#clojure – 2010-06-25

2010-06-25

00:58	vIkSiT	hi all
00:59	vIkSiT	does anyone recommend a way to do a merge on 2 lists?
01:18	Raynes	vIkSiT: What do you mean by "merge"?
01:20	alexyk	how do you error out with a message?
01:22	vIkSiT	Raynes, ah, something like doing a merge sort, except you're taking the union of 2 lists
01:22	Raynes	(throw (Exception. "message"))
01:22	vIkSiT	Raynes, basically, the imperative model has you parallely iterating through the 2 lists, doing a compare, and then adding elements in order..
01:23	vIkSiT	functionally, it looks like a combo of loop-recur and first/last would work.. but not sure if there's a known way to do it
01:32	alexyk	Raynes: was there an (error ...) or something?
01:48	Raynes	alexyk: Not that I know of. There is one in Haskell... It probably wouldn't be very hard to simulate it.
01:49	alexyk	Raynes: yeah. failwith in ocaml. I do all three at once... mixups
01:50	Raynes	alexyk: It might be as simple as throwing a (println "Omgerrors!") in there.
01:51	alexyk	I just do an assert for now after such
01:51	alexyk	it's zomgerrors btw
01:51	Raynes	Indeed.
02:47	jeffhill	Hello! I'm hacking about with generating lazy sequences, and I wanted something like iterate but with something like recursive arguments... so I built a macro around lazy-seq. It seems to work fine, but I feel like I've done something non-idiomatic. Is there anything I'm missing here? Link to source: http://paste.lisp.org/display/111831
03:16	mmarczyk	technomancy: ping! about Lein -- responded to your comment on GitHub
03:17	mmarczyk	gotta be off now, though...
03:56	AWizzArd	,(vec (.getBytes (str (char 55366))))
03:56	clojurebot	[63]
03:56	AWizzArd	wtf
04:00	neotyk	,(java.lang.Character/MAX_VALUE)
04:00	clojurebot	\ï¿¿
04:02	AWizzArd	That's 0xffff
04:03	neotyk	just noticed that is is a range
04:03	AWizzArd	The thing is: when I .getBytes on my string (be it latin1, utf-8, utf-32, etc) I get something different from (map int my-string)
04:04	AWizzArd	I can't (String. (.getBytes my-string "UTF-8") "UTF-8") to get the same result
04:04	AWizzArd	(= my-string (String. (.getBytes my-string "UTF-8") "UTF-8")) ==> false
04:04	neotyk	,(str (char 55366))
04:04	clojurebot	"?"
04:04	cschreiner	is there a way to put @ inside a symbol-name?
04:04	neotyk	,(str (char 63))
04:04	clojurebot	"?"
04:05	AWizzArd	It's fine that it is a questionmark, but I would prefer that it used the same number
04:05	cschreiner	this is relation to generation css; like the @font-face tag
04:05	cschreiner	I cannot do :@font-face
04:06	AWizzArd	,(let [s (apply str (repeatedly 5000 #(char (rand-int 65000))))] [(count s) (count (.getBytes s "LATIN1"))])
04:06	clojurebot	[5000 5000]
04:06	AWizzArd	,(let [s (apply str (repeatedly 5000 #(char (rand-int 65000))))] [(count s) (count (.getBytes s "LATIN1"))])
04:06	clojurebot	[5000 4999]
04:06	cschreiner	(symbol ":@font-face") works though
04:08	cschreiner	nevermind, it's okay
04:09	neotyk	AWizzArd: sorry I wan't be able to help
04:10	AWizzArd	I was just surprised to see such behaviour.
04:11	neotyk	~str
04:11	clojurebot	duck-streams is now clojure.contrib.io
04:11	neotyk	,(source str)
04:11	clojurebot	java.lang.Exception: Unable to resolve symbol: source in this context
04:13	neotyk	it calls .toString on result of (char ...) and that most likely does reencoding
04:15	AWizzArd	Over there in #Java they had a tip, I will study it now.
04:39	dsop	hmm what is a good tool to generate documentation like the ones on clojure.org ?
04:52	tomoj	autodoc
04:54	dsop	tomoj: thanks
04:57	cschreiner	Looking for a more reasonable way to do:
04:57	cschreiner	,(let [items (list 1 2 3 4)] (zipmap items items))
04:57	clojurebot	{4 4, 3 3, 2 2, 1 1}
04:58	cschreiner	or is this sound clojure?
05:06	sergey_miryanov	hi all!
05:07	sergey_miryanov	i have a question about into for transient data - how to use it? if i use (into (transient []) (range 10)) i got an exception...
05:10	AWizzArd	sergey_miryanov: yeah, transients could need a bit more work. For example 'find' does also not work on transient maps.
05:12	sergey_miryanov	AWizzArd: clear
05:15	Licenser	aloa
05:22	tomoj	cschreiner: that seems fine to me
05:22	tomoj	not sure why you'd want to do it..
05:30	cschreiner	tomoj: I just want to create a map from a list where the keys are eq to the vals
05:31	tomoj	right, but why?
05:32	cschreiner	that's another story, not really important here
05:32	tomoj	going to replace some of the values?
05:32	cschreiner	Iit's just for personal testing
05:33	cschreiner	if I would go for (into {} ...) I could'nt quite figure out how
05:34	tomoj	,(let [s (range 10)] (into {} (map vector s s)))
05:34	clojurebot	{0 0, 1 1, 2 2, 3 3, 4 4, 5 5, 6 6, 7 7, 8 8, 9 9}
05:34	tomoj	your way is fine
05:34	cschreiner	guess so ;)
06:28	AWizzArd	Anyone here with a SSD?
06:31	AWizzArd	I would like to see what the performance difference is.
06:32	AWizzArd	Please start a Server JVM and (import '(java.io FileOutputStream BufferedOutputStream DataOutputStream)) and define a string of one million random chars (def s (apply str (repeatedly 1000000 #(char (rand-int 55000)))))
06:33	AWizzArd	And then simply (with-open [dos (DataOutputStream. (BufferedOutputStream. (FileOutputStream. "/data" false)))] (time (dotimes [i 100] (.write #^DataOutputStream dos (.getBytes #^String s "UTF-8") 0 1000000)))) ; writing the string a hundred times to disk
06:54	bartj	Any Russian's here? I want to ask them a slightly off-topic question (:
06:55	bartj	regarding phone-numbers in Russia (:
06:57	AWizzArd	bartj: maybe #C++ ?
06:59	bartj	AWizzArd: why so ?
06:59	AWizzArd	ja nje ponimaju po russki
07:01	sergey_miryanov	i'm a russian
07:06	bartj	AWizzArd: can you please tell me why phone numbers on this page are 12 digits? - http://www.77rosfirm.ru/catalog/card/77.1188629
07:09	silveen	Happy midsummer everyone
08:02	AWizzArd	,"\ud80e\udc80"
08:02	clojurebot	"𓢀"
08:17	mmarczyk	AWizzArd: ouch... exposing my Emacs's failure to display unicode characters properly :-(
08:19	AWizzArd	mmarczyk: yes sure :)
08:20	AWizzArd	,(.codePointAt "\ud80e\udc80" 0)
08:20	clojurebot	80000
08:20	AWizzArd	This is a string of just one character, with Unicode code point 80000
08:20	AWizzArd	,(count "\ud80e\udc80") ;-)
08:20	clojurebot	2
08:20	mmarczyk	is there such a code point?
08:20	AWizzArd	yes
08:21	mmarczyk	oh, look, a bug! :-)
08:21	AWizzArd	no, not a bug
08:21	AWizzArd	this is correct behaviour
08:21	mmarczyk	huh?
08:21	AWizzArd	internally 2 chars are used, to produce one real character
08:21	AWizzArd	this char does not fit into 2 bytes
08:22	AWizzArd	code points exist up to 0x10FFFF = 1114111
08:22	mmarczyk	I'd still kinda sorta expect "logical" characters to be counted
08:23	mmarczyk	,(.length "\u80e\udc80")
08:23	clojurebot	Invalid character length: 3, should be: 4
08:23	mmarczyk	,(.length "\ud80e\udc80")
08:23	clojurebot	2
08:23	mmarczyk	...not going to argue with the host though
08:24	AWizzArd	yes, two 2-byte-chars are used, and they represent one unicode char
08:24	AWizzArd	Was a Java design descision and makes mostly sense.
08:24	mmarczyk	was it before the rise of utf-8?
08:24	mmarczyk	just out of historical curiosity
08:25	AWizzArd	Otherwise they would have to encode each char with 4 bytes under the hood, which is quite wasteful, as 99,99999% of all texts don't contain such very specific chars
08:25	AWizzArd	First Unicode was designed to eat 65k chars
08:25	AWizzArd	Java incorporated this
08:26	chouser	it may not be an implementation bug, but it might well be a design bug
08:26	mmarczyk	I suppose a variable-length encoding would be terrible for indexing into strings etc.
08:26	AWizzArd	Then, later, the unicode consortium discovered this is not enough.
08:26	AWizzArd	They now wanted to support over a million characters.
