#clojure – 2010-05-26

2010-05-26

01:01	bmason	man I'm glad I read this channel... reduce is the bomb
02:59	bartj	er, is there anyone here who knows russian?
03:12	unfo-	i heard google does
03:13	technomancy	unfo-: heyo
03:15	bartj	unfo: apparently google's translate engine is having problem translating a russian page to english - "http://www.yell.ru/all/2000002-%252000125-%251000083/en/146/small"
03:15	bartj	I know this is terribly off topic to the channel but...I wanted to know what language this phrase "bezopasnost' dorozhnogo dvizheniya - oborudovanie" from the above page is in
03:17	bmason	http://encyclopedia.farlex.com/Komitet+Gosudarstvennoy+Bezopasnost
03:17	bmason	perhaps it is a name?
03:19	waterless_cloud	might be ukranian
03:21	bartj	I think it is a business category, because it says so on this page - "http://www.yell.ru/EASY%20PARKING/all/en/7003483/643/entry" but weirdly google translate is not able to find the translation for - "bezopasnost' dorozhnogo dvizheniya - oborudovanie"
03:22	bartj	waterless_cloud: phrase does not work either from "Ukranian to English" on Google Translate for the phrase - "bezopasnost' dorozhnogo dvizheniya - oborudovanie"
03:28	waterless_cloud	GIBDD Gosudarstvennaya Inspectsiya Bezopasnosti Dorozhnogo Dvizheniya (Russian: Federal Road Safety Service)
03:29	bartj	waterless_cloud: I don't understand
03:30	waterless_cloud	and reference here
03:30	waterless_cloud	http://en.russia.edu.ru/tips/sec/1331/
03:30	waterless_cloud	seems to be traffic cops
03:39	vu3rdd	technomancy: would you like to include a maven po.xml in the leiningen repo for other folks who are doing similar things as me?
03:40	technomancy	vu3rdd: no, the pom is an generated artifact of project.clj
03:40	technomancy	generated artifacts generally don't belong in source control; they should just be re-created on-demand
03:41	vu3rdd	technomancy: yes, but building leiningen itself without using lein is a problem in that case, right? And the generated pom does not build a useful jar
03:41	technomancy	oh, what's missing in the generated pom?
03:41	technomancy	I haven't actually tried it in this case
03:42	vu3rdd	it creates an empty jar
03:42	npoektop	hi! is there a function like doseq which collects results?
03:42	vu3rdd	I had to use the maven-clojure-plugin to compile leiningen.core (or anything which needs to be aot compiled)
03:42	vu3rdd	also add resources for the source files to be included
03:43	technomancy	hmm... I'd rather fix the pom task so it generated a useful pom.
03:43	somnium	npoektop: for
03:44	vu3rdd	technomancy: Yes, I thought about that too. pom task definitely needs some care.
03:45	npoektop	somnium, thanks!
03:45	technomancy	vu3rdd: I don't use it myself, so I'm reliant on contributions from others to get it well-polished.
03:45	vu3rdd	technomancy: ok
03:45	vu3rdd	technomancy: I understand the hint
03:46	technomancy	from you or others. =)
03:46	technomancy	IDE-users, etc.
03:46	vu3rdd	technomancy: sure, I will look at it.
03:47	technomancy	vu3rdd: have you thought about waiting for the stable 1.2 release to perform the packaging? it shouldn't be too far off now.
03:47	vu3rdd	technomancy: I don't use IDE myself though. But most java projects use mvn and hence it is quite well supported in GNU/Linux world
03:47	technomancy	vu3rdd: oh, I just meant that in the past the pom task has mostly been used for IDE interop
03:47	vu3rdd	technomancy: I was thinking of packaging 1.1.0. Also debian does not have maven-ant-tasks, so I will have to package that too..
03:49	vu3rdd	technomancy: java packaging has been a pain. I have never faced so many issues in plain old C world.
03:49	technomancy	yes, it's tricky to do systemwide java packages. most of the java ecosystem assumes it's installed on a per-user basis
03:49	vu3rdd	technomancy: do you plan to add some verbosity/logging support in lein so that during debugging it cabe turned on?
03:50	technomancy	vu3rdd: it's crossed my mind, but it's not a high priority right now.
03:50	vu3rdd	technomancy: debian solves it quite nicely. The systemwide jars you install becomes your local maven repo
03:50	vu3rdd	technomancy: ok
03:51	technomancy	hmm... I have often had issues with systemwide debs for things like hudson and solr that have turned me towards using tarballs.
03:51	technomancy	actually no, hudson worked decently, but solr was totally broken last time I checked
03:52	vu3rdd	technomancy: perhaps true. I think it is still evolving. But debian has got some good basic infrastructure for maven in place.
03:52	technomancy	did you say you were planning on packaging the latest lein with clojure 1.1.0? because I don't think that will work; there are some things in lein right now that rely on clojure 1.2-snapshot
03:52	vu3rdd	No, I mean packaging lein-1.1.0
03:53	technomancy	oh, I see. that would probably be better, though I think 1.2 will not be too far away if you want to wait.
03:53	vu3rdd	technomancy: yes, that is what I thought after talking to you yesterday
03:53	technomancy	lein 1.1.0 has a bug in the repl task (due to a bug in clojure) that makes it use the wrong version of clojure for the project repl
03:54	technomancy	but lein 1.2.0 has a bug in the repl task due to a bug in ant that makes it break on some platforms.
03:54	technomancy	very annoying!
03:54	vu3rdd	oh.. ok
03:54	ordnungswidrig	vu3rdd: check if the bug in ant isn't fixed by debian
03:55	vu3rdd	ordnungswidrig: maven-ant0tasks is not yet in debian
03:55	technomancy	ordnungswidrig: the ant developers have no plans to fix the bug. =(
03:55	ordnungswidrig	vu3rdd: oh, debian and maven the long story
03:56	ordnungswidrig	technomancy: maybe the debian maintainers have
03:56	ordnungswidrig	I tried packaging maven for debian 3 years ago. horrible
03:56	vu3rdd	ordnungswidrig: I will check. ant version on debian is 1.8.0. I use unstable tree
03:57	vu3rdd	ordnungswidrig: The current situation is quite ok. Java is sort of a first class citizen in the debian world now
03:57	ordnungswidrig	vu3rdd: can maven be build from source now? IIRC maven needs some heavy "external" bootstrapping
03:58	ordnungswidrig	vu3rdd: even upstream maven used to need some existing maven jars to build itself
03:58	vu3rdd	ordnungswidrig: http://wiki.debian.org/Java/Maven2
03:58	vu3rdd	ordnungswidrig: lots of pain
03:59	technomancy	ordnungswidrig: I'd be thrilled... but it's unlikely.
04:01	ordnungswidrig	now i'm sure it was a good decision to drop this task
04:01	vu3rdd	technomancy: clojure/clojure-contrib 1.1.0 is already in unstable/testing
04:01	technomancy	oh, cool
04:06	vu3rdd	packaging maven-ant-tasks is going to be a bit tough because they build a shaded (uder)jar
04:06	vu3rdd	s/uder/uber
04:06	vu3rdd	I don't think debian "allow" that
04:10	ordnungswidrig	vu3rdd: you mean shading=
04:10	vu3rdd	ordnungswidrig: yes
04:11	ordnungswidrig	vu3rdd: shading does not make sense in a distribution environment as there is only one version of each dependency available anyway
04:12	ordnungswidrig	vu3rdd: but it makes sense for a build tool, where you want to avoid the version blues
04:12	vu3rdd	ordnungswidrig: yes. It is like going back to those old days of static libraries
04:12	ordnungswidrig	vu3rdd: you've heard of nixos linux?
04:13	vu3rdd	ordnungswidrig: no
04:14	ordnungswidrig	vu3rdd: they do package management like a hybrid of git and gentoo :-) Symlinking until your inodes smoke. Every package can have it's one "view" to other packages and versions.
04:14	vu3rdd	ordnungswidrig: oh.. ok. I will look at it
04:28	bozhidar	ll
04:28	bozhidar	ops, sorry - I thought I was in the terminal buffer ;-)
05:25	rava	greetings
05:25	cgrand	hi
05:25	Lajla	rava, bow before me.
05:25	Lajla	I am the second best programmer in the world.
05:26	Lajla	Rise, my friend.
05:26	rava	great and merciful 2nd best
05:27	rava	dost thow know compojure? for yonder #compojure is fairly dead this eve
05:28	Lajla	rava, I actually barely know clojure.
05:28	rava	heh
05:28	Lajla	I'm just a Scheme programmer infiltrating here.
05:28	rava	ew, scheme
05:28	Lajla	Still deciding if I want to learn Clojure or not.
05:28	Lajla	rava, why?
05:28	rava	just prefer common lisp
05:28	rava	no real or objective reason
05:29	Lajla	rava, well, there must be a subjective one?
05:30	rava	asdf for one
05:32	rava	dislike lexical scope only
05:33	rava	couple other odds and ends
05:33	rava	those are what came to mine
05:34	rava	i've not even looked at scheme in 2 + years though
05:37	LauJensen	haha <Lajla> I'm just a Scheme programmer infiltrating here.
05:37	LauJensen	<rava> ew, scheme
05:37	LauJensen
05:39	Lajla	There are some times yes when it would be easier if a variable could be given dynamic scope too.
05:39	Lajla	But still, you shall be executed for your bad taste, just like Oscar Wilde and Alan Turing.
05:45	Chousuke	scheme has fluid-let, doesn't it?
05:49	naeu	could someone clarify a clearly flawed understanding of basic computing that's currently in my head. Why does (byte 2r11111111) produce -1 (or raise an exception in 1.2 snapshot). Doesn't a bit consist of 8 bits, and are they not all allowed to be 1?
05:50	Chousuke	a byte consists of 8 bits
05:50	Chousuke	but a number in base-2 can be as large as you want
05:51	Chousuke	2r... is not notation for bitfields, but numbers in base 2
05:51	Chousuke	though hm
05:51	Chousuke	,2r11111111
05:51	clojurebot	255
05:52	Chousuke	,(byte 2r11111111)
05:52	clojurebot	java.lang.IllegalArgumentException: Value out of range for byte: 255
05:52	cgrand	bytes in java are signed
05:52	Chousuke	ah, right.
05:52	Chousuke	yeah
05:53	naeu	ok, so the max is 127 for byte consisting of 7 bits?
05:53	cgrand	-128 to 127
05:54	naeu	ah ok, so the range is the same (255) just the integer which generates a given list of bits is offset by -128
05:56	cgrand	not really offset: 0 is still 0, not -128
06:00	cgrand	unsigned byte 0..127 = signed byte 0..127; unsigned byte 128..255 = signed byte -128..-1
06:01	hoeck	naeu: http://en.wikipedia.org/wiki/Two%27s_complement
06:03	hoeck	,(bit-and 0xff (byte -1)) ;; to get the unsigned value of a byte
06:03	clojurebot	255
06:09	patrkris	What's the easiest way to run a project set up with leiningen? Using Eric Lavigne's leiningen-run? Seems to get me strange output - [null] prepended to every line i output.
06:14	naeu	hoeck: thanks
06:14	naeu	so what's the simplest way of creating a byte represented by the string 11111111
06:15	cgrand	,(Byte/parseByte "11111111" 2)
06:15	clojurebot	java.lang.NumberFormatException: Value out of range. Value:"11111111" Radix:2
06:17	hoeck	,(.byteValue (Integer/parseInt "11111111" 2))
06:17	clojurebot	-1
06:17	cgrand	thx
06:18	naeu	thanks
06:19	naeu	also whilst I'm in a questioning mood, is there a nice way to iterate over a sequence with for side effects, but with the iteration yielding an index?
06:19	naeu	essentially i'd like a side-effect focussed equivalent of map-indexed
06:20	hoeck	dorun + map-indexed ?
06:23	naeu	hoeck: nice, thanks so much
06:52	Lajla	Chousuke, 'Scheme' doesn't, that is R5RS doesn't.
06:52	Lajla	But it's easy to add it.
06:52	Lajla	The Scheme standard itself is extremely minimal, things like when, let, fluid-let are all just extensions written in Scheme
06:55	bartj	can anyone give a use-case for map-indexed - so that I could write some code using it?
