#clojure – 2009-06-19

2009-06-19

00:08	replaca	rhickey_: are you around?
02:40	Xcalibor	greetings
02:41	Xcalibor	i was wondering... is there any benefit in using :key instead of "key" in a hash-map?
02:57	hoeck	Xcalibor: it's more :lispy?
02:58	hoeck	Xcalibor: If i understand the jvm and clojure correctly, both :key and "key" should be interned, so small strings are effectively the same as keywords
02:59	hoeck	Xcalibor: but keywords implement IFn, they are functions of maps/sets
03:00	hoeck	,(:a {:a 1, :b 2})
03:00	clojurebot	1
03:00	hoeck	,(:c #{:a :b :c})
03:00	clojurebot	:c
04:02	frodef	gen-class/interface is really a hassle for smallish things. I think I'll rather proxy clojure.lang.IFn.
04:42	samuels	does clojure check the type hints at compile time?
04:44	jdz	no, it does not generate all possible programs to test whether the type hints hold for all of them
04:47	samuels	jdz: that's not necessary for checking at compile time
04:48	jdz	well, what is necessary then?
04:48	mapreduce	samuels: I just said that in #scala. Are you an elaborate proxy bot? :)
04:48	samuels	hehe
04:48	samuels	mapreduce: i just didnt' know what to say in reply, so i posted your one instead :)
04:48	mapreduce	All you need for compile time checking is some algorithm to follow. You don't need to generate all possible programs.
04:49	mapreduce	Though you might reject some programs that will never fail, like (if true 5 <type-error-here>)
04:49	jdz	implementation details! :)
04:50	mapreduce	Implementation details are often a language user's business even if they don't want them to be.
04:56	tsdh	Hi. I need to create an ImageIcon in my GUI. The clojure file is in the same directory as the images, so I've thought "image.png" would be ok, but it isn't. The ImageIcon docs say the file should be relative to the class the constructor is invoked...
04:56	tsdh	Absolute paths work, but of course that's no solution...
04:57	dibblego	relative to System.getProperty("user.dir")
04:57	samuels	mapreduce: i think you won thatone
04:57	mapreduce	samuels: I was not competing.
05:00	tsdh	dibblego: Ah, I didn't look at the JavaDoc example. Thanks
05:23	rys	hooray, Clojure book has arrived!
05:27	tsdh	Hm, still no luck with the ImageIcon. The problem is that (ClassLoader/getSystemResource ".") points the the clojure git directory and not to my file...
06:14	frodef	when java code says "foo.class.getResource(bar);" is there a clojure equivalent (to the class member or whatever that is)?
06:15	AWizzArd	getResource is a static method in your example?
06:16	frodef	AWizzArd: I don't know what it is to be honest..
06:18	AWizzArd	frodef: if getResource is a static method then you would do this in Clojure: (foo.class/getResource bar)
06:20	Lau_of_DK	Has contrib changed in some way ? I just updated and now (import '(clojure.contrib.str-utils)) returns nil, but nothing is imported....
06:20	frodef	java.lang.ClassNotFoundException: foo.class
06:20	AWizzArd	frodef: then getResource is not a static method. Maybe you can try: (.getResource (.class foo) bar) ?
06:21	AWizzArd	hmm no, I think that will also not work
06:21	rhickey	frodef: what are you trying to do?
06:22	frodef	rhickey: creating an icon for may swing app.. :)
06:22	frodef	seems I have to getResource, which is a method on java.lang.Class..
06:23	frodef	..and so I have to do "(class-of foo)" it seems to me?
06:23	rhickey	(.getResource TheClass ...)
06:24	rhickey	if you know the name of the class, else (.getResource (class an-instance) ...)
06:24	frodef	rhickey: ah, so clojure "class" is CL "class-of"? thanks!
06:26	frodef	sorry to ask silly questions, I just get quite confused trying to navigate the perimeter between clojure and java.
06:27	rhickey	np
06:39	frodef	is there something like an implicit class that the code lives in?
06:40	frodef	"the code" being my code in a foo.clj..
06:40	rhickey	frodef: every fn is an instance of a different class, but there is no class corresponding to a file, since files aren't proper entities in Clojure
06:41	rhickey	,(class rest)
06:41	clojurebot	clojure.core$rest__3736
06:41	frodef	ok
06:48	frodef	hm.. I guess this java resource stuff doesn't mesh so well with dynamic programming.
06:53	AWizzArd	frodef: you can program dynamically typed in Java too. Declare everything as type Object :-)
06:54	cheddar	Anyone know why ants.clj now seems to hang when I send-off all the ants? If I (send-off (nth ants 1) behave) that ant moves around. But when I (map #(send-off % behave) ants) the whole thing freezes. Is it the new version of clojure? Or something to do with JVM settings? It used to work.
06:54	frodef	programming interactively isn't going to be so easy I guess.
07:02	cheddar	strange, when I increase the sleep delays by 10 times, it works
07:02	cheddar	but it used to work without that change
07:02	cheddar	and CPU usage is not above 20%
07:03	cheddar	is there a way to make behaviour degrade more gracefully?
07:03	frodef	cheddar: sounds race-conditiony, no?
07:04	cheddar	yeah, but that's strange isn't it?
07:05	cheddar	isn't ants.clj written using agents and refs in order to avoid race conditions?
07:05	cheddar	i'm not sure now that it's a race condition - at least it's not locking up
07:05	cheddar	inspecting the ants in the repl, i see their coordinates are still changing
07:06	cheddar	but the display isn't updating
07:09	cheddar	hmmm, it seems maybe it works better when the JVM doesn't have the -server option
07:09	cheddar	does that make sense?
07:13	rhickey	cheddar: I can reproduce - looking now
07:20	cheddar	it seems the ants are still moving around, but the screen redraws aren't occurring
07:20	cheddar	actually it happens without -server as well
07:24	jgracin	hi! how do I dispatch on Java primitive byte array? I mean, my dispatch function is class and I'd like to add method for byte arrays.
07:29	rhickey	cheddar: fixed in master
07:40	cheddar	thanks!
07:49	hoeck	jgracin: I don't know whether clojure has array class-literals, but (defmethod foo (type (make-array Byte/TYPE 0)) [& args] ..) should work
07:55	jgracin	hoeck: thanks, I'll try that.
07:57	rhickey	jgracin: (Class/forName "[B")
07:58	rhickey	http://java.sun.com/j2se/1.5.0/docs/api/java/lang/Class.html#getName()
08:27	jgracin	rhickey: thanks.
09:06	mtd	I have to manipulate and plot not-too-small data series (mostly priceA and priceB on the Y, time on the X) stored in an oracle db and I'd like to graph them. On the verrry off chance, is there any Clojure-REPL-hotness that I could play with?