08:26	mmarczyk	"nobody should ever need more than 65k chars" -- sounds familiar somehow ;-)
08:27	AWizzArd	It could be a design bug, but it probably isn't one
08:29	mmarczyk	actually, how does one replace a "narrow" character with a "wide" character in an array of characters? copy the whole thing while adding an extra char in the middle...?
08:29	AWizzArd	The real problem is that the api doesn't deal with malformed strings.
08:29	AWizzArd	When you want to traverse a(ny) string correctly, you can not traverse its (seq)
08:30	AWizzArd	You need to look at the .codePointAt i
08:30	AWizzArd	and if that value is > than (int Character/MAX_VALUE), then you need to an extra increase of i in the loop
08:30	AWizzArd	+do
08:30	mmarczyk	oh, you mean it's a clojure.lang.StringSeq problem...?
08:30	AWizzArd	no
08:30	AWizzArd	there is no problem
08:31	AWizzArd	The default behaviour is to traverse a string with charAt
08:31	AWizzArd	This works in nearly 100% of all cases.
08:31	mmarczyk	not quite though?
08:31	AWizzArd	It's good.
08:32	AWizzArd	Though, it is not really a correct approach.
08:32	mmarczyk	ok, so an alternative would be a "CodePointSeq" on a string
08:32	AWizzArd	yes, for example
08:32	mmarczyk	which couldn't have characters as members, I guess
08:32	mmarczyk	interesting!
08:32	AWizzArd	In principle such a CodePointSeq could be half the (count of-the-string).
08:33	AWizzArd	(<= (/ (count s) 2) (count CodePointSeq) (count s))
08:34	mmarczyk	wondering I might be tripped up by this: do Chinese characters fit into the lower 65k part of Unicode then?
08:34	AWizzArd	I think all or most of all fit in there
08:34	mmarczyk	ok
08:34	chouser	I don't see how this could possibly be ideal
08:34	chouser	There should be two things -- bytes and codepoints a.k.a. chars
08:35	AWizzArd	It really is not ideal. We need a good bit of inside knowledge if we wanted to write correct code.
08:35	chouser	having a string be made up of something that is neither isn't useful, is it?
08:35	AWizzArd	Not the best thing, I agree. But this allowed all Java programs to continue to run. Also it means that Strings are not even more hungry than they already are.
08:35	mmarczyk	is a big-C Character capable of holding a wide char?
08:35	AWizzArd	An empty string eats 30+ bytes or so
08:36	chouser	utf8 is teh bomb
08:36	AWizzArd	no, native char and the wrapper Character both internally use 2 bytes to represent a character.
08:36	mmarczyk	utf-8... would need to store an extra array of indices indicating where, say, every 32nd char starts
08:36	mmarczyk	for constant-time .charAt
08:38	AWizzArd	All this would be okay.. people who studied this topic can write correct code.. but there still remains one problem, and that is: malformed strings.
08:38	chouser	maybe constant-time .charAt isn't worthwhile
08:38	AWizzArd	Unfortunately those don't throw an exception.
08:39	mmarczyk	chouser: isn't that what enables StringSeq...? though maybe we could do without it
08:39	AWizzArd	While "\ud80e\udc80" is a perfect 2-char / 4-byte / 1-unicode-char string this one is nonsense: "\udc80\ud80e" (just reversed)
08:39	AWizzArd	,(vec (.getBytes "\ud80e\udc80" "UTF-8"))
08:39	clojurebot	[-16 -109 -94 -128]
08:39	mmarczyk	though constant time indexing also enables substring
08:39	AWizzArd	,(vec (.getBytes "\udc80\ud80e" "UTF-8"))
08:39	clojurebot	[63 63]
08:39	mmarczyk	I'm not letting anyone take away my substring :-P
08:40	AWizzArd	,[(map int "\ud80e\udc80") (map int "\udc80\ud80e")]
08:40	clojurebot	[(55310 56448) (56448 55310)]
08:40	chouser	mmarczyk: in it's current implementation, sure. If String's charAt weren't constant-time, you'd want other adjustments to its API
08:41	AWizzArd	Those designers of the spec decided that there have to be low surrogate chars and high surrogate chars, which expect something to their left or right
08:41	mmarczyk	chouser: I suppose, not sure what those would be though; .substring is more important anyway
08:43	AWizzArd	well, substring won't work on such strings
08:43	mmarczyk	AWizzArd: with utf-8 strings supported by array of indexes into the underlying byte array it would
08:43	AWizzArd	It works correctly when looking at the single chars, as in 2-byte-values.
08:43	mmarczyk	I think?
08:44	AWizzArd	Strings are internally represented by a char array.
08:45	AWizzArd	If you know for sure that your string contains no surrogate char at all, then subs and .substring work correctly and as expected.
08:45	AWizzArd	If you want to do a substring as a human would expect it, by visible characters, then you will have to do it manually :)
08:46	AWizzArd	Before each use of subs you can check if it is okay to use it: (re-find #"[\uD800-\uDFFF]" s)
08:46	mmarczyk	yeah, I understand
08:46	mmarczyk	but you could theoretically implement strings differently
08:47	AWizzArd	if those 2048 chars are not included in the string (and this should be true for nearly 100% of all strings), then using subs is fine
08:47	mmarczyk	the above was referring to such hypothetical implementation
08:47	AWizzArd	mmarczyk: yes, and in such a case that would work
08:47	AWizzArd	Or they could make char being 3 or 4 bytes wide.
08:48	mmarczyk	I think I'll make figuring out the performance/memory hit into a little project :-)
08:48	AWizzArd	well sure, that could be interesting
08:50	mmarczyk	thanks for the conversation, this is interesting stuff
08:50	AWizzArd	The very easy solution is to use 3 bytes per char. That allows you to represent up to 16 mio different chars, while the current unicode standard supports just 1,114,112 characters
08:50	AWizzArd	But interesting would be a compressed string, so that "aaaaaaaaaa" would just eat around 10 bytes, not 30
08:51	chouser	seriously. UTF8 is awesome
08:51	mmarczyk	presumably different things are possible on big proper boxen than in the embedded realm
08:52	AWizzArd	I am still pissed, because I am now writing a very fast serialization lib for Clojure and can't rely perfectly on .getBytes, which will produce wrong results for random strings.
08:52	chouser	I mean, it was nice when strings could have fixed-sized chars, but that era has passed.
08:52	mmarczyk	extra byte per char in J2ME -> "surely you are in jest?"
08:53	mmarczyk	chouser: yup, and currently we live in the era where strings cannot / must not have fixed-sized chars, yet they still do ;-)
08:53	mmarczyk	somebody make the next one hurry up
08:57	mmarczyk	alright, I need to be off for now... see you later
09:16	jcromartie	I wish every programmer realized how evil mutable state is
09:16	jcromartie	then I would have apps that set session variables and trigger a chain of event handlers that results in reading the session variable, in the same HTTP request
09:17	jcromartie	wouldn't
09:25	lpetit	hello there.
09:28	lpetit	Regarding the recent "resurgence" of newbie questions such as "why does contains? has the semantics it has ..." et al., I've been thinking about some ways to make it both easy for the newbie to get the answer by himself, and easy to the newbie to contribute back where he found the info
09:30	lpetit	Something directly usable from the REPL. e.g. some 'rationale' namespace. rationale functions would work on the same set of symbols (doc) and (find-doc) work on. rational content for vars/specialchars could be contained somewhere in the cloud (would need the network available).
09:30	lpetit	(rationale contains?) may return a string with an indexed list of texts.
09:31	lpetit	e.g. "(0) http://some.googlegroup.thread\n(1) Rich Hickey once said on #clojure ..."
09:31	lpetit	(rate contains? 0) would mean : me as a user found rationale 0 to be useful
09:32	lpetit	(unrate contains? 0) would mean: me as a user found rationale 0 particularly useless
09:32	lpetit	the 2 last functions would go to the cloud and and a "vote" for the given rationale
09:33	lpetit	finally, (add-rationale contains? "some free text") would contribute a new rationale
09:34	lpetit	Just thinking about it. Thoughts ?
09:37	jcromartie	i don't see the benefit of it over just sending them here or to the mailing list
09:41	lpetit	jcromatie: I was thinking about this as an easy way to accumulate knowledge (especially concerning the rationale of things: why is contains? the way it is ?), and, via the rate/unrate functions, to automatically sort good knowledge. In the end: time saver for both the newbie and the non-newbie. So more time for the non newbie to help the newbie on more interesting questions. But maybe I'm...
09:41	lpetit	...just dreaming up ?
09:41	lpetit	jcromatie: I thought that the immediate access of those functions via the "always open" REPL could help leverage these functionalities
09:42	lpetit	jcromatie: of course, a little help from clojure-bot could also help add knowledge directly from irc, just after having answered for the nth time to the same question
09:43	lpetit	I know the answer generally is in the ml archive and / or in the #clojure logs and / or somewhere on the wikis / blog posts, but it's kind of steganographied knowledge ...
