#clojure – 2012-06-08

2012-06-08

00:17	johnmn3	are there any clj libs that make quick work of compressing and decrompressing zip files?
00:18	hiredman	johnmn3: the jvm comes with the java.util.zip package
00:32	johnmn3	hiredman: yea, i've been reading the docs on it. Just seems like it could use a wrapper in clojure
00:58	muhoo	wow, a week of work, for 140 lines of code (google apps step2 auth workflow for friend)
00:58	muhoo	heh, of which 40 is just the (:import ...) block
00:59	brehaut	holy crap!
00:59	cemerick	for want of an import * ;-(
00:59	cemerick	er, ;-)
01:01	muhoo	should i make a github project for basically 1 file with 140 loc?
01:01	cemerick	been done for much, much less
01:02	muhoo	ok will do. cemerick: had a chance to look at that :session dissoc patch?
01:03	muhoo	i want to deploy an app using friend soon, early next week, and i'd like not to have to cook up my own friend jar
01:04	cemerick	muhoo: sorry, not yet. I killed my open source budget earlier in the week. But: why are you so keyed in on getting that slot dissoc'ed?
01:06	muhoo	paranoia, really. i don't like leaving java objects hanging around in session memory, or in couchdb if i use that backend
01:06	muhoo	when it comes to security stuff, i get really anal-retentive.
01:06	cemerick	ah
01:06	cemerick	well, that particular object is really innocuous
01:07	cemerick	I'm pretty anal myself in wanting to get that sort of thing cleaned up, but I want to make sure the solution is general enough to be applicable to any other post-workflow cleanup, etc.
01:07	muhoo	true. i'll have another look at it myself then.
01:08	cemerick	muhoo: Thanks for bulling through this step2 workflow; I think yours will be the first third-party friend workflow. :-D
01:09	muhoo	cool, i hope someone may find it useful.
01:09	muhoo	other than me, of course.
01:10	muhoo	is there a wiki for friend i could link it from, so that it could get found?
01:11	cemerick	I'll add all workflows to the main readme; the openid one will be broken out onto its own shortly, so the table will have at least two rows. ;-)
01:12	cemerick	muhoo: I'd like to eventually have something like: https://github.com/bnoguchi/everyauth#readme
01:13	cemerick	of course, with actual links to the implementing libraries, etc.
01:14	muhoo	btw, this step2 thing supposedly does oauth2
01:15	muhoo	i like the idea of having an open source budget
01:15	cemerick	yeah, I saw that claim; I don't know what to make of it at all (i.e. openid + oauth = huh?)
01:16	cemerick	It's a new policy experiment on my part. I've been overcommitted w.r.t. open source projects in the past year or so. I'm refocusing business-wise, so certain things need to get trimmed back, at least nominally.
01:17	muhoo	sounds fair. you've done a lot.
01:17	muhoo	right now, when i need money, i go looking for work, then when i have enough, i study or do open source stuff, then when it looks like i might run out of money, or if someone calls me with stuff, i work some more :-)
01:18	cemerick	heh
01:19	cemerick	well, I don't do contract work, so I need to theoretically be unreasonably strict about time expenditures
01:19	muhoo	really? i thought you were a contract guy
01:19	cemerick	I've considered seeing what I could put together re: gigs, but it's perpetually uninteresting to me.
01:19	cemerick	muhoo: Nah; 100% product since ~2004: http://snowtide.com
01:19	muhoo	fantastic!
01:20	muhoo	that's where i want to go next: some kind of cloud service thing where i can get people to pay me $20/mo or something each.
01:20	cemerick	That's a lie actually; more like ~2006, but everything between 2004 - 2006 was product-related (tweaks, value-add, etc).
01:22	muhoo	the key to wealth is to make money while you are sleeping :-)
01:23	cemerick	The wealthy people aren't hanging in irc while grinding away at the next idea. :-)
01:23	cemerick	But, yeah.
01:29	muhoo	ah, what to name this thing. i refuse to name it clj-step2
01:31	cemerick	may I suggest friend-step2? Or perhaps that's too presumptuous. :-)
01:32	muhoo	QQS! if github will let me do that
01:32	muhoo	that's the texas 2-step: quick-quick-slow
01:32	cemerick	never would have figured that one :-P
01:33	muhoo	i lived a few years in dallas, learned some things.
01:35	LoganLK	will a multiple core processor give a benefit (over a single core) to a large amount of futures running?
01:37	muhoo	well i could have named it bultitude but that's already taken
01:38	cemerick	LoganLK: yes, as long as each future is doing enough work to warrant the coordination overhead
01:38	LoganLK	the future is just a web request.. but i am sending out a lot of them
01:39	cemerick	yup, perfect
01:39	brehaut	IO counts as work
01:40	cemerick	LoganLK: you may want to create N * 4 agents, and send-off those requests instead of eagerly creating futures for your entire work queue, etc.
01:45	LoganLK	cemerick: i'll look into that. thanks. I'm relatively new to Clojure. I watched your screencast then bought your book. Loving it so far :)
01:45	cemerick	aw, thanks :-)
01:46	cemerick	LoganLK: There's an example in the concurrency chapter using agents to efficiently saturate CPU and network with web requests.
02:10	seanc	anyone tried to install lein-newnew lately? i just tried lein plugin install lein-newew 0.3.1 and got an exception about the ZIP file being empty...
02:21	seanc	went back to lein-newnew 0.2.2 and it worked, just fyi
02:33	cljnewb01234	so I own a copy of "the art of the metaobject protocol:"
02:33	cljnewb01234	now, I know that this book is important
02:34	cljnewb01234	what I don't understand is -- what am I going to learn by reading it?
02:34	cljnewb01234	(it looks litke the type of book that will take time, and I can't just skim)
02:35	muhoo	cemerick: http://github.com/kenrestivo/qqs
02:36	cemerick	muhoo: fabulous, will tweet tomorrow
04:30	kral	namaste
04:31	edoloughlin	kral: fáilte
04:55	bobry	hrm, I have a NullPointerException, resulting from an '(assert ...') in a cljs macro, is this a cljsbuild bug?
04:56	clgv	bobry: hard to tell without the code ;)
05:10	gfredericks	I get a maven error when depending on domina :/
05:10	gfredericks	it says it can't find the clojurescript jar
05:25	clgv	gfredericks: this are the available versions on clojars: https://clojars.org/domina/versions
05:30	gfredericks	clgv: yeah, I'm using 1.0.0-beta4
05:30	clgv	gfredericks: well that's strange then
05:32	gfredericks	indeed
05:32	gfredericks	computers: they don't work
06:11	muhoo	hmm, doesn't appear to be any way to force lein to grab the latest snapshot version of a jar, if there's no way to get access to the .m2/repository directly, i.e. if on heroku
06:13	clgv	muhoo: maybe you can use lein2 as a lib to get the resolve+download-from-clojars part for free
06:13	clgv	or is that maven already?
06:14	muhoo	i think it's a maven-ism
06:15	muhoo	note to self: snapshot means DAILY snapshot. no less.
06:15	clgv	oh. only dates are compared?
06:16	muhoo	only the day portion, not the time, it appears
06:16	clgv	ah kk
06:25	ejackson	huh, that's interesting
07:18	borkdude	did they really expect static typing would -improve- devtime? http://www.cs.washington.edu/education/courses/cse590n/10au/hanenberg-oopsla2010.pdf
07:27	ordnungswidrig	borkdude: i think that consequent static typing like in haskell or ocaml attracts better programmers that build better software. Which they would do in a dynamicaly typed language as well.
07:32	borkdude	Raynes yes, release a new version (does one get notified of closed issues on github)?
07:33	diso	Hi, in a REPL started with clojure-jack-in using (.println System/out "Hello, world!") returns nil but doe'snt print anything, whilst (println "Hello, world!") works as expected. Can somebody reproduce this?
07:33	borkdude	diso maybe the reply doesn't use System.out but some other stream?
07:34	borkdude	repl
07:35	ordnungswidrig	diso: slime/swank binds out to a different stream thatn System/out. This is to send the output to the slime repl.
07:36	ordnungswidrig	you can look in the buffer swank where System/out should go
07:36	diso	it uses out, yes but in a repl startet with java -cp clojure.jar clojure.main both statements print to the repl
07:36	ordnungswidrig	diso: swank rebinds out
07:36	diso	ah okay i will look there thank you.
07:37	ordnungswidrig	diso: evaluate out in the repl and you will see
07:37	diso	ordnungswidrig: sorry was typing and not looking to your answer
07:38	borkdude	,(identical? System/out out)
07:38	clojurebot	false
07:39	diso	this will help debugging when using java-interop, since calls from the java lib i was using can then say something to me ;-)
07:46	diso	ordnungswidrig: after issuing clojure-jack-in i don't have a buffer called swank, did you mean the repl-buffer slime-repl nil?
07:48	ordnungswidrig	no, it is called swank but I just notices that stdout doesn't go there any more.
07:49	diso	so, searching for stdout output
07:50	clojurebot	Gabh mo leithscéal?
08:46	diso	ordnungswidrig: I'have found the following statement: "in swank, System/out is connected to the terminal of the swank server", so when i start the swank manually with lein swank and connect via M-x slime-connect then I can see my System.out.println output there.
08:46	ordnungswidrig	diso: hmm, i didn't know that.
08:47	ordnungswidrig	diso: jack-in behaves differently
08:48	diso	ordnungswidrig: another statement confirms, that it should go to the swank buffer
08:48	diso	ordnungswidrig: so i'll have to dig deeper
09:44	borkdude	"However, considering the plethora of JVM languages out there, Clojure’s quick capture of even 1% of the JVM developers is somewhat impressive." http://files.zeroturnaround.com/developer-productivity-report/zeroturnaround-developer-productivity-report-2012.pdf
10:01	jedmtnman1	hey, anyone have a link to that heatmap of keyboard characters in clojure.core? google is failing me.
10:01	jedmtnman1	or, i am probably failing google
10:03	TimMc	I'd blame it on Google. Their search stinks these days.
