#clojure – 2008-12-21

2008-12-21

00:06	jawolfe	I've gotta run, never mind ... I'll post in the newsgroup instead. thanks!
00:10	hiredman	(doc concat)
00:10	hiredman	damn it
00:11	jawolfe	Hrm, i suppose apply is evaluating the arguments (duh).
00:12	hiredman	there is clojure.core/concat and clojure.core/lazy-cat both are lazy
00:12	jawolfe	hiredman: thanks
00:12	jawolfe	hiredman: i have looked at both
00:12	dhaya	jawolfe: Looks like concat is lazy except for the first 3 colls, try (take 1 (apply concat (map (fn [x] (prn x) (list x)) (range 1 20))))
00:13	jawolfe	dhaya: that's what it looks like to me too
00:13	jawolfe	for my application, I need something that is lazy for all (but the first, perhaps)
00:14	jawolfe	i.e. a fn that takes a lazy seq of seqs
00:15	jawolfe	and only forces each inner seq when its first element is needed
00:18	lisppaste8	jawolfe annotated #72461 with "solution?" at http://paste.lisp.org/display/72461#1
00:19	jawolfe	Thanks, gotta run
03:32	poiuyt	how can I conj or cons just a parenthesis?
03:33	hiredman	example?
03:33	poiuyt	iwant (conj [+ 2 3] ( )
03:33	poiuyt	-> [+ 2 3 (]
03:34	hiredman	you want the paren character
03:34	poiuyt	ys
03:34	hiredman	character literals in clojure are characters proceeded by \
03:34	hiredman	\(
03:34	hiredman	\)
03:38	poiuyt	ok
03:38	poiuyt	but i cant eval (clojure._PLUS___221@e5ace9 2 3 \( clojure.___247@5ed853 12 3 \))
03:39	hiredman	uh
03:39	hiredman	heh
03:40	hiredman	no
03:40	karmazilla	he want's to conj a list of things
03:40	hiredman	(eval `(+ 1 3))
03:40	hiredman	karmazilla: want he wants is different then what he wants to do
03:41	hiredman	poiuyt: eval takes a list of things, a data structure
03:41	hiredman	(eval (seq [+ 1 3]))
03:51	hiredman	so weird that clojurebot is still here
03:51	hiredman	the server it runs on has been down for hours
03:52	poiuyt	lol
03:52	poiuyt	he has come alive
03:52	poiuyt	kurzweils predictions has been preceeded by reality
03:52	poiuyt	clojurebot
04:05	poiuyt	if i have a vector of arbitrarily nested vectors, what is the best way to visit them all?
04:05	poiuyt	loop recur and call the function again for each subvector?
04:05	poiuyt	is there such a function already?
04:07	hiredman	maybe in contrib?
04:08	poiuyt	what is the best html-parsing lib available in clojure?
04:17	hoeck	poiuyt: http://sourceforge.net/projects/nekohtml worked for me
04:34	poiuyt	how can i ask (operator? x)
04:34	poiuyt	or the common name for +-/*
04:38	poiuyt	can I doc multimethods?
04:42	Lau_of_DK	poiuyt: I dont know of any built in method for testing for operators, but this works
04:42	Lau_of_DK	(= '* (or '* '/ '+ '-))
04:49	meredydd	heya
04:49	meredydd	How do you get a backtrace off a REPL exception these days, without wrapping the whole line in (try (catch...))
04:49	meredydd	?
04:51	leafw	*e
04:51	leafw	or (.printStackTrace *e)
04:51	leafw	or (.printStackTrace (.getCause *e))
04:51	meredydd	Joy. Okay, I'll make a macro.
04:52	leafw	meredydd: all 3 are different levels of detail on the Exception, notice.
04:52	meredydd	I'm aware.
05:01	trasko	is it possible to define my own exceptions and then print the line number hwere it happened?
05:03	leafw	trasko: yes, you can define your own exceptions. And they will automatically show where they were thrown.
