#clojure – 2009-01-07

2009-01-07

01:20	durka	Exception in thread "pool-2-thread-2" java.lang.OutOfMemoryError: Java heap space
01:20	durka	does the agent system recover from this?
01:26	Chouser	I don't think so
01:28	durka	i think this may be a bug...
01:28	Chouser	If it's catchable/preventable, I would agree.
01:28	Chouser	I've got a simple test case here, just a sec.
01:28	durka	i saw the exception in the console, but i can't get at it from the REPL
01:28	durka	(agent-errors nil)
01:29	durka	(.getQueueCount agt) = 1
01:29	durka	i meant, (agent-errors agt) = nil
01:29	durka	and it normally finishes faster than this...
01:29	durka	i'll try giving the jvm more memory
01:30	lisppaste8	Chouser pasted "OOM in agent" at http://paste.lisp.org/display/73224
01:31	Chouser	durka: that's probably worth posting to the google group.
01:32	durka	shall i?
01:32	Chouser	please
01:33	Chouser	I really ought to be in bed.
01:33	durka	goodnight then
01:34	Chouser	:-) g'night
01:34	durka	and can i cite your test case?
01:34	Chouser	sure
02:06	durka	apparently agents can throw Exceptions, but when they throw OutOfMemoryErrors this weirdness happens
02:09	lisppaste8	durka annotated #73224 with "more direct test case" at http://paste.lisp.org/display/73224#1
02:14	lisppaste8	durka annotated #73224 with "more general" at http://paste.lisp.org/display/73224#2
02:14	durka	curiouser and curiouser
02:16	durka	i think maybe there is a "catch (Exception e) {" that should be a "catch (Throwable e) {"
08:06	AWizzArd	clojurebot: max people
08:06	clojurebot	max people is 116
08:08	duck1123	I would be cool if clojurebot could announce when the record is broken. Does it have hooks for join/parts?
08:09	knapr	does it keep track itself?
08:09	knapr	GUIs have never been easier than swing+miglayouy+clojure
08:45	AWizzArd	How can I convert a collection (vector of numbers between -128 and 127) into a byte array? (to-array col) produces an Object[] and (into-array col) produces an Integer[].
08:45	AWizzArd	(make-array Byte/TYPE (count col)) returns a byte array of the right size, but I can't fill it via doseq, as (aset ...) requires an index.
08:46	AWizzArd	rhickey_: would it be difficult to add an optional parameter to to-array that specifies the type?
08:46	gnuvince_	AWizzArd: (doseq [[i e] (map vector (iterate inc 0) col)] ...)
08:48	AWizzArd	thx
08:57	gnuvince_	AWizzArd: you could make that map a function too; I like the name "enumerate" from Python.
09:10	knapr	Im using a database, now derby but will probably switch to mysql or postgresql because it is a bit slow. however I read about some people storing as binary? what does that mean? storing the records as hashes and then accessing them like that?
09:10	AWizzArd	knapr: you should try #sql or read an introduction to sql.
10:18	rhickey	Chouser: (count (reduce #(cons %2 %) [0] (range 4601)))
10:18	rhickey	4602
10:18	rhickey	user=>
10:18	rhickey	(count (reduce #(cons %2 %) [0] (range 46010)))
10:18	rhickey	46011
10:18	rhickey	user=>
10:18	rhickey	(count (reduce #(cons %2 %) [0] (range 460100)))
10:18	rhickey	460101
10:18	rhickey	?
10:20	Chouser	that's what mine does too.
10:23	rhickey	heh, that was Java 5, on Java 6:
10:23	rhickey	(count (reduce #(cons %2 %) [0] (range 46010)))
10:23	rhickey	46011
10:23	aking	must vm dependent - I get: java.lang.StackOverflowError (NO_SOURCE_FILE:0)
10:23	rhickey	oops:
10:23	rhickey	user=> (count (reduce #(cons %2 %) [0] (range 46010)))
10:23	rhickey	java.lang.StackOverflowError (NO_SOURCE_FILE:0)
10:25	durka	did my message to the group go through?
10:32	AWizzArd	Is there an easy way to convert hex numbers between 0 and FF to a java (unsigned) byte?
10:32	rhickey	Chouser: I see this as a non-problem - why consing onto vector like that?
10:32	rhickey	repeatedly consing
10:32	AWizzArd	Byte/parseByte only accepts numbers between 0 and 7F
10:35	Chouser	rhickey: It was reported here, not in my own code. But it seems easy to write code that expects a list and cons's onto it a lot, then to have a user that's using a vector, passes it to your func, and it works a lot of the time but has a landmine lurking.
10:35	Chouser	shrug your call of course.
10:35	Chouser	AWizzArd: hex numbers encoded as one or two chars in a string?
10:36	rhickey	Chouser: consing onto non-lists was originally disallowed specifically because of this, but the bottom line is vectors aren't lists
10:36	rhickey	generally, people should use conj
10:36	AWizzArd	Chouser: as one/two chars in a string. (hex-to-signed-byte "F3")
10:38	drewolson	hey all, are the "correct" svn repos for both clojure and contrib on google code now?
10:39	pd	I think the primary clojure-contrib repo is on github now
10:39	pd	http://github.com/kevinoneill/clojure-contrib/tree/master
10:39	Chouser	AWizzArd: (.byteValue (Integer/parseInt "FF" 16))
10:39	Chouser	AWizzArd: does Java have such a thing as an unsigned byte?
10:39	pd	He also provides a mirror of the clojure svn if you'd prefer a git copy
10:39	drewolson	pd: i saw that a while ago, figured it was just a mirror
10:40	Chousuke	Chouser: apparently not
10:40	Chousuke	Chouser: java doesn't seem to have unsigned anything :/
10:40	AWizzArd	Chouser: cool thanks
10:40	Chouser	pd, drewolson: primary for clojure and contrib are SVN at code.google.com
10:40	drewolson	Chouser: ok, good, that's what i thought
10:40	pd	I didn't even know clojure-contrib was on code.google
10:41	pd	ty
10:41	Chouser	Chousuke: I approve. If the complexity of dealing with unsigned numbers was ever worth that one bit, it certainly isn't now.
10:42	Chousuke	Chouser: unsigned types make more sense for size types though. :/
10:43	Chousuke	AFAIK most sizes in java are ints
10:43	Chousuke	though even having "primitive" types like ints and longs is somewhat fail.
10:44	Chouser	yeah
10:46	Chouser	rhickey: if you're rejecting the patch on the grounds its unnecessary (or for any other reason, really) I'd appreciate very much a firm notification. Here is ok, but on the group is best.
10:47	rhickey	Chouser: sure, just discussing now, trying to understand the use case and options
10:48	Chouser	ok, wasn't sure if you were done or not. :-)
10:48	rhickey	also the difference between jdk 5 and 6 on this
10:49	Chouser	how is it possible that the jvm can avoid the stack overflow. It's clearly a non-tail mutually-recursive function, isn't it?
10:50	rhickey	Chouser: the JVM optimizer is always free to do TCO, just never promises it will
10:50	rhickey	TCO at language level is a promise
10:50	Chouser	but it's not even in the tail position
10:51	Chouser	1+RT.count(_rest)
10:51	rhickey	right, this is just some opt
10:51	Chouser	huh
10:51	rhickey	I agree :)
10:52	Chouser	I think I'm on OpenJDK 1.6 at the moment. I always forget how to check and change it.
10:57	rhickey	durka: fixed
11:00	rhickey	Chouser: in any case, I think we need Counted marker interface
11:02	durka	awesome
11:03	Chouser	rhickey: it would have other uses?
11:04	rhickey	Chouser: yes, there are several places where it could help - vector creation for example.
11:04	rhickey	that was the reason on the todo list
11:05	Chouser	I can't get to it anymore
11:06	Chouser	So Counted would mean its count() is faster than linear time?
11:07	rhickey	Chouser: right, Counted indicates O(1) count
11:08	danlarkin	at the cost of insertion time
11:08	rhickey	danlarkin: ?
11:08	danlarkin	right?
11:08	danlarkin	I mean, certainly a very small cost
11:09	rhickey	no, almost everything in Clojure is Counted
11:09	rhickey	already, just no marker
11:09	Chousuke	I think he means incrementing the counter instance variable.