10:03	TimMc	Or rather, it has gotten very patchy.
10:12	sergey	do I get it right that it's better not to write recursive functions in clojure because there's no tail recursion optimisation in JVM?
10:13	sergey	and always replace them with either loop/recur or lazy-seq?
10:13	S11001001	there are other reasons to be recursive than list building :)
10:13	S11001001	but for those, better yet, the lazy sequence functions
10:14	Iceland_J	And for list building you rarely -need- to use explicit recursion I feel
10:14	sergey	for example I have a function which takes as an argument the number of times of its execution I call it recursively with the decremented argument each time, is that ok?
10:15	Iceland_J	You could zip it together with [n-1 n-2 ... 0]
10:16	S11001001	sergey: perhaps (doseq [_ (range n)] (f ...)) for that; or for functional functions, nth and iterate
10:17	borkdude	recur
10:17	S11001001	loop/recur is the worst
10:17	borkdude	why?
10:17	clojurebot	borkdude: because you can't handle the truth!
10:17	clgv	why???
10:17	lazybot	clgv: Oh, absolutely.
10:18	clgv	;)
10:18	borkdude	S11001001 what is wrong with loop/recur apart from the fact that you don't often need them because of higher order functions already present?
10:18	S11001001	it's anaphoric and gets introduced all the time. Imagine writing all your macros using (if-let [~'it ...] ...).
10:20	borkdude	S11001001 do you mean it's too "low level"?
10:20	S11001001	maybe if this problem hadn't been encountered and fixed before in other lisps it would be forgivable
10:21	S11001001	in short, recur should require a symbolic label which refers to a lexical recur-target
10:23	clgv	S11001001: you are arguing for a goto? :O
10:23	Hodapp	goto and functional programming... hmmmm
10:23	S11001001	clgv: the only difference is that the current recur is a goto (wherever; you pick one, compiler)
10:24	borkdude	S11001001 you mean like (recur [a 10, b 11]) ?
10:24	borkdude	S11001001 or (recur fn-foo) ?
10:24	S11001001	(fn blah #_"intros ordinary lexical binding & recur-target binding" [x] do-stuff (recur blah (inc x)))
10:25	borkdude	S11001001 hmm, I don't see the problem really
10:25	borkdude	S11001001 if you could recur to different targets you could really get a goto-like messy effect
10:25	S11001001	you couldn't
10:26	S11001001	that's not what the feature is for
10:26	S11001001	(fn [x] (macro-that-may-intro-recur-target (recur (inc x))) ; where does recur go?
10:26	S11001001	that's the problem with anaphora
10:26	S11001001	it's easy to just not use aif et al in your macros, because there's a non-anaphoric variant
10:26	S11001001	if-let
10:27	clgv	S11001001: thats specified. it's always the first loop or function that encloses the recur
10:27	S11001001	but there's no variant of fn that doesn't intro an anaphoric recur-target
10:27	Vinzent	how recur-target is anaphoric?
10:27	S11001001	clgv: yes, and that's what's fixed by requiring a label
10:27	gfredericks	S11001001: I think any situation like you describe would make it ambiguous to the programmer that he's in the tail position
10:27	borkdude	S11001001 what would happen here? (fn foo [x] (loop [a 1] (recur foo 10))) isn't that a bit weird that you could do that
10:27	gfredericks	not much consolation though
10:28	joly	isn't 'lexical' a better term here than anaphoric since no new identifier is introduced?
10:28	S11001001	borkdude: compilation failure
10:28	S11001001	joly: it's an identifier namespace with only one identifier
10:28	borkdude	S11001001 so the only thing wrong with loop/recur is being explicit about its target?
10:28	borkdude	not being explicit
10:28	S11001001	borkdude: yes
10:28	borkdude	S11001001 ok understood
10:28	clgv	S11001001: I just compiled a recur form and see that it is transformed into a for loop
10:29	S11001001	yes
10:29	S11001001	recurs would still compile into exactly the same thing
10:29	borkdude	so the message here is: don't use recur in a macro because you don't know what the target will be?
10:29	S11001001	no
10:30	S11001001	the problem is that you can't use fn or loop in your macros without affecting recur calls in bodies given to the macros
10:30	borkdude	yes I kind of meant that
10:30	clgv	S11001001: clojure tells me that 'recur is a special form
10:30	S11001001	yes
10:31	borkdude	S11001001 ah, wait, you are saying don't write a macro that contains recur
10:31	S11001001	a macro call
10:31	borkdude	S11001001 right
10:31	S11001001	so, try to write some lisp without using any macros
10:31	Vinzent	I don't understand - if a macro expansion contains recur, then it also contains enclosing fn or loop. What's the problem?
10:31	clgv	recur is no macro
10:32	S11001001	Vinzent: (fn [x] (macro-that-may-intro-recur-target (recur (inc x))) ; where does recur go?
10:32	borkdude	S11001001 Vinzent yes, I guess a good "bad example" would help here
10:32	borkdude	S11001001 what are the macros that could intro recur targets? only look?
10:32	borkdude	loop
10:32	S11001001	fn
10:33	borkdude	is fn a macro?
10:33	S11001001	yes, but even if it wasn't
10:33	S11001001	any macro using fn to wrap a body would also intro a recur target
10:33	S11001001	like future
10:33	borkdude	ok so (fn [x] (fn [x] … (recur …)) -> this will be clear to the writer what the target is
10:34	Vinzent	S11001001, well, if the doc string of the macro says that it introduces recur target, then to macro-that-may-intro-recur-target; otherwise, to fn. Ideally it should be like that :)
10:34	tmciver	S11001001: I think your example may be a little contrived; why would you write a recur as arg to a macro?
10:34	borkdude	(fn [x] (loop .. (recur ..)) also clear. have a good example where it really goes wrong?
10:35	Vinzent	tmciver, well, one could do that and no exception will be throwed, so I think the example is legal
10:35	S11001001	tmciver: you would write a recur as an arg to if, or cond, wouldn't you?
10:36	clgv	S11001001: so what's your aim? what would you do with your recur issue?
10:36	S11001001	clgv: I've said what the solution would be
10:37	joly	similar to Scheme's named let?
10:37	clgv	S11001001: lost in noise^^
10:37	S11001001	indeed, but with an assertion of tail position
10:37	S11001001	clgv: require a label for recur
10:38	clgv	ah ok. dont know if that is an improvement for most of the programs to be written
10:39	joly	has there been discussion of what tail calls will look like to other functions once Java gets TCO?
10:39	borkdude	ok, so: (defn foo [x] (println "fun" x) (future (println "future" x) (println x) (recur)))
10:39	borkdude	one println too many
10:40	S11001001	lazy-seq intros a recur target
10:40	borkdude	but this indeed has the future as the recur target
10:40	borkdude	and it doesn't print the arg provided to the function
10:40	borkdude	btw
10:40	borkdude	why?
10:40	clojurebot	http://clojure.org/rationale
10:40	S11001001	delay, similarly to future
10:41	S11001001	additionally, many body-taking macros can be written entirely in macro form
10:41	S11001001	however, it is often more convenient and flexible to write most of the functionality in a function variant, and implement the macro by wrapping the body in a fn
10:41	S11001001	that's all future is (it uses future-call)
10:42	S11001001	at this time, binding is not implemented this way
10:42	S11001001	but the decision is more-or-less arbitrary, and can change from version to version
10:44	S11001001	interestingly, while binding does not make a recur-target, with-bindings does (following the strategy of future)
10:46	bordatoue	why is arguments passed as an arrayseq, is there a way to gain indexed access to an element in the arrayseq can i use (args 4) to get the fourth element
10:48	borkdude	(get args 4)
10:48	borkdude	but generally this isn't what you want I think
10:48	bordatoue	borkdude: can i use (args 4)
10:48	borkdude	bordatoue no
10:49	bordatoue	borkdude: why, i thought it was a collection
10:49	borkdude	bordatoue you can use destructuring though
10:49	borkdude	why only vectors, sets and maps behave as functions
10:49	borkdude	-why
10:50	gtrak	can you specify a particular recur target within a recur?
10:52	gtrak	http://stackoverflow.com/questions/8205209/why-argument-list-as-arrayseq
10:53	bordatoue	Is there any easier way to find out documentation for ArraySeq without using google
10:53	borkdude	gtrak yeh, that was me in november 2011… ;-)
10:53	gtrak	lol
10:54	gtrak	bordatoue: I think lein2 can jump to java sources, but I haven't verified
10:59	clgv	bordatoue: why? it is an implementation detail
11:00	clgv	bordatoue: if it's a seq you can use `nth` thats what destructuring do as well
11:01	bordatoue	clgv: how can i tell from a method signature which one implements ISeq and which one is a part of collections
11:01	clgv	ParsingException
11:02	bordatoue	is there an equivalent of type as in python
11:02	clgv	,(type [])
11:02	clojurebot	clojure.lang.PersistentVector
11:02	clgv	,(ancestors (type []))
11:02	clojurebot	#{java.lang.Iterable clojure.lang.IPersistentCollection java.util.Collection java.io.Serializable java.lang.Comparable ...}
11:03	bordatoue	thanks
11:05	clgv	,(g 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20)
11:05	clojurebot	#<CompilerException java.lang.RuntimeException: Unable to resolve symbol: g in this context, compiling:(NO_SOURCE_PATH:0)>
11:05	clgv	lol. nvm
11:06	bordatoue	,(print *e)
11:06	clojurebot	#<Unbound Unbound: #'clojure.core/*e>
11:06	bordatoue	ah ah
11:09	nDuff	Hmm.
11:13	amalloy	e, 1, and friends are only bound in a real repl, not in clojurebot's approximation
11:14	amalloy	gtrak: no, you can't specify a recur target
11:15	gtrak	1.5 needs goto :-)
11:17	duck1123	I bet you could do some amazing things with the inclusion of goto. And every one of them should be punishable by death.
11:17	nDuff	How can I temporarily unbind clojure.lang.Compiler/LOADER?
