#clojure – 2009-06-18

2009-06-18

00:15	jtal	ok, one more related question- if I cant load str-utils, math, etc- is there an env var or something where I need to add a path?
05:02	frodef	is there some built-in mutex-like thingy, or is there a mutex in the java libs?
05:06	eevar2	frodef: http://java.sun.com/docs/books/tutorial/essential/concurrency/index.html
05:07	eevar2	frodef: and check java.util.concurrent in the api docs
05:07	frodef	yep, thanks.
05:10	eevar2	frodef: you should come a long way with the clojure synchronization mechanics, though
05:10	eevar2	didn't notice the first part of your question
05:12	frodef	I just need to do (let [x (new lock-or-whatever)] ... (await x)) and some other thread (that has lexical access to x) will notify x.
05:12	eevar2	read about refs and agents
05:13	frodef	I did, but I couldn't "make them fit" this pattern.
05:16	eevar2	me neither -- then again, my clojure skills blow :)
05:16	frodef	A simple java lock should do the trick, though. .. mine too :)
05:18	jdz	frodef: i used a blocking queue (don't remember the namespace by heart, but somewhere in java.concurrent)
05:19	jdz	so you have threads which put stuff into the queue, and others who just take stuff from it
05:19	jdz	and the nice thing is that the producer thread can put more than a single thing on it (the queue capacity)
06:18	AWizzArd	clojurebot: max people
06:18	clojurebot	max people is 164
07:11	frodef	when using proxy to implement an interface, is there some way to automatically "fill out" the methods you don't care about?
07:13	frodef	Hm.. Perhaps this is what the "fooAdapter" class is for, for each "fooListener"..
07:14	eevar2__	frodef: yup
07:21	frodef	eevar2__: thanks.. it's a strange world, this :)
07:23	eevar2__	wonderful world of Java :)
09:46	frodef	isn't it somewhat strange to not have a version of distinct that takes a test function?
09:52	rhickey	frodef: you want to pass a comparator?
09:52	rhickey	or a keying fn?
09:55	frodef	rhickey: either, or both..? :)
09:56	rhickey	cemerick: so, I thought of a canonic case where I would want pulls vs patches - say pprint goes into core (I hope soon), it's likely to get ongoing enhancement and I'm sure Tom would prefer to keep working in his repo and just tell me when to pull than to create involved patches. true really for any non-trivial subsystem moved from contrib to core (not core ns)
09:57	rhickey	frodef: distinct-by?
09:58	frodef	rhickey: right.. I'm (again) used to CL where distinct would have a number of keyword arguments like test, key, etc.
09:58	frodef	rhickey: but you prefer a separate function?
09:59	rhickey	frodef: I was just looking for conventions already present in Clojure, found sort-by
10:01	rhickey	well, you're still shorter than remove-duplicates :)
10:02	frodef	in terms of characters, yes, but not in terms of operators :)
10:03	rhickey	since Clojure doesn't have lots of equality preds, test isn't that important, we have rseq, subvec, the only key would be :key, so I'm not sure the CL idioms work
10:04	rhickey	i.e. we havent' been doing :key
10:18	cemerick	rhickey: yeah, that's true.I'll bet that patchfiles will disappear eventually, though. I looked at the network diagram associated with your repo, and there's already at least 8 patches that are presumably worth pulling in, and that's just in the first week being on github. I don't think you're going to want to deal with the busywork that patches entail as that snowball builds.
10:19	cemerick	rhickey: BTW, regarding this:
10:19	cemerick	~ticket 79
10:20	clojurebot	Titim gan éirí ort.
10:20	cemerick	hrm, I thought that was the incantation.
10:20	rhickey	cemerick: seems to me to be just git am vs git pull, and in the former case I can peruse the diff first to see if I want to bother
10:20	rhickey	~ticket #79
10:20	clojurebot	{:url http://tinyurl.com/mwfnkq, :summary "GC Issue 76: adding sorted-set-by", :status :new, :priority :low, :created-on "2009-06-17T20:21:15+00:00"}
10:20	cemerick	ah-ha
10:22	cemerick	rhickey: so, Tim filed a patch, but it doesn't apply against current revs. I fixed it and based it against your master, but since I used his patch (and I don't know the status of his CA), what is the legal status of that fix?
10:22	Chouser	His CA is in.
