#clojure – 2010-06-07

2010-06-07

00:00	rava	ty
00:01	trptcolin	sure
00:38	_brian2_	hi, is there a way to convert a string, say "{:foo "bar"}" to its clojure form?
00:39	tomoj	,(doc read-string)
00:39	clojurebot	"([s]); Reads one object from the string s"
00:39	_brian2_	ok, thnks!
01:39	jartur	Hello, everyone. I have another thrilling question.
01:39	jartur	Does anybody know of a way to connect REPL to the running app?
01:39	jartur	Is there a way at all?
01:49	jartur	Looks like No =)
01:51	jartur	Okay
01:51	jartur	clojure.contrib.server-socket should be it
02:00	jartur	It's so easy, I feel embarrassed. But google didn't want to show me the way.
04:27	LauJensen	Morning all
04:39	robink	LauJensen: Good morning.
04:43	TheBusby	morning
04:46	_exterm	morning
05:03	LauJensen	There's still a few seats left for the next Conj Labs (http://conj-labs.eu) so let me know if you're interested
05:23	esj	Morning Good People.
05:35	LauJensen	lypanov: Where there's a will there's a way
05:35	LauJensen	:)
05:36	lpetit	Morning all
05:36	LauJensen	Morning Mr. Petit
05:37	esj	wotcha.
05:37	lpetit	counterclockwise users in the room ?
05:42	lpetit	apparently no ... :-(
05:45	esj	lpetit: I used it briefly but the lack of paredit drove me back to the dark god Emacs.
05:48	zmila	lpetit, me :) as usual
05:49	lpetit	esj: the force is with counterclockwise nowadays = 80% of paredit has been ported to it!
05:50	lpetit	esj: It's called "Structural Editing Mode" (because paredit means nothing to eclipse users), and comes in two flavors: "default" (does not get in your way, e.g. does not prevent you from deleting single open or close paren), and "advanced mode" (my preferred) which tries hard (as paredit does) to prevent you from breaking the structure of your code
05:50	lpetit	zmila: ! :)
05:50	lpetit	zmila: did you update to the last version ?
05:50	esj	lpetit: Great news! I'll give it a spin then :)
05:51	zmila	i have 57.stable
05:52	lpetit	zmila: if you're a windows user and are annoyed by the bug on windows concerning "cannot delete file" .., I encourage you to upgrade to 58.stable and give feedback !
05:54	zmila	lpetit, i did not see this bug
05:54	zmila	till now I deleted no clj file :) only adding
05:55	lpetit	zmila: bug occured when you launch the REPL, and while the REPL is running, you edit and save clojure files
05:56	LauJensen	Im very fond of Emacs, especially with par-edit disabled
05:57	lpetit	LauJensen: ccw paredit tries to be more user friendly in the "default" mode
05:59	LauJensen	But its like training wheels - Its a habit you need to grow out of
06:00	LauJensen	How have you improved it lpetit ?
06:02	lpetit	LauJensen: separated the features which "surprise you". e.g. when you try to delete a paren, intentionally wanting to temporarily break the structure of your code, but paredit doesn't let you do so, "for your own good". In default mode, this command is not active.
06:03	lpetit	LauJensen: and there's a note in the status bar indicating if you're in the default mode or in the status mode. And jumping from one to the other can be made by keyboard shortcut, entry in the contextual menu of the editor, or by clicking on the message on the status bar.
06:04	lpetit	LauJensen: stopped working on porting paredit features yet. Working on improving the overall perf.
06:04	LauJensen	Ah ok, especially the 'non babysitter feature' is nice
06:05	lypanov	LauJensen: if you have another session at a later date i'll have better luck :)
06:05	esj	LauJensen: i like my babysitter, she's cute
06:05	esj	very curvy
06:07	LauJensen	lypanov: We are working on some, still Europe but perhaps a little different in scope :)
06:09	lypanov	hopefully our company will grow a bit and i'll be able to join. at this point i'm still 100% on forging ahead full speed. no time for clojure :(
06:09	lypanov	stuck in RoR land. (though its not that bad with v 3)
06:10	LauJensen	lypanov: I've never dug into RoR - Can you recommend any projects for scouting out the basics?
06:11	lypanov	LauJensen: ror 2 stinks frankly. but there are a few good presentations on 3 by the lead. not sure about canonical good projects alas.
06:12	lypanov	ror 3 is still not out, so not much yet available on it alas. :(
06:12	lypanov	its the first version that doesn't make we want to weep though.
06:13	lypanov	LauJensen: if i spot a talk that doesn't assume existing ror 2 knowledge will pass it along. can't find anything with a very quick google.
06:14	lypanov	anyone know how clojure in clojure is coming along?
06:14	LauJensen	k
06:14	LauJensen	lypanov: no idea, cgrand might have a clue
06:14	lypanov	i'd still love to replace jquery + js with lots of clojure code.
06:15	LauJensen	lypanov: via scripjure, clojurescript, what?
06:17	lypanov	LauJensen: via clojure itself. currently doesn't exist.
06:18	lypanov	i like the gwt model of single language for server / client.
06:18	lypanov	having the same for clojure would make web app devel so much more fun.
06:19	lypanov	LauJensen: basically i'd want aot js compilation
06:20	LauJensen	k
06:33	lypanov	does anyone have any opinions / knowledge on good visually orientated code editors much like the self environment but a bit more modern?
06:34	lypanov	being the complete apple ho that i am, i'd just love to hack out clojure on an ipad.
06:34	LauJensen	haha
06:34	lypanov	(html5, touch, no app store)
06:34	LauJensen	I would like to shake your hand sir, you're the first I've ever met who wants an Ipad
06:35	LauJensen	"Uuuh I wish my Iphone was too big for my pocket and that I was unable to make calls from it"
06:35	LauJensen	Or that my websites looked like this: http://bradcolbow.com/archive/view/the_brads_test_how_your_site_looks_on_the_ipad/
06:35	lypanov	LauJensen: i was in a room last week with 5 of them.. and this was in .nl. freaky.
06:36	unfo-	LauJensen, ahahahhaha :D
06:36	lypanov	LauJensen: it looks terrible at an angle, but straight on, i can't see the smudges.
06:36	lypanov	(nor the light, but i'm used to reflective screens now also)
06:37	LauJensen	seriously, Steve must be laughing his butt off that people are actually buying his products - I wish I could be paid to have so much fun
06:37	lypanov	you have a mac right?
06:37	LauJensen	No - I've gotten rid of all of my Mac gear
06:37	lypanov	ah. ;)
06:38	lypanov	honestly, i understand where you're coming from. i just wish that someone else would have products worth buying.
06:38	lypanov	(not talking about laptops here)
06:38	LauJensen	hmm, actually ipad did win one comparison, I wasnt aware before now
06:38	LauJensen	http://techcrunch.com/2010/01/30/ipad-v-a-rock/
06:38	lypanov	:P that ones great.
06:39	lypanov	ah, saw a better one. this ones silly.
06:39	lypanov	the other one said "battery: unlimited"
06:39	eevar2	apple's releasing new stuff today, no?
06:39	lypanov	eevar2: yeah but not h/w.
06:39	lypanov	well, other than iphone.
06:39	lypanov	anyway. as i said, i'm an apple ho.
06:39	lypanov	i verily admit to insanity.
06:39	LauJensen	I wish I could get my hands on the prototype: http://img.hexus.net/v2/news/apple/ipad-joke-1.jpg
06:40	lypanov	LauJensen: have you used one?
06:40	LauJensen	No of course not
06:40	lypanov	k.
06:40	LauJensen	Do I look like an iTard? :)
06:41	lypanov	you used to be, did you grow a beard recently? :)
06:41	LauJensen	hehe
06:41	lypanov	anyway. i regret my mention of apple ho.
06:41	LauJensen	Why, did I hurt your feelings?
06:42	lypanov	should i say "OMG I WANT A CLOJURE APP ON MAH ANDROID PAD :("
06:42	lypanov	LauJensen: naah, just that it got off topic.
06:42	LauJensen	Ok :)
06:42	lypanov	i was more interested in actual movement on touch interfaces for programming.
06:42	lypanov	i think they could really kick ass.
06:42	lypanov	syntax free langs are the only ones that its possible with.
06:43	LauJensen	Well, in terms of development, anything from Apple is truely uninteresting
06:43	lypanov	nod but i'm talking about a website.
06:43	lypanov	and the ipad is the only readily available large multi touch device atm.
06:43	lypanov	but lets say something else does come out, and i'm talking about that.
06:43	lypanov	lets call it the LauPad
06:44	lypanov	i still think that visual touch orientated development is a progression that would afford many things we currently can't do easily without a fair bit of interaction.
06:44	eevar2	the iphone store is supposedly a lot better than android et al. in terms of making actual sales
06:45	lypanov	yeah. but as i said, i want a clojure dev env.
06:45	LauJensen	eevar2: Due to the large user base no doubt
06:45	lypanov	i'm not interested in selling it. nor in native apps.
06:45	lypanov	apple would block any app i wrote anyway.
06:45	eevar2	LauJensen: and willingness to pay
06:46	lypanov	i won't touch a non open sourced editor again in a few life times, and wouldn't want to force it upon others either.
06:46	lypanov	imagine where textmate could have been by now if it was open.
06:47	lypanov	anyway. i'll go off and look for some papers. someone must have done something interesting in the area.
06:48	eevar2	LauJensen: then again, the people telling me android users won't buy stuff are into add based revenue ;)
06:48	lypanov	ad based < closed source < open source
06:48	lypanov	nothing is worse.
06:48	eevar2	lypanov: and gpl < mit/bsd
06:49	lypanov	yup.
06:49	lypanov	well, for applications sometimes i disagree.
06:49	lypanov	but plugins must always be bsd/mit and libs also.
06:50	lypanov	even agpl is nice. i just wouldn't ever want to be the consumer of it ;)
06:50	lypanov	i'm gonna disappear as #clojure hasn't been more offtopic in a while :P
07:41	rhickey	anyone with examples of Java APIs that require your class have a particularly named static method?
07:43	telmich	hello
07:44	LauJensen	hi telmich
07:47	karmazilla	rhickey: in spring, you can instantiate beans by specifying a static factory method
07:48	rhickey	karmazilla: must it have a particular name?
07:48	karmazilla	rhickey: no
07:49	rhickey	I'm interested in the case where an API says, you must have a static method called "___"
07:51	esj	clojure.lang.IPersistentCollection calls for equiv. Any pointers on what this is and how it differs from Object.equals ?
