#clojure – 2015-12-09

2015-12-09

00:00	dongcarl	justin_smith: wow... clojure syntax is tricky...
00:00	justin_smith	owlbird: eval is like riding a sherman tank down the block to get a soda from the convenience store two buildings away
00:00	owlbird	The fun will be given to :row-fn while querying data from db.
00:00	justin_smith	owlbird: maybe you just want identity
00:00	justin_smith	,(identity 42)
00:01	clojurebot	42
00:02	owlbird	no, it's just a sample code, there may be many styles, each one will has its row-fn to format the result-set
00:02	justin_smith	owlbird: why would you call eval?
00:03	owlbird	if no style fun was found, just return the original value
00:03	justin_smith	owlbird: eval is like the worst possible way to return the original value, replace it with identity
00:03	justin_smith	that's what I was saying
00:03	justin_smith	also if you give eval a list with symbols you probably wouldn't like the result
00:04	owlbird	oh... I got you
00:04	owlbird	many thx
00:06	justin_smith	owlbird: eval is for compiling code, it just happens to act like identity for inputs that are not lists
00:06	justin_smith	or symbols
00:21	dongcarl	justin_smith: question, when it calls one of the defmethods, is the argument the same as the argument for defmulti or is it ch?
00:22	Trioxin	it's too bad this doesn't seem to be actively developed anymore https://github.com/arcadia-unity/Arcadia
00:27	justin_smith	dongcarl: it gets the same argument the dispatch function did
00:31	Trioxin	what would one use in terms of an engine to begin game dev in clojure?
00:33	dongcarl	justin_smith: can you help me understand what's going on here: https://github.com/shaunlebron/parinfer/blob/master/lib/src/parinfer/reader.cljc#L95
00:33	dongcarl	or anyone really
00:33	dongcarl	here's what a "state" usually looks like https://github.com/shaunlebron/parinfer/blob/master/lib/src/parinfer/indent_mode.cljc#L19
00:34	dongcarl	it's calling (select-keys state [:x-pos :ch]), but state doesn't have keys :x-pos or :ch
00:34	dongcarl	stack does
00:34	dongcarl	so is it just zeroing out the values in stack using conj?
00:37	ridcully	Trioxin: libgdx has a wrapper: play-clj
00:39	ridcully	Trioxin: depending on the kind of game you plan, something like quil could be enough. also consider clojurescript
02:55	visof	hi guys
02:56	visof	what is the best effiecient way to read multiple url in parallel?, what i know is make each one in future then use @ for all is this the right solution?
03:16	Guest53639	%CLEAR
03:16	Guest53639	CLEAR
03:23	riceandbeans	why does hello world take 13.28 seconds to run?
03:24	visof	riceandbeans: that's jvm
03:24	visof	riceandbeans: you run it as jar?
03:25	visof	riceandbeans: use lein uberjar to get the jar
03:25	amalloy	it's not the jvm, it's lein
03:26	visof	amalloy: why lein get all of this time?
03:27	amalloy	because it loads a bunch of crap, does a bunch of scanning the classpath? i don't know exactly
03:29	riceandbeans	good god
03:29	riceandbeans	22 seconds to build a hello world jar
03:30	riceandbeans	ok, so java jar hello world takes 5.21 seconds
03:30	riceandbeans	still less than awesome
03:31	riceandbeans	yes, I'm totally judging your language based on hello world
03:37	visof	riceandbeans: https://github.com/technomancy/leiningen/wiki/Faster
04:30	owlbird	why do the clojure.jdbc functions like db-do-prepared, update!, delete! return a seq, while the original jdbc API return int
04:50	Gamer-Pro	Hello Guys
04:50	Gamer-Pro	Download this game from here
04:50	Gamer-Pro	http://www.indiedb.com/games/spaceshooter-x1/downloads/spaceshooter-x1-update-for-windows
04:50	Gamer-Pro	This game is the best game of the world
04:55	BRODUS	anyone know what version of clojure 4clojure uses ?
04:57	powered	1.6 I think or lower
04:57	powered	because some 1.7 functions don't work
05:06	BRODUS	powered: thanks. I think its at least lower than 1.5 since its not recognizing reduced.
05:36	jonathanj	if anyone has some time to spare on some code golf, it would be appreciated: https://pb.codehash.net/5276b95b8f404573b3ff1db95717e4ca
06:06	parsing_clj	Hi there! I want to write a parser and im using insta parse
06:07	parsing_clj	this is my grammar so far ,modified from the README: http://pastebin.com/3wqcZzLc
06:08	parsing_clj	I want to parse THIS IS A HEADLINE, the same as THIS IS A HEADLINE
06:08	parsing_clj	so basically ignore the white spaces, does anyone knoe how to do that?
06:11	ARM9	parsing_clj try '**' for head
06:11	parsing_clj	ARM9: ok
06:15	ridcully	jonathanj: the merge looks fine to me. i'd extract that interop chain into something nicely named
06:22	visof	parsing_clj: you use trim first
06:22	visof	,(trim " THIS IS A HEADLINE ") parsing_clj
06:23	clojurebot	#error {\n :cause "Unable to resolve symbol: trim in this context"\n :via\n [{:type clojure.lang.Compiler$CompilerException\n :message "java.lang.RuntimeException: Unable to resolve symbol: trim in this context, compiling:(NO_SOURCE_PATH:0:0)"\n :at [clojure.lang.Compiler analyze "Compiler.java" 6704]}\n {:type java.lang.RuntimeException\n :message "Unable to resolve symbol: trim in this co...
06:23	visof	,(clojure.string/trim " THIS IS A HEADLINE ") parsing_clj
06:23	clojurebot	" THIS IS A HEADLINE "
06:23	visof	oh
06:24	ARM9	what did you have in mind?
06:24	visof	parsing_clj: you can use tokens lib stanfordnlp
06:24	ARM9	it'd be nice if instaparse had some sort of tokenizer
06:25	visof	stanford-talk has very good example process-text
06:25	l1x	hi guys
06:25	l1x	i am trying to use this java lib in clojure http://basho.github.io/riak-java-client/2.0.1/com/basho/riak/client/api/RiakClient.html#newClient(java.net.InetSocketAddress...)