10:22	cemerick	ah, OK
10:22	cemerick	nevermind then :-)
10:23	Chouser	cemerick: will you attach it to the ticket?
10:23	cemerick	I've never had to deal with the notion of provenance before, and I only thought about it after I applied the fix.
10:23	cemerick	Chouser: sure.
10:23	cemerick	the procedure is to set the status to 'test' if a patch is ready?
10:25	rhickey	cemerick: I'm thinking about that, one caveat right now is that the existing patches weren't produced with git format-patch, which is the way we want to patch moving forward
10:25	Chouser	The doc says "Now you can attach that patch file to the Assembla ticket and please add the 'patch' tag. Please mark the ticket 'ready to test' by checking that option under Choose an action..."
10:26	Chouser	Can't we just grandfather in the raw patches that are already there? I can't think of why that would be a problem.
10:26	cemerick	Chouser: damn, sorry, what doc? Information overload over here lately. :-(
10:26	Chouser	cemerick: yeah, np: http://clojure.org/patches
10:27	cemerick	Chouser: thanks much
10:28	cemerick	rhickey: I'd agree with Chouser, except that I'll bet that a lot of patchfiles that have lingered aren't going to apply at this point.
10:29	Chouser	I think we can add a custom field to assembla tickets. Might be nice to have a "latest proposed patch" field.
10:30	Chouser	that might facilitate, for example, automated scrubbing of tickets to flag those whose patches don't apply cleanly.
10:35	rhickey	I think we need to look through what we've got, it might be a good exercise for people to produce up-to-date patches for non-trivial things. toggling off ready-to-test for non-clean patches might be enough?
10:38	cemerick	rhickey: missed your 'git am vs. git pull' msg from before -- I think it's mostly a tooling/UI issue that github will likely resolve (e.g. tie pull requests to particular sha's, show the diffs easily, maintain status on whether or not the changeset merges cleanly, etc).
10:41	rhickey	cemerick: I think you were gone yesterday when I summed up my feelings about pull requests from unsupervised repos - they feel to me as if someone submitted a patch by giving me a URL to the patch. Those URLs in the ticket system make understanding it dependent upon their continuing validity
10:48	cemerick	rhickey: I'm marking an issue as ready to test with a patch attached -- you don't want such things assigned to you?
10:59	frodef	is there a reason why you can't simply add methods with proxy? I.e. (. (proxy [Object] [] (foo [] 'hello)) foo) could have worked?
11:02	cemerick	frodef: You might be able to invoke foo in that case using the reflector, but . invokes a static java method.
11:02	clojurebot	☕
11:04	Xcalibor	greetings
11:04	cemerick	frodef: nope, nevermind, proxy only reports the methods defined by the interfaces/abstract class it inherits from.
11:05	Chouser	frodef: there are a few reasons that can't work. For one, you'd need a full method signature declaration (return type, arg types, private/public, etc.)
11:06	frodef	a more trivial question.. (gen-class :name foo ..) how do I create an instance of this foo?
11:06	frodef	Chouser, cemerick: ok, thanks.
11:06	Chouser	frodef: if there's no existing class or interface that you can extend via proxy, you'll have to use gen-interface, gen-class, or write some java code.
11:07	rhickey	cemerick: no one should be assigning things other than to themselves, so leave blank
11:07	cemerick	frodef: gen-class produces a java class in bytecode, so it'd just be (com.Foo.)
11:07	Chouser	but only with AOT compilation.
11:08	frodef	Chouser: right, I'm doing gen-class, but I can't seem to be able to instantiate it.. :)
11:08	cemerick	right, I was assuming that :-)
11:08	frodef	so I have to import the class or something like that?
11:09	Chouser	frodef: you'd have to (gen-interface :name com.yoursite.Foo), AOT compile that, then (proxy [com.yoursite.Foo] [] ...)
11:09	Chouser	no need to use import.
11:10	frodef	Chouser: ok, thanks.
11:11	Chouser	hmph. patches attached to assembla tickets are application/octet-stream
11:11	cemerick	I'm pretty convinced the 'daily summary' selection in assembla's notifications is just broken.
11:11	Chouser	I'm having trouble convincing firefox to send them to my editor the way I want it to.
11:27	rhickey	cemerick: your sorted-by set likely has the same problem as sorted-by maps in that empty doesn't preserve the comparator?