05:04	leafw	see (doc throw)
10:03	klinkers	how do i call a method like
10:03	klinkers	Connection con = DriverManager.getConnection
10:03	klinkers	("jdbc:derby:eltesto1;create=true", "flonk", "plonk");
10:03	klinkers	let [conn (new DriverManager.getConnection...?
10:05	leafw	klinkers: use namespace syntax, since it's static
10:06	leafw	(def con (DriverManager/getConnection "jdbc:derby:eltesto1;create=true" "flonk" "plonk"))
10:06	klinkers	ah
10:42	klinkers	how dio im import java.sql.*; ?
10:42	klinkers	(import '(java.sql *)) doesnt work
10:42	kotarak	klinkers: you can't import everything.
10:44	klinkers	how do i append 2 strings?
10:44	klinkers	(+ "hello" " there")?
10:45	kotarak	(str "Hello" " " "there")
10:53	klinkers	how do I break a line?
11:10	kotarak	klinkers: ?? In the output? (newline)
11:52	Lau_of_DK	rhickey: Regarding the new initiatives like the wiki and such, would you consider tracking clojure on http://www.pivotaltracker.com/ ?
14:36	Lau_of_DK	rhickey_: I didnt check my log before I shut down emacs, but did you get a chance to look at pivot tracker?
15:02	Lau_of_DK	Boys, can I provide doc strings for multimethods in some clever way ?
15:02	kotarak	Only we (defmulti #^{:doc "bla"} foo ...)
15:05	Lau_of_DK	Thanks Mr. Brandmeyer
15:07	Lau_of_DK	Hmm, that doesnt connec with (doc foo)
15:17	lisppaste8	drewr pasted "agent queue" at http://paste.lisp.org/display/72487
15:18	drewr	I've created a queueing mechanism for incoming and outbound messages over a socket.
15:19	drewr	The first time a couple of these gets created, everything works fine. Agents are dispatched, actions are called, stuff is logged.
15:19	drewr	The next time around, perhaps when a SocketException is caught, make-queue will get called again.
15:19	drewr	The agent is created and returned.
15:19	drewr	However, the action does not seem to be called (i.e., nothing is logged at least).
15:21	drewr	Is there anything obvious that I'm missing?
15:23	mehrheit	maybe it's blocking indefinitely?
15:24	drewr	What would cause it to block so as not to run f?
15:25	drewr	Specifically, the log on line 14 is printed, but the one on line 4 is not.
15:26	mehrheit	right, I missed that one
15:26	drewr	I even tried a throw in there to make sure there wasn't an I/O issue with stdout on subsequent calls.
15:27	mehrheit	does sending other functions to the returned agent work?
15:30	drewr	mehrheit: I'm not sure I understand.
15:30	drewr	To the agent after it's created?
15:31	drewr	The only way I interact with the agent is (.put (:q @a) "msg...").
15:51	chrisn	hello?
15:52	replaca	chrisn: good afternoon
15:52	kotarak	salut?
15:56	mehrheit	drewr: yes, after creation
15:57	drewr	mehrheit: I found out that the action runs later after I close the connection, so there is definitely something blocking the agent from running dispatchAction().
16:00	drewr	I wonder if LockingTransaction.getRunning() is blocking since this agent gets spawned from another agent.
16:03	rhickey	drewr: how do you know it's not blocked in take?
16:04	rhickey	drewr: I don't see how transactions come into it, but getRunning() doesn't block in any case
16:05	drewr	rhickey: I suppose it could be, but why wouldn't it print the log message? I'm flushing out in my log fn.
16:06	rhickey	drewr: you saw "queue stopped" from prior action?
16:06	drewr	Yes.
16:07	rhickey	drewr: no agent errors?
16:07	drewr	conn is essentially a map like {:sock ... :inq (make-queue ...) :outq (make-queue ...) listener: (listen ...returns an agent too...}
16:08	drewr	The listen agent catches the SocketException and eventually calls (connect) which creates another conn, with new :inq and :outq.
16:08	drewr	It's these new queues that get blocking behavior I don't see the first time.
16:12	drewr	Actually, the log I'm looking at is to a file, but each line is written with (with-open) so it should be flushing on each close.