11:09	rhickey	whatever cost there is to counting is already being incurred
11:12	durka	is clojure intended for use with java 5 or 6 or either?
11:12	rhickey	durka: 5 and up
11:39	knapr	http://paste.pocoo.org/show/98460/
11:39	knapr	im trying to connect to postgresql
11:39	knapr	the database and table is there
11:40	knapr	so i must be doing something els wrong
11:42	knapr	ah be damned colons!
11:56	pjb3	just discovered clojure contrib javadoc, very neat
11:57	pjb3	If you are at a REPL, and you have clojure contrib on your classpath
11:57	pjb3	(use 'clojure.contrib.javadoc)
11:57	pjb3	(javadoc String)
11:57	pjb3	and you get the javadocs for String in a swing frame
12:00	Chouser	hm, maybe that and the stuff from repl-utils should be combined?
12:02	Chousuke	pjb3: I wish it'd open in Firefox though.
12:02	Chousuke	pjb3: the swing thing is slow
12:03	Chousuke	or at least, was last time I tried. :/
12:03	danlarkin	Chousuke: I just tried it... still pretty slow
12:22	triddell	I'm trying to port an existing eclipse plug-in to use Clojure instead of Groovy. At runtime I'm currently getting the following stack trace:
12:22	triddell	java.lang.ExceptionInInitializerError
12:22	triddell	at java.lang.J9VMInternals.initialize(J9VMInternals.java:222)
12:22	triddell	at clojure.lang.Namespace.<init>(Namespace.java:31)
12:22	triddell	at clojure.lang.Namespace.findOrCreate(Namespace.java:116)
12:23	Chouser	(with-alter [my-var inc] ...) vs. (binding [my-var (inc my-var)] ...) ?
12:23	triddell	line 31 of Namespace.java is: mappings.set(RT.DEFAULT_IMPORTS);
12:23	triddell	what is RT in this instance?
12:23	Lau_of_DK	Good evening gentlemen
12:23	Chouser	triddell: paste the whole stack trace to paste.lisp.org?
12:24	danlarkin	good afternoon Lau
12:24	kotarak	Hola, Lau.
12:25	lisppaste8	triddell pasted "untitled" at http://paste.lisp.org/display/73240
12:28	Chouser	triddell: the "Caused by:" sections are often pretty key.
12:29	rhickey	right:
12:29	rhickey	Caused by: java.lang.IllegalArgumentException: URI scheme is not "file"
12:29	rhickey	at java.io.File.<init>(File.java:377)
12:29	rhickey	is the problem
12:30	Lau_of_DK	How difficult would it be to make some type of Fly-make for Clojure?
12:31	triddell	uh, ok, I'm stupid... I had a problem getting the clojure.jar to be found and this was the first error since... didn't pay attention very well... thanks
12:32	Chouser	it's a shame that slime apparently forces you to specifically ask to see anything more than the final cause.
12:37	Lau_of_DK	Whats involved in determining which line in the source is producing a given problem ?
12:38	Chouser	heh, um. It's a process called 'debugging' and is learned via much practice.
12:39	Lau_of_DK	So I cant expect to automate it ?
12:40	danlarkin	So I have a function in one namespace which calls (require ) a variable number of times based on its arguments. Calling (loaded-libs) in the namespace which calls my require-calling-function shows that the required-libs bleed through. Is there a way I can remove the lib from loaded-libs but still return a closure capable of calling a function from inside that lib?
12:40	danlarkin	ugh, that is pretty hard to read isn't it
12:41	Chouser	loaded-libs is global, isn't it?
12:42	Chouser	you asking if there's a way for it to not be?
12:42	danlarkin	yeah
12:42	kotarak	side question: wouldn't it make sense to check for the namespace instead of bookkeeping a list in require?
12:44	Chouser	that would only exacerbate danlarkin's problem.
12:44	cgrand	Chousuke, danlarkin: which JRE are you using and which OS? (clojure.contrib.javadoc/javadoc try to use a real browser before opening the swing browser)
12:45	Chousuke	cgrand: Java 1.5, Mac OS 10.5.6
12:45	danlarkin	cgrand: same as Chousuke
12:46	Chousuke	I guess finding the native browser does not work on OS X java.
12:46	cgrand	well it requires java 6 ...
12:46	Chousuke	right.
12:47	kotarak	On OS X it's just "open <url>"...
12:47	Chouser	ooh, works here -- javadoc comes right up in firefox.
12:47	Chousuke	true.
12:47	Chouser	Ubunutu
12:48	Chousuke	maybe add a check for OS X and use the system "open" instead of java stuff :/
13:04	joma	I must say jdbc+postgresql+pgadminIII+clojure rocks
13:05	Chousuke	yay, works
13:08	mfredrickson	technomancy: ditto. i was looking at that yesterday. neat stuff
13:09	technomancy	seems like you could use it for persistence instead of an RDBMS
13:10	joma	use what instead of rdbms?
13:10	technomancy	use terracotta
13:20	gnuvince_	What's terracotta?
13:21	technomancy	gnuvince_: it's a transparent clustering/persistence mechanism for the JVM
13:22	technomancy	you just tell it what data to track, and it makes it available to any JVMs in the cluster as if it were local
13:23	gnuvince_	That's pretty cool
13:24	gnuvince_	I'm looking at their web page: you just write your program exactly as you would and it automagically becomes clustered?
13:25	technomancy	you need to give it a little help just to tell it which pieces of data you want to share... maybe also tell it what terracotta server to connect to
13:25	technomancy	but it's really simple
13:27	technomancy	haven't actually used it yet; just read about it.
13:27	gnuvince_	ok
13:37	Chousuke	hm
13:38	joma	im using contrib.sql and it seems im evaing the functions outside the scope of the connection which means I should just pass the statement but eval it inside the function. so how can I do that? i pass (do-statement #(statm)) instead of (do-statement (statm))?
13:38	joma	im asking becasue i cant test right now, damn windows
13:38	Chousuke	my hack to contrib.javadoc is far from optimal, but here it is for the mac users: http://code.google.com/p/clojure-contrib/issues/detail?id=2
13:39	Chousuke	I didn't even bother searching for a 1.5 cross-platform solution :/
13:40	Chousuke	now it only needs some way to set the documentation URL to 1.5 instead of 1.6...
13:41	Chousuke	makes me think; would a OS identification library be useful for contrib?
13:47	philscotted	Is there some variable that is always bound to the last expression evaluated at the REPL, like * in CL?
13:48	Chousuke	philscotted: 1, 2, 3, e (last exception)
13:50	philscotted	Cheers! Sorry, I'm sure it's in the documentation somewhere, but I hadn't come across it.
13:50	gnuvince_	technomancy: about terracotta, does an instance of the application pick one of the terracotta server to execute on?
13:50	philscotted	Ack. Just found it. Right at the top of the API page.
13:51	technomancy	gnuvince_: I don't think so.
13:51	technomancy	I've only done the most cursory of examinations though.
13:51	gnuvince_	ok
13:52	gnuvince_	I'm wondering if it's the kind of thing were you add a new box, update the XML file and boom, you have more processing power to accomodate and larger number of concurrent users.
13:56	triddell	rhickey: looking back at this paste: http://paste.lisp.org/display/73240 is it possible that this error is caused by the clojure classloader within an eclipse/equinox/osgi environment? The error occurs when trying to initialize the first clojure-based class: solo.Generator on line 6
14:00	Lau_of_DK	Can someone here show me how to put 'binding' to good use?
14:00	leafw	Lau_of_DK: ?
14:01	leafw	you mean a good usage case?
14:01	leafw	the archetypical is for redirecting stdout
14:01	Lau_of_DK	Ok
14:01	Lau_of_DK	brb
14:02	leafw	(binding [out (.getSomeStdout an-instance)] .... any code .... )
14:03	leafw	any calls to println in any of the code called within the binding block will use that out
14:03	leafw	including from other functions indirectly called by functions within the block.
14:04	Lau_of_DK	k
14:04	Lau_of_DK	Thanks
14:04	knapr	http://paste.pocoo.org/show/98475/
14:04	leafw	Stuart's book calls it "acting at a distance", because it lets you redefine just about anything.
14:04	knapr	how do I replace that annoyingly ugly loop?