07:09	bartj	anyone?
07:27	mikem	looking at clojure.core source, I see something like this: (fn defn (&form &env name & fdecl] ...) -- what does the leading & in &form and &env mean?
07:32	reZo	ohai
07:33	bartj	mikem: my guess - that they are optional arguments
07:34	mikem	bartj: hm I thought optional arguments are suffixed with a *
07:36	bartj	mikem: the general convention is to use earmuffs for things intended for rebinding. Reference: http://www.assembla.com/wiki/show/clojure/Clojure_Library_Coding_Standards
07:43	hoeck	mikem: those are some macro-only?, compiletime? args while expanding the macro, &env resolves to the current lexical environment, &form to the current form
07:44	patrkris	how do I get the Class<T> of Integer? Tried Integer/TYPE, but that is apparently not it.
07:45	hoeck	mikem: sorry, those are arguments the macro-expansion function takes when called
07:45	hoeck	,Interger
07:45	clojurebot	java.lang.Exception: Unable to resolve symbol: Interger in this context
07:45	hoeck	,Integer
07:45	clojurebot	java.lang.Integer
07:46	patrkris	hoeck: just like that? cool :)
07:48	hoeck	yeah, Integer/TYPE returns the primitive type int
07:49	bartj	what is the use of #^ while defining function
08:11	_na_ka_na_	hi, is there any built in fn to convert an ns into file system path ?
08:12	_na_ka_na_	currently I'm have to do -
08:12	_na_ka_na_	,(.replaceAll "a.b.c" "\\." (str java.io.File/separator java.io.File/separator))
08:12	clojurebot	"a//b//c"
08:12	_na_ka_na_	File/separator is two times coz of windows
08:19	mikem	hoeck: ah great, thanks for that clarification
08:20	hoeck	mikem: to see what they contain, you can use a dummy macro like this: (defmacro foo [] (println &env &form))
08:21	hoeck	mikem: and expand in a let: (let [x 10] (foo))
08:38	mikem	also in clojure.core source, I see the ^ symbol used to denote metadata as well as type hints (ie: http://github.com/richhickey/clojure/blob/master/src/clj/clojure/core.clj#L124) -- is the ^ context aware?
08:40	mikem	i guess it must be
08:40	chouser	no
08:41	chouser	type hints are metadata
08:41	mikem	a lightbulb just went on over my head :P of course!
08:41	chouser	"^FooType foo" is just short for "^{:tag FooType} foo"
08:42	mikem	thanks :)
08:48	_na_ka_na_	hey can anyone solve my regex mystery, I'm not so proficient in Java,
08:48	_na_ka_na_	,(let [m (.matcher #"(.+)\.([^\.]+)" "a.b.c")] (.matches m))
08:48	clojurebot	true
08:49	_na_ka_na_	,(let [m (.matcher #"(.+)\.([^\.]+)" "a.b.c")] (.groupCount m))
08:49	clojurebot	2
08:49	_na_ka_na_	,(let [m (.matcher #"(.+)\.([^\.]+)" "a.b.c")] (.group m 0))
08:49	clojurebot	java.lang.IllegalStateException: No match found
08:49	_na_ka_na_	Please tell me why the above fails?
08:50	rhickey	chouser: want to discuss recur?
08:50	chouser	rhickey: sure
08:50	_na_ka_na_	nvm got it
08:50	_na_ka_na_	,(let [m (.matcher #"(.+)\.([^\.]+)" "a.b.c")] (if (.matches m) (.group m 0)))
08:50	clojurebot	"a.b.c"
08:50	chouser	_na_ka_na_: clojure has re- functions that make these things easier.
08:51	rhickey	first and foremost, recur is about no stack usage - goto
08:51	rhickey	with loop, the target is labelled by loop
08:52	chouser	yep, I'm with you so far
08:52	rhickey	with fns, the target is the same fn, with the same arglist. The 'this' is the fn, and implicit
08:52	rhickey	you can't goto another method in the same this, or to another this
08:53	chouser	hmmm
08:53	rhickey	to do so would be a proper method call with stack overhead
08:55	chouser	because the fn is a java method and you can't change the value of 'this' within it.
08:56	rhickey	because you can't recur out of a method
08:57	rhickey	naming this is what's broken as it is really not an argument supplied in the normal way
08:57	chouser	when you recur to a fn, locals are actually re-assigned, right?
08:58	rhickey	but doing so is useful and preferred
08:58	rhickey	chouser: yes
08:58	chouser	and the JVM won't let you reassign 'this' even if you wanted to.
08:58	rhickey	but technically, rebound
08:59	rhickey	chouser: 'this' is not an argument, nor a local in Java
08:59	chouser	right
08:59	chouser	no way to refer to it in the bytecode such that it could be rebound
08:59	rhickey	there's nothing to rebind
09:01	rhickey	placing this in the arglist just makes it appear like something it's not
09:01	rhickey	but we've done that
09:02	rhickey	but that shouldn't swap our thinking around
09:02	rhickey	this is not an argument
09:02	chouser	no, it's more of a debate.
09:02	chouser	sorry, j/k
09:02	rhickey	aargh
09:02	chouser	ok, I'm with you
09:03	rhickey	one way out is top forbid recur to method head in deftype/record method bodies, force using loop
09:03	rhickey	but that seems like a loss
09:04	chouser	My concern was cheifly if it might actually be a argument on some hosts, or be able to act like one, that the behavior of the JVM here would lock us out of capabilities these other hosts might provide.
09:04	rhickey	doing so would make it obvious, because people that tried to loop-rebind this would find they couldn't then use it without using '.'
09:05	chouser	sigh I don't follow.
09:06	rhickey	if recur to head was forbidden in deftype/record method bodies, you'd be forced to say (foo [this x] (loop [x x] ... (recur new-x)))
09:07	chouser	yes, I see that.
09:07	rhickey	but if you wanted to recur with this ... (loop [this this x x] ... (recur new-this new-x))
09:07	rhickey	but why would you do that if you didn't want to use this?
09:08	rhickey	and if you did want to use this, you'd find yourself needing to use '.'
09:08	chouser	ah, instead of plain names to refer to your own fields, for example.
09:08	rhickey	and thus I hope understand why this isn't a bindable recur target, you must call another method to use it
09:09	rhickey	yes, or explicit field ref
09:11	rhickey	so, the this 'arg' is truly special, and it is surprising only recur exposes that
09:12	chouser	ok, it that last point that finally makes it click for me
09:12	rhickey	I think the choice of propagating the specialness to recur is least appeal atm
09:13	rhickey	so either we call out this arg, and thus why it can't be in recur, or disallow recur to deftype/record method head
09:13	chouser	in the method bodies x means (.x this), but if you could rebind 'this' those would mean different things. or something. just broken.
09:13	rhickey	we have the former right now
09:14	djpowell	I'm no fan of throwing the kitchen sink into clojure, but I would have liked to see some version of clojure.contrib.repl-utils/show brought across with clojure.repl - it was very handy to have, especially if you are used to using an IDE to remember all your method names.
09:20	chouser	djpowell: I'm with you on that.
09:21	chouser	rhickey: I think the current behavior of recur is good, now that I understand why.
09:22	rhickey	so, what do people prefer - recur not pass this, or disallowed?
09:22	rhickey	chouser: oh good!
09:22	chouser	disallowed doesn't seem any better
09:23	_na_ka_na_	rhickey: can you explain with an example how the first option would work?
09:23	rhickey	chouser: it would be better only in calling out methods as not first class fns, the heads of the latter being ok recur targets
09:24	chouser	yeah, and a more informative error message, etc.
09:24	rhickey	_na_ka_na_: (foo [this x] ... (recur new-x))
09:24	_na_ka_na_	rhickey: and inside the method body we can use (.a this) (.b this) as usual right?
09:24	chouser	But the "what" is clearly documented as it is, and you can always use 'loop' to get explicit and clear behavior
09:25	rhickey	_na_ka_na_: inside you can just use a and b
09:25	chouser	as you said, you could even do (loop [this this] ...) if you want, but then the meaning of plain field symbols goes all wonky
09:25	rhickey	chouser: right, recur would not be inconsistent in any way
09:26	_na_ka_na_	great I think, then the first option sounds good
09:26	_na_ka_na_	but would require some calling out, as you said
09:27	chouser	some day the error case for recur could notice an attempt to use 'this' and give a more helpful error, but that's a nicety that's a whole different level of concern than I was trying to bring up.
09:27	rhickey	chouser: if you do (loop [this this] ...) I don't see any problem with plain fields. Do you think people will expect them to move to the new this?
09:28	chouser	no, they won't, or at least it'll be immediately obvious why they don't.
09:28	_na_ka_na_	I think people with lesser Java experience would, no?
09:28	chouser	no, I really don't think so.
09:29	chouser	(defrecord C [a] (foo [this] (prn a) (loop [this this] (prn a) (when a (recur other-c))))
09:30	chouser	I think it would seem magical and unexpected for the a's inside the loop to get new values on recur
09:31	rhickey	yeah, and the lack of use of this in the body a sure sign of something wrong
09:31	SinDoc	Any ideas as to making ADTs in Clojure?
09:31	rhickey	SinDoc: deftype
09:32	chouser	really, the only answer I needed for my original concern was "what about plain symbols referring to fields?"
09:32	chouser	I wasn't considering those.
09:32	chouser	With them in the picture, my argument (er, debate) falls apart.
09:32	SinDoc	rhickey: Thanks, I'm on it.
09:34	_na_ka_na_	chouser: Can you explain what the intended behavior is in the above example .. when you recur with other-c ?
09:35	rhickey	chouser: still, that's more of the symptom than the root cause
09:35	chouser	_na_ka_na_: my point is that the intended behavior is a bit non-sensical
09:36	_na_ka_na_	My thinking is if at all people do that (i.e. loop [this this]) they'd expect it to change?
09:36	rhickey	(foo this [x] ... (recur new-x))
09:37	chouser	_na_ka_na_: Maybe it should use (.a this) inside the loop instead of just a
09:37	chouser	rhickey: yeah, that thought occured to me.
09:37	chouser	or maybe :as this up at the top. ;-)
09:37	rhickey	is more akin to fn, but a break from the protocol sigs
09:37	rhickey	chouser: heh
09:38	rhickey	no, it's better than :as at top by a lot
09:38	rhickey	the alignment with self-naming fns is strong
09:38	_na_ka_na_	noobish question, but why is it at all necessary to keep a 'this' arg ?
09:39	_na_ka_na_	is it can never change, is there even any point ?
09:39	_na_ka_na_	s/is/if
09:39	chouser	you need to be able to refer to it
09:39	rhickey	_na_ka_na_: for method self-calls and to pass your this to others
09:39	chouser	so you can pass it as an arg to other functions, that sort of thing.
09:40	_na_ka_na_	hmm, got it!
09:43	_na_ka_na_	what if, this were made a special symbol inside a method call? I mean, (defrecord C [a] (foo [x] (.. x, a, this all available..)))
09:44	_na_ka_na_	just like in Java
09:45	chouser	_na_ka_na_: These options have all been discussed. :-) Previously you could name the 'this' for the whole defrecord by saying ":as this"
09:46	chouser	one problem with that is that your protocol was (defprotocol P (foo [this x])) ...a different number of args
09:46	_na_ka_na_	chouser: oh, I'm late then, nvm
09:47	chouser	I don't know if that's the main reason rhickey changed it or if there was something else more compelling.
09:48	_na_ka_na_	hmm, but can be solved by saying this is special
09:48	_na_ka_na_	just checked, in Java you get invalid VariableDeclaratorId
09:48	_na_ka_na_	void f(int this){}
09:50	Licenser_	hmm anyone using enclojure?
10:01	cemerick	Licenser_: yes
10:10	Licenser_	cemerick: did you ever have the problem that you can't create new projects?
10:11	cemerick	Licenser_: I've never created a clojure-specific project in netbeans/enclojure. Just a regular java project with clojure dependencies is just fine.
10:12	cemerick	(i.e. either a netbeans-specific project, which you can configure through the UI, or a maven project, from which netbeans will pull all configuration / dependency info)
10:13	Licenser_	nods okay too sad
10:27	cemerick	Licenser_: why sad?