11:18	bordatoue	is the ISeq similar to yield in python
11:18	nDuff	(it's using a DynamicClassLoader bound to a classloader that can't actually access some things I'm trying to reach)
11:18	nDuff	bordatoue: ...given as yield is a language keyword and ISeq is a protocol, no?
11:19	nDuff	bordatoue: ...now, if the question is whether ISeq can be used to expose lazily-generated values, the answer is yes
11:19	nDuff	bordatoue: ...but it's used for much more than _only_ that.
11:19	bordatoue	nDuff: that was what i was alluding to
11:19	nDuff	bordatoue: ...sequences which aren't lazy at all can and do also implement ISeq
11:20	nDuff	bordatoue: ...by the way, Python also has an iterator protocol, which would be a closer match
11:20	bordatoue	nDuff: is there a way to reset this behaviour
11:20	nDuff	...reset which behavior?
11:21	nDuff	("reset"?)
11:21	bordatoue	generation of elements
11:21	nDuff	If you mean you want to go back to the head of a sequence, just don't lose the head in the first place.
11:21	nDuff	Sequences aren't stateful.
11:21	nDuff	...so "resetting" them has no meaning.
11:21	nDuff	...well...
11:22	nDuff	...eh. "Don't lose the head in the first place" still holds.
11:22	nDuff	Basically -- iterating through a sequence doesn't change or invalidate a pointer to any other place in the sequence
11:22	nDuff	...it just gives you new pointers to later locations
11:22	nDuff	so there's nothing that ever needs to be reset.
11:23	gtrak	bordatoue: no, you should really treat the contents as values and not use it for side-effects
11:23	bordatoue	how is this sequence implemented internally
11:24	nDuff	bordatoue: ISeq is an interface. Interfaces have no implementations.
11:25	nDuff	bordatoue: ...so different types implementing ISeq have separate implementations.
11:27	bordatoue	nDuff: I am refering to seq
11:27	nDuff	bordatoue: ...and (seq foo) does completely different things depending on what foo is
11:27	bordatoue	nDuff: why should seq be lazy,
11:28	gtrak	fp troll: (defn a [b] (label ::c (d b))) (defn e [f] (let [b f] (goto ::c)))
11:29	nDuff	bordatoue: ...because the interface is defined in such a way that it's possible for implementations to be lazy? Which is not to say that every one of them _is_, simply that they _can be_.
11:29	nDuff	bordatoue: If you're genuinely curious about implementation details, I'd suggest downloading the source to Clojure and reading through a bit. It's actually not that hard, and very much enlightening.
11:30	clgv	bordatoue: it seems to be time for an introductory clojure book, don't you think?
11:30	bordatoue	if seq is used on java arrays, why should it be lazy what advantage does it offer. If i have recursive method that uses seq then it will not work due to single pass behaviour
11:30	nDuff	bordatoue: that's not true.
11:31	nDuff	bordatoue: (that latter part, that is)
11:31	nDuff	bordatoue: ...you might _think_ it's true, because you expect Clojure sequences to behave like Python iterators, which are stateful
11:31	nDuff	bordatoue: ...but they aren't stateful, so you can use them recursively just fine.
11:31	nDuff	bordatoue: ...I'm inclined to agree with clgv here -- some dead-tree reading would be helpful.
11:33	nDuff	s/stateful/mutable/ (to be more accurate)
11:34	gtrak	actually I think it's possible to implement this goto thing
11:34	gtrak	I just have to add some indirection to the lexical bindings within a label
11:35	nDuff	I know it's possible to unbind a var's root definition using Var.unbindRoot(), but is there a sane way to push Unbound onto the stack (ie. to do so temporarily)?
11:38	amalloy	nDuff: i don't think so. i'm curious why you'd want to?
11:39	nDuff	amalloy: I want to temporarily unset clojure.lang.Compiler/LOADER, such that a new DynamicClassLoader is generated based on the current thread context classloader
11:40	nDuff	...actually, thinking about that, it'd be better just to cache one, such that I don't have a bunch of un-GC-able classloaders sitting around in memory...
11:40	amalloy	nDuff: and you can't just bind it to nil because someone is checking whether the var is bound?
11:40	Hodapp	maaaan, some people really love their static type systems
11:41	amalloy	Hodapp: they're pretty cool! i'd be happy to have one like haskell's available in clojure
11:42	nDuff	amalloy: ...right, though I _can_ just call (RT/makeClassLoader) and rebind it myself. Going that route.
11:44	nDuff	Huh. clojure.lang.Compiler/LOADER evals to "#<Var: --unnamed-->", but (binding [clojure.lang.Compiler/LOADER (clojure.lang.RT/makeClassLoader)] true) yields "CompilerException java.lang.RuntimeException: Unable to resolve var: clojure.lang.Compiler/LOADER in this context, compiling:(NO_SOURCE_PATH:1)"
11:46	amalloy	nDuff: try with-bindings*
11:47	amalloy	bindings expects to be given a symbol which ns-resolves to a var; that's an expression which evaluates to a var
11:47	amalloy	or actually with-bindings is probably good enough
11:48	Hodapp	amalloy: well, it boils down to the assertion that the only code that should be written is code that is sufficiently constrained to be amenable to particular types of automated proofs of certain aspects of its behavior.
11:50	amalloy	Hodapp: that sounds like a strawman. why shouldn't i say the same thing about functions? life was better with GOTOs, when the compiler/assembler could trust me to decide what the instruction pointer should be, instead of proving that i only jump to designated "function begin" points
11:51	Hodapp	amalloy: How is this a strawman? This is precisely what static type systems are.
11:51	Hodapp	amalloy: More or less by definition.
11:51	Chiron_	Hi, I want to use clojure to code storm transaction topology such as this one: https://github.com/nathanmarz/storm-starter/blob/master/src/jvm/storm/starter/TransactionalGlobalCount.java
11:51	amalloy	yes, but it's such a general statement that it's also what every other compiler does
11:51	Chiron_	I thought about reify but not sure what about the instance variables
11:51	amalloy	anything more complicated than ASM falls under your criticism
11:52	Hodapp	amalloy: Calling that a criticism is a strawman.
11:52	amalloy	okay, bro
11:52	Hodapp	amalloy: So you're basically telling me that what I said is true, but it's not a valid criticism.
11:53	Hodapp	amalloy: That's fine. It wasn't meant as a criticism, so it not being a valid criticism isn't really relevant.
11:54	Hodapp	amalloy: The point is that it's just one form of constraining the code - in this case, constraining it to that which the type system may prove things about. There are multiple other ways of constraining the code, and multiple other ways of reasoning about the code whether through automated proofs (as in a static type system) or by hand.
11:56	Hodapp	The criticism is not that it's constraining the code, but rather that it's an assertion that that is the de facto correct way of doing it.
11:56	Hodapp	What if you need a level of dynamism it will not allow?
11:56	Chiron_	sorry guys, any help?
11:57	amalloy	Hodapp: i don't think i would want clojure to have a required static-type system, agreed
11:57	Hodapp	amalloy: It would be within the realm of possibility to have macros be generated in a way about which similar automated proof is possible, for instance.
11:57	amalloy	Hodapp: what static-typing languages have you used?
11:58	Hodapp	amalloy: Well, C++, Java, C if that counts... and Scala and a little OCaml.
11:58	Hodapp	I quite like Scala's. OCaml I have not gotten into much detail with.
11:58	hyPiRion	Chiron_: If you receive no answer, chances are people either don't know or are busy doing other stuff :)
11:58	Hodapp	Haskell is on my list.
11:59	amalloy	nod the only one i've liked is haskell, but i haven't tried ocaml and have barely touched scala
11:59	Chiron_	hyPiRion: sorry yes I know
11:59	Hodapp	in Scala it's like it's hardly there half the time; the type inference is quite good
11:59	amalloy	IME you don't actually need the "dynamism" you bring up, in a language like scala or haskell where the static typing isn't crap
12:00	Hodapp	I'm not sure that said dynamism can be ruled out in all cases.
12:00	hyPiRion	Chiron_: No worries, just sayin'.
12:00	gtrak	I can't help but think in terms of contracts anyway, I've had a little trouble adjusting to the looseness of it
12:01	Hodapp	I believe there are probably usages where you are either going to use something dynamic, or you're going to end up inventing your own with the same consequences but a bit uglier.
12:01	amalloy	could be
12:02	gtrak	like... 'function b calls input-stream on its arg' is a lot like 'function b expects this arg to be of the type of things that input-stream can be called on'
12:03	gtrak	I feel like I'm creating anonymous types in my mind when i read the code
12:03	Chiron_	so maybe in another way, how to extend a Java class in Clojure and add some instance variables to the extended class?
12:03	nDuff	Hrm; there's something more than I thought going on here -- setting both the thread context classloader _and_ Compiler/LOADER still doesn't work. https://gist.github.com/dd7cfb33fdcf3a1d3e3a
12:05	hyPiRion	Chiron_: My usual way of doing it would be to create a java-file, compile it and refer to it through require or something like that. Leiningen makes it rather simple to do.
12:06	hyPiRion	Or is the goal to do it in Clojure?
12:06	Chiron_	hyPiRion: i want to do it in Clojure
12:07	Chiron_	if it is not possible to extend and add instance vars in clojure, maybe a better option to use another jvm language -scala/jruby- in this case
12:07	Chiron_	extends sure is doable in clojure but what about adding instance vars?
12:08	hyPiRion	I haven't worked much with java-interop, but have you read http://clojure.org/java_interop#Java%20Interop-Implementing%20Interfaces%20and%20Extending%20Classes ?
12:09	hyPiRion	Look at http://richhickey.github.com/clojure/clojure.core-api.html#clojure.core/gen-class - it seems like there's some way of doing it there.
12:09	lazybot	Nooooo, that's so out of date! Please see instead http://clojure.github.com/clojure/clojure.core-api.html#clojure.core/gen-class and try to stop linking to rich's repo.
12:09	hyPiRion	Sorry lazybot :(
12:10	rvgate	using webnoir sessions, im trying to change a value stored within the session... (+ (session/get :credits) 5), but for some reason this gives me a class cast exception...