07:54	powr-toc	I've just update slime to 20100404 (from elpa) and tab completion has stopped working in the REPL buffer... It just says "No dynamic expansion found for user> foo"
07:54	powr-toc	any idea how to fix it?
07:56	telmich	I'm wondering, how to get the zip-reader running
07:57	telmich	I'm trying to get this file running http://github.com/panter/edu/blob/master/clojure/imdb/src/imdb/movies_by_years.clj and installed clojure via lein
07:58	telmich	when removing the comment and having downloaded the required files before, I get the error "java.lang.Exception: Unable to resolve symbol: zip-reader in this context (NO_SOURCE_FILE:50)"
07:59	telmich	where lein is downloaded from http://github.com/technomancy/leiningen/raw/stable/bin/lein
08:44	rhickey	cemerick: got any examples of Java APIs that require your class have a particularly named static method?
08:46	cemerick	hrm
08:48	rhickey	cemerick: you were suggesting injecting a static method with a particular name might interfere with interop requirements
08:49	cemerick	oh, I was just saying that if you also want to allow the definition of static factory fns, then you'd want to push any automatically-generated statics into some "safe" mangled territory, e.g. Foo/_record, Foo/_ctor, etc.
08:50	rhickey	cemerick: so they have to be then wrapped in order to not be ugly?
08:51	cemerick	rhickey: we're still in the context of the map factory fn?
08:51	rhickey	yes
08:53	cemerick	no, the "wrapping" would be there so as to present a consistent interface, and get out of the territory of Foo-from, etc.
08:54	cemerick	The static factory fn that happens to be a static method is an impl detail.
08:56	rhickey	cemerick: unless you have a recipe for avoiding ongoing non-modular accumulation in (record ::Foo ...), it's definitely on the ropes
08:56	rhickey	its not an implementation detail, it's an implementation impediment
08:57	cemerick	I missed what the modularity issue is. (record Foo ...) just becomes a macro to (Foo/_record ...), no?
08:58	rhickey	if macro, can't be applied
08:58	cemerick	ah, right
08:58	rhickey	can't be used in data-driven manner
08:58	cemerick	well, that's separate from this modularity issue
08:59	rhickey	not really, if you didn't care about apply or data-driven use the macro wouldn't have modularity issues
09:01	cemerick	back in a sec
09:51	cemerick	rhickey: from left field: perhaps the most sensible thing would be to determine how factory fns should work in general, and then figure out how these various default/standard factory fns fit into that.
09:54	rhickey	cemerick: I have pages of notes around this already, just trying to shake out the (record Foo ...) option. If I thought it was the nicest thing ever but couldn't implement well, it won't go in
09:55	rhickey	conceptually, they are very simple - they are functions that call the ctor, but take arguments that differ from the ctor's
09:56	cemerick	rhickey: I understand, but stepping back, those "userland" factory fns presumably wouldn't fit into the (record Foo ...) regime, and they wouldn't fit into a Foo-from option either. The key is not in the implementation, but in the UI, as it were.
09:57	rhickey	cemerick: what do you mean by userland?
09:57	cemerick	factory fns for which the impl is user-defined, as opposed to canned / provided by default
09:58	rhickey	cemerick: having the generated either target or be replaced by user code is easy
09:58	cemerick	couldn't parse that :-)
09:59	rhickey	it is easy to make the canned/provided version either target or be replaced by the user-impl
09:59	cemerick	ah
10:00	cemerick	sure
10:00	cemerick	what I'm driving at is that these canned factory fns should probably be handled just like those userland factories -- in terms of where the vars live (if any) where the static methods are (if any), etc.
10:05	hugod	rhickey: could the data-driven use be covered by (record (some form) ...) expanding to (invoke-factory-using-reflection (some form) ...)? or am I misunderstanding "data-driven"?
10:05	rhickey	hugod: there will be no reflection based things
10:06	rhickey	so data driven implies some sort of self registration by the record into some sort of dispatcher
10:08	chouser	to be modular, that implies the registration is held at the classloader? which suggests a static method of the class itself?
10:08	hugod	to prevent modularity issues, each namespace could maintain some registration data for the dispatcher?
10:09	rhickey	chouser: I'm not sure modular and dispatcher are compatible at all
10:09	a_strange_guy	just a stupid question: What was the original reason to switch from 'Foo as a fn living in a Var to 'Foo beeing a auto-imported java classname?
10:09	cemerick	a_strange_guy: only java.lang.* stuff is auto-imported.
10:10	rhickey	cemerick: no, defrecord Foo imports Foo
10:10	chouser	is modularity a general problem for multimethods dispatching on type?
10:10	cemerick	oh, didn't think a_strange_guy's Q was related to records, my bad :-/
10:11	rhickey	a_strange_guy: because deftype started generating named classes, so too confusing to have user.Foo and user/Foo
10:11	rhickey	chouser: yes
10:12	rhickey	chouser: but not inherent, as with factories, who's fn itself refers to the type
10:13	cemerick	back to basics, perhaps. Strawman time: https://gist.github.com/a455d0afd7478092b141
10:13	cemerick	apply is nice, but direct usage is probably 90% of the game, and is needed for the interop story anyways.
10:14	rhickey	cemerick: ick
10:14	a_strange_guy	bleh, dont like static factories
10:15	cemerick	rhickey: interesting, I thought that's what you were aiming for for user-defined factories anyway.
10:15	rhickey	a_strange_guy: vs what?
10:16	a_strange_guy	static factories and constructors mean that I have to re-eval not only the defrcord form but all it's users
10:16	a_strange_guy	else they keep producing the old version of the record
10:17	cemerick	without static factories and ctors, I think you're left with vanilla maps. :-)
10:17	rhickey	well, the dynamism is an important point, and a difference between (record Foo ...) and (record ::Foo ...)
10:18	rhickey	you can always build dynamic things on static, but not vice-versa
10:18	rhickey	so, I understand a_strange_guy's point
10:19	a_strange_guy	well i really don't think that static things are needed at all
10:19	a_strange_guy	except maybe for Interop
10:19	rhickey	a_strange_guy: what about performance?
10:20	a_strange_guy	is there one?
10:20	rhickey	a_strange_guy: one what?
10:20	a_strange_guy	calling a fn wrapping the ctor or calling the ctor directly?
10:21	rhickey	a_strange_guy: a perf difference? yes
10:22	a_strange_guy	meh, I put too much hope into the JIT ^^
10:23	rhickey	the JIT is awesome, but saying something is dynamic makes it volatile, which inhibits inlining
10:27	a_strange_guy	and what about direct binding?
10:28	rhickey	a_strange_guy: avoids that by taking the value out of the var, remembering it, and thus making it non-dynamic
10:29	a_strange_guy	this would be IMHO preferrable
10:29	a_strange_guy	making the default record construction dynamic
10:29	rhickey	a_strange_guy: same result, must re-eval
10:30	a_strange_guy	sure, but only where we have perf bottlenecks
10:31	a_strange_guy	making the default static seems like premature optimization
10:32	rhickey	a_strange_guy: I don't se what you mean by default. The dynamic must be built upon the static
10:32	rhickey	if you don't want to call the ctor, don't
10:32	rhickey	if it didn't exist, then what?
10:33	a_strange_guy	the way i understand the JIT is:
10:34	a_strange_guy	static -> full inlining
10:34	a_strange_guy	wrapper fn in a var -> no inlining
10:35	a_strange_guy	wrapper fn in a var + directly bound at callsite -> full inlining?
10:35	rhickey	only if there is something static to inline to
10:36	a_strange_guy	the wrapper fn would call the native ctor
10:36	rhickey	defrecord now provides an accessible named ctor. Don't use it if you want dynamism
10:36	rhickey	a_strange_guy: I understand all this, sheesh
10:37	rhickey	the class exists, it has the name you gave it, the constructor has the same name. These are baseline facts. You want a function, you can write one that calls these things. You want one generated by defrecord, propose a name
10:39	a_strange_guy	looking at the IRC logs i don't think that i could provide a better name
10:40	rhickey	not auto-importing the class and commandeering its name for the factory fn would preclude importing it
10:41	a_strange_guy	LISP-1: you only have one namespace, so use it wisely xD
10:42	Twey	Haha
10:43	a_strange_guy	my proposal would be to return to Foo beeing a fn in a var, but having metadata on the var to find the correct class
10:43	a_strange_guy	and some magic for extend and type-hints
10:44	a_strange_guy	but I don't think that this is thought throgh till the end
10:45	a_strange_guy	my english sucks :-)
10:49	a_strange_guy	that would mean that (Foo "some" "args") calls the fn dynamically,
10:49	cemerick	a_strange_guy: we've been there already
10:50	a_strange_guy	hmm
10:50	cemerick	that is, that's what records started with
10:50	Joreji	Hey guys, I have a java app from which I start a new thread which itself loads a clojure class. Is there some way to set the COMPILE_PATH variable before the class is being loaded (i.e. from the java code with something like Var.pushThreadBindings)
10:51	Joreji	?
10:51	cemerick	Joreji: why do you care about compile-path if you've already got a classfile?
10:52	a_strange_guy	jeah, i just liked that simplicity
10:54	cemerick	The current state is strictly simpler than if Foo were a fn. Now you have easy access to the generated class, and using it is just like using any other class.
10:55	rhickey	admittedly hosty and less dynamic
10:55	cgrand	on Foo/record and apply, what about not making Foo/record a static method but a static field pointing to a IFn? (this would require modifying the compiler to support calling a fn in a static field)
10:55	Joreji	cemerick: Well, I want other classes to be compiled before loading the class. Basically during start-up of clojure.
10:56	a_strange_guy	cgrand: there ist still a problem with redefining the defrecord
10:56	Joreji	I've modified clojure to do that, however I need a way to set the COMPILE_PATH to something different before the clojure start-up.
10:56	cemerick	a_strange_guy: the same "problem" exists with macros. shrug
10:56	rhickey	cgrand: it complicates the semantics of (Foo/bar ...) quite a bit
10:56	cemerick	Joreji: do you need to compile a class, or load a class?
10:57	Joreji	cemerick: compile.
10:57	cemerick	Joreji: this is in a deployment environment?
10:57	Joreji	cemerick: ... and then load, I guess.
10:57	Joreji	cemerick: It's on android.
10:58	Joreji	Not sure what you mean by deployment environment.
10:58	cemerick	Joreji: then I'm really confused -- don't you need to AOT-compile everything you want to get onto the device?
10:58	a_strange_guy	rhickey: is making Foo a fn in a var again completely out of question?