14:05	gnuvince_	(doc reduce)
14:05	clojurebot	f should be a function of 2 arguments. If val is not supplied, returns the result of applying f to the first 2 items in coll, then applying f to that result and the 3rd item, etc. If coll contains no items, f must accept no arguments as well, and reduce returns the result of calling f with no arguments. If coll has only 1 item, it is returned and f is not called. If val is supplied, returns the result of app
14:06	gnuvince_	knapr: with that
14:06	leafw	knapr: with map, likely, since you accumulate results into a vector. Just call vec on the resulting list from map.
14:07	Chouser	(reduce conj [] res)
14:08	Chouser	knapr: ^^
14:08	Chouser	knapr: except your loop stops when it finds a nil inside res, or reaches the end of res
14:08	Chouser	...while that reduce will include any nils, stopping only at the end.
14:08	Chouser	oh, even better: (into [] res)
14:11	knapr	what is the standard return-value for sideeffectful functions in Clojure? return nothing? it seems it is to return nil whihc I hate since nil==false
14:11	knapr	returning true makes more sense
14:12	gnuvince_	knapr: a function returns the value of its last expression
14:12	danlarkin	knapr: perhaps have it return what it has affected
14:12	gnuvince_	If your last expression is a println, that'll be nil.
14:13	knapr	but lets say evaling statements into a database
14:13	knapr	examplemin contrib.sql returns nil
14:13	knapr	id rather have true
14:13	knapr	what is idiomatic?
14:15	Chouser	I don't know that there's a strong idiom
14:15	Chouser	print and friends return nil
14:15	Chouser	send and send-off return their first arg
14:16	leafw	Chouser: ref and agent -friendly functions return always the first arg ... because it's the ref itself.
14:18	Chouser	I use the return value from 'send' about as often as the return value from 'print'
14:18	knapr	Chouser: anyway why cant I just return res directly but have to perform that function?
14:20	StartsWithK_	triddell: do you have a working osgi bundle using clojure? if so, i would like to know how you created it
14:20	Chouser	knapr: with-results uses resultset-seq, which is a lazy seq over query results
14:22	triddell	StartsWithK_: not yet, working one right now
14:23	StartsWithK_	triddell: how far did you go? like, can you register clojure.jar as separate bundle and have it work with another that has some clojure code in it?
14:23	StartsWithK_	or you have to include clojure.jar with your own bundle
14:27	triddell	StartsWithK_: I'm building an eclipse plugin right now and I've tried separate bundles an d combined. I think I have a classloading issue as soon as a clojure compiled class is called.
14:27	triddell	RIght now trying to add: Class.forName("clojure.lang.RT") in bundle's start() according to this: http://groups.google.com/group/clojure/browse_thread/thread/d8e577b5bf41c1bb/4202854a7ed14aca?lnk=gst&q=osgi#4202854a7ed14aca
14:28	triddell	But it's not working yet
14:28	StartsWithK_	i had the same problem
14:28	StartsWithK_	"solved" it by including clojure.jar with my bunde, but had to do that for every bundle
14:29	StartsWithK_	and i couldn't make it talk with other bundles, classloader problem
14:29	triddell	were you creating an eclipse plug-in or an osgi bundle?
14:30	StartsWithK_	i tried equinox and knoplerfisg as osgi containers with clojure, none worked
14:30	StartsWithK_	i wanted to use osgi in my app
14:32	StartsWithK_	at the end i was able to create one mega bundle with everything inside it, no dependencies to other bundles.. not much point in that so i gived up
14:34	triddell	StartsWithK_: bummer... well I'm still banging my head on it.. for this app one bundle should work ok, since I'm just one of many bundles inside eclipse... I can see how it wouldn't work out for you... Do you have a Manifest from your mega-bundle that worked for reference?
14:35	StartsWithK_	triddell: sorry i don't, it was couple of months ago and i deleted it.
14:35	triddell	StartsWithK_: np, you've given me some hope though :-)
14:37	StartsWithK_	triddell: biggest problem was using (proxy), when i would create object with it and used it as argument to method with object from another bundle, i would get strange classloader errors
14:37	StartsWithK_	maybe creation of such proxy object should go inside (binding [use-context-classloader true] ..)
14:37	StartsWithK_	i don't know, its not documented anywhere how to use it
14:39	Chouser	StartsWithK_: It's likely the way proxy creates and loads classes may have changed since you tried.
14:39	triddell	I'm not using Proxy so I might not run into that.
14:40	StartsWithK_	Chouser: Problem is that when proxy object (maybe even a normal one) is created in one bundle it gets his classloader from RT, and then when used it another bundle it still remember its old classlader, but osgi will change that during objects lifetime
14:41	StartsWithK_	i don't think that changed, there is still one classloader that RT initializes in its static part
14:49	kotarak	Could this be considered a bug? (use/require :reload some.name.space) does not reload files included with load.
14:50	Chouser	StartsWithK_: but now you have the option of AOT compiling code, including the proxy, and then no special classloader is needed at runtime, even if you use a new proxy definition (as long as it has the same base classes as an already-compiled proxy)
14:51	StartsWithK_	Chouser: is then use-context-classloader unused now? if it was used before for anything
14:52	rhickey	StartsWithK_: yes, it is still used
14:53	StartsWithK_	rhickey: can you explain how it should be used
14:54	rhickey	one of the problems I have with people using osgi for Clojure is I never get a sense of why - i.e. what is the objective, or is some environment forcing it? Also, where is there a short description of how osgi supports dynamic bytecode?
14:55	StartsWithK_	by environment: glassfish v3 and eclipse, apache felix is ported do android
14:56	StartsWithK_	netbeans uses something similar to serviceloader from java6, but serviceloader uses same classloader for 'bundles' as for main app
14:56	rhickey	AOT-compiled Clojure should look like any other Java app
14:57	rhickey	Clojure is not a service, so sharing it between modules would be an encapsulation violation - probably want one instance of Clojure per module
14:58	StartsWithK_	scala is not a service, but you can use scala standard library as osgi bundle (requires small change to manifest file)
14:58	StartsWithK_	it dosn't need to have any special osgi support in it
14:59	StartsWithK_	any .jar can be converted to osgi bundle by adding extra metadata to manifest
14:59	StartsWithK_	it then is not registered as service, more like its on 'classpath' for other bundles
15:00	rhickey	scala is a static language whose namespaces are java packages. Clojure is a dynamic language whose namespaces are in data structures you might not want to share
15:00	rhickey	but if you did you could
15:00	rhickey	seems like most osgi problems relate to dynamic bytecode, not a nissue for scala
15:00	StartsWithK_	only requirement i could find is that any code in it dosn't hold on to his classloader
15:00	Chousuke	hm
15:01	rhickey	StartsWithK_: right, but there is no standard classloader that loads bytecode from bytes publicly, so dynamic bytecode requires custom classloader
15:01	walters	rhickey: basically the java classpath and package concepts just aren't good enough as a module system
15:01	StartsWithK_	well i can take jetty.jar and turn it to osgi bundle, and jetty has no direct osgi support, or log4j or any other 'normal' jar
15:02	rhickey	StartsWithK_: and when you take AOT clojure.jar what happens?
15:02	StartsWithK_	i haven't tired, it was long before AOT
15:02	StartsWithK_	i should create new test suits with AOT
15:03	StartsWithK_	but it was such a pain last time to try all the combinations
15:03	rhickey	I'd much rather someone who understands how osgi supports dynamic bytecode would point me to a short reference
15:04	StartsWithK_	i'm not that someone :) but i'll try to find slides from knoplerfish project that have some problems outlined in them
15:04	joma	http://paste.pocoo.org/show/98494/
15:05	joma	why doesnt that work when (Hmm) does?
15:05	joma	(hmm)
15:05	joma	ah wait
15:07	joma	well that returns nil
15:07	joma	but doesnt happen
15:11	joma	how do I call getNextException ?
15:12	gnuvince_	Is there a quick tutorial on java.util.concurrent so that I could know what's in it?
15:13	gnuvince_	I know Rich often recommends using it
15:16	philscotted	Anyone know if this issue has been resolved? http://groups.google.com/group/clojure/browse_thread/thread/43a413ae55ed25d0
15:16	philscotted	I'm adding a jar file with add-classpath but imports are not working.
15:16	philscotted	However, I can refer to the classes by their fully qualified names.
15:18	danlarkin	lisppaste8: url
15:18	lisppaste8	To use the lisppaste bot, visit http://paste.lisp.org/new/clojure and enter your paste.
