#clojure – 2015-01-28

2015-01-28

00:09	Jabberz	there a good way to mock core.async in tests (say midje tests), so you don't have to create a bunch of channels and the like, but verify say something got put onto a mock channel?
00:10	tomjack	ah, I see https://www.refheap.com/1e30c198d528300fcba9ef24a
00:10	tomjack	reification is exponential in depth?
00:11	TEttinger	Jabberz: ##(str "Your mother was a hamster and your father smelt of elderberries, " "core.async")
00:11	lazybot	⇒ "Your mother was a hamster and your father smelt of elderberries, core.async"
00:11	TEttinger	...I have never used midje or core.async
00:15	tomjack	Jabberz: would a mock channel behave differently from a normal channel?
00:16	Jabberz	tomjack: it would be nice to say something like (some-fun) => (>! some-data) in a test, rather than go through the whole ceremony of setting up the plumbing
00:18	Jabberz	afaict, you can't do a (provided (>! ...) in midje, because it's not a function, it's the from-hell-macro, so that's kinda leaving me at setting up the channels unless someone's already solved this type of problem elsewhere and wrapped it in a nice little library? :D
00:20	Jabberz	tomjack: and yeah, I don't want to deal with lots of async stuff running all over the place in the tests, and having to wait, or join or latch on, so that's kind of the motivation -- stub out the core.async, and just verify that the business logic that puts stuff on/takes stuff out of the queues is all correct
00:21	tomjack	you want to check that some-data was put into _any_ channel?
00:22	tomjack	otherwise, you're going to have to pass something (a mock channel?) in? at that point I fail to imagine how the mock channel would behave differently from a normal channel
00:22	Jabberz	in midje terms, just that channel put was called with some args
00:23	tomjack	you can check put! I think then
00:24	Jabberz	I'm basically treating core.async similar to a db -- wanting to mock out the interactions with it, so all the business logic gets unit tested, and the pick up exercising the async and db stuff all together in end-to-end functional tests
00:26	tomjack	ok, just check for clojure.core.async/put! ?
00:26	tomjack	dunno, good luck :)
01:01	catern	http://blog.venanti.us/ultra/
01:02	catern	>About two months ago I decided I'd been writing Clojure long enough that the time had come to optimize my development environment.
01:02	catern	web devs should be banned :(
01:02	catern	from everything
01:02	catern	hipsteerrrrrrs
01:02	catern	(that is all)
01:04	sharkz	Hi, how can I create a list '((10 20) (11 21)) from '(10 11) and '(20 21)
01:05	sharkz	I want to create a one-by-one pair of two equally-sized lists
01:05	slipset	,(list '(10 11) '(20 22))
01:05	clojurebot	((10 11) (20 22))
01:06	slipset	,(map list '(1 2) '(2 3))
01:06	clojurebot	((1 2) (2 3))
01:07	sharkz	that's is cool
01:07	sharkz	*that's cool
01:07	slipset	:)
01:08	sharkz	I didn't know map can iterate multiple lists
01:08	sharkz	thank you so much :)
01:08	slipset	No problem!
01:08	profil	sharkz: map can: "(map f c1 c2 c3 & colls)"
01:09	slipset	~map
01:09	clojurebot	map is LAZY
01:37	Kneiva_	,(mapv list '(10 11) '(20 21))
01:37	clojurebot	[(10 20) (11 21)]
01:37	Kneiva_	sharkz: isn't this what you wanted? --^
01:41	teajoe	Hey guys! So I'm reading about macros @ Brave & True, and I stumble upon the postfix macro: evaluates a function defined in postfix notation. This is pretty cool, but can I do it for any entire application (evaluate the entire app as postfix)? I've heard lisps are very oriented to tweaking themselves, but am still new to this. danka
02:38	clj123	how can I merge all seq within seq. For eg [[1 2 3] [4 5 6] [7 8]] => [1 2 3 4 5 6 7 8]
02:40	TEttinger	clj123: ##(apply into [] [[1 2 3] [4 5 6] [7 8]])
02:40	lazybot	clojure.lang.ArityException: Wrong number of args (4) passed to: core/into
02:40	TEttinger	clj123: ##(into [] [[1 2 3] [4 5 6] [7 8]])
02:40	lazybot	⇒ [[1 2 3] [4 5 6] [7 8]]
02:40	TEttinger	dammit I can't remember
02:40	havenwood	,(vec (flatten [[1 2 3] [4 5 6] [7 8]]))
02:40	clojurebot	[1 2 3 4 5 ...]
02:40	ucb	,(apply concat [[1 2 3] [4 5 6] [7 8]])
02:40	clojurebot	(1 2 3 4 5 ...)
02:40	TEttinger	&(apply concat [[1 2 3] [4 5 6] [7 8]]) is correct, flatten is...
02:40	lazybot	⇒ (1 2 3 4 5 6 7 8)
02:41	TEttinger	~flatten
02:41	clojurebot	flatten is rarely the right answer. Suppose you need to use a list as your "base type", for example. Usually you only want to flatten a single level, and in that case you're better off with concat. Or, better still, use mapcat to produce a sequence that's shaped right to begin with.
02:41	havenwood	interesting
02:41	havenwood	TEttinger: thanks, good to know!
02:41	TEttinger	havenwood, I use flatten all the time, but only when I have total control of the data I am using
02:41	TEttinger	usually only for things I know are small and shallow
02:42	clj123	cool thanks
03:20	muhuk	Hi all.
03:20	SagiCZ1	morning..
03:25	hellofunk	zacts: those books look interesting, i'll check them out
03:33	ianhedoesit	SagiCZ1: it's 00:31!
03:33	SagiCZ1	ianhedoesit: ever heard of time zones?
03:33	ianhedoesit	no!
03:33	SagiCZ1	reminds me of some of my american friends
03:33	ianhedoesit	I live in a world where I reject the notion of timezones.
03:34	hellofunk	ianhedoesit: and yet, you make a big deal out of the time
03:34	SagiCZ1	ianhedoesit: at least you don't deal with jet lags!
03:34	hellofunk	:0
03:34	ianhedoesit	I figure date time messes are too messy in the first place, even if I ignore timezones.
03:35	ianhedoesit	and my data is never more than a day off, so it's not really an issue.
03:35	SagiCZ1	ianhedoesit: i am fascinated by half our or quarter hour timezones.. that must really weird
03:35	SagiCZ1	*hour
03:36	ianhedoesit	SagiCZ1: I've never understood those. I haven't looked into why any place has those sorts of time zone variations though.
03:36	SagiCZ1	ianhedoesit: i think it's places that literally don't have any other interesting thing..
03:37	daniel`	sud afrika
03:37	ianhedoesit	http://www.timeanddate.com/time/time-zones-interesting.html
03:38	SagiCZ1	i hope joda time knows about all these
03:38	ianhedoesit	the first thing it talks about is India and why it has a 30 minute offset from UTC. actually kinda interesting.
03:40	ianhedoesit	neat, only two places in the world have quarter hour offsets!
03:40	ianhedoesit	how informative to the clojure community!
03:42	hyPiRion	SagiCZ1: yup, it does
03:42	SagiCZ1	hyPiRion: cool
03:51	Glenjamin	the UK is the worst: if you forget to handle timezones your software still works 50% of the year
03:52	hyPiRion	If you forget to handle time, it'll still work twice a day. etc
03:53	Glenjamin	do people have this problem in other timezones? I've only ever worked in GMT
03:54	Glenjamin	9/10 callsites convert to UTC - software only broken from march-october.
03:55	hyPiRion	There is DST many places on earth
03:55	Glenjamin	right, but the non-dst zone isn't === UTC ?
03:56	Glenjamin	so if you're storing UTC, you'd notice if you miss one
03:56	hyPiRion	oh, sure thing. Ish
03:56	wink	Glenjamin: we're one off with CET - so glancing at logs it can be mistaken for UTC easily
03:56	wink	hyPiRion: are you goin to FOSDEM?
03:57	clgv	Glenjamin: I like the time zones with offsets of additional 30min or 45min
03:57	wink	I found http://qntm.org/abolish an interesting read
03:58	hyPiRion	wink: nope, unfortunately
03:58	Glenjamin	clgv: i would imagine anyone working in those zones gets good at time quite quickly
03:58	Glenjamin	certainly people working in those zones communicating times to people in other zones
04:14	SagiCZ1	Glenjamin: i dont think people in those timezones communicate with the outside world that much
04:15	ianhedoesit	yeah, Nepal and New Zealand are pretty lonely.
04:44	clgv	SagiCZ1: australia? :P
04:44	SagiCZ1	clgv: they dont have full hour timezones?
04:44	clgv	SagiCZ1: check Darwin
04:44	SagiCZ1	clgv: oh nevermind then.. that's embarassing.. sorry :X
04:45	clgv	SagiCZ1: the other exotic ones are in southern asia
04:46	clgv	I'd never guessed that there are time zones with offset +13/+14
04:46	cfleming	Here in New Zealand we use a full hour offset, but we're still pretty lonely down here.
04:47	Empperi	cfleming: sorry saying this from here, not from official channels but latest Cursive update broke debugging both in Mac and Windows
04:48	SagiCZ1	cfleming: do you kiwis get to visit Australia often? they seem pretty close but I imagine some people may never leave the island..
04:48	Empperi	Idea 14
04:48	cfleming	Empperi: Do you mean 0.1.44? 0.1.43 broke it, 0.1.44 should have fixed it
04:48	Empperi	dunno, just the update Idea automatically told me to update :)
04:48	cfleming	SagiCZ1: Sure, Kiwis travel a lot
04:49	Empperi	but good to know
04:49	SagiCZ1	cfleming: cool//
04:49	Empperi	will check that I have the latest version
04:49	Empperi	missing my debugger... :P
04:49	cfleming	Empperi: If you have 0.1.44 and it's broken let me know. It should be much better in 44 than 43 - more reliable breakpoints and expression eval now finally works, so breakpoints can have conditions etc
04:50	cfleming	Empperi: I haven't forgotten about your action issue either BTW, just busy, sorry
04:50	SagiCZ1	cfleming: btw, i am struggling to understand how REPL history works.. for example when i work in REPL, then close it and open a new one, i dont have the latest history from the just closed REPL but some arbitrary history from the past.. i also tried to check the "Show history" window, and that was somewhat confusing as well
04:50	Empperi	cfleming: actually, I might have an idea what is causing it and it is not Cursive related
04:50	Empperi	it's about how Idea reads the keybindings
04:51	Empperi	for example, I have "load namespace into REPL" in keybinding ctrl+alt+l
04:51	Empperi	if I happen to press alt down before ctrl, or if I press them at the same time then it doesn't work
04:51	cfleming	SagiCZ1: That sounds weird, it should have a relatively straightforward linear history
04:51	Empperi	if I press ctrl down first, then alt and then l, then it works
04:51	SagiCZ1	cfleming: even when working across multiple REPLs? how should that work?
04:51	Empperi	annoying but should point to the fact, that the problem is not in Cursive
04:52	Empperi	and this problem is not there with Mac
04:53	cfleming	Empperi: That sounds really weird - you're on Windows, right? Does this happen on a Mac too?
04:53	cfleming	Empperi: Oh, no, ok
04:53	Empperi	I use both and this happens only on Windows
04:53	TEttinger	Empperi: try also binding alt-ctrl-l
04:53	TEttinger	if it distinguishes between ctrl-alt and alt-ctrl, maybe it's configurable
04:53	Empperi	will do when I'm on pc again, right now doing work at home with my work laptop (mbp)
04:54	TEttinger	still the strangest IDE quirk I've heard of in a while
04:54	Empperi	yeah
04:54	Empperi	anyway I can live with that now that I know what is causing the problem
04:55	Empperi	I was getting sick and tired of going to menu just to load the file into repl...