09:07	Chouser	I have no idea how it handles large data sets, but I found this easy to use: http://www.markmfredrickson.com/code/
09:08	liebke	incanter should work, http://incanter.org
09:09	mtd	Chouser, liebke: thank you very much
09:09	Chouser	as for pulling the data out of oracle, I haven't done any of that yet. There are a couple sql apis you could look at.
09:10	Chouser	clojure.contrib.sql
09:10	Chouser	http://github.com/Lau-of-DK/clojureql/tree/master
10:36	stuarthalloway	is there a separate irc node for clojure-dev conversation, or is that all happening here? I have the day free to work on clojure, but don't want to clutter this form with dev talk if that is inappropriate
10:37	rhickey	stuarthalloway: here is fine
10:37	stuarthalloway	cool!
10:37	stuarthalloway	(1) just sent email to clojure-dev proposing the "simple way" to move clojure's tests into clojure
10:38	stuarthalloway	but I can imagine reasons that the simple way might cause problems, so willing to pursue some harder way if necessary
10:40	stuarthalloway	if that is the most substantial problem we can probably reach an accomodation
10:40	Chouser	it was noted here recently that test-is has dependencies on several other contrib pieces.
10:40	rhickey	is there consensus around test-is? Is everyone in contrib using it for their tests? I'm fine with moving to clojure, but should be part of a broader plan to move some contrib bits to clojure, including pprint
10:40	rhickey	right, and sorting out deps
10:40	stuarthalloway	those are both bigger problems
10:41	rhickey	this would be a good clojure-dev topic
10:41	rhickey	stuarthalloway: bigger how?
10:42	stuarthalloway	the dependencies issue could lead to weeks of conversation going in circles
10:43	stuarthalloway	and, while important, is less important than making testing simple
10:43	stuarthalloway	I would even prefer the temporary expedient of a private subset of contrib buried in clojure for testing purposes only
10:43	stuarthalloway	to give the community time to consider dependencies at leisure
10:44	rhickey	stuarthalloway: well, someone should scope it so we're not speculating, I'm for test and pprint
10:44	Chouser	2 versions of code, one in core and one in contrib, sounds a bit scary
10:44	stuarthalloway	Chouser: didn't say it was good
10:45	Chouser	:-)
10:45	stuarthalloway	but freezing (and endorsing) a testing library is a separate step from wanting to test core
10:45	Chouser	Are there problems now that would be solved by having the clojure tests in clojure instead contrib?
10:45	Chouser	instead of
10:46	stuarthalloway	YES!
10:46	rhickey	I think the number one issue is not being able to submit a patch with both code and test
10:47	stuarthalloway	having to coordinate patches in two different repos is a big hurdle to community contribution
10:47	stuarthalloway	rhickey: by "scope it out" do you expect (1) a writeup of what it would look like, or (2) a fork so you can see the ideas develop in code?
10:48	rhickey	stuarthalloway: a list of deps would be a start
10:48	stuarthalloway	rhickey: seems like the easiest way would be to try--I will do it
10:49	rhickey	Chouser: I mentioned yesterday I saw the #1 use case for pulls versus patches ong-running contrib project, like pprint, brought into clojure
10:49	rhickey	long-running
10:50	stuarthalloway	rhickey: the "git way" is for me to just fork and go -- "fork" not having the negative connotations it does with other SCMS
10:51	stuarthalloway	so my plan is to fork Clojure, move contrib bits in, and get tests to run -- understanding that this may be throwaway work
10:51	stuarthalloway	then we can take a look, argue about names of things, etc.
10:51	stuarthalloway	that sound good?
10:51	rhickey	this will trash test-is history?
10:52	stuarthalloway	test-is history will be unavailable in the clojure repos
10:52	rhickey	I think everyone can presume pulling the bits in can be made to work
10:53	stuarthalloway	but still available in the contrib repos if needed
10:53	eevar2	you can merge git repos while preserving history
10:53	stuarthalloway	eevar2: this isn't a fully merge, though -- just grabbing piece until things work to sort dependencies
10:54	rhickey	stuarthalloway: I want it to end up one place or another, so Stuart can continue to work on it, so it is unlikely your fork will be merged unless it deal with these details
10:54	eevar2	yes. still, you'd filter out the paths you don't want beforehand, then merge/rename whatever pieces of contrib you want to add to core
10:55	eevar2	hmm.. filtering out paths may ruin some of the history in what you're merging in, and might be a bad idea
10:55	eevar2	*of what you're..
10:55	stuarthalloway	eevar2: I don't care (much) about the history in this case.
10:56	drewr`	Is there a way, possibly through meta info, to access information about a fn's caller?
10:56	stuarthalloway	Stuart Sierra is the primary (only) developer, and he knows where to find the history if he needs it
10:56	eevar2	okies. just letting you know it can be done
10:57	drewr	I guess what I'm looking for is something like elisp's defadvice.
10:57	rhickey	looks like just template/walk/stacktrace as deps
10:58	AWizzArd	And anyway: how often is the history older than a few weeks really needed? How often does rhickey look at what he did at revision 631?
10:58	stuarthalloway	rhickey: the cool thing about git is the ease of showing you my ideas, vs. just discussing them. I think the fork will spur a conversation. If it never gets merged, then that is fine
10:58	drewr	I want to say (defn #^{:deprecated true} foo [& args] ...) and throw an exception if foo is called anywhere.
10:59	rhickey	stuarthalloway: of course there's no harm, but I'm not seeing too much benefit either, a 2 minute perusal of the source shows 3 deps
11:00	Chouser	drewr: You can use a java exception (does not need to be thrown) to examine the stack
11:00	stuarthalloway	drewr: why not a macro that implements (deprecated foo)?
11:00	Chouser	drewr: but I don't think you'll be able to get from the java stack to the fn or var metadata.
11:00	Chouser	at least not easily
11:01	Chouser	drewr: why not just replace the body of foo with (throw (Exception. ...))?
11:02	stuarthalloway	rhickey: we'll see. I am a big believer in the difference between "could" and "did"
11:03	drewr	Chouser: That's what I originally did, then I got into thinking about what was possible in capturing fn invocation.
11:04	Chouser	drewr: I guess I don't see the benefit of doing it via metadata. Is there some use case where just calling (throw ...) is less desirable?
11:05	drewr	I went from that to just putting (deprecated!) as the first line in the fn where deprecated! does whatever I need at the time.
11:05	Chouser	you could also use (binding [foo (fn [& x] (throw ...))] ...) to only check certain call chains.
11:06	drewr	Ah, that's a good idea.
11:06	Chouser	or alter-var-root to change it after-the-fact
11:06	drewr	I'm not quite sure which fns will be called though.
