#clojure – 2009-07-10

2009-07-10

02:03	thearthur	(partition dropps partial blocks at the end. I need one that doesn't do that
02:03	thearthur	is this in contirb somewhere?
02:04	thearthur	???? (partition 4 [1 2 3 4 5 6 7]) => ((1 2 3 4))
02:05	thearthur	(my-partition 4 [1 2 3 4 5 6 7]) => ((1 2 3 4) (5 6 7))
02:10	thearthur	nm found that partition can do this with its pas arg
02:10	thearthur	(partition 4 4 () [1 2 3 4 5 6 7])
02:24	Raynes	Is there a predicate to check whether or not a sequence is empty?
02:24	arbscht_	,(empty? [])
02:24	clojurebot	true
02:24	Raynes	Thank you. :)
02:27	thearthur	(push and pop mention working on queues. how do i create a clojure.lang.PersistentQueue
02:33	mrpika	there doesn't appear to be a clojure wrapper for it
02:34	mrpika	you'd have to use it via java interop
02:35	mrpika	or conj
02:35	mrpika	http://markmail.org/message/brvozelsgmr5a3ba
04:01	thearthur	how do i undefine a var from the repl?
04:01	thearthur	dont want to reaload everything
04:06	Chousuke	ns-unmap should help
04:12	Raynes	Why would you need to undefine a var from the REPL?
07:55	Holcxjo`	I am lost with Java interop. I have a class that has a static class as its member. How do I instantiate that inner class?
08:02	Holcxjo`	To be specific: I am trying out the Evernote API. The Java demo does: import com.evernote.edam.userstore.*; ... UserStore.Client userStore = new UserStore.Client(...args..)
08:03	Holcxjo`	I can do (import '(com.evernote.edam.userstore.UserStore)) but anything after that fails...
08:03	Holcxjo`	com.evernote.edam.userstore.UserStore/Client (. com.evernote.edam.userstore.UserStore Client) (com.evernote.edam.userstore.UserStore/Client. <some arg>) ...
08:04	Holcxjo`	Any ideas?
08:04	liebke	Holcxjo`: you access inner classes with $, e.g. Class$InnerClass
08:04	Holcxjo`	Oh!
08:04	liebke	yeah, not really obvious :)
08:04	Holcxjo`	Also not in the book I believe...
08:05	Holcxjo`	It's on http://clojure.org/java_interop if you know what the correct term (inner class) is...
08:05	Holcxjo`	thanks!
08:06	liebke	no problem :)
09:52	Raynes	I accidentially wrote a faster (apply max ..) :o
10:08	Chouser	I think they're making fun of us. http://lambda-the-ultimate.org/node/3488
10:15	Chousuke	nah :)
10:16	cemerick	Chouser: I think they were aiming at NewLisp
10:16	Chouser	ah
10:17	liebke	yeah it doesn't look like it's aimed at clojure, read the paper (pdf) http://jfm3.org/phosphorous.pdf
10:18	Chouser	I guess it was the JVM comments that made me wonder
10:18	Chousuke	I like how their CamelCase convention extends to the paper as well.
10:25	cemerick	Chouser: JVM:lispers :: Microsoft:free-software-advocates
10:25	Chouser	heh
10:37	Raynes	Like parent's tell their children. "Don't mind them, they are only upset because you're so special!"
10:41	Chouser	does anyone have a link to the little outline about new-new plans?
10:42	cgrand	Chouser: http://www.assembla.com/wiki/show/clojure/New_new ?
10:42	Chouser	ah, yes. thanks!
10:44	achim`	oh, the NewLisp presentation has a nice font selection
10:45	achim`	does anybody happen to know the monospaced font? :)
10:49	lisppaste8	ceninan pasted "macro/seq practice "dsl"" at http://paste.lisp.org/display/83369
10:49	ceninan	I thought I'd learn some macro-fu by doing something that changes lisp semantics, and not just simple syntax "prettification"
10:49	ceninan	so I "designed" (lol) something that I thought would be interesting to implement
10:50	ceninan	anyone familiar with the Icon language?
10:52	cemerick	seems like assert should take an optional failure-msg arg
10:52	Chouser	cemerick: you're playing with assert? Have you seen the new pre/post conditions?
10:52	Chouser	ceninan: not me, sorry.
10:54	cemerick	Chouser: oh, no -- someone mentioned them to me, but I'd forgotten
10:54	cemerick	we're on v1.0 + patches for a while, though
10:55	Chouser	oh, ok.
10:56	cemerick	I suppose aggressively tracking head might be a good idea, but that'd take some CI work.
10:57	Chouser	I'm using 1.0 at work myself, but I also haven't touched our .clj code in a while.
10:57	cemerick	it's the only reasonable way to do things atm, but I do certainly loathe upgrade cycles.
10:58	lisppaste8	ceninan annotated #83369 "untitled" at http://paste.lisp.org/display/83369#1
11:00	Chouser	ceninan: it looks like you're mutating !tries in the middle there
11:00	ceninan	Chouser: that part isn't really correct
11:01	ceninan	the :! should have an optional let form
11:01	Chouser	ceninan: and some minor adjustments to your proposed syntax would make it much easier to implement
11:01	ceninan	I'm all ears :)
11:01	Chouser	looks like an interesting experiment
11:01	Chouser	well, macros are applied at compile time when the compiler sees a (macro-name ...) form
11:03	Chouser	so with your proposed syntax, you either need to define 'integer?', 'between?' and '=' as macros, or wrap the whole thing in a macro call that will walk the tree looking for your :! :<- annotations
11:03	ceninan	is it possible to do a macro macro generator?
11:03	Chouser	ceninan: yes
11:03	ceninan	then there could be some sort of "def" for these types of "functions"
11:04	ceninan	correct?
11:04	Chouser	both the options I described are possible, but neither is particularly simple
11:04	Chouser	yes
11:04	Chouser	but writing such a macro is several steps removed from accomplishing your goal -- unless this specific syntax is your goal, I suppose.
11:05	ceninan	not really stuck on the syntax, no
11:05	ceninan	do you have a simpler proposal?
11:05	Chouser	if you instead wrapped another layer, maybe... (icon-step (integer? (read-line)) :<- ... :! ...)
11:05	Chouser	then you could define a single macro 'icon-step' that did your heavy-lifting.
11:06	ceninan	hmm, yes - that does sound like a more practical approach
11:06	Chouser	and actually, if you got that working and still wanted to write a 'def-icon-fn' macro, it could produce macros that expand to 'icon-step' forms and your life would be more pleasant. :-)
11:07	ceninan	right :)
11:07	ceninan	thanks, hadn't thought of that
11:07	Chouser	as with any macro, I'd highly recommend writing out code the code you want your example to expand to before you start trying to write the macro
11:08	ceninan	and that leads to my next question :)
11:08	mccraig	Chouser: are pre/post conds described/exampled anywhere ?