10:59	rhickey	cemerick: I've made that argument also (same re-eval as macros), but people expect otherwise. I feel we are approaching a real divide between people needing perf and those desiring dynamism at all times
10:59	Joreji	cemerick: No. Android has its own compiler for the dalvik vm which compiles java vm bytecode code to dalvik vm bytecode. I can thus do: .clj (clojure source) -> .class (compiled java) -> .dex (compiled dalvik)
10:59	bartj	reading russian data present in a mysql table and spewing it to the output, all I get is garbage
11:00	cemerick	Joreji: right, I understand that, but doesn't that workflow need to happen before the app is loaded onto the device?
11:00	rhickey	cemerick: putting a good knob/swithc on that is tricky
11:00	bartj	using clojure.contribl.sql -> does that take an encoding parameter?
11:01	Joreji	cemerick: Not really. All required components are available on the device, so it is possible to do it there.
11:01	cemerick	rhickey: I'm not particularly sympathetic to those that push for hardcore dynamism. That makes a lot of stuff really hard across the board, it seems (and again, you're tied up with macros anyway).
11:01	bartj	nope, none of the clojure.contrib.sql takes any encoding parameters?
11:01	a_strange_guy	it's more a divide between "perf as default and dynamism when needed" and "dynamic as default and perf when needed"
11:02	cemerick	Joreji: Interesting. Well, anyway, yes, you can use pushThreadBindings from Java. Grep the clojure source for how to do that. :-)
11:02	cemerick	(not a very common requirement)
11:03	rhickey	if people have to annotate every thing explicitly in order to be fast, then fast programs will be very ugly (see CL)
11:03	Joreji	cemerick: Hmm. I've already tried using: Var.pushThreadBindings(RT.map(Compiler.COMPILE_PATH, path));
11:03	Joreji	cemerick: Alas that didn't work, so I expected it to require clojure already being initialized.
11:04	cemerick	a_strange_guy: perf must be the default. No one puts up with slow anything for very long.
11:04	cemerick	Joreji: well, yes, clojure must be fully initialized for that to work, I'd presume.
11:04	cemerick	Although just referencing RT should do that.
11:04	candera	cemerick: Rubyists might say otherwise
11:05	cemerick	candera: exactly! :-P
11:05	a_strange_guy	cemerick: The reverse argument is also true
11:05	candera	:)
11:05	a_strange_guy	The python and ruby people rewrite stuff in C to make it fast
11:05	Joreji	cemerick: As in: import clojure.lang.RT ?
11:06	cemerick	a_strange_guy: no, it's not. You can always layer on dynamism if you really need it, as rhickey has said. You can't make a heavily-dynamic system as fast it it would be if it were static, no matter how hard you try.
11:06	candera	"Make easy things easy and hard things possible"
11:06	cemerick	Joreji: Probably on first reference (i.e. RT.map will init the class, not the import)
11:07	cemerick	though I suppose dalvik might be different in subtle ways
11:07	Twey	Python doesn't any more so much… you've got Cython and RPython and stuff instead
11:07	Twey	a_strange_guy: ^
11:09	hugod	Joreji: compile-path is initialised from the clojure.compile.path property - is that what you want?
11:10	a_strange_guy	cemerick: sure you cant make a dynamic system as fast as a static one, but that is a straw man. The goal is to provide the knobs to make a dynamic system more static, so that it can become as fast as a static system
11:12	a_strange_guy	and adding dynamic layers on a static system isn't always a good idea
11:12	cemerick	a_strange_guy: I'll refer to rhickey's comment about CL. Further, defaults matter a lot. You should not have to work harder to get a system that performs to its fullest. Advanced users that want über-dynamism can be expected to jump through a few hoops though.
11:13	cemerick	Besides, are you looking for macros to operate at runtime? If not, then referring to defrecord ctors aren't a problem either.
11:13	fogus	candera: I think that the Ruby guys have made a conscious decision to trade speed in execution for speed in development time.
11:14	a_strange_guy	there is a big difference between macros and static ctors
11:14	a_strange_guy	macros do stuff that can't be done at call time, only at compile time
11:14	Joreji	hugod: I'll try that out now, thanks for pointing it out!
11:15	Twey	fogus: False dichtomy
11:15	Twey	fogus: In doing so, they throw away the chance to do both
11:15	candera	fogus: Yep. My point was simply to counter cemerick's statement, as it's dependent on what you're doing. I'm actually happy to have a language that defaults more in the perf direction.
11:15	cemerick	a_strange_guy: the point is that if the macro definition changes (or if a defrecord class changes) you need to reload all your usages in order to get the new behaviour/classes.
11:17	fogus	Twey: It's only a false dichotomy if you think Ruby can be fast. ;-)
11:17	Twey	Haha.
11:17	Twey	It probably can, though. Haskell manages to be pretty fast, after all, and it's much higher-level than Ruby.
11:17	candera	Wasn't someone in here the other day claiming that JRuby was going to get faster than Clojure "pretty easily"?
11:17	fogus	Twey: But my point was not that severe. I think speed in development time is just as legitimate as speed of execution... thankfully Clojure allows both.
11:17	Twey	Aye
11:17	a_strange_guy	I just don't accept the "macros and ctors are similar cases" argument
11:18	a_strange_guy	normally your code isn't littered with macros everywhere
11:18	cemerick	heh
11:18	Chousuke	well
11:18	Chousuke	not with macros that YOU wrote :P
11:18	cemerick	I was going to say. :-)
11:19	a_strange_guy	having many defrecords in a namespace is realistic
11:19	cemerick	The whole point of them is that they are classes with fields, permitting host-esque performance. There's a tradeoff there compared to regular maps.
11:20	a_strange_guy	yeah
11:20	a_strange_guy	but you can't use regular maps instead
11:20	a_strange_guy	because they don't allow protocols
11:20	cemerick	right, because regular maps can only be dispatched dynamically
11:20	cemerick	that's what multimethods are for
11:21	a_strange_guy	Protocols are dispatched dynamically too...
11:21	cemerick	no, they're dispatched based on the static type of the first argument
11:22	a_strange_guy	s/dynamic type/static type/
11:22	sexpbot	Protocols are dispatched dynamically too...
11:24	lypanov	fogus: heya. just wondering if manning has said anything about epub editions of the clojure book? i only read on my iphone and it hates pdf's :(
11:24	fogus	lypanov: I think epub will be available when the book is finally published.
11:24	a_strange_guy	IMHO you shouldn't be punished to reloading your code if you wan't to do explorative programming with defrecords
11:24	candera	lypanov: I've been reading on iPod as well. It's not an awesome experience, no. But GoodReader's text mode does reasonably well with it.
11:25	lypanov	fogus: :( nod thought so.
11:25	fogus	lypanov: Sorry.
11:25	cemerick	a_strange_guy: punished?
11:25	lypanov	candera: it seemed to really mess up the text i was reading. example code was seriously messed up for example.
11:25	lypanov	maybe i should try latest stanza.
11:26	candera	lypanov: Yeah, I have to flip back over to the regular view for code.
11:26	lypanov	fogus: not your issue. just hope they change the way they publish meaps in the future.
11:26	a_strange_guy	cemerick: I fix a typo in a defrecord form
11:26	lypanov	if the dang ipad was here already i'd be reading :)
11:26	candera	lypanov: +1 on both
11:27	a_strange_guy	but have to reeval every defn where the type gets created
11:27	cemerick	a_strange_guy: again, same situation as with macros. Again, defrecord and its kin generate a class.
11:27	lypanov	candera: i even wrote a new app to do it. but the width of the content is just slightly too high to fit in landscape even with pixel tweaking.
11:27	lypanov	candera: (pdf reader with page turn via swipe)
11:28	lypanov	great for papers though. but i'd prefer an ipad or higher res iphone to most content on the iphone currently.
11:28	candera	lypanov: PDF on iPhone is definitely suboptimal. iPad would be perfect if a) I could fold it up to fit in my pocket, and b) someone other than Apple made it. :)
11:28	a_strange_guy	cemerick: most macros just wrap some forms in a thunk, and expand to a call to a fn
11:29	a_strange_guy	so you can actually fix most macros by just fixing the underliing fn
11:29	cemerick	some are like that yes -- many, many others are now
11:29	cemerick	not*
11:30	lypanov	candera: yeah... god the m$ thing looked beautiful.
11:30	a_strange_guy	I don't want to get in an argument here, but most people try to make macros fixable at runtime
11:30	cemerick	a_strange_guy: FWIW, absolutely no one here is against dynamic development. I think the consensus is that we'll accept a dent in dynamism for maximal perf, though.
11:31	cemerick	There's massive classes of macros that exist to do work at compile time rather than runtime. That's not a hit on them, that's their entire purpose.
11:33	a_strange_guy	what I am trying to say is that static constructors are a big pain in the ass in Javaland
11:34	cemerick	heh, along with all manner of other things! :-D
11:34	a_strange_guy	they try to work around that stuff with all the DI bullshit
11:34	cemerick	indeed
11:34	a_strange_guy	so why import static constructors fom Java
11:34	cemerick	Because we want to be fast.
11:35	a_strange_guy	xD
11:36	cemerick	There's lots to hate in the JVM world. Lots more to love IMO. One of the biggest parts to love is that the JVM is smokin' fast. If you didn't want to take advantage of that, then there's far, far less of a reason to use the JVM.
11:36	a_strange_guy	nobody stops you fom writing (new namespace.Foo <args>) if perf is needed
11:36	cemerick	Same with the CLR, BTW.
11:37	cemerick	Sure, but defaults matter. You line everyone up, and they'll care about perf far, far, far more than development-time dynamism.
11:37	a_strange_guy	actually 'new could do some magic and allow (new Foo <args>)
11:37	cemerick	Again, macros present the same issues.
11:37	cemerick	Oy.
11:38	cemerick	magic == indirection
11:38	a_strange_guy	indirection == bad?
11:39	a_strange_guy	actually this would seem logical
11:39	cemerick	I remember when I suggested (a long time ago) to rhickey that some conditional be added to new. I got taken to school. :-D
11:40	cemerick	indirection has a perf penalty
11:40	Twey	(= 'indirection 'abstraction)
11:40	cemerick	The language should not get in people's way.
11:40	a_strange_guy	'new is a special form that allows fast and hosty object creation
11:40	Twey	And abstraction is good
11:40	cemerick	Twey: actually, that's not true. Macros are a good abstraction, but need not introduce any indirection.
11:41	Twey	cemerick: Macros are indirection by definition
11:41	Twey	They introduce an extra layer between the code you write and what actually happens
11:41	cemerick	only at compile time :-)
11:41	candera	"There is no problem in CS that cannot be solved by adding a layer of indirection. There is no problem in performance engineering that cannot be solved by removing one."