09:45	eevar2	lpetit: imo the best way to learn the essentials is "programming clojure" + "the joy of clojure"
09:46	chouser	it would be nice to have something better. right now we have official docs that are mostly up-to-date, but pretty dense and minimal. And we have "live help" on irc and ggroup which is up-to-date and as verbose as needed.
09:46	stuarthalloway	has anyone ever written/seen a Clojure (into-bean bean map)
09:46	chouser	if you want something more verbose but less live, I think the main difficulty to overcome is keeping it up-to-date. All the books, and both wikis have this problem.
09:47	stuarthalloway	like clojure into but reflective to get stuff into random java beans?
09:47	chouser	stuarthalloway: I've seen unbean somewhere I think.
09:47	stuarthalloway	chouser: thx, works as a search term
09:48	chouser	into-bean's a better name
09:49	stuarthalloway	and implies the right semantics (starting with an existing bean)
09:49	chouser	'bean' may be misnamed as suggested in the ggroup thread. you found that?
09:49	stuarthalloway	yes
09:50	stuarthalloway	I am going to put into-bean into contrib somewhere
09:57	lpetit	chouser: I thought about somtheing totally "open to the community", just as a wiki with read/write access to anybody, but not a wiki, just 3 functions from the REPL to consult (rationale), add content (add-rationale), remove content (remove-rationale) (and/or rate/unrate). A simple "wikipedia" indexed by var
09:58	lpetit	chouser: but more an aggregator of links (per var) than containing the real content. But maybe it would just be yet another out of date source of info :-(
09:59	chouser	lpetit: well, it is a social problem not a techincal one. if lots of people use it and it's easy to update, it's likely to stay up to date.
10:00	chouser	perhaps having it in-repl would be enough to solve that.
10:00	lpetit	chouser: yes. And it's it does not cost too much to introduce the feature, ...
10:00	lpetit	chouser: yes, I bet that having it in repl could greatly help make the dream come reality
10:02	lpetit	chouser: I guess a minimalist approach could leverage the most useful part of the features very quickly
10:03	lpetit	there is no http client in contrib, right ?
10:05	hugod	lpetit: a repl based tool would probably reduce the amount of spam too.
10:05	hugod	I have the start of a repl/command line feedback tool that uses google appengine
10:06	hugod	lpetit: clojure.contrib.http.agent
10:06	lpetit	hugod: yay !
10:08	lpetit	hugod: I was testing the idea. I don't intend to work on this for real in the next few weeks. If you like the idea, feel free to go ahead !
10:08	hugod	I've been put off by appengine startup times though, 10s to 20s. I assume it gets better with volume, but...
10:09	lpetit	hugod: oh yes, this would be a breaker for using app engine, I guess
10:10	lpetit	hugod: the evil will be in the detail (as for what differenciates an iphone to an ordinary mobile phone). If it ain't fast (as in "instantaneous"), it won't be used.
10:11	hugod	lpetit: asynch requests help to keep it feeling fast
10:25	lpetit	hugod: can you tell us more 'bout your "repl/command line feedback tool that uses google appengine" ? Especially, I don't understand why it is tied to appengine. Just using GET/POST HTTP functions should suffice, no ?
10:26	hugod	lpetit: well, the data has to go somewhere, but it would be easy to move to some other storage
10:27	hugod	it would be trivial to change it to couchdb or similar
10:28	lpetit	hugod: ok, but can't you make the client/server protocol independent of the server's underlying storage techno ?
10:29	hugod	lpetit: of course the the protocol is independent of the storage
10:29	lpetit	hugod: sorry then
10:30	hugod	lpetit: I'm probably not explaining it well :)
10:31	hugod	lpetit: at the moment it uses hash tags to identitfy individual comments, and allows replies on particular tags
10:32	hugod	lpetit: but adding recoginition of clojure function names would be simple
10:32	lpetit	hugod: let us know when you've implemented it ! :)
10:32	hugod	lpetit: each user is identified by an opaque handle
10:33	hugod	lpetit; which the user can attach to aomething less anonymous
10:33	lpetit	hugod: I don't think we need to identify people. I'm thinking about a "no-config" "no-brainer" system for the "rationale" feature
10:34	hugod	lpetit: ranking would be per item then?
10:34	lpetit	hugod: well, yes
10:34	hugod	lpetit: I think allowing people to identify their comments would be an incentive for some
10:36	lpetit	hugod: I was more thinking about keeping it as simple as possible. You find something useful, you push it, done.
10:37	lpetit	hugod: or maybe a "loose" identification, not via an id of any sort, but just an optional argument to the function : (add-rationale "my text" "lpetit")
10:37	hugod	lpetit: I'm all for the simpler the better, at least initially
10:37	lpetit	,(System/getProperty "user")
10:37	clojurebot	nil
10:37	lpetit	,(System/getProperties)
10:37	clojurebot	#<Properties {java.runtime.name=Diablo Java(TM) SE Runtime Environment, sun.boot.library.path=/usr/local/diablo-jdk1.6.0/jre/lib/i386, java.vm.version=10.0-b23, java.vm.vendor=The FreeBSD Foundation, java.vendor.url=http://www.freebsd.org/java/, path.separator=:, java.vm.name=Diablo Java HotSpot(TM) Client VM, file.encoding.pkg=sun.io, sun.java.launcher=SUN_STANDARD, user.country=US, sun.os.patch.level=unknown, java.vm.spe
10:38	lpetit	,(System/getProperty "user.home")
10:38	clojurebot	"/home/hiredman"
10:38	lpetit	,(-> (System/getProperties) bean)
10:38	clojurebot	{:empty false, :class java.util.Properties}
10:38	lpetit	argh
10:39	lpetit	hugod: was thinking about maybe using the System env context, but I fear complaints for violation of privacy ....
10:40	hugod	lpetit: that's why I am using an opt-in scheme for identification
10:43	hugod	lpetit: I was aiming for a repl based version of something like uservoice
10:44	lpetit	hugod: what is uservoice ?
10:44	hugod	lpetit: uservoice.com
10:45	lpetit	hugod: ok
10:45	hugod	lpetit: but using #tags similar to thedeadline
10:45	lpetit	hugod: I must leave, keep up the good work, cu later !
10:46	hugod	lpetit: :)
10:46	jimduey	Hi folks. Anyone have a favorite webhost company to serve a site of static pages.
10:46	jimduey	?
10:49	bhenry	jimduey: nearlyfreespeech.net unless your volume will be big
10:49	jimduey	Thanks, I'll check it out. Minimal volume
10:59	arohner	before I start writing, are there any lein plugins that start a daemon process?
11:14	chouser	stuarthalloway: have you tried making the laprepl example in enclojure recently?
11:22	AWizzArd	,(instance? Long/TYPE (long 10))
11:22	AWizzArd	$(instance? Long/TYPE (long 10))
11:22	sexpbot	=> false
11:22	AWizzArd	Why is that?
11:23	AWizzArd	$[Long/TYPE Long (class? Long/TYPE) (class? Long)]
11:23	sexpbot	Command not found. No entiendo lo que estás diciendo.
11:23	AWizzArd	$(vector Long/TYPE Long (class? Long/TYPE) (class? Long))
11:23	sexpbot	java.lang.SecurityException: Code did not pass sandbox guidelines: (#'clojure.core/class? #'clojure.core/class?)
11:24	cemerick	what happened to clojurebot?
11:24	stuarthalloway	AWizzard: long doesn't promise it will return a Long/TYPE
11:24	AWizzArd	anyway, (identical? Long Long/TYPE) ==> false
11:24	stuarthalloway	and the assumption that such things hold is making rhickey crazy as he fixes overspecified tests in the various num branches
11:25	cemerick	In any case, the (long 10) is going to get boxed going into instance?
11:25	stuarthalloway	cemerick: I don't think the doc promises that either
11:26	stuarthalloway	if you call (long x) on the way into a Java interop form, you don
11:26	stuarthalloway	't want boxing
11:27	cemerick	Right, but if instance is a clojure fn, then everything is boxed -- modulo prim
11:33	arkh	What's a good way to loop over something x times or until (testfunc) returns false
11:34	hiredman	~ping
11:34	clojurebot	PONG!
11:36	chouser	,(for [i (range 10) :while (even? i)] i)
11:36	clojurebot	(0)
11:37	arkh	chouser: thank you
11:37	chouser	arkh: note that's a lazy seq, not a loop in the imperative sense
11:38	arkh	chouser: so wrap with (doall) ?
11:38	arkh	chouser: ... assuming I wanted the whole thing realized
11:38	chouser	you want side effects?
11:39	chouser	you could try doseq instead of for
11:39	chouser	,(doseq [i (range 10) :while (even? i)] (println "Mutate the universe:" i))
11:39	clojurebot	Mutate the universe: 0
11:41	arkh	chouser: doseq looks good; side effects are involved
11:46	cemerick	clojurebot: nice to have you back :-)
11:47	clojurebot	excusez-moi
11:47	cemerick	exactly
11:48	bhenry	chouser: i can't figure out why those code snippets don't display 0 2 4 6 8.