12:10	S11001001	Chiron_: in case this isn't clear, java classes are not the usual tool for modeling problems in clojure; they are avoided except where needed for interfacing with some bit of java
12:11	Chiron_	Yes I know but I'm trying to code this in pure Clojure https://github.com/nathanmarz/storm-starter/blob/master/src/jvm/storm/starter/TransactionalWords.java
12:11	Chiron_	as you see, KeyedCountUpdater define a couple of instance variables
12:12	S11001001	Chiron_: so don't do a one-to-one translation; use clojure-appropriate abstractions instead, like maps to represent data structures
12:12	hyPiRion	rvgate: use session/swap! instead?
12:12	hyPiRion	(session/swap! :credits + 5)
12:13	Chiron_	my problem is how to code KeyedCountUpdater in Clojure? no problem with extending/implementing other classes/interfaces . but what about defining new instance variables?
12:13	S11001001	{:counts {} :collector (stuff here) :id (stuff here)}
12:14	S11001001	don't extend any classes, don't implement any interfaces
12:14	rvgate	hyPiRion, same thing.. classcast exception
12:14	hyPiRion	Weird. You sure the value is a number?
12:14	S11001001	if you need to capture data in the implementation of a thing that must be a java interface, the lexical context of a reify will satisfy that need
12:15	rvgate	hyPiRion, yes
12:16	technomancy	muhoo: if you need snapshots to update you should probably use frozen timestamped snapshots
12:16	technomancy	muhoo: unless you're not in production; then you can use lein2's :snapshots :always policy in project.clj
12:16	rvgate	hyPiRion, (if (nil? (session/get :credits)) (session/put! :credits 10))
12:17	hyPiRion	And that doesn't work?
12:17	rvgate	hyPiRion, just showing that its a number
12:18	hyPiRion	Ah. Then I have no clue.
12:19	bordatoue	is anyone using counterclockwise eclipse plugin
12:20	bordatoue	I have included :gen-class but I can't see any class files generated
12:20	bordatoue	is there any other commands that needs to be specified
12:24	muhoo	technomancy: i've got a new problem now. the upstream maintainer of this library (ahem, google) doesn't do version numbering. it's just "1-SNAPSHOT". but lein is unhappy with that when making jars
12:24	muhoo	it's been 1-SNAPSHOT for 4 years now :-(
12:25	technomancy	good hell
12:25	muhoo	so, should i fork it and publish it with a new groupid?
12:25	muhoo	or should i put it up on clojars with the original groupid, but a proper non-snapshot number?
12:25	technomancy	first closure releases and now this; what is wrong with google?
12:26	technomancy	if they deploy correctly you should still be able to use frozen timestamped snapshots
12:26	technomancy	but I don't put a lot of hope into them getting that right
12:26	muhoo	is there a tut somewhere on dealing with frozen timestamped snapshots?
12:26	muhoo	also, will lein let me build a jar if i depend on one?
12:26	technomancy	definitely
12:27	technomancy	in lein1 you just do `ls lib` and you can find the frozen snapshot numbers
12:28	technomancy	in lein2 `lein deps :tree` will show the frozen version number
12:28	muhoo	oh, ok. so i depend on [foo "0.1.0-20120608.093814-4"] then?
12:29	technomancy	yeah that should do it
12:32	muhoo	that worked great for my lein-built library. as for my maven-built google library, not so much. it's literally "1-SNAPSHOT" in lein2 tree and in lein1 ~/m2/repository
12:32	technomancy	yeah, it means google screwed up their deploy
12:32	technomancy	I assume you've reported a bug? =\
12:32	muhoo	well, i did the deploy. so i screwed it up
12:33	muhoo	i built it form their source. they screwed up the version non-numbering :-)
12:33	technomancy	oh, they don't even put it in a repository?
12:33	technomancy	eep
12:33	technomancy	that's several times worse
12:33	muhoo	it's in svn
12:33	muhoo	come on, this is google we're talking about: the IBM of the 21st cen
12:34	technomancy	bloody unprofessional
12:34	muhoo	afaict, this thing isn't in maven anywhere
12:34	technomancy	make some noise on the bug tracker; they should feel guilty about this
12:35	technomancy	if you use `mvn deploy` you should have numbered snapshots I believe
12:36	bordatoue	is it possible to check if :gen-class is working from the repl
12:37	muhoo	hahaha, mvn deploy does this: https://www.refheap.com/paste/3056
12:37	muhoo	it wants me to sign in in order to deploy! presumably to their internal mvn repo
12:38	muhoo	whiskey. tango. foxtrot.
12:40	technomancy	oh yeah, you'd have to switch up the target repository
12:41	muhoo	i was in the target repo. there are several stacked targets, with a parent that is kind of a stub with common deps
12:41	muhoo	and... in a sub repo, same authentication message Authentication realm: <https://step2.googlecode.com:443> Google Code Subversion Repository
12:42	muhoo	oh nevermind
12:42	muhoo	there's got to be a maven way to tell it which repo to use. i'll have to read up on this
12:43	muhoo	technomancy: thanks, i think i got it now
12:55	muhoo	technomancy: wait, no, this sucks. if i'm depending on a library that is perpetually in 1-SNAPSHOT, then i'm always going to have to have people depend on my snapshot dated jars, i.e. "0.1.0-2012xxxx"
12:57	amalloy	i don't see why. you don't have to make your own library a snapshot version
12:57	technomancy	muhoo: pretty sure the alternatives are worse =\
12:57	gfredericks	which tool is it that won't let you depend on snapshots unless your own lib is a snapshot?
12:58	gfredericks	there must be something because I defenitely had to push some else's lib up to clojars once so I could make a non-snapshot release
12:59	technomancy	gfredericks: you can make releases that depend on frozen snapshots with lein
12:59	diso	quit
12:59	technomancy	just not nondeterministic snapshots
12:59	gfredericks	ah okay; is it lein that makes that restriction?
12:59	gfredericks	or maven?
12:59	technomancy	lein
12:59	gfredericks	okay
13:00	gfredericks	(inc lein)
13:00	lazybot	⇒ 1
13:00	muhoo	ok, i think i got it. i can make my lib depend on a dated frozen snapshot. then my lib doesn't have to be a snapshot, it can be a release. problem solved, i think.
13:00	muhoo	the b0rken google lib can be a snapshot, i don't give a damn. as long as mine doesn't have to be.
13:01	technomancy	aye
13:07	muhoo	every minute that i have to deal with maven, i give thanks to leiningene
13:08	gfredericks	Abraham Lincoln felt that so strongly he made it a national holiday
13:08	Hodapp	muhoo: curiously, I was in #java and had an Ant question. They promptly chased me up a tree and lit it on fire to try to get me to not only use Maven but to swear allegiance to Maven.
13:09	muhoo	well maven > ant, but leiningen > maven, imho
13:11	nDuff	muhoo: ...depends on the use case. Building JIRA plugins, for instance, Leiningen isn't so much the right tool.
13:12	borkdude	if you guys like playing hangman on heroku programmed in clojure, some of my students just sent their link: http://clojurehangman.herokuapp.com/
13:12	borkdude	=)
13:13	Hodapp	I liked the simplicity of Ant, really.
13:13	technomancy	borkdude: pretty cool =)
13:13	bordatoue	could anyone please guide me on how to create a fat jar using counterclockwise plugin
13:14	bordatoue	i can't imagine the number of steps required to attain simple things such as a executable jar
13:14	borkdude	bordatoue just "lein uberjar" from the console
13:15	bordatoue	borkdude: please can you tell me how to do it with eclipse
13:15	technomancy	IIRC task invocation hasn't been implemented yet in CCW?
13:15	borkdude	bordatoue I wouldn't know how to do it with eclipse, I never needed to
13:15	bordatoue	borkdude: lein seems to work , but i need to work with eclipse as well
13:27	josteink	about clojure and web
13:27	josteink	is noir considered the standard web framework for clojure or should I also look into other stuff? :)
13:27	ejackson	josteink: pretty standard
13:28	josteink	I must admit I didnt pay that much attention. NDC has a "overflow" room with projectors covering ALL sessions in paralell
13:28	josteink	so I was jumping between that, billy hollis on UX and async in C#
13:29	josteink	guess Ill go look for some more detailed samples
13:29	amalloy	josteink: noir is easiest. it's also popular to just use the layer underneath it (compojure) directly
13:29	josteink	while C# suffers from inability for language users to -extend- the language itself, the async stuff was pretty good
13:29	josteink	but would be neat to fool around with something which isnt my day to day language ;)
13:30	josteink	amalloy: thanks for the tip. I guess Ill start at the top and work my way down if things starts feeling icky ;)
13:30	borkdude	josteink noir is easy - I teached some students only 5 lessons of clojure and they already could come up with Hangman in noir: http://clojurehangman.herokuapp.com/
13:30	ystael	Is there a general principle for what causes "IllegalStateException: Attempting to call unbound fn" at runtime in code that was perfectly happy at compile time? I don't think I'm doing anything clever with fn bindings.
13:30	josteink	borkdude: nice
13:31	borkdude	(others are building an irc client using seesaw)
13:32	S11001001	ystael: look at the frame where the throw happens; the var get should be under where you use it
13:32	amalloy	well, for example (declare x) (defn y [] (x)) (...never define x...)
13:32	borkdude	have their ever been 500 people in this room at the same time?
13:32	amalloy	yes
13:33	josteink	borkdude: Ive written enough raw protocol level irc-code in enough languages to be very decided on the fact that Im not doing that again ;)
13:33	borkdude	josteink I saw a tweet about NDC/Clojure, but I'm curious who presented what
13:33	josteink	borkdude: some chick I have not heard about before. bodill stokke or something.
13:33	borkdude	josteink they keep it simple and probably unfinished proof of concept
13:33	josteink	borkdude: she was trying to overplay cool/humerous IMO
13:33	josteink	but ah well
13:33	josteink	she kept the code fairly managable
13:34	borkdude	at least clojure picks up chicks, maybe clojure is a brogrammer language?