15:19	lisppaste8	danlarkin pasted "refactoring help" at http://paste.lisp.org/display/73247
15:20	danlarkin	so in the patterns function I just pasted I have the same (let ) block on lines 16 and 23
15:20	danlarkin	but I am not sure how to remove the duplication
15:22	StartsWithK_	rhickey: i don't know if it helps, http://www.eclipsecon.org/2008/?page=sub/&id=145 link to pdf is under the title
15:24	kotarak	I think I found a bug: when using AOT compilation (require :reload-all 'some.name.space) does not correctly reload all "loaded" files! Only the master file of the namespace gets reloaded.
15:31	joma	how do I call getNextException ?
15:31	kotarak	How is the general procedure regarding the Issue Tracker? Should I post this to the list or should I add an issue?
15:32	kotarak	joma: I'm not sure, what need, but that should give you a start: http://clojure.org/java_interop
15:32	Chouser	kotarak: it seems like that may be the current intended behavior. If there's any doubt at all as to whether it should be fixed, I think the list the right place to start.
15:32	rhickey	kotarak: for that one, please create an issue, then post on the list with a link to issue - thanks
15:33	rhickey	StartsWithK_: not any help
15:33	kotarak	rhickey: ok, will do.
15:34	rhickey	Chouser: I'm sure it doesn't work, but reloading needs to be addressed - right now the built-in make-like logic prevents reloading
15:34	triddell	StartsWithK_ and rhickey: I think I just found a major issue with clojure on equinox/osgi. I now have a working eclipse plugin with clojure.
15:34	rhickey	Chouser: you
15:34	rhickey	Chouser: your advice still generally correct, I pre-approve this report
15:34	Chouser	heh, ok.
15:35	rhickey	triddell: what's the trick
15:35	rhickey	?
15:35	StartsWithK_	triddell: yes, share please
15:36	triddell	rhickey: In RT.java, lastModified function... url.getProtocol() checks for "jar" and if it doesn't get it falls through to file... this produces:
15:36	triddell	Caused by: java.lang.IllegalArgumentException: URI scheme is not "file" at java.io.File.<init>(File.java:377) at clojure.lang.RT.lastModified(RT.java:351)
15:36	triddell	if I force the "jar" line: I get:
15:37	triddell	Caused by: java.lang.ClassCastException: org.eclipse.osgi.framework.internal.core.BundleURLConnection incompatible with java.net.JarURLConnection at clojure.lang.RT.lastModified(RT.java:347)
15:37	triddell	if I just return 1 it works!
15:38	rhickey	do BundleURLConnection and JarURLConnection share any superinterfaces?
15:39	triddell	I'm only within one bundle at this point but I was getting the same initial error when using multiple bundles
15:39	triddell	not sure, didn't check into that yet... will do though
15:39	StartsWithK_	clojure.jar is inside your bundle? or separete one?
15:39	triddell	the protocol also must be returning something other than "jar" as well
15:40	triddell	StartsWithK_: inside right now, but was getting same initial before
15:40	triddell	initial error
15:44	triddell	rhickey: no interface but base class java.net.URLConnection it appears
15:46	rhickey	triddell: fixed hard else to File in lastModified - svn 1200
15:47	zakwilson	Is anybody using Clojure with the current CVS version of Slime? I'm having a problem.
15:48	Chousuke	zakwilson: let me guess: no fancy repl?
15:48	joma	HOW DO I CALL getNextException?
15:48	zakwilson	Chousuke: yes.
15:48	Chousuke	zakwilson: wait a moment
15:49	Chousuke	http://github.com/Chousuke/dotfiles/tree/master/Aquamacs-customizations.el#L59 you need that
15:49	kotarak	joma: 21:30 joma: I'm not sure, what need, but that should give you a start: http://clojure.org/java_interop
15:50	zakwilson	Chousuke: Thanks. I'll give that a try.
15:54	zakwilson	Problem solved. Adding slime contrib and running slime-repl-init fixed it. Thanks Chousuke.
15:54	triddell	rhickey: that did it, thanks!
15:55	Chousuke	cgrand: your fix for OS X works too.
16:00	joma	how do I call printStackTrace?
16:00	Chousuke	just call it?
16:00	Chousuke	(.printStackTracke myexception)
16:30	knapr	when passing a string of "2008-12-02" to contrib/sql into a column with the type DATE it works in derby but it postgresql I get nosuchfieldexception. anyone know why?
16:36	knapr	i need to pass '2008-12-02' instead of "2008-12-02", how?
16:36	hiredman	I wonder if it might want a date object instead of a string?
16:37	hiredman	I haven't used contrib/sql
16:38	Chouser	knapr: are you using do-prepared?
16:38	danlarkin	ugh it's so hard to choose a good name for software :-/
16:39	Chouser	it's hard to choose a good name for anything
16:49	Chouser	rhickey: print-dup output has a whole lot of PersistentArrayMap/create in it now
16:52	joma	I wish github would format clojure better
16:52	rhickey	Chouser: yep - the problem is I can't tell when it is important to someone that they get back an ArrayMap. To be correct it needs to know the threshold for when using {} yields ArrayMap
16:52	rhickey	those cases can be turned into {} literals
16:54	Chouser	print-dup of course has to be careful, I just thought that if you were no guaranteeing {} would default to PersistentArrayMap below a certain size, print-dup could take that into account.
16:54	rhickey	8 items or less
16:54	joma	why does clojure code look like it is leaning?
16:54	Chouser	might also just be unnecessary fragility.
16:55	Chouser	joma: why does java code look like it's trying to consume as many lines as possible?
16:55	rhickey	user=> (class {1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8})
16:55	rhickey	clojure.lang.PersistentArrayMap
16:55	rhickey	user=> (class {1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9})
16:55	rhickey	clojure.lang.PersistentHashMap
16:55	Chouser	joma: sorry, that sounded testier than I meant it. Clojure has this is common with most lisps
16:57	_hrrld	Coming from a C++/Ruby/Scheme background (never having used java)... I want to write the result of my clojure program out to disk and I can't figure out how. Do I need to use java facilities to write files with clojure?
16:58	Chouser	_hrrld: it's fine to use Java classes directly. You have a string you want to write out?
16:59	_hrrld	Yup.
17:00	Chouser	_hrrld: I don't think there's anything in clojure.core to help you, so you can either look at clojure.contrib.duck-streams, or just use the Java calls directly.
17:00	joma	Can't infer the SQL type to use for an instance of java.util.Date.
17:01	_hrrld	So it would be a process of learning some Java file I/O, then translating it into clojure?
17:01	Chouser	_hrrld: yes, or looking at existing clojure examples, or using duck-streams.
17:02	hiredman	actaully the I/O thing is in the faq on the website
17:02	hiredman	well
17:02	_hrrld	Can you point me to an example of file I/O, I didn't see anything in the reference docs.
17:02	hiredman	I thought it was in the faq
17:02	Chouser	joma: looks like you want java.sql.Timestamp
17:02	hiredman	where did I see that
17:02	Chouser	joma: http://github.com/stuarthalloway/programming-clojure/tree/master/examples/snippet.clj
17:04	hiredman	_hrrld: you can (doto (java.io.FileWriter. "somefile") (.write "string") (.close))
17:04	Chouser	Which is essentially what 'spit' does: http://code.google.com/p/clojure-contrib/source/browse/trunk/src/clojure/contrib/duck_streams.clj#177
17:04	hiredman	there is also with-open
17:06	joma	ime - milliseconds since January 1, 1970, 00:00:00 GMT. A negative nu
17:06	_hrrld	what does the '&' mean in the docs for with-open? (from: http://clojure.org/api)
17:06	joma	yeah so how do I convert?
17:07	Chousuke	_hrrld: it means it's a variable-length argument
17:07	Chouser	varargs
17:07	Chousuke	_hrrld: your code is given as a parameter to the with-open macro
17:08	Chousuke	_hrrld: (with-open [myfile (java.io.FileWriter "out.txt")] (do-stuff-with myfile) (more-stuff-with myfile))
17:08	Chousuke	oops
17:08	Chousuke	missing a .
17:08	_hrrld	That's really helpful. Much appreciated. (:
17:09	Chousuke	with-open also supports multiple name, init pairs
17:10	_hrrld	And closes them all automatically, I presume?