16:13	rhickey	drewr: "should be" famous last words :)
16:13	drewr	Heh, true.
16:14	rhickey	agents don't use locks/blocking, so I'd look carefully at the use of blocking queues
16:16	drewr	And you don't see any inherent danger in agents calling other agents ad infinitum?
16:16	chrisn	Am am building a rendering system. I know Rich hates mock objects, but I need to make a mock jogl opengl interface. This requires managing some minimal state (textures allocated as well as FBOs allocated for now). I could do this using atoms behind the scene to handle the state. Is this a valid way to go?
16:17	rhickey	drewr: no, the whole point is that, because they can't wait on each other, they can't deadlock
16:17	kbarros	besides performance, is there a semantic difference between "3" and "(int 3)"?
16:17	drewr	rhickey: OK, that's what I figured.
16:19	kbarros	it looks like operations on int's promote them to more general numeric values
16:19	rhickey	kbarros: yes, if you say (let [x (int 3)] ...) then arithmetic involving x and other ints will be 32-bit primitive math
16:19	Chousuke	rhickey: I sent a CA to you on friday. Hopefully it'll arrive safely.
16:20	rhickey	user=> (* 1000000 1000000)
16:20	rhickey	1000000000000
16:20	rhickey	user=> (* (int 1000000) (int 1000000))
16:20	rhickey	java.lang.ArithmeticException: integer overflow (NO_SOURCE_FILE:0)
16:20	drewr	rhickey: If I look at the LinkedBlockingQueue, it's filled with stuff from the socket's input stream. It seems to indeed be that the agent's action isn't running to start .take-ing.
16:23	kbarros	user=> (def a (int 1000000))
16:23	kbarros	user> (* a a)
16:23	kbarros	does not overflow
16:25	rhickey	kbarros: the types have to be accessible to the compiler - there's no type on a, so generic arithmetic
16:25	rhickey	primitives are limited to let locals and args
16:26	rhickey	drewr: you need to make sure the queue you are looking at is the one the agent is
16:27	kbarros	ok, thanks. when using literal numbers could the compiler, as an optimization, generate primitive operations? i.e., could it fall back to generic arithmetic when necessary?
16:30	drewr	rhickey: I'm (fairly) sure of it. All the messages flush out after I "stop" the queues.
16:30	rhickey	kbarros: that's kind of what the generic math is, also there is an interaction with boxing, which dominates over the dispatch. A local with a primitive type is unboxed, a literal is boxed
16:31	drewr	I just don't understand why the ones for the first connection act "normally" and then subsequent ones block.
16:31	drewr	They're different agents, actions, and states.
16:31	drewr	Each call to make-queue creates a different BlockingQueue.
16:34	rhickey	drewr: you sure you're not being fooled by the logging not flushing?
16:37	drewr	Yes, because the logging acts like I expect with a fresh connection -- (read-line) returns line, .puts it into queue, agent .takes it, dispatches on it, logs it, etc.
16:38	drewr	Besides, I put a (throw) in the action just to make sure it wasn't a buffering issue, and the agent didn't report any errors.
16:42	rhickey	drewr: can you make an isolated reproducing test case?
16:42	drewr	I'm thinking of how I can do that. :-)
16:42	chrisn	why are there specific setters for arrays of different types, but not different getters?
16:55	klinkers	hmm lets say you have a bunch of functions that operate on some structure. lets say they all take 4 params. now for all these functions the 1st param will always be the same. can i somehow write a macro that changes function-definitions? hmm does this make any sense?
16:55	kbarros	I get a "class cast exception" for the following expression: (double 1e500) . Is this the appropriate compiler error? It confused my emacs mode.
17:03	RSchulz	kbarros: I think it might be a bug. Witness:
17:03	RSchulz	(double 1e500)
17:03	RSchulz	java.lang.ClassCastException: clojure.lang.Symbol cannot be cast to java.lang.Number
17:03	RSchulz	user=> 1e500
17:03	RSchulz	Infinity
17:04	RSchulz	user=> (class 1e500)
17:04	RSchulz	clojure.lang.Symbol
17:04	RSchulz	I think that's the problem. You get a symbol when you give a floating (or double) numeric literal that is out-of-range for a double.