11:41	a_strange_guy	new could do this 'magic' at compiletime
11:41	Twey	candera: Hehe
11:42	a_strange_guy	if the first form is not a class but a Var
11:42	Twey	In some cases, abstraction can lead to better performance
11:42	Twey	Good abstractions add semantic information to operations, which allows the compiler to make optimisations based on it
11:42	Twey	Also, see lazy evaluation
11:43	a_strange_guy	ill sketch my idea up...
11:44	cemerick	a_strange_guy: so you're going to propose radically changing one of the core host interop operators?
11:45	a_strange_guy	there is no change, 'new still performs fast, hosty and direct object creation
11:46	cemerick	a_strange_guy: in any case, my hat's off to you. Good luck. :-)
11:50	a_strange_guy	no offence taken
11:54	hugod	cgrand: to run the enlive tests, I had to add clojure-contrib as a dev-dependency in project.clj
11:55	cemerick	contrib isn't a direct dep anyway?
11:55	cemerick	hugod: feh, nm, I thought we were in #pallet :-P
11:56	cgrand	hugod: thanks, I should update the tests to use clojure.test too
11:58	rhickey	poll: if you had to qualify a defn with some metadata tag in order for it to have these characteristics: direct-linked, could take/return primitives, non-closure, no self-reference as object/HOF (self-call ok), what tag would you use?
11:58	rhickey	:static?
11:58	rhickey	:fast?
11:59	AWizzArd	:static is okay, and it will remind people of static typing
11:59	cgrand	:static
12:00	a_strange_guy	:static
12:01	rhickey	note that you could still use the resulting fn as a HOF/IFn
12:01	a_strange_guy	although i wouldn't use defn to define it
12:01	AWizzArd	I would be glad about any optional information that is known at compile time.
12:02	candera	I've always thought static was a horrible keyword. The semantics for those of us who have been doing C-family programming for 20 years are right, but in and of itself I think there must be something better.
12:03	cgrand	rhickey: is that an asnwer to my kludgy defhintedfn?
12:03	rhickey	cgrand: perhaps, I'm measuring the difficulty now. But it covers a lot of things, including the 'knobs' for direct-linking, etc
12:03	hiredman	:metal
12:04	hugod	:native
12:04	candera	:bare?
12:05	a_strange_guy	I would like :native, but that would confuse Java-people
12:06	a_strange_guy	so :static or :statically-linked-and-really-fast-do-not-use-if-not-needed xD
12:06	Twey	Haha
12:06	AWizzArd	Well, I think this is all about static information, that won't change at runtime, and are known at compile time.
12:07	rhickey	also static as in the opposite of dynamic
12:07	AWizzArd	As long this is optional it's great.
12:07	rhickey	it would be an important difference in direct linking, moving the decision to the fn definition and away from the caller. Not that you have control over that right now, that's the current architecture
12:09	a_strange_guy	meh...
12:09	rhickey	cgrand: I've had statics and defstatic on the todo list for a while, this would unify several things and deliver in the context of fns.
12:09	AWizzArd	Whatever the key is, be it :static or whatever, what is the value? Just a true/false? Or a record of information?
12:09	rhickey	true
12:10	AWizzArd	ah ok
12:10	AWizzArd	Well, in that case :fast is also fine.
12:11	rhickey	:fast was a joke
12:11	cgrand	I guess that most of core fns would be marked as :static then and the special-case for direct-binding removed
12:11	rhickey	cgrand: exactly
12:13	rhickey	since primitive args/return would only be supported for :statics, using them without saying :static could either: A) be an error, B) be implicitly :static true
12:13	lpetit	rhickey: what would happen if someone tries to rebind such a var ?
12:14	rhickey	lpetit: nothing. you might do this as part of fixing code interactively, with the caveat that you must re-eval any consumer code to see the effect, just like with macros
12:14	AWizzArd	A) ==> the compiler should signal an error
12:16	lpetit	rhickey: new semantic (direct linking, non closure, no self reference apart from itself) => new "public API" keyword ? that is why not :static as the metadata , but the preferred usage be via e.g. defstatic / defnstatic ?
12:17	lpetit	s/keyword/command/
12:17	sexpbot	rhickey: new semantic (direct linking, non closure, no self reference apart from itself) => new "public API" command ? that is why not :static as the metadata , but the preferred usage be via e.g. defstatic / defnstatic ?
12:17	a_strange_guy	lpetit: agreed
12:17	rhickey	lpetit: I'm not quite sure I understand
12:18	a_strange_guy	defn shouldn't hve a semantic change
12:18	a_strange_guy	if you see that a corefn is defined with defstaticfn or similar
12:18	rhickey	I'm sure people will get tired of (defn {:static true} foo ...) just like with :private
12:19	lpetit	rhickey: yes, defn having a semantic change seems odd to me. Maybe internally (or for power users), why not have the :static keyword metadata, but for "normal" users, having defstatic will clearly prompt in their mind that they are dealing with a new kind of definition
12:20	lpetit	rhickey: this looks more like a new kind of definition than a variant to me
12:20	lpetit	rhickey: :private was ok, a variant with a different scope. But once you have the var in your hands, everything happens like usual.
12:21	a_strange_guy	rhickey: do you plan to compile this down to static methods?
12:22	rhickey	lpetit: well, we're back to layering. It is a variant, and I'm not opposed to a defsomethingshort, but lots of things yield defns, and metadata flows...
12:22	rhickey	a_strange_guy: I'm not planning anything yet
12:24	lpetit	lpetit: just like defmacro is a defn with a :macro true metadata, but nobody in current code writes it with anyhting but defmacro, I would see defstatic, with still (as for defmacro) the :static metadata keyword. Oh this would be really consistent: defmacro / :macro , defstatic / :static
12:25	rhickey	lpetit: but defmacro could turn into something completely different tomorrow
12:25	rhickey	the fact that a macro is a fn is an implementation detail
12:26	lpetit	rhickey: wouldn't it be the same for defstatic ?
12:26	rhickey	No, the things inside defstatic vars would be guaranteed to be IFns
12:26	lpetit	rhickey: with some constraints at definition time, but not visible to the caller
12:27	lpetit	rhickey: but the need to reeval user code if the definition changes ...
12:27	rhickey	right, and some linkage directives
12:27	rhickey	higher-order usage would go through the var, as would (#'my-static-fn ...)
12:33	lpetit	rhickey: higher-order function via the var would then "loose" the "no boxing" property ?
12:33	rhickey	lpetit: right
12:38	lpetit	does somebody know if there is a version of try-clojure.org which works with 1.2.0 snapshot ?
13:14	LauJensen	(clojure-version)
13:14	LauJensen	,(clojure-version)
13:14	clojurebot	"1.2.0-master-SNAPSHOT"
13:24	bartj	reading this string - "Детские сады, детские центры, дошкольное образование, прогимназии, ясл" from a MySQl database prints jibberish in Java
13:24	bartj	I am using the clojure.contribl.sql library
13:25	bartj	does one have to decode UTF-8 strings while displaying on the console?
13:25	bartj	what I see is this - "ÐÐ±Ñ€Ð°Ð·Ð¸Ð²Ð½Ñ‹Ðµ Ð¸Ð½ÑÑ‚Ñ€ÑƒÐ¼ÐµÐ½Ñ‚ Ð¾Ð¿Ñ‚Ð¾Ð¼
13:25	bartj	"
13:26	mefesto_	bartj: can you read the value in the database by some other means to make sure it was inserted w/out error?
13:26	bartj	mefesto: yes, I am able to do so in Perl
13:27	bartj	with no charset supplied...
13:32	mefesto_	bartj: have you tried manually specifying the character encoding in the jdbc url connect string?
13:33	bartj	mefesto_: no, I will look in the documentation right now...
13:40	bartj	mefesto_: yes, retried specifying the encoding as follows - "useUnicode=true&characterEncoding=utf8"
13:40	bartj	mefesto_: does not work as expected...
13:40	mefesto_	bartj: and you are sure that the server itself is using utf8 too?
13:41	bartj	mefesto_: er, you mean the mysql server?
13:41	mefesto_	yeah
13:41	bartj	mefesto_: yes
13:42	bartj	bartj: "CHARSET=utf8" is specified while creating the mysql tables
13:43	bartj	mefesto_: I am able to read the db values from a Perl program....
13:44	mefesto_	bartj: yeah, i was just thinking that maybe for some reason connector/j wasn't able to auto detect the correct encoding but perl was...
13:45	bartj	mefesto_: which should not be the problem, since we have specified the encoding in the URL string?
13:45	mefesto_	bartj: yeah
13:53	tcrayford	is printing it out the problem?
13:54	tcrayford	iirc slime/swank might not work that well with utf8 (it works, but needs some configuration set up_
13:54	bartj	tcrayford: no, because, I am able to print it out in Perl?
13:54	tcrayford	when you display the string in the repl I mean
13:54	chouser	good question. Since the String should be unicode-aware, it would be helpful to know if it made it correctly into the String or not.
13:54	bartj	This URL quite nicely explains what the problem is - http://linux.kieser.net/java_pg_unicode.html
13:54	bartj	though the problem and the solution both are for PostgreSQL
13:55	tcrayford	that's somewhat nasty :/
13:55	chouser	bartj: maybe check the length of the string to see if it is right or not?
13:56	candera	You should be able to write it to a file and view it, no? That will bypass display-related issues.
13:57	chouser	if the length is right, the problem is in printing the string. If the string is too large, the problem is getting it from the db into the string.
13:57	bartj	no, writing to a file doesn't help
13:58	bartj	tcrayford: you refer to the URL I pasted?
13:58	candera	Meaning I'm wrong and that doesn't tell you anything, or you tried writing it to a file and it's still garbage in the file?
13:58	tcrayford	bartj: yes
13:59	bartj	candera: I tried writing to a file and it's still garbage
13:59	candera	Are you sure the editor you used to view it knows what the encoding of the file is?
13:59	candera	e.g is there a BOM at the beginning
13:59	chouser	that doesn't prove much, if the thing writing the file or the thing reading the file chooses the wrong encoding.
13:59	bartj	candera: BOM?
14:00	candera	Byte Order Mark. Seen often when serializing XML. A couple of bytes that help readers of the file detect the encoding.
14:00	chouser	I'd check the length of the string while it's still in java, no IO having been done since reading from the db connection.
14:01	candera	bartj: are you on Windows?
14:02	bartj	candera: No, Linux - Ubuntu
14:02	bartj	regarding the string length, doing a length(column) on mysql gives a value of 261
14:03	candera	Ah. I don't know the editor landscape there. On Windows at least, some editors fail to deal with encoding and some get it right.