11:08	ceninan	what would be a good structure for the "seq"?
11:10	mccraig	found a description : http://github.com/richhickey/clojure/commit/0ac482878a1dd520cbee2faa0f5f6ab1082ffa76
11:10	Chouser	mccraig: sure! in the commit message. :-)
11:10	Chouser	there ya go
11:13	Chouser	ceninan: not sure -- seems like figuring that out is the task you've chosen for yourself. :-)
11:14	ceninan	well, that's true ;p
11:14	ceninan	thanks a lot for the help!
11:15	Chouser	I'm not quite sure I grok your syntax, but if you get stuck feel free to ask more.
11:16	ceninan	I'll try to figure it out without dealing with the syntax first :)
11:16	ceninan	thanks again
11:18	ceninan	(btw, they allow you to react to failure conditions; failure can have different causes, eg '=' have the fail conditions '>' and '<')
11:18	ceninan	(tbh, I don't really grok it myself yet ;))
11:19	Chouser	ah! hm.
11:20	Chouser	that makes it sound more like a sort of cond applied to each element of the seq
11:24	ceninan	I'm entirely sure that I follow; the "fail conditions" primary purpose is to produce side effects or change local bindings
11:25	ceninan	they shouldn't escape to a parent operation
11:25	ceninan	but like some sort of (loop (cond)), yes
11:26	Chouser	I'll be fascinated to see what you come up with. :-)
11:26	ceninan	thanks :)
11:28	ozzilee	Chouser: Forgive me for not just trying it or digging through the changelogs, but... the lazy-xml I have doesn't automatically convert tag contents to strings, instead throwing NPEs. Has that changed?
11:29	Chouser	lazy-xml hasn't changed in ages
11:29	Chouser	ozzilee: you're talking about when emitting?
11:30	ozzilee	Chouser: Yeah. Something like {:tag :foo :content [5]} throws an NPE.
11:31	Chouser	ah. I don't think I'd noticed that. A contrib ticket and patch for that would be welcome.
11:33	ozzilee	Chouser: Okay, will do. Probably. At some point :-) I need to work on getting some tests going on this mess before I start updating dependencies.
11:34	Chouser	a ticket without a patch would be fine too
11:35	ozzilee	Any recommendations on that front, btw? As far as a testing framework. Fact maybe? It's hard to keep up.
11:35	hiredman	test-is got moved into core
11:35	ozzilee	Chouser: Ok, I'll do a ticket now.
11:43	ozzilee	hiredman: That looks like it will work. Do you use it?
11:43	ozzilee	Chouser: Ticket submitted.
11:43	Chouser	ozzilee: thanks!
11:44	hiredman	ozzilee: nope
11:45	hiredman	I guess I could write tests for clojurebot
11:45	hiredman	bleh
11:45	Raynes	I wish I knew Java. :\ I might understand what this :state stuff is all about.
11:45	hiredman	Raynes: it creates a field where you can store state information
11:46	hiredman	generally a ref or a atom
11:46	Raynes	Oh, like a field in Java.
11:46	ozzilee	Raynes: Except you only get one per proxy object.
11:46	hiredman	eh?
11:46	hiredman	proxy has a .state?
11:46	Chouser	one per gen-class instance
11:47	Raynes	Works for me. :D
11:47	hiredman	yeah
11:47	ozzilee	Er, gen-class. Yeah. That's what I meant.
12:34	eyeris	When I want to use destructuring to bind a symbol to a map key of the same name, e.g. (let [{k :k} my-map] (...)), is there an abbreviated form of `a :a` that assumes the same name?
12:38	hiredman	(let [{:keys [a]} {:a 1}] a)
12:38	hiredman	,(let [{:keys [a]} {:a 1}] a)
12:38	clojurebot	1
12:41	eyeris	Thanks
13:26	ataggart	does anyone have a clear sense of which datastructure manipulation functions are meta-data preserving?
13:27	ataggart	e.g., conj is, but drop isn't
13:27	Chouser	hm
13:27	Chouser	I'm under the impression that everything on vectors and maps is supposed to preserve metadata
13:27	ataggart	I know I can just look it up, but I thought there might have been some intention behind the choices that I wasn't aware of
13:29	Chouser	lists and seqs ... well I guess I'm not sure
13:30	Chouser	it does seem a bit odd that conj, pop, and next on a PersistentList all return PersistentLists, and conj keeps metadata but pop and next do not
13:31	ataggart	also depends on the undelrying type
13:31	ataggart	,(:foo (meta (conj (with-meta '(1 2 3) {:foo "bar"}) 4)))
13:31	clojurebot	"bar"
13:31	ataggart	,(:foo (meta (conj (with-meta [1 2 3] {:foo "bar"}) 4)))
13:31	clojurebot	nil
13:31	Chouser	I wouldn't expect fns that take a non-seq and return a seq (like drop) would transfer metadata from the coll to the seq
13:31	ataggart	but drop does take a seq
13:31	ataggart	or am I totally bonkers
13:32	Chouser	well, it can take a seq, but it can also take any other collection
13:32	Chouser	,(drop [1 2 3])
13:32	clojurebot	java.lang.IllegalArgumentException: Wrong number of args passed to: core$drop
13:32	Chouser	,(drop 2 [1 2 3])
13:32	clojurebot	(3)
13:32	Chouser	vector in, seq out. I'd be pretty surprised if metadata came through that.
13:34	ataggart	though conj is preserrving, depending on the original type
13:34	Chouser	on the other hand, I'm surprised that pop on a PersistentList loses the metadata. PL in, PL out, just like conj. I wonder if pop's behavior there is intentional.
13:34	Chouser	ataggart: right
13:34	Chouser	I don't know if this is valid, but I tend to think of a seq as a series of things, and wouldn't expect the same metadata on all elements
13:35	ataggart	heh this is funny:
13:35	ataggart	(:foo (meta (conj (with-meta (seq [1 2 3]) {:foo "bar"}) 4)))
13:35	ataggart	,(:foo (meta (conj (with-meta (seq [1 2 3]) {:foo "bar"}) 4)))
13:35	clojurebot	nil
13:35	Chouser	vs. collections like map, vector, and PersistentList where I tend to think of the collection as stable and the metadata applying to the whole thing
13:35	ataggart	,(:foo (meta (conj (with-meta (seq '(1 2 3)) {:foo "bar"}) 4)))
13:35	clojurebot	"bar"
13:35	ataggart	though you can with-meta a seq
13:35	ataggart	as above
13:35	Chouser	hmph.