06:26	l1x	https://gist.github.com/l1x/98a242724dc5762c0677
06:26	owlbird	how to display method name in logs while using tools.logging with log4j
06:26	l1x	i was wondering why i could not pass in an InetSocketAddress
06:27	owlbird	I use the %M pattern, but all methods was shown as "invoke"
06:27	visof	have you (import '[com.basho.riak.client.api RiakClient])
06:27	visof	?
06:27	l1x	yes
06:28	l1x	it says no matching method
06:28	l1x	what does ... mean in java? isnt that a linkedlist?
06:28	ARM9	probably varargs
06:29	visof	l1x: RiakClient.newClient("192.168.1.1","192.168.1.2","192.168.1.3"); but you pass b
06:29	visof	is it working with this example?
06:29	l1x	no this is the java way
06:30	l1x	(RiakClient/newClient 10017 (java.util.LinkedList. '("127.0.0.1" )))
06:30	l1x	the clojure version works
06:30	l1x	i am wondering what is addresses
06:30	l1x	or how could i figure out its type
06:32	schmir	l1x: you need to pass something like (into-array java.net.InetSocketAddress [...])
06:32	schmir	(as the only argument)
06:33	l1x	schmir: thanks!
06:35	l1x	schmir: how would you do that? (into-array (InetSocketAddress. "127.0.0.1" 10017)) Don't know how to create ISeq from
06:36	l1x	(RiakClient/newClient (into-array [(InetSocketAddress. "127.0.0.1" 10017)]))
06:36	l1x	like this :)
06:36	l1x	thanks!
06:37	schmir	(into-array java.net.InetSocketAddress [(java.net.InetSocketAddress. "foo" 3123)])
06:44	jonathanj	so ring.util.io/piped-input-stream uses a nested future to do it's thing
06:44	jonathanj	Source: https://github.com/ring-clojure/ring/blob/1.4.0/ring-core/src/ring/util/io.clj#L11
06:45	jonathanj	the problem with that is that anything that goes wrong in `func` goes completely unnoticed and unhandled
06:47	jonathanj	would it make sense to do something like: (let [rethrow #(throw %)] (piped-input-stream (fn [output] (try (... some code ...) (catch Exception e (rethrow e))))))?
06:48	jonathanj	is it possible to attach a listener that is executed when a future is realized?
06:49	jonathanj	deref-ing a future without knowing if it's going to block or not is not ideal
06:52	tdammers	if you don't want it to block, then maybe a future isn't the right primitive to use?
06:55	jonathanj__	tdammers: what are my other options?
06:57	tdammers	well, you could roll your own code using atoms and lambdas
06:58	tdammers	but (realized? f) would probably fit the bill
06:58	tdammers	i.e., only deref the future if realized? says it's ready
07:41	jonathanj	tdammers: i guess the question then is what do you do if it's not realized by the time you test it?
07:43	tdammers	jonathanj: well, what can you do? fail hard, or mitigate. It depends on what you want it to do in that case
07:43	jonathanj	there's a third choice: wait some more
07:43	tdammers	I'd file that under "mitigate"
07:44	tdammers	but yeah
07:44	jonathanj	i see manifold can coerce futures to deferreds, which means i can use (chain) add a callback
07:44	tdammers	another option would be to restructure your code such that the part that blocks on the future runs in a separate thread
07:44	jonathanj	it's kind of annoying there isn't anything in the core library that is effectively (wait-for-this-dereffable-then-do derefable f)
07:45	jonathanj	i suppose it's probably a lot of complexity
07:49	l1x	how can i translate the chained java methods to Clojure the easiest? new StoreValue.Builder(riakObject) .withLocation(location)
07:49	l1x	.withOption(Option.W, new Quorum(3)).build();
07:49	l1x	(.something (.withOption ....) right?
07:50	l1x	there is the .. macro
07:54	jonathanj	i think: (.. (StoreValue.Builder. riakObject) (withLocation location) (withOption Option/W (Quorum. 3) build)
07:54	l1x	nice!
07:54	l1x	thanks jonathanj
07:54	jonathanj	i missed a ) before the last build
07:55	l1x	isnt StoreValue.Builder supposed to be StoreValue/Builder?
07:55	jonathanj	i don't know, possibly, it kind of depends on what it is
07:56	jonathanj	if it's a static, then yes
07:56	l1x	thank you
07:56	jonathanj	if it's an inner class then it gets a bit hairier
07:57	jonathanj	i think you basically turn the . into $ for inner classes
07:57	jonathanj	http://blog.jayfields.com/2011/01/clojure-using-java-inner-classes.html
07:57	l1x	huh, never done that
07:58	l1x	i am trying out
08:00	jonathanj	from the docs, it looks like it is an inner class
08:00	jonathanj	so you need to import StoreValue$Builder and address it like that
08:07	l1x	https://gist.github.com/l1x/f78b129367a1a6ce80b3
08:07	l1x	like this?
08:08	jonathanj	yes
08:09	l1x	cool now i need to figure out how to chain them
08:10	l1x	store (.build (.withLocation riak-location (StoreValue/Builder. riak-object)))
08:10	l1x	something like that
08:10	jonathanj	the name is StoreValue$Builder
08:10	l1x	this works -> (StoreValue$Builder. riak-object)
08:11	jonathanj	you can use .. to pull that structure inside out
08:11	l1x	yes
08:12	l1x	(.withLocation (StoreValue$Builder. riak-object) riak-location)
08:12	l1x	this also works
08:12	yenda	Does it make sense if I have a clojure app to build a cljs interface for it ?
08:12	l1x	jonathanj: yeah
08:14	jonathanj	i'm assuming withLocation mutates the object and returns `this`, in which case (doto) might also be useful
08:15	gfredericks	doto doesn't need the object to be returned
08:15	gfredericks	methods that return `this` can be used with ->
08:16	jonathanj	what i meant was: since the return value is always the same object you can use doto instead of .. to chain the calls
08:16	l1x	https://basho.github.io/riak-java-client/2.0.0/com/basho/riak/client/api/commands/kv/StoreValue.Option.html
08:16	jonathanj	which gives you a bit more flexibility because if you have something that is not part of a chained call, you can put it in your (doto) form, whereas with (..) you need to start a new form
08:17	l1x	i guess i need to import this
08:17	l1x	with the $ way
08:17	jonathanj	gfredericks: my experience with Java interop is that -> is not often useful because the order of the arguments wobbles around a lot
08:18	jonathanj	but i guess those chained calls would work perfectly well
08:22	ridcully	also doto would not return the build-result, but the builder?
