#clojure – 2011-06-02

2011-06-02

00:11	tomoj	clojure support?? wow
00:18	kornys	tomoj: yep - see https://gist.github.com/1001206 - it's pretty awesome
00:34	offby1	$36/month isn't that bad.
00:34	offby1	EC2 micro is a bit cheaper, but not much
00:50	dnolen	hiredman: ping
00:50	kornys	offby1: yep, though for a tiny demo app, free would be better :)
00:52	hiredman	dnolen: pong
00:53	dnolen	hiredman: I've addressed some but perhaps not all of the issues. LVar and LCons now take metadata for doing custom things.
00:53	dnolen	hiredman: (unifier '(?x . ?y) [1 2]) (binding-map '(?x . ?y) [1 2]) work for first and rest matching.
00:54	hiredman	ah, cool
00:55	dnolen	hiredman: one thing I'm on the fence about is allowing people to directly pass LCons instances to the unifier ...
00:56	dnolen	hiredman: if you wan to create something that you can spit out from a macro, will emitting (lcons 1 (lcons 2 ...)) not work for you? core.logic also provides convenience macro llist for doing this.
00:56	hiredman	well, if they have a pre-prepped thing, and the pre-prepping happens to create an LCons
00:57	hiredman	dnolen: possibly, since you mention prepping is expensive in the readme, and my patter is static at compile time I have the macro do the prepping
00:57	hiredman	patterns
00:57	hiredman	are
00:58	hiredman	since I am not familiar with lcons I went with what seemed like the highest fidelity (pulling the fields and passing them to the constructor again)
00:59	dnolen	yeah I have to think about LOGIC-7 some more. let me know if (unifier '[(?x . ?y) 3] '[(1 2) 3]) support satisfies 5 & 6.
01:01	hiredman	seems like it would, but yes, I'll kick it around
01:01	dnolen	latest is on github and clojars.
03:15	chouser	there. pushed a very early version of data.xml
03:15	chouser	Hm, needs a README. https://github.com/clojure/data.xml
03:20	brehaut	heh "; blame Chouser for anything broken or ugly."
03:22	brehaut	chouser: is it 1.2 compatible?
05:08	andrewclegg	chouser: is data.xml fully lazy, i.e. can you parse really big files as a stream of chunks?
05:08	andrewclegg	v. handy if so
07:48	manutter	lol, I so cheated on 4clojure question #65
07:50	manutter	https://gist.github.com/1004288
07:50	mrBliss	manutter: quite clever!
07:51	manutter	I couldn't resist
07:52	manutter	expect one more function to be added to the Do Not Use list very soon now...
07:56	tufflax	Does swap! "block" the thread calling swap! until it is successful? Or is it a new thread?
08:35	chouser	tufflax: swap! spins. It will not return until it is successful.
08:35	tufflax	ok
08:38	chouser	andrewclegg: That's the intention, but it needs more testing before I'm willing to claim that.
08:42	tufflax	I've got a really nasty but on my hands, has to do with java socketchannels. If anyone want to take a look it's here: http://pastebin.com/Xymx9ZgE
08:42	tufflax	thank you chouser btw :)
08:48	tufflax	s/but/bug and s/want/wants :p
08:57	timvisher	hey all
08:58	tufflax	hi
08:59	timvisher	I'm trying to serialize a seq to disk in a file using `pr` and then read it back in and I'm encountering `error: java.lang.ClassFormatError: Invalid method Code length 361053 in class file wallpaper_manager_core/core$eval2183`
08:59	timvisher	My only guess is that it's trying to execute a function because the seq is serialized as `(...)`
09:00	timvisher	Any thoughts?
09:00	stuartsierra	What are you using to read it back in?
09:00	tufflax	Paste the code
09:00	timvisher	(load-file)
09:00	stuartsierra	There's your problem. :)
09:00	stuartsierra	`load-file` means read & eval. You just want `read`
09:00	timvisher	Ah
09:01	timvisher	so should I make the file into a string and then call read on it?
09:01	timvisher	or is there a convenience function for that?
09:01	tufflax	in that case read-string
09:01	raek	timvisher: no, 'read' reads from a stream.
09:01	timvisher	raek: nice!
09:02	raek	summary: pr and read work on streams (out and in, which can be rebound with 'binding'), pr-str and read-string work on strings
09:21	andrewclegg	chouser: I'll test it by throwing some gigs at it and seeing what happens
09:30	markskil1eck	Any idea why get-datasets for incanter is returning some html?
09:31	kephale1	what is the best way to download files w/in clojure (files of arbitrary type incl binary)?
09:32	kephale1	a la curl
09:33	drewr	kephale1: you can do something with (clojure.java.io/reader "http://foo")
09:34	raek	kephale1: if you don't need to control the headers, the simplest way is probably to pass a URL to clojure.java.io/reader (text) or clojure.java.io/input-stream (binary)
09:34	chouser	andrewclegg: cool. let me know what you find. Do beware of head-holding, though.
