds like reduce
16:20	hiredman	,(reduce conj #{} '(a b a c d g g w))
16:20	clojurebot	#{w a c b g d}
16:20	somnium	it does, doesn't it
16:21	chouser	,(into #{} '(a b a c d g g w))
16:21	clojurebot	#{w a c b g d}
16:21	chouser	I still don't get it.
16:22	somnium	I was thinking [ x y & fs] (let [f (apply comp fs)] ... (conj x (f y)))
16:22	hiredman	chouser: yes, but with reduce you can specify to the function
16:22	hiredman	that's called map
16:23	hiredman	,(into #{} (map inc (range 10)))
16:23	clojurebot	#{1 2 3 4 5 6 7 8 9 10}
16:23	somnium	into isn't lazy
16:24	hiredman	correct
16:24	djork	I think I want to implement a slideshow lib. Would that fit in contrib?
16:24	chouser	nothing that returns a vector can be lazy
16:24	djork	Or is that too high level?
16:24	hiredman	the only that is lazy are seqs, into produces things that arn't seqs
16:24	somnium	map is lazy
16:24	somnium	so it wouldn't be exactly the same, or I am I confused?
16:24	djork	PLT Scheme has the Slideshow language. Clojure could do the same thing.
16:25	Chousuke	what does that do?
16:25	djork	Chousuke: you mean Slideshow?
16:25	Chousuke	yes
16:26	djork	it lets you define slideshow presentations (think Powerpoint) in Scheme
16:26	hiredman	somnium: map returns a seq
16:26	hiredman	always
16:26	hiredman	seqs can be lazy, so map is lazy
16:26	somnium	(source map)
16:27	hiredman	~def map
16:27	somnium	~def into
16:27	hiredman	I already told you the difference
16:28	hiredman	Sequences are the only non-explicitly lazy thing in clojure
16:28	hiredman	map returns a sequence, into returns whatever type of collection you prime the pump with
16:29	hiredman	,(into {} '([a b] [c d]))
16:29	clojurebot	{a b, c d}
16:32	chouser	I can never ever remember which arg comes first for 'with-meta'
16:33	chouser	Hm, the opposite order that #^ uses. Maybe that'll stick this time.
16:35	arsatiki	here's my mnemonic: with-meta consists of two words, the latter of which is meta. it takes two arguments, the latter of which is meta.
16:35	hiredman	chouser: just think of it as a call to the .withMeta method
16:35	chouser	arsatiki: ok! thanks guys, maybe next time I can avoid running 'doc' again.
16:36	danlei	just use emacs
16:37	djork	heh, yes, slime solves all of my argument-order woes
16:37	tomoj	except for struct
16:39	chouser	emacs has already caused me more pain than any piece of software has the right to.
16:39	somnium	:)
16:40	chouser	then again, so has a bunch of software I still use, so I guess I should quit whining.
16:40	danlei	yes, it did that to me, too. but it also gave me more pleasure than any other :)
16:40	djork	I picked up Emacs to cut down on pain.
16:40	djork	Editing Lua on Windows.
16:40	djork	With no other significant tools.
16:41	danlei	when emacs is set up, everything is fine. it's just that it takes a lifetime :)
16:41	hiredman	lua is a neat little language
16:41	hiredman	I love that the lua build process doesn't involve autotools
16:45	djork	Lua is truly lovely. The function environment features is really cool. Clojure could use something like it for sandboxing.
16:45	djork	but it goes a bit too far for clojure, probably
16:46	djork	woah, paged things in github have vim shortcuts
16:46	hiredman	my mp3 player has a lua interpreter
16:50	djork	a lot of things do... it's like sqlite
16:51	chouser	ooh, are there lua bindings for sqlite?
16:51	hiredman	there must be
16:51	chouser	gah, what am I saying. I retract the question.
16:52	leafw	xD
16:52	hiredman	http://github.com/hiredman/odds-and-ends/blob/master/lisp.lua
16:52	chouser	clojure is great for embedding in existing Java apps to make them scriptable, dynamic, whatever. We just need a version that will do the same for C-based apps.
16:53	hiredman	I'm pretty sure I never finished the interpreter
16:53	tmountain	chouser: won't that be a real possibility once clojure in clojure is done?
16:53	chouser	tmountain: maybe, but there would still be real engineering work to do.
16:54	leafw	is there any simple way to take two maps and return the common keys?
16:54	chouser	leafw: ignoring the values?
16:54	leafw	short of #{(keys m1) (keys m2)}
16:54	leafw	yes
16:54	leafw	just the keys
16:54	leafw	that wouldn't work ...
16:55	leafw	(into (keys m1) (keys m2)) perhaps
16:55	chouser	leafw: that gets you both
16:55	chouser	(filter #(contains? m1 %) (keys m2))
16:55	hiredman	leafw: look at clojure.set
16:55	hiredman	or
16:56	chouser	,(filter #(contains? {:a 1 :b 2} %) (keys {:b 3 :c 4}))
16:56	clojurebot	(:b)
16:56	leafw	thanks :)
16:56	hiredman	,(doc clojure.set/intersection)
16:56	clojurebot	"([s1] [s1 s2] [s1 s2 & sets]); Return a set that is the intersection of the input sets"
16:56	leafw	probably clojure.set would have an intersection
16:56	leafw	there, thanks
16:57	hiredman	http://github.com/hiredman/clojure/blob/readerII/src/clj/clojure/reader.clj started pulling stuff out into macros
16:57	chouser	but you'd have to make a set from the maps which would be inefficient.
16:57	hiredman	,(.keySet {:a 1})
16:57	clojurebot	#< [:a]>
16:58	hiredman	and most clojure.set functions just call seq on the sets anyway :(
16:59	leafw	chouser: I am not concerned about efficiency right now, thanks.
