#clojure – 2008-06-17

2008-06-17

02:22	jcrites	what class should I run to run a file rather than epl?
04:48	StartsWithK	how would I do multimethod dispatch based on type?
05:11	cgrand	jcrites: clojure.lang.SCript
05:13	cgrand	StartsWithK: if you want to dispatch on parent class or interfaces, you can't
05:18	StartsWithK	hi
05:20	StartsWithK	well i am not sure do i need this at all
05:20	StartsWithK	but i think i would like to dispatch on [float float float] or [double double double]
05:20	StartsWithK	or something like that
05:21	StartsWithK	is something like that posible?
05:22	cgrand	[float float float] => your mutimethods take 3 args and they are all floats?
05:22	StartsWithK	(defmulti fun (fn [x y] [(class x) (class y)]))
05:22	StartsWithK	something like that
05:23	StartsWithK	is something like that right way to do it?
05:23	cgrand	it could work as long as you are dispatching on the actual type
05:24	StartsWithK	is there maybe some better way to do it?
05:24	cgrand	none that I'm aware of
05:26	StartsWithK	i was planing to wrap parts of jogl, and for instance there are vertex[234][fdis..] functions
05:26	StartsWithK	and i would like to have just one vertex function in my code
05:27	StartsWithK	is this smart way to go about it?
05:27	cgrand	hmm I'm afraid that wrapping with multimethods is going to kill perfs
05:27	StartsWithK	i think so to
05:28	StartsWithK	but maybe it won't be that bad
05:28	StartsWithK	its just i cant think of any other way to do it at all
05:29	cgrand	macro or inline may be a better way
05:30	StartsWithK	well if i use macro or definline for (defn color [r g b a] (. gl (glColor4? r g b a))
05:31	StartsWithK	so ? has to be i, f, d..
05:31	StartsWithK	i guess i could declare them all floats :)
05:32	cgrand	you're right: macro or inline won't cut (my OpenGL is getting old)
05:33	StartsWithK	i blame jogl for having two places with type info for its arguments
05:53	cgrand	a java helper class can allow you to get rid of the type suffixes, allowing you to use inlines (and the helper class methods would get inlined at runtime) but it requires writing some java :-(
06:42	StartsWithK	colud i generate them with genclass?
06:42	StartsWithK	i didn't look to much into genclass lib but it soudls like it could be used for something like this
08:45	asbjxrn	rhickey, Hi, there was a commute question here yesterday about how it differs from eg. alter.
08:45	rhickey	alter is a read-modify-write operation, so if another transaction does the same, one will roll back...
08:46	asbjxrn	Am I right that if two sync blocks each do an alter on the same ref one (or both) ro....
08:46	asbjxrn	rigt.
08:46	asbjxrn	but commute does not roll back?
08:46	rhickey	commute presumes the fn is commutative, so applies the fn to the current value of the ref at commit time, if another transaction does the same, it just lines up to do so, no rollback
08:47	rhickey	so transactions that just inc counters or unconditionally add to maps need not have read-modify-write characteristics
08:47	asbjxrn	Ok, but for that to be safe commute must be atomic, or am I wrong?
08:48	rhickey	it is atomic, part of transaction commit
08:49	asbjxrn	? I thought a transaction commit only happened at the end of the sync?
08:50	rhickey	right, your transaction will see a value for commit while running that might not be the value at transaction end, i.e. the commute will be done twice, once on the transaction temp value and finally on the ref during commit
08:51	rhickey	so it can proceed seeing a non-zero counter or a map containing the added value
08:51	rhickey	all sets are done to transaction temps until commit
08:52	asbjxrn	What I mean, if I do a (commute foo fun1 23) and while fun1 is running I do a (commute foo fun2 23) and fun2 finishes updataing foo before fun1 finishes.
08:52	rhickey	these are separate transactions?
08:52	asbjxrn	yes.
08:52	rhickey	then they don't see each other
08:53	rhickey	they will commit in an order and the commutes will happen in that order
08:53	asbjxrn	Yes, I think I understand now. Both are using the temp value seen at the start of the transaction...