04:56	cfleming	Empperi: Yeah, I bet. That sounds like a JDK issue to me.
04:56	Empperi	cfleming: that doesn't happen on any other software, just with Idea
04:57	cfleming	SagiCZ1: The REPL history is partitioned by type, so local REPLs have a separate history from remote ones.
04:57	Empperi	and my keybindings for Cursive happened to bring it up
04:57	cfleming	SagiCZ1: But within a particular type, it should do what makes sense - I'd have to check the code to remind myself exactly when it syncs what to the history
04:58	cfleming	SagiCZ1: But if you do "Show REPL history" it should show your recent history from the current REPL, definitely
04:59	cfleming	SagiCZ1: BTW the closest point in Australia is still 3.5 hours on a plane from here, it's like Paris to Tel Aviv maybe - still a long way
04:59	engblom	I just solved Martin Gardners coconut problem. I wonder if this is how everybody would solve it or if there is a better clojure solution. The problem and my solution is here: http://pastebin.com/0axr8rZW
05:22	TMA	engblom: it is an excercise for chinese remainder theorem and modular arithmetics -- you are not supposed to brute force the solution
05:25	clgv	engblom: as TMA says ;)
05:26	clgv	engblom: when you don't know about CRT you could at least restrict the search space by reasoning about divisibility of the numbers ;)
05:27	tomjack	(I'm 80% certain that) my problem was a perf bug in core.logic! :)
05:36	tomjack	I believe it had been there since 2012 and would, I think, have caused a significant penalty upon getting answers out whenever the answers were relatively deep (perhaps less deep if also wide?) trees
05:39	tomjack	which is somewhat troublesome, because even in Agda I believe the state of the art for programs whose types verify their complexity is "semi-formal" :(
05:48	Glenjamin	anyone know why the "j" is clojure is italicised?
05:48	engblom	TMA: This was neither a school task or any excersise I would be forced to solve. I just coded this one so I can show the benefit of knowing a bit of programming as you can by programming solve (brute force) problems you would not be able without plenty of skills otherwise.
05:49	engblom	What I wanted to know if the coding style for brute forcing this problem is good, not how to solve it mathematically.
05:50	engblom	I have the mathematical solution. This brute force solution I created is from the point that you should not need more maths knowledge than what elementary school gives.
06:00	TMA	Glenjamin: my take on that: 'tis some kind of reference to Java or JVM
06:02	lachenmayer	Glenjamin: to emphasise that clojure is not closure or clozure? ;)
06:03	TMA	engblom: as for the coding style [disclaimer: I come from common-lisp/scheme background] ... I would personally split the printing and computation into separate functions
06:06	Glenjamin	always good advice imo ^
06:07	engblom	But you would use list comprehension for brute forcing?
06:27	clgv	engblom: but the problem can be easily scaled such that the brute force attempt fails ;)
06:33	TMA	engblom: yes. having the data fed to the bruteforcer as a lazy sequence [or a generator function in vanilla CL] seems like a good concern separation
06:33	TMA	engblom: also: no space before ) :)
06:40	TMA	engblom: (defn compute-solution [candidates] (for [i candidates ...] ...))
06:42	engblom	TMA: So you generate a list of canditates and then feed it to compute-solution?
06:43	TMA	engblom: not a list -- a lazy seq
06:44	engblom	(range) ?
06:44	TMA	engblom: in Common Lisp... (defun compute-solution (candidate-fn) (do ((i (candidate-fn) (candidate-fn)) ...) ...))
06:45	TMA	engblom: range produces a lazy seq, not list
06:45	TMA	engblom: in other words: yes
07:00	H4ns	justin_smith: as sad as it may be, i've decided to use imagemagick because me.raynes.conch is awesome and the java image libraries are not.
07:04	TEttinger	imagemagick is awesome though
07:04	H4ns	TEttinger: much awesomer than the java image stuff
07:06	TEttinger	here, try saving a file as pbm, with options -compress none -depth 8. remove the first two lines of the pbm's text output. stick a [ and a ] on the ends and now you have a valid clojure vector that you can parse with read-string
07:06	TEttinger	if there are less than 11 colors, at least
07:06	TEttinger	which for PBM should be always
07:08	arossouw	are there clojure forums, where i can ask skilled clojure coders to criticize my code?
07:08	TEttinger	arossouw: link to refheap or gist with your code and people will try here
07:08	arossouw	i see, thanks
07:08	TEttinger	there probably are forums
07:09	Glenjamin	there's codereview on stackexchange
07:09	arossouw	cool, thanks
07:09	TEttinger	(inc Glenjamin)
07:09	lazybot	⇒ 11
07:10	arossouw	i've tried something like this, not sure how to access outer block (#(take %1 %2) 3) '(1 2 3 4))
07:11	justin_smith	arossouw: there is a misplaced paren in there somewhere
07:11	justin_smith	what is that trying to do?
07:12	justin_smith	&(#(take %1 %2) 3 '(1 2 3 4))
07:12	lazybot	⇒ (1 2 3)
07:12	Kneiva	,((partial take 3) '(1 2 3 4))
07:12	clojurebot	(1 2 3)
07:12	arossouw	lets say, x is argument 1 for amount of elements to drop y is argument 2, should do ((take x y) 3) '(1 2 3 4))
07:12	justin_smith	arossouw: again, you have an extra paren
07:12	arossouw	i see
07:12	justin_smith	that doesn't even make sense
07:13	arossouw	ok, i'll play around abit
07:13	TEttinger	(doc split)
07:13	clojurebot	Huh?
07:13	TEttinger	(doc split-at)
07:13	clojurebot	"([n coll]); Returns a vector of [(take n coll) (drop n coll)]"
07:13	arossouw	(= (__ 3) '(1 1 2))
07:13	arossouw	4clojure, dont want answer just hint
07:13	arossouw	where __ is function
07:14	TEttinger	fibs
07:14	justin_smith	oh yeah, that's definitely fibs
07:15	arossouw	so i should create an anonymous function that generates fibs and only print first 3?
07:16	justin_smith	not print, return
07:16	arossouw	ah, ok, thanks
07:16	justin_smith	arossouw: you need a function that takes a number n, and returns the first n fibs
07:19	zacts	helgikrs: http://mitpress.mit.edu/books/cognition-basic-musical-structures
07:25	TEttinger	from clojuredocs: ,((fn [n] (take n ((fn fib [a b] (cons a (lazy-seq (fib b (+ b a))))) 1 1))) 3)
07:25	TEttinger	,((fn [n] (take n ((fn fib [a b] (cons a (lazy-seq (fib b (+ b a))))) 1 1))) 3)
07:25	clojurebot	(1 1 2)
07:25	arossouw	cool
07:25	arossouw	i'll have to study lazy-seq, still new at that
07:25	TEttinger	see for yourself how to do it http://clojuredocs.org/clojure.core/lazy-seq
07:25	TEttinger	it's very powerful
07:26	TEttinger	,(defn fibonacci [n] (take n ((fn fib [a b] (cons a (lazy-seq (fib b (+ b a))))) 1 1)))
07:26	clojurebot	#'sandbox/fibonacci
07:26	TEttinger	,(fibonacci 30)
07:26	clojurebot	(1 1 2 3 5 ...)
07:26	Glenjamin	i completely failed to write a function using lazy-seq in an interview this week :(
07:26	TEttinger	,(clojure.string/join " " (fibonacci 30))
07:27	clojurebot	"1 1 2 3 5 8 13 21 34 55 89 144 233 377 610 987 1597 2584 4181 6765 10946 17711 28657 46368 75025 121393 196418 317811 514229 832040"
07:27	TEttinger	Glenjamin, yeah I would have too
07:27	arossouw	it would crash if you did it recursively right? (with python)
07:27	Glenjamin	i managed to write the loop..recur
07:27	zacts	hellofunk: http://mitpress.mit.edu/books/ant-colony-optimization <- cool my next clojure project is going to utilize ideas from this
07:28	Glenjamin	but then didn't look at the docs enough to write the lazy-seq, and got a bit tied up :(
07:28	TEttinger	arossouw, lazyseqs are often infinite. you only evaluate a finite portion by using take
07:28	zacts	helgikrs: sorry man, I meant that for hellofunk
07:28	arossouw	awesome
07:28	justin_smith	arossouw: yeah, if it were evaluated eagerly it would go into a loop until it ran out of stack space
07:28	arossouw	i think data scientists would love that
07:28	justin_smith	arossouw: what lazy-seq does is it creates something that acts like a list, but to get the next item you need to call a function
07:29	justin_smith	arossouw: and python does have something like that, it just has a more confusing way of using it
07:29	arossouw	so it releases memory on each iteration?
07:29	TEttinger	it keeps it around.
07:30	TEttinger	however java will garbage collect unused stuff
07:30	justin_smith	it actually uses more memory each iteration (but it can be released if you don't hold onto the beginning of the series)
07:30	arossouw	interesting
07:31	arossouw	it would be cool if you could download a pdf of clojuredocs.org
07:31	justin_smith	like if you have (def r (range)) and then (nth r 1000000000) you'll use a huge amount of memory, but if you just have (nth (range 1000000000)) that won't
07:31	justin_smith	err (nth (range) 1000000000) that is
07:32	arossouw	ok, nth vs get ,which to choose?
07:32	justin_smith	get is for associative things
07:32	justin_smith	nth is for things that are ordered
07:32	justin_smith	vectors are both
07:32	arossouw	ah
07:32	justin_smith	but otherwise the choice is clear, because only one of the two makes sense
07:33	justin_smith	if a coll supports both get and nth, they are likely identical
07:34	justin_smith	(for that coll)
07:34	justin_smith	also ##(nth [1 2] 22 :not-there)
07:34	lazybot	⇒ :not-there
07:34	justin_smith	nth takes an optional "not found" arg, like get does
07:35	arossouw	oh
07:38	arossouw	i think this might be a bit inefficient, https://gist.github.com/arossouw/2adb82ad208295be31a2
07:40	justin_smith	arossouw: one small thing - (complement nil?) is the same as some?
07:40	arossouw	possibly, i'll try
07:40	justin_smith	they are identical
07:40	arossouw	oh
07:40	arossouw	ok
07:40	arossouw	thanks
07:40	justin_smith	I mean some? is the name of the function, the question mark is part of the name :)
07:41	arossouw	yeah, get it, seen that function
07:41	justin_smith	and (keep f coll) is the same as (filter some? (map f coll))
07:41	TEttinger	(remove nil? collection)
07:41	justin_smith	so you can just return (keep #(last (re-find re-imap-login %)) line)
07:41	justin_smith	and skip two let bindings
07:42	arossouw	nice
07:42	justin_smith	that makes things more clear, but doesn't change the efficiency
07:42	TEttinger	err, one let binding justin_smith
07:43	justin_smith	TEttinger: it eliminates two, because the second is just the return value now :)
07:43	TEttinger	the line with filter is outside the bindings
07:43	justin_smith	arossouw: also, (apply str (map (fn [x] str x "\n")) ...) could just be (clojure.string/join "\n" ...)