10:30	somnium	I'm a bit disappointed that (ahem) 'Fantom' has made it to the available plugins list ahead of Clojure
10:31	cemerick	somnium: this is in netbeans?
10:32	cemerick	oh, yup, there it is
10:40	chouser	ugh, I hate discovering I've been building on an incorrect assumption
10:40	bozhidar	most people do :-)
10:41	Hodapp	well, some people will argue for hours in the hopes that the assumption will prove correct or that people will start to believe it is
10:48	chouser	turns out this thing is an identity not a value.
10:49	chouser	speaking of which, it's interesting that clojure's all about the split between identity and value, and it has functions 'identity' and 'val'.
10:52	lpetit	rhickey: reminder for you to add clojure.com ref in #clojure 's topic :)
10:55	drewr	quick macro question
10:55	drewr	how can I get `((foo) (bar) (baz)) to expand to (foo) (bar) (baz)?
10:56	lpetit	drewr: you can't
10:56	bozhidar	clojure.com?
10:56	clojurebot	clojure is not groovy
10:56	drewr	the normal idiom is `(do ~@(....)), but the wrapping do won't work in this case
10:56	somnium	clojurebot: mechanic?
10:56	clojurebot	No entiendo
10:56	lpetit	drewr: but you can have `(do (defn (...)) (def ...)) at the top level
10:57	chouser	drewr: a macro returns a single thing, it can't return 3 things
10:57	lpetit	drewr: if the macro is called from top level, the do will be read as if each of its child were emitted in sequence
11:00	drewr	ok
11:01	drewr	I should be able to expand it in a parent macro
11:01	drewr	just wanted to make sure I wasn't missing an obvious hack
11:07	cgrand	,(-> (range 33) vec transient pop! count )
11:07	clojurebot	java.lang.IllegalAccessError: Transient used after persistent! call
11:08	cgrand	,(-> (range 42) vec transient pop! count )
11:08	clojurebot	41
11:09	Plouj	(doc source)
11:09	clojurebot	"clojure.contrib.repl-utils/source;[[n]]; Prints the source code for the given symbol, if it can find it. This requires that the symbol resolve to a Var defined in a namespace for which the .clj is in the classpath. Example: (source filter)"
11:09	Plouj	(source doc)
11:09	Plouj	for some reason my slime-repl doesn't find source
11:10	Plouj	,(source doc)
11:10	clojurebot	java.lang.Exception: Unable to resolve symbol: source in this context
11:10	bozhidar	have you required it?
11:10	Plouj	yeah
11:10	bozhidar	only required or used?
11:11	Plouj	humm
11:11	Plouj	actually it seems that (clojure.contrib.repl-utils/source doc) works
11:11	Plouj	(clojure.contrib.repl-utils/source doc)
11:11	Plouj	,(clojure.contrib.repl-utils/source doc)
11:11	clojurebot	java.lang.ClassNotFoundException: clojure.contrib.repl-utils
11:11	Plouj	weird, I used to get that ClassNotFoundException too
11:13	bozhidar	I do a
11:13	bozhidar	(use 'clojure.contrib.repl-utils)
11:13	bozhidar	and (source source) for example works just fine for me
11:14	bozhidar	clojure 1.1
11:14	Plouj	humm, that works for me too, now
11:16	bozhidar	when you do an import you have use the full namespace before the symbol unsless you used :as
11:18	stuarthalloway	anybody here write clojure code that uses Strings?
11:19	stuarthalloway	if so, take a look at https://www.assembla.com/spaces/clojure/tickets/359-promote-contrib-string and tell me what's right/wrong about this plan
11:24	fogus_	I might find better names for chop and chomp
11:29	stuarthalloway	fogus_: well do it then! :-)
11:30	stuarthalloway	those names are common to the scripting languages, but if you've better ones I am all ears
11:30	stuarthalloway	maybe EndOfStringCleanupFactoryCreator.getInstance.create.removeStuff(options)
11:31	fogus_	I suspected that they were.
11:31	fogus_	I suppose if they are ubiquitous then I'm the weird one here
11:31	stuarthalloway	still, if you have any other names at all I would like to hear them
11:31	candera	There are only two hard problems in computer science: naming things, cache invalidation, and off-by-one errors.
11:31	eikke	does clojure have a 'zip' equivalent?
11:32	somnium	I think code like: (-> nomnom chomp slurp chop chop spit) is rather elegant
11:32	cgrand	eikke: (map vector coll1 coll2 ...)
11:32	fogus_	Is options a concrete realization of IDoWeirdStuffToStringsKeySet?
11:32	eikke	cgrand: ah, makes sense...thanks!
11:33	dnolen	stuarthalloway: why promote chomp if String.trim() is "similar and faster" ? I'm curious as to why that's not just called trim and have it call String.trim
11:33	stuarthalloway	if their semantics are identical then chomp will get dumped
11:34	fogus_	dnolen: chomp works on the end only
11:35	bozhidar	btw should upper-case and lower-case actually be uppercase & lowercase?
11:35	dnolen	fogus_: ah
11:35	stuarthalloway	bozhidar: don't think so
11:35	stuarthalloway	the Java versions camel case at the 'c'
11:35	bozhidar	according to the dictionary upper case and uppercase are both valid
11:36	bozhidar	and I think that for a function name the single word should be favoured
11:36	stuarthalloway	so we are being "consistent" with Java, in some odd sense
11:36	bozhidar	I assumed so
11:36	arohner	does congomongo support sending JS to the server?
11:36	bozhidar	I'd prefered names like upcase/downcase - they sound more groovy ;-)
11:37	stuarthalloway	so nobody is going to say "you forgot my favorite string function"?
11:37	somnium	arohner: there's a method, db.eval on one of the db classes.
11:37	bozhidar	other than that I think the library's got what is generally needed
11:37	arohner	somnium: and when querying?
11:37	bozhidar	sometimes there are separated ltrim/rtrim methods though
11:37	somnium	arohner: I found its quite easy to crash a mongod with it though
11:38	arohner	somnium: oh, fun. Thanks for the warning
11:38	stuarthalloway	bozhidar: those exist in contrib, but are not (yet) on the list for promotion,
11:38	somnium	arohner: I think there's a $js or $eval key
11:39	stuarthalloway	in other news: we are going to start doing sublibrary builds for contrib
11:39	bozhidar	stuarthalloway: where are you promoting this string functions to?
11:40	stuarthalloway	to address the problems people have expressed around unnecessarily coarse dependencies + churn in contrib
11:40	stuarthalloway	bozhidar: clojure.string in clojure.jar
11:40	bozhidar	great
11:40	stuarthalloway	the sublibrary thing seems like boring detail work for someone (hoping to sucker sierra into it :-) )
11:40	fogus_	I got nothing for renames, the best I could do was remove-last for chop... chomp is so narrow in focus that any other name would be just as opaque
11:41	bozhidar	btw I think I read in your book that your opinion was that stuff like String manipulation don't need clojure wrappers
11:41	stuarthalloway	... but the sublibraries should be an easy win for library users ... or am I missing some danger here?
11:42	stuarthalloway	bozhidar: string manipulation doesn't need clojure wrappers, but the community seems to like them, and some of the ones in contrib have usability or performance problems
11:42	stuarthalloway	so if they are going to be there, they might as well be right
11:43	bozhidar	sure
11:43	bozhidar	I'd probably prefer them myself if there are part of clojure.jar
11:44	bozhidar	since I generally avoid using stuff from contrib, fearing the extent to which they were tested
11:44	somnium	is (upper-case "foo") => (.toUpperCase "foo") ?
11:45	bozhidar	yes
11:48	fogus_	Oh I see now. chop and chomp are Perlism.
11:49	bozhidar	fogus_: and Rubism as well
11:50	somnium	I wonder if haskell is the only language with a function called 'intercalate' in its std-library
11:52	fogus_	If we change chomp to rstrip, then we can add an additional lstrip, and then add ltrim and rtrim. too much? :p
11:53	stuarthalloway	fogus_: not seeing them used much in code to date...
11:54	fogus_	Fair enough. It might still be useful to rename chomp to rstrip
12:01	bozhidar	I think chomp will do fine
12:01	bozhidar	and there is no problem to have a couple of sinonyms for some function I guess
12:02	bozhidar	this was one of the things I liked a lot about Ruby
12:02	chouser	I very very much dislike that about Ruby
12:02	bozhidar	for me in Clojure a bad name choice was the count function
12:02	yacin	i want to add jpcap to a lein project i'm working on. but it isn't in a mvn repo and it requires a C-Java JNI library to be compiled
12:02	bozhidar	I remember when I first needed it
12:02	yacin	how should i go about packaging it?
12:02	bozhidar	trying (size ..)
12:02	bozhidar	(length ...)
12:03	bozhidar	without consulting the docs I'd never have figured out that I needed count...
12:04	bozhidar	chouser: I dislike the many leftovers from Perl and the end for end statements
12:04	fogus_	rstrip is much more descriptive and does not require that you have experience in some esoteric language. ;)
12:04	bozhidar	but I don't think that the freedom of choice that Ruby's authors have given developers is that bad
12:05	bozhidar	though it might be confusing indeed if you're not familiar with the language
12:05	bozhidar	fogus_: Perl is far from esoteric - after all it's far more popular than Clojure
12:06	bozhidar	from a JAPH's point of view probably Clojure is just as esoteric ;-)
12:06	fogus_	plus rstrip leaves the door open for lstrip... what does chomp provide? lchomp? champ?
12:06	somnium	has anyone ported Acme::Buffy to clojure?
12:06	fogus_	bozhidar: Just kidding
12:07	bozhidar	fogus_: just kidding back ;-)
12:08	bozhidar	I used to do a lot of programming in perl some 5-6 years ago, but know I'm having hard time remembering even the basics
12:08	bozhidar	it was a fun ride while it lasted, though :-)
12:09	bozhidar	now*
12:10	fogus_	Finally, rstrip works better in an overall StrippingProtocol (not what you think).
12:10	rhickey	fogus_: rstrip is just about newlines?
12:10	bozhidar	you should think more about strippers ;-)
12:10	fogus_	rhickey: yes
12:11	rhickey	fogus_: and lstrip is also?
12:11	fogus_	yes
12:11	rhickey	is that a real use case - leading newlines?
12:11	bozhidar	rhickey: fogus_ is not correct
12:12	fogus_	:(
12:12	bozhidar	rstrip/rstrip should generally strip every whitespace character
12:12	cgrand	rhickey or stuarthalloway: did you see https://www.assembla.com/spaces/clojure/tickets/358 ?
12:12	bozhidar	including newlines
12:13	fogus_	bozhidar: I thought that was trim/ltrim/rtrim's job?
12:13	stuarthalloway	cgrand: just claimed it
12:13	bozhidar	fogus_: true
12:13	stuarthalloway	thanks!
12:13	bozhidar	that's why chomp is not exactly rstrip
12:14	bozhidar	since at least in Perl it removed only the trailing newline
12:14	bozhidar	if any
12:14	replaca	rhickey, stuarthalloway: are we going to do something about the mismatch between with-in-reader and line-seq? This messes up a nice file reading idiom.
12:14	cgrand	stuarthalloway: ok, thanks
12:14	rhickey	replaca: which mismatch?
12:14	bozhidar	in most languages - strip == trim
12:15	replaca	with-in-reader returns a PushbackReader and line-seq wants a BufferedReader (I think that's right)
12:16	replaca	rhickey: yeah, that's it
12:17	rhickey	replaca: what's the fix?
12:19	replaca	rhickey: dunno, I'm not sure I'm the one who knows all the issues (understanding what kind of readers we want where). Maybe with-input-reader should open a buffered reader (but I think other funcs want pushback readers) or maybe line-seq should be more relaxed about what kind of reader it takes.
12:19	stuarthalloway	build on top of a language with good IO abstractions? :-)
12:19	fogus_	bozhidar: Not in contrib.string, which is where I tried to constrain my focus. To me chomp and chop are meaningless names, but I do not have a heavy shell text processing background. strip and trim are descriptive and represent names that I might actually look for.