17:10	Chousuke	yes
17:10	Chousuke	in reverse order of initialisation
17:12	_hrrld	"java.lang.IllegalArgumentException: No matching method found: .write" (Am I missing an import, or something?)
17:13	Chouser	that more indicates incorrect parameters to .write
17:14	Chousuke	you probably forgot to give the file as the first argument
17:14	Chouser	make sure the first arg really is a Writer, and the the rest are of types supported by the .write method
17:14	hiredman	or you are calling a write method on something that has no write method
17:14	Chousuke	or something.
17:14	_hrrld	Trying this: (doto (java.io.FileWriter. "a.txt") (.write "foo") (.close))
17:14	Chousuke	right
17:14	Chousuke	hmm
17:15	hiredman	wait
17:15	hiredman	what version of clojure?
17:15	Chouser	hiredman: ah, good call.
17:15	_hrrld	clojure_20080916
17:15	Chousuke	old
17:15	_hrrld	(!)?
17:15	Chousuke	update from google code :)
17:16	_hrrld	haha. The way to write files out has changed in the last two months?
17:16	Chousuke	clojurebot: download?
17:16	clojurebot	It's greek to me.
17:16	Chousuke	clojurebot: download is http://code.google.com/p/clojure/downloads/list
17:16	clojurebot	Ok.
17:16	Chousuke	_hrrld: no, doto has changed
17:16	Chousuke	_hrrld: the old version of doto did not need the . for java methods
17:16	_hrrld	Neat. I'll get the latest and try again.
17:17	Chousuke	_hrrld: many other changes too
17:17	hiredman	which is why the exception shows the method as ".write" instead of "write"
17:17	Chousuke	_hrrld: for example, trying to open multiple files in with-open with that version would have worked, but there was a bug so that the files other than the first one never got closed
17:18	Chousuke	_hrrld: and there's precompilation, clojure.main, and all kinds of neat stuff.
17:18	Chousuke	or was clojure.main after that release? :/
17:18	_hrrld	Cute. Hey, it worked.
17:19	hiredman	clojure.main?
17:20	Chousuke	hiredman: the new startup thing
17:20	Chousuke	hiredman: clojure.lang.Repl and Script are deprecated now.
17:21	drewolson	Chousuke: hrm, didn't know that. does main handle both scripts and repl with no arg?
17:21	Chouser	I suppose that means I should switch over.
17:21	Chousuke	try java -cp clojure.jar clojure.main --help
17:22	hiredman	oh really
17:22	Chousuke	or java -jar clojure.jar --help I guess
17:22	hiredman	check that out
17:26	powr-toc	Chousuke: cool... I hadn't realised that had been implemented
17:30	drewolson	Chousuke: ahh, so much better this way, thanks.
17:30	red_fish	where is documentated #()?
17:31	vogelrn	red_fish: it's mentioned in the reader section of the webpage
17:32	rhickey	Chouser: got your test.html from you appletviewer test handy?
17:32	red_fish	vogelrn: thx
17:32	rhickey	trying to get applets working
17:34	Chouser	rhickey: dunno what code you're looking at, but here's what's in my clojure-applet directory:
17:34	Chouser	<applet code="net.n01se.HelloWorld" archive="hello.jar" width="800" height="800"></applet>
17:35	rhickey	Chouser: thanks - looking at: http://groups.google.com/group/clojure/msg/6bc1c0fe0165053f
17:37	Chouser	ok
17:40	Chouser	well, shoot. clojure.main doesn't do what I want, and I've accidentally overwritten my old clojure startup script.
17:40	drewolson	Chouser: i just actually rewrote mine, want me to pastie?
17:40	Chouser	heh. if you want.
17:40	Chouser	clojure.main is close
17:41	drewolson	http://pastie.org/355135
17:41	Chouser	but I want to provide a startup script for when it runs a repl
17:41	Chouser	and I want to be able to add to the classpath on the commandline
17:42	drewolson	ah, gotcha
17:42	Chouser	I'm amazed by how personal these startup scripts are.
17:42	drewolson	heh, yep
17:42	Chouser	I assumed we all wanted roughly the same thing (nobody has their own python or ruby startup script), but in fact they're all quite different.
17:44	lisppaste8	rhickey pasted "applet load error" at http://paste.lisp.org/display/73258
17:44	rhickey	anyone know applets well enough to explain above error?
17:45	walters	rhickey: i'm guessing it means you can't refer to any class sun.awt.*
17:45	hiredman	interesting
17:45	walters	rhickey: maybe the AOT compiler is traversing the class hierarchy in a way that javac doesn't?
17:45	Chouser	I saw this.
17:45	powr-toc	rhickey: are you trying to popup a window or anthing like that?
17:46	Chouser	dangit. I can't remember what it was.
17:46	Chouser	proxy tries to implement things it shouldn't. or something.
17:47	cmvkk	by the way, if you guys are working at making clojure applets work, then that is very exciting.
17:47	cmvkk	i was just trying to do it earlier today and it didn't work.
17:47	rhickey	just this simple class:
17:47	rhickey	(ns clojure.examples.applet
17:47	rhickey	(:gen-class
17:47	rhickey	:extends java.applet.Applet))
17:48	Chouser	something descends into the Applet heirarchy and finds platform-specific classes that it's not supposed to touch
17:49	Chouser	I'm very agitated. I had this, and I think I found a work-around.
17:49	fffej	I've got a confusing error - No matching method found: indexOf for class java.lang.String - can anyone point me into the right direction?
17:50	rhickey	Chouser: cuase it seems like I'm past the classloader problem
17:50	rhickey	cause
17:50	Chouser	rhickey: excellent!
17:50	Chousuke	fffej: are you giving it the right argiuments?
17:50	rhickey	seems being the operative word :)
17:50	Chousuke	arguments*
17:51	fffej	chousuke: it's taking a JTextFIeld.getText() for both arguments and works most of the time until an exception is thrown
17:51	aking	,(.indexOf "Hello" "o")
17:51	clojurebot	4
17:51	Chouser	hm, I have :excludes #{"applyCompoundShape"} in my gen-class example
17:51	rhickey	basically I've made ROOT_CLASSLOADER load-on-demand, will never create a dynamic loader if no dynamic code
17:52	hiredman	the javadoc for runtime permission has a description of the accessClassInPackage permission, but it is greek to me
17:52	Chouser	apparently I had a patch against gen-class, but I don't seem to have it in my git repo
17:53	Chouser	I'm pretty sure that :excludes is to work around the access error
17:55	durka	rhickey: where does add-classpath go then?
18:04	fffej	i'm not understanding re-find. (re-find #"bar" "bar") => "bar" whereas (re-find #"(foo)\|(bar)" "foo bar") => ["foo" "foo" nil] - why does one return a vector and one just the result directly?
18:05	durka	looks like the second is capturing groups?
18:06	fffej	but shouldn't the return type be consistent? (e.g. a vector of 1 in the first example), otherwise my code has to check whether the return val is a collection which doesn't feel right
18:06	hiredman	,(re-find #"(foo)\|(bar)" "foo bar")
18:06	clojurebot	[foo foo nil]
18:07	hiredman	,(re-find #"bar" "foo bar")
18:07	clojurebot	bar
18:07	hiredman	,(re-find #\w" "foo bar")
18:07	clojurebot	Eval-in-box threw an exception:No dispatch macro for: \
18:07	hiredman	,(re-find #"\w" "foo bar")
18:07	hiredman	clojurebot: well?
18:07	clojurebot	f
18:07	clojurebot	excusez-moi
18:08	cmvkk	f?
18:08	hiredman	,(prn "f")
18:08	clojurebot	nil
18:08	clojurebot	"f"
18:09	hiredman	I guess clojurebot should capture a prn into a stringwriter instead of just calling .toString
18:10	fffej	Is the weirdness with re-find me being daft, or should I file a bug report?
18:11	joma	Why can't I load a namespace into the repl? #'progs.netflix.parsing/parse-intervals works but when I try to us it it fails
18:11	joma	(parsing/str-to-date "2008-12-02")
18:11	joma	java.lang.Exception: No such namespace: parsing (NO_SOURCE_FILE:44)
18:11	joma	user=> (str-to-date "2008-12-02")
18:11	joma	java.lang.Exception: Unable to resolve symbol: str-to-date in this context (NO_SOURCE_FILE:45)
18:11	joma	user=>
18:12	hiredman	how are you trying to "load" the namespace?