07:44	arossouw	awesome
07:44	justin_smith	TEttinger: yes, and I am replacing it with keep
07:44	justin_smith	TEttinger: oh, now I get it, never mind :)
07:44	justin_smith	you are right
07:44	TEttinger	heehee
07:44	justin_smith	I was confused momentarily by the indentation
07:44	TEttinger	(inc justin_smith)
07:44	lazybot	⇒ 173
07:44	TEttinger	(inc indentation)
07:44	lazybot	⇒ 1
07:45	justin_smith	arossouw: in that case, the string/join will be faster than apply / str / map combo
07:46	justin_smith	since you don't need to create the temporary lazy seq before making the string
07:49	arossouw	justin_smith: how would i apply join "\n" to line-seq (io/reader file)?
07:49	justin_smith	,(clojure.string/join "," [1 2 3])
07:49	clojurebot	"1,2,3"
07:49	arossouw	ah
07:49	justin_smith	just make the line-seq the last arg to the join
07:52	slipset	,(apply str (interpose \, [1 2 3]))
07:52	clojurebot	"1,2,3"
07:52	justin_smith	slipset: that has the correct output, but string/join is always better
07:53	slipset	most probably true, but I'm somewhat fond of interpose :)
07:54	TEttinger	interpose and interleave are both quite useful
07:55	justin_smith	(defn interject [e coll] (let [position (rand-int (count coll))] (concat (take position coll) [e] (drop position coll)))
07:55	slipset	besides, this is kind of cool
07:55	slipset	(apply str (interpose \, "123"))
07:56	slipset	,(apply str (interpose \, "123"))
07:56	clojurebot	"1,2,3"
07:56	justin_smith	,(clojure.string/join \, "123")
07:56	clojurebot	"1,2,3"
07:56	arossouw	https://gist.github.com/arossouw/2adb82ad208295be31a2 , looks slightly better now, thanks for the help
07:56	slipset	(inc justin_smith)
07:56	lazybot	⇒ 174
07:56	slipset	didn't know about that one :)
07:57	justin_smith	arossouw: don't forget you need (require '[clojure.string :refer [join]]) for that to work
07:57	justin_smith	or my preference, (require '[clojure.string :as string]) and change join to string/join
07:57	arossouw	ok
07:57	justin_smith	also, that map needs to be keep
07:58	justin_smith	otherwise you still get nils
07:58	arossouw	oh, cool, thought you ment like keep, like keep that part
07:59	arossouw	see now, great stuff
07:59	justin_smith	,(keep next [[1] [12 13] [42] [5 7 9 11]])
07:59	clojurebot	((13) (7 9 11))
08:00	arossouw	so much to learn
08:00	justin_smith	if I had named keep, I likely would have called it winnow
08:00	arossouw	refactored now :-) https://gist.github.com/arossouw/2adb82ad208295be31a2
08:00	justin_smith	(getting rid of the husks :))
08:02	arossouw	anyone here using http-kit with nginx in production?
08:02	justin_smith	yes
08:02	justin_smith	not via the plugin though
08:02	arossouw	do you have to lein uberjar everytime you submit your code?
08:02	justin_smith	yes
08:02	arossouw	justin_smith: do you use midje to test the stuff or deftest?
08:03	justin_smith	clojure.test (deftest)
08:03	arossouw	ok
08:03	arossouw	hows the performance sofar?
08:03	justin_smith	it's been excellent
08:03	arossouw	of nginx + httpkit
08:03	justin_smith	haven't had to scale beyond one server yet
08:03	arossouw	interesting
08:04	arossouw	planning on rewriting python web app to http-kit + compojure + nginx
08:04	justin_smith	cool
08:04	arossouw	don't like orm's for some reason
08:04	justin_smith	that's a smart thing to dislike
08:04	arossouw	lol
08:05	arossouw	is clojure gaining market share?
08:06	justin_smith	I think so, yes
08:06	justin_smith	though being a lisp, I don't know how big a market share it can reach
08:07	arossouw	does lisp have a bad reputation?
08:07	justin_smith	it doesn't look like math the way you learn it in highschool
08:07	tbaldridge	arossouw: no, it just requires you to think a bit more before you start to code
08:07	justin_smith	people are reluctant to learn the syntax
08:08	tbaldridge	arossouw: with great power comes great responsibility
08:08	arossouw	i see, its unfamiliour to most and people usually prefer to stick to known stuff
08:08	arossouw	i'm afraid i can't like any other programming language now
08:09	justin_smith	yeah, it's a simple language barrier issue, and most people are not motivated enough to get past the small amount of learning needed for prefix notation (especially with mathematical stuff)
08:09	justin_smith	hehe
08:10	justin_smith	after you gain some competency, try something in the ml family too - syntactically very different, and strong typing is a different reality, but they share many of clojure's positive qualities
08:10	arossouw	do you think it might persaude people if you give a clear overview of the benefits of functional coding?
08:10	justin_smith	(OCaml, sml, haskell maybe)
08:10	justin_smith	arossouw: perhaps, I think the main thing that will convince people is if we show them great things we can make (and how easy it is to make more like that)
08:10	arossouw	tried haskell, will try again, tricky to learn
08:11	justin_smith	yeah, haskell is a challenging language
08:11	justin_smith	but it's worth knowing at least one ml and at least one lisp, imho
08:11	arossouw	ok
08:12	justin_smith	but yeah, I think one thing at a time, and the best way to convince anything is cool is to make something awesome
08:12	arossouw	text parsing is slow in haskell, but i guess i'd need more experience
08:12	justin_smith	haskell has multiple string types, and the default string type is very inefficient
08:13	arossouw	ok
08:13	tbaldridge	and if you want to learn a good ml (not a bad idea) stick with F#. It's like Clojure, but based on ML (instead of Lisp)
08:13	muhuk	if you're interested in haskell but don't want to leave JVM world; try frege
08:13	justin_smith	tbaldridge: yeah, F# is a good one (it's heavily derived from OCaml, my favorite ml)
08:14	tbaldridge	I started playing with ML awhile back, but was put off by how bad the library support was. That's mostly fixed by F#
08:14	luxbock	tbaldridge: when's your next video coming out? I haven't seen any new ones in a while
08:15	arossouw	muhuk: i'll check it out, thanks
08:15	tbaldridge	luxbock: today or tomorrow, I've hit a bit of a writer's block, going to get back into it
08:15	luxbock	tbaldridge: cool. I really like your transducer ones
08:29	dysfun	what's the clojurescript idiom for when you need to initialise a component through a browser API that takes a function?
08:29	dysfun	e.g. getUserMedia(function() {})
08:30	Glenjamin	pass it a fn
08:30	dysfun	currently i've got it putting onto a core.async channel
08:30	Glenjamin	oh right, i see what you mean
08:30	dysfun	i'm trying to get it out of callback mess :)
08:31	dysfun	i'm using it with om
08:36	Glenjamin	you could probably wrap up the idiom into a channel that's immediately dereffed?
08:38	dysfun	yeah, it's looking like that's the way
08:39	dysfun	doesn't seem the cleanest, but it has the benefit of being straightforward
08:39	clgv	"wrap up the idiom"?
08:41	dysfun	yeah, once i've got it working, i'll see how to tidy it up
08:53	arossouw	is it possible to create a popup calandar with clojurescript?
08:53	R0B_ROD	arossouw: Its possible to do many things with clojurescript
08:53	arossouw	oh, ok
08:54	R0B_ROD	arossouw: If you layout your problem definition and design a simple algorithm Im sure you will be able to code what you want.
08:54	arossouw	i see
08:55	R0B_ROD	arossouw: I say it as theory since I have no experience with coljurescript or clojure.
08:55	R0B_ROD	+ I mistype all the time
08:56	justin_smith	arossouw: clojurescript has full access to javascript
08:56	clgv	arossouw: you can narrow it down to "is X possible with JavaScript & HTML?"
08:57	justin_smith	arossouw: so you can use whatever javascript libraries are apropriate (and there are calender libs for javascript)
08:57	michaelr525	hi
09:00	arossouw	ok, thanks
09:02	michaelr525	hmm
09:02	Glenjamin	clgv: i should really have said "define a function"
09:02	michaelr525	I need to do this in Clojure: new GenericType<ChunkedInput<String>>() {}
09:02	michaelr525	any idea?
09:03	michaelr525	it's from here: https://jersey.java.net/documentation/latest/async.html#d0e10298
09:03	justin_smith	michaelr525: generics don't exist in the jvm
09:03	justin_smith	they only exist in the java compiler, which we are not using
09:03	justin_smith	ignore the generic part, and just make a ChunkedInput
09:03	michaelr525	justin_smith: I'm not sure it's going to work in this case..
09:04	hyPiRion	justin_smith: GenericType, rather.
09:04	justin_smith	hyPiRion: oh, right, thanks :)
09:04	michaelr525	justin_smith: that code expects to have that type information in runtime..
09:04	hyPiRion	but this is some anonymous class magic instantiation
09:04	justin_smith	michaelr525: generics are not enforced by the vm, in fact they don't even exist
09:04	justin_smith	hyPiRion: so that calls for reify, right?
09:05	hyPiRion	justin_smith: right
09:05	michaelr525	err..
09:05	michaelr525	from the above url: "The entity is read as a ChunkedInput entity. In order to do that the GenericEntity<T> is used to preserve a generic information at run time. If you would not use GenericEntity<T>, Java language generic type erasure would cause that the generic information would get lost at compile time and an exception would be thrown at run time complaining about the missing chunk type definition."
09:06	michaelr525	that's why I don't think it would work ;)
09:08	justin_smith	michaelr525: what that means is that an instance of GenericType is provided as an argument to redEntity
09:08	justin_smith	as hyPiRion said, use reify to create that instance as the first arg to readEntity
09:08	justin_smith	GenericType is what is important here, the generics are just noise
09:10	justin_smith	reify (and sometimes proxy) are how we create anonymous class instances like in that code snippet
09:10	zacts	hi clojurists
09:11	R0B_ROD	Hello justin_smith
09:11	justin_smith	hello
09:11	michaelr525	justin_smith: ok, thanks.. i'll give it a try before wraping that in a java class..
09:12	michaelr525	i'd have to override some of GenerType's methods to return the parametrized type name at runtime
09:12	justin_smith	michaelr525: yes, reify is capable of doing that
09:13	zacts	I feel like my designs in clojure are lacking
09:13	zacts	I want to gain more experience
09:13	zacts	with functional programming
09:13	justin_smith	zacts: you should make something
09:14	zacts	what to make? :-)
09:14	TMA	zacts: almost anything, really
09:14	zacts	I feel like I know how the pieces move in chess, but don't know how to play the game
09:14	zacts	well
09:15	slipset	zacts: if you have a working webapp, implement either the front-end or the back-end in clojure(script)
09:15	slipset	If you work in a company which uses irc/slack/hipchat or something, implement a simple bot which looks up jira issues for you
09:16	justin_smith	or, in fact, you could even make a chess engine (save the AI part for later, just do something that can accept valid moves, and reject invalid ones)
09:16	zacts	hum.. nah I don't want to do chess right now
09:16	zacts	it was a metaphor for my clojure understanding
09:16	zacts	I guess I may just start with the music ideas
09:16	TMA	zacts: I have found that at least for me the main point was to change the thinking. the designs then came spontaneously
09:17	zacts	I like this book on ants too though
09:17	zacts	http://mitpress.mit.edu/books/ant-colony-optimization
09:17	slipset	IMHO, when learning a new language/framework it's nice to be able to only focus on the language/framework and not worry about how to implement something.
09:17	zacts	maybe I could make an ant simulator
09:17	justin_smith	zacts: there are a few ways to drive music generation from clojure, you could pick one and maybe make a counterpoint engine, or a drum machine, or whatever style of music generation you are looking for
09:17	augustl	is there something like datomic's squuid available as a stand alone lib?