17:17	leafw	I am getting an error trace that makes no sense: java.lang.ClassCastException: java.lang.String cannot be cast to clojure.lang.IObj (identify.clj:1
17:18	leafw	... where at identify.clj:1 there is only a commented out line
17:18	leafw	no other lines of the exception stack trace refer to the file itself.
17:18	kotarak	leafw: you call meta somewhere on a string
17:19	kotarak	... maybe by passing a string to something expecting a symbol.
17:19	leafw	thanks kotarak -- checking
17:22	leafw	kotarak: you are right. In a defn, I had the string before the fn name.
17:22	leafw	the doc string, that is.
17:22	leafw	I would have never guessed so from the Exception stack trace.
17:22	kotarak	leafw: IObj is the meta interface
17:23	chouser	that's a compile-time exception which often has to be read a little differently
17:24	chouser	the final exception which is usually listed first (not the root cause, listed last) should have a file name and line number in the message itself
17:24	chouser	java.lang.ClassCastException: java.lang.String cannot be cast to clojure.lang.IObj (<filename>:<line number>)
17:25	kotarak	chouser: which, as leafw wrote above, is not always the correct spot... ;P
17:25	chouser	oh
17:25	chouser	you're right
17:25	chouser	hmph
17:26	chouser	leafw: was the bad defn in identify.clj?
17:26	leafw	same for missing [] for the arguments of a fn that takes no arguments: the stack trace points to the right line, but that's all the help it provides.
17:26	chouser	was the message cut off at :1 or was there a close paren there?
17:26	leafw	chouser: yes
17:26	leafw	at :1 there was a comment.
17:27	chouser	well, pointing to the right line can be a huge help.
17:27	chouser	leafw: was there a close paren after (identify.clj:1 ?
17:28	leafw	chouser: yes, my bad copy/paster.
17:28	chouser	hmph.
17:28	chouser	when I tried it just now I got the correct line number. I wonder what's different about your case.
17:28	dakrone	hey everyone, can someone help me understand how to get (str-utils2/replace "1d6" #"(\d+)d(\d+)" #"\1") to work with replacing regex groups?
17:28	leafw	try it: create a file with (defn "bla bla" fn-name-here [] (printl "hello"))
17:29	leafw	I have no idea :)
17:29	chouser	leafw: yeah, I did that, but it pointed me at exactly the right line.
17:29	kotarak	OT: Is there a Windows/VimClojure user around?
17:30	hiredman	dakrone: I kind of doubt replace takes two regexs
17:30	hiredman	,(require '[clojure.contrib.str-utils2 :as s2])
17:30	clojurebot	nil
17:30	dakrone	hiredman: according to the docs the replacement is allowed to be a pattern
17:30	hiredman	(doc s2/replace)
17:30	clojurebot	"([string pattern replacement]); Replaces all instances of pattern in string with replacement. Allowed argument types for pattern and replacement are: 1. String and String 2. Character and Character 3. regex Pattern and String (Uses java.util.regex.Matcher.replaceAll) 4. regex Pattern and function (Calls function with re-groups of each match, uses return value as replacement.)"
17:31	chouser	dakrone: (str-utils2/replace "1d6" #"(\d+)d(\d+)" "\\1x\\2")
17:31	kotarak	clojurebot: regex
17:31	hiredman	dakrone: [regex regex] is not listed as an allowed pattern
17:31	clojurebot	Sometimes people have a problem, and decide to solve it with regular expressions. Now they have two problems.
17:31	chouser	that double-escaping on replace is gross, but I don't know what would be better
17:32	dakrone	hiredman: okay, I read it wrong, I need to use a function then?
17:32	hiredman	chouser: are you sure that is right?
17:32	chouser	hiredman: worked here
17:32	hiredman	ok
17:32	chouser	or this, but it might get you in more trouble later: (str-utils2/replace "1d6" #"(\d+)d(\d+)" (str #"\1x\2"))
17:32	dakrone	(replace "foo" #"(\d)" (fn [g] g)) would work then?
17:32	hiredman	generally java uses $1 $2 for groups instead of \1 and \2
17:32	chouser	oh!
17:32	kotarak	,(s2/replace "1d6" #"(\d+)d(\d+\)" "\\1x\\2")
17:32	clojurebot	Unclosed group near index 12 (\d+)d(\d+\) ^
17:32	chouser	yeah, do that: (str-utils2/replace "1d6" #"(\d+)d(\d+)" "$1x$2")
17:33	kotarak	,(s2/replace "1d6" #"(\d+)d(\d+)" "$1x$2")
17:33	clojurebot	"1x6"
17:33	leafw	I have an autogenerated .clj file with 350000 lines, describing a map. And it fails to load. This map was pprint from an actual clojure map. Is there an internal limit to how big a clojure file can be to be loaded?
17:33	dakrone	kotarak: awesome, that works
17:33	dakrone	thank you very much
17:33	kotarak	dakrone: well, that's just what chouser said. :)
17:34	kotarak	leafw: what is the error you get?
17:34	leafw	kotarak: let me paste it somwehere
17:34	hiredman	the reader (written in java) has recursive calls to read, so at somepoint you will blow the stack
17:35	lisppaste8	leafw pasted "error loading gigantic .clj file with a map in it" at http://paste.lisp.org/display/89392
17:36	leafw	hiredman: the map is not very deep, only 4 or 5 levels max
17:36	leafw	i.e. map of maps etc.
17:37	hiredman	leafw: does it contain references to lineage/identify$eval__224
17:37	hiredman	?
17:38	leafw	not that I know of
17:38	leafw	the map should be plain data
17:38	leafw	no functions
17:38	hiredman	well, uh, can you grep it?