12:19	replaca	rhickey: all I know is that they don't play well together :)
12:20	replaca	rhickey: I'm happy to start a discussion on the list and try to surface all the issues and drive it to resolution if you'd like
12:21	bozhidar	fogus_: I tought we were have a more general discussion, about the naming -> what should it do. Sorry about that
12:21	rhickey	replaca: that would be great, thanks
12:21	replaca	rhickey: k, I'll do it later on today
12:22	bozhidar	were having*
12:22	bozhidar	I hate my typing...
12:22	patrkris	why is [null]
12:22	patrkris	... prepended to all output from leiningen?
12:25	stuarthalloway	patrkris: level 1 answer: it's an ant thing
12:25	patrkris	stuarthalloway: ah... any way to get around it?
12:25	stuarthalloway	maybe -- years since I forgot ant
12:25	patrkris	:)
12:26	stuarthalloway	if there is a way, it probably would require some small change to lein to expose it
12:31	Chousuke	I think it actually has something to do with the logging facilities
12:32	angerman	it's not possible to have a tree-shaker for clojure, right?
12:36	arohner	angerman: I don't think it's possible to do purely in clojure. it'd have to understand the host language as well
12:37	angerman	arohner: well I guess cleanly designed and separable librarys would make it not such much a problem :D
12:38	angerman	and otherwise I guess it could get really trick to figure out what is dynamically loaded ...
12:40	mefesto	is there an existing function that will provide a lazy-seq of chars for a Reader?
12:44	arohner	mefesto: what are you trying to do?
12:44	arohner	,(seq "foo")
12:44	clojurebot	(\f \o \o)
12:44	arohner	somnium: are you interested in a patch to congomongo that allows you to send arbitrary JS to the server?
12:45	mefesto	arohner: parsing some file and i need to be aware of newlines depending on if im in quotes or not
12:45	Licenser_	Writing clojure in windows is a sad experience :(
12:45	mefesto	so line-seq wouldn't work since a quote could be open on line 1 and closed on line 5
12:45	arohner	mefesto: oh, a java stream Reader, not a clojure compiler reader
12:45	candera	Licenser_: agree. It's the ghetto.
12:45	mefesto	arohner: yeah, sorry for the confusion.
12:46	Licenser_	I tried emacs, ccw and enclojure all make me cry :/
12:46	Licenser_	Why can't my boss give me a MacBook :P
12:46	Chousuke	mefesto: (repeatedly #(.read thereader)) should work
12:47	candera	I settled on emacs, and I like it well enough, but it's definitely got more friction than on other platforms. But VirtualBox and Ubuntu are both free, so often I just go virtual.
12:48	mefesto	Chousuke: i went this route since i wanted them to be chars, overkill?: https://gist.github.com/6d7c8a2c9bcd748b22fb
12:49	mefesto	,(char -1)
12:49	clojurebot	java.lang.IllegalArgumentException: Value out of range for char: -1
12:49	Licenser_	candera: I use emacs on my mac but I don't like to do all the jar management by hand it makes me unhappy :P and I can't use maven/lein/whatever else
12:49	Quiesce	Licenser_: Yeah? What would you use on this hypothetical gift MacBook?
12:49	Chousuke	mefesto: that's essentially what my suggestion does
12:49	Licenser_	Quiesce: AquaEmacs :P and a usefull internet connection hopefully
12:49	Chousuke	mefesto: except it doesn't check for EOF, it includes in in the stream
12:51	angerman	yea, Aquamacs is pretty nice. ...
12:51	angerman	except when you run into C-; making the meta switch :D
12:51	somnium	arohner: sure
12:53	drewr	isn't there a flatten-1 somewhere that only flattens a sequence one list deep?
12:54	drewr	ie: (flatten-1 '(1 2 (3 (4 5)))) -> (1 2 3 (4 5))
12:54	drewr	or is it called something else?
12:54	chouser	apply concat
12:54	drewr	hm
12:54	chouser	hm, or maybe not in that case.
12:55	chouser	,(mapcat #(if (coll? %) % [%]) '(1 2 (3 (4 5))))
12:55	clojurebot	(1 2 3 (4 5))
12:55	chouser	dunno if there's a flatten-1 somewhere or not
12:56	drewr	my contrived example isn't quite right
12:57	drewr	'((1 2) (3 4) ((5 6) (7 8))) -> ((1 2) (3 4) (5 6) (7 8)) is more like what I need
12:57	drewr	my macro needs to end up with a list of single sexps
12:58	chouser	I assume it's not always just the last one that's grouped an extra level?
12:58	drewr	right
12:59	chouser	,(mapcat (fn [[x :as a]] (if (coll? x) a [a])) '((1 2) (3 4) ((5 6) (7 8))))
12:59	clojurebot	((1 2) (3 4) (5 6) (7 8))
13:00	drewr	ah, thanks, I think that works
13:01	KirinDave	I finally got around to reading "Multiple Dispatch in Practice"
13:02	KirinDave	Man, it's really interesting how there seems to be this consistent but small desire for multiple dispatch
13:03	drewr	what's strange is that my solution worked if I called it as (foo '((...) (...))), but idn't when called as (foo (map bar '(....))); bar being what transformed the original forms into the nested forms
13:04	drewr	...which shouldn't have changed the behavior of the flatten
13:04	drewr	but oh well
13:04	zakwilson	Given enough time, almost every project needs multiple dispatch or some sort of ugly hack to simulate it.
13:05	KirinDave	zakwilson: I was just surprised at how consistent that demand is
13:05	KirinDave	And it's not for like 7 way dispatch, the majority is 2 term dispatch
13:07	rhickey	KirinDave: and you can do clean, extensible 2-way dispatch using double dispatch and protocols, which don't have the problems of using interfaces for the same task
13:08	KirinDave	rhickey: When you say double dispatch you mean?
13:08	rhickey	KirinDave: dispatching on 2 terms via 2 single-dispatches
13:08	KirinDave	Ah
13:09	KirinDave	So yeah, bouncing back and forth a la visitor?
13:09	rhickey	normally that's a problem using interfaces, leading to the dreaded visitor
13:09	rhickey	but protocols are not type intrusive, so it's clean and open
13:09	KirinDave	rhickey: I must confess, from a strictly aesthetic perspective I don't like that solution.
13:09	KirinDave	protocols DO make it easier.
13:10	KirinDave	And the 2 dispatch system is more amenable to trace tree optimizations.
13:10	KirinDave	Still, it just sticks in my teeth. :)
13:10	rhickey	not just easier, substantively different in that it is open, double dispatch with interfaces can't be
13:11	KirinDave	I understand and agree. :)
13:11	KirinDave	I just like reading code where the double dispatch is clear
13:11	KirinDave	So md methods seem like a better notation to me. It's very clear what's happening.
13:27	KirinDave	Oh McCarthy
13:27	KirinDave	srry
14:18	islon	is there a way to set the number of agent threads used by send-off?
14:18	chouser	no, it grows as needed.
14:21	islon	hmmm... strange... I used send-off to run a bat (java.lang.Process) that blocks forever, the second send-off I called (another different bat) never started
14:21	chouser	to the same agent?
14:22	islon	yes... ok I got it...
14:28	cemerick	sending to one's own threadpool/executionservice is on my 1.3 list
14:28	cemerick	send-to, perhaps
14:30	rhickey	cemerick: send-to vs (agent ... :on-pool pool) ?
14:31	rhickey	I don't think consumers want/need to know about the pool
14:31	chouser	that would influence only send-off not send?
14:31	rhickey	chouser: both
14:32	chouser	huh
14:32	cemerick	rhickey: that's why my first thought was send-to
14:32	cemerick	send and send-off are thin abstractions for what pool to use
14:33	chouser	it's not uncommon for me to use send and send-off on the same agent because it's the action that controls that decision.
14:33	cemerick	right
14:33	islon	the :on-pool is implemented?
14:33	cemerick	no
14:33	chouser	seems weird to remove that option in order be able to use your own pool.
14:33	rhickey	If you are choosing its pool you must know how you are going to use it
14:33	cemerick	the usage can be mixed, though
14:34	fogus_	I try not to think about pools
14:34	rhickey	cemerick: so choose a pool that works for both, certainly we could only have send-off
14:35	rhickey	but providing a pool at every send-to is a catastrophe from an app modularity standpoint
14:35	cemerick	yeah, I can see that
14:36	rhickey	really, looking back I'd rather have had only send and force you to choose a pool at agent construction
14:36	cemerick	In that case, the intended semantics of send and send-off would get blurry IMO. Blocking vs. non-blocking, ok, but again, you really do end up needing to know the specific characteristics of the corresponding pools.
14:36	cemerick	right, yes
14:37	rhickey	cemerick: if you always use blocking-is-ok there is only one semantic
14:37	rhickey	even if it doesn't ever block
14:38	chouser	I think that'd be fine with me.
14:38	rhickey	so send == send-off, supply a pool to get a compute-only agent
14:38	rhickey	deprecate send-off
14:38	chouser	sure
14:39	islon	what about (binding [agent-pool my-pool] (do stuff...))
14:39	chouser	binding-only args are baaad.
14:39	chouser	especially in such close proximity to theads
14:40	chouser	(binding [agent-pool my-pool] (send foo #(do ... (send foo ...)))) ; sends to two different pools!
14:40	islon	right
14:43	remleduff	Do you think dynamic bound variables have been net positive to clojure? They seem to be where most of my head-banging problems have come from
14:44	remleduff	Once you're used to them, it's less of an issue, but seems that (map some-fn-that-depends-on-dynamic-vars ...) is something everyone has tripped on at least once
14:44	cemerick	rhickey: sounds good to me
14:45	chouser	remleduff: I do wonder. My current take is that they're useful, but at the top of the list for ways to accidentally produce unintended behavior.
14:45	hiredman	binding is the doom of the world
14:45	hiredman	(unless you use it to emulate multiple returns)
14:45	chouser	heh
14:46	remleduff	I think that would be the top check I'd like to see in a clj-lint project ;)
14:49	technomancy	hiredman: botsmack
15:13	bartj	anyone from an information extraction background and using clojure here?
15:20	cemerick	bartj: hi :-)
15:20	cemerick	very broad concept there
15:59	remleduff	Is there a difference between butlast and drop-last?
16:00	chouser	hardly. drop-last is lazy
16:01	remleduff	Is there a difference between chomp and butlast/drop-last?
16:02	remleduff	Oh, chomp checks if it's a newline
16:02	chouser	and returns a string, presumably. butlast and drop-last return seqs
16:02	chouser	oh, I see where you're ogin
16:02	chouser	goin
16:02	chouser	going
16:03	chouser	clojure.string/butlast instead of chop or less
16:03	remleduff	Probably slightly too big of a semantic difference
16:04	chouser	I think it's the same as chop -- drops one thing regardless of what it is
16:04	chouser	chomp/rtrim/rstrip are something else
16:06	remleduff	Oh yes, I meant chop ;)
16:09	fogus_	Is butlast defined by its drop-iness or its seq-iness, or are they inseparable?
16:09	tomoj	huh?
16:10	chouser	butlast should continue to return a seq, as it does now
16:10	chouser	,(butlast "abc")
16:10	clojurebot	(\a \b)
16:11	chouser	but (clojure.string/butlast "ab") could certainly return "ab"
16:11	fogus_	That's kinda confusing no?
16:11	chouser	isn't that what clojure.string is all about?
16:12	fogus_	confusion?
16:12	chouser	right!
16:12	fogus_	:p
16:12	chouser	wait, no.
16:12	chouser	reusing names fron clojure.core, but applied to strings
16:12	fogus_	Well, that I'm not sure.
16:13	chouser	like replace, and ... others I can't think of.
16:13	tomoj	what does anyone think of the idea of having a single set of functions/macros paralleling map, filter, for, etc but which return the same type as their args, use transients or whatever speedups are available, and are implemented with protocols?
16:14	tomoj	or is the seq abstraction too important to optionally abandon? hmm
16:14	remleduff	I really like the way protocols let you provide a default implementation and then specialize where appropriate, so I hope something like that happens
16:15	chouser	tomoj: I think there have been a couple efforts in that direction
16:15	tomoj	excellent
16:15	chouser	clojure.contrib.generic or something, and I think sean devlin has done something else.