11:21	Chouser	bah.
11:22	Chouser	rhickey: would it be terribly difficult to change my clojure-dev member address from n01se.net to gmail.com?
11:22	Chouser	my mail client (gmail!) is insufficiently powerful to handle the permission requirements well on my end.
11:25	Chouser	or add my gmail.com address as well, I suppose.
11:27	rhickey	Chouser: sure, just apply to the group using your other identity, I don't manage emails, just users from google's perspective
11:28	Chouser	oh, ok.
11:28	j-dot	Chouser: I was able to change my email address on the "Edit my membership" page ... try here:
11:28	j-dot	http://groups.google.com/group/clojure-dev/subscribe
11:29	Chouser	j-dot: than what? I don't see any way to change my email address there.
11:37	j-dot	Chouser: Hmmm ... I have a "Which address do you want to use for this group" option
11:38	j-dot	perhaps that's because I've registered multiple email accounts in my Gmail settings
11:38	Chouser	j-dot: ah, ok. thanks.
11:38	j-dot	np
11:45	eevar	stuarthalloway: +1 for moving clojure-core tests into the clojure repo, btw
12:57	dhaza	can anyone point me towards information on jars and packaging in java?
12:57	dhaza	im coming at clojure from a lisp background and the entire jar process is pretty opaque to me
12:58	j-dot	I don't have any URLs, but a jar is just a zip file with your code/classes
12:58	slashus2	Jars are just zip archives as far as I understand it.
12:58	Chouser	jars can also have a specially-named file with a bit of extra info in it
12:59	DougC	java packages are similar to namespaces in Clojure.
13:00	DougC	The class java.lang.String is in the package java.lang and the source is in a file called String.java
13:00	DougC	In the file system this would be in a directory structure as java/lang/String.java
13:01	dhaza	you need to jar the compiled files to have the code loaded?
13:01	dhaza	or does it read the source upon load?
13:01	DougC	When you compile a java source file the bytecode ends up in a .class file
13:01	clojurebot	the unit of compilation in clojure is the namespace. namespaces are compiled (not files). to compile a namspace the namespace needs to be on the classpath and so does ./classes/ (and the directory needs to exist) because clojure writes the class files to that directory. http://clojure.org/compilation
13:01	Chouser	for Clojure, compiling is independent of bundling into a .jar
13:02	Chouser	a .jar can have .clj or .class files and still work fine.
13:02	DougC	So java.lang.String would compile into a file in the filesystem as java/lang/String.class
13:03	dhaza	i get it
13:03	DougC	The CLASSPATH tells the jvm where to look for classes and it can contain path elements which are either directories or .jar files.
13:04	DougC	If a path element is a directory then it defines the directory in which those relative paths live. So one element of the classpath might be ~/classes and in there you would find java/lang/String.class
13:05	DougC	For jar files, as other have said, they're just zip files where the class directory structures are all zipped up (there are other conventions for things but they're not important for this discusssion)
13:09	dhaza	okay, so if my package name is 'org.foo.blah.MyClass', it would be in the jar under org/foo/blah/MyClass.class
13:09	Chouser	exactly
13:10	Chouser	and your .clj file for that would usually be in some element of the classpath at org/foo/blah/MyClass.clj
13:11	Chouser	though it's fine to have it in a different element of the classpath, and therefore not in the same directory as MyClass.class
13:12	Chouser	For any given project I like to have a 'src' and a 'classes' directory, with both in my classpath. That way I can blow away the contents of 'classes' anytime I want to be sure I'm getting a clean build.
13:12	Chouser	...and I don't have to look at .class files littering up up 'src' dirs.
13:13	Chouser	then if I make a .jar for that, I can pick either just the class files, just the clj files, or both.
13:13	mattrepl	stuarthalloway: Cluje, eh? =)
13:14	stuarthalloway	I think I liked "Frak" best
13:15	Chouser	I like verily!
13:15	Chouser	(verily (= (foo) 1)))
13:18	hiredman	is it name game again?
13:18	hiredman	time
13:20	stuarthalloway	verily, I say unto thee all: I am stepping away for 30 minutes to do a video interview with @coreyhaines
13:21	stuarthalloway	if thou wouldst agree on a name for thy core testing framework while I am going, I will implement it, and submit the fork to our glorious emperor
13:28	j-dot	haha, I like verily ... but if I was that confident, why would I bother testing?
14:17	r2q2	Hello
14:18	Chouser	Hi
14:21	kefka	,(inc (Long/MAX_VALUE))
14:21	clojurebot	java.lang.ArithmeticException: integer overflow
14:21	stuarthalloway	rhickey: I will rename clojure.contrib.test-is to clojure.test
14:22	stuarthalloway	what about the dependencies: stacktrace, template, and walk?
14:22	kefka	Is this a bug, or should we explicitly not use inc / dec if we want integer promoption?
14:22	stuarthalloway	each in their own namespace under clojure?
14:22	kefka	,(inc 12345678901234567890123)
14:22	clojurebot	12345678901234567890124
14:22	stuarthalloway	or maybe put walk into core?
14:22	stuarthalloway	(walk is way cool and I end up using it on every project)
14:23	Chouser	Long/MAX_VALUE is a primitive
14:23	Chouser	,(inc (Long. Long/MAX_VALUE))
14:23	clojurebot	9223372036854775808
14:25	r2q2	I was wondering if there is a plan for clojure to use the new java 7 features that allow for dynamic languages?
14:26	r2q2	http://openjdk.java.net/projects/mlvm/
14:27	hiredman	r2q2: mlvm is not java 7
14:28	hiredman	mlvm is an experimental vm
14:28	hiredman	clojure targets java 5, because that is everywhere
14:28	brett_h	Does anyone have a link to an explanation on what it means for something to be "composable", I'm specifically referring to the "locks don't compose, but STMs do" here: http://www.slideshare.net/jboner/state-youre-doing-it-wrong-javaone-2009
14:29	r2q2	hiredman: Yea thats what I thought.
14:29	r2q2	hiredman: I meant JSR 292 but yea I was guessing its too far out there to support.
14:30	rhickey	,(inc (num (Long/MAX_VALUE)))
14:30	clojurebot	9223372036854775808
14:32	stuhood	brett_h: Mark Volkmann's Clojure article contains "The Bank Example": http://java.ociweb.com/mark/clojure/article.html#Refs
14:32	hiredman	brett_h: if you have a locking strategy for A and a locking strategy for B, just mashing those two locking strategies together doesn't work
14:32	dhaza	brett_h, composable is similar to nestable, you can nest STM transactions, you cant nest locks
14:33	dhaza	well, i said that wrong
14:33	dhaza	you can nest locks, but not arbitrarily
14:33	brett_h	hiredman: you mean it doesn't always work, where transactions do? like you don't have to think about it?