14:03	bartj	and doing a .length() after reading it from the Db in Java gives a 42
14:03	chouser	whoa
14:03	bartj	I mean 52
14:03	chouser	so there's definitely something broken there. too short? maybe it's tripping over a null or something.
14:03	bartj	candera: It displays quite nicely when I output it on the bash command line using a Perl progam
14:03	chouser	not a display issue
14:04	bartj	candera: so, I am pretty certain it is not a display issue
14:04	candera	Yeah, if the length is goofed up, that's not it.
14:04	candera	Sorry for the red herring.
14:05	chouser	bartj: you don't have a .getBytes() method you can try instead of .getString() or seomthing?
14:06	chouser	bartj: you've seen this page? http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-charsets.html
14:08	bartj	chouser: yes
14:08	bartj	I suspect the problem is that I am encoding characters twice!
14:09	bartj	but decoding the db output (in Java) does not seem to solve it... :(
14:09	chouser	not if it's already too short
14:10	bartj	chouser: I am not sure if the comparision I did is apt?
14:11	chouser	the column is a string in the db, not bytes or something else?
14:11	chouser	if so, then length(col) should be your gold standard.
14:12	chouser	Java strings can contain nulls and such without the .length() method failing, so whatever chars make it into the string will be counted.
14:13	chouser	the question being then the format of the data sent by the server to the jdbc driver, and then how the driver interprets that to create the Java string. Surely those are the only two places left to mess it up.
14:15	bartj	the string length of this is - "Детские сады, детские центры, дошкольное образование, прогимназии, ясли": 71
14:15	chouser	I see nothing but blanks and commas between those quotes. :-)
14:16	bartj	chouser: ha
14:16	chouser	but we're not trying to solve my IRC client's unicode problems, so that's ok.
14:18	neotyk	,(count "сады")
14:18	clojurebot	4
14:20	neotyk	,(.length "сады")
14:20	clojurebot	4
14:20	KirinDave	well played, java
14:20	neotyk	bartj: what's the problem?
14:21	neotyk	I had quite some problems with PL encoding while ago
14:22	bartj	neotyk: I am unable to display unicode strings (russian for now) from mysql database
14:22	neotyk	what is your connection string? do you set client encoding (jdbc driver)?
14:24	bartj	jdbc:mysql://localhost/db?useUnicode=true&characterEncoding=UTF-8
14:24	bartj	neotyk: yes
14:25	bartj	the string size is 71, but when I read it from the mysql database it shrinks
14:26	neotyk	and you see those characters in mysql client
14:26	bartj	, (count "Детские сады, детские центры, дошкольное образование, прогимназии, ясли")
14:26	clojurebot	71
14:26	bartj	yes, absolutely
14:30	mozinator	hi guys, fetching clojure from leiningen doesnt seem to work anymore for me, anyone else noticed this ? [WARNING] repository metadata for: 'snapshot org.clojure:clojure:1.2.0-master-SNAPSHOT' could not be retrieved from repository: clojure due to an error: Error transferring file
14:32	lancepantz
14:32	chouser	mozinator: Can you fetch it from build.clojure.org instead?
14:34	mozinator	chouser, I can download it from here: http://build.clojure.org/snapshots/org/clojure/clojure/1.2.0-master-SNAPSHOT/
14:34	mozinator	chouser, maybe its the same error I encountered in the past on debian linux, a missing java keystore
14:35	ysph	mozinator: i am able to fetch with [org.clojure/clojure "1.2.0-master-SNAPSHOT"]
14:36	mozinator	strange.... openjdk works, but sunjdk does not work on debian linux
14:36	zakwilson	Ahh... a channel that isn't full of Apple stuff.
14:36	zakwilson	Should I be using 1.1 or 1.2 for my personal project?
14:36	chouser	1.2
14:36	zakwilson	No hesitation there. Alright, I'll switch over.
14:36	ysph	mozinator: what version of sunjdk?
14:37	chouser	:-)
14:37	mozinator	sun-java6-jdk 6.20-dlj-1
14:37	ysph	1.6.0_12-b04 here
14:37	ysph	the new oracle vms are known to reject previously valid signatures
14:37	zakwilson	What's the appropriate contrib version to go with that?
14:37	ysph	may or may not be a factor
14:38	mozinator	ysph, I am going to check if I can reproduce the problem with a manually downloaded jvm
14:38	chouser	zakwilson: latest snapshots of each should be ok
14:41	mozinator	ok, I can reproduce the problem with a manualy downloaded JVM from oracle
14:41	mozinator	ysph, I think your theory is right
14:45	LauJensen	Im wondering what happens under the hood when I do (drop 150000 (read-lines "log"))
14:46	LauJensen	contrib.io.read-lines
14:48	ysph	LauJensen: how far under the hood?
14:48	ysph	mozinator: it seems plausible, our Eclipse RCP build stopped working after a routine upgrade to 6.20 because of rejected certs
14:49	LauJensen	ysph: Far enough to see how wasteful it is
14:49	kotarak	LauJensen: it opens a reader on log
14:50	LauJensen	kotarak: Sure, but what goes on with the 15000 lines?
14:50	chouser	technomancy: ?
14:50	kotarak	LauJensen: they are obviously read when you call seq on it
14:50	technomancy	chouser: we have a game at work where every time we run into issues with Oracle's JVM, I get a point, but whenever there's a problem with OpenJDK, they get a point.
14:50	technomancy	chouser: so far we are tied.
14:50	chouser	heh
14:51	ysph	technomancy: this latest round might put you over the top
14:51	technomancy	ysph: it does sound pretty wretched
14:51	zakwilson	After upgrading to Clojure and contrib 1.2, I cannot connect to Swank on a remote machine. I don't think I changed anything else.
14:52	LauJensen	kotarak: So drop doesn't have any tricks for lazy-seqs where it moves the line-pointer instead of actually consuming each dropped line
14:52	LauJensen	?
14:52	chouser	zakwilson: oh, man, I'm sorry -- I should have qualified my recommendation. I don't use swank, and it often lags a bit. I dunno if 1.2 works properly there yet or not.
14:52	kotarak	LauJensen: how could it know how far to read into the file for line without knowing in advance where the linebreaks are?
14:53	LauJensen	Good point :)
14:53	zakwilson	chouser: no problem. I'll poke at it a bit and if it doesn't start working, it won't be hard to downgrade.
14:54	LauJensen	kotarak: But I just dont get it, its currently getting to the last line of a 65MB file in 400ms
14:55	ysph	zakwilson: there is a swank-clojure 1.2.0 in the repos that works for me locally
14:56	zakwilson	ysph: I have 1.2.1. Starting it with lein swank works locally, but I can't connect to the remote.
14:56	kotarak	LauJensen: do you have it in cache? (read: memory)
14:56	LauJensen	no
14:56	zakwilson	Wait... maybe I did something stupid.
14:57	zakwilson	Nope. Connection refused
14:57	LauJensen	~source group-by
14:58	zakwilson	nmap of local from remote shows port 4005 closed. nmap of remote from remote shows port 4005 open. I have not changed any firewall settings on remote.
14:59	kotarak	zakwilson: did you start with 127.0.0.1 to listen?
14:59	a_strange_guy	I think that #'ns-unmap -ing imported classes has some wierd consequences
14:59	a_strange_guy	http://gist.github.com/429027
15:00	zakwilson	kotarak: I ran "lein swank". If it takes any options, I'm not aware of them.
15:01	tomoj	zakwilson: don't you want to look at nmap of remote from local?
15:02	zakwilson	tomoj: Uhh... yes. I said it backwards, but did it forwards. nmap of remote from local shows port 4005 closed
15:02	tomoj	do you really want it to be open, anyway?
15:02	ysph	zakwilson: what does netstat -tln on remote say regarding 4005?
15:02	zakwilson	tomoj: both machines are on my LAN. I do want it open.
15:03	tomoj	ah
15:03	zakwilson	ysph: netstat -tln\|grep 4005 comes up blank.
15:04	zakwilson	ysph: I should note that remote is running Mac OS and might have a slightly different netstat than you expect.
15:05	ysph	if you run it without grep is should say something like "Active Internet connections (only servers)"
15:06	ysph	zakwilson: how about netstat -l -p tcp
15:08	zakwilson	ysph: that doesn't show a line with 4005, but it does show the active ssh connections. netstat -an does show a line with 4005:
15:08	zakwilson	tcp4 0 0 127.0.0.1.4005 . LISTEN
15:11	ysph	zakwilson: this looks somewhat promising, http://github.com/hugoduncan/lein-remote-swank
15:13	zakwilson	ysph: I tried that when using Clojure 1.1 and it failed to start Swank on the remote machine.
15:14	troutwine	Would someone be so kind as to comment on this: http://groups.google.com/group/clojure/browse_thread/thread/6225680aad195dc0
15:14	troutwine	I swear that lein was working just fine not but a few days ago. :/
15:14	cschreiner	sudo?
15:15	mozinator	troutwine, I just had the same error
15:15	mozinator	troutwine, its the new oracle jdk
15:15	Twey	troutwine: Looks to me like it can't download swank-clojure-1.2.1.jar from anywhere
15:15	jdmmmmm	troutwine: try this: :dev-dependencies [[swank-clojure "1.2.1"]]
15:15	troutwine	Oh good, I've not gone mad.
15:16	troutwine	jdmmmmm: I'll give it a go.
15:16	troutwine	mozinator: How is the new Oracle JDK causing this?
15:16	hugod	zakwilson: I haven't tried the lein-remote-swank plugin in a while - I'll take a look to get it working with 1.2
15:17	mozinator	troutwine, I am not really sure , but when I switch between openjdk and the new oracle jdk things go from working to not working
15:17	TimMc	LauJensen: I took a look at your website's code... I guess what I really need to look at is a dynamic site's code.
15:17	zakwilson	hugod: cool. Currently, it's failing to rsync.
15:17	troutwine	jdmmmmm: That did it. I don't understand why, though. Can you explain the difference?
15:18	mozinator	troutwine, ok, clearly you have another problem :)
15:18	TimMc	LauJensen: I'm interested in finding a site that uses data from the DB to generate HTML pages based on templates.
15:18	hugod	zakwilson: probably a host key checking problem, or maybe invalid rsync options for your os
15:19	ysph	apparently lien swank as of 1.2 takes args such as lein swank 4005 "localhost"
15:19	LauJensen	TimMc: check out templates.clj. All the templates take arguments, they can come from a DB if you prefer. There's really nothing more to it
15:19	troutwine	mozinator: Hanging out as I do on Debian Stable, I don't think I'm using Oracle's Java yet.