13:36	ataggart	which makes sense in the idealized abstract ... thing
13:36	ataggart	where functions might only care about seqiness
13:37	ataggart	maybe this is just an area for improvement ;)
13:37	Chouser	,^(next (cons 1 '#^{:my :meta} (2 3 4)))
13:37	clojurebot	{:my :meta}
13:37	Chouser	there metadata is kept on the individual cell of the persistentlist
13:37	ataggart	aha! I was lookig for that with-meta reader macro
13:38	Chouser	it's not exactly the same as with-meta
13:38	ataggart	orly?
13:38	Chouser	it applies metadata to the form read by the reader and passed to the compiler
13:39	Chouser	with-meta happens at runtime and applies to the collection at that point
13:40	Chouser	,(meta #^{:my :meta} (conj [1 2] 3))
13:40	clojurebot	nil
13:40	Chouser	,(meta (with-meta (conj [1 2] 3) {:my :meta}))
13:40	clojurebot	{:my :meta}
13:40	ataggart	hm, still not grokking why that's different
13:41	Chouser	in the first example, the form (conj ...) has metadata. Could provide a line number, type hint, etc. In this case, the compiler doesn't know what to do with :my and it gets thrown away
13:41	ataggart	oh is that first one appplying the metadata to the conj, and not the result of the conj?
13:41	Chouser	right
13:41	ataggart	excellent
13:42	Chouser	,(meta #^{:my :meta} [1 2 3])
13:42	clojurebot	{:my :meta}
13:42	ataggart	,^(next (cons 1 '#^{:my :meta} [2 3 4]))
13:42	clojurebot	nil
13:42	Chouser	but in the case of a literal collection like this, the compiler transfers the metadata from the read form onto the resulting collection
13:43	ataggart	,^(next (cons 1 '#^{:my :meta} (seq [2 3 4])))
13:43	clojurebot	{:my :meta}
13:43	Chouser	right! because 'next' on a vector returns a seq -- a different thing
13:44	Chouser	but next on a seq returns a list returns a list, and the metadata is kept per-cell on a list
13:44	Chouser	but next on a list returns a list, and the metadata is kept per-cell on a list
13:46	Chouser	but the mental model I'm defending here doesn't explain pop (or next) on a list, so either there's a bug or I'm just wrong.
13:47	ataggart	letting this all marinade
13:47	Chouser	hm
13:47	Chouser	actually, list in general is a bit confusing
13:48	Chouser	conj on a list preserves metadata, cons does not?
13:48	Chouser	oh, cons on a list returns a Cons not a PersistentList
13:49	danlarkin	+1 for metadata review
13:49	ataggart	,^(conj '#^{:my :meta} '(1 2 3) 4)
13:49	clojurebot	{:my :meta}
13:49	ataggart	,^(butlast '#^{:my :meta} '(1 2 3) )
13:49	clojurebot	nil
13:49	ataggart	I'm guessing butlast has the same issue drop does
13:51	ataggart	this is coming up for me because I''m splitting a url up into its segments via re-split, then adding meta-data to that seq, but the later manipulating that seq
13:51	ataggart	using butlast
13:51	Chouser	I use PersistentLists so rarely for data in Clojure.
13:53	ataggart	this is the problem I have:
13:53	Chouser	gah. pop on vector also loses metadata
13:53	Chouser	that can't be right
13:53	ataggart	^(butlast (with-meta (re-split #"/" "foo/bar/baz") {:my :meta}))
13:54	ataggart	,^(butlast (with-meta (re-split #"/" "foo/bar/baz") {:my :meta}))
13:54	clojurebot	java.lang.Exception: Unable to resolve symbol: re-split in this context
13:54	ataggart	d'oh
13:54	ataggart	well that woudl output nil
13:54	Chouser	yeah, I'm with you
13:54	Chouser	I don't know if butlast or drop-last should preserve metadata or not
13:55	ataggart	so I've been doing stuff like (with-meta (butlast orig-seq) {meta orig-seq))
13:55	Chouser	but I would be shocked if pop losing metadata on vectors is intentional
13:55	ataggart	pop is next, right?
13:55	Chouser	no
13:55	ataggart	oh
13:55	ataggart	ioening the src
13:55	Chouser	it is for lists, but for vectors and queues it pops off the right
13:55	ataggart	*opening
13:55	ataggart	ah yeah
13:55	Chouser	,(pop [1 2 3 4])
13:55	clojurebot	[1 2 3]
13:56	Chouser	so that would be ideal for you case
13:56	Chouser	a thing the naturally grows and shrinks on the right
13:56	Chouser	but is walked left-to-right
13:56	ataggart	,(pop (take 3 (iterate inc 1)))
13:56	clojurebot	java.lang.ClassCastException: clojure.lang.LazySeq cannot be cast to clojure.lang.IPersistentStack
13:56	Chouser	it's a vector. And most modifications to a vector preserve metadata (conj, assoc)
13:57	ataggart	so I'd need to turn the seq from re-split into a vector...
13:57	ataggart	smells funny
13:57	Chouser	nah, that's common
13:57	Chouser	re-split doesn't know what you're going to do. And it can be lazy (though I don't think it is).
13:57	Chouser	so a seq is a natural return type
13:58	Chouser	but you don't care about lazy and want to treat it as a vector. that's exactly what 'vec' is for
13:58	ataggart	hmm fair point
13:58	Chouser	but the only way I know to drop an item off the end of a vector is 'pop'
13:58	Chouser	and that still loses your metadata. :-P
13:58	ataggart	butlast works fine except for the losing of metadata
13:58	ataggart	heh
13:59	Chouser	right, but butlast is O(n) and more importantly an argument could be made that it should not preserve metadata.
13:59	Chouser	maybe. I dunno.
14:00	ataggart	the O(n) is a good point
14:00	ataggart	ya, swithcing to vectors
14:00	Chouser	vec is O(n) but it would happen once and then you can pop/conj in O(1) as needed
14:00	ataggart	regardless of the metadata issue
14:00	ataggart	right
14:01	Chouser	and I'm haveing a very hard time imagining the argument for dropping metadata on vector pop
14:01	ataggart	maybe Rich can clarify the rules for when manipulation functions should be meta-data preserving
14:01	Chouser	unless there's some other fn for dropping off the last item of a vector. I can't think of one.
14:01	ataggart	pop sounds right
14:02	Chouser	huh
14:02	Chouser	oh
14:03	ataggart	this is all coming up because I'm porting (sort-of) a rest framework I originally wrote in java. got tired of just toying with clojure.
14:03	ataggart	REST, not '(rest ) :)
14:04	Chouser	:-)
14:04	Chouser	,^(pop (with-meta (conj [1 2 3] 4) {:my :meta}))
14:04	clojurebot	{:my :meta}
14:04	Chouser	,^(pop (with-meta [1 2 3 4] {:my :meta}))
14:04	clojurebot	nil
14:04	Chouser	it's a bug
14:04	Chouser	,(class (with-meta [1 2 3 4] {:my :meta}))
14:04	clojurebot	clojure.lang.LazilyPersistentVector
14:05	ataggart	ah finally a good excuse to register with assembla
14:05	Chouser	the metadata gets lost in the transition from LazilyPersistentVector to PersistentVector
14:10	ataggart	do we need to send in a CA in order to be able to post tickets in assembla?