17:39	leafw	there is no "identify" neither "defn" neither "lineage" in it
17:39	hiredman	well, certain things like various reader macros can cause the reader to try and interpret something as a classname
17:39	leafw	there are String literals like this: "BLD6 [lineage] #550"
17:39	hiredman	are you sure the output is well formed?
17:40	leafw	AFAIK, whatever is inside the "" ought to be ignored
17:40	hiredman	yes
17:40	leafw	I can't be sure, was geenrated from pprint
17:40	leafw	but ok, I'll grep some more
17:40	hiredman	well just dump a subset of your data through pprint
17:41	hiredman	is there a reason you are using pprint? if you aren't going to personally read the file, why bother pretty printing
17:41	leafw	vim via vimclojure colors it all correctly, that's a good sign.
17:41	hiredman	:(
17:41	kotarak	:)
17:41	leafw	hiredman: the reason was that I have to go through it a bit to see that its printing the right thing.
17:41	hiredman	leafw: can you replicate the whole thing outside of vimclojure?
17:42	leafw	hiredman: you mean running it standalone, not through nailgun? Sure, that is eactly what I did.
17:42	leafw	I am using latest master. Perhaps I should fall back to clojure 1.0
17:42	spuz	does clojure have functions for getting data over http? Ie. making a call to a webservice API?
17:43	kotarak	pfew wishingsweataway
17:43	hiredman	spuz: java.net.*
17:43	bradford	what is the best way to send a fn across process boundaries and get the correct interned var on the other side?
17:43	hiredman	bradford: magic pixie dust
17:43	bradford	argh!
17:44	bradford	I'm playing with lots of symbol, var and ns stuff
17:44	leafw	what a show stoppe.
17:44	kotarak	bradford: (let [the-val var] (fn [foo] (binding [var the-val] ...)))
17:44	spuz	hiredman: URL.openStream?
17:44	jdz	well, a function is just an instance
17:44	bradford	spuz: loot at contrib.http
17:44	leafw	hiredman: I'll try printing without pprint
17:44	kotarak	bradford: or the yet-to-be-included bound-fn macro
17:44	jdz	and instances can be serialised/deserialised, right?
17:45	bradford	kotarak: what the hell is that? :-)
17:45	hiredman	fns and vars are completely different
17:45	bradford	(let [the-val var] (fn [foo] (binding [var the-val]
17:45	bradford	...)))
17:45	kotarak	bradford: that's the magix pixie dust
17:45	hiredman	vars are just storage
17:46	hiredman	spuz: sure
17:46	hiredman	jdz: a fn can close over data
17:46	jdz	hiredman: and how does that not make it an instance?
17:47	hiredman	jdz: it doesn't
17:47	bradford	aren't fns just using vars as part of namespace resolution? symbol name represents fn instance, which is bound to the var in the slot in teh global namespace map that is identified by the namespace/symbol combo?
17:47	kotarak	bradford: say you need out bound to something special in function foo in another thread. Then you do (let [saved-out out] (fn [] (binding [out saved-out] (foo)))) And use the created fn to start the other thread.
17:47	hiredman	it just makes instanciating a new fn complicated
17:47	spuz	bradford: Why didn't I think of checking clojure.contrib? :)
17:48	leafw	internal println was worse: String got printed as its contents, not as a double-quoted String. IS there any other preferable way of getting a {} into a file for later storage?
17:48	hiredman	leafw: don't use println
17:48	kotarak	leafw: prn
17:48	leafw	hiredman: ok
17:48	kotarak	,(print "foo")
17:48	clojurebot	foo
17:48	kotarak	,(pr "foo")
17:48	clojurebot	"foo"
17:48	hiredman	,(doc pr)
17:48	clojurebot	"([] [x] [x & more]); Prints the object(s) to the output stream that is the current value of out. Prints the object(s), separated by spaces if there is more than one. By default, pr and prn print in a way that objects can be read by the reader"
17:49	bradford	when i load a new process now, i resolve by looking up symbols from namespace and fn string...like so
17:49	bradford	(defn retrieve-fn [namespace sym]
17:49	bradford	(let [ns-sym (symbol namespace)]
17:49	bradford	(apply use :reloadall [ns-sym])
17:49	bradford	((ns-resolve ns-sym (symbol sym)))))
17:50	hiredman	bradford: why?
17:50	bradford	the issue now is that the client code (the code that is distributing functions onto remote processes) would like to extract the symbols for the ns and fn from teh fn instance itself.
17:50	kotarak	bradford: ??? doesn't namespace/sym suffice?
17:50	leafw	hiredman, kotarak : same error java.lang.ClassFormatError: Invalid this class index 3171 in constant pool in class file ini/trakem2/vector$eval__64 (SAT-lib-2.clj:1)
17:51	hiredman	hmmmm
17:51	hiredman	well that is a different classfile
17:51	leafw	no, sorry, the ns was different
17:51	hiredman	are these aot compiled classes?
17:51	leafw	the file itself has just a map in it
17:51	leafw	a big, large map
17:52	bradford	so, given a fn instance, what is the simplest way to get its namespace and symbol?
17:52	hiredman	yeah, I mean the namespace calling load
17:52	kotarak	bradford: ah. You are talking about processes while I was thinking threads. Nevermind.
17:52	Chousuke	bradford: it may not have any
17:52	dakrone	I understand how to do something like (dotimes [d dice] (roll-dice sides)), but I don't understand how to collect & operate on the results of all the calls, can someone explain how to add all the results together to me?
17:52	bradford	right, im talking processes
17:52	kotarak	bradford: a fn instance doesn't have that.
17:53	bradford	so i can only get it from the var, right, i.e. ^#'foo
17:53	chouser	yes, named vars know their namespace and name
17:53	kotarak	dakrone: (take dice (repeatedly #(roll-dice sides)))
17:53	chouser	(not all vars are named)
17:53	hiredman	indeed
17:54	bradford	so can i look up a var from an instance that hte var is bound to?