11:28	cemerick	rhickey: right -- it uses PTM/EMPTY, so it inherits the problem. I figured that was a separate issue to be dealt with separately.
11:28	rhickey	patch welcome, or at least an issue
11:29	rhickey	if there isn't one
11:29	rzoom	i would like to replace particular elements in a list of lists with the value 1. Here is what I have: (map #(replace {%1 1} %2) '(2 3) '('(0 0 0 0 0) '(0 0 0 0 0)))
11:29	cemerick	rhickey: OK. I've never actually hit that problem, so I figured I'd leave well enough alone
11:30	rzoom	but that doesn't seem to work. would like element 2 in the first list and element 3 in second list to be set to 1.
11:30	rzoom	yet alas: ((quote (0 0 0 0 0)) (quote (0 0 0 0 0)))
11:30	rzoom	perhaps I am missing something
11:31	Chouser	rzoom: vectors are better at this sort of thing than lists. You're sure you need a list?
11:32	jdz	the double quoting is not related to that, though
11:32	rzoom	vectors are fine, still not working as i would expect.
11:32	rzoom	(map #(replace {%1 1} %2) [2 3] [[0 0 0 0 0] [0 0 0 0 0]])
11:32	rzoom	returns ([0 0 0 0 0] [0 0 0 0 0])
11:34	Chouser	,(map #(assoc %2 %1 1) [2 3] [[0 0 0 0 0] [0 0 0 0 0]])
11:34	clojurebot	([0 0 1 0 0] [0 0 0 1 0])
11:34	rzoom	!!!
11:34	rzoom	thanks!
11:34	rzoom	is there some reason that the replace shouldnt work?
11:35	Chouser	replace isn't about indexes, but about replacing values that are equal to a search value.
11:36	Chouser	,(replace {0 1} [0 0 0 0 0])
11:36	clojurebot	[1 1 1 1 1]
11:36	rzoom	well played.
11:36	rzoom	thanks for the insight, must have missed that one.
11:37	Chouser	huh, replace doesn't return a seq. I'm surprised.
11:37	rhickey	Chouser: as documented, no?
11:37	Chouser	indeed
11:38	rhickey	the idea was to make a sparse replace on a vector efficient
11:38	Chouser	ah, ok. Just not something I knew about.
11:42	cemerick	one thing that surprised me a bit about using sorted-set-by (and therefore PTM with a non-default comparator) is that object that sort equally are treated as duplicates. In my usage, that means that the comparator I provide ends up falling back to = to ensure that 0 is only ever returned for objects that are actually equal (as opposed to sorting identically given the primary comparison I'm using).
11:45	rhickey	cemerick: ?
11:45	cemerick	rhickey: just scribbling into the ether, apropos of nothing, really.
11:47	cemerick	there's just consequences to using an RB tree for a sorted set, as the set semantics can change compared to regular sorted sets.
11:47	Chouser	cemerick: I think I've heard others express surprise about that before, but it's certainly a more general behavior -- allows you to control what is equal for that set, not just the order.
11:48	cemerick	ok
11:48	cemerick	e.g. (sorted-set [1 2] [-5 10] [1 5]) contains three values
11:48	cemerick	however, one would initially use sorted set like so:
11:48	cemerick	sorted-set-by, I mean
11:49	cemerick	(sorted-set-by (comparator #(< (first %) (first %2)) [1 2] [-5 10 [1 5]) ==> a set with only two values
11:50	cemerick	when actually, the naive (common sense?) expectation is to end up with a set with the same values as the default-comparator sorted-set, just ordered differently
11:51	cemerick	bah, dropped a bracket and a paren on that last one: (sorted-set-by (comparator #(< (first %) (first %2))) [1 2] [-5 10] [1 2])
11:55	Chouser	,(sorted-map-by (comparator #(< (first %) (first %2))) [1 2] :a, [-5 10] :b, [1 5] :c)
11:55	clojurebot	{[-5 10] :b, [1 2] :c}
11:55	Chouser	for those of us without your patch. :-)
11:56	cemerick	heh
11:56	cemerick	Of course, none of this is really a problem, but I think sorted-set-by is a bad name, as the membership semantics are different than sorted-set
12:04	Drakeson	are there plans to include swank-clojure right into upstream slime?