08:24	jonathanj	yeah, that's true
08:24	jonathanj	how the heck does (is (thrown? ...)) work?
08:25	luma	it's magic
08:25	jonathanj	i went looking for a function `thrown?` and found nothing, then read the docs for clojure.test
08:26	l1x	jonathanj: thank you very much for your help
08:26	l1x	it works! :)
08:26	l1x	https://gist.github.com/l1x/f78b129367a1a6ce80b3
08:26	luma	the is-macro checks if the first item of the list is 'thrown?, and acts accordingly if it is
08:26	l1x	do you think I should rewrite it using the .. ?
08:26	jonathanj	how do i use thrown? with ex-info?
08:27	jonathanj	l1x: or -> like gfredericks suggested
08:27	luma	well, the only thing you can check is if an exception of a specific class is thrown
08:27	luma	or specific class and message with thrown-with-msg?
08:27	jonathanj	luma: yeah... (ex-info) kind of hides that from me
08:28	luma	ex-info creates an exception of class ExceptionInfo, you can check for that
08:28	jonathanj	can i test the info on an ex-info?
08:28	luma	no
08:28	jonathanj	so i guess i write my own (try)
08:29	luma	or yes, but then you have to manually catch it with (try ... (catch ExceptionInfo ex ...))
08:30	jonathanj	i think maybe it's time to use midje
08:31	l1x	thanks
08:32	jonathanj	l1x: :)
08:45	BRODUS	any ideas why this simple try catch block isn't working
08:45	BRODUS	(try (keys [1 2 3]) (catch Exception e (println "error")))
08:45	opqdonut	,(keys [1 2 3])
08:45	clojurebot	#<ClassCastException java.lang.ClassCastException: java.lang.Long cannot be cast to java.util.Map$Entry>
08:45	BRODUS	right, I expect it to fail
08:46	opqdonut	I just checked what the exception is
08:46	opqdonut	an Error wouldn't get caught, but it's an Exception
08:46	beaky	,(try (keys [1 2 3]) (catch Exception e (0)))
08:46	clojurebot	beaky: I don't understand.
08:46	beaky	:(
08:47	opqdonut	(0) is wrong anyway
08:47	beaky	,(try (keys [1 2 3]) (catch Exception e (nil))))
08:47	clojurebot	beaky: Cool story bro.
08:47	BRODUS	Caused by java.lang.IllegalArgumentException
08:47	BRODUS	Can't call nil
08:47	opqdonut	beaky: you have one ) too much
08:48	opqdonut	BRODUS: very interesting
08:48	opqdonut	it indeed doesn't work
08:48	ARM9	contemplate why (nil) is erronous
08:49	opqdonut	BRODUS: ah, I know, keys returns a lazy sequence!
08:50	opqdonut	for some reason (doall (keys [1 2 3])) doesn't help though
08:50	opqdonut	but e.g. (first (keys [1 2 3])) does work
08:50	opqdonut	,(try (first (keys [1 2 3])) (catch Throwable e :error))
08:50	clojurebot	opqdonut: I don't understand.
08:50	BRODUS	i understand lazy sequences mostly, but its not clear why thats the reason its failing
08:50	opqdonut	clojurebot has probably banned try-catch
08:51	luma	because it's a lazy sequence, the error isn't thrown inside the try
08:51	opqdonut	the error is thrown when somebody tries to use the result
08:51	luma	the lazy sequence is returned out of the try, and the error is thrown when that lazy sequence is later realized
08:51	opqdonut	,(do (keys [1 2 3]) 1)
08:51	clojurebot	1
08:51	opqdonut	see, no error
08:52	opqdonut	,(do (first (keys [1 2 3])) 1)
08:52	clojurebot	#error {\n :cause "java.lang.Long cannot be cast to java.util.Map$Entry"\n :via\n [{:type java.lang.ClassCastException\n :message "java.lang.Long cannot be cast to java.util.Map$Entry"\n :at [clojure.lang.APersistentMap$KeySeq first "APersistentMap.java" 166]}]\n :trace\n [[clojure.lang.APersistentMap$KeySeq first "APersistentMap.java" 166]\n [clojure.lang.RT first "RT.java" 660]\n [clojure....
08:52	opqdonut	,(do (doall (keys [1 2 3])) 1) -- I have no idea why this doesn't break though
08:52	clojurebot	1
08:54	MJB47	doall is only used for side effects
08:54	MJB47	it returns nill iirc
08:54	MJB47	while first returns the value
08:54	opqdonut	MJB47: doall should still realise the sequence and give me the error
08:55	opqdonut	,(doall (map #(/ % 0) [1 2 3])) -- e.g.
08:55	clojurebot	#error {\n :cause "Divide by zero"\n :via\n [{:type java.lang.ArithmeticException\n :message "Divide by zero"\n :at [clojure.lang.Numbers divide "Numbers.java" 158]}]\n :trace\n [[clojure.lang.Numbers divide "Numbers.java" 158]\n [clojure.lang.Numbers divide "Numbers.java" 3788]\n [sandbox$eval122$fn__123 invoke "NO_SOURCE_FILE" 0]\n [clojure.core$map$fn__4541 invoke "core.clj" 2625]\n [cl...
08:55	opqdonut	,(do (map #(/ % 0) [1 2 3]) 1) -- for comparison
08:55	clojurebot	1
08:56	luma	opqdonut: it's because of how keys works
08:56	luma	dorun/doall just continuously calls next on the sequence
08:57	luma	and KeySeq (which is what implements keys) only actually produces the key when first is called on it
08:57	opqdonut	but clojure sequences are only lazy in the spine, not the values
08:57	luma	not when next is called on it
08:57	opqdonut	oh, KeySeq is special
08:57	opqdonut	bah
08:57	luma	https://github.com/clojure/clojure/blob/clojure-1.7.0/src/jvm/clojure/lang/APersistentMap.java#L165-L171
08:58	opqdonut	thanks
08:58	yenda	Is it possible to have a clojurescript frontend for a clojure program ? is there something simpler than using ring to make them talk to each other ?