09:34	kephale1	drewr & raek: aha! ty
09:35	drewr	was thinking reader would polymorphically give you an input-stream, but that wouldn't really make sense
09:36	tufflax	If anyone was trying to figure my bug out, i solved it.
09:36	raek	kephale1: it is probably a good idea to pass the url string through clojure.java.io/as-url and then verify that the scheme is http or https
09:37	raek	kephale1: my friends helped me find a bug/feature in my title-printing IRC bot but asking it to look up the title for the URL file:///dev/random
09:38	raek	s/but/by/
09:38	sexpbot	<raek> kephale1: my friends helped me find a bug/feature in my title-printing IRC bot by asking it to look up the title for the URL file:///dev/random
09:38	raek	tufflax: what caused it?
09:38	kephale1	raek: oh good point
09:40	raek	kephale1: another point: if you want to read text, you specify the encoding of the text (the "file format" of the characters) with the :encoding option: (clojure.java.io/reader url :encoding "UTF-8")
09:40	raek	if it is left out, UTF-8 will be used
09:41	kephale1	cool, i'm mostly going to be shuffling jars around, but might have some text
09:42	tufflax	raek: I was reading into the wrong buffer. On line 128 in my paste i should have saved the new c in the global players, because i was relying on :read-tmp to be set right. Now if I didnt get any more data after reading the length of the message i forgot the new read-tmp and next time read into the buf for getting the lengths
09:47	tufflax	I gotta come up with a better way to debug my programs. I ran the server in a future, and prints dont show from the new thread. What I ended up doing was conjing debug msgs into a global. btw i tried bound-fn but it didnt help. In other languages ive mostly used prints to debug but it's not very convenient in clojure. is there a good debugger?
09:47	raek	tufflax: are you using emacs, btw?
09:47	tufflax	vim
09:48	raek	the prints should appear somewhere
09:48	tufflax	not before my eyes ;)
09:48	raek	in emacs, the repl and stdout are distinct, so prints from the repl thread appears in the repl and prints from other threads appear in stdout
09:49	raek	tufflax: you could try (intern 'clojure.core 'out out) in the repl where you want the prints
09:49	tufflax	oh, that explains why bound-fn didnt work
09:49	raek	this will change the root binding
09:49	raek	tufflax: strange, bound-fn should work too
09:50	tufflax	but still, prints is not so convenient, because sometimes one needs to add a (do ... ) or something like that. what do you use for debugging?
09:51	raek	for stateful code? prints I guess... :)
09:51	tufflax	hehe
09:51	raek	for purely functional code, you can always test things in isolation in the repl
09:52	tufflax	yeah of course, functional is better :p
09:52	raek	I tend to build a purely functional layer first that does parsing and operations on domain data
09:52	kephale1	is there a pretty way to empty an input stream into an output stream?
09:52	kephale1	copy into, i should say
09:52	tufflax	raek did you read my code? any tips on style?
09:53	raek	kephale1: yes, 'copy' :-) http://clojuredocs.org/clojure_core/clojure.java.io/copy
09:53	kephale1	raek: yay, ty again
09:54	raek	tufflax: I briefly looked though it. I'm not very familiar with how you program using the java.nio classes. my first idea was that this would probably be simpler using a thread/connection and streams
09:55	raek	but I know that one thread per connection does not scale very well...
09:55	tufflax	Yeah thats why i didnt want that
09:58	raek	one advice is to try to limit the number of kinds of side-effects per function
09:59	raek	for example read-from-client both reads data and changes the atom
10:00	raek	if it would return the data instead, it would be easier to test it in isolation
10:00	tufflax	yeah that migth be a good idea
10:01	tufflax	s/migth/might/
10:01	sexpbot	<tufflax> yeah that might be a good idea
10:01	Cozey	Which Mustache library would you recommend? clostache, stencil, mustache.clj ?
10:01	raek	in general, try moving state-changing stuff out from functions if possible
10:01	raek	I think add-new-client is fine here, since it only does mutation
10:12	Cozey	what would you recomment for app configuration? property file? file with clojure code ?
10:13	abedra	Cozey, it depends on what you are configuring and if that configuration needs to change while the program is running
10:14	Cozey	i'd say it won't be update by the app, but maybe it would be nice to change the conf and notify an application - do You have something in mind?
10:14	Cozey	this would actually not be easy to implement in a smart way
10:15	abedra	Cozey, having a clj file as configuration data makes it a little easier to reach in and modify configuration
10:15	Cozey	yes
10:15	Cozey	true
10:16	Cozey	it would be in lisp philosophy
10:16	chouser	but not if there are comments in the config file
10:16	abedra	I tend to use properties files for configuration data that I know won't change
10:16	abedra	:)
10:17	oelewapperke	if you know it won't change don't you want the compiler to optimize it out ?
10:18	abedra	oelewapperke, I mean won't change after it is configured
10:18	abedra	oelewapperke, I want it to be configurable by others though
10:19	abedra	that may not know how to edit a clj file
10:28	tufflax	Does bound-fn only install the bindings in the lexical scope?