12:07	Drakeson	is it actually possible?
13:08	s450r1	/join #clojure
13:09	Chouser	you're already here!
13:11	s450r1	excellent, this is where I want to be :-) Too bad I don't know how to use IRC
13:11	Chouser	:-)
14:12	hiredman	~contrib ticket #1
14:12	clojurebot	{:url http://tinyurl.com/ntftnj, :summary "Add chunk support to map filter et al", :status :accepted, :priority :normal, :created-on "2009-06-13T14:38:41+00:00"}
14:13	hiredman	hmmm
14:15	hiredman	~contrib ticket #1
14:15	clojurebot	{:url http://tinyurl.com/lytzu7, :summary "Move Clojure Contrib issues from GC to Assembla", :status :new, :priority :highest, :created-on "2009-06-10T18:34:55+00:00"}
14:33	mrsolo	is assembla better than github?
14:33	hiredman	assembla is different
14:34	mrsolo	hmm aseemba is subversion?
14:34	mrsolo	isn't clojure moving to git?
14:35	dnolen_	mrsolo: assembla is for issue tracking. clojure is on github already.
14:35	hiredman	mrsolo: read http://groups.google.com/group/clojure/browse_thread/thread/d2bf650ab97b204f#
14:36	mrsolo	clojure_dev group is limited i assume
14:36	mrsolo	membership that is
14:39	Chouser	mrsolo: anyone who has submitted a CA can be a member: http://groups.google.com/group/clojure-dev/browse_thread/thread/99e2728f665c52fa
14:41	mrsolo	good grief google ad is blocking on view on that link
14:41	mrsolo	ah refresh fixes it
14:42	cemerick	heh, no one's volunteered for the clojure-contrib issue migration yet, eh?
14:42	hiredman	it is off the highest priority...
14:42	hiredman	of
14:43	Chouser	cemerick: I spent about 15 minutes looking at what it would take to automate.
14:44	hiredman	you can get an rss feed of gc issues I think, right?
14:45	cemerick	it might be sorta easy, if it weren't for attachments, which are always finicky.
14:45	hiredman	hmmm, rss feed is not helpful
14:46	Chouser	cemerick: well, I was aiming for a 70% solution.
14:46	cemerick	I managed to do about 1.5 issues/minute. Wouldn't take long.
14:46	Chouser	but such annoying work. Thanks for doing it, but my goodness.
14:47	cemerick	manual is the way to go, if you're objective about cost/benefit
14:47	technomancy	cemerick: what we need is a newbie to come in here wondering, "What's a good project I could use to learn Clojure"
14:48	cemerick	I'll probably break down and do it this weekend if no one else steps up. :-/
14:48	technomancy	and then we conscript him
14:48	Chouser	cemerick: unless you place a high cost multiplier on the "annoying work" factor.
14:48	cemerick	technomancy: heh, that's my position AFAIC. A little copy-paste action is the least I could do to help out.
14:49	cemerick	Chouser: random scripting jobs are way more painful for me than a little bit of data entry.
14:49	Chouser	Next time somebody complains that contrib doesn't have a release out, they're the one doing it.
14:49	technomancy	nice
14:52	bhurt_	\?
14:52	bhurt_	Ignore that
14:52	bhurt_	Anyone else here?
14:52	technomancy	definitely not
14:52	Chouser	not me
14:53	bhurt_	So, quick stupid question: is there a way to call an overloaded Java constructor with a variable number of arguments?
14:54	Chouser	sure. did you try it?
14:56	bhurt_	Yes. I mean, I'm passed a seq of the arguments I want to pass on to the constructor. apply doesn't seen to work.
14:57	bhurt_	I'm willing to assume there are a valid number and they're of valid types.
14:57	bhurt_	Nor does playing games with partial seem to work.
15:01	achim-_	bhurt_: constructors and method calls aren't functions, although they look like some in most cases. but you can't use them as "higher order" functions, i.e. as arguments to other functions
15:01	bhurt_	Sigh. I was afraid of that.
15:02	achim-_	you'll have to take apart you argument seq or pull off some macro tricks ...
15:02	bhurt_	Or just wrap the constructor in a variable arg function, and then apply it.
15:03	achim-_	right, that's easier
15:04	achim-_	... but how would it work? you get the optional args as a seq ...