08:59	dnolen	yenda: yes. I can't think of anything much simpler than ring etc.
09:01	BRODUS	thanks for the help everyone
09:04	jonathanj	luma: how do actually address ExceptionInfo?
09:05	jonathanj	clojure.lang.ExceptionInfo it seems
09:07	yenda	dnolen: so ring+compojure is the way to go ?
09:07	j-pb	yenda: bidi is nicer when you do cljs as well
09:08	j-pb	it allows you to share routes between cljs and clj and generate urls from those shared datastructures
09:08	j-pb	it's much more declarative than compojure
09:08	j-pb	you can also use a websocket, if you want more real live communication between your ab and it's ui
09:09	j-pb	but then you are more constrained in terms of browser support
09:09	yenda	j-pb: Browser support is not a problem
09:11	j-pb	setting up a websocket connection is not that hard to do, and it's somewhat nicer than posts
09:11	j-pb	I'd pick something like sente for that
09:12	dnolen	yenda: there is no "way to go"
09:12	dnolen	but yes that will work
09:13	powered	you still need ring for Sente
09:14	j-pb	powered: yeah true, but it's a lot less exposed
09:14	powered	true
09:14	yenda	j-pb: I'm not sure I need websockets though
09:15	yenda	I just need to call functions that alter structures or have side effects and return results
09:15	powered	if you don't want low latency or server -> client messages then don't use websockets
09:19	yenda	powered: sente could be nice to watch components state I could add it later
09:19	yenda	powered: but for now I just want to have a UI to call functions and move stuff in a graph and send it back again to other functions
09:20	powered	right so you need a simple and quick middleware setup?
09:20	powered	sente's good for that too
09:23	yenda	powered: ok, I settled for om frontend-wise so I'm gonna see how it fits with sente
09:24	jonathanj	if the body of a (future) throws, that exception is wrapped in ExecutionException, is it possible to not do that?
09:28	j-pb	jonathanj: you could catch the exception within the future
09:42	QU35t	Fellas
09:42	opqdonut	jonathanj: or you can just unwrap it when forcing the future
09:48	jonathanj	j-pb: i actually want the exception though
09:48	jonathanj	opqdonut: yeah, that's a little bit inconvenient since i'm writing tests for something that returns a future
09:49	jonathanj	on another note, does anyone use cider and midje together? i can't really seem to get the cider test running stuff working with midje
09:58	heero_	Hi, is there a function to reduce over multiple coll ?
09:59	heero_	(reduce coll init list1 list2 ...)
10:00	j-pb	heero_: (reduce fn init (map vector col1 col2 col3 ...)) ?
10:00	powered	(remove even? {:a 1 :b 2 :c 3})
10:01	powered	what is the function I'm looking for instead of remove ?
10:02	j-pb	powered: clojure had something like that as a contrib library
10:02	heero_	j-pb: ok thx
10:04	MJB47	wouldnt filter work?
10:05	MJB47	,(filter #(even? (second %)) {:a 1 :b 2 :c 3})
10:05	powered	this works: (remove #(even? (second %)) {:a 1 :b 2 :c 3})
10:05	clojurebot	([:b 2])
10:05	powered	ah yes
10:06	j-pb	,(remove (comp even? val) {:a 1 :b 2 :c 3})
10:06	clojurebot	([:a 1] [:c 3])
10:09	j-pb	(apply into ((juxt empty (partial remove (comp even? val))) {:a 1 :b 2 :c 3})
10:09	j-pb	,(apply into ((juxt empty (partial remove (comp even? val))) {:a 1 :b 2 :c 3})
10:09	clojurebot	#<RuntimeException java.lang.RuntimeException: EOF while reading>
10:09	j-pb	,(apply into ((juxt empty (partial remove (comp even? val))) {:a 1 :b 2 :c 3}))
10:09	clojurebot	{:a 1, :c 3}
10:09	j-pb	srcr
10:10	j-pb	scnr lol
10:21	kungi	How do I do an &nbsp in hiccup?
10:22	kungi	nevermind I found " " :-)
11:03	visof	hi guys
11:15	sdegutis	I don't know why I signed onto IRC and into #clojure. It was just habit I guess.
11:15	sdegutis	So.. um.... hi?
11:20	Shambles_	sdegutis: Hi.
11:20	sdegutis	Hi.
11:20	sdegutis	I take it justin_smith is asleep then?
11:28	Shambles_	sdegutis: I have no idea.
11:28	sdegutis	:)
11:38	WorldsEndless	So, I thought the doseq family of functions ran only as long as the shortest supplied vector, but was proved wrong by(doseq [n (increment 1) v [\a \b \c]] (println n v))
11:38	j-pb	doseeq is like for
11:39	j-pb	so if you have
11:39	WorldsEndless	So what if I just want to assign a number to each item of another thing (i.e. for file output)
11:39	j-pb	(map vector col (range))
11:39	j-pb	or (map-indexed vector col)
11:39	j-pb	I prefer the first, but others would probably disagree
11:40	j-pb	,(map vector [:a :b :c] (range))
11:40	clojurebot	([:a 0] [:b 1] [:c 2])
11:40	j-pb	like this?
11:40	WorldsEndless	I guess with the latter, I'd then be able to extract the index with something like (doseq [[n v] indexed-vector]) ?
11:40	j-pb	why do you want to do that?
11:40	j-pb	yes, but
11:41	j-pb	you can force lazy evaluation, with dorun and doall :)
11:41	WorldsEndless	j-pb: Ah, map is the one that truncates on teh shortest, not the comprehensions. Ok
11:41	j-pb	yeah
11:41	j-pb	the comprehensions will walk sequences later provided for each element before
11:42	j-pb	,(for [x (range 3) y (range 3) z (range 3)] [x y z])
11:42	clojurebot	([0 0 0] [0 0 1] [0 0 2] [0 1 0] [0 1 1] ...)
11:42	j-pb	,(for [x (range 2) y (range 2) z (range 2)] [x y z])
11:42	clojurebot	([0 0 0] [0 0 1] [0 1 0] [0 1 1] [1 0 0] ...)