10:30	stuartsierra	bound-fn is for dynamic, not lexical scope
10:31	raek	tufflax: if I understand it correclty, it should remember the values of the dynamically bound vars in the dynamic scope where bound-fn is called. when the resulting fn is called, the body is evaluated in a dynamic scope with those remembered bindings in place
10:33	raek	(def x 1) (def f (binding [x 2] (bound-fn [] x))) (f) => 2
10:35	tufflax	Hm what are the dynamically bound vars? How come (future-call (bound-fn [] (println "arst"))) prints in my repl but not (future-call (bound-fn [] (fn-which-calls-println))), and not (future (println "asr"))
10:39	raek	tufflax: vars that are intended to be dynamically rebound are those named with earmuffs (there is no need for special naming of constants in clojure, since vars should not be metutaded for maintaining application state)
10:40	raek	they are often used to as "implicit" arguments to functions
10:41	raek	e.g. print prints to the stream in out. to print to another stream, one temporarily and dynamically rebinds the value: (binding [out ...] (print ...))
10:41	tufflax	what about earmuffs on global refs and atoms btw?
10:42	raek	sip them if you don't intend to rebind them with 'binding'
10:42	tufflax	ok
10:42	raek	in (def r (ref ..)), r is a constant (it always points to the same ref)
10:42	tufflax	i see
10:43	raek	tufflax: anyway, there should be no difference in writing (binding [out ...] (print ...)) and (binding [out ..] (function-that-calles-print ...))
10:43	tufflax	hmm, strange :p
10:44	raek	...unless that function was made with bound-fn
10:45	raek	(boudn-fn* f) can be seen as (fn [& args] (binding [...all the bound vars...] (apply f args)))
10:48	raek	s/metutaded/mutated/ (in one of my previous lines)
10:49	tufflax	i made a new fn that prints, and it worked as you described, but not my fns in my pasted code hmm. (yes i evaluated them again)
10:49	Cozey	Hmm. I'd like to add a template name to a compojure route - so a wrapper function can render route's response with a proper temlate... I thought maybe about using metadata for this, but I'm not sure how to implement this. any ideas?
10:50	tufflax	maybe it has to do with where they were evaluated... or something :p
10:51	tufflax	i mean, defined
10:52	raek	tufflax: yes, bound-fn remembers the bindings from the point where it is evaluated
10:53	raek	tufflax: also, there are debuggers for clojure: http://georgejahad.com/clojure/cdt.html
10:54	tufflax	I mean defined as in where the defn of the fn that prints where evaluated, all bound-fns were evaluated in the same repl
11:01	tufflax	So, to clarify: I write (ns some.ns) (defn test [] (println "dsrtd")) in a file and eval it. Then in the repl in ns some.ns i write (defn test2 [] (println "rst")). Then in the same repl (future-call (bound-fn [] (test))) does not print but (future-call (bound-fn [] (test2))) does
11:01	tufflax	gotta go afk a bit...
11:08	raek	tufflax: what does #'test return in the repl? #'some.ns/test or #'clojure.core/yesy
11:08	raek	*test
11:09	raek	I think the file was somehow not loaded correctly and you're accidentally using clojure.core/test
11:23	tufflax	raek #'game.networking/test
11:25	raek	tufflax: so calling it directly work, but (future-call (bound-fn* test)) does not?
11:25	tufflax	yup
11:26	raek	and the body just contains (print "asdf")?
11:26	tufflax	or wait, that works
11:26	tufflax	but not (bound-fn [] (test))
11:26	tufflax	inside a future
11:26	raek	oh, of course...
11:27	tufflax	? :)
11:27	raek	(future ((bound-fn [] (test)))) expands to (future-call (fn [] ((bound-fn [] (test)))))
11:27	raek	the bound-fn ends up beining evaluated in the wwring thread
11:27	raek	*wrong
11:28	raek	(let [f (bound-fn [] (test))] (future (f))) should work though
11:29	raek	maybe it would be nice to have something like (bound-future ...code...)
11:29	raek	(defmacro [& body] `(future-call (bound-fn [] ~@body)))
11:30	tufflax	but wasnt that exactly what i did? dont u mean bound-fn*?
11:33	raek	tufflax: which example?
11:33	tufflax	the macro
11:34	raek	tufflax: you said that (bound-fn [] (test)) didn't work inside a future, so I assumed you were doing something like (future ((bound-fn [] (test))))
11:34	Cozey	Help! I cannot (require) clojure.contrib.condition - it can't find Condition__init method but it's there in the jar (clojure 1.3). How can I deal with that?
11:35	raek	Cozey: what toold do you use to start clojure?
11:35	Cozey	it's cake
11:35	Cozey	but let me check with normal repl
11:36	tufflax	raek Sorry for the confusion, what I was doing was exactly what I said when i said "So, to clarify". Also i still have not understood the difference between test and test2
11:37	raek	hrm. there shoudn't be any difference between them...