16:18	tomoj	wonder what good names would be for those functions/macros
16:18	fogus_	That kinda leads into my whole reason for jumping on the string functions. I would like to see a different categories of protocols: 1) some that do something to X and return seqs and 2) some that do something to X and return things of type X 3) ???
16:19	chouser	hm... I wonder if it's the wrong approach.
16:19	tomoj	you would like to see #3 but don't know what it would be?
16:19	fogus_	not sure. It's not really formulated in my brain yet
16:20	remleduff	Why do you need functions that return a seq, as long as everything knows how to produce a seq when needed?
16:20	chouser	A chain of seq-processing fns on a vector with vec around them all is almost certainly going to be more efficient than producing a vector at each step
16:20	tomoj	even with transients?
16:20	tomoj	hadn't considered that possibility
16:20	chouser	transients would still be eager
16:20	tomoj	ah, hmm
16:21	fogus_	darn... gotta go
16:21	chouser	lazy seqs, esp with the cell stuff that rhickey has mentioned, may be as fast as a hand written loop that processes each item fully before proceeding to the next.
16:21	tomoj	I guess the use case I am thinking of is doing a small chain near the inner loop and not wanting to do hashmap->seq->hashmap or whatever
16:21	chouser	such that intermediate objects not even be created.
16:21	tomoj	nice
16:22	chouser	even 1.1 has chunked seqs that can beat some naive hand-written loops.
16:22	chouser	so perhaps something that conveniently wraps the whole chain for you...
16:23	chouser	(same-type->> my-vector (map ...) (filter ...) (partition ...) ...)
16:23	tomoj	when we're working with seqables qua data not qua seqs, I wonder if this would still be faster
16:23	tomoj	e.g. transforming a bunch of small maps with filter or something in the inner loop
16:24	tomoj	if the seqables are small I guess it won't matter a whole lot anyway
16:24	hsjunnesson	So, hm..
16:24	tomoj	or maybe it would matter more? gah, I'm just confused now
16:24	chouser	heh
16:24	hsjunnesson	Okay, so I asked this in #leiningen but no one there replied. I figured people here might have a bead on this.
16:25	hsjunnesson	I uploaded a few jars to clojars and it took a few tries before I "got it", so now I need to delete a couple of jars.
16:25	tomoj	as the seqable grows, the O(n) conversion overhead (like (into {} ...)) becomes less visible, I guess
16:26	chouser	hsjunnesson: I wouldn't actually be surprised if there were no way to delete jars from clojars
16:26	hsjunnesson	That's kind of what I'm afraid of.
16:26	chouser	hsjunnesson: you might send email to the maintainer
16:26	hsjunnesson	I'll have to do that, thanks.
16:29	tomoj	I wonder if clojure is a good choice for a first language to teach to non-programmers
16:30	technomancy	hsjunnesson: yeah, you have to get _ato to do it for you =(
16:30	chouser	tomoj: I wonder that too. The biggest worry I have is the Java interop -- so good for getting things done, but such a bag of complexity for someone who doesn't already know OOP
16:32	tomoj	ah, hmm, hadn't even thought about java interop :(
16:32	tomoj	learning only enough java to get by at the same time you learn clojure sounds terrible
16:33	SynrG	i know oop, but mostly via ruby. i'm only learning enough java to do clojure :/
16:34	chouser	if you could find enough fun and interesting exercizes that didn't require much (or any?) java calls, it might be perfect.
16:34	chouser	and now you could learn OOP via records and protocols before confronting all of Java's complexity
16:35	SynrG	mind you, i've read some java texts ... i've just never actually used it for anythign real
16:35	SynrG	so i don't have what i'd consider a working knowledge
16:36	chouser	I'd especially love to see if a new programmer has any trouble at all with immutable collections, locals, and even HOF -- things that those of us who started in BASIC/Pascal/C/etc. have had to struggle through.
16:37	mefesto	i think my first reaction to someone that has only programmed in clojure would be, "you spoiled brat!" ;)
16:37	chouser	heh
16:38	chouser	I would show them for(i=0; i<foo.length; i++) { if(skip(i)) { i++ } } ... and watch them panic
16:43	_na_ka_na_	Hi
16:43	_na_ka_na_	Ok, I come from the family of imperitive languages so macros and I are not really friends till now. I have a few basic qs reg. macros.
16:43	_na_ka_na_	My understanding is macros (only?) live to make code shorter and sweeter by spitting out mechanical code for you.
16:43	_na_ka_na_	For example if I'm repeatedly printing diff of two numbers, (println (- x y)),
16:44	_na_ka_na_	I can (defmacro print-mah-diff [x y] `(println (- ~x ~y))) ; and (print-mah-diff x y)
16:44	_na_ka_na_	But I've also seen macros such as (defmacro print-mah-diff [x y] (println (- x y))) (eg: gen-class)
16:44	chouser	that is one use-case, yes
16:44	_na_ka_na_	What would be the purpose of such macros? Wouldn't a fn work too?
16:45	mefesto	_na_ka_na_: i think the general rule is that if you can do it with just a function... then you don't need macros
16:45	mefesto	_na_ka_na_: macros let you manipulate the forms before they are evaluated which can be very powerful
16:46	mefesto	_na_ka_na_: think manipulating an AST
16:46	chouser	gen-class is a macro just so you don't have to quote things a bunch. It's a thin wrapper around generate-class which is a function
16:47	remleduff	_na_ka_na_: Your example calculates the substraction at compile time... it might be used for optimization of an expensive constant
16:48	Borkdude	When is a array-map going to turn into a hash-map when conj-ing elements to it?
16:48	_na_ka_na_	chouser: Didn't quite get that, gen-class couldn't be a fn ?
16:49	chouser	gen-class could be a function, but you give it a lot of symbol names for functions and classes that don't exist yet
16:49	_na_ka_na_	Hmm got it
16:49	chouser	if those were evaluated before being passed to a function, you'd get errors. So instead you'd have to quote all of them.
16:50	chouser	if generate-class weren't private, you could just call (generate-class '{:name ... :methods ...}) ...it would work fine.
16:50	remleduff	_na_ka_na_: macros are also for selective evaluation. You couldn't implement a short-circuiting "or" function as a function.
16:50	chouser	so that's another use-case for macros -- to simplify the interface to a function
16:51	_na_ka_na_	remleduff: we could quote the stuff to be ored?
16:51	chouser	what macros are is a hook into the compilation process. What they're for is anything you can dream up that needs to hook into the compilation process.
16:52	chouser	_na_ka_na_: not quite, because then you'd have to 'eval' those quoted things, which isn't quite the same
16:52	_na_ka_na_	Ok so I'm noting all of these points down, and will ponder over them next few days :P
16:53	chouser	without macros, 'or' could be a function that takes a bunch of no-arg functions.
16:53	chouser	(or-fn #(foo) #(bar) #(baz)) instead of (or (foo) (bar) (baz))
16:53	_na_ka_na_	chouser: how not same? I mean evaling the quoted things.
16:54	chouser	but that would still be more work at runtime than what 'or' does now. So that's another use-case -- performance improvements by doing some work at compile time that would otherwise have to be done at runtime.
16:54	chouser	not the same because eval doesn't have access to the local environment where it's called.
16:55	chouser	eg. this doesn't work: (let [a 1] (eval '(prn a)))
16:55	_na_ka_na_	Hmmm
16:55	Borkdude	_na_ka_na_: are you perhaps that guy who asked a Scala call-by-name/Clojure macro question on Twitter? :)
16:56	_na_ka_na_	One curious q : so how is (x \|\| y) implemented in imperative langs w/o macros?
16:56	chouser	_na_ka_na_: direct compiler support, usually. you can't write your own.
16:56	_na_ka_na_	Borkdude: Nope
16:57	chouser	_na_ka_na_: some langauges have explicit support for delayed arg evaluation (scala), otherer are always lazy (haskell)
16:57	_na_ka_na_	Haven't touched Scala (yet)
16:57	_na_ka_na_	Hmm, that sounds interesting
16:57	Borkdude	chouser: he (the guy on Twitter) said he could do most macro's with Scala call by name... I wonder how far he would get, don't know Scala though
16:58	chouser	Borkdude: not far at all
16:59	humphrej	Hi
16:59	humphrej	here's an add-watch puzzler http://clojure.pastebin.com/N4TauWpV
16:59	_na_ka_na_	So how does the compilation work? Does the reader first macroexpand all macros and then compile stuff?
17:00	humphrej	should a call to await cause a watcher-fn to update?
17:00	_na_ka_na_	What if there are nested macro calls like or, .. when is the next time macroexpand is called?
17:00	chouser	humphrej: old value is the same as the new value -- you can ignore if you want.
17:01	mefesto	_na_ka_na_: i believe macro expand continues until all macros are fully expanded
17:01	chouser	_na_ka_na_: each top-level form goes through several stages before the next top-level form is touched.
17:01	chouser	1. read (text to nested data structures)
17:02	chouser	2. macroexpand of the outermost form, repeatedly until it is a special operator
17:03	chouser	3. analysis of that outermost special operator. depending on the op all kinds of special things may be done. frequently steps 2 and 3 are done recursively for one or more args to this outer special form
17:03	humphrej	chouser: yes but I'm thinking the equality test could be expensive if the agent state is large
17:04	chouser	4. walk the tree that results from the analysis, emitting bytecode
17:04	chouser	5. (optionally) save bytecode to .class files
17:04	chouser	6. load and run bytecode.
17:05	chouser	1 is "read time", 2-5 are "compile time" and 6 is "runtime"
17:05	_na_ka_na_	thanks! chouser take makes a lot of things clear, are these details somewhere on the clojure.org page?
17:06	Borkdude	Am I right that an array-map of 8 kv's turns into a hash-map when conjed a new kv pair?
17:06	_na_ka_na_	one doubt: macroexpand won't expand inner level forms .. like for or .. right? they are expanded only at runtime?
17:06	chouser	note that step 2 involves calling functions provided by the user -- so what is "compile time" for the current top-level form is "runtime" for the macros being expanded.
17:06	chouser	_na_ka_na_: no, those inner levels are expanded as parts of the recursion at step 3
17:07	Borkdude	Wait, 9.
17:07	_na_ka_na_	Hmm so after step 3. macros no longer exist
17:07	Borkdude	,(type (conj (array-map 0 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8) [9 9]))
17:07	clojurebot	clojure.lang.PersistentHashMap
17:08	chouser	Borkdude: yeah, but you shouldn't rely on that.
17:08	chouser	_na_ka_na_: essentially, yes!
17:08	Borkdude	chouser: it could differ per OS or smth?
17:08	_na_ka_na_	lol that smiley looks nice in a data structure
17:08	chouser	Borkdude: per version of clojure
17:09	Borkdude	chouser: aha
17:16	_na_ka_na_	I was trying to do something like ... (defmacro m1 [x y] (println (+ x y)))
17:16	_na_ka_na_	(defmacro m2 [x y] (do (m1 x y) (println (- x y))))
17:16	_na_ka_na_	and it was not compiling .. I couldn't understand .. but now I do :)
17:16	chouser	well! good.
17:17	_na_ka_na_	Is it idiomatic to do something like .. (defmacro m2 [x y] (do (macroexpand `(m1 ~x ~y)) (println (- x y))))
17:17	chouser	I don't think so. I have used macroexpand in a macro for the very first time last week, and it seems very specialized.
17:18	_na_ka_na_	Didn't quite follow the very specialized part
17:18	chouser	the reason your m2 doesn't compile is because you're passing the symbols x and y to m1, which then tries to add them.
17:18	_na_ka_na_	yup
17:19	chouser	you might want: (defmacro m2 [x y] `(do (m1 ~x ~y) (println (- ~x ~y))))
17:20	Chousuke	except that will evaluate x and y twice
17:20	_na_ka_na_	ya for that one can use let
17:20	chouser	using eval, macroexpand, or anything with side-effects in a macro definition is pretty suspect.
17:21	chouser	actually, to a certain extent using defmacro is a bit suspect. :-P
17:21	remleduff	Hmm, doesn't any macro that expands to def have side effects?
17:21	Chousuke	remleduff: no
17:21	Chousuke	remleduff: the code they expand to has side-effects, not the macro
17:22	remleduff	ah, makes sense
17:22	Chousuke	it is possible to use the whole language in a macro. It's just not generally done. But I think for example using reflection to generate bindings for an API could be a good application of more complicated macrology.