15:05	hoeck	bhurt_: or use clojure.lang.Reflector
15:05	hoeck	lisppaste8: URL
15:05	lisppaste8	To use the lisppaste bot, visit http://paste.lisp.org/new/clojure and enter your paste.
15:06	Chouser	If you don't know the number of args at the point where you call the constructor, you'll have to use reflection.
15:06	lisppaste8	hoeck pasted "jctor" at http://paste.lisp.org/display/82075
15:07	bhurt_	achim-_: yes, but then I can call apply on the normal function and pass in the seq.
15:08	bhurt_	I think I like the idea of using a multiarg function and apply better than using reflection. Maybe that's just me.
15:09	Chouser	like (defn foo ([x] (Ctor. x)) ([x y] (Ctor. x y)) ([x y z] (Ctor. x y z))) ?
15:09	bhurt_	Chouser: exactly.
15:09	hoeck	bhurt_: the only downside I know of is, that it is slow
15:09	bhurt_	Oh well.
15:09	Chouser	"it" being reflection
15:10	Chouser	the multibody foo there should be very fast as long as there's only one Ctor per arg count.
15:10	bhurt_	Chouser: also, reflection is generally complicated, and easy to get wrong.
15:10	hoeck	mhh, one could write a macro which uses reflection at compile time to generate those functions
15:10	achim-_	(defmacro applym [this to-that] `(~this ~@to-that))
15:10	bhurt_	There is.
15:10	achim-_	(applym .toUpperCase ["quick brown fox"])
15:10	achim-_	-> "QUICK BROWN FOX"
15:11	Chouser	bhurt_: clojure.lang.Reflector handles a lot of the complexity, but it's reasonable to avoid it when you can.
15:12	bhurt_	In this case, it allows me to drop a let expression and just fold it into the pattern matching on the arguments of the function.
15:12	bhurt_	Which actually simplified the code.
15:13	cemerick	~ticket #129
15:13	clojurebot	{:url http://tinyurl.com/nxbzf7, :summary "Add documentation to sorted-set-by detailing how the provided comparator may change set membership semantics", :status :new, :priority :low, :created-on "2009-06-18T19:14:02+00:00"}
15:14	cemerick	rhickey, Chouser: ^^^, FYI
15:14	Chouser	achim-_: yeah, but you usually use apply when you don't know the size of your arg seq until runtime.
15:14	cemerick	that's my best writeup of it
15:14	Chouser	cemerick: thanks
15:16	bhurt_	Chouser: in this case I don't, I'm being passed in the seq. Bad programming style, I know.
15:21	achim-_	Chouser: argh, of course. i think i need to get some sleep
15:33	hoeck	may someone briefly explain how a given comparator may change the set semantics of sortet-set-by?
15:33	mrsolo	Re: Ticket #128: Ensure that PersistentTreeMap.empty() is getting fixed...woot!
15:34	cemerick	mrsolo: heh, yeah, it's in the queue :-)
15:36	cemerick	hrm, it occurs to me that I should probably be adding unit tests to contrib as I start pushing patches related to core stuff.
15:36	technomancy	cemerick: yeah, jumping between repos is no fun
15:36	cemerick	well, they're always going to be in different repos, no?
15:37	cemerick	(though, actually, merging clojure and contrib doesn't sound so bad....)
15:37	technomancy	oh heavens no
15:37	technomancy	I mean just the test suite
15:38	mrsolo	merging clojure and contrib? no......
15:38	mrsolo	i assume contrib will get very big
15:39	george_	is clojure fast than java or common lisp ?
15:39	technomancy	I just mean it's unfortunate that when you're working on Clojure itself you have to switch to another repo to work on/run your tests
15:39	cemerick	well, I wasn't saying that only one jar file should be produced.
15:39	technomancy	george_: It Depends™
15:39	mrsolo	george: clojure can't be faster than java i assume
15:40	technomancy	it can certainly be faster than poorly-written Java
15:40	cemerick	mrsolo: the magic wand of pmap makes it functionally faster than java in many circumstances
15:40	mrsolo	well but that is 'person behind the keyboard' argument
15:41	george_	what is great about clojure so ?