09:18	justin_smith	zacts: one of Rich Hickeys fist big demos of clojure concurrency was an ant simulator
09:18	zacts	oh really? coolio
09:18	dnolen	ClojureScript 0.0-2740 released, bunch of Windows fixes
09:18	justin_smith	augustl: iirc there is an under-documented method in java.util.UUID for generating SQUUIDs
09:19	slipset	https://gist.github.com/spacemanaki/1093917
09:19	slipset	ant-demo
09:19	tbaldridge	zacts: video of the ant-demo is in the last 30 min of this video: https://www.youtube.com/watch?v=dGVqrGmwOAw
09:19	zacts	oh nice! I'll check it out
09:20	justin_smith	zacts: make your own ants demo and then look at all the ways it is inferior to the one Rich Hickey mad :)
09:20	augustl	justin_smith: ah, will see what I can find :)
09:21	justin_smith	augustl: I think the key is how one constructs the mostSigBits and leastSigBits to the constructor
09:21	zacts	lol justin_smith
09:22	Glenjamin	that is a looong video
09:24	augustl	`new UUID(new Date().getTime(), new SecureRandom().nextLong())` seems to work
09:24	justin_smith	cool, didn't realize it would be that simple, but makes sense I guess
09:25	augustl	it's different from the one in datomic though
09:25	justin_smith	and that returns the right .timestamp value?
09:25	augustl	I get 0000014b-30eb-c56a-cba6-ff9a806cbef0, 0000014b-30eb-c56b-eacf-04128f2c0e06, etc
09:25	augustl	there are fewer zeroes in the one from datomic, fwiw
09:26	augustl	s/zeroes/leading zeroes/
09:27	clgv	zacts: I am implementing ACO algorithms in Clojure ;)
09:27	justin_smith	,(.timestamp (UUID. (.getTime (java.util.Date.)) (.nextLong (java.security.SecureRandom.)))
09:27	clojurebot	#<RuntimeException java.lang.RuntimeException: EOF while reading>
09:29	justin_smith	augustl: another thing is that it doesn't actually create a time-based UUID (I don't know if that matters to you or not)
09:29	justin_smith	,(.timestamp (java.util.UUID. (.getTime (java.util.Date.)) (.nextLong (java.security.SecureRandom.))))
09:29	clojurebot	#<UnsupportedOperationException java.lang.UnsupportedOperationException: Not a time-based UUID>
09:32	zacts	clgv: oh nice
09:32	zacts	clgv: perhaps I'll hang out with you here then
09:33	zacts	so hellofunk for music, clgv for ACO ants project =)
09:33	zacts	#clojure is really a cool channel so far with really smart people
09:41	samiswellcool	zacts: agreed
09:46	michaelr525	reify only supports interfaces hmm
09:46	michaelr525	maybe proxy instead..
09:49	donbonifacio	hum.. I have constant values that I use all over my app, each one is a map. Tried to use a record instead, and the tests take almost twice as much. Was expecting the opposite
09:54	justin_smith	donbonifacio: are you associating new keys that aren't part of the record, or dissociating keys that are part of the record?
09:54	donbonifacio	nop, they are totally constant, and only read data
09:55	donbonifacio	I store them all (about 30 objects)
09:55	donbonifacio	and then just read info from them
09:57	michaelr525	justin_smith: reify doesn't work with classes, only with interfaces and I can't seem to override the constructor with (proxy).. maybe I should try (gen-class) ?
09:58	justin_smith	http://cemerick.com/2011/07/05/flowchart-for-choosing-the-right-clojure-type-definition-form/
09:59	justin_smith	so you need a named type, that can be referred to statically from java that extends an existing base class?
09:59	justin_smith	according to cemerick's page that means you need gen-class, which is too bad
10:00	michaelr525	it doesn't have to be a named class..
10:00	michaelr525	I just need an instance of that classs
10:00	justin_smith	in that case proxy should work, if you can figure out how to override the default constructor?
10:01	michaelr525	i'll dig deeper ;)
10:04	zarkone	Hello all. i use figwheel started in repl for clojurescript. I add :source-map-true and :source-map-timestamp compiler options. CIDER can point the source location, but only the old one -- seems like :source-map-timestamp doesn't work...
10:05	zarkone	both for Chrome and FF
10:13	michaelr525	no
10:13	michaelr525	doesn't seem like proxy can do it..
10:13	michaelr525	damn
10:22	justin_smith	,(.newInstance (class (proxy [Object] [] (toString [] "silly"))))
10:22	clojurebot	#<Object$ff19274a sandbox.proxy$java.lang.Object$ff19274a@4bb1e2ee>
10:22	justin_smith	given a proxy, you can get its class and invoke the zero argument constructor
10:23	stuartsierra	augustl: Time-based "Version 1" UUIDs don't put the most significant time bits first.
10:24	stuartsierra	Datomic squuids follow the format of random "Version 4" UUIDs but with the leading bits as time.
10:28	augustl	stuartsierra: ah, good to know, will look into that
10:29	clgv	zacts: hehe
10:44	michaelr525	justin_smith: well, I actually wanted to prevent the default contructor from being called. I thought that if I'd override it then it might work...
10:46	justin_smith	michaelr525: how did you go about implementing the constructor?
10:46	zacts	hum.. clgv what kinds of applications are you working on with ACO?
10:47	zacts	(if I may ask)
10:53	dysfun	hrm, how do i use libraries that don't use cljx from clojurescript if crossovers are deprecated?
10:53	dysfun	in this case, i want to use flatland/useful
10:55	dysfun	i know that the module i require doesn't do anything that clojurescript doesn't support
11:09	clgv	zacts: the classical ones - combinatorial optimization
11:09	zacts	oh neato
11:49	denik	Using, test.check, is it possible to generate hash-maps with matching, unique properties (e.g. ids)?
11:50	denik	Two maps should share the same val at k
11:50	cemerick	denik: yes; see `bind` in its API
11:50	denik	ok
11:50	reiddraper	denik: use gen/fmap to add that key after generating the two values
11:51	cemerick	or that :-P
11:51	denik	what do you mean by two values
11:51	denik	it would be a shared value
11:51	denik	a pool of unique ids, for example
11:51	Glenjamin	(gen/fmap #(assoc % :k2 (:k1 %)) your-generator)
11:52	Glenjamin	oh, i thought you wanted one hash-map with two values the same
11:52	Glenjamin	well, the general idea is the same anyway
11:52	denik	okay, I'll try to figure it out
11:52	dweave	what does * mean around symbol names ie: my-symbol
11:52	denik	thanks everyone :)
11:53	Glenjamin	reiddraper: while you're here - if i want a number > or < some constraint, am i better off using choose, or gen/fmap with int?
11:53	reiddraper	denik: a good rule of thumb is that if something is statically generating (ex: apply unique, range-based ids), use fmap to deterministically add those values, instead of generating them
11:53	reiddraper	Glenjamin: probably choose
11:54	denik	reiddraper: copy that
11:54	Glenjamin	reiddraper: thanks - it's been super easy to far to expose stuff to JS
11:54	nic77	so guys im havin some trouble with graph theory i think
11:54	reiddraper	Glenjamin: happy to hear that :)
11:55	Glenjamin	i'm interested to try and put individual generators into npm modules somehow - like test.chuck but more granular
11:55	nic77	the wikipedia doesnt really explain the explanations lol i think im missing some building blocks
11:56	Glenjamin	although download size isn't really a constraint, being able to mix&match and move versions easily is pretty useful
11:58	nic77	these vertex and nodes and edges, whats with all of that
11:58	nic77	im not imagining something correctly
11:59	nic77	i thought it was a graphics thing
12:00	nic77	ill piece it together
12:00	nic77	this sucks though, its not clicking like things usually do
12:03	SagiCZ1	oh.. i was like.. this channel fell awfuly quiet today.. and then i noticed my scrollbar is stuck
12:03	nic77	http://en.wikipedia.org/wiki/Vertex_%28graph_theory%29
12:03	nic77	hey sag
12:03	nic77	yeah its gonna get sad here in a min
12:03	SagiCZ1	nic77: how come?
12:04	nic77	so that wiki link makes graph theory seem more like a venn diagram with arbitrary
12:04	nic77	im just a n00b with weak math thats why sag
12:04	nic77	but theres something about it i do understand, how to use it
12:04	SagiCZ1	nic77: that wouldnt sadden me
12:04	nic77	once i limp along
12:05	nic77	then i was projecting, giving
12:05	nic77	heck ill be back, have to tend to some animals
12:06	nic77	and turbotax
12:06	SagiCZ1	nic77: dont pay taxes to your turtle and feed the state some salad
12:07	nic77	hmm, is that a phrase i dont know of?
12:07	nic77	has a ring to it
12:07	SagiCZ1	nic77: no i just made that up, sorry
12:07	nic77	never heard that :D
12:07	nic77	well im in, good one
12:08	nic77	turtles would be easier tbh
12:08	nic77	were talkin old english mastiffs
12:08	nic77	im all for a high priced turtle bloodline
12:08	nic77	with registration papers
12:09	nic77	probably galapagos
12:09	nic77	hardly enough to start a registry though
12:09	SagiCZ1	nic77: ;)
12:09	nic77	id get into that biz
12:10	nic77	maybe
12:10	nic77	i dunno, depends on if it would end up animal smuggeling
12:10	nic77	or if it ends at uncle bills
12:10	nic77	thats how you know your into something one step above drugs
12:10	nic77	when product arrives at uncle bills
12:11	nic77	2 years ago in ohio next to me a wild animal refuge place had nearly all of its animals set loose
12:12	SagiCZ1	nic77: how many did you catch_
12:12	SagiCZ1	?
12:12	nic77	none im in indiana
12:12	nic77	its just related a bit
12:12	nic77	like in one of those graphs
12:14	nic77	the laws and stigma with having tigers and crap, the dark element and shady dealings
12:14	nic77	that lead to lawsuits and crap
12:15	nic77	county and state came down on the guy so he turned them loose and i think killed himself
12:15	nic77	its a bad industry like the drugs, comes with alot of baggage
12:17	nic77	like stolen art or something that comes along with opposing armys/ideals
12:17	nic77	but the laws in ohio were laxed so people were able to carry out 80's cocaine cowboys type behavior when it came to zoo animals
12:17	nic77	i dont think they killed any animals but did have to round them up
12:17	nic77	it took a few days, maybe even a few fews
12:17	nic77	big cats and stuff
12:18	nic77	lion on the highway, bears even tigers oh my
12:18	arrdem	#clojure-offtopic please <3
12:19	clgv	arrdem: do you think the bot will follow those orders?
12:20	arrdem	clgv: I wasn't paying enough attention to decide it's a bot :P
12:20	arrdem	just glanced over and saw a couple paragraphs of offtopiuc
12:20	clgv	arrdem: either a bot or a child ;)
12:20	nic77	around here a guy used to have a refuge, ran IMI irving materials
12:20	nic77	he got sued for capping the market and scaming extortion crap
12:20	nic77	concrete guys :D
12:20	nic77	they are just how they seem in the movies it seems
12:20	nic77	a guy in new york bought a dog from us and is currently slandering, hes a concrete bull too
12:20	nic77	is attempting to slander and blackball us currently
12:20	nic77	so becarefull when you mention turtles
12:20	arrdem	oh good.
12:20	nic77	hah, a bit of both
12:20	nic77	sorry guys
12:20	arrdem	amalloy_: fingerguns please. nic77.
12:20	nic77	what?