15:19	LauJensen	And the dynamic bit is in admin.clj/webview.clj where all the urls served are dynamic
15:19	ysph	ref: http://stackoverflow.com/questions/2752720/how-to-reconnect-to-slime-swank-clojure-session
15:20	zakwilson	hugod: It stopped working after I rewrote project.clj. I may have somehow broken something with remote-user and remote-path, but I'm not sure how I'd do that as they seem pretty simple.
15:21	zakwilson	Problem solved, I think. Just before ysph posted that, I looked up the source for lein-swank and tried specifying a host. Evidently, it was using the loopback interface.
15:21	jdmmmmm	troutwine: I have no idea. It could have been uploaded with a different groupId previously and then removed. I just go with what's in clojars.
15:24	TimMc	LauJensen: Oh, the admin stuff... I'll take a look as soon as my work machine stops crapping itself.
15:25	troutwine	jdmmmm: I believe I understand. I was instructing lein to request swank-clojure from clojure.org, rather than from where-ever availiable.
15:26	zakwilson	My problem had nothing to do with the Clojure version. I ran lein clean to get rid of the old Clojure version and also got rid of the old lein-swank version.
15:26	jdmmmmm	troutwine: It's all based on how it's identified (groupId/artifactId in clojars or other public maven repos).
15:32	zakwilson	New problem: "lein swank" doesn't seem to be honoring jvm-opts
15:34	korre	zakwilson: what is jvm-opts?
15:35	zakwilson	A project.clj key that supposedly passes arguments to the java command line
15:35	korre	a ok
15:36	korre	whell so you can patch swank or use a shell script to start upp swank
15:37	zakwilson	Or I can edit the leiningen shell script, which is what I've done in the past. Still, that option was created to avoid having to do so and I'd like to use it.
15:39	korre	so what are you using the comand line args for in swank mode?
15:40	bartj	to all the kind souls who helped with the encoding issue, thanks! - the problem is solved
15:41	Licenser	hello my lispy friends
15:41	LauJensen	Hi Licenser
15:41	Licenser	greetings LauJensen
15:41	korre	who told you i had a lisp?
15:41	Raynes	korre: McCarthy.
15:41	Raynes	He sensed it.
15:42	a_strange_guy	cemerick: you remember us dicussing new+dynamic-fns-in-vars+defrecord?
15:42	a_strange_guy	this was my idea: http://gist.github.com/429080
15:42	zakwilson	korre: the one I need is -Xmx2048M, but I've been playing with other optimizations to see if I can get more speed.
15:42	a_strange_guy	horray for metadata in clojure
15:44	bartj	chouser: The primary problem with the encoding issue was that the database and the 'server' were not utf8...
15:45	bartj	chouser: in mysql you could do - "show variables like '%char%';"
15:46	korre	zakwilson: i thought lien started up new process's... if that is the case then modifying the script shouldnt help
15:47	chouser	so it was stored right in the table, but the server was putting it on the wire wrong.
15:47	zakwilson	korre: I killed the old one first
15:49	cemerick	a_strange_guy: I'm sunk in elsewhere, can't look at that too closely atm. If you throw in some comparative benchmarks, and they're comparable with the existing way, you might want to run it by rhickey. I doubt he'll want to change defrecord so significantly (again), but anything's possible if your idea is better.
15:50	mmarczyk	good morning (ugt)
15:51	mmarczyk	LauJensen: hurray on your open-sourcing of your site-baking code :-)
15:51	bartj	chouser: yes!
15:56	Licenser	fear! I'm bac!
16:04	LauJensen	mmarczyk: hooray :)
16:06	mmarczyk	lol, I misspelled hooray :-)
16:07	kotarak	well, it was almost hurry
16:07	kotarak	hurra
16:07	mmarczyk	yup
16:07	mmarczyk	and apparently "hurrah" is a word in English too
16:08	kotarak	hurra is german :)
16:08	mmarczyk	and also Polish :-)
16:09	LauJensen	and Danish
16:10	mmarczyk	well now, we've just found some linguistic common ground ;-)
16:11	Twey	Ah, but do you have ‘huzzah’?
16:12	mmarczyk	incidentally, how's ClojureQL coming along? I was thinking of providing a direct mapping from CQL query expressions to the internal structures I'm using in this project: https://www.assembla.com/wiki/show/adqc at some point
16:12	LauJensen	mmarczyk: Eventually 'hurra' will also be adopted into the Swedish language, but unfortunately they have been unsuccesful in teaching the population words with more than 4 letters. There's progress though.
16:12	mmarczyk	(that's Clojure in GSoC -- I'm going to post to the mailing list about this soonish, but probably after wrapping up the piece I'm working on right now)
16:13	LauJensen	mmarczyk: We're hit a small bump in the road, which Budu and I are trying to help kotarak resolve. Hard to give a deadline
16:13	LauJensen	(hard=impossible)
16:13	mmarczyk	oh, not really asking for that, just pinging for progress
16:13	LauJensen	progress is slow, but fortunately there's not many miles left
16:14	mmarczyk	sounds optimistic enough
16:15	LauJensen	Cynicism never had a chance :)
16:15	mmarczyk	well, s/slow/not happening/ would be worse :-)
16:27	mjg123	Hi. If I have a list of numbers-as-strings, how to convert it to a list of numbers?
16:27	mjg123	something like: (map java.lang.Integer/parseInt ("1" "2" "3")) ?
16:27	a_strange_guy	,(map #(Integer/parseInt %) ("1" "2" "3"))
16:27	clojurebot	java.lang.ClassCastException: java.lang.String cannot be cast to clojure.lang.IFn
16:27	candera	,(map #(java.lang.Integer/parseInt %) ["1" "2" "3"])
16:27	clojurebot	(1 2 3)
16:28	mjg123	oh, so the java call is not a function by itself?
16:28	candera	Correct.
16:29	LauJensen	(map read-string ["1" "2" "3"])
16:29	LauJensen	$(map read-string ["1" "2" "3"])
16:29	sexpbot	=> (1 2 3)
16:29	mmarczyk	oh, welcome back, sexpbot
16:29	a_strange_guy	java calls expand to either the 'new or the '. (dot) special form
16:29	mmarczyk	and that's if the source the strings are coming from can be trusted
16:30	mmarczyk	$(map read-string "#=(println :foo)" "2" "3"])
16:30	sexpbot	java.lang.Exception: Unmatched delimiter: ]
16:30	mmarczyk	$(map read-string ["#=(println :foo)" "2" "3"])
16:30	sexpbot	=> :foo (nil 2 3)
16:30	mmarczyk	but, um
16:30	mmarczyk	$(require '[clojure.contrib.java-utils :as ju])
16:30	sexpbot	java.lang.SecurityException: Code did not pass sandbox guidelines: (#'clojure.core/require clojure.contrib.java-utils)
16:30	mjg123	in fact I have lists of Character, not String...
16:31	mmarczyk	ouch. never mind
16:31	mmarczyk	(that's was about the SecurityException)
16:32	a_strange_guy	$#=(String. "foo")
16:32	sexpbot	Command not found. No entiendo lo que estás diciendo.
16:32	a_strange_guy	$[#=(String. "foo")]
16:32	sexpbot	Command not found. No entiendo lo que estás diciendo.
16:32	a_strange_guy	$(vec #=(String. "foo"))
16:32	sexpbot	java.lang.Exception: EvalReader not allowed when read-eval is false.
16:33	LauJensen	$(map #(-> % str read-string) [\1 \2 \3])
16:33	sexpbot	=> (1 2 3)
16:34	mjg123	annnd that's project euler #8 done, thanks :)
16:34	LauJensen	http://clojure-euler.wikispaces.com
16:35	mjg123	I thought you weren't supposed to do that!
16:35	LauJensen	Well...
16:35	mjg123	:)
16:35	mjg123	I won't look
16:35	LauJensen	Gotta jet - Good night all
16:37	mmarczyk	good night!
16:40	ska2342	Reference types are supposed to hold one of Clojures datastructures. Has there been any discussion on holding e.g. POJOS, and what pitfalls await the programmer?
16:42	mmarczyk	mjg123: as an alternative solution, (require 'clojure.contrib.java) (for Clojure 1.2; used to be java-utils), then (clojure.contrib.java/wall-hack-method clojure.lang.LispReader :matchNumber [String] nil "123") => 123
16:42	mmarczyk	ska2342: you're forgoing all the safety guarantees if you put something mutable behind a reference type
16:45	ska2342	mmarczyk: yes, thank you. Actually I am more interested in some readable material that sums the reasons for not doing so up. Ah, whatever, not so important, I gues..
16:50	mmarczyk	hm, you could do some rummaging around the google group, this does come up occasionally (although I regret to say I have no links at hand)
16:50	mmarczyk	but basically, the whole idea behind the reference types is that they allow you to operate on snapshots of your data
16:51	ska2342	mmarczyk: it's OK, thank you. :-)
16:51	mmarczyk	for a Ref r, you can say, at any time, @r, and you're sure to get something consistent back; that's clearly no longer the case if that thing can change
17:27	chouser	it seems that dynamic typing allows me to solve problems that I don't undestand as fully as I would need to if I were to solve them with static typing
17:28	chouser	I suppose there might be differing opinions as to whether that's good or not.
17:28	Bootvis	the question is: is that a good thing?
17:28	chouser	:-)
17:29	TakeV	Is there a tutorial for using interfaces? I've seen a demo, but I don't quite understand them.
17:33	TakeV	Er, *protocols
17:34	chouser	Halloway has a very good video on them.
17:34	chouser	TakeV: http://vimeo.com/11236603
17:35	TakeV	Thank you.
17:35	chouser	TakeV: let me know if that's sufficient for them to make sense to you
17:35	chouser	I'm considering doing a talk on protocols, but if that video is sufficient, maybe I need to find another topic.
17:37	ska2342	chouser: it is, if you are familiar with the expression problem. I had to read some other sources to get that and then it made sense to me.
17:43	zakwilson	I want to convert a number to its ascii equivalent. What's the Clojure version of CL's code-char?
17:46	zakwilson	Oh, it's simply char
18:07	TakeV	chouser: More information is always good.
18:37	Licenser	:)
19:22	vIkSiT	lo all. is there a way to pass "default" values to a fn definition?
19:25	ysph	vIkSiT: typically you provide multiple defs as in (defn foo ([x] (foo x 5)) ([x y] (* x y)))
19:29	Raynes	ysph: He's gone.
19:29	ysph	snap
19:29	ysph	rather impatient eh?