08:53	rhickey	plus their own mods to it
08:53	rhickey	including commutes
08:54	asbjxrn	Right, I was thinking one commute would update the ref in all the refs in all the transactions. Didn't occur to me that it could be postponed until the end of the transaction.
08:55	rhickey	nothing shared changes until commit
08:59	asbjxrn	Clojure really force me to think about variables in a different way from what I'm used to.
08:59	asbjxrn	Takes some time to get all the concepts lined up in my mind.
09:05	rhickey	commute is definitely different, actually an old database thing called field calls:
09:05	rhickey	http://books.google.com/books?id=S_yHERPRZScC&pg=PA431&lpg=PA431&dq=database+%22field+calls%22&source=web&ots=JJluVOMDxm&sig=Pw2OK3MnzamxuUcCNEBHuRuGY6c&hl=en&sa=X&oi=book_result&resnum=5&ct=result
09:07	asbjxrn	Heh, I wonder what your reference database looks like, you seem to throw links like that out all the time.
09:07	rhickey	:) that's a great book, highly recommended
09:09	cemerick	is anyone aware of anyone working on generating javadoc-esque documentation from clojure source?
09:09	Chouser	rhickey: thanks for explaining commute. I should have realized it was just deeper magic than I was allowing for.
09:09	rhickey	Chouser: sure, sorry I missed it yesterday, was in a rare (for me) meeting
09:10	rhickey	cemerick: the API page of the docs is generated from the source
09:10	Chouser	rhickey: np. nothing urgent about it. I suppose I could have tried to read the source.
09:11	rhickey	Chouser: seems a shortcoming of the docs if it wasn't clear
09:11	Chouser	are the "for Java Programmers" slides up anywhere? Mind if I steal a few?
09:11	blackdog	would it be useful to have a javadocs thing with @author etc interpreted and xrefs? could be a good project
09:12	asbjxrn	I tried... But the (commute ref fun args) just became (commute fun args) :) I guess the magic was in the java source which really is magic to me.
09:13	blackdog	rhickey, or would that be better left to metadata?
09:14	rhickey	blackdog: metadata
09:14	blackdog	ok
09:14	cemerick	rhickey: Yeah, I figured. I'm turning around various documentation ideas. Perhaps it'd be more reasonable to generate javadoc stuff from gen-class output...
09:15	rhickey	gen-class could have more gen definitely, including Clojure stubs
09:15	blackdog	cemerick, the enclojure plugin for netbeans has code which generates it all in a tree for you
09:15	blackdog	and it's a stand alone swing jpanel
09:16	blackdog	cljNamespaceBrowser.clj
09:16	StartsWithK	rhickey, is something like this http://pastebin.com/d3b2658f6 a bad idea?
09:16	blackdog	if you run that you get most of what you're looking for
09:16	rhickey	Chouser: http://groups.google.com/group/clojure/web/clojureforjava.pdf
09:17	Chouser	thanks
09:18	blackdog	rhickey, have you defined standard meta tags then ? for author etc, or documented them so we all use the sasme
09:18	blackdog	keys
09:19	Chouser	pretty tedious to tag every var in a file with the same author string
09:19	rhickey	blackdog: just by usage, if using your own, use a ns prefix.
09:19	blackdog	yes but for 3rd party tools
09:20	blackdog	so docs could be cross referenced etc
09:20	rhickey	metadata for namespaces has been requested
09:20	Chouser	you could put file-level stuff in per-namespace map. foo/meta or something
09:20	rhickey	StartsWithK: what is your goal with that?
09:21	StartsWithK	i would like to wrap parts of jogl
09:21	StartsWithK	so there woudn't be color3f color4f color3d color4d...
09:21	rhickey	ugh
09:21	StartsWithK	just one color, normal, vetex etc.. function
09:22	rhickey	they letters at the end are types?
09:22	StartsWithK	yes
09:22	rhickey	that is goofy for a Java lib
09:22	StartsWithK	function[number-of-args][arg-type]
09:22	asbjxrn	Standard openGL naming.