15:41	technomancy	mrsolo: right, my point is that you can't measure two languages without taking into account other factors
15:41	cemerick	doing anything similar is Java is not really workable, IMO, so it's more than the programmer knowing what he's doing
15:41	technomancy	s/measure/compare/
15:41	hiredman	~git 6e9d29d51acef05e033cd0bed1ac23dc962943c6
15:41	clojurebot	Add .gitignore file. Covers classes, jars, and pom.xml.
15:41	technomancy	that's my patch!
15:42	dnolen	george_: haha, that's a big question. have you gone over the rationale at clojure.org?
15:42	george_	dnolen: nope
15:42	george_	dnolen: i will take a look
15:43	mrsolo	technomany: but you can give some generalized indication in term of language performance
15:43	hiredman	clojurebot: rationale?
15:43	clojurebot	rationale is http://clojure.org/rationale
15:43	mrsolo	assuming that human factor is more or less..cough equal..
15:44	technomancy	mrsolo: I still maintain that it depends entirely on the problem domain.
15:44	hiredman	,(/ 300 5)
15:44	clojurebot	60
15:44	hiredman	,(/ 300 5 4)
15:44	clojurebot	15
15:44	hiredman	blarg
15:48	george_	isn't there any other lisp designed for concurrency ?
15:50	hiredman	most don't even have standard threading apis
15:51	dnolen	george_: there aren't many languages designed for concurrency much less lisps ;)
15:53	dnolen	george_: clojure really shines when it comes to single machine concurrency. it doesn't really have distributed concurrency story (some work has been done in this area)
15:54	george_	that's why lisp is not popular as it deserves ... it is too busy looking backward, instead of looking to the future
15:54	george_	how about clojure sinthax ? is it similar to common-lisp ?
15:55	dnolen	george_: well that's only partly true. many parts of common lisp seems to be pretty futuristic to me. it simply suffers from being largely invented some 20-30 years ago, and it shows.
15:55	dnolen	george_: syntax in clojure has learned a lot from python/ruby, syntax support for data structures.
15:55	dnolen	,[1 2 3]
15:55	clojurebot	[1 2 3]
15:56	dnolen	for example is a vector
16:07	Lau_of_DK	Speaking of Compojure - I've got a question, when I launch my route-handlers, how do I go about providing them with a sessionID in an elegant way - I want something like this to happend (with-my-binding [foo 5] (defn bar [] (println foo)), which prints 5, regardless of it being defined out of the local scope
16:19	Lau_of_DK	any takers?
16:26	hiredman	http://mailinator.blogspot.com/2008/02/kill-myth-please-nio-is-not-faster-than.html interesting
16:30	dnolen	Lau_of_DK: i wonder if this can be handled with the middleware feature? It's called decorate
16:31	Lau_of_DK	Im not sure - Im more thinking a macro perhaps, like (with-results) from ClojureQL - its not quite what I want, but close
16:31	Lau_of_DK	How would you decorate?
16:34	dnolen	Lau_of_DK, never done it myself, but compojure has something called decorate-bind, it takes all your routes and wraps them.
16:34	dnolen	it looks the decorate macro is the thing to use.
16:34	Lau_of_DK	I'll have a look
16:35	cemerick	hiredman: yeah, that particular myth is going to be around forever
16:35	cemerick	the lesson is profile, profile, profile
16:40	bhurt_	"It's more complicated, so it has to be faster!"
16:43	mrsolo	NIO is not faster? what? hmm
16:44	cemerick	it can be (and is on certain platforms)
16:44	cemerick	but in the most common deployment scenarios and platforms, it's on par, or slower
16:44	hiredman	most common being linux?
16:45	cemerick	(of course, everything is app-dependent)
16:46	cemerick	hiredman: yeah...most unixes end up being slower. I seem to remember someone telling me that nio screams on solaris, but I've never profiled it myself. I think it was a toss-up on windows when we looked at it.
16:46	cemerick	that was two years ago, though, so who knows at this point.
16:47	cemerick	jeez, that was two years ago. :-/
16:48	mrsolo	hmm why are we switching to mina then... boggle
16:49	hiredman	because twitter used mina that one time
16:49	cemerick	well, mina is event-driven, right? Having a pleasant programming model could be worth the penalty.