17:54	dakrone	kotarak: awesome, very elegant, thank you much
17:54	hiredman	nope
17:54	leafw	ok, then time to report a bug in 1.1.0-alpha-snapshot
17:54	hiredman	a var is just a sort of pointer to some object
17:54	bradford	so none of this: (fn [x] (:ns (meta (find-var x))))
17:54	bradford
17:55	chouser	you could walk through ever var of every namespace looking for a value that matches
17:55	chouser	maybe you'd find one. maybe not.
17:55	bradford	i suppose i could just quote the function and lass the symbol around rather than the fn instance itself
17:56	rhickey	is the code on the remote site? if so why not just send the vars?
17:56	rhickey	are you actually serializing fn bytecode?
17:56	hiredman	why not just send the whole .clj file
17:56	bradford	yes, the jar is on every machine with teh same code
17:56	leafw	rhickey: is there any know limitation in file size when load-file a very large .clj file?
17:57	hiredman	bradford: what!?
17:57	rhickey	then sending the var itself will unify with the var on the other side, since named vars are interned
17:57	bradford	hiredman: this is hadoop
17:57	rhickey	note, that is not the same thing as the symbol, the var itself
17:57	hiredman	bradford: so why bother?
17:58	hiredman	I don't see why you need to do all this lookup stuff, if you have the samecode on every machine
17:58	chouser	presumably he wants to tell the other process which function to apply to a given task or data
17:58	rhickey	just send vars
17:59	rhickey	that's what they are for, they do this already
17:59	bradford	right, i am just sening around ns/sym for function lookup
17:59	bradford	so i jsut send the var
17:59	rhickey	when read that var will be ready to call if it has the fn on the remote side, no lookup
18:00	bradford	as in... (eval-remote #'foo)
18:00	rhickey	simple reading #'rest will get you the var object
18:00	rhickey	,(read-string "#'rest")
18:00	clojurebot	(var rest)
18:01	rhickey	,((read-string "#'rest") [1 2 3])
18:01	clojurebot	java.lang.ClassCastException: clojure.lang.Cons cannot be cast to clojure.lang.IFn
18:02	hiredman	and clojurebot doesn't allow eval
18:02	rhickey	ack, yes, you will have to eval it
18:02	bradford	and this works because of the way var-lookup / interning works?
18:03	rhickey	but no lookup, since vars look themselves up when interning
18:03	rhickey	right, every var with the same intern name is the same (identical) var
18:03	leafw	so, there are no known file-size limitations for load-file? How can a 6 Mb .clj file fail?
18:03	bradford	right, that's waht i mean, the interning process is built to do this lookup, so i shold be using it to do lookup rather than doing it myself
18:03	rhickey	i.e. (identical? (eval-remote #'foo) (eval-remote #'foo)) -> true
18:04	rhickey	also identical to foo loaded remotely
18:04	rhickey	in same ns
18:04	rhickey	bradford: yes, let interning do it
18:05	bradford	great, this should simplfy some hadoop code we have.
18:05	rhickey	also note that the var is callable
18:05	chouser	could use 'resolve' instead of 'eval' -- would that be a poor choice?
18:05	rhickey	so vars are a good way to communicate about fn identity rather than value, presuming they are stored in fns
18:06	hiredman	rhickey: maybe you could weigh in on leafw's read problem?
18:07	rhickey	if it's a data file you'll hit the constant pool limit for class files, use read for data files, not load
18:08	leafw	( hiredman: I can wait. The current conversation is instructive)
18:09	rhickey	chouser: resolve is about symbols, not vars, but could resolve a fully-qualified symbol to get var, yes
18:09	rhickey	leafw: that's for you above ^^ use read, not load, for data files
18:09	hiredman	oh
18:10	hiredman	oh!
18:10	hiredman	I see read vs. load
18:10	leafw	rhickey, hiredman : thank you.
18:11	leafw	and you mean read and not load-reader?
18:15	hiredman	load-reader and load both go through the compiler
18:15	leafw	load-reader also fails. Ok.
18:15	leafw	(just tested)
18:19	leafw	yay read survived, with a PushbackReader that had 4096 size.
18:26	yuras	what is the opposite fun of (keyword x)?
18:27	somnium	yuras: you're probably looking for 'name
18:27	djork	hey are there better ways to configure jline?
18:27	djork	I'm missing a good bit of the features I am used to
18:28	djork	like Emacs-style editing (C-d delete, M-b or M-f jumping)
18:28	leafw	I am impressed by the way at how fast clojure can parse a 6 Mb map from a file. Its almost subsecond.
18:32	hiredman	Clojure: It's More Fn
18:32	StartsWithK	djork: you can configure jline if you write your own clojure launcher with jline api, there was/is a project that does that on google code, but i can't remeber the name of it or thoes is still exist
18:33	djork	Jline docs are nonexistent as far as I can tell
18:33	StartsWithK	does*
18:33	somnium	hiredman: will that be the T-shirt slogan for clojure-conf '10 ?
18:33	hiredman	chouser could put that on a shirt
18:34	djork	hah nice
18:34	hiredman	clojurebot: zazzle?
18:34	clojurebot	excusez-moi
18:34	hiredman	http://www.zazzle.com/clojure+tshirts
18:35	somnium	that logo would look good on coveralls
18:36	StartsWithK	djork: http://wrost.blogspot.com/2009/08/clojure-tab-completion-on-windows.html maybe this will help
18:36	leafw	continuing ... what is the best way to create a sorted map from an existing map? sorted-map requires elements unpacked from a list, so (sorted-map (interleave (keys m) (vals m))) cannot work. So how?