11:37	tufflax	:p
11:37	raek	(future-call (bound-fn [] ...))) should do the right thing
11:37	raek	tufflax: and you can call (test) and (test2) at the repl and both works?
11:38	tufflax	yes
11:38	raek	and it's not something basic, like that you forgot to save the file?
11:39	raek	also, you said this behaves differently when you are are using bound-fn* instead of bound-fn?
11:39	tufflax	yes
11:40	tufflax	well
11:40	tufflax	hmm
11:40	raek	,(macroexpand-1 '(bound-fn [] (test)))
11:40	clojurebot	(clojure.core/bound-fn* (clojure.core/fn [] (test)))
11:40	tufflax	it gets stranger
11:41	raek	tufflax: and you didn't miss the parentheses around (test)?
11:44	tufflax	raek hmm sometimes it works and sometimes not, maybe it's vims fault http://pastebin.com/qT3VXA4Y
11:45	raek	tufflax: another idea: try using println instead of print (and maybe even add a flush call). maybe this is a buffering issue.
11:46	tufflax	it is all printlns
11:48	raek	tufflax: does (defn test [] (println "arst") (flush)) make any difference?
11:49	tufflax	no, sometimes it still doesnt show up
11:50	tufflax	im getting tierd of this problem :p
11:50	tufflax	tired
11:52	Cozey	raek: some dependency pulled contrib 1.2 - that was the problem i think
11:54	Dantas	Hi all, reading a nice post about metaprogramming, im wondering what ns means (intern ns 'hello (fn [] println "Hello World"))
11:57	andrewclegg	dantas: that's your current namespace
11:57	raek	Dantas: it's a var that contains a reference to the current namespace of the repl
11:57	andrewclegg	user=> (in-ns 'clojure.core)
11:57	andrewclegg	#<Namespace clojure.core>
11:57	andrewclegg	clojure.core=> ns
11:57	andrewclegg	#<Namespace clojure.core>
11:57	andrewclegg	etc.
11:58	tufflax	So a final question about bound-fn; what gets bound in the new function? "bindings in effect" -- things bound with binding? But im not binding anything, and apparently out gets bound anyway. Does every regular var count as a binding in effect?
12:03	raek	tufflax: you didn't bind out yourself, but your repl runs inside a binding form that binds it
12:05	Dantas	raek: andrewclegg : the var symbol is ns ??
12:05	Dantas	is the var symbol ns ?
12:05	raek	something like (binding [in ..., out ...] (loop [] (-> (read) (eval) (println)) (recur)))
12:06	Dantas	thanks
12:06	raek	Dantas: in the source code, ns is an ordinary symbol, just like intern and fn are
12:06	Dantas	yeah, thank u
12:07	raek	the symbol names the var #'clojure.core/ns
12:08	raek	just like intern names the var #'clojure.core/intern
12:09	raek	tufflax: i.e. your repl (but not the threads you start) runs in a dynamic scope where out is rebound
12:13	tufflax	raek ok, so not every regular var counts then
12:13	tufflax	i guess that makes sense
12:14	tufflax	raek ok now im gonna get on with my project. thank you very much for always helping me
12:29	timvisher	is clojure.contrib.logging deprecated in favor of tools.logging?
12:29	stuartsierra	yes
12:29	timvisher	awesome
12:30	stuartsierra	Basically, everything in the monolithic clojure-contrib is deprecated after 1.2
12:30	timvisher	what if i'm on 1.2?
12:30	stuartsierra	up to you then
12:30	timvisher	so tools.logging should work even on 1.2?
12:31	stuartsierra	It should, yes.
12:31	technomancy	timvisher: that's what we're doing
12:31	technomancy	it's a lot more convenient than contrib.logging
12:37	mabes	technomancy: why do you find it more convenient? isn't it the same API as contrib.logging?
12:38	technomancy	mabes: no, it doesn't make you to call format everywhere
12:38	timvisher	is it easier to set logging levels in tools.logging?
12:38	technomancy	it takes var-args and makes a string for you
12:38	technomancy	timvisher: no, that is implementation-dependent. there's no way to do it across all the loggers tools.logging supports.
12:39	mabes	technomancy: ah, that is nice
12:39	timvisher	gotcha
12:40	timvisher	so is this information still relevent if i were to use tools.logging? http://www.paullegato.com/blog/setting-clojure-log-level/
12:41	technomancy	probably. I haven't used j.u.logging or apache logging.
12:42	timvisher	cool. thanks for the help
12:48	zitchie	list
12:48	zitchie	,(defn foo 'adsf')
12:48	clojurebot	Unmatched delimiter: )
12:52	wastrel	hi
14:46	pdk	,(/ 1000 16.0)
14:46	clojurebot	62.5
15:01	gfrlog	(fn [_] (do (foo) (bar))) is redundant, right?