12:20	clojurebot	what is not a bug
12:23	Mr0	was that me? im having a keyboard issue at the moment
12:24	justin_smith	nic77: we have an off topic channel, please don't fill up this channel with off-topic stuff
12:24	justin_smith	we have a separate off-topic channel, that is
12:25	nic77	started out on topic just so you know
12:25	nic77	:D
12:27	nic77	motives are easy for some
12:27	nic77	i do want to know about the graph theory question
13:10	{blake}	What is "com.s
13:10	{blake}	un.jndi.cosnaming.CNNameParser.CNCompoundName"? I just got a class not found on it. It went away as mysteriously as it arrived, but Google turns up nothing and I'm curious.
13:10	{blake}	(Or at least, Google turned up nothing I could understand.)
13:11	justin_smith	{blake}: typically anything in com.sun is an implementation detail that shouldn't usually be used directly
13:11	{blake}	justin_smith: Yeah. I'm just wondering what I did to get a ClassNotFound on it!
13:14	tehgeekmeister	is there a core clojure thing or idiom for applying a bunch of functions to one collection?
13:14	H4ns	i'm wondering: how do other people deal with their dependency issues? we're spending hours and hours getting things sorted out with the help of :pedantic? true, but it is a very painful process and every single library update potentially creates hours of work fixing the dependency tree.
13:14	H4ns	is our project just too big? we've not even started, really, but it already is a royal pain in the back.
13:15	{blake}	tehgeekmeister: Wouldn't that be comp?
13:15	tehgeekmeister	{blake}: i want to apply them independently, not sequenced
13:15	justin_smith	H4ns: I tend to be pretty conservative about using or updating libs
13:17	Glenjamin	tehgeekmeister: you mean like ##(->> [1 2 3] (map inc) (map #(/ 2 %))) ?
13:17	lazybot	⇒ (1 2/3 1/2)
13:17	H4ns	justin_smith: that basically shifts the cost to later, but it is not really a solution, i'd think?
13:17	{blake}	tehgeekmeister: So you have collection C and you want a result of C1, C2, C3, C4 where f1, f2, f3 and f4 have been applied to C, respectively.
13:18	justin_smith	H4ns: why would I ever need to pay that cost? if a library has changed API I don't feel obligated to update all my code for the new API
13:18	Glenjamin	justin_smith: security updates, cascading upgrades
13:19	Glenjamin	larger change when you do finally upgrade being harder than many incremental changes
13:19	Glenjamin	dependency management is generally just hard
13:19	justin_smith	Glenjamin: fair enough. My security mostly rests on nginx and ssl.
13:19	justin_smith	sure, that's true
13:19	tehgeekmeister	{blake}: no, one set of functions, one collection. apply each function, and return a seq of the results
13:19	Glenjamin	if ring had a security vuln, you'd patch nginx to block the attack vector?
13:20	H4ns	we're now switching to flattening out our dependencies so that instead of relying on the version information that is in the jars that we use, we put all libraries with explicit versions into our own project as dependency.
13:20	Glenjamin	tehgeekmeister: something like ##(map apply [inc dec] (constantly 1))
13:20	lazybot	java.lang.IllegalArgumentException: Don't know how to create ISeq from: clojure.core$constantly$fn__4178
13:20	justin_smith	Glenjamin: nginx, in practice, blocks most of the http attack vectors. I of course would update ring if/when something in ring was a security issue
13:20	H4ns	not sure if that is a great idea, but it sure sounds better than endless exclusions lists
13:21	Glenjamin	tehgeekmeister: sorry: ##(map apply [inc dec] (repeat 1))
13:21	lazybot	java.lang.IllegalArgumentException: Don't know how to create ISeq from: java.lang.Long
13:21	Glenjamin	dammit, repl hates me today
13:21	Glenjamin	oh, i see why that doesn't work
13:21	Glenjamin	anyway, the general idea would be to map function application over the list of functions
13:22	justin_smith	Glenjamin: in my clojure experience, the lib updates that cause dep management issues are the feature / API updates, a security update will be a drop in replacement and not cause code to break.
13:22	justin_smith	not that we can always count on backports if we are stuck on an old version of course
13:23	H4ns	justin_smith: like i said, you're shifting the cost to later.
13:23	Glenjamin	the cascading updates problem is more common i reckon
13:23	Glenjamin	you need feature X from lib A, causing the whole tree to be invalid
13:24	justin_smith	H4ns: I have yet to need to change an app because of a security vulnerability in a clojure lib.
13:24	{blake}	Clojure does seem to be less susceptible to this than other languages, at least.
13:24	Glenjamin	{blake}: what makes you say that?
13:25	Glenjamin	justin_smith: right, but there's no funamental property of clojure that makes that true
13:25	{blake}	Glenjamin: Ruby and Python experience. =P
13:25	{blake}	Ruby, in particular, is like a kid with ADD.
13:25	tehgeekmeister	Glenjamin: ah, that makes sense
13:25	tehgeekmeister	Glenjamin: getting back into the functional mindset still
13:25	justin_smith	Glenjamin: immutability actually has some nice consequences regarding security holes, actually.
13:26	Glenjamin	it's just that there's a relatively small number of libraries, and the maintainers are still using them
13:26	Glenjamin	give it another 5 years of dependency fragmentation
13:26	{blake}	Everything is deeply intertwingled. And minor revisions tend to invalidate reams of code.
13:26	justin_smith	and immutable data structures (at least their pervasiveness), is definitely a property of clojure
13:26	Glenjamin	could you elaborate? i think we get benefits from dynamic typing in this space
13:26	{blake}	Glenjamin: Fair point, but I do think there's a priority in Clojure of interoperability.
13:27	{blake}	I love the way Noir split itself, e.g.
13:28	justin_smith	Glenjamin: also in terms of upgradability, idiomatic clojure code tends to be coded against interfaces and not concrete types, which tends to reduce brittleness.
13:29	justin_smith	but yes, dependency management can be a pain in the ass, even in clojure
13:29	Glenjamin	do protocols have to do complete implementations?
13:29	justin_smith	not at all
13:29	Glenjamin	that's a big +1 for this then
13:29	Glenjamin	new interface method doesn't break existing users
13:29	justin_smith	right
13:30	justin_smith	in pure clojure (not using java), interfaces don't even need complete implementations
13:30	Glenjamin	oh? i thought you had to declare all methods?
13:30	justin_smith	not from clojure, no
13:31	llasram	Welllll, but unimplemented protocol methods will throw an exception
13:31	llasram	So it isn't that useful a feature (IMHO)
13:31	justin_smith	llasram: if you call them
13:31	justin_smith	llasram: we are talking about libs breaking because a spec changed
13:31	llasram	Ah,
13:31	llasram	nm
13:31	llasram	Apologies for not reading full context before jumping in
13:32	H4ns	i think the basic problem is that one cannot express imprecise dependencies in leiningen
13:32	H4ns	(or maven, ftm)
13:32	justin_smith	Glenjamin: I don't want to get too arrogant and say that we don't have version problems or security issues in Clojure, that would be foolish
13:32	Glenjamin	i think its relatively accurate at the moment
13:32	Glenjamin	but not because of anything that's a property of the platform
13:32	justin_smith	but there are measurable ways that Clojure does things better than average, and I think if you are doing a lot of frequent upgrading you might be doing unneeded work
13:33	Glenjamin	https://groups.google.com/forum/#!topic/clojure/WuS31RSiz_A
13:33	Glenjamin	"Dependencies suck, so stop sharing code please"
13:33	Glenjamin	to tl;dr it somewhat uncharacteristically
13:34	perplexa	heya
13:35	perplexa	i have a midje test and a function with 2 external dependencies that i want to test. i see that there is the (provided) statement to do that stuff, but i don't quite understand how i would use that to rewrite functions in different namespaces
13:36	perplexa	my function (execution-epochs) calls (h/last-execution) and (t/now) and i want to mock those, but (provided (h/last-execution :test) => (t/date-time 2015 2 27 23 0 0 0)) does not seem to give the desired rresults
13:36	perplexa	and tips where to start? ;/
13:36	Glenjamin	anyway, the above problem is why i keep thinking about writing a package manager for utility functions
13:38	tbaldridge	perplexa: instead of trying to test that way, build your system via modules, if you want to mock date-time, make date-time a protocol that you can pass a implementation of to execution-epochs
13:39	tbaldridge	you gain two things a) more extensible code, b) less reliance on wacky macros like providing
13:39	tbaldridge	*provided
13:39	perplexa	hmyeah the code is bad ;p
13:40	perplexa	my actual problem is something else, i just wanted to write a test for the function to be able to write something new that still behaves the same, but i guess i'll have to do that without a test ;x
13:43	perplexa	was using clj-time 0.9.0's periodic function, it somehow stopped working for me, i tried to figure out why as it was working previously. eventually downgraded to 0.8.0, but that unfortunately doesn't support an end-datetime, but the funtion itself didn't cause errors anymore. so i upgraded again and usddenly 0.9.0 was working in my repl again, but still fails on a production machine. the function's code: ...
13:43	perplexa	... https://github.com/clj-time/clj-time/blob/master/src/clj_time/periodic.clj and i was getting "ClassNotFoundException clj_time.periodic$periodic_seq$fn__4142 java.net.URLClassLoader$1.run (URLClassLoader.java:366)" in the repl and the caught exception on the other machine is "No matching method found: multipliedBy for class org.joda.time.Period".
13:44	perplexa	the class not found exception magically disappeared after down and upgrading clj-time, though ;/
13:44	justin_smith	perplexa: maybe you had some stale compiled artifacts that didn't get cleaned properly?
13:45	perplexa	justin_smith: lein clean should do the trick, though?
13:45	justin_smith	yeah
13:46	justin_smith	you could run lein deps :tree to verify you are seeing the version you expect pulled in
13:49	perplexa	justin_smith: that looks horrible
13:49	perplexa	hadoop ftl ;<
13:49	zacts	hellofunk: yo man
13:50	hellofunk	zacts: 'sup dude
13:50	zacts	hey, so what do you think about those computer music books?
13:50	justin_smith	perplexa: bad dep tree?
13:51	hellofunk	amalloy_: hickey's strange loop transducer's talk and the wikipedia page for folds seem to provide different explanations of the difference between foldl and foldr
13:52	hellofunk	amalloy_: hickey's seemed to imply that foldr was basically a cons operation, and foldl is a loop operation, but in both cases the supplied initial value was used from the start of the list. the wiki page diagrams suggest foldr uses the supplied default at the end of the list?
13:53	perplexa	justin_smith: tells me about lots of exclusions that i should consider
13:54	dnolen	Facebook announced React Native - target Android & iOS w/ JS - too bad ClojureScript isn't on board yet w/ React ;)
13:54	justin_smith	hellofunk: if you cons, the first value you add ends up at the end
13:54	justin_smith	,(reduce cons () (range 10))
13:54	clojurebot	#<IllegalArgumentException java.lang.IllegalArgumentException: Don't know how to create ISeq from: java.lang.Long>
13:54	justin_smith	,(reduce #(cons %2 %) () (range 10))
13:54	clojurebot	(9 8 7 6 5 ...)
13:55	justin_smith	or is that a non-sequitor?
13:55	hellofunk	,(reduce #(cons %2 %) [] (range 10))
13:55	clojurebot	(9 8 7 6 5 ...)
13:55	arrdem	hellofunk: conj
13:56	arrdem	fast append fts
13:56	arrdem	*ftw
13:56	justin_smith	arrdem: I was making a point about cons / fold
13:56	justin_smith	it would have been simpler code, but less clear a point, if I used conj
13:57	hellofunk	justin_smith: but reduce in clojure is a foldl so not sure you can really get the proper operation expressed
13:57	arrdem	justin_smith: understood. my read of hellofunk's last was that s?he was trying to show the appending behavior.