14:11	ataggart	well, sicne I'd also want to try writing a patch, I guess so. nvm
14:13	Chousuke	you can use the "support" thing to report a bug I think.
14:14	Chousuke	but you need to be a project member to have full access to the ticket system, and that requires a CA
14:14	ataggart	yup, I'll do the CA thing once I have access to a printer
14:14	ataggart	and stamps
14:21	ataggart	chouser: btw thanks for the help
14:22	Chouser	ataggart: sure. I'd still be pleased to have a high-level description from rhickey about when metadata should be preserved vs. not
14:24	cemerick	shouldn't all objects be able to support metadata in JDK7?
14:24	cemerick	(totally orthogonal to the preservation issue)
14:24	Chouser	cemerick: how? got a link?
14:24	cemerick	Chouser: well, invokedynamic allows for method injection, or so I thought.
14:25	Chouser	huh. like add a method to Integer that ... closes over a variable that points to the metadata?
14:26	technomancy	that's the first I've heard of invokedynamic supporting something like that
14:26	cemerick	right...totally naievely, I'd think with-meta would just swap out another meta method
14:26	technomancy	Rich seemed to imply that it's mostly for supporting dynamic OOP on the JVM
14:27	danlarkin	I think it's to aid in languages where foo.bar might mean "get instance variable bar" or "call instance method var"... or something
14:27	danlarkin	s/var/bar
14:27	cemerick	sorry, the terminology is "interface injection": http://challenge.openjdk.org/projects/mlvm/subprojects.html
14:27	cemerick	not technically part of invokedynamic, but still part of JSR292
14:29	Chouser	seems like a stretch to me
14:29	Chouser	doesn't sound like it would let you change the class of an existing object
14:29	Chouser	hm
14:29	cemerick	Chouser: no need to change the class -- just adding the IMeta interface is enough.
14:30	Chouser	no, you need a place to store the meta value
14:30	cemerick	right, so the method that is injected along with the interface would just be a thunk that closed over the meta value
14:31	Chouser	but now you're talking about a new method per instance
14:31	technomancy	cemerick: that's pretty awesome
14:31	cemerick	shrug Better than not having any metadata on java.lang.Objects at all.
14:31	Chouser	and classes like Integer are final -- could you still inject a method?
14:32	cemerick	that I don't know. I think so, as I don't think there's anything special about a final class (e.g. the final keyword is for javac's benefit...?)
14:33	cemerick	alternatively, there could be a single meta() method that looked up the meta value in a global WeakHashMap (if you were worried about creating a meta method per instance)
14:33	hiredman	cemerick: if that was true then you could just use the asm library to generate your own bytecode that ignores final
14:33	ataggart	yes final is for the compiler, and hotspot
14:34	ataggart	thats why you can change final fields via reflection
14:34	hiredman	final means different things in different places
14:34	cemerick	hiredman: I'm always on shaky ground w.r.t. final and bytecode issues.
14:34	Chouser	you could use WeakHashMap today
14:35	Chouser	just have meta and with-meta fns use that instead of (or in addition to) .meta() and .withMeta() methods.
14:35	cemerick	Chouser: right, but I presume Rich wouldn't put a big honking switch in the meta fn.
14:36	Chouser	I think his resistance would actually be more toward the WeakHashMap -- he's actually got plenty of bit switches.
14:37	cemerick	well, who knows, maybe pushing a new method into an object on every with-meta will end up being cheap :-)
14:37	Chouser	we can hope!
14:38	hiredman	that would be rather down the line I imagine
14:38	Chousuke	http://blogs.sun.com/jrose/entry/interface_injection_in_the_vm reading this, it seems it would be done for classes, not objects.
14:38	hiredman	(since clojure targets java 5)
14:39	cemerick	hiredman: I think we're going to see some conditional execution macros and a lot of bifurcation as soon as invokedynamic is available (perhaps as an option in a JDK 6 update)
14:39	Chouser	Chousuke: anything about field injection there? :-)
14:40	hiredman	cemerick: I remain skeptical
14:40	Chouser	bifurcation of .. the clojure sources?
14:40	cemerick	Chouser: no, just different code paths for invokedynamic
14:40	Chouser	oh
14:42	cemerick	Chouser: John Rose indicates that traits are a core use case for interface injection, so yeah, injecting fields is on the table
14:42	cemerick	e.g. an IMeta trait
14:42	cemerick	(which carries a meta field)
14:43	Chouser	clojure.test confuses me
14:43	Chouser	oh...
14:50	ataggart	is there a way to add doc to a defstruct?
14:56	Chouser	(defstruct #^{:doc "this is foo"} foo :a :b)
14:57	Chousuke	Maybe I should bookmark this JRose blog :)
14:57	Chousuke	fun stuff
14:58	Chousuke	Also, the JVM apparently has an interesting method to check for null pointer derefs.
14:59	eyeris	Is there a standard function in the api for flattening a list?
15:00	Chousuke	basically, it assumes that they don't happen doesn't do any checking (probably not true for some cases, but anyway). Instead, it traps them when they happen, recovers, and recompiles the code at runtime to include a null check :P
15:00	hiredman	cute
15:00	Chousuke	+and
15:00	Chousuke	eyeris: there is flatten in contrib
15:01	eyeris	Thanks
15:03	cemerick	Chousuke: yeah, I've subscribed jrose's blog :-)
15:04	cemerick	Let's hope he stays hale, hearty, and healthy. Quite a positive voice in the jvm circus.
15:05	ataggart	chouser: yup that did it. thanks.
15:06	Chouser	ataggart: so you're planning on filing a bug, sending in a CA, and writing a patch, right?
15:07	ataggart	I woudl but it'll take me a few days to get to a printer. feel free to open it in the meantime.
15:07	Chousuke	hmm
15:08	Chousuke	,(remove coll? (tree-seq coll? seq [[1] {1 2} [2 [5 4]] 3]))
15:08	clojurebot	(1 1 2 2 5 4 3)
15:08	Chouser	ataggart: I wouldn't want reduce your incentive to register and contribute. :-)_
15:08	Chouser	Chousuke: nice!
15:09	ataggart	hehe fair enough, just didn't want to hold you up.
15:11	Chouser	ataggart: you don't need a CA to file an assembla ticket, though. Just use the Support tab
15:11	ataggart	ah excellent. doing it now.
15:14	Chousuke	I have to say, searchable menu entries in OS X Cocoa apps are sheer genius.