11:49	bhenry	ah nevermind
11:50	chouser	bhenry: right, for that you'd use :when instead of :while
12:02	bhenry	chouser: does :while withhold laziness so that (range 10) never sees the 1?
12:06	chouser	no, :while has to realize 1 in order to see that it's not even. But it won't realize any more than that.
12:07	chouser	,(for [i (iterate inc 0) :while (even? i)] i)
12:07	clojurebot	(0)
12:15	arohner	technomancy: did you get my github email last night?
12:16	technomancy	arohner: yeah, still working through a slight backlog on pull requests. hope to get to it this weekend.
12:17	arohner	technomancy: no problem. I had another thought, which is a plugin that hooks lein deps would work if src/ was on the lein classpath
12:17	arohner	technomancy: nm. something else must have been broken
12:30	defn	lancepantz: sorry if you already answered this, but did you get your JRuby + Rails + Clojure project up for viewing?
13:18	Kjellski	Hi there, I'm trying to build a function that blocks a seq like this: [1 2 3 1 4 5 1 6 7] into these: [[1 2 3] [1 4 5] [1 6 7]] with a signature like this (defn block-by [pred coll] ...) is there already something that does this?
13:19	AWizzArd	$(map vector (partition 3 [1 2 3 1 4 5 1 6 7]))
13:19	sexpbot	=> ([(1 2 3)] [(1 4 5)] [(1 6 7)])
13:20	mmarczyk	s/vector/vec/
13:20	chouser	Kjellski: http://clojure-log.n01se.net/date/2010-06-23.html#12:40
13:20	bhenry	kjellski do you want to make a new group whenever you get to something lower than the previous number or do you want to actually do it by 3's?
13:21	AWizzArd	or do you want to start a new group whenever you hit a 1?
13:21	bhenry	that too
13:21	AWizzArd	mmarczyk: identity would also be an option
13:22	chouser	he asked this a couple days ago
13:22	mmarczyk	AWizzArd: better not use map at all then
13:22	AWizzArd	chouser: how do you know this? You don't remember everything you read in here, do you? ;)
13:22	Kjellski	I want it to split the vector by 1, whatever pred is should split...
13:22	mmarczyk	$(doc split-with)
13:22	sexpbot	=> ------------------------- clojure.core/split-with ([pred coll]) Returns a vector of [(take-while pred coll) (drop-while pred coll)] nil
13:22	AWizzArd	mmarczyk: yesss lol
13:22	Kjellski	AWizzArd: exactly!
13:23	mmarczyk	AWizzArd: :-)
13:23	chouser	Kjellski: I posted a couple answers for you on Wednesday, after you left. :-P
13:23	AWizzArd	Kjellski: you can not answer an or-question with yes/no/exactly
13:23	Kjellski	mmarczyk: but split with looses the relation between the elements and groups them into one
13:23	mmarczyk	well, that's what the questions above are about...
13:24	chouser	,(map #(apply concat %) (partition 2 (partition-by #{1} [1 2 2 1 3 3 1 4 4])))
13:24	clojurebot	((1 2 2) (1 3 3) (1 4 4))
13:24	mmarczyk	there'll be different solutions depending on what you want to apply the predicate to
13:24	Kjellski	chouser: Oh sorry mate... but thanks for that, I think this would be useful as a standard funtion...
13:24	AWizzArd	mmarczyk: I was thinking of matrix transposition, such as (map list [1 2 3] [4 5 6] [7 8 9]) and started out with map, doh!
13:25	chouser	Kjellski: what would you name it?
13:25	mmarczyk	AWizzArd: ah, then prefixing your snippet with apply would work
13:26	Kjellski	chouser: something like "block-by" or "block-by-pred" ...? I would use this for parsing, but maybe the name would be good for other viewpoints?
13:27	Kjellski	,(map #(apply concat %) (partition 2 (partition-by #{1} [1 2 3 4 1 5 6 7 1 8 1 9])))
13:27	clojurebot	((1 2 3 4) (1 5 6 7) (1 8) (1 9))
13:28	Kjellski	chouser: what a dream... I'm always thinking way to complex to get to such solutions...
13:28	chouser	Kjellski: well, me too, pretty often. did you scroll up from that link and see my first solution?
13:28	mmarczyk	oh, something like Haskell's groupBy then?
13:28	mmarczyk	groupBy (\x y -> if y == 1 then False else True) [1,2,3,1,4,4,1,5,5]
13:29	mmarczyk	[[1,2,3],[1,4,4],[1,5,5]]
13:29	chouser	(doc group-by)
13:29	clojurebot	"([f coll]); Returns a map of the elements of coll keyed by the result of f on each element. The value at each key will be a vector of the corresponding elements, in the order they appeared in coll."
13:29	Kjellski	chouser: yes
13:29	mmarczyk	good function, to be sure, but we've already got a group-by which does something completely different
13:29	mmarczyk	ah, right.
13:29	mmarczyk	so not sure what to call it now...
13:30	chouser	it's closes to partition-by
13:30	chouser	closest
13:30	chouser	perhaps partition-starting-when :-)
13:30	Kjellski	mmarczyk: I thought of "block-by" but that could be confusing too...
13:30	mmarczyk	chouser: yeah
13:31	mmarczyk	shall we hold a naming contest? ;-)
13:31	Kjellski	chouser: what about "collect-blocks"
13:31	mmarczyk	or maybe call it partition-with and laugh diabolically each time somebody confuses it with partition-by :-P
13:31	Kjellski	mmarczyk: ^^
13:32	chouser	hm, I thought we already had a partition-with, but I guess not
13:33	Kjellski	,(map #(apply concat %) (partition 2 (partition-by #{1} [1 2 3 4 1 5 6 7 1 8 1 9 1 1 1 2 1])))
13:33	clojurebot	((1 2 3 4) (1 5 6 7) (1 8) (1 9) (1 1 1 2))
13:33	Kjellski	chouser: see the end?
13:33	chouser	ooo, you broke it!
13:34	chouser	,((fn [p s] (->> [nil s] (iterate (fn [[_ [_ & xs]]] (split-with (complement p) xs))) (partition 2 1) (map (fn [[[_ [i]] [r]]] (cons i r))) (take-while first))) #{1} [1 2 3 4 1 5 6 7 1 8 1 9 1 1 1 2 1])
13:34	clojurebot	((1 2 3 4) (1 5 6 7) (1 8) (1 9) (1) (1) (1 2) (1))
13:35	Kjellski	chouser: actually, that rocks
13:36	Kjellski	chouser: thanks
13:36	chouser	a bit of a mess, but you're welcome.
13:36	Kjellski	chouser: thinking: I´ll try to understand it the next two weeks ^^
13:37	chouser	I deny allegations of destructuring abuse
13:39	Kjellski	chouser: could you do me favour? I would love to see it indented the best, so I could try to figure out what it does...
13:39	chouser	yeah, sure.
13:40	rhickey	updated docs for equiv branch, ready for people to check it out: https://www.assembla.com/wiki/show/b4-TTcvBSr3RAZeJe5aVNr/Enhanced_Primitive_Support
13:41	chouser	Kjellski: http://gist.github.com/453174
13:42	Kjellski	chouser: thanks a lot!
13:47	chouser	Kjellski: just updated the gist to a simpler version
13:48	Kjellski	chouser: I´ll have a look...
13:52	Kjellski	chouser: this function comes in really handy happy
13:52	eivindgl	q
13:53	Kjellski	chouser: how nice to split logs with it... ^^
13:53	chouser	Kjellski: :-) great.
13:54	chouser	cemerick: I played with stack traces some yesterday, basing a lot on clojure.stacktrace
13:55	cemerick	chouser: any breakthroughs?
13:56	chouser	no, but I segmented the code differently -- have a fn that returns a seq of stack-trace lines, so that you can then filter, map, etc. to get what you want
13:56	chouser	then another fn that prints them once you've filtered
13:57	chouser	I'll keep playing with it, but I wanted to mention that your code was a nice starting place.
13:57	cemerick	my code?
13:57	chouser	didn't you write clojure.stacktrace?
13:58	cemerick	No -- I think SS did?
13:58	cemerick	I would never produce something so useful.
13:58	chouser	oh. sure enough named right there at the top. I wonder how I got confused.
13:59	chouser	sorry.
13:59	cemerick	no worries
13:59	Kjellski	there is no way to step through clojure code right? ^^ like good old debugging?
13:59	cemerick	People often think I'm responsible for awesome stuff. I just make sure I'm next to people that do real work, or I think out loud and often someone will do something I can take credit for. ;-)
13:59	chouser	anyway, I've got shorter stack by throwing away RestFn patterns that show up on every apply, and non-core .clj lines highlighted
13:59	cemerick	like, shell highlighted, or otherwise?
14:00	chouser	yeah, shell. ANSI color codes. I live in the 70s
14:00	chouser	or was that 80s
14:01	cemerick	Nah, I'm jealous of good shell highlighting, etc.
14:01	cemerick	Not jealous enough to actually use a shell for real work, of course.
14:01	chouser	I want to throw away multi-line patterns that show up on seq-walking. probably collapse compiler lines since I normally don't care.