15:03	wastrel	hi
15:04	ataggart	gfrlog: the do is redundant, yes
15:15	gfrlog	ataggart: thanks
15:15	gfrlog	wastrel: hi
15:18	wastrel	vampire day massacre
15:18	wastrel	sorry connection lag
15:27	bartj	why can't this be done:
15:27	bartj	(fmap #(+ 1 %) (for [i (range 10)] i))
15:27	chouser	what's fmap?
15:28	hiredman	,(doc fmap)
15:28	clojurebot	No entiendo
15:28	hiredman	clojurebot: jerk
15:28	clojurebot	No entiendo
15:29	hiredman	,(require 'clojure.contrib.generic.functor)
15:29	clojurebot	java.io.FileNotFoundException: Could not locate clojure/contrib/generic/functor__init.class or clojure/contrib/generic/functor.clj on classpath:
15:29	hiredman	,(require 'clojure.contrib.generic.functors)
15:29	clojurebot	java.io.FileNotFoundException: Could not locate clojure/contrib/generic/functors__init.class or clojure/contrib/generic/functors.clj on classpath:
15:29	hiredman	bleh
15:29	hiredman	there is an fmap somewhere in contrib vaguely like that
15:29	bartj	hiredman, yes, I know
15:29	bartj	Applies function f to each item in the data structure s and returns
15:29	bartj	a structure of the same kind.
15:30	hiredman	bartj: I wasn't pointing it out to you
15:30	hiredman	(for [i (range 10)] i) is redundant btw
15:31	hiredman	bartj: I imagine it doesn't work because there is no fmap implemented in that library for seqs
15:32	bartj	hmm
15:33	bartj	seems strange though
15:34	hiredman	fmap is not used very often, so missing functionality is too be exepected
15:35	bartj	raek, help?
15:35	stuartsierra	You want ordinary `map`
15:39	bartj	stuartsierra, yes, I guess, fmap is supposed to be applied to maps, etc only
15:39	bartj	stuartsierra, but, I am puzzled and curious why it doesn't work in this guess
15:39	bartj	*case
15:39	stuartsierra	because the author didn't implement it.
15:40	stuartsierra	`fmap` and the other c.c.generic stuff never went through a design review.
15:40	stuartsierra	They're not part of the language.
15:41	hiredman	bartj: the only real advantage of fmap is it can map over values in maps, most people are likely to just roll their own
15:42	hiredman	,(let [a {:a 1}] (zipmap (keys a) (map inc (vals a))))
15:42	clojurebot	{:a 2}
15:42	hiredman	,(into {} (for [[k v] {:a 1}] [k (inc v)]))
15:42	clojurebot	{:a 2}
15:42	hiredman	etc
15:43	bartj	stuartsierra, thanks!
15:43	bartj	hiredman, thank you for your time and help
15:45	zakwilson	I want a library for handling HTTP requests that lets me define handlers with typed arguments and define how the string parameters get converted in to the appropriate types. I don't know of such a thing and expect I'll have to write it myself, but I figured I'd ask first.
15:46	zakwilson	To give an example of the sort of code I want to write: (defhandler foo [id :int name :string img :file] ...)
15:49	hiredman	zakwilson: if defhandler is a macro you could just as easily use type hints instead of keywords, fwiw
15:51	hiredman	and I would suggest just generating a function that takes a map of params or whatever and returns a map coerced to the correct types
15:51	zakwilson	hiredman: Would type hints work if I wanted to make the type conversions a multifn dispatching on the name of the desired type where the return value might be an arbitrary data structure (e.g. a map representing a database row)?
15:52	hiredman	zakwilson: oh, the macro would examine the tag metadata on the symbol and do the right thing, it would look like type hints to the user, but they would never make it to the compiler
15:53	zakwilson	I'm also considering an optional throw-on-null - e.g. (defhandler foo [id [:int :required]] ...)
15:53	hiredman	that sounds more like validation
15:54	hiredman	(defvalidator foo ...) (defhandler bar ...) ((comp foo bar) some-request)
15:54	hiredman	er, reverse those
15:54	hiredman	(comp bar foo)
15:54	zakwilson	Yes. I'm not sure if I should conflate validation and ensuring that the handler doesn't have to worry about types in its body.
15:54	stuartsierra	When in doubt, separate.
15:56	zakwilson	Some languages treat null checks as part of the type system, and it seems like that might go well with this little island of static typing, but it does sound more flexable to have separate validators that can do anything.
16:04	zakwilson	The other question that comes to mind for me is whether and how to deal with session variables (sandbar stateful-session). My instinct is to not do it, but it could be kind of cool to have handlers take everything a web request might need to deal with as arguments.
16:06	hiredman	the functional way to do it would be with the state monad
16:07	hiredman	but mentioning that tends to make people break out in hives
16:08	pjstadig	clojure is not a pure functional language, you're free to use vars, atoms, etc.
16:08	pjstadig	that's why they're there
16:09	gfrlog	there are still best practices about when to use them though
16:09	pjstadig	yeah but there is no reason to have to use monads
16:10	pjstadig	rich took the practical road with concurrently semantic reference types
16:10	pjstadig	but you're free to use monads too if you want
16:10	pjstadig	what do i care?