09:22	StartsWithK	yes, just wrapped opengl
09:22	rhickey	are the args of uniform type, e.g. 3 floats or 3 doubles?
09:22	StartsWithK	yes
09:23	rhickey	you can just dispatch on the type of the first then
09:23	rhickey	save some overhead
09:23	StartsWithK	but you have color3f color3d and color4f color4d
09:24	rhickey	right, still arity overloaded
09:24	StartsWithK	yes :/
09:25	StartsWithK	another suggestion was i write thin java wrapper to remove types from names..
09:25	rhickey	if you do that, you can combine it with definline or macros for optimal speed
09:26	StartsWithK	and i was looking at genclass to use it instead of java
09:26	StartsWithK	but i can't figure out how it should be used
09:26	asbjxrn	If you're writing a wrapper for parts of openGL, would it be acceptable for you to just select one type and stick with that?
09:26	rhickey	genclass is not for from-scratch Java class implementation
09:27	rhickey	it's for class/interface extension/implementation
09:27	pjb3	hey, the website looks great!
09:27	StartsWithK	asbjxrn, it would work i guess..
09:27	rhickey	else it would have to have all the cruft of Java, field types, access modifiers etc
09:27	StartsWithK	i see
09:27	rhickey	pjb3: thanks, my brother Tom did it
09:28	StartsWithK	then java + definline would work the best i guess
09:28	rhickey	StartsWithK: yes
09:28	StartsWithK	thanks
09:29	rhickey	StartsWithK: but what about asbjxrn's suggestion, just use one type?
09:31	StartsWithK	rhickey, i have something like that right now, but i call dem colorf colord and then i can just (def color colorf) if floats will be used
09:33	asbjxrn	I haven't used openGL enough to understand the tradeoffs between the different typed functions. Can you explain? (if it can be done in a concise manner.)
09:34	asbjxrn	(kinda off-topic here, I guess.)
09:35	StartsWithK	i wish i could explain, but i just started with opengl, so just wanted to duplicante the api without messing something up with my assumptions
09:36	asbjxrn	Ok, never mind. I was just curious.
09:36	pjb3	Where is load-file defined?
09:37	rhickey	pjb3: RT.java
09:39	pjb3	rhickey: Why not in boot.clj?
09:40	asbjxrn	Sounds like you would be ok just going with floats: "OpenGL uses floats internally, an using anything other than the single-precision floating-point functions adds a performance bottleneck because the values are converted to floats anyhow before being processed by OpenGL"
09:40	rhickey	pjb3: what would load boot.clj?
09:40	asbjxrn	Makes you wonder why they added the other typed functions..
09:40	yrb	asbjxrn: I belive it comes from the "performance" crowd, memory space vs precision trade offs mainly
09:40	pjb3	ah
09:41	pjb3	I knew there would be a good reason :)
09:41	asbjxrn	That quote is from OpenGL SuperBible. (I had a plan of learning openGL once..)
09:43	yrb	asbjxrn: I am porting a couple of the examles and doing a port of ants.clj, but man is jogl ugly
09:44	yrb	asbjxrn: looks like a c library but worse because they have shoehorned bits into classes
09:44	asbjxrn	Obviously StartsWithK is then responsible for making a edible wrapper.
09:45	StartsWithK	asbjxrn, thanks, so i guess ill just create it with floats only
09:47	yrb	I have ditched it all for maps of "floats" basically is my little wrapper
09:48	yrb	I don't think the way I am heading is going to preform all that well, but I am mainly aiming for it being good for toys
09:55	asbjxrn	Even toys might be hampered with bad performance. My graphic toy lib lags on my once top-of-the-line 667 Mhz powerbook if I'm not careful.
10:09	cemerick	would I be better off using building enclojure from svn head, instead of using the current public build?