14:33	brett_h	dhaza: yeah, arbitrarily is what I was thinking
14:33	brett_h	ok that helps
14:33	brett_h	stuhood: thanks I'll check it out
14:34	stuhood	brett_h: no problem... not really an explanation, but a great example
14:34	brett_h	oh yeah, I've read that
14:35	brett_h	I was kind of hoping for a "see, locks fail here because they aren't composable" but I think I get what dhaza means
14:35	brett_h	you can't arbitrarily combine things that lock and expect them to work
14:36	stuhood	brett_h: right, so in that example the withdrawals and deposits need to be atomic, so they would have to be protected with locks traditionally. but if you want to transfer money, you need another lock around the first two.. which means there is deadlock potential unless you start thinking about lock ordering
14:36	brett_h	right, thanks
14:38	rhickey	search for compose in: http://en.wikipedia.org/wiki/Software_transactional_memory
14:38	brett_h	rhickey: perfect, thanks
14:42	stuarthalloway	rhickey: I have renamed clojure.contrib.test-is to clojure.test on the fork: http://github.com/stuarthalloway/clojure/tree/c718835af09f238f8f9d57c330951e1a8e906475
14:43	stuarthalloway	remaining todo:
14:43	stuarthalloway	(1) rename the dependencies from clojure.contrib to ????
14:43	stuarthalloway	(2) fix contrib to match
14:45	rhickey	stuarthalloway: you can use one of the assembla wiki pages for this kind of punchlist
14:46	stuarthalloway	rhickey: ok doke
14:46	stuarthalloway	if you are ok with the renaming, I could just be done in another 30 mins or so
14:47	stuarthalloway	packages in question are stacktrace, template, and walk
14:47	rhickey	putting the issues in a single place will help me make a decision and others to chime in on any missing pieces or things you haven't thought of
14:48	stuarthalloway	should I create a ticket too, and link out to the wiki page?
14:48	rhickey	sure
14:52	rhickey	I'm not sure how I feel about template, stacktrace and walk seem fairly straightforward
14:52	rhickey	I'm talking about the contents, not the names
14:52	stuarthalloway	agreed!
14:52	stuarthalloway	I was about to move template into test
14:52	stuarthalloway	clojure.test.template
14:53	stuarthalloway	as a way of marking it to be internal detail
14:53	stuarthalloway	could also make the bits private to clojure.test
14:53	stuarthalloway	which would be a stronger statement of same
14:53	stuarthalloway	Ticket is at http://www.assembla.com/spaces/clojure/tickets/131-Move-Clojure-tests-from-contrib-into-Clojure
14:54	stuarthalloway	wiki page at http://www.assembla.com/wiki/show/clojure/Getting_Tests_Into_Clojure
14:54	rhickey	iirc Stuart said he needed template for 'are'?
14:54	stuarthalloway	yep
14:55	rhickey	are there other consumers of template?
14:56	stuarthalloway	checking...
14:57	stuarthalloway	only contrib.condt
15:04	stuarthalloway	ok, everything is renamed into clojure except template
15:04	stuarthalloway	gonna fork contrib and pull out the items added to clojure, and make sure that works
15:23	r2q2	,{:a 1,,,,,,,,,,,,, :b 2,,,,,,,,,,,}
15:23	clojurebot	{:a 1, :b 2}
15:24	Lau_of_DK	Compojurefreaks - Anybody worked out, how one can benefit from the (with-session) macro ?
15:40	rzoom	what does max-key do if there are multiple keys with the max value?
15:41	hiredman	(doc max-key)
15:41	clojurebot	"([k x] [k x y] [k x y & more]); Returns the x for which (k x), a number, is greatest."
15:41	rzoom	seems to take the last key
15:41	shoover	there is a lot of forking going on here
15:42	gnuvince	here where?
15:42	shoover	this room, clojure in general
15:43	Chouser	so far nobody has forked with the intention to do anything other than either merge back into clojure, or throw it away. So no need to worry.
15:44	shoover	understood, it's interesting to see all the activity
15:44	hiredman	did we ever get that flag to stop read from evaling stuff?
15:54	Chouser	hiredman: looks like it: https://www.assembla.com/spaces/clojure/tickets/38
15:54	Chouser	,read-eval
15:54	clojurebot	true
15:54	Chouser	,(doc read-eval)
15:54	clojurebot	"; When set to logical false, the EvalReader (#=(...)) is disabled in the read/load in the thread-local binding. Example: (binding [read-eval false] (read-string \"#=(eval (def x 3))\")) Defaults to true"
16:12	rhickey	saw some tweets re: test-is inclusion implying the end of testing frameworks, thoughts?
16:13	Chouser	I have no alegience, but the current clojure regression tests use test-is. That's a bit inescapable.
16:13	stuarthalloway	ruby having test::unit didn't stop rspec, shoulda, and micronaut...
16:14	Chouser	it can't be worth re-writing the tests to use something else.
16:14	dhaza	find/replace
16:15	stuarthalloway	also, the testing library used to test clojure needs to be (relatively) minimal compared to frameworks people might create to test clojure application code
16:15	Chouser	I suppose the core version of test-is could be made private, but that seems a bit dumb since it will be used and maintained.
16:15	rhickey	Chouser: no, but there may be other ways to organize things without seeming to bless one
16:15	stuarthalloway	I don't think it is a problem
16:16	Chouser	namespace doc string that says, "there are many clojure testing frameworks -- one of the others may fit your needs better than this one"?
16:16	stuarthalloway	+!
16:16	hiredman	~ticket 38
16:16	clojurebot	Gabh mo leithscéal?
16:16	hiredman	clojurebot: why?
16:16	clojurebot	why not?
16:16	Chouser	~ticket #38
16:16	hiredman	~ticket #38
16:16	clojurebot	{:url http://tinyurl.com/mngulq, :summary "GC Issue 34: A Lisp reader without access to EvalReader()", :status :fixed, :priority :low, :created-on "2009-06-17T19:11:53+00:00"}
16:16	clojurebot	{:url http://tinyurl.com/mngulq, :summary "GC Issue 34: A Lisp reader without access to EvalReader()", :status :fixed, :priority :low, :created-on "2009-06-17T19:11:53+00:00"}
16:16	Chouser	whee!
16:16	dhaza	clojurebot, Trej buil listileace misceal
16:21	suresh	hello all.. am new to clojure.. can anyone help me understand these lines..
16:21	suresh	(defmacro #^{:private true} -?>
16:21	suresh	([form] form)
16:21	suresh	([form next-form & forms]
16:21	suresh	`(when-let [x# ~form] (-?> (-> x# ~next-form) ~@forms))))
16:21	r2q2	suresh: What are the lines supposed to do?