18:13	joma	ah (progs.netflix.parsing/str-to-date "2008-12-02")
18:17	leafw	joma, how about (load-file "progs/netflixs/parsing") ?
18:18	hiredman	(doc require)
18:18	clojurebot	Loads libs, skipping any that are already loaded. Each argument is either a libspec that identifies a lib, a prefix list that identifies multiple libs whose names share a common prefix, or a flag that modifies how all the identified libs are loaded. Use :require in the ns macro in preference to calling this directly. Libs A 'lib' is a named set of resources in classpath whose contents define a library of
18:26	zakwilson	I'm having trouble getting Slime to eval expressions in the correct namespace. Has anybody else had a similar problem?
18:38	powr-toc	will clojure ever ship with contrib?
18:43	durka	(doc re-groups)
18:43	clojurebot	Returns the groups from the most recent match/find. If there are no nested groups, returns a string of the entire match. If there are nested groups, returns a vector of the groups, the first element being the entire match.; arglists ([m])
18:44	durka	fffej: so if it is a bug, it is a documented bug anyway
18:44	hiredman	,(prn "f")
18:44	clojurebot	nil
18:44	clojurebot
18:44	clojurebot	"f"
18:45	hiredman	hmmm
18:45	hiredman	,(identity "f")
18:45	clojurebot	f
18:45	clojurebot
18:45	clojurebot
18:46	hiredman	we will call that a regression
18:49	hiredman	,(identity "f")
18:49	clojurebot	f
18:49	vladsharp	can you see how a macro was expanded?
18:50	danlarkin	(doc macroexpand)
18:50	clojurebot	Repeatedly calls macroexpand-1 on form until it no longer represents a macro form, then returns it. Note neither macroexpand-1 nor macroexpand expand macros in subforms.; arglists ([form])
18:51	vladsharp	ah ha, thanks
18:54	hiredman	,(identity nil)
18:54	clojurebot	nil
18:54	Chousuke	fun to see you developing clojurebot :P
18:54	Chousuke	something magically changes and then the bot behaves differently!
18:55	hiredman	so the result is now (prn ...) to a stringwriter so we get quotes and stuff
18:57	joma	(defn split [string delim] (seq (.split string delim))) , into core?
18:59	hiredman	oops
19:00	hiredman	I think I just commited to github with the wrong commit message
19:04	joma	why does clojure code look like it is leaning?
19:04	joma	(defn split [string delim] (seq (.split string delim))) , into core?
19:05	joma	can anyone add to clojurebot?
19:10	powr-toc	hiredman: if you've not pushed the commit you can git commit --amend
19:11	hiredman	powr-toc: I am figuring that out
19:12	powr-toc	hiredman: but if you've pushed it things are uglier
19:13	hiredman	yeah
19:13	hiredman	well it's ugly now
19:13	powr-toc	lol
19:13	hiredman	I think I will leave it
19:13	hiredman	damn it
19:13	hiredman	now I have two duplicate commit messages
19:13	hiredman	yeah, time to leave be
19:13	powr-toc	hiredman: is it your project? And is it highly unlikely anyone has pulled?
19:14	hiredman	yes, and yes
19:14	hiredman	I can just leave it be
19:14	hiredman	I don't mind too much
19:15	danlarkin	too late, hiredman, it's already in my feed reader as a duplicate message :)
19:16	hiredman	I know
19:16	dreish	People will be talking about it for years.
19:16	hiredman	there are 2 duplicate messages now
19:16	hiredman	4 total messages
19:17	powr-toc	hiredman: it's probably reasonably easy to recover from by deleting the remote branch, rebasing it (so long as no one else has pulled it) and pushing back again
19:17	joma	anyone have an example of doing a prepared statement with contrib.sql?
19:18	hiredman	the "merge branch live" thing is new
19:18	hiredman	live being the name of the branch I run clojurebot out of
19:18	hiredman	must be because I made the commits in that branch?
19:18	rhickey	Applet started!
19:18	hiredman	!!
19:18	durka	:D
19:19	danlarkin	great success!
19:19	hiredman	,(Date.)
19:20	clojurebot	#<Date Wed Jan 07 16:18:00 PST 2009>
19:20	hiredman	clojurebot: Wed Jan 07 16:18:00 PST 2009 is <reply>"Applet started!" -- rhickey
19:20	clojurebot	You don't have to tell me twice.
19:20	rhickey	last prob was that gen-class and proxy both gen stubs for package-private methods, shouldn't
15:20	durka	when java is looking through the classpath for imports, how does it decide which classes to make available?
15:21	durka	it kind of looks like it ignores the versions of plain .class files, but checks the versions in .jar manifests against the current JVM version
15:21	durka	does anyone know if this is the case?
15:47	rhickey	Chouser: any tests for binding-assertions patch? looks ok, but I haven't exercised it
15:47	pjb3	rhickey: Applet started? What was that about?
15:48	rhickey	pjb3: got a dummy applet running using AOT, resolved classloader and security issues
15:48	rhickey	svn 1201
15:48	pjb3	cool, does that mean you can run Clojure code in an applet?
15:49	rhickey	yes, when AOT compiled
15:49	pjb3	Cool, it would be cool if we had something like this for Clojure: http://sisc-scheme.org/sisc-online.php
15:50	pjb3	A REPL in an Applet, would that still work? Because you would be defn'ing stuff from a REPL in an applet
15:50	hiredman	clojurebot: Jan?
15:50	clojurebot	"Applet started!" -- rhickey
15:50	rhickey	pjb3: that is not going to happen, at least not without a cert - Clojure is not interpreted and the default security prohibits custom classloaders needed for eval
15:51	pjb3	Ah, that's what I thought
15:51	rhickey	clojurebot: svn?
15:51	clojurebot	svn is http://clojure.googlecode.com/svn/trunk/
15:51	pjb3	Step one would be making a swing REPL I guess
15:51	albino	why not do what all the cool kids do and put a REPL as a server process talking over a socket to an ajax front end
15:52	pjb3	then futzing around with a signed applet or whatever to get the custom classloaders, assuming that is possible
15:52	pjb3	albino: Yeah, I thought about that, but then it's hard to have a general purpose "Try Clojure Online!" thing
15:52	hiredman	albino: hmmmmm
15:52	pjb3	right?
15:52	pjb3	because then everybody is connected to the same repl, defing things, etc.
15:53	powr-toc	how would you create a constant unmodifiable reference in clojure?
15:54	powr-toc	i.e. an immutable pointer to an immutable object.
15:54	albino	IIRC that's how this one works http://tryruby.hobix.com/
15:54	rhickey	powr-toc: in a ref?
15:54	albino	pjb3: well you can have a process per user, or thread per user...whatever
15:55	pjb3	albino: right, then you got to host that and everything
15:55	albino	pjb3: perfect, so you're volunteering then? :)
15:55	pjb3	Where as if you had an applet, you can server it as a static file
15:55	hiredman	albino: the main diff between that and an applet is an applet runs on the user's computer and uses their resources
15:55	hiredman	so you can let them do everything clojure does
15:55	albino	hiredman: right
15:55	albino	I was just speaking on the fastest way to get a "try it now in a browser" clojure thing going
15:56	hiredman	but, uh, you need to sandbox a repl running on your own machine
15:56	hiredman	and effective sandboxing blows
15:56	pjb3	What about that web start stuff? That just downloads a java app and runs it, right?
15:56	albino	hiredman: I assume you're already sandboxing clojurebot?
15:56	powr-toc	rhickey: not sure... I'm just trying to grok the difference between var's, ref's, agent's etc... I really just want a constant value equivalent to a final reference to something
15:56	hiredman	albino: yes
15:56	hiredman	albino: but not very effectively
15:56	pjb3	Well, either way, step 1 is a swing gui REPL
15:56	albino	hiredman: so you've already got a start at it
15:56	rhickey	powr-toc: to prevent accident or malice?