10:41	StartsWithK	hg clone http://freehg.org/u/kresimir/neman/
10:42	StartsWithK	you will need lib.clj from clojure-contrib
10:42	StartsWithK	and there is simle-jogl that show how it will look one day :)
10:42	StartsWithK	and jogl in classpath
10:56	yrb	looks good :)
10:56	yrb	A
10:57	StartsWithK	ill try to do rest soon, and if you wish just browse the source http://freehg.org/u/kresimir/neman/
10:58	yrb	have you looked at java3d
10:58	StartsWithK	yrb, how did you deal with arrays? i now get te same problem with vertex3v and vertex4v variations..
10:59	StartsWithK	yrb, i did a little, but it looked to complicated
11:01	yrb	I used a small map for my vecs
11:05	yrb	using a map instead of a vector is very convenient
11:06	rhickey_	yrb: how so?
11:09	yrb	well if i call Vertex2 it just pulls out the :x and :y keys and can us the :or to provide defaults, so if you call Vertex3 with an map that only has :x, :y it can default :z
11:09	yrb	use*
11:10	rhickey_	ok
11:18	yrb	and the (:key map) syntax is very nice :)
11:19	rhickey_	yrb: if you are going to have a lot of those, consider using a StructMap
11:21	pjb3	Can you have a function that has keyword parameters?
11:22	pjb3	so you can call it like (it :foo 1 :bar 2) or (it :bar 1 :bang 2)
11:22	yrb	I have a StructMap for points, what are the performance characteristics of additional keys like
11:23	rhickey_	yrb: like map's generally
11:24	rhickey_	pjb3: I take a rest arg (&) of key val key val..., then apply hash-
11:24	rhickey_	map to it and destructure in the body, e.g. see refer or gen-class.
11:24	pjb3	ok, that
11:24	pjb3	's what I thought
11:43	drewr	rhickey_: I implemented a function similar to FILTER which takes a sequence and returns a lazy seq.
11:43	drewr	I found that the (when (seq coll) ...) form was needed to read in the seq lazily.
11:43	drewr	I don't understand why.
11:44	drewr	You're not storing the value of the SEQ call. How does it affect coll?
11:44	rhickey_	that's a test to see if there's anything to do
11:44	drewr	Yes, but without it, my function would blow the heap.
11:45	drewr	It would try to consume the whole coll.
11:45	rhickey_	want to paste it?
11:45	drewr	Sure.
11:46	lisppaste8	drewr pasted "LAZY-CONS in WHEN SEQ" at http://paste.lisp.org/display/62376
11:47	drewr	If I don't wrap the LAZY-CONS in that WHEN, the function works, but it tries to consume all the messages (at least, that's my perception because it runs out of heap space).
11:48	rhickey_	your fn is same as (partial map attach-message-details)
11:49	drewr	Thanks, I figured there was a built-in way to do something in the middle of the stream.
11:49	rhickey_	be careful testing on long inputs right in the repl, as your args are on the repl's stack (i.e. it keeps the head), unless you wrap with (let [] ...) or times
11:49	drewr	Hm. They're coming from a database.
11:50	drewr	I created a function which runs a query in batches and returns the rows as lazy stream.
11:50	rhickey_	when has nothing to do with the laziness, just tells you when you're done, maybe an infinite loop caused heap exhaustion?
11:51	rhickey_	i.e. without when, how will it stop?
11:51	drewr	I figured the seq coming in would let it know. :-)
11:51	drewr	That's probably what it was then.
11:52	drewr	I knew the SEQ didn't have side effects so I wasn't sure what was happening.
12:05	yrb	StartsWithK: Misread your question before...(too much api exposure) I have chosen not to really deal with arrays yet, since I didn't need them for the particle rendering I was going
12:11	StartsWithK	yrb, thats ok, i was just trying for a while to picture how i should do that :) i'll skip them too for now
12:11	yrb	There are 2040 constants for GL ...