14:01	cemerick	Sounds awesome.
14:02	cemerick	Is it reasonable to screen out clojure.* frames by default in general?
14:02	cemerick	(however much that might irritate rhickey? ;-)
14:02	chouser	dunno. I'll keep playing with it.
14:03	chouser	the seq thing is pretty key though -- allows modular filtering.
14:03	chouser	I want to play with reversing the causes and/or lines within each cause.
14:03	chouser	maybe the root cause and it's line should be immediately above the prompt
14:04	chouser	walking up the screen would be walking back in time
14:05	cemerick	are you working on a patch to be grabbed somewhere?
14:05	chouser	hrm. no. dumb.
14:06	cemerick	chouser: oh, #322 is done AFAIK. Not sure if you remember, but we were poking at some solutions a few months ago. http://www.assembla.com/spaces/clojure/tickets/322
14:07	chouser	cool. I've been watching the conversation, but haven't looked at the code.
14:07	cemerick	It was pretty simple, once I had a clue as to what was going on.
14:08	chouser	http://gist.github.com/453211 stack-trace-playground.clj
14:09	chouser	so the idea is people can write up different filters, colorizers, etc. then mix and match as desired.
14:10	technomancy	chouser: have you tried http://github.com/mmcgrana/clj-stacktrace ?
14:10	chouser	technomancy: once long ago
14:10	technomancy	we just started using it, and it makes me very glad.
14:10	chouser	hm... ok!
14:10	technomancy	I would love to see it in clojure 1.3
14:10	cemerick	"\033[31m" == red?
14:10	technomancy	it's pretty easy to hack into clojure.test
14:11	technomancy	(with some monkeypatching)
14:12	technomancy	http://p.hagelb.org/clj-stacktrace-monkeypatch.html
14:12	chouser	nice. mmcgrana writes good code.
14:14	chouser	oh, man. I clearly need to start with clj-stacktrace and work from there.
14:15	chouser	alpha-punc-subs, trim-redundant, color-codes, all good stuff I would have ended up doing roughly the same way, but he's already done it better.
14:16	chouser	but I do want to do filtering which I don't see a way to plug in very cleanly.
14:39	arkh	being unable, at this point in my clojure education, to comprehend the source for 'for', does anyone know if 'for' realizes all seq-exprs on each iteration, regardless of modifiers, or will it short circuit evaluation of a seq-expr if a :while is false on an earlier listed seq?
14:39	arkh	e.g., does for behave like and ?
14:42	chouser	,(for [x [1 2 3], y [4 5 6] :while (< y 6)] [x y])
14:43	clojurebot	([1 4] [1 5] [2 4] [2 5] [3 4] [3 5])
14:43	chouser	I guess I'm not quite sure what you're asking.
14:43	arkh	you know ... I might not either ...
14:44	chouser	I have a vague recollection of :while not short-circuiting in quite the way one might expect, but I'm having trouble remembering the details
14:44	chouser	,(for [x [1 2 3], y [4 5 6] :while (odd? x)] [x y])
14:45	clojurebot	([1 4] [1 5] [1 6] [3 4] [3 5] [3 6])
14:46	arkh	,(for [x [1 2 3], y [4 5 6] :when (odd? x)] [x y])
14:46	clojurebot	([1 4] [1 5] [1 6] [3 4] [3 5] [3 6])
14:47	mmarczyk	,(for [x [1 2 3] :while (odd? x) y [4 5 6]] [x y])
14:47	clojurebot	([1 4] [1 5] [1 6])
14:47	chouser	so, :while only will skip the rest of the seq it's applied to
14:47	mmarczyk	clj-stacktrace looks awesome.
14:48	arkh	chouser: thank you again
14:48	briancarper	mmarczyk: Cool, I didn't know `for` worked that way.
14:49	briancarper	I never tried sticking a :while in the middle.
14:49	mmarczyk	aye... still, nowhere near as funky as PLT Scheme/Racket's list comprehensions :-)
14:50	arkh	mmarczyk: I meant to thank you also! Thanks.
14:50	mmarczyk	arkh: np :-)
15:08	dnolen	heh "equiv the revenge of num"
15:16	jcromartie	Why would a Common Lisp user say that Clojure can be "incredibly slow?"
15:16	jcromartie	Is that just the "Java is slow" myth extended?
15:17	Raynes	Probably.
15:17	jcromartie	SBCL ranks pretty far behind Scala on the benchmarks game
15:17	jcromartie	but SBCL is JVM too
15:17	jcromartie	weird
15:17	qbg	No...
15:18	qbg	SBCL is native
15:18	Raynes	Common Lispers can be "incredibly elitist".
15:19	qbg	Well, CL does lay waste to many other languages...
15:19	jcromartie	oh wait, I see, it was comparing it to Java... I got it confused with ABCL
15:20	technomancy	CL'ers might be looking for excuses to stick with CL and not try new stuff
15:20	technomancy	since that strategy has worked for them since the mid-80s
15:21	jcromartie	he says it's because it's all because they "have to make refs"
15:21	jcromartie	to interop with Java
15:23	jcromartie	and that Kawa (a JVM Scheme) is "over 1000 times faster than optimized Clojure"
15:23	jcromartie	"All because it doesn't have to make refs"
15:24	technomancy	he's clearly not interested in having a discussion; he just wants to Be Right.
15:24	qbg	Wow...
15:24	dnolen	just another troll looking for a good time it sounds like
15:53	yacin	i've got a project in emacs
15:53	yacin	and i moved some functions to a separate file
15:53	yacin	now, i'm obviously going to run into namespace/functionname clashes
15:53	yacin	how can i remove the defined functions from the old file?
15:53	yacin	without having to restart my swank session?
15:54	bhenry	yacin: i am also curious about this. i don't think you can.
15:54	bhenry	i always just restart
15:55	yacin	restarting isn't a big deal, but it'd be nice to not lose my current state
15:55	mmarczyk	ns-unmap
15:56	bhenry	,(doc ns-unmap)
15:56	clojurebot	"([ns sym]); Removes the mappings for the symbol from the namespace."
15:56	bhenry	mmarczyk nice.
15:56	mmarczyk	I use this for convenience: (defmacro undef [& syms] `(doseq [s# '~syms] (ns-unmap ns s#)))
15:57	mmarczyk	(undef foo bar baz) / (clojure.contrib.with-ns/with-ns my.old.ns (undef foo bar baz))
15:58	mmarczyk	got it in my init scripts (user.clj on the classpath)
15:59	yacin	mmarczyk: thanks!
16:00	mmarczyk	(doseq [old-name (keys (ns-interns 'my.old.ns)) :when (contains? (ns-interns 'my.new.ns) old-name)] (with-ns my.old.ns (undef old-name)))
16:00	mmarczyk	for some generalised cleanup
16:00	mmarczyk	(wrote it here, so sorry for any typos etc. missed)
16:01	mmarczyk	ah, the name 'old-name' for the local doesn't make any sense, should be 'sym' or sth
16:08	yacin	mmarczyk: for setting up a user.clj, do i need any (ns ...) declaration? or just a bunch of functions?
16:08	mmarczyk	yacin: no (ns ...) form, no
16:08	yacin	kk
16:08	yacin	that's what i figured
16:08	yacin	thanks
16:09	mmarczyk	whatever you put in there will be loaded into the user namespace on startup
16:09	mmarczyk	I've got a bunch of requires... [clojure [set :as set] [walk :as walk]] etc.
16:09	mmarczyk	^requires in there...
16:10	polypus	~ping
16:10	clojurebot	PONG!
16:10	yacin	gotcha
16:10	mmarczyk	the script goes into any directory present directly on the classpath; clojure.jar has "Class-Path "."" in its MANIFEST, so if you put user.clj alongside it, it should work fine
16:12	mmarczyk	...at least I think that's why my setup for experimental repls involving ~/.clojure/{clojure.jar,clojure-contrib.jar,swank-clojure.jar,user.clj} works :-)
16:13	yacin	yeah, i just dumped it in ~/.clojure
16:13	polypus	if i define a seq using cycle or repeat and hold on to the head while bang on it for a while, is the seq caching aware of the cyclic nature of the seq, i.e. does it cache all the values, or just the period of the cycle? if that makes any sense
16:15	polypus	i.e. (def s (repeat 4)) (take 1000000 s) are 1000000 values cached or just 1 because it's a repeat
16:16	mmarczyk	it's aware of its cycling value
16:17	mmarczyk	if I understand you correctly
16:17	mmarczyk	(def t (cycle [(promise)])) (deliver (first t) :foo) @(second t)
16:17	mmarczyk	gets back :foo
16:18	mmarczyk	(meant to say "cyclic nature" above)
16:19	mmarczyk	ah, wait, I guess I did misunderstand...
16:19	polypus	mmarczyk: what i mean is will the seq cache consume 1000000 int's worth of memory or just 1
16:20	polypus	because i've held on to the head
16:20	mmarczyk	yeah
16:20	polypus	yeah 1000000
16:20	polypus	?