18:37	djork	that site is the only site on the entire internet that documents jline.words
18:37	leafw	(I am trying to avoid (TreeMap. m), which would work, but adds a java dependency)
18:37	rhickey	,(into (sorted-map) {:c 1 :b 2 :a 3})
18:37	clojurebot	{:a 3, :b 2, :c 1}
18:37	leafw	into. The magic word of the day. Thanks.
18:38	yuras	into made my day today. very magical thing and yet works nicely
18:38	leafw	into seems to emulate the constructor that most java collections have, combined with the addAll
18:42	kotarak	btw: does anyone else have problems with the group? My posts take sometimes over six hours to show up. :(
18:43	leafw	haven't noticed anything like that kotarak, but I have not checked properly.
18:44	rhickey	kotarak: I moderate several times a day, and I think chouser does as well
18:44	kotarak	It started sometime ago, when I was suddenly banned. :(( chouser fixed that. But since then I have trouble with terrible delays.
18:44	leafw	another quirk: clojure.set/intersection can't work with a key set obtained from a map: clojure.lang.APersistentMap$KeySeq cannot be cast to clojure.lang.IPersistentSet
18:44	StartsWithK	i am using clojure 1.0.0, to do unit test i have to load 16 test namespaces and there dependencies, it takes 4 seconds to load them? is this normal? (there is no aot)
18:44	rhickey	we've had a problem with spammers spoofing legit members. I had to moderate myself!
18:44	kotarak	o.O
18:44	hiredman	StartsWithK: 4 seconds from a cold jvm?
18:45	StartsWithK	hiredman: no, only to (require) test namespaces
18:45	hiredman	hmmm
18:46	hiredman	well, you have to go to disk to load each file, and each file has to be read and compiled
18:46	StartsWithK	i know, but still, 4 sec to load something like 32 files
18:47	hiredman	clojurebot has the largest set of namespaces of any of my projects, and it takes a few beats to load, but I've never checked to see where that time is spent, could just be waiting to make the irc connection
18:47	leafw	and a pseudo-quirk: I would expect the collection returned by vals on a sorted map to implement List interface, since it's ordered. But java.util.TreeMap doesn't either.
18:49	leafw	same for the sorted keys.
18:50	leafw	and I am wrong: it does!
18:50	leafw	,(ancestors (class (vals (into (sorted-map) {:a 1 :b 2}))))
18:50	clojurebot	#{java.lang.Object clojure.lang.Seqable clojure.lang.IPersistentCollection java.util.Collection clojure.lang.ASeq clojure.lang.IMeta clojure.lang.Streamable clojure.lang.Sequential java.lang.Iterable clojure.lang.IObj :clojure.contrib.generic/any clojure.lang.Obj java.util.List clojure.lang.ISeq}
18:51	StartsWithK	there is total of 34 namespaces and 16 namespaces with tests, non compiled
18:51	jhawk28	another favorable article from Tim Bray: http://www.tbray.org/ongoing/When/200x/2009/10/27/Recur
18:53	StartsWithK	funny thing is, i (read) every file of my project, at least until i hit first ns declaration in it, and it takes 1ms to do that for all the files
18:53	StartsWithK	so i think its a compiler
18:57	djork	"But it seems obvious to me that recur is better than tail-call-optimized recursion." true dat
18:58	djork	so don't take his word for it
18:58	hiredman	I need to finish messing around with my reader so I can start messing around with the compiler
19:00	StartsWithK	a new reader? in clojure or java?
19:00	hiredman	clojure
19:01	hiredman	yeah
19:01	StartsWithK	that would be nice, clj->html, lint, formating..
19:01	hiredman	just rewriting LispReader.java into clojure/reader.clj
19:02	spuz	I have a question about code style. In java I often declare local variables even if they are only referenced once so that I can break up the code and also make it more self documenting. In clojure which of the following styles is preferred?
19:02	lisppaste8	spuz pasted "Local variables in clojure" at http://paste.lisp.org/display/89396
19:02	hiredman	http://github.com/hiredman/clojure/blob/readerII/src/clj/clojure/reader.clj a few missing pieces and needs to be cleaned up and better build integration
19:04	hiredman	do you really want to self-document google's auth api?
19:04	spuz	hiredman: ?
19:05	hiredman	you are basically documenting google's auth-token api
19:05	hiredman	from the client side
19:06	hiredman	which, I mean, who cares? if you have some hairy function it's nice to break it up into chunks, but if it is just hitting some other standard api, shrug
19:07	spuz	I wouldn't say it's a standard API, it's not as if you can hover over it and see what it expects as arguments...
19:07	hiredman	spuz: I can just google it
19:07	spuz	heh
19:07	hiredman	I would document what is userdetails is supposed to be
19:08	spuz	it's not just the auth api, it's all of the parameters which have to be pieced together into the request string
19:08	hiredman	which is just a detail of the api
19:08	spuz	yep ok
19:08	jhawk28	hiredman: line 189 - misspelling ;)
19:09	hiredman	jhawk28: huh?
19:09	hiredman	in eval-reader?
19:09	jhawk28	allowed instead of allow
19:09	hiredman	ah
19:09	hiredman	shrug
19:09	jhawk28	hehe
19:09	hiredman	I guess that is important
19:10	StartsWithK	hiredman: will the final reader also be regex based?
19:10	hiredman	eh?
19:11	hiredman	I'd hardly call it regex based
19:11	spuz	hiredman: so would you say it's fair to say that in clojure or any lisp, the way to document your code is to break it up into smaller functions, rather than into local variables?
19:11	hiredman	it uses 4 regexs in 4 or 5 functions
19:12	StartsWithK	hiredman: yes, your right, i just looke at the top of the file
19:12	hiredman	spuz: no
19:12	jhawk28	any idea when a function will allow more than 20 parameters?