15:14	fffej	with the latest release, i'm having a little problem using zip_filter.clj. I get this message "descendants already refers to: #'clojure.core/descendants in namespace" - any ideas? I see in the source it is excluding descendants, so I'm a little confused
15:15	Chouser	fffej: you're doing (use 'clojure.contrib.zip-filter) ?
15:15	Chousuke	I wasn't able to find the menu item for adding an RSS feed in mail but then I remembered the help system, wrote "add rss" into it and it instantly showed me what I wanted <3
15:15	fffej	chouser: I'm doing (ns mynamespace (:use [clojure.contrib.zip-filter]))
15:16	Chouser	Try (:require [clojure.contrib.zip-filter :as zf]), then use zf/descendents or whatever
15:16	cemerick	Chousuke: whoa, I had no idea Mail could consume RSS!
15:16	Chousuke	cemerick: heh
15:17	fffej	chouser: thanks, that works
15:17	texodus	How can I determine if a something is an array?
15:18	Chouser	texodus: are you sure you want to be using an array? If you're trying to determine membership, usually a set or map works better.
15:18	texodus	yes, im positive - im writing a specific Writer proxy that needs to dispatch differently on byte arrays
15:19	ozzilee	Chouser: Submitted a patch for lazy-xml. Pretty simple, unless I missed something.
15:19	texodus	so, proxy to a defmulti - but then how to determine an array ...
15:19	texodus	,(class (to-array [1 2 3]))
15:19	clojurebot	[Ljava.lang.Object;
15:20	texodus	,(= (class (to-array [1 2 3])) [Ljava.lang.Object)
15:20	clojurebot	Unmatched delimiter: )
15:20	texodus	clearly doesnt work
15:20	ataggart	,(Class/isArray (class (to-array [1 2 3])))
15:20	clojurebot	java.lang.IllegalArgumentException: No matching method: isArray
15:20	Chouser	,(some #{(byte 2)} (into-array Byte/TYPE (map byte [1 2 3])))
15:20	clojurebot	2
15:20	ataggart	hmm
15:21	ataggart	,(.isArray (class (to-array [1 2 3])))
15:21	clojurebot	true
15:21	hiredman	!
15:21	clojurebot	CLABANGO!
15:21	hiredman	beat me to it
15:21	ataggart	lol
15:21	texodus	,(. (class (to-array [1 2 3])) isArray)
15:21	clojurebot	true
15:21	texodus	bingo
15:21	texodus	thanks alot
15:22	Chouser	wow, I completely misread the question
15:22	eyeris	I am taking a first stab at writing a macro. I have a seq of bindings I want, but (let) wants a vector. How can I create a vector from that seq?
15:23	hiredman	vec
15:23	Chouser	I had an extra "in" in there. "if a something is in an array"
15:23	ataggart	,(vec (seq [1 2 3[))
15:23	clojurebot	Unmatched delimiter: )
15:23	ataggart	,(vec (seq [1 2 3]))
15:23	clojurebot	[1 2 3]
15:23	hiredman	or just pass a vector
15:23	hiredman	like most other bindings
15:25	Chouser	ozzilee: thanks! BTW, issues have moved to assembla, and code to github. See http://clojure.org/patches
15:26	ozzilee	Chouser: Oh, contrib too? Neato.
15:26	Chouser	yep. We're trying to create as long a list of out-of-date project hosting services as possible
15:26	ozzilee	Chouser: Heh no kidding :-)
15:27	eyeris	Is %1 valid inside (fn) as it is inside #()?
15:27	Chouser	eyeris: no
15:27	Chousuke	,((fn [] %1))
15:27	clojurebot	arg literal not in #()
15:27	Chousuke	,((fn [] %))
15:27	clojurebot	arg literal not in #()
15:27	Chouser	although the error case has changed a bit recently
15:28	Chousuke	hmm
15:28	Chousuke	old clojure? :P
15:28	Chousuke	should simply be an unresolved symbol.
15:28	Chouser	right, to support :post
15:28	eyeris	Is there a compliment for (keyword)?
15:29	hiredman	:post?
15:29	hiredman	eyeris: name
15:30	ataggart	,clojure-version
15:30	clojurebot	{:interim true, :major 1, :minor 1, :incremental 0, :qualifier "alpha"}
15:30	Chouser	hiredman: ((fn [] {:post [(= % 5)]} 6)) ==> java.lang.Exception: Assert failed: (= % 5)
15:30	Chouser	wow, that's a whole lot of parens
15:31	hiredman	Chouser: oooooo
15:31	Chouser	and various other grouping chars
15:31	hiredman	no :pre yet?
15:31	Chouser	:pre is there too, but doesn't need %
15:31	hiredman	ah
15:38	eyeris	Is :as not allowed in the defn arg list?
15:38	Chouser	right.
15:40	Chouser	,(let [foo (fn [& [a b :as all]] [a b all])] (foo 1 2 3))
15:40	clojurebot	[1 2 (1 2 3)]
15:40	Chouser	but you can do that, I suppose
15:41	eyeris	Yeah, I've been using it in let/
15:42	Chouser	sure, but you see I'm doing it in fn args there, just destructuring the "rest" part
15:42	eyeris	Oh, you mean I could do (def foo (fn... and use the destructuring there
15:42	Chouser	or just (defn foo [& [a b :as all]] [a b all])
15:45	eyeris	I see. I tried that, but I was doing it wrong :/
15:45	Chouser	nested destructuring can get a bit convoluted
15:46	eyeris	I am using it with a nested map, so I was trying (defn foo [& {{:keys [a b]} :sub-map} :as my-map] ...)
15:46	eyeris	But it shoudl be (defn foo [& {{:keys [a b]} :sub-map :as my-map}] ...)
15:50	zakwilson	I'm trying to decide between Haskell and Clojure for a project. I know Clojure better, but Haskell seems like it will make FRP easier and I intend to make fairly heavy use of that.
16:04	ataggart	chouser: what clojure implementation code were you looking at when we were talking about pop and meta-data?
16:04	ataggart	I haven't quite grokked how to get from clj function defs in clojure.core over to clojure java
16:05	hiredman	ataggart: when you see a java method call
16:05	ataggart	I'm referring to clojure's java implementation
16:05	ataggart	e.g., where pop, conj, etc are implemented
16:06	hiredman	yeah
16:06	hiredman	~def pop
16:06	hiredman	so you look at the pop method in the RT class
16:07	hiredman	~def c.l.RT
16:07	ataggart	yes, but chris was referring to something earlier andni'm trying to find out exactly what he was looking at
16:08	ataggart	in order to write a bugfix patch
16:08	hiredman	oh
16:08	hiredman	pardon
16:08	ataggart	np, I should have been more clear
16:10	guille_	hi there
16:18	ataggart	how does one add further comments on a ticket in assembla?
16:19	ataggart	hmm must be because I'm not CA'd yet
16:20	guille_	if i'm implementing a interface in a single file, should I use the last name of the namespace (using same convention as java), that's the interface name, or the whole package name in :name too?