16:10	hiredman	pjstadig: monads and mutable reference types are orthogonal
16:11	pjstadig	how so?
16:11	pjstadig	you're talking about mutable state
16:11	robbe-	If I make-array an array, and share it amongst thread, the array is copied I presume? i.e. if I want to make changes that should be reflected in other threads, I could use e.g. an array of refs?
16:11	hiredman	gotta run, rides here
16:11	robbe-	threads*
16:12	pjstadig	an array shared amongst threads will need some kind of locking
16:13	robbe-	That means the array is shared amongst threads, and no thread-local copy is made when a thread starts
16:13	pjstadig	i guess you could use an array of refs, but why not a vector
16:13	robbe-	right?
16:13	pjstadig	right no copying unless you do it
16:13	robbe-	thanks
16:14	robbe-	Considering alternatives at the moment. :-)
16:15	robbe-	Ah, I the requirement (not set by me) is O(1) access time, and clojure documentation states: "Vectors support access to items by index in log32N hops."
16:15	robbe-	Which is still plenty fast, but not constant time.
16:16	robbe-	And I can't change the requirements because it's a school assignment. :P
16:16	gfrlog	robbe-: you could argue that it is constant, given bounds on the size of the computer and all that
16:17	gfrlog	although I suppose by that argument any halting algorithm is constant :-/
16:17	robbe-	Indeed, but I expect the assistant to be nitpicking about it. I'll just ask to be sure.
16:17	gfrlog	clojure is not a good tool for school assignments that are biased against clojure
16:18	raek	robbe-: I'd say: renegotiate the O(1) requirement to include the practically constant time of clojure's data structures
16:18	robbe-	Yeah, but you probably mean "not constant but incredibly low bound, which is in this case low enough to be considered constant." or something along those lines. :-)
16:19	robbe-	Thanks for the advice. :-) Will mail him shortly.
16:19	raek	for a vector with 2^32 elements, the lookup requires 6 or 7 steps, IIRC
16:19	gfrlog	robbe-: Also make the case that arrays are not quite constant either due to caching and paging and such
16:19	raek	and for hash-maps you can't have more than 2^32 distinct hashes
16:20	gfrlog	raek: what's the collision algorithm?
16:20	robbe-	gfrlog: <devils advocate> but you can fix that with more ram
16:20	gfrlog	robbe-: the point is that in practice the distinction is quite murky
16:21	robbe-	I agree. Just trying to come up with anything he might have to counter the arguments. :-)
16:21	gfrlog	well I imagine it's his perogative to say "cuz I sed so"
16:22	robbe-	:D
16:22	robbe-	Well, I have to go now, it's getting late over here, so I should mail and go to bed. Thanks everyone for the input.
16:23	gfrlog	enjoy your time zone
16:23	robbe-	Surely am, 20 past 10 pm and still light outside, I like that. Winter sucks though. :-)
16:24	gfrlog	enjoy your latitude as well then :)
16:25	raek	gfrlog: I think it has a special kind of node in the tree for hash collisions that linearly searches through an array of the objects with the same hash value
16:26	gfrlog	raek: that sounds like it would allow >(2^32) entries then, just with quickly decreasing performance?
16:26	raek	yeah, that's true
16:27	pjstadig	yes
16:28	pjstadig	so if you have a terrible hashing algorithm the data structure is actually log(n)
16:28	gfrlog	I'd think it'd be O(n)
16:29	gfrlog	if "terrible" means "worst case"
16:29	gfrlog	given raek said "linearly searches"
16:30	pjstadig	hehe
16:30	pjstadig	oops
16:30	pjstadig	i meant O(n)
16:30	raek	gfrlog: https://github.com/clojure/clojure/blob/master/src/jvm/clojure/lang/PersistentHashMap.java#L719
16:30	raek	https://github.com/clojure/clojure/blob/master/src/jvm/clojure/lang/PersistentHashMap.java#L784
16:31	ordnungswidrig	,(extend Object (ns-resolve ns (defprotocol Z)))
16:31	clojurebot	DENIED
16:31	ordnungswidrig	(extend Object (ns-resolve ns (defprotocol Z)));; this works
16:32	ordnungswidrig	(extend Object (ns-resolve ns (defprotocol Z (x [this]))){:x (constantly 42)}) ;; -> #'user/Z is not a protocol
16:32	ordnungswidrig	any clues
16:34	pjstadig	i believe defprotocol returns a symbol
16:34	pjstadig	not the protocol
16:35	pjstadig	oh
16:35	pjstadig	right but you're trying to resolve
16:35	pjstadig	ah but there is a #' there
16:37	pjstadig	yeah add an @ in front of the defprotocol
16:37	pjstadig	err
16:37	pjstadig	the ns-resolve
16:37	pjstadig	(extend Object @(ns-resolve ns (defprotocol Z (x [this]))){:x (constantly 42)})
16:37	ordnungswidrig	pjstadig: thanks, nice catch.