17:09	RSchulz	Perhaps it's deliberate to ensure that no computation takes place using an unrepresentable number.
17:09	RSchulz	It could be argued that it should be converted to one of the arbitrary-precision formats.
17:14	gnuvince_	RSchulz: by the way, did you start a blog? I saw you asked about blogging services on the mailing list.
17:15	kbarros	rschulz: thanks
17:15	RSchulz	gnuvince_: Not yet. I think it might have been a passing fancy.
17:16	kbarros	i was expecting (double 1e500) to be the double representation of infinity
17:16	gnuvince_	That's too bad, you seem to have very good knowledge, it would've been a nice boost to the community.
17:23	Lau_of_DK	In a try/catch, how do I print the exception trapped?
17:25	Lau_of_DK	Nevermind, I got it
17:28	RSchulz	gnuvince_: I haven't definitely decided not to. I'm just worried about being able to sustain a blog long enough for it to be meaningful.
17:29	drewr	rhickey: FYI, I switched that .take to a .poll (which shouldn't block), and get the same behavior.
17:29	klinkers	if i make a new namespace, cant i def taken names there?
17:29	RSchulz	Part of my problem (psychologically...) is that as soon as I've accomplished something, it no longer seems special to me and hence no longer worth writing about.
17:30	gnuvince_	RSchulz: you're not alone ;)
17:30	drewr	Is there any reason why an agent won't fire an action ASAP after instantiation?
17:37	rhickey	drewr: agent sends don't dispatch until agent action returns
17:38	rhickey	see #5 here: http://clojure.org/agents
17:39	drewr	From what I'm seeing, #1 is what is postponed.
18:16	replaca	how do I add stuff from a jar in the classpath to the repl? I've been trying stuff like (load "/clojure/contrib/lazy-xml") but this doesn't seem to get it.
18:16	replaca	(Up til now, I'd just been going straight from Slime)
18:17	dreish	add-classpath
18:17	rhickey	replaca: the stuff is a lib?
18:17	rhickey	(load 'clojure.contrib.lazy-xml)
18:18	replaca	yeah, in a jar. The jar is already on my classpath
18:18	rhickey	load takes the name of the lib as above
18:18	Chouser	load takes a string, doesn't it?
18:19	rhickey	but you really want require don't you?
18:19	Chouser	(require 'clojure.contrib.lazy-xml) or (use 'clojure.contrib.lazy-xml) should work.
18:19	replaca	there we go!
18:20	replaca	rhickey: thanks for the noob answer
18:20	replaca	my lisp skills and my java skills haven't quite jelled into the clojure mindset :-)
18:21	replaca	require is exactly the button I was looking for
18:21	rhickey	Chouser: you're right, so would need slashes and -/_ translation
18:22	Chouser	but not the .clj extension. ...far too low-level to bother explaining. :-)
18:30	replaca	Another Q: is there a standard way now of doing AOT compiles from the command line (or ant tasks)? I saw the group thread, but it still seems unresolved.
18:30	replaca	Or should I just write a clj helper for ant?
18:35	Nafai	replaca: That would be useful, if there isn't one already
18:38	replaca	Nafai: yeah, it seems necessary eventually for packaging things. But it might not be today's most important problem for anybody. :-)
18:38	Nafai	replaca: I was considering doing that myself
18:39	Nafai	Because I was going to try to write Android apps with Clojure
18:39	Nafai	But I ran into some snags :)
18:39	replaca	Yeah, to be really useful it needs to be in the core
18:40	replaca	I think we need command line tools that let you "ignore" the jvm
18:40	replaca	This has been discussed at some length in the group and I don't really have a lot to add to that discussion
18:41	replaca	nafai: what kind of snags did you run into?
18:41	Nafai	replaca: A couple: 1) the compiler for dalvik, the Android vm, won't compile the existing Clojure jar
18:42	Nafai	2) Clojure still uses a runtime dynamic classloader and that won't work with dalvik either
18:42	Nafai	So my snags weren't related to making the ant task :)
18:42	replaca	oh, you wrote about that in the group, right?