16:24	ataggart	chouser: I figured it out. pop is meta-data preserving, but it's just that we see the meta-data associated with that sublist, which in the example is nil.
16:25	ataggart	,(meta (pop (conj (with-meta '(1 2 3) {:my :meta}) 4)))
16:25	clojurebot	{:my :meta}
16:27	ataggart	,(meta (pop (with-meta '(1 2 3) {:my :meta})))
16:27	clojurebot	nil
16:27	ataggart	because (meta '(1 2)) is nil
16:28	ataggart	so ticket #149 should probably be marked as invalid
16:42	eyeris	Will someone here help me figure out where I am going wrong with this macro? http://pastebin.ca/1491100
16:44	eyeris	If I load it into the repl and use macroexpand, it just prints out the input, without any expansion.
16:44	eyeris	That must be a sign that I've used defmacro incorrectly, but I can't figure out how
16:46	Chousuke	hm
16:47	eyeris	The (one) and (two) in the example usage are incorrect, but that isn't what the exception refers to.
16:49	Chousuke	you can't pass runtime values to a macro
16:49	Chousuke	you need to call it as (with-map-bindings {:one 1 :two 2} (+ one two))
16:50	eyeris	I see.
16:51	eyeris	Is there a way to pass a runtime vector to let instead of a literal vector, so that I could write that macro as a fn?
16:51	Chousuke	a macro is a function that takes clojure code as its parameters, and does not evaluate them; so, in your example, you're passing it a plain symbol :)
16:51	Chousuke	that would require eval I think.
16:52	Chousuke	but... are you sure you need to do that at all?
16:52	eyeris	I don't need to, but I'd sure like to.
16:52	eyeris	I have a map that has 6 keys, my function uses all of them.
16:52	Chousuke	in which case would you need to pass a binding vector to let at runtime?
16:53	eyeris	I could use :keys [...], but that just feels overly verbose when I am binding them all
16:53	Chousuke	hmm
16:53	eyeris	It also is misleading to me, because when I see :keys [...] I instinctively look at what isn't bound from the map
16:54	Chousuke	will simple-map be static?
16:54	eyeris	No
16:54	Chousuke	hm.
16:55	Chousuke	I still don't get your use case though. :/
16:56	Chousuke	you can't bind locals based on runtime values, unless you also generate the code at runtime and use eval
16:56	Chousuke	and that's not what you want to do.
16:57	eyeris	Oh, wait.
16:57	eyeris	The keys in my map will be static.
16:57	eyeris	The values will change though.
16:57	Chousuke	in that case there's no problem
16:59	eyeris	I just have to use this map in a lot of functions, so I'd like to write a function or macro that binds the keys to local symbols so that I don't have to write out the destructuring for every function
17:00	Chousuke	just write (defmacro defspecialfn [& body] `(let [{:keys ~'[your keys here]} the-global-map] ~@body))
17:00	Chousuke	assuming the-global-map will be bound with (binding ...) here.
17:01	Chousuke	otherwise, it would be static and the values would not be changing :)
17:02	eyeris	Okay
17:02	eyeris	I think I understand that.
17:02	Chousuke	alternatively, it could be a ref
17:02	Chousuke	in which case you'd want to do:
17:02	Chousuke	just write (defmacro defspecialfn [& body] `(let [{:keys ~'[your keys here]} (deref the-global-map)] ~@body))
17:03	Chousuke	forgot to remove the just write.
17:04	Chousuke	note the funky ~' unquote thing there. that needs to be used so that let won't complain about binding to qualified names.
17:04	Chousuke	it also serves as documentation to the reader that the macro introduces local names implicitly.
17:05	eyeris	I think this is one step removed from what I want to do.
17:05	eyeris	The map I am using is always passed into the function. I don't rebind the global symbol to update the values.
17:07	Chousuke	well, in that case you can add the map name as a parameter to the defspecialfn macro
17:08	eyeris	And, inside the function, that would be a local symbol just like a normal function parameter?
17:08	Chousuke	(defmacro defspecialblock [mname & body] `(let [{:keys ~'[your keys here]} ~mname] ~@body))
17:08	Chousuke	then (defspecialblock global-map (do-stuff using your keys))
17:08	eyeris	http://pastebin.ca/1491131
17:09	eyeris	That is what I am doing
17:09	eyeris	(dispatcher) returns the next function to pass the state map into
17:10	eyeris	that function returns the next map to bind to the state symbol
17:10	eyeris	repeat
17:11	Chousuke	investigate the macroexpansions of these functions, you'll see what kind of code they produce :)
17:16	eyeris	Where does the name of the special block go?
17:16	Chousuke	well, that above thing doesn't actually define a function.
17:16	Chousuke	just a let block
17:16	Chousuke	but you could easily extend it to a defn
17:19	eyeris	This still doesn't make sense, because I can't call it from inside my state loop, because I don't have a symbol for the map outside the state loop
17:20	eyeris	can't call it = can't call the special fn defined by the macro
17:20	Chousuke	it doesn't define a fn :P
17:21	eyeris	if does if I make if (defspecialfn [fname mname & body] (defn fname `(let ...
17:22	eyeris	Wait a minute
17:22	eyeris	I think I got something
17:23	Chousuke	the ` should go before defn, not let.
17:25	Chousuke	http://pastebin.ca/1491144
17:25	Chousuke	I used [] around map-name because giving the binding in a vector to the macro is more idiomatic
17:30	eyeris	I want the map symbol inside the generated fn to always be the same
17:31	eyeris	Removing map-name from the defmacro parameters and then removing the ~ from before each use of it yields an exception though
17:32	Chousuke	well, of course.
17:32	eyeris	Why?
17:32	Chousuke	because it's passed to defn as the parameter name
17:32	Chousuke	and inside ` it becomes qualified, and qualified names can't be used to bind to.
17:33	Chousuke	if you want to force the name, do ~'whatever
17:34	eyeris	This whole "the macro language is the language" is bs
17:34	Chousuke	hm?
17:34	Chousuke	no it's not.
17:34	Chousuke	a macro must return a data structure.
17:34	Chousuke	which is clojure code.
17:34	eyeris	but it must do so using different evaluation rules
17:35	Chousuke	not really. the only differing thing is that macro parameters are not evaluated
17:35	Chousuke	otherwise a macro is a normal function
17:35	Chousuke	it may look different because of `(foo bar whatever
17:36	Chousuke	but syntax-quote is not restricted to macros
17:36	Chousuke	,(let [x 1] `(foo bar x ~x))
17:36	clojurebot	(sandbox/foo sandbox/bar sandbox/x 1)
17:37	Chousuke	you could write every macro without syntax quoting
17:38	eyeris	http://pastebin.ca/1491155
17:38	eyeris	Why is that wrong?