17:23	chouser	Chousuke: yes
17:23	chouser	but you ought to be able to justify at each step why you're doing it in a macro instead of a function.
17:24	Chousuke	Doesn't penumbra do something like that btw?
17:25	Chousuke	IIRC reflects the jogl stuff and generates clojure functions/macros... Or maybe I'm confusing it with something else.
17:25	chouser	of course once you're in that deep you might want to do like clj-native and just call into the ASM lib directly.
17:27	bozhidar	I was wondering what is the idiomatic way in Clojure to cache a function return values
17:28	chouser	memoize
17:28	bozhidar	in a trivial example say we want to cache factorials between invocations
17:29	bozhidar	chouser: so I should basically just wrap my existing function with memoize and that's all?
17:30	chouser	for trivial cases, that's it.
17:30	_na_ka_na_	bozhidar: even shorter is defn-memo (c.c.def)
17:34	goodmike_mh	Hey, I'm the numbskull who asked on the Google group about how Agents are lock-free...
17:34	goodmike_mh	Are either of the authors of Agent.java online, and would you be willing to humor me?
17:35	chouser	Pretty sure Agent.java is all rhickey.
17:35	chouser	but others may be able to help
17:36	goodmike_mh	git blame tells a different tale, chouser. :-)
17:36	chouser	hmph.
17:36	Borkdude	chouser: why is it that a conjed array-map sometimes returns a hash-map?
17:36	goodmike_mh	But I'll take any help i can get
17:36	bozhidar	chouser, _na_ka_na_ : thanks
17:36	chouser	Borkdude: because array-maps are only efficient when they're small.
17:36	goodmike_mh	I can tell it's all about the agent's queue
17:37	goodmike_mh	The queue can only be updated (e.g. given a new value) in an atomic compareAndSet
17:38	goodmike_mh	Then the actions in the queue are executed
17:38	Borkdude	chouser: so 9 (varying over Clojure versions) is kind of the flipping point of efficiency?
17:38	chouser	Borkdude: that's the conclusion currently implemented, yes.
17:39	Borkdude	chouser: still you can use array-maps bigger if you want, because of ?
17:40	chouser	goodmike_mh: almost. if enqueue takes the queue size from 0 to 1, it starts a chain of executions
17:41	goodmike_mh	chouser: Yeah, I saw that! I was a little confused by why there's an explicit action.execute in case the queue was empty before the compareAndSet
17:41	chouser	goodmike_mh: after each action is done, it's popped off the queue (CAS again), and if there is another action to be done it is executed
17:41	chouser	so each action on an agent triggers the execution of the next action on the agent
17:42	goodmike_mh	chouser: I assume there is a case in which the queue was non-empty at the end of Agent#enqueue. I'm not sure what happens then.
17:43	chouser	goodmike_mh: nothing happens then. it relies on the fact that whatever action is currently in the executor will trigger the next when it's done.
17:43	goodmike_mh	chouser: Whoa. OK. I think I see that.
17:44	goodmike_mh	chouser: so it's the atomic updating of the queue that keeps actions from overlapping. It looks like it makes the agent's actions execute in serial order. Is that right?
17:45	chouser	agent's actions execute in serial order, yes.
17:46	goodmike_mh	chouser: Thanks. I really appreciate being able to ask someone who wrote the code.
17:46	goodmike_mh	chouser: Even if he doesn't remember doing it. :-)
17:46	chouser	the actions are kept from overlapping by the CAS on the queue and the use of that to make sure that only one place (either enqueue or the current action) triggers the next action.
17:47	chouser	I didn't write it! I tweaked the error handling a bit.
17:48	goodmike_mh	chouser: The CAS on the queue is really sweet.
17:48	chouser	yeah, that was definitely not me
17:48	goodmike_mh	chouser: Regardless, thanks for the explanation.
17:48	chouser	you're welcome
17:50	_na_ka_na_	what does point 5. clojure.org/agents - "If during the function execution any other dispatches are made (directly or indirectly), they will be held until after the state of the Agent has been changed." mean?
17:53	tomoj	_na_ka_na_: if you send or send-off during the action you send(-off) to an agent, those will not happen until after the current agent is finished
17:53	tomoj	I don't know the vocabulary to explain it better :(
17:55	_na_ka_na_	Hmm got that tomoj, I wanted to understand why that is so?
17:55	tomoj	oh, hmm
17:55	tomoj	well, it's useful :)
17:56	tomoj	not sure if there is a better reason... can agent actions be retried?
17:58	_na_ka_na_	don't think agent actions are ever re-tried .. but one explanation might be that the other agent to which there is a dispatch might want to see the updated state? doesn't sound like a good reason tho.
17:59	goodmike_mh	_na_ka_na_: I think the reason for holding the dispatch is to prevent contention. The conversation chouser and I just had was about how agents serialize the actions that update their state. So they have to happen one after the other.
18:01	goodmike_mh	_na_ka_na_: It's fairly common for the fn sent to an agent to include a send of the same fn to the agent, with the understanding that "this will happen later"
18:01	tomoj	like ants.clj
18:01	goodmike_mh	yep
18:02	_na_ka_na_	goodmike_mh: you mean an agent in its action fn sends an action to itself again?
18:02	kasperlanger	I think it's to make actions on agents atomic (the A in ACID)
18:02	goodmike_mh	_na_ka_na_: exactly. the fn contains a line like (send agent foo)
18:03	goodmike_mh	agent being the Var holding a reference to the current agent.
18:03	arkahn	covered in http://blip.tv/file/812787 with relevant bits at 71:43, specifically 72:50
18:03	goodmike_mh	_na_ka_na_: Are you suggesting a case in which the function sends a function to a different agent?
18:04	_na_ka_na_	hmm ok that might be an explanation then, but I can't immediately see why an agent would send to itself again
18:04	_na_ka_na_	goodmike_mh: yes, it can be a common scenario no?
18:05	arkahn	_na_ka_na_: example code at the bottom of ants.clj http://tinyurl.com/322p5kf
18:06	goodmike_mh	_na_ka_na_: I'm not sure I would do that. :-) I prefer to coordinate state outside the current agent by opening a transaction (dosync) and updating refs. Let the coordinated state stay in refs.
18:06	tomoj	an agent action can send send itself to its agent again to set up a sort of infinite loop
18:06	tomoj	and the action can do some check for whether to send again or not in order to terminate
18:06	arkahn	_na_ka_na: a function sends off to itself after checking for whether it should continue running and executing a sleep so it doesn't sit and spin too fast
18:07	tomoj	so it's a way to do the run loops you see in java threading stuff with agents
18:07	_na_ka_na_	goodmike_mh: not all situations would require coordination, for eg: agent a got a message, it just wants to tell agent b that it got a msg, doesn't care when agent b gets that msg
18:08	arkahn	it's my understanding that agents only provide for coordination so state change (i.e. the function in send or send-off) can happen safely
18:09	arkahn	you're also guaranteed on order (it's fifo)
18:10	arkahn	you can read an agent at any time but aren't guaranteed you'll see the latest changes
18:10	goodmike_mh	_na_ka_na_: Hmmm. An agent is primarily a way of encapsulating some state and controlling its updating. You have to provide behavior. They're not like objects in other languages with their own behavior. So I'm not sure that message-passing is the right approach here.
18:11	goodmike_mh	_na_ka_na_: In Clojure >= 1.1 promises are used to pass values between threads. Agent B might block on a promise until agent A sets it. You'd want to send-off the fn to B that blocks.
18:13	goodmike_mh	Just a thought. Please, no one go out and architect their systems based on my semi-informed comments.
18:15	arkahn	Ever since I learned about functions self-sending, I've wondered if it's currently common/idiomatic. I guess it's safe to say that Rich thought it was back in the day, whenever ants.clj was written.
18:16	arkahn	it seems like a working contortion of the original intent for agents, but maybe I just don't know enough ; )
18:18	nDuff	I've got a loop/recur where I'm getting "recur arg for primitive local: remaining must be matching primitive"; however, to me, the types look like they should match. Is there a way to determine what the expected vs actual types are?
18:18	_na_ka_na_	goodmike_mh: what would be a good way to implement the following scenario: agent A, agent B are doing independent work.. A reaches a state when it wants to inform B about it, but B might be busy, so it doesn't want to be synchronous
18:20	_na_ka_na_	kasperlanger: I didn't quite follow the atomicity part you mentioned
18:21	kasperlanger	If the function you send to an agent has no side effect except possibly mutating the agent and sending of stuff to other agents you know the whole thing will either happen completely or not at all
18:21	kasperlanger	That's a really nice property to have
18:24	arkahn	nDuff: could you use 'class' to verify type?
18:25	_na_ka_na_	kasperlanger: but isn't is possible that after the agent state is mutated, action is sent to another agent which fails ?!
18:25	_na_ka_na_	i mean the act of sending fails, not the action of other agent
18:25	nDuff	arkahn, ...explicitly wrapping the argument in question with (int ...) resolved the issue; I'm presuming the other value was being boxed to an Integer, perhaps?
18:26	kasperlanger	Yes the action on the second agent may fail
18:26	kasperlanger	The atomicity property only holds on a single send not the whole chain of sends
18:27	kasperlanger	oh
18:27	arkahn	nDuff: not sure without seeing the code
18:27	_na_ka_na_	didn't quite get, but how does point 5. ensure that? or if point 5. wasn't there how can the whole action not be atomic?
18:28	kasperlanger	I don't think sending fails since it's all local stuff but I'm speculating now :)
18:28	SinDoc	I'm using Clojure 1.2.0-master and have trouble with deftype
18:29	SinDoc	(deftype point [x y])
18:29	arkahn	kasperlanger: I don't think sending ever fails - but the assumption of state may have changed so you can determine at the time the function executes whether you'd like the function to complete
18:30	arkahn	kasperlanger: ... trying to remember what that precondition check is called ...
18:31	nDuff	arkahn, http://gist.github.com/415170; the comments at the end of the file point where the issue occurs
18:32	herdrick	question: lein new project_name gives you a project.clj with dependencies that are out of date, I think. org.clojure/clojure "1.1.0
18:32	herdrick	oops
18:32	nDuff	(this is some of my first clojure code, by the way, so I'm fully expecting -- and would welcome -- any critique / comments re obvious failings re practices / style failings)
18:33	herdrick	ok, so the default org.clojure/clojure version according to lein new is 1.1.0
18:33	herdrick	isn't that obsolete?
18:34	herdrick	ah, I see that's the newest version in Clojars
18:34	remleduff	1.2 isn't out just yet
18:36	_na_ka_na_	chouser: the eval behavior of clojure struck me, coz eval in Perl happens in lexical scope, what implications do the two behaviors have ?
18:37	herdrick	okremleduff: ok, thanks
18:41	islon	"lein help" gives me Exception in thread "main" java.lang.RuntimeException: java.lang.ClassCastException: clojure.lang.Cons cannot be cast to clojure.lang.Named (help.clj:5)
18:42	SinDoc	I pasted a simple deftype interaction; http://paste.lisp.org/display/100539
18:44	SinDoc	The syntax is probably not correct but this is what I understand from (doc doctype)
18:44	SinDoc	*deftype
18:44	tomoj	don't use new
18:44	tomoj	but your example still doesn't work for me after fixing that
18:45	tomoj	but e.g. this works: (deftype Point [x y]) (def pt (Point. 1 2)) (.x pt)
18:45	remleduff	deftype defines no functionality, you want defrecord
18:45	tomoj	also, (defrecord Point [x y]) (def pt (Point. 1 2)) (:x pt)
18:45	tomoj	yeah
18:46	islon	how do I start swank with my lein project? i tried "lein swank" but it gives me "swank is not a task. Use "help" to list all tasks."
18:46	remleduff	islon: Do you have lein-swank in your dev-dependencies, and have you run lein deps?
18:46	skeeterbug	islon, i had that problem, upgrading lein worked for me
18:46	Raynes	islon: You need to add swank-clojure to your project.clj file.
18:46	SinDoc	i thought fields were keywords
18:46	Raynes	remleduff: lein-swank is obsolete.