11:44	WorldsEndless	So with the comprehensions, I should avoid infinite sequences, or just position them later?
11:45	WorldsEndless	rightwards, that is
11:45	j-pb	with doseq don't do infinite sequences
11:45	j-pb	beause it's eager
11:46	WorldsEndless	I'm guessing that as long as you respect the laziness, (for) should be okay?
11:46	j-pb	with for, you can happily have infinite sequences, but if you have one nothing before it will advance
11:46	j-pb	yeah
11:46	j-pb	,(for [x (range) y (range 2)] [x y])
11:46	clojurebot	([0 0] [0 1] [1 0] [1 1] [2 0] ...)
11:46	j-pb	however
11:46	j-pb	,(for [x (range 2) y (range)] [x y])
11:46	clojurebot	([0 0] [0 1] [0 2] [0 3] [0 4] ...)
11:47	WorldsEndless	Cool. I'm still not as fluent as I would like to be in laziness
11:47	j-pb	you might as well had written
11:47	j-pb	,(for [x [0] y (range)] [x y])
11:47	clojurebot	([0 0] [0 1] [0 2] [0 3] [0 4] ...)
11:47	j-pb	because it will never get passed the first element
12:22	WorldsEndless	regexp aren't behaving as I expect. Why is #"^" not matching "beginning of line"?
12:23	ridcully	,(re-seq #"(?m)^\w" "a b c\nd e f")
12:23	clojurebot	("a" "d")
12:24	WorldsEndless	ooh. ?m is new to me
12:24	WorldsEndless	is there a doc page for that (googling for regexps is hair-whitening)?
12:25	luma	WorldsEndless, http://docs.oracle.com/javase/7/docs/api/java/util/regex/Pattern.html
12:25	justin_smith	WorldsEndless: all the (?*) flags are documented in the javadocs
12:25	justin_smith	or what luma said
12:27	WorldsEndless	beautiful. That link will come in handy, too. Thanks.
12:42	WorldsEndless	basic (probably java) question here: I have a dir-path string (not file). If it doesn't exist, I need to create the file. make-parents doesn't seem to do the trick, since it's a dir. What function do I need here?
12:44	WorldsEndless	Ah. Files.createDirectory
12:45	WorldsEndless	I learn much more of Java from working in Clojure than I did in years of classes in Java...
13:15	yenda	haha same for me
13:16	yenda	and I learned more of Clojure in 3 months than in java in 3 years
13:40	ecmike	Use Clojure to learn Java, use Java to learn Boilerplate
14:15	tolstoy	Learn Clojure to have fun at home, and Java to have someone else decide what and when to do a job-jar task based on a 10-year-old outdated architecture.
14:15	tolstoy	"For 15 years we've known inheritance is a bad way to model your domain." Scribbles: Not a team player. Doesn't get Agile.
14:15	tolstoy	Doesn't understand "the business".
14:15	tolstoy	</rant>
14:16	tolstoy	More positively: I wonder if there's a place for a "lein" like script that uberjars up clojure + that fs library for scripting (if you don't mind slow JVM start up, which I don't).
14:23	justin_smith	tolstoy: it's not the jvm that is slow to start up, it's clojure
14:24	tolstoy	I've got a "clj" script that starts up the repl. Plenty fast for me. I'd love something with some extra file-system utils.
14:25	justin_smith	sure, make a project with the right deps, uberjar, then you can do "java -cp your.jar clojure.main" to get a repl
14:25	justin_smith	you don't even need any main ns or aot for that to work
14:25	tolstoy	Right!
14:25	justin_smith	for bonus points put pomegranate in there too, so that maven deps can be pulled in at runtime
14:26	justin_smith	(if you really need them)
14:26	tolstoy	Exactly. You can have the "mh" of build systems.
14:26	justin_smith	heh
14:26	alive876	hi newbie here, i have a project that "lein run" works "lein test" ok , but when i run "lein repl" i get Var null/null is unbound. the project file and main file are http://pastebin.com/2pWGAP0u and http://pastebin.com/Pn721weh thanks much
14:27	tolstoy	A simple app to build a classpath. Another app to run a repl. Then just compose them. No complicated "boot" thing. ;)
14:28	tolstoy	Really, all you need is something to resolve deps and the rest: bash (or whatever).
14:28	tolstoy	Hm. How do you get the org mode manual inside an emacs buffer? I don't know how to work the help system.
14:28	justin_smith	alive876: I don't know noir but :ns 'jcrit is suspicious - there is no ns called jcrit there
14:29	justin_smith	tolstoy: M-x info
14:29	alive876	ok , ty
14:29	tolstoy	justin_smith Thanks!
14:29	justin_smith	tolstoy: after that you can click the hyperlinks or nav via keyboard
14:30	justin_smith	tolstoy: info is pretty cool - for example incremental search will look into subchapters that aren't visible in the buffer
14:30	justin_smith	and using org, you can create followable links to places in the info system!
14:30	justin_smith	(full circle)
14:31	tolstoy	justin_smith: That's great. I always need to look up in buffer options.
14:31	alive876	(ns jcrit.server
14:33	justin_smith	alive876: right, I was referring to the args to the noir server start. Also, just fyi lib-noir is still maintained, but noir itself is deprecated and they recommend using ring + your routing lib of choice directly
14:34	alive876	ok thanks
14:35	tolstoy	Hm. I always have `:main ^:skip-aot foo.bar`. I wonder if that metadata tag is even needed anymore?
14:35	alive876	but for my edification , this does or doesn't create the jcrit ns?
14:35	alive876	i mean (ns jcrit.server
14:35	justin_smith	alive876: no, it creates the jcrit.server ns
14:36	alive876	ok,thanks
14:36	tolstoy	alive876 Try commenting out the (server/load-views ...) form and see if that fixes it.
14:36	rhg135	tolstoy: AFAIK not till lein 3
14:37	rhg135	For now aot is the default for main
14:37	tolstoy	rhg135: Ah. I've been including that so long, the problem it solved for me has receded into the dim mists of time.