19:13	hiredman	I would just say who cares about the details of google auth? if you are doing something interesting or tricky, break it up and document it by all means
19:13	spuz	hiredman: that sounds like a yes :p
19:16	StartsWithK	hiredman: this style you use with <string> sufix and such, is that something special in clojure 1.1 or new naming convention?
19:16	hiredman	if you are passing a map, I recommend map destructering
19:17	Chousuke	jhawk28: max parameter count will actually go down, not up :P
19:17	Chousuke	jhawk28: if you need more, use a rest parameter
19:18	hiredman	StartsWithK: nameing convention
19:18	hiredman	LispReader has two static methods with the same name that vary on the type of the first param
19:19	StartsWithK	i have some macros that have more than one variant, so foo expands or to foo-1 or foo-2, is foo<1> prefered in general?
19:19	StartsWithK	its hard to find any style guides on clojure code :)
19:20	hiredman	I dunno
19:20	hiredman	<foo> might not technically be a valid symbol
19:20	Chousuke	it is.
19:21	Chousuke	StartsWithK: I'd find more descriptive names than -1 and -2 :P
19:21	hiredman	are you sure? clojure.org/reader says alphanumeric plus a handful of symbols
19:22	hiredman	which don't include < or >
19:22	Chousuke	except < and > exist as functions :)
19:22	hiredman	so does /
19:22	Chousuke	yeah, but that receives special treatment anyway
19:23	hiredman	StartsWithK: aslong as you are consistent, I just like <1> visually
19:23	StartsWithK	Chousuke: its hard when they are only used to parse two different syntax variants of the same macro, so foo -> (or foo-1 foo-2) -> foo* is the best i can think of in such a case
19:24	hiredman	I'm not sure I even need those particular macros
19:24	hiredman	but I am sticking with a direct translation of LispReader
19:24	StartsWithK	i don't think it gets more descriptive with that, i tried something like foo-with-only-properties-defined foo-with-properties-and-dependencies, but that is just too long
19:24	Chousuke	so wait, you're actually working on your reader now ?:P
19:25	hiredman	yeah
19:25	hiredman	I stopped working on my old gen-class reader
19:25	Chousuke	I still have to find that bug with my reader :P
19:25	StartsWithK	how will clojure use reader in clojure?
19:25	hiredman	I've been grinding on thew new fractal letfn reader for the past week
19:26	somnium	you guys should agree on some tests and benchmarks so you can compete :P
19:26	Chousuke	StartsWithK: it's not easy but it's possible.
19:26	StartsWithK	old one will need to be there to bootstrap it?
19:26	Chousuke	only once )
19:26	StartsWithK	or will clojure keep its a jar of older version for that?
19:27	hiredman	I have ant compile java+clj, compile my reader, then delete compiled clojure (keep the compiled java), then recompile clojure using my compiled reader
19:28	hiredman	my clojure prints "falling back to java reader" to System.err at startup if it can't load the cinc reader
19:28	Chousuke	I don't think I'll have time to work on mine for a while
19:29	hiredman	:(
19:29	Chousuke	and that bug I'm seeing... I have no clue.
19:29	hiredman	my syntax reader doesn't work either
19:29	Chousuke	the stacktrace mentions macroexpand and definline and whatnot but nothing there looks wrong.
19:30	Chousuke	so I bet its some error that just build up until it finally triggers there.
19:30	Chousuke	builds*
19:30	hiredman	yeah
19:30	Chousuke	I think it might be because I'm using a syntax-quote macro instead of just expanding it at read time.
19:31	Chousuke	but that would suck.
19:31	hiredman	it turns out I had a bunch of bugs, and because of my incremental fall back to LispReader they weren't exposed until I replaced read-delimited-list
19:31	Chousuke	;(
19:31	hiredman	which exposed two or three bugs in other readers, which drove me crazy
19:31	Chousuke	:)
19:32	Chousuke	I think I might be encountering additional problems due to the pushbackreader (ew state :()
19:32	hiredman	because for a while I thought I was just missing something in read-delimited-list
19:32	Chousuke	I originally wrote the reader to work on a seq of characters
19:32	hiredman	hmmmm
19:32	Chousuke	then adapted it a bit to use a pushbackreader
19:32	hiredman	wouldn't it be easy to just turn a pbr into a seq of chars?
19:33	Chousuke	yeah, but turns out you need some "read-ahead"
19:33	Chousuke	at least one character
19:33	Chousuke	for example, if the input is "foo\n" then to read the symbol foo you need to consume the \n too
19:33	Chousuke	but that confuses the repl )
19:33	Chousuke	so you need to put the \n back
19:35	hiredman	you just cons the character back on :P
19:35	Chousuke	hiredman: except that won't work.
19:35	Chousuke	hiredman: as it's gone from the stream, which the repl uses
19:35	hiredman	:(
19:36	Chousuke	also for some reason attaching line metadata to objects causes the init method code to grow to ~90k which is over the 64k limit
19:36	hiredman	hmmm
19:37	Chousuke	if I change the metadata attaching function to identity, it works :P
19:37	Chousuke	but I get no line metadata.
19:38	Chousuke	but not having line metadata makes it so much harder to debug ;(
19:38	Raynes	I think I just found the thing that will make me stop using VimClojure. :\
19:39	Raynes	The weirdness of namespaces.
19:40	StartsWithK	Raynes: what weirdness?
19:40	Raynes	You can't just open any file apparently, it tosses an error because of namespace crap. I have to open the file (it has to be on the servers classpath) and then \rf the file, and then close and then reopen the file. Apparently.
19:40	somnium	huh, just tried to build 1.0 branch of contrib with clojure 1.0 and it blew up
19:42	Raynes	At least, according to this: http://www.mail-archive.com/clojure@googlegroups.com/msg10657.html, it's the expected behavior. It's annoying behavior to me. :\
19:44	Raynes	It's a bit of a dream killer.