16:49	cemerick	that, and hell, maybe NIO will be faster for you
16:50	hiredman	oh hey, I'm at 100 urls
16:50	mrsolo	we switched mainly because threads spawning was out of control
16:50	mrsolo	and of course..expected performance boost
16:52	hiredman	http://www.usenix.org/events/hotos03/tech/full_papers/vonbehren/vonbehren_html/index.html <-- Why Events Are A Bad Idea (for high-concurrency servers)
16:53	mrsolo	duh
16:53	cemerick	ouch.
16:54	cemerick	still, that was 6 years ago. Things change (they used JDK 1.4).
16:56	hiredman	yeah
16:56	hiredman	but things always change
16:56	hiredman	lisppaste8: url?
16:56	lisppaste8	To use the lisppaste bot, visit http://paste.lisp.org/new/clojure and enter your paste.
16:56	lisppaste8	hiredman pasted "ignore this" at http://paste.lisp.org/display/82090
16:57	hiredman	well that did not work
17:00	wavister	is there a way to exit a repl without going (. java.lang.System exit 0)
17:01	hiredman	(System/exit 0)
17:01	wavister	thanks
17:01	hiredman	control c or control d
17:02	hiredman	wavister: classes in java.lang.* are essentially "auto" imported, just like they are in java
17:02	hiredman	and the modern way to call static methods is (Class/method args)
17:07	wavister	ok spiff
17:08	wavister	so if i did (def x (new StringBuilder))
17:08	wavister	i wouldn't do (x/append "stuff")
17:08	wavister	still learning so pardon mistakes if they are there
17:10	hiredman	oh no
17:10	hiredman	append is not a static method
17:10	wavister	so the / is like a namespace thing? and static methods are essentially defined in the namespace of the class?
17:10	wavister	right
17:10	hiredman	no
17:10	hiredman	the Class/method is just syntactic sugar
17:11	hiredman	it makes it look that way, but (Class/method args) expands to (. Class method args)
17:12	hiredman	for calling methods you use (.method Object args)
17:12	hiredman	,(macroexpand '(.method Object args))
17:12	clojurebot	(. (clojure.core/identity Object) method args)
17:12	hiredman	huh
17:12	hiredman	that identity is new, I think
17:21	mrsolo	why is it there?
17:25	Lau_of_DK	From suns website: javax.mail.internet.MimeMessage ... Why does this fail: (import '(javax.mail.internet MimeMessage))
17:28	hiredman	does it throw an exception?
17:28	Lau_of_DK	Sure does
17:28	Lau_of_DK	Class not found
17:28	hiredman	is it in your classpath?
17:29	hiredman	javax.mail is not shipped with the jre I believe
17:29	Lau_of_DK	oh.. then that would be the cause
17:29	slashus2	It would be nice if they would ship it with the jre.
17:29	Lau_of_DK	incredible that java doesnt come with some kind of sendmail functionality
17:30	hiredman	don't worry, we'll all switch to Wave and email will be obsolete
17:30	Lau_of_DK	I gotta jet, thanks for chiming in hiredman
17:30	hiredman	nice to see you around
17:46	bhurt_	OK, Stupid Question of the day #2: Java Arrays are seq'able, right? I.e. if I have a Java function that returns a Foo[] I can just treat that as a seq of Foo's, right?
17:47	Chouser	you can call seq on it, yes.
17:47	Chouser	,(into-array [1 2 3 4])
17:47	clojurebot	#<Integer[] [Ljava.lang.Integer;@a96fd2>
17:48	Chouser	,(filter even? (into-array [1 2 3 4]))
17:48	clojurebot	(2 4)
17:48	bhurt_	Thanks.
18:39	hiredman	I hate slideshare
18:40	hiredman	is it too much to ask people just let me download a pdf?
18:40	hiredman	http://www.slideshare.net/hlship/clojure-functional-concurrency-for-the-jvm-presented-at-open-source-bridge?type=presentation
18:41	rhickey	didn't assembla tickets have tags yesterday?
18:43	durka42	~ticket #123
18:43	clojurebot	{:url http://tinyurl.com/lese6n, :summary "GC Issue 119: require doc out of date", :status :test, :priority :low, :created-on "2009-06-17T21:23:16+00:00"}
18:43	sudish	Would it be possible to have clojure-dev gatewayed to Gmane? http://gmane.org
18:44	sudish	I added the main clojure group long ago, but clojure-dev subscriptions have to be authorized, so I didn't take the liberty. The sub URL is http://gmane.org/subscribe.php
18:45	sudish	The primary advantage over other mailing list archives being NNTP gatewaying. Newsreaders make for a nice mailing list read interface.