13:34	borkdude	:P
13:34	josteink	https://twitter.com/#!/bodiltv
13:34	ystael	S11001001: I see the call that fails, I just don't understand why it fails... The fn in question is the ->Type of a defrecord which is required and properly namespace-referenced at the call site
13:35	borkdude	josteink cool tnx
13:35	Hodapp	curiously enough, I found out on date #2 that my current girlfriend is a Linux user.
13:35	pepijndevos	what does Clojure use to transform * to __STAR__ or something?
13:35	josteink	borkdude: NDC typically publishes videos of ALL sessions some time after the conference
13:35	Hodapp	she never thought much of it. Her brother put it on her netbook after her Windows install killed itself, and since she could still browse and play her music she didn't care.
13:35	S11001001	ystael: check ->Type and (meta #'->Type)
13:35	josteink	borkdude: in case you are interested. but it was way more clojure talks last year
13:35	ystael	S11001001: ok, will do, thank you!
13:36	borkdude	Hodapp well that is how some linuxens are intended right
13:37	borkdude	josteink this looks interesting https://github.com/bodil/catnip
13:38	josteink	I mean... not linux itself, just the crappy desktop environments :P
13:38	josteink	borkdude: nice
13:39	Hodapp	borkdude: I just find it kind of hilarious since I have plenty of friends who, if they were asking about said girlfriend, the first question they'd get out of the way is "Does she use Linux?"
13:39	borkdude	Hodapp ah well
13:39	josteink	the first thing my GF asked when she saw my KDE machine was "That looks like a computer from the stone age"
13:39	josteink	cant say I blame her :P
13:40	Hodapp	when I lugged my newly-built computer into college, some people there had literally never seen a tower before
13:40	josteink	uh
13:40	Hodapp	Some thought it was from the stone age. Others thought it was a goddamn Gibson.
13:40	josteink	laptops havent been the standard for -that- long
13:40	josteink	unless you are 15 years old
13:40	borkdude	Hodapp I met my girlfriend/wife in a chatroom, we added each other on ICQ to continue our conversation about topology (math)
13:40	technomancy	josteink: pffft; that's nothing: http://p.hagelb.org/glasstty.png
13:40	borkdude	Hodapp math girls…
13:40	josteink	technomancy: heh
13:41	josteink	technomancy: is that like emacs OS or what? :P
13:41	technomancy	josteink: "emacs OS" is redundant =)
13:41	technomancy	it's just emacs in a full-screen terminal with the glasstty font
13:41	ystael	S11001001: interesting. the var metadata is {:ns #<Namespace the.ns>, :name ->Type} and (prn ->Type) prints as #<Unbound Unbound: #'the.ns/->Type>
13:42	Hodapp	technomancy: is this what the Overtone demo used?
13:42	ystael	This is an aot-compiled jar loaded into a groovy interpreter, I expect that has something to do with it
13:42	technomancy	Hodapp: I think that was a custom terminal program. this is just gnome-terminal with a fancy font.
13:42	Hodapp	ah, okay
13:42	borkdude	this is pretty cool, has anyone tried it? https://github.com/bodil/catnip
13:42	Hodapp	I did like that Overtone demo. Too bad I'm music-tarded.
13:43	septomin	there's some terminal for mac that distorts the screen like a CRT
13:43	amalloy	ystael: you have to init the clojure runtime, probably
13:43	Hodapp	why not just use a real CRT >:(
13:43	Hodapp	better color!
13:44	josteink	CRT :(
13:44	septomin	i actually considered looking for a CRT after buying my first atari 2600 recently
13:44	ystael	amalloy: before any non-aot-compiled call will work?
13:44	josteink	oooh. I tried hooking my old amiga 1200 up to me LCD 40" HDTV
13:44	josteink	let me just now NOT recommend that
13:45	Hodapp	if I had the desk space I'd get one of those old Sun or SGI ones that needed a forklift to move and set off car alarms outside when you turn them on
13:45	josteink	I think the LCD TV came out confused of that occation too. I mean... the odd interlacing and all :P
13:45	Hodapp	those had the best image and color I've ever seen
13:45	septomin	my tv was fine, it just seemed weird to have these perfectly defined rectangles
13:47	Hodapp	septomin: yeah, this is why things like scale2x were made I guess
13:48	Hodapp	or hq3x... https://en.wikipedia.org/wiki/Scale2x#hqnx_family
13:48	Hodapp	those techniques are neat from the standpoint of basically being LUTs, very easy to do
13:48	borkdude	I'll say it one last time ;-) : this is cool! https://github.com/bodil/catnip
13:49	josteink	so out of curiousity.... I saw this talk about hadoop today. saw it was JVM based and all
13:49	josteink	anyone tried clojure with hadoop? :P
13:49	Hodapp	borkdude: could you repeat whatever you just said?
13:49	borkdude	I'll say it one last time ;-) : this is cool! https://github.com/bodil/catnip
13:49	Hodapp	BAHAHAHA
13:49	borkdude	:P
13:49	technomancy	josteink: most people who need hadoop use cascalog
13:49	technomancy	you can do raw hadoop, but it isn't fun
13:50	S11001001	ystael: try loading the clojure module before waking up the class
13:50	S11001001	oh, n/m
13:51	ystael	I'm guessing I'm in some state where anything that requires the binding of a var to be found will fail?
13:52	pepijndevos	How does Clojure turn symbols with non-alphanumeric characters into something Java understands?
13:52	amalloy	pepijndevos: it's in Compiler somewhere
13:53	pepijndevos	amalloy, erm, ok, so not something I'm supposed to actually use I guess :(
13:53	amalloy	&clojure.lang.Compiler/CHAR_MAP
13:53	lazybot	java.lang.SecurityException: You tripped the alarm! class clojure.lang.Compiler is bad!
13:53	amalloy	,clojure.lang.Compiler/CHAR_MAP
13:53	clojurebot	{\@ "_CIRCA_", \! "_BANG_", \" "_DOUBLEQUOTE_", \# "_SHARP_", \% "_PERCENT_", ...}
13:53	pepijndevos	I was hoping something like clojure.core/garble
13:54	pepijndevos	nice
13:54	amalloy	,(clojure.lang.Compiler/munge "empty?")
13:54	clojurebot	"empty_QMARK_"
13:54	technomancy	so ring requires its session secret for encrypting cookie data be 16 bytes
13:55	ystael	this is extra interesting because it works from jruby but not from groovy
13:55	technomancy	would it be appropriate to use post-conditions to enforce that?
13:55	ystael	so some aspect of how things are getting initialized is different
13:55	pepijndevos	amalloy, thanks
13:57	technomancy	also, why is the issue tracker on ring disabled?
13:57	amalloy	pepijndevos: btw, you could find that just by searching for _QMARK_ in the clojure source
13:59	hyPiRion	One question there though
13:59	pepijndevos	amalloy, hm, true. I couldn't remember when clojure generated these, and which ones.
13:59	hyPiRion	would (def a_QMARK_ ...) and (def a? ...) conflict?
14:00	amalloy	hyPiRion: you can try that faster than you can ask the question
14:01	ystael	S11001001, amalloy: i think i figured it out, thanks for your help
14:03	hyPiRion	,(fn a? [])
14:03	clojurebot	#<sandbox$eval27$a_QMARK___28 sandbox$eval27$a_QMARK___28@62d22770>
14:04	hyPiRion	,(fn a_QMARK[])
14:04	clojurebot	#<sandbox$eval55$a_QMARK__56 sandbox$eval55$a_QMARK__56@2375b4dd>
14:04	hyPiRion	They're equivalent without the gensym-append.
14:05	borkdude	hyPiRion try with defn, does it matter?
14:05	hyPiRion	Oh, I know their names refer to different objects in Clojure.
14:06	S11001001	so even aoted you wouldn't get a conflict
14:07	borkdude	if I defn a_QMARK_: #<user$a_QMARK_ user$a_QMARK_@12b8501d>
14:07	borkdude	and defn a? #<user$a_QMARK_ user$a_QMARK_@1330b13c>
14:07	gfredericks	:/
14:07	borkdude	so how does this work?
14:08	gfredericks	probably just "don't do that"
14:08	borkdude	gfredericks still I don't get a conflict in the repl
14:08	hyPiRion	It hides the gensym-number, I suppose.
14:08	S11001001	,first
14:08	clojurebot	#<core$first clojure.core$first@25a81332>
14:09	gfredericks	the gensym-number?
14:09	S11001001	maybe not
14:09	S11001001	,(class first)
14:09	clojurebot	clojure.core$first
14:09	S11001001	well now
14:09	S11001001	make class, instantiate, replace class, instantiate
14:10	hyPiRion	,[(fn a?[]) (gensym "a_QMARK_") (fn a?[])]
14:10	clojurebot	[#<sandbox$eval133$a_QMARK___134 sandbox$eval133$a_QMARK___134@6baacbb8> a_QMARK_139 #<sandbox$eval133$a_QMARK___136 sandbox$eval133$a_QMARK___136@67eb04cf>]
14:10	gfredericks	ah yeah
14:10	gfredericks	the two classes are not identical?
14:11	hyPiRion	the functions, you mean?
14:11	gfredericks	(identical? (class a?) (class a_QMARK_))
14:11	S11001001	,(identical? (do (defn x[]) (class x)) (do (defn x[]) (class x)))
14:11	clojurebot	#<Exception java.lang.Exception: SANBOX DENIED>
14:11	S11001001	heh
14:12	borkdude	S11001001 false
14:12	hyPiRion	,(class a_QMARK_)
14:12	clojurebot	#<CompilerException java.lang.RuntimeException: Unable to resolve symbol: a_QMARK_ in this context, compiling:(NO_SOURCE_PATH:0)>
14:14	gfredericks	it's odd the jvm allows it
14:14	solussd	is there a way to 'dump' your current namespace (in the repl) so you can restore require/use 'd libraries?
14:15	gfredericks	(ns user1234)
14:15	hyPiRion	just change from one ns to another
14:16	solussd	by dump I meant print out all the currently required/used namespaces, not 'get rid of'. :)
14:16	technomancy	it would be awesome if namespaces printed as their ns form
14:17	borkdude	so how does clojure distuingish between a defned function called a? and a_QMARK_ ?