12:11	yrb	2640*
12:14	yrb	StartsWithK: I thinking of moving over to j3d over jogl
12:16	StartsWithK	yrb, maybe you should check out http://www.jmonkeyengine.com/ too
12:17	StartsWithK	i never used it, but it looks like it has everything j3d
12:20	yrb	hrmmm, I am mainly looking to do visualisation
12:22	yrb	though it looks very cool
12:24	yrb	man j3d still exposes you to the the whole blah{2,3,4}{d,f}
12:31	yrb	Javamonkey looks quite sane however
12:33	yrb	JMonkey Engine*
13:01	yrb	StartsWithK: I really like the pattern matching dispatch :)
14:44	Lau_of_DK	Is there any documentation on Swing Eventlistners vs. Clojure ?
14:47	Lau_of_DK	mainPanel.addMouseListener(new java.awt.event.MouseAdapter() {
14:47	Lau_of_DK	public void mouseClicked(java.awt.event.MouseEvent evt) {
14:47	Lau_of_DK	onPanelClick(evt);
14:47	Lau_of_DK	}
14:47	Lau_of_DK	});
14:47	Lau_of_DK	There curlies are throwing me a bit off, is there a clever way to implement this in Clojure ?
14:50	cgrand	proxy?
14:50	Lau_of_DK	I dont think that you proxy listners, Im not trying to override something like I would with Paint for instance
14:52	Lau_of_DK	wait
14:52	Lau_of_DK	I think you're right
14:52	cemerick	I think proxy is the way to go.
14:52	Lau_of_DK	lemme think this through
14:53	cemerick	Lau_of_DK: Note the "simple example gui application" here: http://en.wikibooks.org/wiki/Clojure_Programming
14:54	cemerick	I think I saw a sample somewhere that used a macro that allowed you to pass a fn instead of explicitly using proxy, and the macro would line the fn up with the interface to be implemented
14:54	cemerick	Of course that'd only work for an interface that has a single function.
14:55	Lau_of_DK	(def panel (doto
14:55	Lau_of_DK	(JPanel. (MigLayout.))
14:55	Lau_of_DK	(addmouseEventListener
14:55	Lau_of_DK	(proxy [MouseAdapter] []
14:55	Lau_of_DK	(on-click)))))
14:56	Lau_of_DK	This works, but I need to ship some MouseArgs to MouseAdapter, but any args added causes an Unable to Resolve Symbol error... why? :)
14:59	Lau_of_DK	I dont fully understand the syntax, but this works
14:59	Lau_of_DK	(def panel (doto
14:59	Lau_of_DK	(JPanel. (MigLayout.))
14:59	Lau_of_DK	(addMouseListener
14:59	Lau_of_DK	(proxy [MouseAdapter] []
14:59	Lau_of_DK	(on-click [evt])))))
14:59	Lau_of_DK	Nevertheless, it does not seem to fire (on-click) when I click the panel
15:04	cgrand	I presume you defined a on-click function and expect it to be called?
15:04	Lau_of_DK	yes sir
15:04	Lau_of_DK	(defn on-click
15:04	Lau_of_DK	[g]
15:04	Lau_of_DK	(println "Click"))
15:05	cgrand	(mouseClicked [evt] (your) (code) (here)))))) instead of (on-click [evt])))))
15:07	Lau_of_DK	That compiles as well, but nothing is firing
15:11	Lau_of_DK	cgrand, does it fire in your camp ?
15:24	cgrand	yes: (def f (javax.swing.JFrame.)) (.show f) (.addMouseListener f (proxy [java.awt.event.MouseAdapter] [] (mouseClicked [e] (println "Click") (flush))))
15:29	Lau_of_DK	That works, thanks alot cgrand
15:49	rhickey_	anyone got a regex that I can use to remove non-double newlines? For the generated API docs I'd like to get rid of single newlines so it flows better.
15:51	drewr	rhickey_: #"[^\n]\n[^\n]"
15:51	drewr	Or..
15:51	drewr	#"[^\n]\w?\n\w?[^\n]"
15:52	drewr	Perhaps. Depends on how predictable the input is.
15:53	Chouser	Java and emacs do regex very poorly, worse even than sed
15:53	drewr	You could do it with a DFA, but that would be a little overkill.