16:20	mmarczyk	(let [foo (cycle [:foo])] [(first foo) (first (drop 100000000 foo))]) ; => see if memory runs out :-)
16:24	mmarczyk	having looked at the code I'd still say it knows it's cyclic
16:26	mmarczyk	however that's not to say the generated lazy seq cannot get bigger... I guess it will, due to bookkeeping (multiple pointers to the same seq bundled with instructions to go on cycling when the end is reached)
16:26	mmarczyk	that's as far as I can tell, anyway.
16:27	polypus	mmarczyk: ok ty, the code you just posted did not run out of memory but i'll mess around a bit more. would be good to know for sure though, cuz you don't seem %100 convinced
16:28	mmarczyk	well the seq cycle returns will doesn't make copies of the cycled seq as it goes
16:29	mmarczyk	it sorts of "chains together" multiple pointers to the same seq
16:29	mmarczyk	and at the end there's a "terminator" which, if you hit it when traversing the cycled seq, will produce extra elements
16:29	mmarczyk	so whatever was in the cycled seq will not be replicated (as the promise/deliver example demonstrates)
16:30	mmarczyk	but there will be a growing lazy seq scaffolding
16:30	polypus	mmarczyk: thanks again
16:30	mmarczyk	np
16:30	mmarczyk	I am, say, 99% convinced ;-)
16:30	polypus	:)
16:31	Kjellski	is it deprecated to use (defstruct ...) ?
16:31	polypus	Kjellski: yeah, use defrecord or deftype
16:31	Kjellski	polypus: thanks =)
16:32	polypus	np
16:33	Kjellski	polypus: ... they´re not in the cheatsheet at clojure.org/cheatsheet =(
16:35	mmarczyk	Kjellski: actually they haven't been released... 1.2 stuff
16:35	polypus	Kjellski: http://clojure.org/datatypes
16:36	mmarczyk	Kjellski: so, if you're on 1.1, you won't be able to use them, but transitioning from structs to records should be entirely painless
16:37	Kjellski	thanks to both of you, but in that case I could use (defstruct ...) nevertheless?
16:38	polypus	Kjellski: yeah, like mmarczyk says, it'll be easy to upgrade later
16:38	polypus	and np :)
16:38	Kjellski	polypus: alrightyyyy thennnn ^^
16:38	mmarczyk	:-)
16:47	dnolen	rhickey: equiv isn't quite as zippy for me as the latest equal. but perhaps the more important issues are addressed.
16:53	arohner	stupid question: in paredit mode, how do you split strings? i.e. "foo bar" -> "foo" "bar"
16:53	polypus	arohner: option shift s
16:53	mmarczyk	M-S
16:54	arohner	polypus mmarczyk: thank you
17:00	polypus	best performing clojure datastructure for a stack?
17:01	polypus	i assume it's a list
17:05	mmarczyk	core functions seem to encourage vector usage -- peek/pop
17:07	polypus	mmarczyk: did you see that in commented code?
17:07	mmarczyk	I just mean that the mere existence of functions called peek/pop suggests that whatever they work on should be a good fit for a stack
17:08	polypus	but they're defined on lists too
17:08	mmarczyk	or a queue, I guess; and indeed, they also work with clojure.lang.PersistentQueue
17:08	mmarczyk	hey... never noticed that :-D
17:08	mmarczyk	thanks! :-)
17:08	polypus	np :)
17:09	mmarczyk	I thought a transient vector could do well in a single-threaded context, but there is no peek!, only pop!
17:09	mmarczyk	and peek doesn't work
17:09	mmarczyk	so probably a list after all, though definitely use conj over cons
17:10	riddochc	p WhyteShadow=Well, there's a few particular "parameters" you can tweak to adjust the perception of height in a figure...
17:10	mmarczyk	hm, really no peek! ...? checking again... nope
17:10	riddochc	Ack! Sorry.
17:10	riddochc	Wrong window.
17:13	polypus	mmarczyk: i am in a single threaded context and don't need peek so maybe it'll do. ty
17:14	mmarczyk	polypus: you don't need to know what's on top of the stack?
17:17	ataggart	since peek doesn't modify anything, why would there need to be a transient version of it?
17:17	mmarczyk	because the regular version works with IPersistentStack instances
17:17	mmarczyk	nothing transient is IPersistent*
17:18	polypus	mmarczyk: oops forgot pop returns the rest not the popped
17:18	mmarczyk	polypus: right
17:18	ataggart	ah, good point
17:18	mmarczyk	polypus: actually, why not use java.util.Stack?
17:19	ataggart	hopefully all things that will become nonissues when cic is done
17:19	mmarczyk	ataggart: agreed
17:19	polypus	mmarczyk: that's an idea
17:20	mmarczyk	ataggart: though in this case, I don't see a reason not to provide a transient counterpart to IPersistentStack or to use a persistence-agnostic interface
17:20	mmarczyk	that's far from saying there isn't one -- just wondering :-)
17:21	ataggart	mmarczyk: agreed. As Rich says: "patches welcome" :)
17:21	mmarczyk	polypus: it's likely to have been beaten to death and back with optimisation sticks
17:21	mmarczyk	ataggart: I might oblige when my CA gets processed :-)
17:22	mmarczyk	appears to be stuck in the queue
17:25	mmarczyk	is there any way to see pull requests one has sent off on GitHub? my memory is failing me and I'd like to check one...
17:36	SirNick	How can I use apply with an arbitrary symbol?
17:36	polypus	SirNick: elaborate
17:37	polypus	mmarczyk: trying out java's stack now. ty 4 tip
17:37	SirNick	polypus: Given I have a variable holding a symbol, how can I call apply with that symbol, for example (apply the-variable [])
17:37	mmarczyk	polypus: cool, np :-)
17:37	qbg	(apply (resolve the-variable) ...)
17:38	SirNick	qbg: Resolve is exaclty what I was looking for, thank you
17:38	replaca	does anyone know why the IntrenalReduce version of reduce is commented out currently?
17:38	replaca	in core.clj: #_(defn reduce... )
17:39	mmarczyk	replaca: which branch?
17:39	qbg	It isn't for me
17:40	replaca	mmarczyk: master
17:41	mmarczyk	hm, so it is
17:41	mmarczyk	no idea
17:41	mmarczyk	a08eac88766fa5eca96d: disable direct binding (and internal reduce with it, for now)
17:41	replaca	oh well. It breaks the doc string, but that's not the end of the world as long as it's temporary
17:41	mmarczyk	no rationale
17:42	mmarczyk	but if it's somehow connected to direct binding, pre-:static stuff maybe...?
17:42	replaca	mmarczyk: yeah, that would make sense
17:43	mmarczyk	oh? personally, I can't make heads or tails of it ;-)
17:43	mmarczyk	planning on a mystery-hunting trip to find out why there's any relation :-)
17:48	mmarczyk	while of course I jest, I notice that equal has reinstated internal-reduce (with no mention of :static in core.protocols... oh well)
17:49	Kjellski	gn8 all =)
17:49	Kjellski	thanks for the help
17:49	Kjellski	;)
18:01	SirNick	Say I have a function defined in namespace A and I call the function from namespace B, resolve then resolves from namespace B but how can I make it resolve from the place it was defined, namespace A?
18:03	AWizzArd	ns-resolve maybe
18:03	qbg	Or you could namespace-qualify the variable
18:03	SirNick	yea I was thinking ns-resolve, but is there anyway I can get around writing out the namespace name explicitly? Something like ns but that doesn't change perhaps?
18:04	rhickey	dnolen: go a git of something that's slower?
18:05	SirNick	I meant to say ns (with astericks) not bold ns
18:05	dnolen	rhickey: ahh, all this different operators because of your research, was using something that got slower (unchecked-multiply), replacing with unchecked-multiply-long fixed it. no equiv is light speed like equal.
18:05	SirNick	Well it did it again even with added spaces... that's annoying
18:06	ataggart	Am I the only one who never seems to pass around symbols? Is this a CL thing?
18:07	AWizzArd	SirNick: would it work that your function in A becomes a closure, capturing its own ns at compile time and using that then with ns-resolve?
18:07	mmarczyk	SirNick: functions can have metadata attached currently, so you can put the ns name there
18:07	mmarczyk	in fact
18:08	mmarczyk	$(meta #'reduce)
18:08	sexpbot	=> {:ns #<Namespace clojure.core>, :name reduce, :file "clojure/core.clj", :line 773}
18:08	SirNick	AWizzArd: Hmm that might work, but I am not sure how to accomplish such a thing in clojure
18:08	mmarczyk	here you go
18:08	mmarczyk	(meta #'your-var)
18:08	SirNick	mmarczyk: Oh interesting idea... I'll look into that as well, thanks
18:09	mmarczyk	ah, no, actually that presupposes an ability to resolve the Var
18:09	SirNick	mmaryczyk: would there be a way to get the metadata of the function you're currently in perhaps?
18:09	mmarczyk	I guess so
18:09	qbg	SirNick: Any reason why you can't namespace-qualify the symbol in the variable?