16:22	suresh	those lines are from Ring.. i'm curious what the macro does
16:24	slashus2	It only chains while the forms are true?
16:25	suresh	slashus2: can you help me understand by deconstructing those lines?
16:26	hiredman	suresh: it is a short circuiting version of →
16:26	r2q2	suresh: What is the name of the macro? It seems to be missing that?
16:26	hiredman	"->"
16:26	hiredman	r2q2: -?>
16:27	r2q2	Oh
16:27	hiredman	it's in contrib these days
16:27	clojurebot	contrib is http://code.google.com/p/clojure-contrib/
16:27	hiredman	(doc -?>)
16:27	clojurebot	"clojure.contrib.core/-?>;[[x form] [x form & forms]]; Same as clojure.core/-> but returns nil as soon as the threaded value is nil itself (thus short-circuiting any pending computation). Examples : (-?> \"foo\" .toUpperCase (.substring 1)) returns \"OO\" (-?> nil .toUpperCase (.substring 1)) returns nil "
16:28	suresh	ah.. didnt know it was part of contrib
16:28	dnolen	i think it would look better if it was ->? or ?-> to preserve the arrow. -?> looks to much like line noise.
16:28	hiredman	it is now, it is possible that ring started using before it was in contrib
16:29	r2q2	(doc ->)
16:29	clojurebot	"([x form] [x form & more]); Threads the expr through the forms. Inserts x as the second item in the first form, making a list of it if it is not a list already. If there are more forms, inserts the first form as the second item in second form, etc."
16:35	rhickey	technomancy: you had some reservations about assembla the other day?
16:36	technomancy	rhickey: no, I don't think so. unless it was about the volume of email, which is pretty solvable.
16:36	rhickey	ah, ok
16:37	technomancy	rhickey: any further suggestions for CS papers? I got my Kindle yesterday and just finished Out of the Tarpit.
16:37	technomancy	great stuff.
16:39	technomancy	hah
16:39	technomancy	ls ~/documents/papers \| lisppaste
16:39	rhickey	unfortunately they are almost universally badly named
16:40	technomancy	heh, right.
16:40	technomancy	well I've got enough to keep busy for the time being.
16:41	technomancy	spent twenty minutes yesterday just clicking madly on download links at feedbooks.com for classics.
16:45	rhickey	43 on STM, 257 on functional programming, 20 on persistent data structures, 54 on purely functional something, 111 on the cochlea ...
16:46	rhickey	49 on RDF, 256 on concurrency...
16:47	rhickey	69 on datalog, 169 on locks, 379 on lisp
16:48	rhickey	etc
16:48	stuarthalloway	how many on FP in dynamic languages?
16:48	rhickey	100 on erlang
16:49	Chouser	what percentage of these have you read?
16:49	stuarthalloway	technomancy: not cs, but read Daniel Suarez's Daemon if you haven't already
16:49	rhickey	all
16:49	rhickey	I read them as I get them
16:49	rhickey	these go back to 2003 on this machine
16:49	technomancy	stuarthalloway: will check it out; thanks
16:50	rhickey	a lot of technical papers are only 12 pages
16:50	technomancy	oh, it's only available as a DRM'd ebook. =(
16:51	rhickey	but I've also read, e.g. Joe Armstrong's PhD thesis, quite good
16:52	rhickey	obviously some of those numbers overlap, just doing a raw search
16:54	rhickey	unfortunately there aren't many like Out of the Tarpit, research papers tend to be very narrowly focused, and in order to read one you almost always have to read one or more of the papers it references etc
16:55	stuarthalloway	technomancy: loving emacs-starter-kit, thank you!
16:55	technomancy	stuarthalloway: awesome.
16:55	rhickey	I should put together a list of at least the ones that mattered most in the end
16:55	dysinger_	stuarthalloway - for those of us on clojure 1.0 and depending on clojure-contrib for test-is and other things - If you refactor it out - then we have to freeze clojure-contrib for a while ......
16:55	stuarthalloway	still undecided about paredit-mode -- do most Clojurians use it?
16:55	dysinger_	no more march straight onward
16:55	hiredman	dysinger_: or cut a release
16:55	dysinger_	y
16:56	technomancy	stuarthalloway: it takes some getting used to, but I will never voluntarily go back.
16:56	dysinger_	technomancy showed me paredit-mode - I am hooked
16:56	dysinger_	wouldn't code without it now
16:56	rhickey	yes please, a contrib release corresponding to 1.0 would be hugely useful
16:57	technomancy	stuarthalloway: the big hurdle is there's no way to just explore it; a lot of its features are hidden until you just look through the list of keybindings it provides
16:57	technomancy	maybe I should do a short focused screencast on paredit... hrm.
16:57	technomancy	it's one of those things that's far easier to learn by example
16:57	stuarthalloway	technomancy: that would be a great screencast
16:57	stuarthalloway	esp. since people keep whining about IDE support
16:58	dysinger_	they'll never be happy :)
16:58	stuarthalloway	would be nice to see the benefits of coding in sexprs
16:58	dysinger_	y it's really cool
16:58	technomancy	yeah, most of the other features you see ported to IDEs, but I've never seen paredit duplicated.
16:58	dysinger_	you don't have to track down your paren ending out of ))))))))) anymore - just cut/paste/move etc
16:58	stuarthalloway	it would be lot harder to write a structural edit mode for a language with ... more structures
16:59	rhickey	does it understand [] and {}?
16:59	technomancy	rhickey: recent versions do
17:00	stuarthalloway	and it uses the close versions ] } ) to mean different things
17:00	stuarthalloway	...regardless of what form you are closing
17:00	r2q2	rhickey: yes.
17:00	dysinger_	I love that they call it barfage and slurpage
17:01	r2q2	rhickey: Documentation is sort of lacking the cheatsheet or the comments in paredit.el are your best bet in understanding what it does.
17:01	dysinger_	It's easy when someone shows you it in action. Technomancy showed it to me and I was like "ah" and mimicked like a monkey in 5 minutes
17:02	technomancy	yeah, sounds like a screencast is in order. =)
17:02	r2q2	rhickey: Actually it doesn't understand {} but it does understand []
17:02	dysinger_	the editing and the cut and paste without worrying about () {} [] is awesome
17:03	dysinger_	ah - haven't tried with {} yet
17:03	r2q2	Hrm oops I'm running an older version
17:04	dhaza	why cant the editor call into clojure to get help with the parse tree and how to indent?
17:04	technomancy	dhaza: totally possible. emacs does that with CL, but nobody's hooked up the wiring to do it w/ clojure yet.