15:54	Chouser	#"(?<!\\n)\\n(?!\\n)"
15:54	Chouser	but you want to remove them? re-seq does the opposite of what you want.
15:55	rhickey_	String.replaceAll ?
15:56	Chouser	hm, takes a String, not a Pattern.
15:57	rhickey_	docs for it say:
15:57	rhickey_	An invocation of this method of the form str.replaceAll(regex, repl) yields exactly the same result as the expression
15:57	rhickey_	Pattern.compile(regex).matcher(str).replaceAll(repl)
15:57	rhickey_	so I could do that
15:58	drewr	If the docs are generated, why not fix whatever's generating the single newlines?
15:58	rhickey_	the newlines work in the source, and for the doc function
15:58	drewr	So you are munging the docstring as you output it.
15:59	rhickey_	yup
15:59	Chouser	yeah, that looks good: (.replaceAll (.matcher #"(?<!\\n)\\n(?!\\n)" text) "")
15:59	rhickey_	thanks all
15:59	Chouser	or drewr's regex of course would be fine too.
16:03	drewr	If I've got something long-running that I'd like to kick off, but still have control of my REPL, is the best technique to accomplish that manually creating a Java thread for the function to run in?
16:03	Chouser	you could use send-off
16:05	rhickey_	FYI, API page in docs now has anchors: http://clojure.org/API#map
16:05	drewr	Chouser: Ah yes, I remember that from ants.clj. Thanks.
16:05	drewr	rhickey_: Very nice. I like the new site a lot.
16:07	Chouser	rhickey_: thanks, that'll be useful.
16:08	Chouser	partition won't generate a short list at the end. Is that intentional?
16:09	Chouser	(partition 2 [1 2 3 4 5]) => ((1 2) (3 4)) not ((1 2) (3 4) (5))
16:09	rhickey_	yes, because it now takes steps as well
16:10	Chouser	ah!
16:11	rhickey_	user=> (partition 3 2 (range 10))
16:11	rhickey_	((0 1 2) (2 3 4) (4 5 6) (6 7 8))
16:11	Chouser	yeah, I saw the step, but here's what I thought I wanted:
16:12	Chouser	(map first (partition 2 (range 5))) from which I wanted (0 2 4)
16:13	Chouser	but (map first (partition 1 2 (range 5))) gets it
16:13	Chouser	(apply str (map first (partition 1 2 (re-split #"(?<!\\n)\\n(?!\\n)" text))))
16:13	Chouser	heh. ok, anyway...
16:14	rhickey_	partition mimics Mathematica's partition, for these simple cases
16:15	rhickey_	(range 0 5 2)
16:15	Chouser	Oh, Mathematica! right!
16:15	Chouser	hm, tone of voice doesn't carry very well over IRC.
16:16	rhickey_	Mathematica is very cool
16:16	rhickey_	Every programmer should see it, just as they should a good Common Lisp environment
16:17	rhickey_	see how the other half lives :)
16:18	Chouser	Back when I still did math more complicated than flubbed attempts to apply the Rule of Cosines (college, in other words), I took a run at Mathematica.
16:18	Chouser	But it was a short run, and I lost. I don't think I had good use case.
16:18	rhickey_	it's really a Lisp, with great data visualization tools
16:20	Chouser	so -> is a port of //
16:20	rhickey_	I can't grok their syntax, so use long names
16:21	Chouser	I don't know what I'm talking about. I just had a vague recollection of Mathematica being postfix.
16:21	Chouser	Apparently I'm wrong in the general case, but it provides a//b as a form of b[a]
16:22	rhickey_	#(%1 %2) is Mathematica-esque, %& being my addition
16:22	Chouser	just like (-> a b) is (b a)
16:22	rhickey_	-> is most like Mathematica Thread, IIRC
16:28	Chouser	surely visualiztion functions could be dropped into clojure
16:29	rhickey_	I would hope so, given all the graphics options available to Java
16:33	Lau_of_DK	How is an Unsupported Operation Exception ?