14:37	justin_smith	but if your jar / uberjar isn't meant to be a point click run app, you don't need a :main-ns
14:39	rhg135	Yeah, I guess the maintainers assumed lein 3 would come soon
14:40	rhg135	The bug report is old
14:41	alive876	<justin_smith> that doesn't seem to work. and at the beginning i keep getting this too:Consider using [enlive "1.0.0" :exclusions [org.clojure/clojure]].
14:46	justin_smith	alive876: that's a separate issue - you are using an old version of enlive which asks for a different version of clojure than your other libs
14:46	justin_smith	noir probably asks for some ancient version of clojure like 1.3 or something
14:47	alive876	it seems a lot of broken tutorials out there
14:47	justin_smith	alive876: yeah, clojure moves a little faster than the google magic that would lead you to the up to date stuff
14:49	justin_smith	but that's just a warning, the real issue is that noir isn't being started up properly. As I mentioned that 'jcrit arg is pointing it to a namespace that doesn't exist, but what does noir actually want that namespace to be, or to have in it?
14:50	alive876	yea, i just looked and the tutorial is 4 years old
14:50	justin_smith	maybe you could find a newer tutorial, maybe one using bidi or compojure instead of noir
14:56	princeso	is there a function to map data like this {:a 1 :b 2} into some complex object like {:c :a :d {:e :a :f :b} } to result in {:c 1 :d {:e 1 :f 2}} ?
14:57	ecmike	tolstoy: just saw your rant, spot on
14:57	clojurebot	Huh?
14:58	justin_smith	princeso: what is the relationship between the inputs and outputs there?
14:58	justin_smith	is it just mapping the vals by your {:a 1 :b 2} ?
14:59	princeso	justin_smith yes
14:59	justin_smith	,(require 'clojure.walk)
14:59	clojurebot	nil
15:00	justin_smith	,(clojure.walk/postwalk (fn [o] (if (map? o) (into {} (map (fn [k v] [k ({:a 0 :b 1} v v)]) o)) o)) {:c :a :d {:e :a :f :b} })
15:00	clojurebot	#error {\n :cause "Wrong number of args (1) passed to: sandbox/eval49/fn--50/fn--51"\n :via\n [{:type clojure.lang.ArityException\n :message "Wrong number of args (1) passed to: sandbox/eval49/fn--50/fn--51"\n :at [clojure.lang.AFn throwArity "AFn.java" 429]}]\n :trace\n [[clojure.lang.AFn throwArity "AFn.java" 429]\n [clojure.lang.AFn invoke "AFn.java" 32]\n [clojure.core$map$fn__4541 invok...
15:00	justin_smith	oops
15:00	justin_smith	,(clojure.walk/postwalk (fn [o] (if (map? o) (into {} (map (fn [[k v]] [k ({:a 0 :b 1} v v)]) o)) o)) {:c :a :d {:e :a :f :b} })
15:00	clojurebot	{:c 0, :d {:e 0, :f 1}}
15:01	justin_smith	,(clojure.walk/postwalk (fn [o] (if (map? o) (into {} (map (fn [[k v]] [k ({:a 1 :b 2} v v)]) o)) o)) {:c :a :d {:e :a :f :b} })
15:01	clojurebot	{:c 1, :d {:e 1, :f 2}}
15:01	princeso	omg you rock justin_smith! thanks
15:01	justin_smith	princeso: of course you could name things and make it not a one-liner
15:03	justin_smith	the main tricks there are using postwalk in order to do nested updates, using into {} in order to map over a hash-map and create a new hash-map, and the usage of the hash-map as a lookup function (with the optional not-found value, so that the nested hash-map stil comes through))
15:03	justin_smith	and the key/value destructuring from the map entry tuple
15:04	ecmike	justin_smith: that's really great... do you have any suggestions for learning clojure that well? I find myself doing a lot of thrashing before getting to anything elegant like that
15:04	justin_smith	ecmike: hanging out here and trying to read / recreate /improve the one-liners helps :)
15:05	justin_smith	ecmike: also, there is of course http://conj.io as a reference, and seeing other people's answers on 4clojure
15:05	ecmike	justin_smith: doing that already, wondering if there were any books/sites/etc you recommend
15:05	ecmike	justin_smith: oh great, I'll check that out
15:05	princeso	justin_smith: thanks for that. am almost new to clojure, so ill figure out how it all works.
15:05	ecmike	bookmark*2
15:06	justin_smith	princeso: as I recommended to ecmike above, conj.io helps - as does the community knowledge here in the channel if you try to follow the discussion
16:07	WorldsEndless	I am segmenting a file into many files based on a regexp; it works if I slurp and use string/split, but this will have problems for large files. How can I make this large-file friendly?
16:09	WorldsEndless	Is the best way to iterate through the lineseq and check for re-match, and then split and send the splits to the appropriate place?
16:15	jonathanj	i see people talking about using nrepl in production to embed a repl for investigation etc. purposes
16:15	jonathanj	how do you prevent any random person from connecting to your repl?
16:15	justin_smith	WorldsEndless: that seems reasonable, yeah, though you could try variations like reading N bytes followed by cleaning up the odds and ends by combining any left over from the previous (if any) before running the regex
16:16	justin_smith	jonathanj: by default nrepl only allows connections from localhost (which is a good idea), and then you need to use an ssh tunnel to use it from another host
16:16	jonathanj	ah
16:17	jonathanj	any reason not to embed an nrepl in a production server?
16:17	justin_smith	memory overhead I guess - nrepl is pretty easy to launch from an app
16:18	celwell	jonathanj: that's a terrifying sentence. what's the reason, just curious.
16:18	justin_smith	celwell: well we already discussed the fact that it won't be available without a secure connection
16:27	TimMc	celwell: It is both terrifying and incredibly usful.
16:28	TimMc	jonathanj: We use liverepl instead. The protection is that you have to be the same user who launched the JVM process in order to inject the agent.
16:28	TimMc	I once tried to add nrepl to an app using drawbridge but I couldn't get password protection working.
16:29	justin_smith	TimMc: my take was that realistically every OS has local holes, and the realistic barrier to enforce is logged in / not logged in, and once logged in you can't prevent escalations
16:30	justin_smith	(though I guess they don't always have to be as easy as they are via nrepl)
16:31	TimMc	Yeah, but there's always a chance your kernel is up to date. ;-)
16:31	TimMc	Heck, if you're inside the datacenter at most shops you're probably as good as in.