17:04	clojurebot	emacs is best configured for Clojure with instructions at http://technomancy.us/126
17:05	technomancy	dhaza: it's preferable to have static indentation rules imho though since if you auto-indent in Emacs (where it understands those rules) and someone else on the project uses a different editor, you won't be able to achieve consistency.
17:05	dhaza	cool. because reader macros are groovy o_o
17:06	wlr	paredit dwim re: () {} [] "" for me at least
17:06	technomancy	oops. I bundled an older version of paredit in the starter kit so it doesn't auto-insert balancing {} chars.
17:07	technomancy	it understands them when enforcing sexp validity, but it just doesn't insert closing ones for you. will update.
17:10	wlr	wrapping n exprs with M-n ( or M-n [ or M-{ or M-" as appropriate is really handy
17:12	technomancy	r2q2: you need paredit 21 or higher
17:13	r2q2	I self installed v21
17:13	r2q2	I even evaluated and loaded paredit 21
17:14	technomancy	oh... clojure-mode is expecting paredit to be loaded before it loads, which is not always the case.
17:15	r2q2	Oh yea thats probably the problem. I was manually enabling paredit as a minor mode.
17:16	technomancy	looks like it needs to be enabled in a hook
17:16	r2q2	(add-hook 'clojure-mode-hook (lambda () (paredit-mode +1)))
17:16	technomancy	which is what it says in the docs at the top, but it's easy to miss.
17:17	r2q2	If you have questions about paredit riastradh is on #scheme.
17:18	dysinger_	I had to add a clojure-mode hook for paredit too
17:18	dysinger_	and I installed paredit with elpa
17:18	r2q2	As long as that is evaluated before clojure-mode that mode hook should work.
17:20	dysinger_	I have autoload for clojure based on file extension and then the paredit clojure-mode hook after
17:20	r2q2	Well
17:20	dysinger_	which I think accomplishes the same
17:20	dysinger_	anyway it works
17:21	r2q2	Yea thats fine I was probably doing something funny. Nevermind
17:26	r2q2	Anyone use redshank ?
17:27	technomancy	r2q2: I think redshank is pretty cl-specific
17:27	lisppaste8	dysinger pasted "my el init" at http://paste.lisp.org/display/82154
17:28	dysinger	kool lisppaste is nice
17:28	dysinger	r2q2 there's my el init (based on emacs-starter-kit) (as tim.el)
17:32	r2q2	technomancy: Yea I know some of it is. The mouse copy function was usable in scheme.
17:32	r2q2	technomancy: It was the only feature that was actually useful.
17:32	technomancy	what's it for?
17:33	r2q2	Well
17:34	r2q2	It allows you to write a lisp expression and insert it at an arbitrary point
17:36	r2q2	It copies an sexpr from the point that you click at to the point of your cursor
17:37	r2q2	I'm doing a horrible version of actually telling you what it does. http://lispm.dyndns.org/news?ID=NEWS-2007-11-18-1 is much better
18:08	rhickey	Clojure bookshelf: http://www.amazon.com/Clojure-Bookshelf/lm/R3LG3ZBZS4GCTH/ref=cm_lm_byauthor_title_full
18:09	technomancy	rhickey: nice; thanks!
18:10	liebke	rhickey: great list!
18:10	arohner_	rhickey: that is really cool
18:10	arohner_	I only have 1/7th of that list...
18:12	rhickey	I'm not sure I got the latest edition of every book in the links
18:13	technomancy	Dylan, huh? wouldn't have guessed that.
18:14	mrsolo	thats a lot of books
18:15	technomancy	a good portion of them you can find as PDFs
18:15	mrsolo	lisp in small pieces is one expensive book
18:16	technomancy	mrsolo: there was a mistake on amazon.ca that had it being sold for $4 for a while
18:16	mrsolo	missed that heh
18:16	technomancy	they didn't deliver, but they sent a coupon to everyone who ordered it. =)
18:16	technomancy	n
18:16	mrsolo	purely functional data structures was way over my head firt time i read it
18:23	ataggart	rhickey: do you have a personal site, where you stick links like the one to the bookshelf?
18:23	rhickey	ataggart: no
18:29	r2q2	I'm probably being nitpicky but the semantics of (identical? "foo" "foo") has changed since the video Clojure Data Structures has been made. At the time of the video was made (identical? "foo" "foo") => false , while in clojure 1.0 (identical? "foo" "foo") => true . The time that this occurs is at 32:00.
18:31	dreish	My guess would be that the difference is strings passed to eval are all getting interned now.
18:31	dreish	More of an eval change than an identical? change.
18:32	rhickey	r2q2: it wasn't guaranteed then (string literal interning), it is now
18:33	r2q2	Oh okay.
18:34	rhickey	,(identical? (str "str" "ing") "string")
18:34	clojurebot	false
18:46	mrsolo	why is clojure designed in the way so that (+ nil) doesn't raise NPE?
18:47	mrsolo	,(:a nil)
18:47	clojurebot	nil
18:47	mrsolo	,(nil :a)
18:47	clojurebot	java.lang.IllegalArgumentException: Can't call nil
18:47	mrsolo	,(+ nil 3)
18:47	clojurebot	java.lang.NullPointerException
19:13	banister	hey what's the diff bw multimethods and functional overloading
19:14	ataggart	the former uses an arbitrary dispatch function, and the latter is based on arity (if I understand your meaning of "functional overloading")
19:15	banister	overload
19:16	banister	thanks
19:16	banister	hey can type hints be used to check for type correctness
19:16	Chouser	no
19:17	banister	hmm ok
19:17	hiredman	~deft
19:17	clojurebot	deft is http://gist.github.com/128259
19:17	banister	why not?
19:17	hiredman	because no one has written something to do it yet
19:17	Chouser	well, it will catch some circumstances, but not all.
19:18	banister	do you think they will ? cos it seems the only major criticism I hear about clojure to date
19:18	Chouser	I wrote up an example once showing a case where it doesn't catch the type difference. I wonder where i put it.
19:19	Chouser	banister: are you talking about runtime or compile time type checking?
19:19	banister	compile time
19:20	banister	can't you just take the system from scala?
19:21	technomancy	open source doesn't work like that. =)
19:21	ataggart	sounds like an IDE feature waiting to be written
19:22	banister	but I mean both scala and clojure blatantly take from erlang why doesn't clojure take this one thing from scala?
19:22	r2q2	You can't just blindly import a system from an opensource language and plop it to another. You would have to rewrite it.
19:22	hiredman	banister: deft does very simple compile time type checking
19:23	hiredman	(and completely ignores types at runtime)
19:23	drewr	What has clojure "blatantly" taken from erlang?
19:23	Chouser	technomancy: I think it's more of languages not working that way than open source.