19:45	technomancy	if you ever feel bummed out by the fact that clojure is a relatively new/immature language, I recommend http://planet.lisp.org to set you right again.
19:45	technomancy	particularly things like http://trittweiler.blogspot.com/2009/10/0affff.html
19:46	Raynes	Are you talking to me? >_>
19:46	technomancy	sort of
19:48	Raynes	Nothing to do with Clojure. I have no earthly idea why VimClojure works like that. I was just trying it out anyways. I just doubt I'll continue using it if this is supposed to be a "feature".
19:52	tomoj	come to the dark side
19:59	Raynes	tomoj: If you were implying that I should use Emacs, I already do. I've just been playing with Vim for a few days to see what all the fuss was about. But after discovering this retarded little 'feature', I'm not sure what the fuss /is/ about.
20:00	tomoj	there's fuss about vimclojure?
20:01	somnium	I like vim, but I never managed to get vim clojure to work after two tries
20:01	somnium	is the clojure-1.0 jar on google code broken?
20:01	leafw	somnium: it's no big deal really.
20:01	tmountain	anybody know if there's a library floating around to do https requests?
20:03	StartsWithK	tmountain: apache commons httpclient can
20:03	tomoj	I imagine java can help cook one up pretty easily
20:03	tomoj	ah, :)
20:05	jhawk28	tmountain: the Jetty client can too, but it doesnt support https through a proxy
20:06	tmountain	I'm trying with technomancy's client now, and pointing it at https://www.google.com/ gives me a 302 moved response
20:06	tmountain	I wonder if that's verification that it works...
20:07	technomancy	word of warning: that is not a particularly well-tested library.
20:07	technomancy	I'm only now starting to use it in real code as of like a week ago.
20:07	tmountain	no worries, I'm using it for a pet project
20:07	tmountain	trying to hack together a google wave bot in clojure
20:07	technomancy	perfect; that's exactly what it was designed for
20:08	technomancy	(pet project, not wave)
20:08	tmountain	yeah, it seems to work fine at the level I'm using it, so I don't forsee any problems
20:09	tmountain	rewriting old Java code in Clojure really makes you realize how much unnecessary ceremony Java has
20:10	tmountain	like... I can just rip out these 15 lines and replace them with this one-liner
20:10	tmountain	it's crazy
20:11	technomancy	if that's surprising you either haven't been doing java very long or you've been doing it waaaaay too long. =)
20:13	somnium	tmountain: I just had the opposite experience, I needed to write a little java proxy class and was amazed at how excruciatingly painful it was after getting used to clojure
20:17	jhawk28	tmountain: you probably need to import the cert into the store
20:20	tmountain	I think you guys misunderstood me
20:20	tmountain	I'm saying Java code is too freakin' long
20:20	tmountain	heh
20:21	funkenblatt	misunderstanding implies that they actually read what you wrote
20:21	yuras	:)
20:21	funkenblatt	which i think may be assuming too much
20:21	yuras	i found even python code too long after clojure idioms
20:22	tmountain	agreed
20:29	yuras	can somebody help me to make this func more clojurish http://gist.github.com/220108 ? or is it already okay?
20:39	somnium	yuras: you could use reduce instead of loop
21:42	bmidgley2	are there any working applet demos out there?
21:43	chouser	yeah, there was a thread on the group recently
21:46	bmidgley2	chouser: the google group? I couldn't find it
21:47	chouser	ok, I'll look.
21:49	chouser	http://en.wikibooks.org/wiki/Clojure_Programming/Examples/Creating_an_Applet
21:52	tomoj	comes up for me when googling "clojure applet"..
21:52	tomoj	maybe my google knows I want to see that
21:53	tomoj	one might wrap some reads in a dosync to ensure a consistent snapshot, even though there's no writes, right?
21:56	rlb	Is there already something that lets you get a sub-command's stdout as a java stream? i.e. (line-seq (:stdout (sub-process ["find" "-type" "f" ...])))
21:57	chouser	tomoj: yes
21:58	tomoj	there's the thing in java
21:58	bmidgley2	ok, I guess I need a newer java6... none of the examples I found would run
21:58	tomoj	if it's wrapped in clojure I haven't seen it
21:58	chouser	rlb: I don't think so. there's shell-out in contrib, but it reads the whole output before returning to help uncomplicate resource management
21:59	rlb	right -- I need a stream
22:00	rlb	perhaps I'll look in to it...
22:00	chouser	you can consult the source there to see how to go about it
22:01	chouser	I've considered tweaking it to allow for an api like (with-process-stream [] ...)
22:01	bmidgley2	hmm... http://www.xenoage.com/extern/zongblog/eartraining/ gives me "Loading Java Applet Failed..."
22:01	bmidgley2	in firefox on osx 10.5.8
22:01	chouser	rlb: that would close and clean up when you leave that dynamic scope. Would that fit your use case
22:01	bmidgley2	java 1.6.0_15
22:02	chouser	bmidgley2: yeah, macs give clojure applets trouble.
22:02	bmidgley2	too bad
22:02	chouser	bmidgley2: there's some discussion here: http://tinyurl.com/yfbbfjs
22:02	bmidgley2	our usergroup is in a clojure discussion now
22:03	rlb	chouser: might.
22:03	tomoj	,(line-seq (java.io.BufferedReader. (java.io.InputStreamReader. (.getInputStream (.exec (Runtime/getRuntime) "echo hello")))))
22:03	clojurebot	java.security.AccessControlException: access denied (java.io.FilePermission <<ALL FILES>> execute)
22:03	tomoj	darn
22:03	chouser	tomoj: but note that leaves the streams open and the process running or in a zombie state
22:04	rlb	chouser: though perhaps I'll handle this particular problem differently -- might skip the sub-process and have clojure handle everything. Though I think I'll still want a way to get a lazy sub-process stream eventually.