14:17	solussd	one of these days I'll give consolidating all the nice things I want in my repl into a library.. (un-intern symbols, dump ns, etc)
14:18	hyPiRion	solussd: I'd probably look at the ns- functions
14:18	S11001001	borkdude: because it doesn't use class names for stuff
14:19	technomancy	(defmethod print-method clojure.lang.Namespace [x writer] (print-method `(ns (.getName x) [...]) writer)) ; plz
14:19	borkdude	S11001001 clojure doesn't use the emitted classes?
14:20	technomancy	hm; I wonder if namespaces have enough info in their fields to reconstruct the ns form from them...!
14:20	amalloy	technomancy: java.lang.ClassCastException: clojure.lang.PersistentList cannot be cast to java.lang.String
14:21	amalloy	or no, you were cleverer than that
14:21	amalloy	sneaky
14:21	solussd	hyPiRion: what's "ns-functions"?
14:22	borkdude	S11001001 when would this become problematic, only when calling these functions from Java or what?
14:22	amalloy	technomancy: i don't think they do
14:22	hyPiRion	,(ns-imports ns)
14:22	hyPiRion	etc
14:22	clojurebot	{ProcessBuilder java.lang.ProcessBuilder, Enum java.lang.Enum, SuppressWarnings java.lang.SuppressWarnings, Throwable java.lang.Throwable, InterruptedException java.lang.InterruptedException, ...}
14:23	solussd	ah
14:23	technomancy	amalloy: yeah it looks like it has refer info but not require-as
14:23	technomancy	at least according to the slime inspector
14:23	S11001001	borkdude: and you don't call clojure functions from java by referring to the class that happens to implement the function, particularly as it probably doesn't exist while you're compiling java
14:24	technomancy	oh, not true; ns-aliases
14:24	amalloy	technomancy: funny, i read it as the other way around. maybe it has both
14:25	technomancy	C-c S-i on a namespace instance shows the refers; on the class it shows the aliases
14:26	stuarthalloway	any org-mode users want to upvote this for me? http://dev.clojure.org/jira/browse/CLJ-1009
14:28	gfredericks	stuarthalloway: yes.
14:28	gfredericks	I have a pile of code I could delete if that got in
14:29	technomancy	gfredericks: have you tried doric?
14:29	gfredericks	technomancy: nope
14:29	technomancy	IIRC it's org-compatible
14:29	nDuff	What's the point of building a stack of DynamicClassLoader instances? clojure.lang.RT.baseLoader() returns Compiler.LOADER if it's bound, and makeClassLoader builds a new DynamicClassLoader on top of the current baseLoader... but why?
14:30	gfredericks	technomancy: it looks orgy-mody-compatible-y
14:31	timvisher_	would the right place to grab slime from related to clojure development be technomancy/slime on github?
14:31	technomancy	timvisher_: if you have to work from source for some reason, sure.
14:32	timvisher_	i'm just trying to build an info manual for it
14:32	timvisher_	but it would make sense for me to have docs for it related specifically to the version i care about
14:33	timvisher_	i'm assuming your copy is the bundled copy
14:34	technomancy	oh sure; that makes sense
14:53	solussd	I'm using enliven w/ noir and it appears that I cannot use dynamic vars (e.g. errors) inside an enliven template selector handler function. The dynamic vars are unbound in the template. Supposedly there was a fix for this in v1.0.1, but it appears to still not work. :/ anyone else have any familiarity with this issue?
14:53	solussd	*enlive (damnyouosxautocorrect)
14:56	cemerick	solussd: it really, really should work. Are you certain you're using 1.0.1?
14:56	solussd	cemerick, I am. :/ at least my repl spits out the right source
14:56	solussd	e.g. (source net.cgrand.enlive-html/snippet*)
14:57	cemerick	hrm
14:58	cemerick	solussd: I'll see what I can see.
14:58	solussd	thanks
15:08	muhoo	cemerick, technomancy: bumped version to release, got maven squared away, fixed demo app so it actually works. whew
15:08	technomancy	cool
15:09	muhoo	technomancy: thanks especially for your help.
15:09	technomancy	no problem. I understand the build environment isn't always transparent, and we're working on improving that.
15:10	muhoo	google apps login in clojure/noir through friend on heroku http://young-cloud-3820.herokuapp.com/
15:10	technomancy	there are OSS replacements for the entire build chain, but they're not quite polished and ready to become default yet
15:10	muhoo	(if you have a google apps account, that is)
15:10	cemerick	muhoo: I don't, but awesome that you have a demo app up too :-D
15:11	muhoo	maven isn't oss? it's from apache...
15:11	technomancy	muhoo: I mean for the heroku-specific parts
15:11	muhoo	oh, i had no trouble with the heroku parts
15:11	technomancy	well, you did have an issue with the build's cached .m2 dir
15:11	muhoo	it was google's openid that was giving me the shits
15:12	technomancy	where if it was local you could have just gone in and done rm -rf on the old snapshot
15:12	borkdude	muhoo what am I supposed to fill in?
15:12	nDuff	Whoa.
15:12	borkdude	my password is going straight into your db? ;)
15:13	muhoo	borkdude: that's not how opneid works :-)
15:13	borkdude	ah, I can just fill in google.com
15:13	cemerick	borkdude: google is asking, not muhoo :-)
15:14	borkdude	gmail.com
15:14	muhoo	gmail.com should work too
15:14	borkdude	Internal error. It has been reported.
15:14	cemerick	well, only if you have google apps gmail
15:14	amalloy	muhoo: pretty cool
15:15	amalloy	i don't really know what it's for, but hurrah
15:15	borkdude	cemerick I have a gmail address.. should it work?
15:15	cemerick	borkdude: gmail != google apps
15:15	cemerick	I don't have google apps either, but a fair number of others do.
15:15	technomancy	cemerick: turns out the oauth workflow I needed was too trivial for friend; I was overcomplicating it and it turned out to be like five lines of compojure+clj-http.
15:15	muhoo	i haven't tried that. my use caase is a corporate intranet with a google apps auth setup
15:16	amalloy	i have a few google apps accounts, and this works fine. now muhoo's app knows my first and last name!
15:17	kaoD_	I can't believe this
15:17	borkdude	I have done oauth too, with http://whosnotfollowingme.heroku.com
15:17	kaoD_	I'm generating an image and outputting it through Ring
15:17	kaoD_	it doesn't work
15:18	kaoD_	but when I run it using lein-ring
15:18	kaoD_	and make any change to the namespace so it reloads
15:18	kaoD_	it works
15:18	kaoD_	what the hell is going on
15:18	kaoD_	even if the change is adding a space or removing a newline!
15:18	cemerick	technomancy: presumably you had to learn about oauth? Sounds painful enough to want to use friend anyway ;-)
15:19	jedmtnman1	is (defn- …) valid syntax - if it is, i dont know how it diffs from (defn ..)
15:19	gfredericks	defn- makes it private
15:19	cemerick	borkdude: that's cool, now I know who to not send holiday cards to! :-D
15:19	jedmtnman1	ahh
15:19	technomancy	cemerick: the only complicated bit was actually registering my application with github and figuring where the tokens go: http://develop.github.com/p/oauth.html
15:19	gfredericks	== (defn ^:private foo ...)
15:19	borkdude	:-)
15:20	borkdude	Young-cloud-3820.herokuapp.com is asking for some information from your Google Account michielborkent@gmail.com
15:20	borkdude	Allow -> Internal error. It has been reported.
15:20	borkdude	should this not just work then? google has an openid provider thing
15:21	wink	openid is horrible in its own way
15:22	technomancy	worked for me
15:22	borkdude	ah wait, maybe it is a third party cookie thing
15:23	muhoo	internal error means a discovery exception i didn't catch :-)
15:23	muhoo	you don't have a google apps account, i bet
15:24	borkdude	muhoo it doesn't work with whatever openid provider?
15:24	borkdude	muhoo ok
15:25	muhoo	not my thing. the built-in openid workflow of friend does.
15:25	borkdude	muhoo I don't have google apps then I guess…. I do have a google appspot account
15:25	muhoo	mine is a special case, written and tested for google apps only
15:25	cemerick	I think a friend-demo webapp is called for. Include all known friend workflows. :-P
15:27	solussd	cemerick: if I wrap my call to the template in a 'doall', it works… not very elegant or intuitive for someone writing the body of a defpage though. :/
15:27	josteink	borkdude: google apps is overrated. provides you with a vanity email, and no early access to any new google stuff :P
15:28	borkdude	josteink at least you can search in your mail properly and don't have stupid 1 GB mail limits I now have at my work
15:28	josteink	heh
15:29	josteink	i love the spam filter we have at work
15:29	nDuff	...oh, you mean as opposed to a gmail.com account.
15:29	josteink	it has like 100% false positives
15:29	borkdude	I never think about spam filters since my gmail
15:29	nDuff	I like Google, but willing to lock my online identity into them forever? No thanks.
15:30	josteink	and whenever you whitelist something it has caught which it shouldnt, it says "OK dude!" and then promptly forgets about it and categorizes everything from that sender as spam also in the future
15:30	josteink	nDuff: fair enough
15:30	josteink	it means you owe your email identity, which can also be argued to be a oath identity
15:31	josteink	or openid identity or whatever
15:31	gfredericks	nDuff: why do you need your own domain when you could get a microblog on a transient app made by a hipster startup?
15:31	borkdude	josteink I remember doing those kinds of things as well, it was 2006 and we had a sysadmin installing special spam software and spam plugins in our outlook clients
15:31	josteink	yeah
15:31	josteink	I probably spent more time tuning spamassasin than reading mail before I gave up on doing my own email server
15:32	borkdude	josteink I think that was the software
15:32	cemerick	solussd: I dunno, some quick REPL tests with enlive 1.0.1 look good here.
15:32	cemerick	Using both snippet and deftemplate, bound vars appear to be used as I'd expect.