17:38	Chousuke	(defmacro [x] (list 'clojure.core/+ x x)) <- equivalent to (defmacro [x] `(+ ~x ~x))
17:39	hiredman	eyeris: map-name is not specified anywhere
17:39	eyeris	Right, so I have to use 'map-name
17:40	Chousuke	~'map-name
17:40	clojurebot	map is LAZY
17:40	Chousuke	:P
17:40	eyeris	Chousuke so what does ~'map-name evaluate to?
17:40	Chousuke	if you do 'map-name, you will see (quote user/map-name) in the expansion
17:40	hiredman	eyeris: 'map-name
17:40	Chousuke	eyeris: just the symbol map-name
17:40	hiredman	~ is unsyntax quote and ' is quote
17:40	clojurebot	Alles klar
17:41	hiredman	erp
17:41	Chousuke	eyeris: in other lisps, you'd just write "map-name" without the unquote-quote thing.
17:41	Chousuke	eyeris: but in clojure, map-name will be qualified in that case, and you can't bind to qualified names
17:41	ataggart	eyeris: are you just trying to reimplement (binding [...] ...)?
17:42	Chousuke	to get around this, you unquote a quoted, plain symbol. thus the result is the plain symbol
17:43	Chousuke	compare:
17:43	Chousuke	,`sym
17:43	clojurebot	sandbox/sym
17:43	Chousuke	,`~sym
17:43	clojurebot	java.lang.Exception: Unable to resolve symbol: sym in this context
17:43	Chousuke	,`~'sym
17:43	clojurebot	sym
17:43	eyeris	I get that part
17:44	eyeris	but you syntax quote the macro to prevent evaluation, right?
17:44	Chousuke	no
17:44	Chousuke	the evaluation is already prevented by the macro expansion.
17:44	hiredman	Chousuke: you kinda do
17:45	Chousuke	well, hm.
17:46	hiredman	you quote the form the form you want to be the result of the macro
17:46	hiredman	and you use syntax quote because you want all its snazzy features
17:47	eyeris	Yeah, but that form is unevaluated until it is called
17:47	hiredman	…
17:47	eyeris	e.g. the defn isn't evaluated before defmacro returns
17:47	Chousuke	eyeris: no.
17:48	Chousuke	well, in a way yes.
17:48	Chousuke	but there's a more fundamental point.
17:48	Chousuke	eyeris: the macro, when called, returns the defn form
17:48	Chousuke	as a value
17:48	Chousuke	and then the compiler calls eval on it.
17:48	eyeris	Right
17:49	eyeris	Normally, to prevent the evaluation of a symbol, you quote it, right?
17:49	eyeris	(let [x 1] 'x) returns the symbol x
17:49	eyeris	but (let [x 1] x) returns 1
17:50	eyeris	s/returns/evaluates to/
17:50	Chousuke	you can also try the following in a repl: (defn f [x] (println x) x) (defmacro m [x] (println x) x) and try (f (+ 1 1)) (m (+ 1 1))
17:51	Chousuke	the former will print 2 and return 2, the latter will print (+ 1 1) and return 2
17:51	Chousuke	(because of the implicit eval)
17:52	eyeris	Right, but inside defmacro, x is not resolved when it is an argument to println
17:52	Chousuke	after that, you can try using syntax-quote on x in both the macro and the function
17:52	eyeris	whereas outside defmacro, it would be
17:53	Chousuke	yes it is resolved.
17:53	Chousuke	it's just that its value is '(+ 1 1)
17:53	Chousuke	not 2, like in the function
17:53	eyeris	erm
17:54	eyeris	I misspoke
17:54	eyeris	it is not evaluated
17:55	Chousuke	yes.
17:56	Chousuke	or to make it painfully clear, x is evaluated inside the macro, but when you pass in arguments to the macro, the arguments are not evaluated, and thus the value of x becomes the literal list '(+ 1 1)
17:57	Chousuke	in a similar way, the local x is evaluated inside a function, but also the arguments you pass to the function are evaluated; and (+ 1 1) evaluates to 2, so the value of x becomes 2
17:58	eyeris	Okay
17:59	eyeris	So
17:59	eyeris	So 'x in a fn evals to the symbols x.
18:00	Chousuke	yes.
18:00	eyeris	So what does 'x eval to inside a macro?
18:00	Chousuke	the same as in a function.
18:00	Chousuke	to the symbol x
18:01	eyeris	So the syntax-quote function is doing the namespace resolution?
18:01	Chousuke	yes.
18:01	eyeris	Why? What problem does that solve?
18:01	Chousuke	hygiene
18:02	Chousuke	it's more difficult to accidentally introduce names
18:02	Chousuke	that could shadow names in other code.
18:02	Chousuke	very useful for macros, which is why ` is very popular when writing them
18:02	Chousuke	but ` is not restricted to macros
18:03	Chousuke	,((fn [x] `(x :is ~x :unqualified ~'x)) 5)
18:03	clojurebot	(sandbox/x :is 5 :unqualified x)
18:04	eyeris	So I can write macros without `?
18:04	Chousuke	sure.
18:05	Chousuke	(defmacro add-to-y [x] (list '+ x 'y))
18:05	Chousuke	next (def y 3); (add-to-y 4) -> 7
18:06	Chousuke	because the macro will return '(+ 4 y) which is 7 when evaluated in that context
18:09	Chousuke	the problem comes when you do something like : (defmacro eval-with-y [code] (list 'let '[y 5] code)); (def y 1) (eval-with-y (+ y 10)) -> 15
18:10	Chousuke	however, if you do (list 'let `[y 5] ...), you will get an exception
18:10	Chousuke	because y is qualified and let does not like that.
18:11	Chousuke	so you won't mistakenly override names in the context.
18:12	eyeris	Is there a contrib that function as syntax quote without the ns resolution?
18:12	Chousuke	no.
18:13	Chousuke	but as show, you can override syntax-quote's ns resolution by prefixing the symbol you want to be bare with ~'
18:13	eyeris	I get that.
18:13	eyeris	That is just plain not intuitive for me.
18:14	Chousuke	well, it's not supposed to be :)
18:14	eyeris	' = quote
18:14	eyeris	~ = unquote
18:14	clojurebot	Excuse me?
18:14	Chousuke	yeah
18:14	eyeris	reversing something should yield the same thing
18:14	eyeris	erm
18:14	Chousuke	and when you quote a symbol, and unquote it, you get the symbol back.
18:14	eyeris	(inc (dec 1)) == 1
18:15	Chousuke	but see
18:15	Chousuke	,`'y
18:15	clojurebot	(quote sandbox/y)
18:15	eyeris	Right, but if ~'x evaluates to x, why doesn't it get resolved?