20:04	kensanata	I have a little swing app that lets me draw tiles. When the user presses a key, the app should switch to a different tile. Coming from Emacs Lisp, I want to do something like this: (def current-tile FLOOR_TILE) (defn get-tile [] current-tile) (defn set-tile [tile] (set! current-tile tile)) -- what's the Clojure way of doing this?
20:05	kensanata	I tried various variants resulting in IllegalStateException... :)
20:05	technomancy	kensanata: probably an atom, unless you need to be able to coordinate changes across threads.
20:05	technomancy	atoms are the simplest of the reference types
20:06	kensanata	Ok, will read up on atoms. Thanks. :)
20:33	charliekilo	noob compojure question: is it ok to use agents in Compojure? ... I remember that creating threads (which to my knowledge agents do), is not allowed in containers like Tomcat, Jetty.
20:35	boredomist	How could I create an instance of a nested Java class? Say I want to use Y from some.package.X.Y where X is also a class, for example
20:39	lancepantz	you (import '[some.package X]') then refer to it as X$Y
20:40	lancepantz	err (import '[some.package x])
20:41	boredomist	Great, thanks :)
20:41	lancepantz	np
20:49	ataggart	charliekilo: so long as you do it properly, creating new threads from webapps running tomcat, etc. is fine.
21:23	herdrick	question: is there a way to make memoized functions ignore, for the purpose of memoization, some of it's arguments
21:23	herdrick	i.e. have the cache be keyed only on certain args?
21:32	ataggart	not with the default implementation, but I recall a good post that made memoize a bit more flexible. lemme see if I can pull it up
21:34	ataggart	http://kotka.de/blog/2010/03/The_Rule_of_Three.html
21:37	defn	Where would (identity form) give you something you don't necessarily expect?
21:39	tomoj	where you're expecting something strange?
21:39	defn	heh, cgrand had a little clojure riddle he posted on twitter
21:39	tomoj	primitives maybe?
21:39	defn	where is ((identity X) arg1 .. argN) equal to (X arg1..argN)
21:40	tomoj	oh, I see it
21:40	tomoj	interesting
21:42	vIkSiT	hi all
21:42	tomoj	is his next tweet the answer?
21:42	hugod	zakwilson: I have brought lein-remote-swank up to date. Let me know if you have any issues.
21:42	vIkSiT	If I've got a fn called myfn, invoked as (myfn a b) - whats the best method to run this function with values of in (range 0 10)?
21:43	vIkSiT	values of a*
21:43	vIkSiT	(like, repeatedly run this function, sort of a test harness)
21:43	mmarczyk	well
21:44	mmarczyk	taken literally, that's not a very difficult riddle :-)
21:44	mmarczyk	no extra constraints?
21:44	vIkSiT	mmarczyk, none :) so, I can think of two ways of doing it..
21:44	tomoj	presumably only interesting answers will be accepted
21:45	mmarczyk	vIkSiT: map, for, doseq... :-)
21:47	mmarczyk	for cgrand's riddle, there are two huge obvious classes of answers... it would be quite interesting to discover whether there are any possibilities outside those
21:47	vIkSiT	ah, is there a way to do it using loop?
21:47	tomoj	clojurebot: loop?
21:47	mmarczyk	vIkSiT: well, sure, but why?
21:47	vIkSiT	mmarczyk, ah, thats the one I couldn't get to work..
21:48	mmarczyk	(loop [i (range 0 10)] (myfn i b)), with b to be supplied by the surrounding lexical context
21:48	vIkSiT	(btw, which riddle was just referenced here?)
21:48	mmarczyk	ouch, I mean
21:48	tomoj	loop doesn't work that way
21:49	vIkSiT	mmarczyk, clojure.lang.LazySeq cannot be cast to java.lang.Number :) loop won't work that way..
21:49	vIkSiT	thats what I was assuming too..
21:49	mmarczyk	yeah, typed a for form, then switched to "loop" by replacing the initial symbol :-P
21:49	vIkSiT	hehe
21:49	tomoj	vIkSiT: if you're throwing away the return values, use doseq
21:49	mmarczyk	(loop [i 0] (when-not (== i 10) (myfn i b) (recur (inc i))))
21:50	KirinDave	So we gotta write a clojure to objc combiler.
21:50	KirinDave	Err, compiler
21:50	mmarczyk	here you go, but it's best accomplished in a different way
21:50	KirinDave	So that we can get something resembling clojure onto the iPhone. :\
21:50	vIkSiT	tomoj, mmarczyk - ah in either case, the return value won't be available..
21:51	tomoj	if you want the return value, use for
21:51	mmarczyk	also, this is assuming you're throwing away the results; if you want, say, to build a vector of them, (loop [i 0 r []] (if-not (== i 10) (recur (inc i) (conj r (myfn i b))) r)) is one possibility, though for normally be better
21:52	mmarczyk	vIkSiT: then doseq is indeed your best bet: (doseq [i (range 0 10)] (myfn i b))
21:52	vIkSiT	aaaah. that last loop is quite enlightening, thanks.. yes, the for and doseq are cool.
21:52	tomoj	,(= (range 10) (range 0 10))
21:52	mmarczyk	tomoj: oh, right :-)
21:53	mmarczyk	KirinDave: that's not really going to work with the current wording of the dev agreement
21:53	KirinDave	mmarczyk: Can they tell?
21:53	KirinDave	Probably not.
21:53	mmarczyk	sure, if they care to
21:54	KirinDave	Well, I am not sure.
21:54	KirinDave	I've written code generating compilers that people couldn't tell before
21:54	KirinDave	For C++ even
21:54	Raynes	$(= (range 10) (range 0 10))
21:54	sexpbot	=> true
21:54	KirinDave	Fairly complex.
21:54	mmarczyk	well, I'm wishing you luck, of course
21:54	vIkSiT	hrm, clj to objc sounds like fun :)
21:54	mmarczyk	I've seen some machine-generated code and it was always pretty obvious that no human's written it
21:54	Raynes	tomoj: When at first, clojurebot doesn't succeed, try, try sexpbot. :p
21:55	mmarczyk	but I'm not saying that you can't do better
21:55	mmarczyk	the "common wisdom" seems to be that that's actually pretty hard
21:55	tomoj	I would be incredibly surprised if objc compiled from clojure looked anything like human-written objc
21:56	KirinDave	tomoj: It might not be idiomatic.
21:56	KirinDave	And it might require a very comprehensive library
21:56	mmarczyk	plus all this is without bringing up the feeling of insecurity I'd have if I knew I was consciously doing things in a way incompatible with the dev agreement for the platform I was working on
21:56	KirinDave	But guys, I'm blue skying here. Come on.
21:56	KirinDave	mmarczyk: Actually it's really unclear what apple wants to do there.. I mean, are they really gonna kill games using Lua?
21:56	KirinDave	They haven't done that entirely yet.
21:57	mmarczyk	not to say I don't believe such legal shenanigans as Apple is performing right now should be banned due to the way in which they complicate (make impossible?) interoperability, because I do :-P
21:57	Lajla	mmarczyk, how do you specy ranges in clojure.
21:58	tomoj	banned by whom?
21:58	Lajla	Something I, as the greatest programmer except the MCSA should surely be told.
21:58	Lajla	Like, (range 0 10) does that work?
21:58	mmarczyk	tomoj: I meant "made illegal"
21:58	Lajla	,(range 0 10)
21:58	tomoj	crazy
21:58	mmarczyk	where it's not illegal already.
21:59	Lajla	Oh, it's above there coindidentally.
21:59	Lajla	$(range 10)
21:59	sexpbot	=> (0 1 2 3 4 5 6 7 8 9)
21:59	mmarczyk	well, I don't care enough to really argue that particular point
21:59	Lajla	$(range 0 4 2)
21:59	sexpbot	=> (0 2)
21:59	Lajla	What's with the new bot?
22:00	tomoj	if there's anyone I'd be more afraid of than apple/MS/google/whoever to have in control of such things, it's the government :)
22:00	herdrick	ataggart: thanks
22:00	mmarczyk	tomoj: not sure what you mean by this
22:00	herdrick	that is interesting
22:00	tomoj	Lajla: "specy" ?
22:00	Lajla	tomoj, oui
22:01	mmarczyk	the government mandating which languages we can use -- I don't even want to think about that
22:01	Lajla	We used to have a different one that worked by a comma
22:02	mmarczyk	but a law being passed to prevent companies from these kinds of outrageous interference with people targetting their platforms in languages they don't control seems rather different
22:02	tomoj	heh i
22:02	mmarczyk	to the point where I can't see that the two ideas are really related
22:02	tomoj	I imagine a C++ country, a clojure country, a ruby country, etc, all going to war in cyberspace
22:03	Lajla	C++ would win, stupid people are surprisingly powerful in large groups.
22:03	mmarczyk	in fact, isn't it already explicitly legal to reverse engineer where that is done for interoperability reasons, right? both in US and EU
22:05	mmarczyk	of course the current trend is to prevent consumers from circumventing DRM even where this would be for purposes compatible with fair use... locking up content being sold to consumers in formats which might well become a problem in and of themselves (a server goes down and you can't play your track...) is apparently perfectly legal
22:05	mmarczyk	I wonder if I should perhaps discontinue ranting at this point :-)
22:12	zakwilson	hugod: I will give it a try tomorrow. I'm currently away from my LAN.
22:32	vIkSiT	hmm. so I've got this tree traversal function here: http://paste.lisp.org/display/111243
22:32	vIkSiT	is there an idiomatic way to conver this to loop/recur, and somehow conj the results into a seq?
22:32	vIkSiT	convert*
22:33	mmarczyk	you don't need the do around the or, for one thing
22:33	mmarczyk	also, how about in-order for a name :-)
22:34	vIkSiT	hehe, yes, the name will change ;)
22:34	vIkSiT	ah yes the do is gone now.
22:35	mmarczyk	but, um, this will just print the leaves left-to-right, right?
22:35	vIkSiT	yes?
22:36	vIkSiT	in-order should do left, root, right?
22:36	mmarczyk	yup
22:37	tomoj	won't that always return false?
22:37	mmarczyk	how about (->> bt (clojure.walk/prewalk (fn [{:left left :right right}] [left right])) flatten (map #(partial println "Visited"))) or some such thing
22:38	mmarczyk	oh, sorry, that's (fn [{left :left value :value right :right}] [left value right])
22:39	Raynes	clojure.walk?
22:39	mmarczyk	for loop/recur, you'd have to rewrite this in continuation-passing style
22:39	Raynes	:\|
22:40	mmarczyk	Raynes: hi! nice to see sexpbot again :-)
22:40	vIkSiT	hmm clojure.walk. must look that up.