15:57	hiredman	I mean, no IO out of the jvm
15:57	hiredman	but you can still dealock the bot if you try hard
15:57	albino	what would be fun (maybe not practical) is forking a openvz container per user
15:57	hiredman	deadlock
15:57	albino	OS level sandboxing
15:57	powr-toc	accident
15:57	hiredman	albino: or just find someone who really understands jvm sandboxing
15:58	pjb3	Of Course I guess it should be that hard for people to download a jar file and run "java -jar clojure.jar" from the command line
15:58	hiredman	I just cargo culted the jvm sandboxing that clojuebot uses
15:58	albino	hiredman: who is the expert on that?
15:58	hiredman	no clue
15:58	rhickey	powr-toc: most Clojure data structures are immutable
15:59	albino	ha, just found this in my bookmarks http://calumleslie.blogspot.com/2008/06/simple-jvm-sandboxing.html
15:59	albino	I wonder if you could build something with this http://www-jpc.physics.ox.ac.uk/index.html
16:00	albino	maybe with JPC you could get it to run on the client machine without jumping through applet hoops
16:00	hiredman	ugh
16:00	powr-toc	rhickey: yeah, I understand that... I'm really just after the idiomatic way of doing it... the value will be accessible across a variety of threads. I'm wondering if there is effectively a way to freeze a reference to a particular object
16:00	powr-toc	do I just use def?
16:01	rhickey	powr-toc: how is it getting passed to the threads?
16:01	rhickey	people don't usually accidentally say alter/swap!/set! etc
16:02	rhickey	but you could set up a validator that will reject all changes like this:
16:02	rhickey	(let [aval {:never :change}]
16:02	rhickey	(def aconst (ref aval :validator #(identical? % aval))))
16:02	rhickey	(dosync (alter aconst assoc :a 1))
16:02	rhickey	java.lang.IllegalStateException: Invalid reference state (NO_SOURCE_FILE:0)
16:02	powr-toc	rhickey: it's not yet :-) - I'm thinking it'll be initialised immediately at start up though and I'd like it to be visible to all threads
16:02	rhickey	same technique could work for vars/atoms
16:04	powr-toc	ok... It's no biggy really... it's just when I'm in java I'm used to using the final modifier not only to help enforce immutability, but to also indicate intent... i.e. this reference won't ever be changed.
16:04	zakwilson	If you create a var with def, isn't it only rebindable thread-locally?
16:05	rhickey	powr-toc: that's kind of the default intent in Clojure
16:05	powr-toc	rhickey: lol - good point! :-)
16:07	powr-toc	rhickey: In which case presumably idiomatic clojure is simply to declare such values with something like (def pi 3.14159) and be done with it
16:07	powr-toc	which suits me
16:07	rhickey	yeah
16:08	pjb3	But if you can do (ns-unmap 'user 'pi) (def pi 42), right?
16:08	hiredman	is there a lazy-seq of the digits of pi yet?
16:09	mmcgrana	Anyone have a good solution for the problem of setting up different "environment" in e.g. Compojure? For example in a webapp I have a "host" config option that needs to be different in deployment and production. I've thought through several possibilities but none were satisfactory, hopefully someone has encountered this problem before.
16:09	mmcgrana	* different "environments"
16:09	pjb3	mmcgrana: specify the environment as a startup option, load a .clj matching the environment?
16:09	pjb3	production.clj, develoment.clj?
16:10	pjb3	environment.clj loads for everything?
16:10	mmcgrana	how do you suggest "specifying the environment as a startup option"?
16:10	danlarkin	mmcgrana: I haven't used compojure, but in django it's commonly solved with having settings.py import settings_local.py -- where settings.py defines settings common to dev, prod, stage, etc. and settings_local.py defines database connection settings etc
16:10	mmcgrana	excuse my ignorance in advance...
16:10	pjb3	command-line argument
16:10	pjb3	assume development if not present
16:11	pjb3	Does compojure run in a servlet engine like Jetty?
16:11	pjb3	Then maybe a JNDI or web.xml setting
16:12	mmcgrana	I think so, I'm not actually using compojure I just thought people who had been using it would have thought about the issue
16:12	pjb3	mmcgrana: not using it either
16:13	mmcgrana	btw the basic environment.clj, production.clj etc approach sounds basically right, i just have to get the mechanics down
16:14	powr-toc	mmcgrana: the environment.clj development.clj production.clj approach is used heavily in the ruby/rails community if you want any further inspiration
16:16	mmcgrana	in fact rails/merb is what pushed me away from ruby into clojure (did rails full time for a while)
16:16	powr-toc	mmcgrana: do you mean the rails/merb fracture?
16:17	powr-toc	or do you just dislike those frameworks? :-)
16:17	mmcgrana	just the style of the frameworks: crazy side effects and excessively coupled apis in particular
16:18	powr-toc	mmcgrana: yeah, I've been getting increasingly dissillusioned with OOP for many problems
16:19	danlarkin	"you know you've been programming in an FP language too long when"... side effects start to confuse you
16:20	danlarkin	that has been happening to me :-o
16:20	Chousuke	has there been a time when side effects have not confused you?
16:21	powr-toc	danlarkin: don't you mean you know you've NOT been programming in an FP language when side effects start to confuse you? ;-)
16:21	Chouser	absolutely
16:22	mmcgrana	i must be missing something obvious: say i'm in a boot script an i have an env value, like :dev. I'd like to require myapp.config and end up with myapp.config/env bound to :dev and the various myapp.config/setting{a,b,c} bound according to the env. I don't know how to communicate the env value to the myapp.config namespace so that it sets everything correctly. I guess i could put the env-determination logic in myapp.conf
16:24	hiredman	.conf? sounds like something that should be a .clj
16:25	mmcgrana	hiredman: when i say myapp.config i mean a namespace defined by /myapp/config.clj
16:26	hiredman	you want a seperate library for reading the the config that you can including in other projects?
16:27	mmcgrana	i'm not sure what you mean: myapp.config is just a namespace to hold config values
16:28	hiredman	you appear to be concerned with where to put the logic, and I am asking why
16:30	mmcgrana	I just want config values to be in a single place, that's all. I don't want to put the env-determination logic in the config namespace because I will want to use multiple mechanism to make such determinations.
16:31	mmcgrana	lisppaste8: url
16:31	lisppaste8	To use the lisppaste bot, visit http://paste.lisp.org/new/clojure and enter your paste.
16:32	lisppaste8	mmcgrana pasted "config" at http://paste.lisp.org/display/73269
16:33	lisppaste8	mmcgrana annotated #73269 with "untitled" at http://paste.lisp.org/display/73269#1
16:33	hiredman	urm
16:34	hiredman	you just need (eval (read ...))
16:35	Chouser	a.k.a. load-file
16:35	hiredman	Yes.
16:35	hiredman	(doc load-file)
16:35	clojurebot	Sequentially read and evaluate the set of forms contained in the file.; arglists ([name])
16:39	mmcgrana	i like the idea of root config values and then additional/changed values for different envs, but i dont think i want to commit to requiring different files for each env. for example in the simple case in the paste everything could just be in the myapp.config file. sorry if i'm being thick, maybe i just need to sleep on it.
16:39	powr-toc	what's the clojure predicate that returns whether or not a supplied item is present in a seq?
16:39	hiredman	(doc some)
16:39	clojurebot	Returns the first logical true value of (pred x) for any x in coll, else nil.; arglists ([pred coll])
16:40	powr-toc	mmcgrana: can you not just represent the whole config as a hashmap?