18:15	Chousuke	hmm
18:15	Chousuke	,`~'y
18:15	clojurebot	y
18:15	eyeris	See
18:15	eyeris	it's different
18:16	Chousuke	,`~'sandbox/y
18:16	clojurebot	sandbox/y
18:16	Chousuke	(quote sandbox/y)
18:16	Chousuke	,(quote sandbox/y)
18:16	clojurebot	sandbox/y
18:17	Chousuke	well, you can just take this behaviour for granted now :)
18:17	Chousuke	it's useful.
18:18	Chousuke	it is a bit funky though.
18:18	clojurebot	0
18:19	Chousuke	but you can think of it as unquoting a quoted form yielding the same form.
18:20	eyeris	It seems to me that syntax-quote must do resolution before any evaluation and simply doesn't recurse into an unquote form
18:21	Chousuke	possibly.
18:21	Chousuke	syntax-quote is a special thing implemented in java, after all
18:21	Chousuke	it has some special semantics.
18:22	Chousuke	but it's undeniably useful, and not at all that difficult after you understand how it works.
18:22	eyeris	For Symbols, syntax-quote resolves the symbol in the current context, yielding a fully-qualified symbol (i.e. namespace/name or fully.qualified.Classname).
18:23	eyeris	That should be followed by "except when preceded by ~'" or something more complete.
18:24	eyeris	Thanks for your help
18:24	eyeris	I'll need a lot more practice, but it's a start
18:28	ataggart	rule 1 of macros: don't write macros.
18:29	Chouser	such a good rule.
19:09	ataggart	chouser: did you see my comments from earlier regarding pop?
19:17	Chouser	I hadn't. But yes, I guess I was failing to describe it.
19:17	Chouser	glad you figured it out anyway, though.
20:34	newbie	hi, i am iterating pixels on the screen using 2 doseq loops one for x one for y then i cam doing some calculations on the pixels i end up with some valid pixels how can i return a vector of pixels?
20:35	hiredman	newbie: use for
20:37	newbie	but how can i create a vector and add on it like in loop?
20:38	hiredman	vectors can be created with the literal [] or the function vec, or vector
20:40	newbie	yeah but vector is not persisted between iterations so i can add on it
20:41	hiredman	if you are looping, you can rebind names in the loop
20:42	hiredman	,(loop [foo 0 r []] (if (= foo 5) r (recur (inc foo) (conj r foo))))
20:42	clojurebot	[0 1 2 3 4]
20:42	newbie	no thats why i am asking is there a way to achive the same effect while using doseq
20:42	hiredman	doseq is for side effects
20:44	newbie	so loop inside loop is my only choice? to iterate x,y coordinates?
20:45	hiredman	for
20:46	hiredman	,(vec (for [x (range 10) y (range 10)] [x y]))
20:46	clojurebot	[[0 0] [0 1] [0 2] [0 3] [0 4] [0 5] [0 6] [0 7] [0 8] [0 9] [1 0] [1 1] [1 2] [1 3] [1 4] [1 5] [1 6] [1 7] [1 8] [1 9] [2 0] [2 1] [2 2] [2 3] [2 4] [2 5] [2 6] [2 7] [2 8] [2 9] [3 0] [3 1] [3 2] [3 3] [3 4] [3 5] [3 6] [3 7] [3 8] [3 9] [4 0] [4 1] [4 2] [4 3] [4 4] [4 5] [4 6] [4 7] [4 8] [4 9] [5 0] [5 1] [5 2] [5 3] [5 4] [5 5] [5 6] [5 7] [5 8] [5 9] [6 0] [6 1] [6 2] [6 3] [6 4] [6 5] [6 6] [6 7] [6 8] [6 9] [7 0
20:56	newbie	thanks that solved it.
21:11	Raynes	hiredman: Why is :state necessary? The only times I've seen it used are by clojure-generated classes that extend Java classes. http://www.fatvat.co.uk/2009/05/clojure-and-robocode.html being one example. It looks to me like :state could be replaced with any ol' ref.
21:11	hiredman	Raynes: it generally is a ref
21:12	Raynes	I know, but what is the point of it? I'm not really sure why it's needed.
21:12	hiredman	the key point is the is a state per instance
21:13	Raynes	So it's primarily for Clojure classes that will be instantiated by Java classes?
21:13	Raynes	Well, clojure-generated classes.
21:13	hiredman	yes
21:14	Raynes	I've noticed that Stuart's example in Programming Clojure instantiated a Clojure-generated class inside the same Clojure-generated class. That blew my mind for a few minutes. :p
21:14	Raynes	Thanks.
21:37	Knekk	I am looking for something new to read, related to CS/Programming, in case anyone wants to recommend something they've found enlightening or interesting.
21:37	Knekk	thanks
21:42	gnuvince	Knekk: SICP?
21:42	Knekk	http://mitpress.mit.edu/sicp/?
21:43	gnuvince	Aye
21:43	hiredman	clojurebot: delicious
21:43	clojurebot	delicious is http://delicious.com/clojurebot
21:43	Knekk	nice, didn't know about it. Thanks :)
21:44	gnuvince	Knekk: as far as CS books go, this one is way high up there.
21:44	Knekk	$114 at amazon, free on the site. Yay
21:44	Knekk	hiredman: thanks
23:43	flying_spaghetti	anyone can help with a noob question?
23:46	fUD	Greets
23:47	flying_spaghetti	hail
23:50	flying_spaghetti	i seek a clue
23:52	Chouser	go
23:54	flying_spaghetti	well, i am learning clojure, writing a raytracer.. i want to know, is there an easy way to get a list representation of the content of a struct or vise versa
23:54	flying_spaghetti	because i have, e.g. colour structures with members r, g, b, and sometimes it would be nice to apply the same operation to all of them using map
23:55	flying_spaghetti	and then convert the resulting list back into a colour structure
23:55	flying_spaghetti	i am wondering if there is a generic means to do that
23:56	texodus	,(into {} (map (fn [[_ v]] (+ 1 v)) [1 2 3])
23:56	clojurebot	EOF while reading
23:57	texodus	,(into {} (map (fn [[_ v]] (+ 1 v)) [1 2 3]))
23:57	clojurebot	java.lang.RuntimeException: java.lang.UnsupportedOperationException: nth not supported on this type: Integer
23:58	flying_spaghetti	oh great thx i will try
23:58	texodus	oh right, struct
23:58	texodus	(into {} (map (fn [[_ v]] (+ 1 v)) {:1 1 :2 2 :3 3})
23:58	texodus	,(into {} (map (fn [[_ v]] (+ 1 v)) {:1 1 :2 2 :3 3})
23:58	clojurebot	EOF while reading
23:59	texodus	,(into {} (map (fn [[_ v]] (+ 1 v)) {:1 1 :2 2 :3 3}))
23:59	clojurebot	java.lang.IllegalArgumentException: Don't know how to create ISeq from: Integer