16:31	justin_smith	TimMc: kernel, plus userland - there are local exploits via games, random file conversion utilities, libraries...
16:50	TimMc	jonathanj: One downside of liverepl is that it doesn't give you all the nice nrepl features, you just get a raw repl with no line-editing.
16:50	TimMc	liverepl can load additional jars, so that might be a path through if that's a problem for you
17:06	jonathanj__	TimMc, justin_smith: Thanks for the discussion, I'll do some reading.
17:13	glosoli	Hey what's the name of the some tool that could scan through the project and point out the stuff that is being done unconventionally ?
17:13	celwell	kibit
17:13	justin_smith	eastwood is good for this too
17:15	glosoli	thanks
18:22	irctc	Hi everyone! :)
18:23	WorldsEndless	Hi, irctc
18:23	irctc	I need help with a concept in functional programming paradigm. I have a function that connects to an API and returns an xml file. Then I need to parse that file, extract info from the parsed result, do some calculations etc, etc. Now I want to make my code functionally as pure as possible.
18:24	irctc	As things stand now I have a sequential stack of calls that goes all the way back to the connection that relys on each function passing through a user ID with wich to connect to the api in the connection function.
18:24	irctc	How do I change this to be more functionally pure?
18:24	irctc	Hi WorldsEndless. :)
18:25	irctc	I am thinking that a more functionally pure way would be to just call this connection function once and then save the returned xml in a global (def) and then pass that def arround to other pure functions.
18:26	irctc	Am I right in approaching this problem that way or is there another way that it should be done?
18:26	WorldsEndless	Interesting. I recently watched Stuart Sierra's talk on functional design patterns that might be of interest to you: https://www.youtube.com/watch?v=etr08mExAI0
18:27	WorldsEndless	irctc: Using React-based web libraries, the global atom idea is pretty standard
18:28	irctc	What does that mean. Sorry I'm new to functional programming.
18:28	WorldsEndless	irctc: Although I recently had to redo one of my codebases to pass the value around rather than look it up globally, which brought a lot of benefits
18:28	tolstoy	irctc I think the idea is to do all your side-effects at the front (or end), so don't bury the web request in the middle of calculations.
18:29	WorldsEndless	So, Facebook's "React" framework is a popular javascript engine that has proven very powerful in the hands of Clojurians. The clojurescript approaches, like Om or Reagent, rely heavily on that idea of a global Atom
18:29	tolstoy	irctc: Just make the request, then pass the result to pure transforms, then save (or whatever) the result.
18:29	WorldsEndless	Agree with tolstoy
18:29	irctc	tolstoy, my web request goes all the way through my entire algorithm. This makes it very hard to do unit testing as well.
18:30	WorldsEndless	irctc: Just make the request once, as you were thinking, and pass the needed parts around to what needs them
18:30	tolstoy	irctc: Actually making the request is buried in there somewhere?
18:30	justin_smith	if you put it in a def, then you are stuck with only being able to handle one xml file, with pure handlers and no global you can process as many as your computer can handle
18:30	irctc	So, the request should be made up front and then saved in a def?
18:30	WorldsEndless	irctc: you may not need to save it in a def at all
18:30	irctc	Ok, so how do I go about that aproach?
18:31	justin_smith	irctc: how many different xml files would you ever handle? if it's exactly one for the entire lifetime of the app a def may be OK, if you would ever need a second xml file, definitely don't use a def
18:31	tolstoy	irctc: Using a "def" is just an implemenation detail. You could: (let [xml (make-request...)] (save-to-db (transform xml))). No global.
18:31	irctc	>=1
18:31	irctc	Nah, I don't use a db here.
18:31	justin_smith	yeah, if 2 would ever happen, a def would just make things more complicated
18:32	irctc	I agree justin_smith. You have a point there.
18:32	tolstoy	irctc: Again, it's just to illustrate that the network retrieval and saving are at the top level, not in the `transform` bit.
18:32	WorldsEndless	So, guys, I'm a bit tangled up on an idea. Spliting one file to many using readers and writers, I open with "with-open [reader ..]" but am not sure how to alternate writing to a current file with writing to a new file when I match my break-regex.
18:33	irctc	So when I do (connection userid) where userid is some number like 123, I get my xml file back. How do I then pass it to pure functions without calling (connection userid) function from other functions?
18:33	justin_smith	The functional concept is that a function given the same args should return the same result. So focus on breaking the task up to stages that are 100% dependent on their arguments.
18:34	justin_smith	irctc: by calling those functions
18:34	tolstoy	irctc: (let [xml (connection userid)] (transform xml))
18:34	justin_smith	WorldsEndless: with loop or reduce you can have a binding that is passed for each item, and when you see that a new output file is needed you pass that along as the new binding
18:35	WorldsEndless	Wait... you can do that with reduce?
18:35	irctc	Ok, the way I have it set up now is I make (top-function-call userid) --calls--> (mid-function-call userid) --calls--> (bottom-function-call userid) --calls--> (connection userid).
18:35	justin_smith	WorldsEndless: something like (reduce (fn [[out-file acc] el] ... [new-out-file new-acc) lines)
18:35	tolstoy	WorldsEndless: Hm. You could use a multimethod? read-line, charaterize it, then pass it to a multimethod, even method of which knows which file to append it to?
18:36	justin_smith	WorldsEndless: the idea being on a given line you know whether to make a new file or not, and that's the file you pass to the processing of the next line
18:36	tolstoy	irctc: Right. I'd reverse that, if you can. Then pass the XML through your other functions as needed.
18:36	irctc	tolstoy, you mean I should do a let [xml (connection userid)] at the top-level-function and then just pass the local xml binding to other calls within the top-level-function?
18:37	tolstoy	irctc: Yes.
18:38	irctc	tolstoy, I think that makes sense. :) Thank you.
18:38	tolstoy	irctc: At least one benefit of that is that when you read each function, the only thing that function can see are the params.
18:38	irctc	Thank you everyone for your help. I'm gona try it this way.
18:38	WorldsEndless	irctc: good luck
18:39	irctc	Well the main benefit for me will be that I'll be able to do unit testing. Right now I would have to rely on an external api call and that's a real mess if you have large xml files.