18:10	SirNick	qbg: well the function takes one argument and that argument is converted to a string for one purpose and is also used to call other functions in the same namespace
18:10	mmarczyk	$(^{:foo :foo} (fn foo [] (meta foo)))
18:10	sexpbot	=> nil
18:11	mmarczyk	:-(
18:11	mmarczyk	but it works for me at the repl :-/
18:12	mmarczyk	$(^{:foo true} (fn foo [] (meta foo)))
18:12	sexpbot	=> nil
18:12	mmarczyk	an equal-branch repl, but I don't see how that would be relevant
18:13	qbg	,(name 'clojure.core/+)
18:13	clojurebot	"+"
18:13	SirNick	Hmm interesting, I thought name would include the namespace
18:14	SirNick	well then I suppose that would work... makes things a bit more bulky to call the function though since I would have to write the function twice
18:14	SirNick	Sorry, I mean the namespace twice
18:14	qbg	`+
18:14	qbg	,`+
18:14	clojurebot	clojure.core/+
18:14	mmarczyk	,(.namespace 'clojure.core/+)
18:14	clojurebot	java.lang.IllegalArgumentException: No matching field found: namespace for class clojure.lang.Symbol
18:15	mmarczyk	,(.getNamespace 'clojure.core/+)
18:15	clojurebot	"clojure.core"
18:15	qbg	,(namespace `+)
18:15	clojurebot	"clojure.core"
18:16	mmarczyk	oh, missed the wrapper
18:16	qbg	find-doc is useful
18:16	SirNick	see what I'm trying to avoid is having to call it like this (feeds/fetch-cached feeds/feed-name)
18:16	mmarczyk	right :-)
18:17	mmarczyk	well, if you're just going to evaluate something in a different namespace
18:17	mmarczyk	have a look at clojure.contrib.with-ns/with-ns maybe
18:17	mmarczyk	(with-ns feeds (fetch-cached feed-name))
18:18	SirNick	hmm that might work
18:18	mmarczyk	not a good idea if this should cause side effects in the current namespace, of course
18:18	mmarczyk	but for a pure function it works fine
18:20	rhickey	hrm, clj: http://github.com/liebke/clj, but the shell script to run clojure is likely to be called clj...
18:21	mmarczyk	SirNick: going funky (probably nonsensical, too) now: (clojure.walk/prewalk-replace #(if-let [v (and (symbol? %) (get (ns-publics 'feeds) %))] v %) '(fetch-cached feed-name)) ...use in a macro
18:22	SirNick	heh, wow, haven't been using clojure long enough to follow exactly what's going on there
18:22	mmarczyk	rhickey: indeed it is... plus, is this significantly different from lein + plugins?
18:23	technomancy	mmarczyk: it's remarkably similar to what I had in mind for lein 1.3 but haven't gotten around to implementing yet. =)
18:24	mmarczyk	technomancy: I see... so now not only do we not use the well-established JVM-land build tools (because mvn is icky), but also have two build tools of our own? hm :-P
18:25	mmarczyk	um, not really build tools, more like "dev environment setup tools"
18:26	Raynes	technomancy: Speaking of leiningen, put out a new version already. You're slow, and I'm impatient. You've only got like 75 projects to work on, it isn't like you have time. ;)
18:26	mmarczyk	ah, scratch that... doesn't if lein doesn't need to duplicate that functionality
18:27	technomancy	Raynes: RC2 this weekend, dude
18:28	Raynes	I don't do release candidates, because I'll bork something. I just can't wait to do "lein upgrade". :D
18:33	technomancy	it's actually pretty easy to use a from-source lein now; the bootstrap process is way simplified.
18:33	technomancy	course you can always wait too
18:35	SirNick	mmarczyk: This seems to work, but is it a good idea to do this way? http://gist.github.com/453545
18:37	mmarczyk	SirNick: ah, that would be AWizzArd's suggestion
18:37	mmarczyk	sure, if it makes life easier for you somehow
18:37	SirNick	Ah, so that's what he was meaning... well good then, thanks to both of you
18:38	riddochc	Hm. It looks to me like liebke's clj project overlaps slightly with leiningen or maven? This is interesting.
18:39	riddochc	Or not, I suppose. Looks like it's not doing build-systemy things.
18:40	dnolen	riddochc: it's much more like gems it seems to me. simpler though less flexible then lein. but perhaps easier for people new to clojure
18:43	mmarczyk	its name clashes with my Python launcher for Clojure though -- grrr :-P
18:43	dnolen	also simplifies writing a decent bundle for an editor like TextMate it seems.
18:45	riddochc	Its name clashes with my use of hashdot, too, but I suppose if clj becomes an "official" launcher, I wouldn't complain much.
18:45	technomancy	riddochc: you're using hashdot? sweet.
18:46	riddochc	Alternatively, we could look for a word for some mineral or rock that has the letters 'cl' in it somewhere.
18:46	riddochc	Yeah, hashdot helps me keep my classpath sane.
18:48	technomancy	riddochc: does it have a sane build yet? I wanted to fix it, but the madness of makefiles plus .deb creation scared me off.
18:49	riddochc	One thing I haven't done yet is figure out good way to develop a hashdot-based script with slime/swank...
18:50	riddochc	Sane build? I don't know, it's not the simplest system possible, that's for sure, but it solves the problem for clojure, jruby, and sisc all at once.
18:50	mmarczyk	riddochc: clinohumite :-P
18:50	technomancy	when I was using it you had to manually edit the makefile to point to your java home... this was basically right after it came out though.
18:51	technomancy	I used to work with its author.
18:51	riddochc	For most things I build from scratch, like hashdot, I do something like ./configure --prefix=/usr/local/stow and then use Gnu stow to manage a ton of symlinks from /usr/local/bin/x etc. to /usr/local/stow/package/bin/x etc.
18:51	technomancy	cool
18:51	mmarczyk	whoa.
18:52	riddochc	I don't recall having to tell it to point to my java home, but it was a while ago I did it. Not sure.
18:53	riddochc	Yeah, someone pointed me to gnu stow a few years ago, and it's become a seriously indespensible tool for me.
18:53	riddochc	I use it and a shell script to switch between a zillion versions of ruby, for example.
18:54	mmarczyk	what OS?
18:55	riddochc	Linux, mainly. But I don't see why it wouldn't work on mac.
18:56	riddochc	So, for example, I have /usr/local/stow/clojure-07f05862c2362f6b51e7c92ccd9476c45c9dff6e/lib/clojure.jar which is pointed to by a symlink, /usr/local/lib/clojure.jar To switch to a different version of clojure, I use stow like a package manager.
18:56	mmarczyk	I'm a penguin too ;-)
18:57	riddochc	Cool, mmarczyk. A lot of my fellow linux friends RL switched to apple systems over the years, I feel like a holdout.
18:58	mmarczyk	yeah... a friend of mine who as good as introduced me to Linux has bitten the devilish apple too... now he calls my notifications service "growl" :-(
18:59	riddochc	Anyway, that's my system. Usually, my hashdot scripts reference jarfiles in /usr/local/lib/ so they follow the updates as I build new versions of things. When I need to refer to a specific version, I refer to files within /usr/local/stow/...
19:00	riddochc	Though now that I think of it, maybe I should have /usr/local/jar instead. I suppose jarfiles don't really belong in a lib directory.
19:04	lpetit	thinking about the fact that currently, in counterclockwise (and I guess in enclojure as well ?), we always AOT compile namespaces, "to be certain" that the relevant bits that must absolutely be AOT compiled "by essence" (generating java classes/interfaces/protocol interfaces/type classes for java interop) are AOT compiled
19:04	Raynes	That's horrifying.
19:06	lpetit	wouldn't it make sense to have a mode where only those parts I qualified as "relevant" are AOT compiled ? that is only compile in memory everything, but genclasses, geninterfaces, protocol interfaces, definterfaces, type classes (close world of type of things which need to be AOT compiled for java interop)
19:06	lpetit	?
19:06	mmarczyk	lpetit: could you possibly delegate some of the work to a build manager? say, use lein as a library?
19:07	mmarczyk	or maybe clojure-maven-plugin if that's required to play nice with the ide
19:07	lpetit	mmarczyk: I don't see how lein could help. I want something totally automatic.
19:07	riddochc	I still seem to be tripping over issues related to AOT compiled stuff and version stuff.
19:07	mmarczyk	well, you could just allow the user to specify what needs to be aot'd
19:08	lpetit	I don't want my users to have to specify the list of namespaces to be AOT compiled. I even don't want "namespaces" to be AOT compiled. Only the relevant parts. If a namespace defines 3 regular functions and 2 protocols, I just want the interfaces related to the protocols
19:08	mmarczyk	ah, that's actually what you suggest
19:08	mmarczyk	hmmm
19:08	mmarczyk	oh, so it isn't.
19:08	lpetit	mmarczyk: that's exactly what I want to avoid :-)
19:08	mmarczyk	lpetit: got that now :-)
19:09	mmarczyk	I wonder if (binding [compile-files true] (eval '(definterface ...))) would do that...?