16:56	TimMc	Well, I wrote a thing to do dynamically scoped "binding" in Java.
16:56	TimMc	Now I'm passing it around for my coworkers to poke holes in. :-P
17:01	gfrlog	TimMc: wut
17:06	raek	TimMc: does it use ThreadLocal? http://download.oracle.com/javase/6/docs/api/java/lang/ThreadLocal.html
18:29	TimMc	raek: InheritableThreadLocal, yeah.
18:30	TimMc	I'll see if I can get permission to post it -- I coded it on the clock, so by default it is company code.
18:30	TimMc	My test rig for multi-threaded well-behavedness is amusing.
18:31	TimMc	gfrlog: try { LocalConfig.push("key", "val"); /* stuff */ } finally { LocalConfig.pop("key") }
19:06	miwillhite	hey all…anyone got an app running on Heroku yet?
19:58	miwillhite	hey…I'm trying to run my clojure app with `lein run -m stammer.core` (relevant files: http://pastebin.com/wfQJ3C9c)…but I keep getting the exception in the pastebin…any ideas? (using lein 1.5.2)
20:00	sritchie	hey all -- what protocol would I extend to cover all sequences?
20:00	sritchie	I had thought it would be aseq, but this misses vectors
20:01	hiredman	vectors are not seqs
20:02	sritchie	I think iterable is what I'm looking for
20:18	miwillhite	can I access environment variables in clojure? (like RACK_ENV in ruby for example)
20:18	technomancy	,(System/getenv "HOME")
20:18	clojurebot	java.security.AccessControlException: access denied (java.lang.RuntimePermission getenv.HOME)
20:19	miwillhite	awesome thanks!
20:21	technomancy	sure
20:23	__name__	,(Runtime.getRuntime)
20:23	clojurebot	java.lang.ClassNotFoundException: Runtime.getRuntime
20:23	__name__	,(Runtime/getRuntime)
20:23	clojurebot	#<Runtime java.lang.Runtime@12996d6>
20:24	__name__	,((. Runtime/getRuntime availableProcessors))
20:24	clojurebot	java.lang.Exception: Unable to find static field: getRuntime in class java.lang.Runtime
20:24	__name__	,((. Runtime getRuntime availableProcessors))
20:24	clojurebot	java.lang.Exception: Unable to resolve symbol: availableProcessors in this context
20:24	__name__	,((.. Runtime getRuntime availableProcessors))
20:24	clojurebot	java.lang.ClassCastException: java.lang.Integer cannot be cast to clojure.lang.IFn
20:24	__name__	,(.. Runtime getRuntime availableProcessors)
20:24	clojurebot	1
20:25	__name__	Sorry.
20:42	TimMc	__name__: You can also privmsg clojurebot
20:43	__name__	TimMc: I know.
20:43	__name__	I always thought it would work and wanted to share the result.
20:43	__name__	#scala's bot still is cooler.
20:43	__name__	It gives you 42 :)
20:44	TimMc	weird
20:52	korny	hi folks - is there something like partition-by that splits a seq by a boolean fn into two seqs, rather than repeatedly splitting?
20:52	korny	&(partition-by odd? [1 2 3 4 5 6])
20:52	sexpbot	⟹ ((1) (2) (3) (4) (5) (6))
20:52	korny	whereas I'd like ((1,3,5)(2,4,6))
20:53	hiredman	,(group-by odd? (range 10))
20:53	clojurebot	{false [0 2 4 6 8], true [1 3 5 7 9]}
20:53	korny	hiredman: thanks
20:59	pickles	I am about to tear my hair out over this: (def x [obj obj]), (.method (first x)) -> true, (map '.method x) -> [nil nil]
21:01	hiredman	do you understand what quoting a symbol does?
21:01	brehaut	,(map #(.toString %) [(Object.) (Object.)])
21:01	clojurebot	("java.lang.Object@1c0ae76" "java.lang.Object@fde7ba")
21:02	pickles	hiredman: i have an idea, but no, i do not
21:02	brehaut	,(type 'foo)
21:02	clojurebot	clojure.lang.Symbol
21:02	TimMc	,(type '.method)
21:02	clojurebot	clojure.lang.Symbol
21:02	TimMc	grah, I am slow
21:03	brehaut	pickles: symbols are ifn's, so it can be passed it
21:03	brehaut	as an argument to map
21:03	brehaut	,(ifn? '.method)
21:03	clojurebot	true
21:03	brehaut	but its not a handle on a method
21:03	hiredman	pickles: http://www.apl.jhu.edu/~hall/lisp/Lisp-Quote-Hints.text is written for common lisp, but similar enough
21:04	brehaut	,(map 'foo [{'foo 1} {'foo 2}])
21:04	clojurebot	(1 2)
21:04	pickles	when I do (map .terminates? x) i get an exception
21:04	hiredman	correct
21:05	TimMc	pickles: .method is a Java interop form
21:05	hiredman	because the name '.terminates?' is not bound locally or globally to anything
21:05	TimMc	It can't be passed around.