15:32	cemerick	solussd: can you boil it down to a small example?
15:33	cemerick	nDuff: that way lies madness :-)
15:33	nDuff	cemerick: ...by the way, I think I've had a message sitting in the moderation tool for clojure-tools for weeks.
15:33	cemerick	Oh? crap.
15:33	nDuff	s/tool/queue/
15:34	cemerick	nDuff: no pending messages…
15:34	nDuff	...hmm; bizarre.
15:35	borkdude	haha, I took a look at appspot and there's still an app I made once on there: http://borkent-app2.appspot.com/ (self tutoring for learning hebrew)
15:35	amalloy	maybe you accidentally sent it to rec.games.baseball, and they're still scratching their heads over it
15:35	nDuff	cemerick: ...on a related note, would you have any experience worth sharing on the madness in question, or would that be a good topic for a mailing list post? :)
15:35	cemerick	nDuff: do resend :-)
15:36	borkdude	unfinished though
15:36	cemerick	nDuff: ML, please. I'm vaguely aware of your struggling with classloaders and REPLs and atlassian kit, and god knows what else, but I've not been following closely. :-)
15:38	borkdude	I guess this page needs updating with respect to dynamic vars: http://clojure.org/lisps
15:39	solussd	cemerick: https://www.refheap.com/paste/4fd25367e4b0e13ddf181b31
15:39	solussd	pardon the indentatino
15:39	muhoo	overall friend demo does sound like a good idea, but the workflows do have their differences
15:43	cemerick	muhoo: yeah, each workflow would have its own form, etc.
15:45	borkdude	technomancy issue 592 can be solved now just by changing one character in project.clj :-)
15:50	muhoo	well i won't promise anything, but i suppose i might hack something like that up at some point (using noir)
15:51	cemerick	muhoo: oh, I wasn't suggesting that you do it :-)
15:52	muhoo	i kinda got halfway there when i was trying to figure out how this stuff worked anyway
15:52	muhoo	i implemented interactive-form, openid, and now this step2. the only one i didn't do is basic
15:52	borkdude	,(or "friend" "foe")
15:52	clojurebot	"friend"
15:53	muhoo	the mock_app.clj was clear enough
16:11	dnolen	getting close to optimizing away protocol dispatch in CLJS when we have type information ...
16:16	lpetit	nDuff: interested in reading you. We may have related problems: when working on Counterclockwise itself with live REPL, I have to do some tricks to get the code sent to nrepl server evaluated from the right class loader context.
16:17	lpetit	nDuff: I was thinking about adding some mechanism to deal with that, so I was reluctant to do it in an OSGi-specific only way
16:21	nDuff	lpetit: nod. Atlassian/Felix does things differently enough from Eclipse/Equinox that I haven't been able to use clojure.osgi directly, but it's been very useful reading.
16:21	lpetit	nDuff: so if you can bring a second use case to the table, then it'll help find the right abstraction to not hardwire things in Counterclockwise anymore. Currently, it relies on clojure.osgi providing some OSGi specific way of getting a bundle's class loader and injecting it before evaluation (System's context class loader, RT.baseLoader .. -out of my head).
16:22	lpetit	nDuff: it is certainly possible, tho, to decomplect (yeah, really) those 2 concerns from clojure.osgi to make it more general: mechanism to inject classloader, being passed either the class loader to inject, or a function to be called to return the class loader
16:23	Bronsa	i'm wondering why we have both `second` and `fnext`
16:24	lpetit	nDuff: the only "concern" is that with clojure.osgi, changing the class loader is side effect full, like changing the current ns is. Though it may not be a concern, after all. One must just understand that it's changed for the current thread (like is also the current ns, after all, lol)
16:24	dnolen	Bronsa: the reasons are mostly historical I believe.
16:24	lpetit	,(source fnext)
16:24	clojurebot	Source not found
16:25	Bronsa	lpetit: they are exactly the same :)
16:25	Bronsa	dnolen: ok, thanks
16:25	lpetit	indeed
16:27	josteink	linked in, now last.fm
16:27	josteink	and then you have the whole "flame" malware with a whole new kind of md5 busting
16:27	josteink	this is a bad week for internet security
16:27	borkdude	o no, what if someone will scrobble a track of an artist I dislike!
16:27	Bronsa	lol
16:28	josteink	borkdude: you say that now
16:28	devn	lol lastfm got hacked too? jesus. another password i have to change.
16:28	josteink	borkdude: but when nicki minaj will be forever stitched to your last.fm profile, following justin bieber
16:28	neotyk	stuarthalloway: clj-1009, should keywords be printed in header, rather than it's name?
16:28	josteink	you WILL know better
16:28	borkdude	:-)
16:29	devn	This really needs to get changed: http://richhickey.github.com/clojure-contrib/
16:29	lazybot	Nooooo, that's so out of date! Please see instead http://clojure.github.com/clojure-contrib/ and try to stop linking to rich's repo.
16:30	stuarthalloway	neotyk: I don't understand...
16:31	neotyk	stuarthalloway: (print-table [{:a 1}])
16:31	neotyk	\| :a \|
16:31	neotyk	+----+
16:31	neotyk	\| 1 \|
16:31	stuarthalloway	is that desired, or current?
16:31	neotyk	current
16:32	amalloy	printing the name instead sounds disastrous, neotyk. if you want to print something other than a keyword, don't give it a keyword?
16:33	devn	re: stuart sierra's post on the list about additions to clojure.string -- would it not be possible make a clojure.string.format or something that lives out in the world as a "contrib" library?
16:34	stuarthalloway	devn: sure
16:34	stuarthalloway	neotyk: that looks right to me -- name cuts off options
16:34	neotyk	stuarthalloway: just asking
16:35	stuarthalloway	neotyk: np -- opinions most welcome
16:35	neotyk	amalloy: I see your point
16:35	devn	stuarthalloway: btw, too bad you had to take off after your talk at euroclojure, would have liked to chat
16:35	devn	oh well, the next clojure conf/j
16:36	stuarthalloway	devn: agreed -- my spousal travel vouchers expired :-)
16:36	devn	stuarthalloway: :D
16:36	stuarthalloway	devn: I will be at StrangeLoop
16:37	devn	stuarthalloway: Well then, see you there. :)
16:38	borkdude	"Clojure, the cool newcomer" -> http://files.zeroturnaround.com/developer-productivity-report/zeroturnaround-developer-productivity-report-2012.pdf
16:47	solussd	cemerick: do you know how to 'select' the sibling of an element using enlive templates? e.g. If I had <label …> <input .. >, how to select input right after label?
16:50	michaelr525	he;;o
16:50	Raynes	h; ;here
16:51	michaelr525	whats up?
17:14	zaka	Is there a way to create an (lazy)-sequence that adds the previous item from the sequence + 3. The result is for example: [1 4 7 10 13 16] or [2 5 8 11[ (doesn't matter if it return an array, list or something else, just as long it iterable)
17:15	nDuff	Is there an equivalent to RT.var(String ns, String name) that fails, rather than returning a new unbound var, if nothing exists in the desired namespace?
17:15	gfredericks	zaka: yes.
17:16	gfredericks	zaka: your example looks like you're just incrementing everything?
17:16	zaka	gfrederiks: yes
17:17	gfredericks	,(map inc [1 4 7 10 13 16])
17:17	clojurebot	(2 5 8 11 14 ...)
17:17	jaley	zaka: (defn foo [x] (iterate (partial + 3) x)) ;; something like that?
17:17	amalloy	gfredericks: he's asking for iterate, not map
17:18	gfredericks	amalloy: I must be drunk because I have no idea what's going on here
17:18	gfredericks	oh I see
17:18	gfredericks	ha yeah I misread that
17:53	cemerick	solussd: try [(left [:label])]
17:53	cemerick	solussd: Haven't had a chance to look at your paste yet, will have to wait until tonight/tomorrow. Maybe start a ML thread on it if you get impatient. :-)
18:05	gfredericks	man this monet library
18:06	livingston	if I'm trying to call a specific java function that I can pass a null pointer to but it wants a null X and not a null Y I tried type-hinting ^X nil in the call, but does that not actually work as I expect?
18:07	amalloy	livingston: you need to provide enough hints to resolve the entire thing statically, or all the hints get ignored (so specifically, you're probably missing a hint on the target object)
18:07	amalloy	(.foo ^Target target ^X nil) should work
18:07	livingston	OH I didn't know that
18:08	amalloy	or...can you put metadata on nil? you might need to do something wonky like (let [^X x nil] (.foo ^Target target x))
18:11	livingston	hrm... that didn't work either.
18:11	gfredericks	amalloy: surely you can't put metadata on nil??
18:11	lazybot	gfredericks: Definitely not.
18:12	gfredericks	ain't it java null?
18:12	livingston	basically I'm trying to call the airity 3 form of contains here: http://jena.apache.org/documentation/javadoc/jena/com/hp/hpl/jena/rdf/model/Model.html
18:12	gfredericks	oh you mean the symbol
18:12	amalloy	gfredericks: i don't think nil is a symbol, is the point
18:12	amalloy	livingston: you shouldn't need any hint at all
18:12	amalloy	&(class (read-string "nil"))
18:12	lazybot	⇒ nil
18:13	gfredericks	amalloy: ooh sorry you were questioning in the opposite direction
18:13	gfredericks	I should stop talking for the remainder of the evening
18:13	livingston	I wouldn't think so, the call is unambiguous, I think I'm getting the right call signature, but the library is complaining about the "type" of null thing I passed in.
18:14	amalloy	library sucks, bro
18:14	amalloy	if you get as far as calling the method, and something inside there complains, ain't nothing clojure can do to pass a nil that is somehow different
18:14	livingston	I hinted to the thing in the call signature ^RDFNode nil and it says null Literals not allowed (literals are a subtype...)
18:14	livingston	ok
18:14	gfredericks	surely the lib couldn't be examining the type of a null at runtime
18:15	gtrak	I don't have to use deref on a dynamic var do I, and why not?