19:10	lpetit	things that must be AOT compiled for java interop will always (in java) be things that can only be loaded once per JVM, I guess. I just want to minimize the disk I/O work while recompiling the whole project. And also minimize the number of files in the disk after a day full of work !
19:10	riddochc	Well, have fun clojuring. Gotta run.
19:11	lpetit	(because everytime a ns is recompiled, new function names appear in the classes/ folder ...)
19:11	lpetit	mmarrczyk: maybe. but, again, code cluttering ....
19:11	mmarczyk	lpetit: ah, so you want a new, really fine-grained compile function
19:12	lpetit	mmarczyk: yes, balancing the potential virtues of that
19:12	lpetit	s/balancing/contemplating/
19:12	sexpbot	mmarczyk: yes, contemplating the potential virtues of that
19:13	mmarczyk	that would be cool, especially since I guess interfaces don't really depend on Clojure, so they could be AOT'd and not break cross-clojure.jar compatibility
19:13	mmarczyk	right?
19:15	lpetit	right for interfaces.
19:16	lpetit	for types, I guess not right. for gen-classes, which generate stubs, I guess not right. :-(
19:20	lpetit	time to go to bed, let's continue this discussion later :)
19:23	mmarczyk	good night :-)
19:23	ataggart	Other than nil punning is there a rationale for using next vs rest? I tend to use next, but much of the code I see online is using rest.
19:25	mmarczyk	rest is lazier
19:26	mmarczyk	next evaluates the head of its return value, whereas rest doesn't
19:26	defn	http://www.youtube.com/watch?v=KrfpnbGXL70
19:27	defn	lol.
19:27	rfg	defn: Yeah, it's great :)
19:27	ataggart	mmarczyk: that was my thinking as well.
19:29	defn	They enable us to send XML messages through SOAP. Through SOAP!"
19:31	mmarczyk	defn: that... is... AMAZING!!! :-DDD
19:31	defn	mmarczyk: :)
19:31	defn	I wish they would have mentioned Clojure in there
19:31	defn	they only have Scala Johannson
19:32	technomancy	clojurebot: soap is <reply>"They enable us to send XML messages through SOAP. Through SOAP!!!" (http://www.youtube.com/watch?v=KrfpnbGXL70)
19:32	clojurebot	c'est bon!
19:32	defn	hahaha, well done technomancy
19:35	defn	lol I missed how his dad actually dies
19:36	defn	His dad rips his shirt and finds the Java tattoo with a heart around it, bahahaha
19:36	defn	ah man -- brilliant
19:37	ataggart	the shutdown chime when the guest gags at the kids announcement: priceless.
19:37	mmarczyk	then he has "Where do you want to go today?" on his gravestone
19:37	mmarczyk	just watched again, still can't believe this :-D
19:49	ataggart	technomancy: do you foresee any reason to avoid augmenting lein to optionally process/emit dependencies a la ivy? There are a few cases I'm running into where maven's all-or-nothing approach is causing problems.
19:53	technomancy	ataggart: I evaluated Maven, Ivy, and maven-ant-tasks before starting Leiningen, and the latter was the only one I could get working at all. I think it would be doable with Ivy, but it would be a lot of work, and I don't know what the benefits would be.
19:53	technomancy	that said, it would be easy to write a plugin that could emit an ivy.xml file which you could use for certain ivy-specific operations.
19:53	ataggart	ivy's dependency system is far more flexible than maven's
19:54	ataggart	basically ivy is a superset of maven w/r/t dependency management
19:54	technomancy	sure, but maven's is a lot more flexible than leiningen's, and I've never found myself needing more flexibility than that. =)
19:54	ataggart	heh I do
19:55	ataggart	maven's dependency stuff is pretty limiting for us
19:55	technomancy	I guess I would need to see some use cases.
19:57	ataggart	one hypothetical example would be c.c.logging emitting multiple configurations depending on what implementation you wanted to use
19:57	ataggart	so I could depend on a particular configuration of the logging lib, and only those transitive deps for that configuration would be pulled in
19:57	ataggart	not every single possibly needed transitive dependency
19:58	technomancy	hmm... I still haven't figured out why you'd want to use c.c.logging against anything other than log4j, but maybe I'm missing something.
19:58	technomancy	it's not like you can have a medium-to-large-sized clojure project without already pulling in log4j. =)
19:58	ataggart	it was an example of the filtering of transitive dependencies
20:00	ataggart	for a more real world example, we have a rest framework that supports representing resources as xml, json, etc. each of those are underpinned by a transitive lib. if one doesn't care about the json stuff, then one could depend on the rest framwork's "runtime" and "xml" configurations
20:00	ataggart	and only get those dependencies
20:01	ataggart	it's not a big deal when one has a shallow dependency tree, but for more complex systems without that kind of specification someone could end up pulling in many dozens of unneeded jars
20:02	technomancy	it seems like you could get close to the same effect by just splitting it into multiple artifacts
20:03	technomancy	I don't know... I'm unlikely to implement it myself. if you would like to experiment with it and try to determine if it'd be possible without doubling the complexity of the deps task, I might be convinced if it's clear people would use it.
20:03	ataggart	yes you could move the cost into the project organization, and break things up as needed, but sometimes that's not very useful when it's one new local class that needs some lib that in turn needs a dozen other libs
20:03	ataggart	it's already possible since I made it work
20:03	ataggart	I'm just trying to guage the pushback
20:05	technomancy	why don't you ask on the mailing list to see if that problem is affecting many other users?
20:06	ataggart	k
20:11	yacin	are there already jars set up for clojure 1.2 for lein?
20:12	technomancy	yacin: [org.clojure/clojure "1.2.0-master-SNAPSHOT"] should do the trick
20:23	technomancy	(def foo (constantly (bar))) ;; <= bar is only evaluated at compile-time, right?
20:38	hiredman	technomancy: eh, functions are called at runtime
20:39	technomancy	hiredman: I mean bar isn't called for every call of foo.
20:40	technomancy	,(let [a (atom 0) foo (constantly (do (swap! a inc) :hi))] (foo) (foo) (foo) @a)
20:40	clojurebot	1
20:42	tomoj	constantly is a function, so
20:42	technomancy	(pardon the gratuitous do)
20:42	technomancy	yes, that. =)
21:28	polypus	anybody know why peek is not supported on a transient vector?
21:58	ataggart	polypus: no one cared enough to submit a patch?
21:59	ataggart	but since you know it's a vector: (let [t (transient [1 2 3])] (get t (dec (count t))))
22:06	polypus	ataggart: ok, ty
22:09	ataggart	btw why do you need peek if you're the one modifying the transient?
22:11	polypus	ataggart: cuz i'm using it as a stack and it get recursed over
22:12	ataggart	k
22:13	polypus	s/get/sets
22:13	polypus	s/sets/gets :)
22:26	Raynes	Is there a way to run uncompiled clojure code inside of a jar? I think we need to get rid of jars and start using special cljars. :D
22:26	tanob	Raynes: I would name "clars" :)
22:27	Raynes	>_>
22:28	lancepantz	Raynes: you're actually not supposed to compile clojure inside of jars
22:28	lancepantz	wars you have to
22:29	Raynes	Huh?
22:29	Raynes	I'm not sure I caught that right.
22:29	lancepantz	i think i may be misunderstanding
22:29	lancepantz	but you shouldn't put class files generated from cljs inside of jars
22:30	tomoj	uncompiled clojure code in a jar works fine
22:30	lancepantz	the idea being that your jars are not dependent on a specific version of clojure
22:30	Raynes	Oh, I don't want to.
22:30	Raynes	tomoj: Not if you're trying to run the code.
22:30	polypus	Raynes: if you do 'lein jars' and unzip the jar you'll see just plain uncompiled clojure files
22:31	tomoj	oh, I see
22:31	tomoj	no main class?
22:31	hiredman	lancepantz: pretty sure you don't really need to in wars either
22:31	Raynes	I don't want to compile the code.
22:31	Raynes	I'm just saying. There is no way to package up a Clojure application without compiling the code, as far as I know.
22:31	hiredman	you just need a compiled servlet class and it can load and compile clojure at runtime
22:32	polypus	Raynes: just try lein jars. no compilation involved
22:32	Raynes	That doesn't help.
22:32	tomoj	certainly need compilation
22:32	tomoj	of _something_ anyway
22:32	lancepantz	hiredman: that may be true
22:32	hiredman	Raynes: https://www.assembla.com/spaces/clojure/tickets/315-add-support-for-running--main-namespace-from-clojure-main-without-aot
22:34	tomoj	that wouldn't help for a runnable jar, would it?
22:34	Raynes	Yeah, it would.
22:34	tomoj	run some wrapper instead?
22:35	Raynes	It's not hard to work around, but it sucks to have to work around it.
22:35	Raynes	It will be really cool to be able to make use of jars without AOT.
22:35	Raynes	For applications, I mean.
23:18	Raynes	Is there a way to stop Clojure from printing the already refers to warnings?
23:35	Raynes	,(clojure.contrib.string/ltrim " blah")
23:35	clojurebot	"blah"
23:58	defn	Anyone here meditate?