16:33	Lau_of_DK	How = What
16:34	Lau_of_DK	(. frame
16:34	Lau_of_DK	(addPropertyChangeListener
16:34	Lau_of_DK	(proxy [PropertyChangeListener] []
16:34	Lau_of_DK	(propertyChangeEvent [evt]
16:34	Lau_of_DK	(println evt)))))
16:34	Lau_of_DK	This for instance - Will throw one such
16:35	rhickey_	with proxies, chances are you are failing to implement a method that the consumer of the proxy calls
16:36	Lau_of_DK	k
16:37	Lau_of_DK	yea, you're right rhickey, it expected propertyChange()
16:47	drewr	Where does agent come from in ants.clj?
16:47	drewr	It doesn't get defined anywhere, only used in SEND-OFF a couple times.
16:48	Chouser	It's bound to the current agent.
17:00	Chouser	http://clojure-log.n01se.net/date/2008-04-05.html
17:04	drewr	Heh, thanks.
18:05	rhickey_	I've added experimental static member access classname/member, work for fields with no parens Math/PI, or static method invocations (Math/sqrt 42)
18:10	drewr	Sweet. Consistent with package naming.
18:11	rhickey_	feels right, so far
18:11	pjb3	Can you add jars to the classpath after the JVM has started?
18:12	rhickey_	pjb3: http://clojure.org/API#add-classpath
18:12	pjb3	excellent
18:12	rhickey_	but if it's something you'll use often, best to add it to your real classpath at some point
18:14	pjb3	Can you say "put all the jars in this directory" on the classpath, or you do you have to list them individually when you start the JVM?
18:28	drewr	lisppaste8: url
18:28	lisppaste8	To use the lisppaste bot, visit http://paste.lisp.org/new/clojure and enter your paste.
18:28	pjb3	When I do (add-classpath "file:///Users/pbarry/projects/clojure/mysql-connector-java-5.1.6.bin.jar")
18:28	pjb3	then (.newInstance (.forName Class "com.mysql.jdbc.Driver"))
18:28	pjb3	I get class not found
18:29	lisppaste8	drewr pasted "partial function with side effect" at http://paste.lisp.org/display/62396
18:30	drewr	rhickey_: I'm trying to interject a function in the middle of a stream to increment a ref, but it doesn't increment, and the messages don't come out the other side. Any thoughts?
18:35	rhickey_	pjb3: add-classpath won't work for JDBC drivers, which need to be visible to System Classloader
18:35	drewr	rhickey_: Actually, scratch that. Those functions seem to work. I think my problem lay elsewhere.
18:35	drewr	Although, is that invocation of PARTIAL what you had in mind earlier?
18:36	pjb3	rhickey_: Ok, so I have to specify those with -cp when I start the JVM?
18:37	drewr	No, I needed to return the message in INC-MESSAGE-COUNT.
18:37	rhickey_	drewr: no the partial call shouldn't include all args, and I don't think you need it here at all, make sure you force the evaluation of the lazy map call with doall or dorun
18:37	drewr	rhickey_: Can I keep it lazy as it passes through?
18:38	drewr	I don't want it to consume the seq. I need to chain more stuff to it.
18:58	drewr	OK, my PARTIAL solution works; I just have to make sure to return the object from the seq in my mapped function.
19:04	shizzy0	Hello
19:05	shizzy0	I'm getting familiar with Clojure, and I was curious about what the '#' means.
19:08	shizzy0	[Finds it.] Ok, so #'x means (var x). How similar is this to Common Lisp's use of #'function?
19:35	rhickey_	shizzy0: it means something totally different, as Clojure is a Lisp-1, and the use it almost the opposite, where in CL you'd use #' to get the function value, in Clojure you use #' to avoid getting the value, getting the containing var instead
21:03	mnestic	is there a chat log of #clojure anywhere?
21:07	Ycros	mnestic: I think there is because a google search of mine once hit one
21:10	mnestic	Ycros: ahh, found it
22:03	shizzy0	thanks, rhickey_ for the response (a few hours ago).