19:24	technomancy	Chouser: yeah, it was a joke.
19:24	technomancy	sort of.
19:24	Chouser	banister: I expect clojure will evenutally have a system for making assertions about the code at compile time.
19:24	ataggart	can someone remind me how odl clojure is again?
19:24	ataggart	*old
19:25	Chouser	banister: I imagine it will not be based solely around Java classes
19:25	technomancy	banister: that's the joy of young languages; you get to be involved in writing that kind of stuff. =)
19:25	hiredman	younger than the mountains, older than the trees
19:25	mrsolo	drewr: function dispatch based on arity very handy
19:25	Chouser	mrsolo: also very fast on the JVM.
19:25	mrsolo	sorta..although erlang does a lot more
19:25	hiredman	mrsolo: hardly exclusive to erlang
19:25	mrsolo	hardly
19:26	mrsolo	inspired i guess
19:26	hiredman	even java can do that
19:26	ataggart	"even" lol
19:26	technomancy	"even my grandma can do that"
19:26	technomancy	"... and she's been dead for seven years"
19:26	mrsolo	only rick can answer what language inspired that part <shrug>
19:27	hiredman	I think C++ can do that too
19:27	Chouser	ataggart: I think Nov. '98 was 1 year since it was first "released"
19:27	hiredman	but my c++ is shakey
19:27	mrsolo	erlang is listed in the book list.. clojure must have taken something out of that langauge
19:28	mrsolo	may be distrubted programming in the future release? :-)
19:28	Chouser	ataggart: I believe rhickey was working on it for 1.5 to 2 years or before that initial "release"
19:28	ataggart	chouser: 1998?!
19:28	Chouser	heh
19:28	Chouser	2008
19:28	Chouser	sorry
19:28	ataggart	ahh ok
19:36	mrsolo	,(/ 1)
19:36	clojurebot	1
19:36	mrsolo	,(/ nil)
19:36	clojurebot	java.lang.NullPointerException
19:37	mrsolo	,(* nil)
19:37	clojurebot	nil
19:41	ataggart	the one-arg arithmetic ops seem to have alternate behavior
19:41	r2q2	Yea
19:41	r2q2	I was going to say that
19:41	r2q2	,(*)
19:41	clojurebot	1
19:41	ataggart	(-)
19:41	r2q2	,(+)
19:41	clojurebot	0
19:41	r2q2	,(-)
19:41	clojurebot	java.lang.IllegalArgumentException: Wrong number of args passed to: core$-
19:41	r2q2	,(/)
19:41	clojurebot	java.lang.IllegalArgumentException: Wrong number of args passed to: core$-SLASH-
19:42	r2q2	Hrm + and * like being called without anything.
19:42	ataggart	as (+ 5 1) =>6 and (- 5 1) =>4, I would expect (+ 5) and (- 5) to both => 5
19:42	ataggart	,(- 5)
19:43	clojurebot	-5
19:43	r2q2	,(+ 1)
19:43	ataggart	so it's more like operator overloading
19:43	clojurebot	1
19:43	banister	do you guys think of scala as the enemy
19:43	ataggart	it's a language
19:43	r2q2	I have no opinion of scala. I know its a language I know it has static typing.
19:44	hiredman	~scala
19:44	clojurebot	Unfortunately the standard idiom of consuming an infinite/unbounded resource as a stream can be problematic unless you're really careful -- seen in #scala
19:44	drewr	I find Scala less interesting than Clojure, but that's mostly out of ignorance.
19:44	hiredman	#scala is a scary place
19:44	banister	competing with scala clojure for the functional concurrency crowd on jvm?
19:44	Chouser	banister: not at all. It's got pretty different goals around dynamism, type systems, metaprogramming, etc.
19:45	technomancy	banister: they target very different audiences. to appreciate scala you have to think that on some level, Java (the language) is almost a good idea.
19:45	drewr	I know that I have questions about a functional language whose programming tome has mutable state in its first code example. :-)
19:45	hiredman	banister: I cannot concieve of competition between clojure and scala, in my mind there is no competition
19:45	mrsolo	use both of them <shrug>
19:45	banister	I own the programming scala book but it's size scares me
19:45	hiredman	I hang in #scala too, and my impression is that the scala language is a train wreck
19:46	mrsolo	hiredman: is it? hmm
19:46	mrsolo	i went through scala book and tutorial. thats about it
19:46	hiredman	mrsolo: that is just my impression from the irc channel
19:47	mrsolo	i didn't find any major flaw other than the toolchain was a bit buggy
19:47	ataggart	count me among the cadre that has no opinion of scala
19:47	hiredman	clojure is just so much more elegant and simpler
19:47	r2q2	Is lazy-seqs prefered now to lazy-cons?
19:47	Chouser	I did a couple dozen projecteuler problems in scala before learning clojure.
19:47	Chouser	r2q2: lazy-cons is gone
19:47	ataggart	r2q2: it replaces it
19:47	r2q2	Yea thats what I thought.
19:48	technomancy	I read through a lift tutorial, but I got turned off by the fact that there was more XML than scala code in the first chapter.
19:48	hiredman	~literal [0] scala
19:48	clojurebot	also<reply>"we are completely screwed on ==." -- seen in #scala
19:48	hiredman	bah
19:48	r2q2	Because the website says that it is the prerelease in SVN trunk
19:48	r2q2	The video says to use lazy-cons.
19:48	ataggart	old video is old
19:48	hiredman	svn is old
19:48	technomancy	too bad you can't patch videos like you can code. =\
19:48	hiredman	~literal [1] scala
19:48	clojurebot	<reply>Unfortunately the standard idiom of consuming an infinite/unbounded resource as a stream can be problematic unless you're really careful -- seen in #scala
19:49	Chouser	hiredman: is that one not true in Clojure as well?
19:49	hiredman	~literal [2] scala
19:49	clojurebot	2 is out of range
19:49	hiredman	Chouser: well clojue nulls locals for you
19:49	hiredman	clojure
19:49	hiredman	scala does not
19:49	Chouser	hiredman: oh, they're not just talking about trying to consume the whole thing?
19:50	Chouser	oh. my.
19:50	r2q2	How should I learn clojure other than using the website or the videos?
19:50	hiredman	r2q2: the videos are fine, just old
19:50	ataggart	write some
19:50	hiredman	euler it up some
19:50	technomancy	r2q2: read some. then write some more.
19:51	technomancy	~mire
19:51	clojurebot	mire is http://github.com/technomancy/mire/tree/master
19:51	technomancy	^an example project that might be helpful
19:52	hiredman	nah, you should make clojurebot less of sprawling disaster
19:52	hiredman	:)
19:52	ataggart	d'oh
19:53	ataggart	,(< 5)
19:53	clojurebot	true
19:53	ataggart	,(> 5)
19:53	clojurebot	true
19:53	ataggart	it's a miracle
19:53	hiredman	,(doc >)
19:53	clojurebot	"([x] [x y] [x y & more]); Returns non-nil if nums are in monotonically decreasing order, otherwise false."