22:40	mmarczyk	what's wrong with clojure.walk?
22:41	Raynes	mmarczyk: <3 sexpbot :D
22:41	Raynes	mmarczyk: Nothing wrong with it. I just have no clue what it is.
22:41	mmarczyk	oh :-)
22:41	vIkSiT	mmarczyk, hmm yes, was wondering if thats the most efficient or not. also, I wanted to re-learn consing recur resulsts :)
22:42	vIkSiT	,(doc clojure.walk/prewalk)
22:42	mmarczyk	I find it superbly useful in ast transformations where all disambiguation can be performed based on the looking down the tree :-)
22:42	mmarczyk	$(doc clojure.walk/prewalk)
22:42	sexpbot	=> ------------------------- clojure.walk/prewalk ([f form]) Like postwalk, but does pre-order traversal. nil
22:42	Raynes	mmarczyk: dysinger ran sexpbot out of #clojure for a while. sexpbot overcame and burned down his house. <3
22:42	mmarczyk	clojurebot's gone for a walk apparently
22:42	vIkSiT	aah
22:42	mmarczyk	Raynes: way to go! :-)
22:42	dysinger	are we still talking about this ?
22:42	vIkSiT	well, I guess I should look at how prewalk is implemented then :) trying to do this from first principle
22:43	Raynes	dysinger: sexpbot and I still love you. <3
22:43	mmarczyk	dysinger: no, by this point we're just sorry about your house ;-)
22:43	mmarczyk	vIkSiT: oh I can tell you that -- it's implemented in a wicked cool way :-)
22:43	dysinger	sorry guys I have real work to do
22:43	dysinger	thanks for calling my attention though
22:44	Raynes	You're welcome.
22:44	Raynes	Sourpuss. :(
22:45	vIkSiT	oh that IS nice
22:45	vIkSiT	I like the inner outer implementation
22:45	Raynes	I should make sexpbot's doc special like clojurebot's. That line of -'s at the beginning is rather offputting, especially for long docstrings.
22:46	vIkSiT	Raynes, +1
22:46	Raynes	It's better if the documentation is longer than 300 characters though, because sexpbot gist's evaluation results that are over 300 characters (or is it 200?).
22:46	Raynes	$(range 1 1000)
22:46	sexpbot	=> (1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 ... http://gist.github.com/429531
22:46	Raynes	gists*
22:47	vIkSiT	this is pretty cool too - http://rosettacode.org/wiki/Tree_traversal#Clojure
22:47	vIkSiT	is the "walk" idiom a functional thing? or is that more clojurian?
22:48	tomoj	I wonder how deep your tree has to be to blow the stack
22:48	vIkSiT	tomoj, yes thats the reason I was trying to convert this into loop/recur
22:51	tomoj	not sure how that would be possible
22:52	mmarczyk	(loop [bt bt q (clojure.lang.PersistentQueue)] (let [{:keys [left right value]} bt items (drop-while nil? [left value right])] (condp == (count items) 1 (do (println "Visiting" (first items)) (if-not (empty? q) (recur (peek q) (pop q)))) 2 (recur (first items) (conj q (second items))) 3 (recur (first items) (into q (rest items))))))
22:52	hiredman	~scala {((x: Any, y: Any) => (f: Function2[Any, Any, Any]) => f(x, y))(1, 2)((x: Any, y: Any) => x)}
22:52	mmarczyk	vIkSiT: I actually wrote this right here, no testing, but maybe you can give it a try ;-)
22:53	mmarczyk	Raynes: cool idea with the gist, though I trust egregious misuse is being prevented somehowb :-)
22:54	Raynes	mmarczyk: How would someone misuse that?
22:54	mmarczyk	um, (range) ...?
22:54	Raynes	mmarczyk: It would time out.
22:54	hiredman	clojurebot: ping?
22:54	clojurebot	PONG!
22:54	mmarczyk	how large would it get though
22:55	Raynes	mmarczyk: If it times out, it doesn't give anything to gist.
22:55	mmarczyk	then you could use (range n) for just the right n
22:55	Raynes	And it still wouldn't be large enough to cause problems, I imagine.
22:55	Raynes	And I could lower the timeout if necessary.
22:56	Raynes	Oh noes, clojurebot is back. sexpbot's claim to fame has been interrupted. :p
22:56	mmarczyk	I wonder, ranges materialise pretty quickly
22:58	Raynes	mmarczyk: I could count the size of the output, if it was an actual problem, but that wouldn't be very efficient. Unless gist has a very low limit on what size a gist can be, it shouldn't ever pose a problem.
22:58	riddochc	Maybe if sexpbot choose a more risque name people would use it instead? ;)
22:59	Raynes	I was kidding. <3 clojurebot
22:59	Raynes	But sexpbot does prove useful in situations like the one a moment ago where clojurebot disappeared briefly. :D
23:00	mmarczyk	Raynes: if you feel this way... I promise not to try to break it :-)
23:00	Raynes	mmarczyk: If you can break it, by all means, go for it. :p
23:00	clojurebot	forget Clarity of mind means clarity of passion, too; this is why a great and clear mind loves ardently and sees distinctly what it loves. -- Pascal
23:01	mmarczyk	Raynes: all of a sudden I feel evil :-)
23:02	mmarczyk	$(doall (map #(* 2 #^Integer %) (range 4000000)))
23:02	sexpbot	Execution Timed Out!
23:03	mmarczyk	oh come on.
23:03	mmarczyk	$(doall (map #(* 2 #^Integer %) (range 2000000)))
23:03	mmarczyk	$(range 2500000)
23:03	sexpbot	Execution timed out.
23:03	sexpbot	java.lang.Exception: Thread stopped!
23:03	_brian2_	noob question > from the tree exmple, I m just trying to understnd the first function , what is f in (defn walk [node f order] ...
23:03	mmarczyk	Thread stopped?
23:03	mmarczyk	$(range 2500000)
23:04	sexpbot	Execution timed out.
23:04	sexpbot	java.lang.Exception: Thread stopped!
23:04	mmarczyk	$(range 1000000)
23:04	sexpbot	java.io.IOException: Error writing to server
23:04	mmarczyk	ohhh, that's something new :-)
23:05	mmarczyk	oh, alright... admittedly I should have queried the bot
23:06	mmarczyk	_brian2_: were are you getting that from?
23:07	_brian2_	somebody just cited http://rosettacode.org/wiki/Tree_traversal#Java
23:09	Raynes	mmarczyk: I suspect that java.io.IOException: Error writing to server is the breakage you were looking for.
23:09	mmarczyk	_brian2_: apparently the idea is that node is a binary tree with data on interior nodes
23:09	mmarczyk	a node looks like {:left left-branch :val value-at-current-node :right right-branch}
23:10	mmarczyk	f is a function which you wish to call on every element of the tree
23:10	Raynes	mmarczyk: I'll catch that error and handle it specially. <3
23:10	mmarczyk	and order is a vector of keywords
23:10	mmarczyk	possible entries: :left :visit :right
23:10	mmarczyk	walk does these things in order at each node
23:10	mmarczyk	:left means walk the left subtree, :right similarly
23:11	mmarczyk	:visit means call the given function on the current node
23:11	mmarczyk	Raynes: :-)
23:11	_brian2_	I take it its a clssic comp sci problem (?)
23:12	_brian2_	I never took a comp sci class
23:12	Raynes	danlarkin: I don't think I'm on a bot ego trip. :\ I just like to unsuccessfully mess with hiredman about clojurebot.
23:12	mmarczyk	_brian2_: it's also one of the basic day-to-day programming problems
23:12	_brian2_	ok, cool
23:12	mmarczyk	with all the stuff being represented as trees
23:12	_brian2_	thats better ;-)
23:13	mmarczyk	:-)
23:21	mmarczyk	vIkSiT: http://gist.github.com/429563
23:22	mmarczyk	here's a loop/recur-based tree walk
23:22	mmarczyk	not pretty. :-(
23:25	_brian2_	danlarkin : I am having a problem with getting clojure-json write to file example
23:26	danlarkin	oh! yes I saw your issue
23:26	danlarkin	then I got distracted :)
23:26	_brian2_	ok, thanks
23:26	vIkSiT	mmarczyk, aah. looks interesting though :)
23:26	danlarkin	I think the problem is you're not opening the FileWriter in a with-open, so it never gets closed (which means the buffer isn't flushed)
23:27	danlarkin	at least in the example you pasted
23:28	_brian2_	ok, I dont see any with-open in your posted exmple code
23:29	danlarkin	it may be poor sample code :)
23:29	_brian2_	;-)
23:35	vIkSiT	hmm how do I print out a clojure.lang.PersistentQueue?
23:35	vIkSiT	mmarczyk -to debug?
23:36	danlarkin	_brian2_: let me know if that works out for you
23:36	_brian2_	ok
23:36	_brian2_	thnks!
23:38	vIkSiT	hrm.
23:39	mmarczyk	vIkSiT: just wrap it in a print(ln)
23:39	vIkSiT	mmarczyk, well, that just shows #<PersistentQueue clojure.lang.PersistentQueue@a2272bf8> :)
23:39	vIkSiT	During my do, I do a (prn (peek q))
23:39	vIkSiT	which works fine .. but just wanted to see the entire thing at once..
23:40	mmarczyk	vIkSiT: really? in that case, perhaps (println (str "#:PQ" (seq the-queue)))
23:40	vIkSiT	aha
23:40	mmarczyk	or skip the #:PQ if you're not worried about confusing it with a list
23:40	vIkSiT	(seq q) does the trick
23:40	mmarczyk	yup
23:40	vIkSiT	yes in my case its a nice tree
23:40	vIkSiT	whew. level order is complex
23:41	vIkSiT	https://gist.github.com/aa0a0095fe386f490b6f
23:43	vIkSiT	interesting I didnt know cond could get a :done
23:46	tomoj	huh?
23:48	mmarczyk	vIkSiT: hm? :-)
23:48	mmarczyk	(looking at the new gist now)
23:52	trptcolin	is there a way to use binding to replace Java methods?
23:53	mmarczyk	trptcolin: nope. you might be able to use proxy to mock them out though
23:54	mmarczyk	not that that's likely to help a lot...
23:54	trptcolin	mmarczyk: it definitely confirms that i'm not crazy :)
23:55	mmarczyk	trptcolin: happy to hear that ;-)
23:56	_brian2_	danlarkin : is it supposed to look something like: (with-open [o (FileWriter.) ] (json/encode-to-writer [1 2 3 4 5] o "/tmp/foo.json"))
23:59	danlarkin	no, pass the path as an arg to the FileWriter constructor