18:42	Nafai	I've just discussed that here
18:43	replaca	I'd love to see android work too. It seems like it would be awesome to write clojure apps targeted at mobile devices
18:43	replaca	Nafai: I guess I read it while lurking one day :-)
18:44	Nafai	Yeah, me too
18:45	klinkers	can someone give me an example of how i should name my namespace
18:45	klinkers	i stil dont get how to refer tot hem from others
18:46	klinkers	i want like python: import numpy as N; N.f
18:47	replaca	klinkers: mostly you don't use import to pull in other clojure stuff, you use (require) or :use if you're in an ns statement
18:48	replaca	Like this:
18:48	replaca	(ns foo (:use com.infolace.format))
18:49	replaca	or
18:49	replaca	(require 'com.infolace.format)
18:49	replaca	the java babble of the inverted domain name is completely optional (just convention)
18:50	hiredman	I just put everything under hiredman
18:50	hiredman	hiredman.clojurebot
18:51	klinkers	but where does :require look?
18:52	klinkers	im a in a dir datstructures now
18:52	klinkers	im a in a dir datastructures now
18:52	klinkers	i have one file (ns datstructures.stack)
18:52	replaca	In the classpath
18:52	klinkers	ah i see
18:53	replaca	so the "." is typically a directory boundary
18:53	replaca	so that would look for datastructures/stack.clj
18:54	replaca	(again the inheritacnce of java/jvm)
20:04	replaca	Answering my own earlier question (in case anyone cares): You can compile from the command line with java -cp ... clojure.lang.Compile foo bar
20:04	replaca	where foo and bar are the names of namespaces to compile
20:05	replaca	There is an ant task in the latest clojure build.xml
20:36	Chouser	rhickey: {} constructs an ArrayMap -- is that documentable, or incidental?
20:37	rhickey	Chouser: when does it matter?
20:42	RSchulz	user=> (class {})
20:42	RSchulz	clojure.lang.PersistentHashMap
20:43	RSchulz	user=> (ancestors (class {}))
20:43	RSchulz	#{clojure.lang.IObj java.lang.Iterable java.util.concurrent.Callable clojure.lang.IPersistentCollection java.lang.Object clojure.lang.APersistentMap clojure.lang.Obj java.lang.Runnable clojure.lang.AFn java.io.Serializable java.util.Map clojure.lang.IFn clojure.lang.Associative clojure.lang.IPersistentMap}
20:45	Chouser	rhickey: It doesn't much, but the ArrayMap docs could mention that {} is one way to construct one.
20:45	Chouser	RSchulz: try (class {:a 1 :b 2 :c 3})
20:46	rhickey	Chouser: apparently it isn't, currently
20:46	Chouser	depends if it has any entries. I guess I'll skip it.
20:47	rhickey	Chouser: and number of entries
20:47	rhickey	user=> (class {:a 1 :b 2 :c 3 :d 4 :e 5 :f 6 :g 7 :h 8})
20:47	rhickey	clojure.lang.PersistentArrayMap
20:47	rhickey	user=> (class {:a 1 :b 2 :c 3 :d 4 :e 5 :f 6 :g 7 :h 8 :i 9})
20:47	rhickey	clojure.lang.PersistentHashMap
20:48	Chouser	ok
21:42	Dmisener	How do I programmatically what the current version of clojure that is running?
21:43	Dmisener	As you can imagine Google "Clojure version" isn't much help {grin}
21:46	Chouser	Dmisener: did you download a release, or use svn?
21:50	Dmisener	Downloaded the release... I know what version it is ... but was wonder if I was to make version sensitive code.. how would I determine the current running version.
21:50	Chouser	no such capability yet.
21:53	Dmisener	Thanks...
23:50	replaca	Q: what sorts of things can out be bound to and have print/pr work?
23:50	hiredman	writers
23:50	hiredman	filewriter, printwriter
23:51	replaca	thanks, hiredman