16:40	hiredman	,(#(some #{%} %2) :a [:a :b :c])
16:40	clojurebot	:a
16:40	hiredman	,(#(some #{%} %2) :d [:a :b :c])
16:40	clojurebot	nil
16:41	mmcgrana	indeed i could, its how i have it sketched now, but you still need some val that holds the env that you use to index into the first level of the map
16:42	dreish	Why that syntax and not (some #{:a} [:a :b :c]) ... (some #{:d} [:a :b :c])
16:42	hiredman	mmcgrana: I have that val
16:43	hiredman	(keyword (first command-line-args))
16:43	mmcgrana	yes i see, but outside of the myapp.config namespace only
16:44	hiredman	dreish: well he asked for a predicate
16:44	hiredman	which is a function that returns nil or (not nil)
16:44	hiredman	#(some #{%} %2) is such a function
16:49	powr-toc	(defn includes? [val seq]
16:49	powr-toc	(not (nil? (some #{val} seq))))
16:49	powr-toc
16:49	hiredman	uh
16:49	hiredman	no
16:50	hiredman	(defn includes? [val seq] (some #{val} seq))
16:50	mmcgrana	hiredman: the (keyword... code would correspond to (determine-env-from-cl) in the paste
16:50	hiredman	,(let [f (fn [val seq] (some #{val} seq))] (f :a [:a :b :c]))
16:50	clojurebot	:a
16:50	hiredman	,(let [f (fn [val seq] (some #{val} seq))] (f :d [:a :b :c]))
16:50	clojurebot	nil
16:52	hiredman	nil? is bascially the same as not
16:52	hiredman	,(= (nil? :a) (not :a))
16:52	clojurebot	true
16:52	hiredman	so you have a double negative
16:55	Chousuke	hiredman: except for (nil? false) (not false)
16:57	hiredman	well
16:57	hiredman	Fine
16:58	hiredman	actually
16:58	hiredman	.(nil? false)
16:58	hiredman	,(nil? false)
16:58	clojurebot	false
16:58	hiredman	sigh
16:59	powr-toc	,(true? :a)
16:59	clojurebot	false
17:00	hiredman	true and false are for java interop, so I don't care, so there
17:02	Chousuke	(true? (Boolean. true))
17:02	Chousuke	,(true? (Boolean. true))
17:02	clojurebot	false
17:03	Chousuke	hmm :D
17:03	powr-toc	lol
17:03	danlarkin	JVM bug
17:03	Chousuke	creating new Boolean instances is apparently bad style anyway
17:04	Chousuke	I wonder why the constructor is even public :/
17:04	danlarkin	rich mentions it in his clojure-for-java-programmers presentation IIRC
17:04	hiredman	,(true? true)
17:04	clojurebot	true
17:04	hiredman	for boxing booleans?
17:04	Chousuke	(true? (Boolean/TRUE))
17:04	Chousuke	gah
17:04	Chousuke	,(true? Boolean/TRUE)
17:04	clojurebot	true
17:05	Chousuke	hiredman: you can just replace them with Boolean/TRUE
17:05	Chousuke	or FALSE
17:06	hiredman	shrug
17:06	Chousuke	or well, I guess the constructor is easier if you don't know which one it is :/
17:07	hiredman	clojurebot: java?
17:07	clojurebot	?
17:09	vogelrn	,(= true (Boolean. true))
17:09	clojurebot	true
17:09	Chousuke	well that's interesting :/
17:10	vogelrn	ah
17:10	vogelrn	it's because true? uses identical?
17:10	vogelrn	it assumes that all trues are the same
17:10	vogelrn	object, that is
17:10	vogelrn	whereas using Boolean to create a new true won't be the same, cached one
17:14	vogelrn	I'm kind of torn because that could lead to some really nasty bugs, but like you said, you shouldn't be using the constructor anyway, should use Boolean/TRUE or Boolean.valueOf
17:14	hiredman	clojurebot: max people?
17:14	clojurebot	max people is 116
17:18	danlarkin	how can I apply to a macro?
17:19	Chousuke	I think you'll have to write a macro-apply :/
17:19	mmcgrana	i don't think that you can, e.g. (apply or '(1 2 3)) doesn't work
17:20	mmcgrana	* i dont think that you can apply to a macro that is, i dont' know about a macro-apply
17:20	hiredman	ugh
17:20	danlarkin	:'(
17:21	hiredman	thanks guys, I totally just burnt dinner
17:25	danlarkin	Hm. I'm not sure macro-apply is possible
17:25	danlarkin	one "can't take the value of a macro"
17:25	Chousuke	danlarkin: you can produce code that expands into a macro given the name of a macro though.
17:25	mmcgrana	what if macro-apply was itself a macro?
17:25	danlarkin	oh.. well I guess maybe if macro-apply were a macro
17:25	Chousuke	well it has to be
17:26	mmcgrana	right
17:26	Chousuke	I'm trying to figure it out
17:26	Chousuke	(defmacro macro-apply [mac & args] `(~mac ~@(apply concat args))) works somehow but fails for (macro-apply or 1) :/
17:27	mmcgrana	(defmacro macro-apply [mac & args] `(~mac ~@args)) ?
17:28	duck1123	How would I, in a macro, take a sequence of forms (from a macro-expand time function), and output it as just those forms.
17:28	mmcgrana	duck1123: i think a macro needs to return a single form
17:28	mmcgrana	(do ~@forms)
17:28	hiredman	^-
17:30	duck1123	ok, what I want to do, is in a compojure defservice form, break the list of my handlers into separate functions. ~@(users-handlers) would do the trick?
17:30	hiredman	uh
17:30	dreish	,(System/identityHashCode (Boolean. true))
17:30	clojurebot	13603674
17:30	danlarkin	Chousuke: you are getting close
17:30	dreish	,(System/identityHashCode (Boolean. true))
17:30	clojurebot	22150794
17:31	hiredman	duck1123: you want a macro to define more than one function?
17:33	duck1123	defservice requires all the handlers to be at the same level, I want to have several functions that each return a seq of handlers that get flattened into the same bottom level.
17:35	duck1123	I think I can make ~@(handler-fn) work.
17:35	Chousuke	danlarkin: I'm going to have to go to sleep now though :/
17:35	danlarkin	Chousuke: aw :)
17:40	danlarkin	Chousuke: (defmacro macro-apply [mac & args] `(~mac ~@(apply eval args)))
17:40	danlarkin	one step better?
17:41	danlarkin	even though it uses eval, which could be avoided I think
17:42	danlarkin	oh and not to mention... java.lang.UnsupportedOperationException: Can't eval locals
18:12	lisppaste8	mmcgrana pasted "contrib broken on head?" at http://paste.lisp.org/display/73274
18:13	mmcgrana	anyone tried recent clojure + clojure-contrib together, particularly w.r.t. clojure.contrib.repl-utils?
18:13	mmcgrana	I'm getting the pasted error.
18:19	Chouser	mmcgrana: huh. yeah, I'm seeing that. hang on.
18:29	Chouser	mmcgrana: fixed, clojure-contrib SVN 352
18:31	mmcgrana	Chouser: wow, thanks for the quick response!
18:34	Chouser	sure, thanks for pointing it out. I'm flailing around with clojure.main and hadn't noticed I wasn't loading repl-utils.
18:41	scottj	Re: snakes w/o globals. Does putting variables that you might want to customize in a let inside a main function make it harder to make customizations than using globals?
18:43	scottj	that is, with globals you could just create a let binding with your customizations and run the program inside that, but with the main function and let inside that, you now have to write your own custom main function.
19:11	Chouser	Clojure applet. Warning: it's rather demanding and may take down your browser: http://chouser.n01se.net/misc/tree.html
19:13	vogelrn	surprisingly it hasn't killed my EEE PC
19:13	Chouser	but does it work?
19:13	vogelrn	yeah
19:13	Chouser	oh! well, cool.
19:14	vogelrn	it probably draws a lot more slowly than it's supposed to, but it does take form :P
19:14	Chouser	Then I'm off. g'night.
19:21	durka	Chouser: OS X: Applet net.n01se.Tree notinited
19:25	durka	Chouser: it works in appletviewer, however
19:31	hiredman	ugh
19:31	hiredman	clojurebot: haskell?
19:31	clojurebot	Yo dawg, I heard you like Haskell, so I put a lazy thunk inside a lazy thunk so you don't have to compute while you compute.
19:32	vogelrn	:P
19:33	danlarkin	haha
19:33	danlarkin	that is a good one
19:34	hiredman	some much good functional stuff in haskell
19:34	hiredman	just looking at the code makes me break out in hives
19:35	lisppaste8	durka pasted "problem with Chouser's applet on OS X" at http://paste.lisp.org/display/73276
19:37	durka	(firefox tosses the same error at the java console)
20:20	hiredman	clojurebot: ping?
20:21	hiredman	ping?
20:22	hiredman	1d6
20:22	clojurebot	5
20:25	hiredman	clojurebot: ping?
20:25	clojurebot	PONG!
20:30	hiredman	2d8+2
20:32	hiredman	2d8+2
20:32	clojurebot	6
20:56	herdrick	question for the crowd:
20:58	herdrick	sorry - answered my own question
23:49	AWizzArd	How can I make a Byte[] and put in some initial values? Something like (make-array Byte [1 2 3 4 5])?
23:52	AWizzArd	(into-array Byte/TYPE [1 2 3 4 5]) throws an exception.