18:46	Raynes	remleduff: lein swank is bundled with the new swank-clojure.
18:46	remleduff	oh, I should upgrade :)
18:47	tomoj	http://github.com/technomancy/swank-clojure explains
18:47	Raynes	[swank-clojure "1.2.1"]
18:47	islon	thanks, is there a way to lein self upgrade? like "lein upgrade"?
18:47	Raynes	Put that in your dev-dependencies
18:47	Raynes	lein update, I believe.
18:47	skeeterbug	yeah, lein update, but i think older versions didn't have that command
18:47	Raynes	No, it's lein upgrade
18:47	tomoj	SinDoc: not on a raw deftype, apparently. for defrecord, yeah
18:48	islon	wow, works =)
18:49	remleduff	SinDoc: deftypes are not maps, so they have no default functionality for keyword lookup. There was an earlier version of deftype that defined some map functionality magically for you if you implemented IPersistentMap or something like that, but Rich wisely split it up into deftype with nothing magic and defrecord which is always map-like.
18:50	SinDoc	Yeah, i saw that defrecord implements IPersisistentMap
18:51	SinDoc	Are there recent examples of deftype?
18:51	SinDoc	http://paste.lisp.org/display/90329
18:52	SinDoc	I found this but doesn't work given the recent changes
18:54	_na_ka_na_	Hey guys I'm writing 2 macros .. first is straightforward .. (defmacro m1 [x] (println "m1 got" x)) ..
18:54	tomoj	SinDoc: https://gist.github.com/ad9cad85dc72c41dd52b
18:54	tomoj	that seems to work
18:54	_na_ka_na_	but in the second, i want to simulate something like :
18:54	_na_ka_na_	(defmacro m2 [y] (macroexpand `(m1 ~y)) (println "m2 got" y))
18:55	_na_ka_na_	only w/o the macroexpand part ..
18:55	_na_ka_na_	how can it be done ?
18:56	SinDoc	tomoj: You're a grace; thanks
18:56	SinDoc	it uses defrecord though
18:56	tomoj	_na_ka_na_: have you tried macroexpanding?
18:56	tomoj	SinDoc: yes, because that's what you should use to do that example nowadays
18:57	tomoj	deftype has similar syntax but no automatic :real :imaginary access
18:57	_na_ka_na_	tomoj: didn't get what you meant ?
18:57	tomoj	_na_ka_na_: I mean, for example, have you tried (macroexpand-1 '(m1 3)) ?
18:57	_na_ka_na_	yes
18:58	tomoj	so you noticed, then, that it expands to nil?
18:58	_na_ka_na_	yup
18:58	tomoj	so why would you want to expand nil and stick that in m2?
18:58	tomoj	anyway.. (defmacro m2 [y] `(do (m1 ~y) (println "m2 got" ~y))) maybe?
18:58	SinDoc	tomoj: I'm looking for a way to write ADTs and currently I was hoping to write an abstraction that does memory management so I should export thinks like cons, car, cdr, etc.
18:59	_na_ka_na_	no but the above definition of m2 work s ... it calls m1
18:59	_na_ka_na_	i mean the macroexpand one
18:59	tomoj	yes, the one I just pasted calls m1 as well
18:59	arkahn	nDuff: it has to do with code in clojure/src/jvm/clojure/lang/Compiler.java near "public static class RecurExpr implements Expr". I'm not understanding it all yet, other than it looks like it's thrown off by the method call you make to #^java.nio.ByteBuffer .remaining which returns a java int. So far, it looks to me like the implementation recur doesn't like mixing boxed and unboxed primitives. Maybe immediately box the return value with Integer
18:59	arkahn	right away? All subsequent clojure manipulations would then treat it like a clojure Number ... maybe
18:59	tomoj	it expands like (m2 5) -> (do (user/m1 5) (clojure.core/println "m2 got" 5))
19:00	tomoj	SinDoc: dunno how to help you, good luck
19:01	SinDoc	tomoj: thanks for the help
19:01	nDuff	arkahn, ...hmm. Would you suggest testing this against a newer upstream (this is 1.1) and filing a ticket if the behavior is unchanged?
19:02	_na_ka_na_	tomoj: but I want both macros to run at compile time, the one you pasted runs at run time
19:02	_na_ka_na_	I mean macroexpand m2 should return null
19:03	tomoj	I see
19:03	arkahn	nDuff: I really don't know. What was the design goal for recur? To be able to mix clojure and java primitives? You could always file a ticket and find out. Unfortunately, I've probably been using clojure for as long or less than you have, so I'm not the best person to ask ; )
19:03	_na_ka_na_	thats how (defmacro m2 [y] (macroexpand `(m1 ~y)) (println "m2 got" y)) will work
19:03	tomoj	yeah
19:03	tomoj	weird
19:03	arkahn	nDuff: fwiw (which isn't much, considering my noob-ness) I like your code
19:04	nDuff	arkahn, heh -- thanks (both for taking the time to look into the issue, and providing at least some kind of feedback that I wasn't making _really_ obvious errors)
19:05	arkahn	nDuff: np! It's always nice to get a sanity check ... I just hope what I dug up was sane
19:05	tomoj	_na_ka_na_: I think you might be stuck with the macroexpand, then, not sure
19:06	tomoj	_na_ka_na_: just curious, what's the point?
19:06	_na_ka_na_	tomoj: I'm trying to write a macro wrapper over gen-class
19:08	tomoj	and you can't wrap gen-class with a normal-style macro? never would have thought otherwise, interesting
19:08	tomoj	I thought you were struggling with macros, but you know them better than I :D
19:11	_na_ka_na_	tomoj: Not sure about that, but I've been struggling alright! This is like my 2nd macro which I've had to write .. I'm done with the macroexpand solution .. but chouser above said that one shouldn't use macroexpand inside a macro
19:14	_na_ka_na_	What's the easiest way to run arbitrary shell commands from Clojure, example I want to, cd into a directory, and kick off a process from there ..
19:25	_na_ka_na_	found c.c.shell
19:28	_na_ka_na_	,(sh :cmd "echo Hello Thr" :return-map true)
19:28	clojurebot	java.lang.Exception: Unable to resolve symbol: sh in this context
19:28	_na_ka_na_	(sh :cmd "echo Hello Thr" :return-map true) throws an exception for me .. java.lang.IllegalArgumentException: No matching method found: exec for class java.lang.Runtime
19:31	chouser	_na_ka_na_: try (sh "echo" "Hello" "There" :return-map true)
19:32	_na_ka_na_	oh I thought you forgot to mention the :cmd key in the docs .. after looking at the source
19:32	_na_ka_na_	wasn't sure how to pass the cmd
19:33	chouser	yeah, the docs could use some help there
19:35	ihodes	anyone in here familiar with Enlive
19:35	ihodes	?
19:36	ihodes	i've been getting "java.lang.NullPointerException ", pointing to a line where i defsnippet, is the issue.
19:38	_na_ka_na_	chouser: maybe you could put your example above in the docs .. and this for Win XP works for me.. (sh "cmd.exe" "/c" "echo" "Hello" "There" :return-map true)
20:05	ihodes	anyone here yet? i'm really interested in learning enlive, but i'm having a small problem i'm sure someone's seen before
20:14	islon	is clojure.contrib.json the best way to read json in clojure?
20:14	ihodes	as far as i know, it's the only way. and if it's in contrib, chances are it does what it's supposed to to, well.
20:20	ihodes	fair enough. technomancy, would you be able to answer a quick question re enlive for me?
20:20	ihodes	i've been getting "java.lang.NullPointerException ", pointing to a line where i defsnippet, is the issue.
20:20	islon	well, json parsing is not that difficult
20:20	danlarkin	ihodes: no!
20:20	danlarkin	he's busy helping me
20:21	technomancy	I've only used enlive once... I was pretty confused the whole time.
20:21	ihodes	ah, ok haha. for some reason i'd thought you'd somehow been involved. wrooong person. know someone who might be able to help? frustrating little problem
20:22	technomancy	cgrand is the author; there's also dnolen who's written some docs for it
20:22	technomancy	neither is around now =
20:22	technomancy	=\
20:22	technomancy	nor lau
20:22	ihodes	noticed that. siiigh. it looks so nice, but mystery error is a pain.
20:22	ihodes	yeah lau did what i'm trying to do with it ahah, though i didn't know until last night
20:25	hugod	ihodes: I think a NPE at that line means that the path to the template is not being found
20:26	ihodes	unfortunately, it's an absolute path right now, so I don'tt hink that's it.
20:28	ihodes	though now i've changed the selector for the snippet and get this: java.lang.ArrayIndexOutOfBoundsException: 1 ---- on the same line
20:31	ihodes	http://pastie.org/979148 is the snippet that is being hated on. there's definitely a file there. i require enlive in as html.
20:44	herdrick	say, can anyone tell me why this: (ns user (:use [incanter.core :only (abs)])) gives me "java.lang.IllegalArgumentException: Don't know how to create ISeq from: clojure.lang.Symbol"
20:44	herdrick	while this: (ns user (:use [clojure.set :only (union)]))
20:44	herdrick	works fine ?
20:45	herdrick	substituting various things for clojure.set works fine too
20:45	herdrick	like clojure.contrib.combinatorics
20:45	herdrick	or whatever
20:45	herdrick	(assuming you replace union with something in one of those packages
20:45	herdrick	)
20:46	tomoj	is that the exact code which caused the error?
20:47	herdrick	yep
20:48	tomoj	can you try both and pastie the repl session?
20:48	herdrick	sure
20:48	herdrick	brb
20:48	tomoj	I'm just boggled
20:50	herdrick	http://pastie.org/979167
20:50	herdrick	this is a fresh install of swank-clojure
20:50	herdrick	and the first time i'm using lein
20:50	tomoj	no idea why that could happen
20:51	herdrick	so probably something wrong with what i'm doing there
20:51	tomoj	does (use '[incanter.core :only (abs)]) also give the error?
20:51	herdrick	yep, identical error
20:51	tomoj	and, does the error come with a stacktrace? I wonder where the error is happening?
20:51	tomoj	maybe it's inside incanter? that's the only explanation I can think of..
20:52	tomoj	your syntax looks perfect to me
20:52	herdrick	whereas (use '[incanter.coreBAD :only (abs)]) gives what you'd expect: "Could not locate incanter/coreBAD__init.class or incanter/coreBAD.clj on classpath: ... "
20:52	herdrick	brb
20:52	hugod	ihodes: the braces around the snippet selector argument look wrong to me
20:53	tomoj	herdrick: what versions of incanter and clojure are you depending on?
20:54	herdrick	here's the stacktrace and it's cause: http://pastie.org/979169
20:54	herdrick	incanter 1.2.3
20:54	herdrick	clojure 1.1.0
20:57	tomoj	hmm
20:57	herdrick	uh oh -
20:57	herdrick	looks like incanter's project.clj file wants clojure 1.2.0
20:57	tomoj	I think incanter 1.2.3 requires clojure 1.2.0
20:57	tomoj	yeah
20:57	herdrick	roger
20:58	tomoj	that's the only way this error makes sense to me, anyway
20:58	herdrick	ok, thanks, that's probably it
20:59	herdrick	i was following this blog post from November: http://data-sorcery.org/2009/11/20/leiningen-clojars/
20:59	herdrick	which is like 14 clojure dog years ago
20:59	herdrick	my bad ;)
20:59	tomoj	hehe
21:01	Drakeson	Is there a facility to work with paths (like file system paths)?
21:01	replaca	Drakeson: I use java.io.File
21:03	Drakeson	replaca: maybe I miss something. How do you do (path :home "foo/bar") using java.io.File ?
21:04	Drakeson	is there something easier than (str (System.getProperty "user.home") "/" "foo/bar") ?
21:11	replaca	Drakeson: More corrrect would be (File. (System.getProperty "user.home) (File. "foo") "bar)
21:11	replaca	I think - I'm not buried in that java space
21:12	replaca	easy to imagine a path func that did (path :home "foo" "bar")
21:16	herdrick	tomoj: ok, that worked
21:16	herdrick	thanks!
21:18	herdrick	By the way, since Clojure 1.2 isn't in lein-space, I just went with the older version of Incanter that is OK with Clojure 1.1
21:19	herdrick	but now I've got a bunch of Incanter 1.2.3 jars in my /lib dir of my lein-created project
21:19	herdrick	those shouldn't be a problem, right?