18:39	irctc	Plus debugging is much easier with pure functions.
18:40	tolstoy	irctc: Yes!
18:40	irctc	I think imperative languages have broken me from the start. lol
18:41	justin_smith	irctc: pure functions are just imperative functions that are only allowed to use data that is in the arg list and can only change things that they return
18:41	justin_smith	you end up returning collections more often, for this reason
18:48	irctc	justin_smith: I feel that because of the lessons I learned while using clojure this made me a better imperative programmer. The notion of pure functions is just not enforced that much from the get go in imperative paradigm. I actually wish I was taught a functional language first before imperative languages.
18:49	justin_smith	irctc: only using arg lists and returning data instead of modifying it are good default behaviors for sure
19:45	domgetter	So I'm trying to take advantage of the JVM doing optimizations, and I'm using dotimes to see timing from multiple runs of a function. Within a dotimes call, I can see it taking less and less time, but a second call to dotimes starts everything over. Is there a way to keep what "work" the JVM did? https://gist.github.com/domgetter/8c0f66b58cbeca135731
19:46	domgetter	As you can see on lines 69, 196, and 1310, the times drop.
19:47	domgetter	When I call dotimes a second time, is it starting over because I'm in a repl?
19:48	domgetter	Or (depending on the optimization of course) can I expect an optimization to stick around in production?
20:02	BRODUS	say i have a complicated expression, and I want to check if the result is either nil? or seq?, how I could I write that while only writing the expression once
20:03	BRODUS	rather than writing it twice in the or expression
20:04	lambda-11235	(let [val ...] (or (nil? val) (seq? val))
20:42	codahale	domgetter: If you’re trying to benchmark something, I’d suggest using Criterium. The JVM makes naive benchmarking effectively impossible.
21:36	jghasjhjkasd	hi can someone tell me why this doesn't work? (map #(1) [1, 2, 3])
21:36	jghasjhjkasd	just trying to turn the list in to a list of 1's
21:36	justin_smith	,'#(1)
21:36	clojurebot	(fn* [] (1))
21:37	justin_smith	that tries to call 1 as a function
21:37	jghasjhjkasd	ohh right ok
21:37	justin_smith	,(map (constantly 1) [1, 2, 3])
21:37	clojurebot	(1 1 1)
21:37	justin_smith	,(map (fn [_] 1) [1, 2, 3])
21:37	clojurebot	(1 1 1)
21:38	justin_smith	my two minor peeves about clojure is that identity is longer than (fn [x] x) and constantly is longer than (fn [& _] x)
21:41	lambda-11235	justin_smith: Indeed, I find myself defining id and const in the REPL when I need them.
21:42	lambda-11235	identity is still two characters shorter than (fn [x] x).
21:44	justin_smith	lambda-11235: oh right, (fn[x]x) isn't nearly as nice
21:47	jghasjhjkasd	hey justin_smith
21:48	justin_smith	yes?
21:48	jghasjhjkasd	is there any way to do that map thing with the shorthand syntax?
21:48	jghasjhjkasd	(map (fn [_] 1) [1, 2, 3])
21:48	jghasjhjkasd	with #
21:48	justin_smith	the closest you can get is #(do % 1)
21:48	justin_smith	because otherwise it will complain about the arg count
21:48	justin_smith	and #(do % 1) is terrible
21:49	jghasjhjkasd	i dont get it.. why do you have to use all the arguments?
21:49	justin_smith	jghasjhjkasd: because if you don't use any args, the compiler generates a function that takes no args
21:49	justin_smith	and you get an error because you get an arg
21:49	BRODUS	is there a function that returns the maps a predicate to a collection and returns the first value that satisfies the predicate?
21:49	jghasjhjkasd	oh ok
21:49	jghasjhjkasd	thx
21:50	BRODUS	*is there a function that maps a predicate to a collection and returns the first value that satisfies the predicate?
21:50	justin_smith	BRODUS: (comp first filter)
21:50	justin_smith	,((comp first filter) even? [1 3 6 42])
21:50	clojurebot	6
21:50	BRODUS	that makes sense
21:54	BRODUS	and because filter returns a lazy sequence that means it will not map over the whole collection
21:57	justin_smith	right
21:58	justin_smith	but I still find it surprising there isn't a function for "first item matching the predicate" built in
22:14	amalloy	that obvious missing function is rich's subtle way of encouraging everyone to run their own fork of clojure with noe added function
23:33	tolstoy	BRODUS "some" kinda does that. (some #(when (pred? %) %) coll).
23:33	BRODUS	right but that just returns a boolean
23:34	tolstoy	The way I wrote it there, it returns the value.
23:34	BRODUS	oh i see
23:34	tolstoy	,(some #(when (even %) %) [1 3 5 6 8 1 2 4])
23:34	clojurebot	#error {\n :cause "Unable to resolve symbol: even in this context"\n :via\n [{:type clojure.lang.Compiler$CompilerException\n :message "java.lang.RuntimeException: Unable to resolve symbol: even in this context, compiling:(NO_SOURCE_PATH:0:0)"\n :at [clojure.lang.Compiler analyze "Compiler.java" 6704]}\n {:type java.lang.RuntimeException\n :message "Unable to resolve symbol: even in this co...
23:34	tolstoy	,(some #(when (even? %) %) [1 3 5 6 8 1 2 4])
23:34	clojurebot	6
23:37	BRODUS	good to know, i need to use 'when' more
23:37	tolstoy	Part of the when, when-not, when-let family.
23:50	jghasjhjkasd	how do you remember when the list/collection is the first or last parameter of clojure functions? just memorization or is there logic behind it?
23:51	jghasjhjkasd	like cons is list last, conj is list first, nth is list first, reduce is list last, etc.
23:52	tolstoy	My guess: if it's operating over a range of values, it's at the end (map, reduce, etc).
23:52	tolstoy	But if you're just changing one thing about the collection, it's first (conj, assoc).
23:54	amalloy	jghasjhjkasd: mostly things that operate on sequences take them last, where things that operate on any kind of collection take it first
23:54	amalloy	also how do you remember your username
23:54	jghasjhjkasd	i just typed in a bunch of gibberish it's not my username i'm using irc web client