21:05	pickles	mmm
21:05	TimMc	pickles: However, #(.method %) can be. :-)
21:05	hiredman	(macroexpand '(.foo bar))
21:05	pickles	so i need to do a lambda
21:05	hiredman	,(macroexpand '(.foo bar))
21:05	clojurebot	(. bar foo)
21:05	pickles	aha
21:05	pickles	i c
21:08	pickles	many thanks all
21:08	pickles	hopefully i'm through being a n00b for the day
21:16	pickles	rock on, it's working, thanks again!
21:52	leif-p	Hi. Anyone know how to make clojure.pprint print metadata like (set! print-meta true) in the REPL does?
22:21	crowbar	Can someone please give me feedback on how I'm handling a point in 3d space? I'm currently trying to hold it as a map with {:x 1 :y 10 :z -4}. To simplify creation of a 3d point I have (defn point [a b c] {:x a :y b :z c}). Is this idiomatic?
22:26	chouser	crowbar: yep, it's fine. Also consider defrecord
22:30	crowbar	thanks chouser. exactly what I was after.
23:07	no_mind	I have some functions returning hash maps. I want to add the returned hash-maps to a sequence and return the sequence. How do I do this ? When I try to use concat or conj or cons, the returned sequence breaks hash-map elements into smaller sequqnces, something I do not want
23:08	hiredman	it means you are doing something that calls 'seq' on your maps
23:08	hiredman	(seq {:a 1})
23:08	hiredman	,(seq {:a 1})
23:08	clojurebot	([:a 1])
23:11	no_mind	hiredman: yes and I want to know how to avois this ?
23:12	no_mind	hiredman: I want result like this ({:a 1, :b 2} {:c 3, :d4}) where {:a 1 :b 2} and {:c 3, :d 4} are returned from two different functions, lets say abc and xyz
23:22	bpr	no_mind: (conj existing-sequence (seq hash1 hash2 ...))
23:23	hiredman	bpr: nope
23:23	bpr	?
23:23	bpr	then list rather than seq?
23:23	hiredman	bpr: what does seq do? how many args does it take?
23:24	bpr	yeah then list
23:26	bpr	lol, god i feel silly now
23:27	bpr	hiredman: thanks for pointing that out
23:48	bdesham	hi all
23:48	bdesham	I'm trying to write a function like flatten, except that it works with sets instead of sequences
23:48	bdesham	I came up with this: https://gist.github.com/1005837
23:49	hiredman	,(flatten #{#{:a} #{:a}})
23:49	clojurebot	Duplicate key: #{:a}
23:49	hiredman	bleh
23:49	hiredman	,(flatten #{#{:a} #{#{:a}}})
23:49	clojurebot	()
23:49	hiredman	huh
23:49	bdesham	it works fine, except that I'd like some way to make sure there are no duplicates in the input... for example, (flatten-sets [:a :b :a]) should generate some kind of output that lets me know there's a problem
23:49	bdesham	hiredman: right ;-) I tried that
23:50	hiredman	,(set (remove set? (tree-seq set? seq #{#{:a} #{#{:a}}})))
23:50	clojurebot	#{:a}
23:50	bdesham	,(filter (complement set?) (rest (tree-seq set? seq (set [:a #{:b} #{:c #{:d #{:e :f} :g}} :h)))))
23:50	clojurebot	Unmatched delimiter: )
23:50	bdesham	damn
23:51	hiredman	,(set (remove set? (tree-seq set? seq #{#{:a} #{#{:a}} #{:a :b :c {:d :e :f}}})))
23:51	clojurebot	3
23:51	bdesham	,(filter (complement set?) (rest (tree-seq set? seq (set [:a #{:b} #{:c #{:d #{:e :f} :g}} :h])))))
23:51	clojurebot	(:c :g :d :f :e :a :b :h)
23:51	bdesham	,(set (remove set? (tree-seq set? seq #{#{:a} #{#{:a}} #{:a :b :c #{:d :e :f}}})))
23:51	clojurebot	#{:a :c :b :f :d :e}
23:51	hiredman	clojurebot: seriously?
23:51	clojurebot	I don't understand.
23:51	bdesham	lol
23:52	bdesham	hiredman: that seems to do the same kind of thing
23:53	bdesham	but I'd like something that would detect duplicates, no matter how deeply nested they are
23:54	hiredman	creating a set removes duplicates
23:56	bdesham	I guess I can check for duplicates at top level with (= v (vec (set v)))
23:56	bdesham	but I could do that with distinct? just as easily
23:56	bdesham	hmm...
23:58	hiredman	sets cannot contain duplicates
23:58	hiredman	thats what a set is
23:59	hiredman	,(conj #{:a} :a)
23:59	clojurebot	#{:a}
23:59	bdesham	hiredman: right
23:59	hiredman	so you cannot possibly check for duplicates in a set, because there will never be any