18:15	livingston	that's what I figured
18:15	gfredericks	gtrak: cuz the value is always available so that'd be needless
18:15	gfredericks	or wait
18:15	gfredericks	no
18:15	gtrak	haha, right?
18:15	gfredericks	the reason is because that would be annoying
18:16	gfredericks	it shouldn't be changing out from under you also
18:16	gfredericks	at least not any more than any other var would
18:16	pepijndevos	Can you modify the play speed of a buffer in overtone?
18:16	gtrak	ah, indeed, since it's thread-local the value would stay the same for the execution of your function
18:16	gfredericks	yeps
18:17	livingston	luckily there is an arity 2 form of the function that is the same as nil for the third and that seems right. although the docs say I can do what I've tried so I was just trying to figure out why I couldn't.
18:17	livingston	thanks
18:26	livingston	if I want to release a maven-ified library, and open source it, what seems to the be the community preferred locations? code on github, and pom/jar where?
18:26	gfredericks	clojars
18:28	livingston	I just have to create an account and it'll let me push there, no questions?
18:28	technomancy	the current implementation accepts nearly anything
18:29	technomancy	in the future you will have to declare a license and stuff to push to the releases repo
18:29	livingston	i'll go EPL - so I'm compatible with clojure itself.
18:31	livingston	does the central repository mirror clojars now? I've notice clojure itself only publishes there now doesn't it?
18:36	amalloy	maven central does not now, and almost certainly never will, mirror clojars
19:05	dnolen	core.logic on CLJS has achieved Racket performance!
19:06	amalloy	you're some kind of superhuman, dnolen
19:07	dnolen	12ms-15ms on V8 to solve zebrao
19:08	dnolen	fingers crossed that PHMs can get us to ~3-4ms
19:35	scottj	if you want to make sure you always have a copy of jars your projects use (in case they get removed from repos) is the best way to create a local maven repo? any special tools there people like or just follow general docs?
19:41	technomancy	scottj: probably best to run a nexus server
20:06	livingston	scottj: your local maven repo should keep them indefinitely, unless you instruct it to delete them.
20:07	aperiodic	scottj: you could also stick them in an s3-maven-wagon (I've found this to be less pain than running a nexus server)
20:07	livingston	if it's just you on one machine periodically backing up ~/.m2/repository is probably good enough, but like technomancy said a server would be even more convenient if you use a couple of machines etc.
20:10	livingston	huh, I've heard the term maven wagon before, but just always glossed over it, when did they start that?
20:11	aperiodic	i think wagon is the maven repo storage protocol?
20:12	livingston	right but I didn't know you could kind of off load them places, eh, that'll go back to the back of my brain to be forgotten again
20:13	aperiodic	yeah, i wrote this shell script that just apes what the s3-wagon-private plugin does to deploy things to an s3 bucket, and it seems to be sufficient
20:14	technomancy	nexus is more convenient in the long haul; just more pain up-front
20:14	frozenlock	Is there a command to open in another software? I would like to save an html file and immediatly after make it open in the user's default browser.
20:14	aperiodic	why's that?
20:14	technomancy	it caches transparently and lets you check only a single repo
20:15	technomancy	publishing a couple artifacts to an s3 repo is easy, but making sure everything your project depends upon is in s3 is more work
20:15	technomancy	frozenlock: try clojure.java.browse
20:15	frozenlock	Will do immediatly, thanks!
20:16	technomancy	sometimes it goofs up the browser selection, but it nearly always opens something
20:16	livingston	fishes through your history, picks a good one ...
20:39	johnmn3	so, there is not already a clj wrapper for java.util.zip.* or some other zip variant?
20:40	johnmn3	I see clj-msgpack, but that's not really zipping
20:41	johnmn3	and I'd like to use a zip file as a database to persist clojure datastructures. Being able to open it with a unzip program would be neat, for my purposes, I think.
20:43	scottj	johnmn3: not sure if it helps, https://github.com/mikejs/ring-gzip-middleware/blob/master/src/ring/middleware/gzip.clj
21:01	frozenlock	johnmn3: I've done something that you might find useful: https://gist.github.com/2898908
21:01	frozenlock	(spit-to-zip "test.zip" "test.txt" "asdfadsfas") will result in a zipped file named test.zip, with inside it a file named test.txt and with the content "asdfadsfas"
21:02	frozenlock	And to spit to a zip with more than a file: (spit-to-zip "test.zip" "test.txt" "asdfadsfas" "test2.txt" "adsfasdfdsa")
21:03	johnmn3	frozenlock: that's fantastic. simple and works as advertised.
21:03	johnmn3	going about making a slurp-from-zip, would it look relatively the same, but with .getNextEntry?
21:04	frozenlock	Probably. Might need ZipInputStream too.
21:05	frozenlock	Feel free to post a link here if you ever make this slurp-from-zip :)
21:05	johnmn3	so for anyone who was here when I was talking about using Chris Granger's simpledb (not the amazon simpledb) for persisting clojure data
21:05	johnmn3	https://www.refheap.com/paste/3063
21:06	johnmn3	I had the idea to just add a watcher to the atom and, whenever an event triggers, to write the atom to disk.
21:06	johnmn3	but I realized that that doesn't work well with io
21:06	septomin	this is the flat file thing?
21:07	frozenlock	johnmn3: Y u no like congomongo?
21:07	johnmn3	(map update-atom-with-write (range 0 10000000)) would be detrimental
21:07	johnmn3	frozenlock: oh yea, I forgot about that. I should def look into that.
21:09	johnmn3	So, I built some logic into it that basically says: "if there has been a write in the last second, wait another second and try again. If there is still contention, give up"
21:09	johnmn3	and then there is also the original timer that persists automatically every 10 seconds (from granger's original 5 min)
21:09	septomin	eventual safety!
21:09	johnmn3	:)
21:10	johnmn3	but it's still not atomic
21:11	johnmn3	and I was thinking about gaurding writes with a ref of some kinds... or maybe I could build some journaling into the atom's map, so that operations can be rolled back if they fail.
21:11	johnmn3	but yea, I should check out congomongo
21:12	johnmn3	one benefit I was thinking, with the current way, was that I could just crack it open with a zip utility and then there clojure datastructures are right there, easy to troubleshoot.
21:12	johnmn3	I'm aware that I'm reinventing the rule.. but my needs are very minimal.
21:13	johnmn3	er, s/rule/wheel
21:13	septomin	i mean if performance is an issue at all it seems like you should use something real
21:13	septomin	but this looks nice for very early stage stuff
21:14	septomin	are your files even large enough that you need to bother zipping them?
21:17	johnmn3	it might just be one file.. that's all I need to persist a whole environment, right? Yea, the file could keep growing and growing to be a few megs, over time.
21:18	johnmn3	though it might be more efficient to store in multiple files in the zip, so as to create a random access situation, rather than re-writing the whole thing every time.
21:18	johnmn3	In which case, each write could be a delta and I could have versioning.. hey look, I just invented chouchdb
21:19	johnmn3	yea, I might as well go with a simple but strong db lib like congomongo
21:19	frozenlock	I'm trying clojure.java.browse/browse-url. It works fine for... well a URL, but how can I make it work with a local file?
21:20	septomin	file:///path ?
21:22	frozenlock	No, it chokes on the path, saying it's not a URI
21:52	frozenlock	Damnit
21:52	frozenlock	I'm such an idiot
21:53	frozenlock	All this time I was trying to open a not-yet-existing file.
22:10	johnmn3	congo mongo is giving me errors
22:11	johnmn3	but I used pomegranate to load it...
22:14	johnmn3	test2.core=> (insert! :robots {:name "robby"}) => Jun 08, 2012 10:07:37 PM com.mongodb.DBTCPConnector fetchMaxBsonObjectSize WARNING: Exception determining maxBSON size using0java.io.IOException: couldn't connect to [/127.0.0.1:27017] bc:java.net.ConnectException: Connection refused: connect
22:20	frozenlock	Is mongoDB running correctly?
22:32	johnmn3	mmmm
22:32	johnmn3	right
22:33	johnmn3	so you're saying "(make-connection ... doesn't actually stand the server side up, and I have to go download mongo seperately? :/
22:35	johnmn3	I think I'm more inclined to just use the spit-to-zip you mentioned. That has the added benefit that I can peer into it as a zip file.
22:52	eslick_	Anyone know of SF-area consulting firms with clojure experience / expertise?
22:56	semperos	https://groups.google.com/forum/?fromgroups#!searchin/clojure/clojurescript$20extend/clojure/-epqvr3pPPg/NPYztkPGWZEJ
22:56	semperos	ClojureScript has extend-type, but no extend
22:56	semperos	is there any equivalent mechanism for passing around default implementations for protocols?
22:56	johnmn3	frozenlock: it looks like the contents are just being dumped into the out, rather than going into the zip file
22:59	semperos	johnmn3: out can be dynamically rebound to point where you want it
22:59	semperos	http://richhickey.github.com/clojure/clojure.core-api.html#clojure.core/out
22:59	lazybot	Nooooo, that's so out of date! Please see instead http://clojure.github.com/clojure/clojure.core-api.html#clojure.core/out and try to stop linking to rich's repo.
22:59	semperos	there's also the built-in with-out-str when you just want to capture things sent to out as a string
23:03	johnmn3	but I don't see a setter method on ZipEntry, so I'm not sure how to put the data in it.
23:08	johnmn3	I think you have to call .write on the ZipOutputStream
23:09	johnmn3	ugh, manually pouring bytes into arrays using offsets just feels so... uncivilized
23:20	aperiodic	yeah, it sucks to have to think about array indices
23:20	frozenlock	johnmn3: Sorry about that, I gave you an older version.. hang on
23:21	frozenlock	https://gist.github.com/2898908 try this one
23:21	frozenlock	(I spent almost a day trying to get this work the first time)
23:21	frozenlock	printer and buffer will be my death :)
23:27	johnmn3	trying
23:30	johnmn3	worked!
23:30	johnmn3	fantastic
23:31	francis	is their a fix for the lein repl so when I use M-p it grabs the last definition I used rather than the line? Or am I going to have to write it