13:57	justin_smith	ahh, rightr
13:58	dino-	I just started working on clojure-koans. When I do `lein koan run`, after a test fails, it just hangs there and doesn't exit and I have to use ctrl-c. Is that normal?
13:58	hellofunk	for example, if your reduce operation is + and you supply init of 5 and a sequence of (range 5) my interpretation is that a foldr would start with 1 + 2, and the 5 doesn't enter the picture until the end
13:59	hellofunk	that's the suggestion from the wikipedia diagram, but perhaps i am interpreting it wrong
14:00	hellofunk	(actually i meant the first 2 would be 0 + 1)
14:00	justin_smith	hellofunk: are foldl / foldr about the order of the operations, or the effective order in which they are applied? Would it not still be a valid foldl if the compiler noticed that + is associative and mixed up the order to optimize it?
14:00	hellofunk	justin_smith: well, that is indeed the question i'm asking
14:01	justin_smith	hellofunk: at least in haskell, order of execution is undefined unless you explicitly use something like a monad that enforces ordering
14:01	justin_smith	I don't know how far we can generalize that though
14:01	hellofunk	justin_smith: checkout these two diagrams for foldl and foldr http://en.wikipedia.org/wiki/Fold_%28higher-order_function%29
14:03	hellofunk	justin_smith: based on these, what would you say are the first two args to a foldr of + over (range 5)
14:04	dino-	Oh, I see, when I edit and save (in another shell) while that `lein koan run` is still running, it unblocks and keeps going. cool!
14:06	justin_smith	hellofunk: 4 and 3. But what I was trying to get at is that "first" doesn't have to determine order of computation, as long as your transformation preserves the correct result
14:07	hellofunk	justin_smith: it would seem to me that a non-lazy version of foldr with reduce would simply be to reverse the sequence first, then proceed with a normal reduce?
14:09	justin_smith	when I read that diagram, foldr preserves order
14:09	justin_smith	also, I think foldl and foldr are terrible names that makes it much too easy to accidentally swap them
14:09	justin_smith	and I would have named them fold and dlof if I invented them
14:10	justin_smith	hellofunk: oh, wait, the operation is cons, so yes, reversing once and then reducing with cons would preserve order
14:10	justin_smith	sorry
14:10	hellofunk	it's not clear to me how laziness relates to this. if this is a foldr approach: 1 + (2 + (3 + (4 + 5))) then what does it mean for this to be lazy? lazy in the sense that you could get just the result of the operation over the first 2 items rather than the whole? in which case, how is that possible when each item is an operation that requires the entire rest of the sequence operated on as well?
14:10	justin_smith	double reverse
14:12	justin_smith	hellofunk: how did laziness come into play here again?
14:13	hellofunk	justin_smith: foldr is recognized as a lazy fold, which does not exist in clojure, while foldl is not lazy. amalloy_ and the hickey video have both pointed this out, but i'm trying to see what that means
14:14	justin_smith	hellofunk: I think that this means the difference between doing a cons at each step, vs. creating a new lazy-seq thunk for each step
14:14	justin_smith	these give you opposite effective orderings
14:14	justin_smith	specifically for the collection building (which can be lazy / non-lazy)
14:15	hellofunk	justin_smith: actually i just saw that laziness is discussed about halfway down that wiki page
14:16	hiredman	laziness in clojure is tied to particular constructs, like a lazy-seq, in haskell were fold is lazy, everything is lazy, so if you do a fold with +, you get a value that is effectively a promise to give you the value of all those additions on demand
14:17	hiredman	in clojure + is not lazy, and numbers are not lazy, so a fold that produces a number cannot be lazy
14:17	Glenjamin	dnolen: do you think cljs would be able to target React Native?
14:18	hellofunk	justin_smith: those diagrams don't really do much to clarify the issue, and in fact i think lead to more confusion
14:19	hellofunk	hiredman: this issue is more about foldl and foldr and why foldr is lazy and foldl isn't, i'm trying to understand where the lazyiness comes into play with a foldr structure like this: 1 + (2 + (3 + (4 + 5)))
14:19	hiredman	hellofunk: that is a bad structure to try and understand with
14:20	hiredman	hellofunk: the order of the reduction doesn't matter in that structure
14:21	justin_smith	hellofunk: they are helpful in that they emphasize the fact that it isn't about order of evaluation, it is about the ordered structure of the result, and which nodes have which other nodes as their tail in the resulting tree
14:21	justin_smith	whether that tree is an evaluation tree or a literal tree (because your function was a collection building one)
14:22	justin_smith	hellofunk: and as hiredman already mentioned, even in a lazy language, you get a thunk that forces all of those additions - you can't consume that partially
14:23	justin_smith	also, if you wanted a reducing function, - may be a better example, since order matters with -
14:24	hellofunk	good point
14:24	hiredman	the laziness doesn't really matter with numbers, the laziness comes in to play when you are building a collection (often a list) in the reduction
14:25	justin_smith	exactly
14:25	hellofunk	i see, so thinking of these examples with cons in mind is the best way to visualize
14:27	hiredman	if you start with [1 2 3 4] and right fold cons you get (cons 1 (cons 2 (cons 3 (cons 4 nil)))) and left fold (cons 4 (cons 3 (cons 2 (cons 1 nil))))
14:27	hellofunk	hiredman: that's helpful
14:28	hiredman	the right fold has the output that corresponds to the first input element first, which, if the whole cons process is lazy, means you can grab the first element without ever doing anything else
14:28	justin_smith	hiredman: wait, wouldn't a right fold consume the leftmost element of the sequence first?
14:28	hellofunk	justin_smith: actuall the wiki diagram lines up with the hiredman example, i think it's clicking with me
14:28	hiredman	the left fold has to descend the entire structure in order to return the first result
14:28	justin_smith	oh, never mind, that's what you said
14:29	justin_smith	I just misread what he was saying
14:29	hellofunk	hiredman: that's actually quite illuminating, thanks
14:42	amalloy	another way to say what hiredman is saying is that foldr can be lazy iff the folding function is lazy in its accumulator. a foldl looks like (f (f (f (f 1 2) 3) 4) 5), which means you can't really get anything useful from the beginning of the sequence, since those are buried in more calls to f
14:43	amalloy	but a right fold looks like (f 1 (f 2 (f 3 (f 4 5)))), so you if f is something like "if the left number is odd, return it, otherwise return the right number), you can do that without even looking at (f 2 ...)
14:48	hellofunk	amalloy: it's confusing to me to call the clojure reduce a foldl since it doesn't at first seem to meet either the foldl or foldr patterns. (reduce (flip cons) [] [1 2 3]) starts with (cons 1 []) and technically you could stop right there if you didn't need the rest.
14:50	amalloy	hellofunk: but you can't, because there are no lazy data structures in use there
14:50	hellofunk	amalloy: so is the burden for laziness on the data or on the function? ie the consumer or that which is consumed?
14:50	amalloy	the language
14:51	amalloy	clojure's evaluation semantics are eager, so no foldl can ever be lazy
14:51	amalloy	(no foldr either)
14:51	hellofunk	amalloy: is not a lazy foldl possible in clojure, maybe not with reduce?
14:51	stuartsierra	reductions!
14:52	amalloy	well, kinda
14:52	justin_smith	I could imagine using reductions to eg. determine with laziness whether the sum of a given collection of positive numbers was greater than some threshold
14:52	amalloy	reductions lets the consumer of the fold choose when to stop folding (by not requesting any more elements), but it doesn't let the fold function itself decide
14:52	stuartsierra	not the same as foldl, but gives you a sequence of "intermediate" reduce results.
14:53	hellofunk	i'm familiar with reductions but it seems like a cheat in the theory of folding
14:53	felixflores	does anyone know how to insert a record's default values using korma?
14:53	gfrederi`	reduce can do that via reduced though, right?
14:54	aaelony	I thought I had heard that Incanter 1.9 or 2.0 was available somewhere but I only see 1.5.7-SNAPSHOT in https://github.com/incanter/incanter... Am I looking in the wrong place?
14:54	felixflores	I tried something like (insert article (values {}))
14:54	felixflores	it doesn't work
14:54	amalloy	gfrederi`: yes. that's the one i think of as a cheat, though
14:54	hellofunk	gfrederi`: well reduced is also a bit of a cheat. i'm curious about the higher theory of how folds work in a pure functional context
14:55	hellofunk	i would probably greatly benefit from studying haskell for awhile
14:56	gfrederi`	oh okay I didn't know we had ruled out cheating
14:57	justin_smith	"#clojure: oh, okay, I didn't know we had ruled out cheating"
14:57	hellofunk	gfrederi`: for example, i don't think haskell requires the notion of either "reductions" or "reduced" to achieve these basic features of lazy folding
14:57	justin_smith	gfrederi`: makes a nice tagline I think
14:57	aaelony	ah... looks like the develop branch is at 1.9.1-SNAPSHOT level ... (nevermind)
14:58	xyz_	Hello, can someone help me iterating over a nested map and pass the path (ie. the [:key1 :key2]) to my mapping function ?
14:59	amalloy	gfrederi` sounds like an italian name. signior frederi
14:59	justin_smith	xyz_: so you want to get to every nested value of a map, and get the path to that thing from the root?
15:00	hellofunk	justin_smith: even if that' not xyz_ 's question, it's one i'd be curious about
15:00	xyz_	@justin_smith exactly
15:01	amalloy	http://stackoverflow.com/a/21769786/625403
15:01	xyz_	the function should receive the key, value and path (but the path could be enough, right ?)
15:01	amalloy	just needs a slight modification to use the vals as well as the keys
15:02	amalloy	or take what's there and map get-in over it
15:02	justin_smith	(inc amalloy)
15:02	lazybot	⇒ 217
15:02	justin_smith	that's awesome
15:02	dino-	I'm very new. A little confusing you can't use nil in place of '()
15:03	justin_smith	(inc puredanger) ; for writing it
15:03	lazybot	⇒ 31
15:03	dino-	I guess it doesn't mean "the empty list"
15:03	xyz_	@amalloy ok I'll try and be back thank you (I'm still a Clojure noob so even if I find it I'll ask for feedback). Thank you
15:03	justin_smith	dino-: () is an empty collection, nil is the null pointer
15:03	justin_smith	dino-: in many languages, a null pointer means "empty list", this is not one of those
15:04	dino-	justin_smith: thank you
15:04	justin_smith	dino-: the function seq will give you nil for an empty list, or an equivalent sequence for any other sequential though
15:04	justin_smith	,(seq ())
15:04	clojurebot	nil
15:04	justin_smith	,(seq '(1 2 3 4))
15:04	clojurebot	(1 2 3 4)
15:05	justin_smith	dino-: there is also the function not-empty, which unlike seq preserves the collection's type, but returns nil for empty input
15:05	justin_smith	,(not-empty {})
15:05	clojurebot	nil
15:05	justin_smith	,(not-empty {:a 0})
15:05	clojurebot	{:a 0}
15:05	justin_smith	,(seq {:a 0})
15:05	clojurebot	([:a 0])
15:05	dino-	huh, ok
15:06	justin_smith	that should suffice for eg. translating common lisp code where they use empty as false in loops
15:07	justin_smith	(inc amalloy)
15:07	lazybot	⇒ 218
15:07	justin_smith	just scrolled down to your answer on that keys-in thread
15:08	amalloy	a lot simpler indeed, but slightly inconsistent
15:08	dino-	Ok, thanks again
15:12	amalloy	justin_smith: i really love that pattern for exploring a decision tree
15:13	amalloy	(fn paths [node] (if (done? node) '(()) (for [choice (choices node), subpath (paths (make-choice node choice))] (cons choice path)))) can apply to all sorts of things
15:14	justin_smith	amalloy: maybe it should have a name?