22:04	tomoj	chouser: noted
22:04	tomoj	hmm
22:04	tomoj	I don't see any "echo" is ps -A
22:05	rlb	(Someone will need to wait() on the sub-process unless that's handled elsewhere.)
22:06	chouser	hm
22:06	chouser	that was my understanding as well, but the process does seem to be gone
22:07	rlb	Perhaps java handles that indirectly.
22:11	bmidgley2	aha!
22:11	bmidgley2	I have 1.6.x at the command line but only 1.5.x in the browser
22:13	slashus2	bmidgley2: Look at your java preferences in your utility folder.
22:15	bmidgley2	it has a 64-bit java se6 first in the list for the applet plugin
22:15	bmidgley2	does the browser choke on 64 bit plugins
22:16	slashus2	try the 32 bit java 6
22:17	bmidgley2	I don't have that in my list
22:17	bmidgley2	downloading the latest java for osx... we'll see
22:21	chouser	I think I may have been misreading the java docs
22:22	chouser	I think the Process instance may wait for the process and handle the clean up for you.
22:24	chouser	If that's the case, then the streams are as safe to use as any other stream and shell-out could be a lot less picky.
22:28	tomoj	so is my line above bad or not?
22:30	chouser	it's bad.
22:30	wooby	has anyone investigated using simple build tool to compile/package clj stuff?
22:32	somnium	wooby: ant works for me
22:32	somnium	wooby: and netbeans will generate a build.xml for you if you create a clojure project, though I haven't done it more than twice
22:33	hiredman	ant has some annoying issues
22:33	hiredman	I really wish ant would not rebuild a dependency unless something was changed
22:34	wooby	yeah i'm not so much into ant
22:34	wooby	i really got used to sbt, and the dep handling especially
22:34	wooby	the transparent ivy goings-on
22:34	hiredman	<3 make
22:34	wooby	make pretty much rules
22:35	chouser	tomoj: this is apparently good enough though: (with-open [s (java.io.BufferedReader. (java.io.InputStreamReader. (.getInputStream (.exec (Runtime/getRuntime) "echo foo"))))] (vec (line-seq s)))
22:35	hiredman	I see a lot of work going into maven+clojure
22:35	hiredman	but I haven't tried that out
22:35	somnium	I wonder how jruby-rake would do
22:38	tomoj	chouser: I see, thanks
22:40	hiredman	xml xml xml everywhere
22:40	hiredman	~xml
22:40	clojurebot	XmL is case-sensitive
22:41	hiredman	~xml
22:41	clojurebot	xml is like violence; if it's not working, you're not using enough of it.
22:41	wavis_	zomg hiredman, that made my evening
22:42	rlb	Does clojure specify anything about the evaluation order of args?
22:42	chouser	yep, they go in order
22:42	rlb	i.e. do you need a let when order is important?
22:42	hiredman	l-to-r
22:43	rlb	hiredman: guaranteed?
22:43	arohner	rlb: yes
22:43	hiredman	really? I know that is how it works, but is that specified somewhere?
22:43	arohner	all lisps I know of do left-to-right
22:43	arohner	scheme specifies it
22:44	hiredman	snort
22:44	arohner	CL probably does too
22:44	rlb	arohner: I don't think it does.
22:44	chouser	yeah, I'm pretty sure it's guaranteed. eval order of literal maps is not guaranteed
22:44	rlb	From r5rs: "The operator and operand expressions are evaluated (in an unspecified order)"
22:45	hiredman	I know I read somewhere that some lisp goes the opposite order, but that was some old paper
22:45	rlb	In the case I care about, it's critical i.e. (vector (read-from-stream) (read-from-stream) (read-from-stream)), etc.
22:45	hiredman	there have doubtedly been lazy evaluation lisps
22:46	rlb	Perhaps I'll just use a let for now and worry about that later...
22:46	hiredman	rlb: I tihnk it would be clearer to put it in a let regardless
22:47	arohner	http://www.lispworks.com/documentation/HyperSpec/Issues/iss279_w.htm
22:47	tomoj	"The subforms in the cdr of the original form are evaluated in left-to-right order in the current lexical and dynamic environments."
22:47	tomoj	CL
22:49	hiredman	clojurebot: what does the hyperspec say about evaluation order?
22:49	clojurebot	what is wrong with you
22:49	hiredman	~botsnack
22:49	clojurebot	thanks; that was delicious. (nom nom nom)
22:50	arohner	http://lib.store.yahoo.net/lib/paulgraham/acl2.txt
22:50	hiredman	clojurebot: hyperspec?
22:50	clojurebot	hyperspec is not applicable
23:05	rlb	Is it possible to do a precise read from java - the equivalent of read(fd, buffer, 32)?
23:05	tomoj	has anyone seen NullPointerExceptions on trying to print something?
23:06	rlb	This has to be binary, and a single read, not a sequence of character reads.
23:06	rlb	(I'm trying to read data from a device on /dev/input/event*.)
23:06	tomoj	I mean like this: (class foo) => c.l.PersistentStructMap, (str foo) => NullPointerException
23:06	rlb	It's trivial in C, but I'm not sure how to get clojure to do it.
23:08	tomoj	oh it's because one of the values in the map is a lazy seq
23:08	tomoj	the backtrace is unfortunately not at all helpful
23:36	rlb	Nevermind -- got it, (.read stream byte-array-size-n) seems to work.
23:46	Licenser	somnium: aloa
23:49	wwmorgan	how can I test whether an object I have is an array?
23:53	scottj	wwmorgan: there are vector? and type
23:53	scottj	wwmorgan: and seq?
23:55	wwmorgan	type is what I'm looking for. Thanks
23:58	hiredman	ok