18:47	eevar	possible to kill off the sourceforge/google code svn repos now?
18:48	eevar	well, delete all contents and leave a file redirecting people to github
18:54	durka42	that seems like not a bad idea
19:11	wavister	how would one evaluate a ratio into a float
19:13	hiredman	,(float 5/3)
19:13	clojurebot	1.6666666
19:14	wavister	ah thanks. how would i have figured that out without asking? i don't much want to read through the whole api every time i don't know how to do something...
19:15	wavister	there's no way to narrow down what i'm trying to look up, it seems
19:16	hiredman	wavister: search the api for "float"
19:16	hiredman	(find-doc "float")
19:17	wavister	my point is that to me the name "float" is not obvious. i looked for "toFloat" and "Ratio" thinking I could get methods meant for ratios
19:17	wavister	now that i know it it's pretty obvious
19:17	hiredman	wavister: how could "float" not be obvious?
19:19	wavister	it doesn't seem like an operation you would do on a ratio to me. maybe i'm obtuse, but that's what I think. the real problem is that i'm intimidated by the api being just a huge list of functions with no segmentation.
19:19	hiredman	wavister: there is segmantation
19:20	hiredman	look, if you had just typed "float" into your browsers search thinger, the second or third hit is "float"
19:20	wavister	ok, no argument
19:20	hiredman	good
19:20	wavister	i appreciate the pointer, also
19:21	mrsolo	http://mailinator.blogspot.com/2008/11/so-our-core-duo-server-now-gets-3.html so he really doesn't like nio
20:01	technomancy	so it looks like nailgun doesn't set up stdin and stdout for you?
20:05	technomancy	hmm; docs seem to say it's supported, but it doesn't get set up for you automatically
20:08	bpattison	I'm trying to call this java function in clojure
20:08	bpattison	StringTemplate hello = new StringTemplate("Hello, $name$", DefaultTemplateLexer.class);
20:08	bpattison	-- where --
20:08	bpattison	(import '(org.antlr.stringtemplate StringTemplate))
20:08	bpattison	(import '(org.antlr.stringtemplate.language DefaultTemplateLexer))
20:08	bpattison	(def x (new StringTemplate "Hello $name$") (. DefaultTemplateLexer class)))
20:08	bpattison	-- but I get --
20:08	bpattison	java.lang.NoSuchFieldException: class (NO_SOURCE_FILE:37)
20:08	bpattison	I'm not only a clojure noob, am a java noob as well -- so any help would be appreciated
20:16	technomancy	looks like it does the right thing with System.in, but in doesn't pick that up. easy workaround
20:19	bpattison	I probably need to understand what DefaultTemplateLexer looks like in Java before I can use it in Clojure
20:21	bpattison	anyway to get the methods/attributes of java class in clojure?
20:22	technomancy	bpattison: use the show function from clojure.contrib.repl-utils
20:22	technomancy	clojurebot: show?
20:22	clojurebot	show is clojure.contrib.repl-utils/show
20:22	bpattison	excellent -- thanks!
20:22	technomancy	though if you use slime it's just C-c I
20:24	bpattison	technomancy: that worked perfectly!
20:25	jtal	when I try to "use" stuff and its not found- where should they be by default?
20:25	mrsolo	slime is great
20:48	jtal	clojure.contrib.* not included maybe?
21:47	blbrown_win2	what is the syntax for varialble args. (defn abc [a b [& c]] (when c (println c))) ...?
21:54	liebke	blbrown_win2: try (defn abc [a b & [c]] (when c (println c)))
22:00	texodus	make sure clojure-contrib is on your classpath
22:04	blbrown_win2	texodus, that is part of clojure-contrib?
22:15	jtal	is everyone eating dinner? :>
22:26	blbrown_win2	programmers don't eat
22:27	jtal	they look like they do
22:29	jtal	my current issue is resolved; the book uses contrib.str-utils and forgets to mention that its a separate thing to get
22:30	jtal	then when I try to find it the first thing I have is the assembla page which doesnt seem to have the code, then I find the google code page which has nothing under "downloads"; then I notice they are in the middle of moving and assume things are in limbo; finally, I find it under source on google page