15:15	amalloy	i've thought about it. it seems like most cases where you'd want to use it, though, it's shorter/easier to inline the logic than to pass in the three necessary functions
15:15	justin_smith	I bet it is the equivalent of some graph algo I don't know the name of
15:15	amalloy	justin_smith: i mean it is literally just DFS
15:15	justin_smith	amalloy: right, but even if it is sufficiently abstracted, we can name it
15:15	justin_smith	OK
15:15	justin_smith	oh, right
15:15	justin_smith	d'oh
15:15	clojurebot	Pardon?
15:15	hellofunk	amalloy: what is DFS mean?
15:15	justin_smith	clojurebot: d'oh
15:15	clojurebot	Cool story bro.
15:16	justin_smith	depth first search
15:16	hellofunk	oh, word
15:16	amalloy	the trick is figuring out how to think of your problem as a search tree so that Depth-First Search makes sense for it
15:16	justin_smith	right, that is often the issue with graph / tree stuff
15:16	justin_smith	the hard part is seeing the graph / tree
15:16	justin_smith	after that it's rote
15:16	hellofunk	"for" is pretty much a depth first logic with cut options, sorta
15:17	justin_smith	hellofunk: plus combinatoric expansion
15:17	amalloy	justin_smith: i feel like this paths function is almost tree-seq
15:17	justin_smith	yeah, it's close, but it preserves some info tree-seq does not
15:17	justin_smith	it has more points of articulation
15:19	SagiCZ	haha.. i finally got rid of that pesky number one behind my nick. Behold!
15:19	hellofunk	,(source for)
15:19	clojurebot	Source not found\n
15:19	justin_smith	hellofunk: oh man, the source of that is insane
15:19	justin_smith	have fun reading it!
15:19	hellofunk	i thought there was a source finder on here
15:19	amalloy	$source for
15:19	lazybot	for is http://is.gd/x33p56
15:19	amalloy	~def for
15:19	benaiah	anyone know of a good tutorial for writing command-line applications in clojurescript?
15:19	justin_smith	woah, interesting use of notification
15:20	amalloy	yeah, i don't really understand why clojurebot sends that as a notice, compared to most of the stuff it says
15:20	slipset	one could argue that the source for for looks quite like the source for doseq
15:21	slipset	anyone in for a slight refactor to pull out the common stuff?
15:21	justin_smith	slipset: they have a lot in common
15:21	slipset	justin_smith: I would imagine
15:21	amalloy	"slight refactor". best joke of the day so far
15:21	hellofunk	amalloy: weird they link to different source files for for
15:21	slipset	~for
15:21	justin_smith	slipset: in clojure.core perf is more important than cleanliness in general
15:21	clojurebot	for is like doseq, but for values instead of side effects
15:21	amalloy	hellofunk: they're running different versions of clojure
15:21	slipset	~doseq
15:21	clojurebot	doseq is like for, but for side effects instead of values
15:21	amalloy	(inc clojurebot)
15:21	lazybot	⇒ 46
15:21	amalloy	and whoever taught him that
15:22	slipset	given the docs I would imagine they have something in common
15:22	hellofunk	(inc whoever)
15:22	lazybot	⇒ 1
15:22	hellofunk	well, just trying to help
15:23	slipset	justin_smith: I've come to understand that perf is quite important, but still.
15:25	justin_smith	slipset: I promise, a PR against core that does not at least keep performance as good as it is currently has very little chance of being accepted (unless it improves correctness of course)
15:25	tbaldridge	personally I think doseq should move to something that uses reduce
15:25	tbaldridge	that way doseq over a vector can use the vector's internal reduce logic
15:26	slipset	tbaldridge: and be faster :)
15:28	{blake}	benaiah: Nope. But if you're willing to take the startup time hit, it shouldn't be any different from any other kind of Clojure app.
15:28	gfredericks	(defmacro doseq "shells out to DOS to determine the equality of x and y" ...)
15:28	justin_smith	{blake}: point is you compile once and run the js, so it eliminates the startup time hit
15:28	benaiah	{blake}: the startup time hit is the reason I wanted to write it in cljs instead of clj. Doesn't seem to be much support for that atm.
15:29	justin_smith	if you are repeatedly using the same utility, which is very common in the command line world
15:29	justin_smith	benaiah: it would be very nice to have a turnkey way to generate a cljs command line util, I agree. I don't know if one exists or not.
15:30	benaiah	justin_smith: perhaps I'll write one. The node CLI tooling is pretty well documented
15:30	justin_smith	benaiah: that would be awesome
15:30	justin_smith	perhaps as a lein or boot task
15:32	justin_smith	benaiah: probably the best approach is to make a template that has the infrastructure for a command line app (simple arg parsing, ready to compile all the cljs into one target, generates a batch file that runs that via node maybe?)
15:35	{blake}	benaiah, justin_smith: Oh, clojureSCRIPT. Duh. Yeah, that makes a difference.
15:36	{blake}	That would be cool.
15:40	zacts	huh, see I'm thinking of using clojurescript for cli apps too
15:40	zacts	jvm clojure is just too slow for startup times
15:40	justin_smith	something in the wind, clearly
15:41	justin_smith	zacts: it can be much faster if you uberjar, or with a repl if you use fast-trampoline or use the standard repl instead of nrepl
15:41	zacts	you know what would be cool would be the clojure equivalent to the scsh (scheme UNIX shell)
15:41	zacts	oh let me check that out
15:41	justin_smith	zacts: that actually would be an awesome thing to port to clojure, I would love that
15:41	zacts	yeah, perhaps I may look at it within the next few months for a fun side project
15:41	justin_smith	zacts: faster lein is documented here https://github.com/technomancy/leiningen/wiki/Faster
15:41	slipset	https://github.com/Raynes/conch
15:42	zacts	I already have too many projects at the moment, and I'm still a newbie
15:42	justin_smith	slipset: yeah, conch would actually be a decent starting point to doing something like scsh
15:43	zacts	nice
15:46	justin_smith	I like this trick: ##(update-in {:a 0} [:b] (fnil identity 42))
15:46	lazybot	⇒ {:b 42, :a 0}
15:46	justin_smith	(here I don't ever want nil as a value, so absence vs. nil value is a non-issue)
15:47	zacts	I wish I had some good books on design
15:47	zacts	designing abstractions
15:48	amalloy	so justin_smith, i tried writing out a generic tree-search algorithm and applying it for the keys-in problem. it's not really better IMO: https://www.refheap.com/866bb537b31215e53e04713c1
15:48	zacts	but perhaps that's something that can really only be learned from experience
15:48	opqdonut	justin_smith: I usually just merge...
15:48	amalloy	needs a million dang parameters
15:48	zacts	SICP I think approaches this though
15:48	opqdonut	,(merge {:b 42} {:a 0 :c 1})
15:48	clojurebot	{:a 0, :c 1, :b 42}
15:49	justin_smith	opqdonut: right, that is similar, but it does the wrong thing if the final map has {:b nil}
15:49	justin_smith	opqdonut: remember I said I didn't want nil values
15:49	opqdonut	oh right
15:49	l1x	if somebody is bored, what is the Clojuresq version of this https://gist.github.com/l1x/9f4c6d68e0a407185277?
15:49	justin_smith	this way I can skip a winnowing step
15:51	justin_smith	l1x: I am 80% sure the answer involves using for
15:52	amalloy	haha yes
15:52	l1x	justin_smith: :)
15:52	amalloy	it is the same dfs, by the way
15:52	amalloy	to solve l1x's problem
15:52	l1x	justin_smith: how can you recur with for?
15:52	justin_smith	amalloy: yeah, that is so parameterized that you may as well type the full thing out and just have the name for the pattern
15:52	amalloy	it's a tree search over a tree with branching factor 10 and depth number-digits
15:52	justin_smith	(regarding that refactor)
15:53	l1x	amalloy: hmm yeah
15:55	amalloy	&(map (partial apply str) ((fn perms [num-digits] (if (zero? num-digits) '(()) (for [perm (perms (dec num-digits)), n (range 10)] (cons n perm)))) 2))
15:55	lazybot	⇒ ("00" "10" "20" "30" "40" "50" "60" "70" "80" "90" "01" "11" "21" "31" "41" "51" "61" "71" "81" "91" "02" "12" "22" "32" "42" "52" "62" "72" "82" "92" "03" "13" "23" "33" "43" "53" "63" "73" "83" "93" "04" "14" "24" "34" "44" "54" "64" "74" "84" "94" "05" "15" "25" "... https://www.refheap.com/96597
15:56	amalloy	or i guess if you want it to do them in the other order, ##(map (partial apply str) ((fn perms [num-digits] (if (zero? num-digits) '([]) (for [perm (perms (dec num-digits)), n (range 10)] (conj perm n)))) 2))
15:56	lazybot	⇒ ("00" "01" "02" "03" "04" "05" "06" "07" "08" "09" "10" "11" "12" "13" "14" "15" "16" "17" "18" "19" "20" "21" "22" "23" "24" "25" "26" "27" "28" "29" "30" "31" "32" "33" "34" "35" "36" "37" "38" "39" "40" "41" "42" "43" "44" "45" "46" "47" "48" "49" "50" "51" "52" "... https://www.refheap.com/96598
16:00	amalloy	l1x: does that make sense?
16:00	l1x	huh
16:03	l1x	amalloy: thanks
16:06	slipset	amalloy: same discussion, different place? http://stackoverflow.com/questions/18246549/cartesian-product-in-clojure
16:09	amalloy	slipset: yes
16:10	benaiah	calling `lein cljsbuild once` on this barbebones repository: https://github.com/Benaiah/cljs-test yields the following error: https://www.refheap.com/96599, which I keep getting every time I try to use cljsbuild on several different starter templates. Anyone have any idea what's going on?
16:13	tcrayford____	anybody ever got ring's wrap-reload middleware working with tools.namespace and component/etc?
16:18	benaiah	does this project.clj look correct for a clojurescript project? https://github.com/Benaiah/cljs-test/blob/master/project.clj
16:31	notbrent	benaiah: https://github.com/swannodette/mies-node-template
16:31	notbrent	you need to make some changes to your project.clj
16:31	notbrent	(sorry if that was answered already, my irc client timed out)
16:32	dnolen	benaiah: looks ok though that build is missing an :id
16:32	xyz_	Hello, this is a follow-up question to getting the path of the keys in a map. Can the value returned from the mapping function replace the value in the previous map (constructing a new map)?
16:32	benaiah	notbrent: no, yours was the first answer. thanks, that's quite helpful
16:33	justin_smith	xyz_: do you need both the paths and to do the mapping, or would just the mapping suffice? because if the latter you may just want to use clojure.walk/post-walk to update leaves
16:34	xyz_	I can't use walk because I need to get a value from another map (a symetrical map if you will)
16:34	clojurebot	It's greek to me.
16:35	xyz_	@justin_smith sorry, the paths and what ?
16:35	justin_smith	xyz_: ahh, OK. In that case I would first get all the paths, then do a reduce over each of the paths of the map, doing the updates as needed
16:35	justin_smith	xyz_: if you just needed to do the updates of leaves, and did not have a separate use for the path to that leaf, postwalk would be simpler. But you do have a separate use for the path.