19:53	slashus2	5 is greater and less than
19:54	hiredman	"5" is in monotonically decreasing order
19:54	hiredman	and increasing too
19:54	ataggart	,(<)
19:54	clojurebot	java.lang.IllegalArgumentException: Wrong number of args passed to: core$-LT-
19:54	Chouser	,(< 2 5 9)
19:54	clojurebot	true
19:54	ataggart	semms like zero- and one-args should behave the same
19:54	mrsolo	,(< nil)
19:54	clojurebot	true
19:55	hiredman	ok
19:55	hiredman	thats cute
19:55	hiredman	~def <
19:55	hiredman	must be a fast path
19:55	ataggart	([x] true)
19:55	slashus2	fast path?
19:56	hiredman	slashus2: fastest path is the shortest
19:56	hiredman	one arg to < is always true
19:56	hiredman	so one arg gets the shorest bit of code possible to always return true
19:56	ataggart	I can't think of a case where you'd care about having one-arg work, but not care about zero args working
19:57	technomancy	I can't think of a case where you'd care about having one-arg work, period
19:57	technomancy	seems like an implementation detail to allow for a more concise definition
19:57	hiredman	technomancy: apply?
19:57	hiredman	,(apply > '(5))
19:57	clojurebot	true
19:58	ataggart	,(apply > '())
19:58	technomancy	hiredman: would need a little more context than that.
19:58	clojurebot	java.lang.IllegalArgumentException: Wrong number of args passed to: core$-GT-
19:58	mrsolo	i like the behavor around nil a bit consistent.. i mean this is very nice suprise
19:58	mrsolo	,(-> {:1 {:a 3}} :2 :b)
19:58	clojurebot	nil
19:58	hiredman	technomancy: if you want to check if a list of numbers is ordered
19:59	technomancy	I guess so... but I don't think that's the intention behind the implementation.
20:00	technomancy	hmm... actually that doesn't seem to be the case; never mind.
20:00	technomancy	the >1 arities don't call the 1-arity body
20:01	ataggart	ah good catch
20:03	hiredman	that would be very silly
20:04	technomancy	yeah, doesn't make sense now that I look at the code.
20:05	hiredman	the fast path is there for no reason but to make (> 5) fast
20:07	slashus2	Why does < put (first more) and (next more) in symbols... they use them so often.
20:08	ataggart	s/does/doesn't ?
20:08	hiredman	hopefully first and next are fast
20:08	slashus2	It seems increases the performance when that is done, but it isn't very much unless you are comparing 10000000 numbers.
20:09	slashus2	900 msecs compared to 660 msecs or so.
20:09	slashus2	I ran them each 10 times.
20:11	slashus2	Tested with (dotimes [_ 10] (time (apply < (range 10000000))))
20:13	slashus2	Am I right? :-(
20:14	rhickey	slashus2: this kind of microbenchmarking really isn't productive at this stage. If you make any decisions based upon it they could be completely blown in the future
20:15	slashus2	rhickey: I know, I was just making sure my observation was correct. I wasn't suggest that we change anything. Premature optimization is not the current goal.
20:17	slashus2	sorry :-(
20:19	rhickey	np, I always wonder about such uses of apply
20:24	rhickey	diff much small for me, jdk 6 os x: 760 vs 620
20:25	slashus2	I just did (let [next-more (next more) first-more (first more) ... and used those throughout the bottom part.
20:25	slashus2	I am running jdk 6 on os x too.
20:25	rhickey	with chunks, 612 vs 513
20:25	ataggart	does let have any more overhead than declaring local variables in java?
20:26	slashus2	:-D
20:26	rhickey	ataggart: not unless you do destructuring, then there's some logic too
20:27	rhickey	better to spend time writing interesting software...
20:31	ataggart	agreed, more curiosity than anything else. plus its good to know when someone asks later on.
22:22	holmak_	In a (dosync ...) expression, does the expression block until the transaction completes?
22:25	holmak_	so, if i "(dosync (alter some-ref inc)) @some-ref", am i guaranteed to get the incremented value of some-ref when i dereference it?
22:33	Chouser	holmak_: inside dosync, things may block, but also the whole transaction may be retried.
22:34	holmak_	Okay... so my question is, when you leave the (dosync ) block, does that mean the change has gone through successfully?
22:34	Chouser	yes
22:35	Chouser	of course it may have been changed by something else in the mean time
22:35	holmak_	Awesome, thanks
22:35	holmak_	Yeah, but only one thing changes this -- i'm using the ref to pass a value back across an async Swing call
22:36	Chouser	if you're only ever changing one value at a time in a dosync, you might want to look at 'atom'
22:36	holmak_	does swap! have the same kind of guarantee?
22:36	Chouser	yes
22:37	holmak_	I know the concurrency stuff guarantees consistency and all that, but I wasn't sure if it just acknowledged the transaction and would keep working at it while your code continues
22:37	holmak_	Thanks for the clarification
22:37	Chouser	this is synchronous behavior -- some of the concurrency is synch, other is asynch.
22:38	Chouser	agents, for example, are async. (send-off my-agent inc) @my-agent ... at the deref, the inc may not have happened yet.
22:38	clojurebot	for is not a loop
22:38	holmak_	I had also considered using an agent, since i knew i could 'send' then 'await', but async wasn't the behavior i needed here
22:39	holmak_	I'm pretty okay with the concurrency stuff, its quite neat. There are just a lot of nuances to concurrency guarantees...
22:41	Chouser	indeed
22:41	holmak_	Good times though, that's why I love playing with Clojure. Doing locks correctly is certainly beyond me. :D
22:55	samuels	hey
23:20	samuels	Chouser: do you have a link to your clojure snakes program
23:22	Chouser	samuels: wow. not sure...
23:23	Chouser	hey look at that, it's on github
23:24	Chouser	http://github.com/Chouser/programming-clojure/blob/02485e760de2cbfa5c58d9f97f4f8205c15f1286/examples/snake.clj
23:26	samuels	thanks!
23:27	Chouser	I haven't touched it for ages -- dunno if it even works anymore.
23:28	samuels	does your version have much mutable state?
23:30	Chouser	samuels: I really don't remember, but looking at it I see very little.
23:31	Chouser	there are two 'dosyncs', both changing a single game-state reference
23:35	samuels	thanks
23:40	durka42	~logs
23:40	clojurebot	logs is http://clojure-log.n01se.net/