21:19	ihodes	hugod: i tried removing them: no luck. it's also a selector from david nolen's tutorial :\
21:19	herdrick	unless i were to use swank-clojure-project (the avoidance of which was the point of this whole switch to lein)
21:23	bhenry	herdrick lein clean && lein deps
21:23	herdrick	bhenry: ok. but is it important?
21:23	bhenry	yes
21:24	herdrick	will swank or anything else pick up those extra jars in lib?
21:25	Drakeson	herdrick: What are you trying to do?
21:25	herdrick	Drakeson: just curious about the issue
21:26	herdrick	my problem was solved above
21:27	Drakeson	oh sorry
21:28	bhenry	herdrick: everything in lib/ is on your class path. so if you (use 'some-project) you won't know which one you'll get (that's if it doesn't error, i'm not sure what would happen, but I know it's important). someone speak up if i'm misdirecting herdrick.
21:29	tomoj	herdrick: what do you mean 1.2.3 isn't in lein-space?
21:29	herdrick	bhenry: i was thinking lein would manage that, and generate a custom classpath using only the jars of the deps and their deps, recursively, etc
21:29	tomoj	no, you need to remove duplicate jars
21:29	tomoj	bhenry is correct
21:30	herdrick	No, Clojure 1.2 isn't in lein-space
21:30	tomoj	oh, hmm
21:30	tomoj	you mean that you can't get it with lein?
21:30	herdrick	well, i couldn't in a couple of tries
21:30	tomoj	you can: [clojure "1.2.0-master-SNAPSHOT"] and [clojure-contrib "1.2.0-SNAPSHOT] or something like that
21:30	herdrick	i'm new to lein tho
21:30	herdrick	tomoj: bhenry: ok, thanks for the info on lein and classpaths
21:31	tomoj	also would need to use 1.2.3-SNAPSHOT for incanter
21:31	ihodes	anyone here an enlive expert yet? ;) still having a fun little NullPointerException http://pastie.org/979148
21:31	herdrick	trying that
21:31	tomoj	but using an old version and 1.1.0 is not too bad either
21:31	tomoj	ihodes: what's {[:.cg-post]} supposed to be?
21:31	tomoj	I'm surprised that doesn't cause an error
21:32	tomoj	,'{[:.cg-post]}
21:32	clojurebot	1
21:32	tomoj	uh, what?
21:32	bhenry	herdrick, if you have a lot of dependencies in your projects, and you're updating them a lot, alias lcd to lein clean && lein deps then just use that every time. it hardly takes any additional time, since lein stores things in a local repository on your computer.
21:32	tomoj	oh, the error message is "1", haha
21:32	ihodes	it's a selector: Noel used it in his tutorial: (def section-sel {[:.title] [[:.content (nth-of-type 1)]]})
21:32	ihodes	nolen*
21:32	ihodes	http://github.com/swannodette/enlive-tutorial/
21:32	tomoj	ihodes: ok, but your {} is wrapped wrong
21:33	tomoj	you've got it just around [:.cg-post]
21:33	tomoj	which looks like a map with a key with no value, which is illegal
21:33	ihodes	true: i did change it to "[:.cg-post]" in the version i'm playing with now
21:33	tomoj	ah
21:33	ihodes	that was a mistake in the pastie when i was messing with stuff, i'm sorry!
21:33	herdrick	say, i've been using [org.clojure/clojure "1.1.0"] instead of [clojure "1.1.0"] does that matter?
21:33	tomoj	no problem
21:33	herdrick	bhenry: ok, thanks looks like a good tip
21:34	ihodes	tomoj: the problem remains, though, unfortunately
21:34	tomoj	herdrick: nope, clojure is just an abbreviation for org.clojure/clojure
21:34	herdrick	ok
21:34	tomoj	well, I'm betting the problem can't be determined without seeing the html, but not sure
21:34	tomoj	and I am not an enlive expert, good luck
21:35	ihodes	it's a compile-time error though
21:35	tomoj	ah
21:35	ihodes	as it occurs when i'm loading the file into the repl
21:35	tomoj	that's odd, must be bad syntax then
21:36	ihodes	it looks exactly like the syntax in nolen's tutorial, i think. gah. ben banging my head against a wall about this for a while. it's what happens when you code in C the first half of the day, and then clojure the other half.
21:37	tomoj	you need something to bind to the keys destructuring map
21:37	tomoj	(I think)
21:37	ihodes	i don't quite understand?
21:37	tomoj	I mean, [{:keys [title author pubdate post-text]} <something needs to go here>]
21:38	tomoj	e.g. the tutorial has [{:keys [title data]} model]
21:38	tomoj	though I can't tell where model came from
21:38	tomoj	oh, maybe that's just an extra param?
21:39	tomoj	yeah, I was wrong, just the {:keys ..} should work too it seems
21:39	ihodes	yeah it's just an additional arg: the {:key..} thing is just destructuring a map
21:41	ihodes	java.lang.NullPointerException (render_index.clj:19) is the exception, too
21:46	tomoj	what's on line 19, the start of the defsnippet?
21:47	ihodes	yep
21:59	tomoj	ihodes: I don't think this problem is your fault
22:00	tomoj	the examples from the tutorial seem to do the same thing
22:01	ihodes	that simultaneously good to hear. and awful! in his examples, it all seems to work. he is using compojure 0.3.2, which might be the cause of the problems, but it's highly unlikely as .4 doesn't change much, and nothing that enlive relies on.
22:01	tomoj	it seems like load-html-resource is being called with nil
22:01	tomoj	which clojure is he using?
22:01	ihodes	the thing is, his examples work in the lein-powered environment he has on hit.
22:01	tomoj	huh, he doesn't have any clojure in his project.clj
22:01	remleduff	NPEs in enlive have always meant "couldn't find your template" for me
22:01	ihodes	i'm going to have to build up his examples (M-w, C-y) in a new project
22:02	tomoj	remleduff: yeah, but..
22:02	tomoj	oh, hmm
22:02	tomoj	does it try loading from the classpath?
22:02	remleduff	What do you get if you just do (html-resource "your_template.html")
22:02	hugod	ihodes: the path should be relative
22:02	tomoj	rather than taking it as an absolute file path
22:02	tomoj	I see
22:02	hugod	it uses the resource loader
22:02	ihodes	ah, let me give that a go
22:02	ihodes	so an absolute path won't work?
22:02	tomoj	so the html file needs to be in resources/ I guess, and you need to have that directory there before you start up swank
22:03	tomoj	(does `lein swank` add resources/ to the classpath?)
22:03	remleduff	resources is a recommendation rather than a requirement. Current lein adds it to classpath, but there were older versions that didn't
22:03	tomoj	swannodette just sticks his html files in src along with the clj
22:04	ihodes	let me try just doing that
22:05	tomoj	you'll need to use the relative path from src/ if you put them there
22:05	tomoj	e.g. "foo/bar.html" for src/foo/bar.html, I mean
22:06	ihodes	ok, so i moved the template to the same folder my render-index.clj is in (the file I've been futzing with) and change the path from absolute to clogs/template_index.html (clogs being my project's name), and it appears to now work.
22:07	ihodes	well, i'm on to other exceptions on other lines anyway. which is progress.
22:07	remleduff	hehe
22:08	ihodes	so is there a nice way I can have the HTML files in other folders..? I had in my project.clj …:resources-path "html") and my html (templates) in that folder. i thought that'd work, but it didn't so I tried absolute path which also didn't. any clues as to why those things don't work?
22:09	remleduff	I'm not sure what resources-path does, but by the time lein reads your project.clj it's too late to change the classpath so I can't imagine it working. Probably just using resources would work
22:11	ihodes	resources-path should be setting the name of the resources to folder to whatever i specify–apparently it doesn't work, as I just tested and I can refer to my template at recoures/template.html
22:11	ihodes	which is nice. odd. now I have another cryptic error.
22:13	remleduff	What's that?
22:14	ihodes	tomoj: before you scroll off the screen, thank you :) remleduff: one of these buggers: java.lang.RuntimeException: java.lang.IllegalArgumentException: Don't know how to create ISeq from: java.lang.Character (render_index.clj:28)
22:16	ihodes	solved.
22:31	ihodes	thanks for the help, all. i'm running into more off errors, so i'm going to call it a night and do some set theory.
22:31	ihodes	odd* errors. finals time is not a good time to try to start a new project. cheers, and thanks again.
22:36	remleduff	Why might M-x slime give me "let: Symbol's function definition is void: swank-clojure-reset-implementation"?
22:37	tomoj	something is looking for swank-clojure-reset-implementation and you don't have it
22:37	tomoj	I guess you probably figured that much out :(
22:38	remleduff	I usually use lein swank to start up, was trying to start up a repl without a project this time
22:39	remleduff	I'm so bad with emacs still
22:40	tomoj	remleduff: do you have swank-clojure installed from elpa?
22:53	defn	technomancy: did you change the URL for your package.el repo
23:00	remleduff	tomoj: I have swank-clojure 1.1.0 from ELPA, yes.
23:03	tomoj	remleduff: I think you need to upgrade
23:04	tomoj	it looks like swank-clojure-reset-implementation is in 1.2.0 but not 1.1.0
23:06	herdrick	so you've got to keep your version of swank-clojure installed in ELPA synched with your swank-clojure jars in your lein projects?
23:06	remleduff	I thought swank-clojure was added to your project. What does swank-clojure in emacs do?
23:06	herdrick	tomoj: is that true?
23:06	tomoj	herdrick: I don't know if that's true or not
23:06	tomoj	swank-clojure from elpa is no longer even required
23:07	herdrick	oh, really?
23:07	tomoj	some stuff will not work without it
23:07	tomoj	presumably M-x slime for example
23:07	tomoj	I don't know what problems a version mismatch might cause
23:07	herdrick	hmm
23:07	herdrick	perhaps better to delete it then
23:07	tomoj	remleduff: swank-clojure in emacs is not needed if you only connect to externally started swank servers
23:08	tomoj	better to ask technomancy
23:08	tomoj	about version mismatches I mean
23:09	herdrick	ive got a related question: when i drop a jar (in this case, an apache commons one) into my /lib under my lein project, it is not picked up by slime/swank until and restart the swank instance
23:09	herdrick	is there a better way?
23:09	remleduff	hendrick: Nope :\
23:09	technomancy	defn: haven't touched that in a while, no.
23:10	technomancy	remleduff: swank-clojure in emacs isn't really needed
23:10	technomancy	you just need slime + clojure-mode
23:10	remleduff	But swank-clojure is what provides M-x slime if I'm just wanting to play with one-off stuff?
23:10	herdrick	tomoj: remleduff: technomancy: ok, thanks
23:13	remleduff	Duh, have to (require 'swank-clojure) if I want to use stuff provided by it (including M-x slime)
23:13	technomancy	remleduff: yeah, that's true
23:14	tomoj	you shouldn't have to require it, I think...
23:14	tomoj	but if it works, who cares
23:15	SinDoc	Can a type refer to itself?
23:16	Raynes	SinDoc: I believe so.
23:17	remleduff	technomancy: How do you usually start up swank if you're hacking on clojure itselfl?
23:17	SinDoc	Raynes: thanks, would you know how?
23:17	Raynes	SinDoc: You have to instantiate the class directly, because the factory functions aren't there yet.
23:17	technomancy	remleduff: I start a repl in swank and then edit clojure in the jar file itself
23:19	SinDoc	Raynes: if I need the name of the class in order to instantiate it then it wouldn't work since the name won't have been bound yet.
23:24	Raynes	SinDoc: http://gist.github.com/415409
23:31	remleduff	Anyone got a good way to check if a deftype has successfully implemented all the abstract methods its inherited, and not forgotten any?
23:47	SinDoc	remleduff: checking the protocols it has implemented using reflection?
23:50	SinDoc	Raynes: Thanks, I tried your solution; it still can't reify itself but it's fine for now
23:51	Raynes	:(
23:51	Raynes	Sorry I couldn't be more helpful. <3
23:52	SinDoc	Raynes: FWIW, I found out that my initial requirement wasn't valid to begin with