16:36	xyz_	justin_smith: yes the problem I had was with the path
16:36	xyz_	justin_smith: I'm not sure I get the reduce idea completely yet though
16:36	justin_smith	so now you have a function that returns all the paths, so do a reduce over all the paths, with the initial map as your init value
16:36	justin_smith	and for each path either update or not as needed with the help of the other map
16:37	xyz_	justin_smith: let me get back to digging ^^ I'll come back again
16:37	amalloy	xyz_: it sounds like you are trying to solve some larger problem, and you've been asking about sub-problems. that's a good instinct, avoid dumping everything on #clojure all at once, but i worry that you're breaking it down into the wrong subproblems. what is the overall thing you are trying to achieve?
16:38	justin_smith	xyz_ (reduce (fn [m path] (assoc-in m path update (get-in m2 path))) initial paths) ; something like this is what I meant
16:38	justin_smith	where the update function looks at the path, and the two maps, and either returns the new value to use at that place (usually the original I would guess)
16:38	xyz_	amalloy: xy problem, I know :) I am trying to convert a map of string values into a map of correct values (correct as defined by the type of the values of the first nested map). Not sure if I'm clear.
16:39	xyz_	justin_smith: ok I'll try to digest that. Thank you
16:39	amalloy	so you have a "template map" or a schema or something, describing the types that another map should have, and you want to walk over the second one and modify its stuff to be the right type
16:40	xyz_	amalloy: exactly. like a schema or a merge
16:40	hiredman	imagine the maps are flat instead of nested
16:41	hiredman	{:a {:b 1}} -> {[:a :b] 1}
16:41	hiredman	do the same thing you would do for the flat maps, but use assoc-in instead of assoc
16:41	amalloy	i don't actually use prismatic/schema myself, but this sounds like something it wants to do: https://github.com/Prismatic/schema#transformations-and-coercion
16:42	xyz_	amalloy: yes. This is an exercise for me, and I can't read cljx yet ^^
16:43	xyz_	hiredman: I think that this is what amalloy hinted at earlier ? with the keys-in function ?
16:43	amalloy	well. you don't need keys-in if your schema map is structured like {[:a :b] String} instead of {:a {:b String}}
16:44	amalloy	but of course you can use keys-in on the latter to produce the former
16:44	xyz_	amalloy: it's structured like a normal map (the latter)
16:44	xyz_	amalloy: yes I'll try that. That's a lot of info right now, let me get back to you guys Thanks again !
16:51	gfredericks	oh man core async depends on a sooper old analyzer
16:51	adnauseaum	clojure is gay and i hate it
16:51	benaiah	notbrent: thanks a lot - that minimal project has made the cljs stuff much clearer than the documentation was able to alone
16:52	notbrent	:) no problem - dnolen did all of the work, i just linked to it ;P
16:59	dbacar	Hi, anyone here with experience regarded to Jsoup?
16:59	adnauseaum	jsoup is gay and i hate it
17:00	justin_smith	amalloy: ping
17:00	dbacar	really?, it seems nice , what would you recommend?
17:00	adnauseaum	justin_smith: ping
17:00	adnauseaum	i'd recommend something that's less gay
17:00	amalloy	so does anyone think it's crazy that i like to respond to pings by just mashing my fingers on the keyboard instead of typing out a proper pong?
17:01	justin_smith	hehe
17:01	justin_smith	I think this guy wants to be restricted from using out channel
17:01	adnauseaum	nah bruh
17:01	amalloy	oh hey, i forgot i can do that
17:01	adnauseaum	we gucci yo
17:01	AeroNotix	oh cool, someone to add to my ignore list
17:01	adnauseaum	woah bruh u talkin shit? talk shit get hit
17:01	AeroNotix	yep. It's offical.
17:01	amalloy	i wonder if i can remember how
17:02	AeroNotix	adnauseaum: peace
17:02	hiredman	I bet you just repsond with garbage to a syn too
17:02	adnauseaum	pzl8
17:02	adnauseaum	reecies peaceis
17:02	adnauseaum	FUCK YOU FAGGOT
17:02	dbacar	there is a static class in Jsoup called Document.OutputSettings, i can't find how can i add it to my ns declaration, any suggestions?
17:03	justin_smith	dbacar: (:import whatever.Jsoup.Document.OutputSettings) I would think
17:04	justin_smith	of course make sure you have your project set up to pull in the dep too
17:05	justin_smith	oh, (:import org.jsoup.nodes.Document.OutputSettings) - or looks like since it is an inner class you need to do (:import org.jsoup.nodes.Document) then use Document$OutputSettings
17:07	dbacar	justin_smith: thank you, it worked, i tried the following and it throws an classnotfoundexc, do you know why?
17:07	dbacar	[org.jsoup.nodes Document Document.OutputSettings
17:07	dbacar	Attribute Attributes Comment DataNode
17:07	dbacar	Element TextNode]
17:07	dbacar	inside an :import
17:07	travisrodman	amalloy: thanks for handling that.
17:07	justin_smith	dbacar: it's not Document.OutputSettings, it's Document$OutputSettings, that's the syntax for an inner class
17:08	justin_smith	and I don't think you can import the inner class (not 100% sure of that though)
17:08	dbacar	i have a long way to go :(
17:08	justin_smith	amalloy: seconding that, thanks
17:09	amalloy	justin_smith: you can import inner classes
17:09	dbacar	justin_smith: it seems you can import inner classes the way you told
17:09	amalloy	they're just ordinary classes with funny names
17:11	justin_smith	oh, that's great
17:11	justin_smith	I just never had to do it yet
17:12	tcrayford____	get used to $ errywhere
17:19	csd_`	If I have a hash map tracking something like users, which its keys indexed to usernames, and I want to check whether a username exists already, would it be more efficient to check a separate hash-set, or to get the hash-map's keys and check those?
17:19	justin_smith	csd_`: contains?
17:20	csd_`	hash-set would contain strings, hash-map would contain maps of user state
17:20	csd_`	keyed by username
17:20	justin_smith	why separate the keys into a hash-set?
17:21	csd_`	im just wondering whether that would be more efficient for purposes of checking if a name exists already
17:21	justin_smith	I am suggesting the function, called "contains?", which should be no slower than hash-set lookup
17:21	csd_`	i'm writing a toy irc server and am thinking about data structures
17:21	justin_smith	in case that wasn't clear and that just looked like a question
17:21	justin_smith	cool
17:22	justin_smith	another options is find
17:22	csd_`	is it poor form to pass global state as a variable to functions, or would it be better to keep it as a global atom
17:22	justin_smith	,(find {:a 0} :a)
17:22	clojurebot	[:a 0]
17:23	dnolen	csd_`: hash set is backed by a hash map, so no won't be faster
17:23	justin_smith	csd_`: well, if you pass it to each function, it doesn't have to be global any more
17:23	justin_smith	you can just call it "accumulated state"
17:23	justin_smith	and the fact that it never branches or backtracks is an implementation detail
17:24	justin_smith	and actually that is better than a global atom, if your design can pull it off
17:24	csd_`	i'm guessing it wouldn't work once i introduce concurrency
17:24	justin_smith	csd_`: the reason I suggest find is that it unambiguously tells you if the value was there, and gives you the existing value if it is
17:24	justin_smith	csd_`: yeah, once you have concurrency you probably want an atom or ref
17:25	csd_`	justin_smith: would you define either as a global?
17:25	justin_smith	the atom, yeah
17:25	justin_smith	if you are just threading a value through, no, not as global
17:25	csd_`	ok
17:25	justin_smith	but I would combine the atom with passing the value, if you do it that way
17:26	justin_smith	rather than doing lookup at each processing step in one request
17:26	csd_`	i havent worked with refs yet, haven't had any need
17:26	justin_smith	and then when done with it, put that back in the atom
17:26	justin_smith	yeah, usually an atom will suffice, but for parallel interdependent modification refs can be a perf boost
17:32	{blake}	Is there a clever way, given the compojure format of (-> (handler) (middleware) (middleware) (middleware) (middleware)), to add some kind of wrapper that tracks each request as it mutates?
17:32	{blake}	I guess I could macro it up.
17:33	justin_smith	{blake}: there is a debug-middleware you can place between any two of the others
17:33	justin_smith	because hey, all the middleware are are wrappers, so throw it in the chain
17:34	justin_smith	or you could hack together a replacement for -> that also debugs the intermediate values
17:34	{blake}	justin_smith: Thanks. I was just thinking "SHOW ME ALL THE THINGS!" but that's probably not worth it.
17:34	{blake}	justin_smith: Ha! Yeah, I'm probably not that ambitious right now.
17:34	justin_smith	{blake}: for that, just throw the debug middleware in between each and every middleware
17:34	justin_smith	and have fun scrolling :)
17:35	justin_smith	in practice, it's usually a specific spot in the chain I am looking at, so I insert the debug middleware in that spot
17:35	justin_smith	{blake}: https://github.com/pjlegato/ring.middleware.logger
17:36	{blake}	Thanks. For some reason my local content is not having the content-type served. Which nobody cares about but IE.
17:37	justin_smith	{blake}: woah, this one looks pretty cool https://github.com/magnars/prone
17:38	{blake}	justin_smith: Woah, indeed.
17:39	justin_smith	also, .ninja domains are now available for $15
17:39	justin_smith	I hope clojure.ninja gets used for something awesome
17:42	TimMc	justin_smith: It won't, it will be recruiters.
17:42	justin_smith	:(
17:42	amalloy	rockstar.clojure.ninja the secret subdomain for the best candidates
17:42	justin_smith	haha
17:43	justin_smith	I hope that at least taco.cat ends up being something good
17:43	amalloy_afk_ninj	ugh c'mon irc, gimme a break
17:43	justin_smith	oh, taco.cat is already up, and it's a silly python program
17:43	tcrayford____	justin_smith: taco.cat is a friend of mine's domain haha
17:44	justin_smith	oh, nice
17:44	tcrayford____	was weird as hell seeing it here
17:44	{blake}	Did they just abandon the TLD system or something?
17:44	justin_smith	{blake}: massive expansion I guess
17:44	tcrayford____	it's a money grab haha
17:45	justin_smith	of course it is
17:45	{blake}	justin_smith: Well, good. I hated to see ninjas left out.
17:45	tcrayford____	lots of folk trying to get new TLDs that are interesting
17:45	TimMc	.cat is Catalan though, not one of the new TLDs
17:45	tcrayford____	I uh, don't think google take TLD into account like they do the actual domain though, for exact matches
17:45	tcrayford____	yeah taco.cat is like 5 year sold
17:46	justin_smith	looks like clojure.sucks is not taken
17:46	justin_smith	but clojure.rocks is
17:46	{blake}	Ooh.
17:46	tcrayford____	another friend has "never.computer"
17:46	justin_smith	that's a good one
17:46	{blake}	"Clojure.sucks but it.sucks less than everything else."
17:46	tcrayford____	(or at least, he used to)
17:47	Ash	clojure.pizza
17:47	justin_smith	such.wow is available (and now I move this to #clojure-offtopic)
17:55	{blake}	Wow, putting in logger resulted in two dozen files being downloaded!
17:56	justin_smith	deps?
17:56	justin_smith	that's a lot for one middleware
17:56	AeroNotix	something something transitives
17:57	{blake}	https://gist.github.com/anonymous/0862bc4fadb769c89f84
17:59	justin_smith	ahh, of course, the logging libs
18:01	{blake}	It seems...excessive. That much code should fix your problem for you. =P
18:02	{blake}	Heh. I love lein under Windows. "Terminate batch job (y/n)"...then whatever you type, terminate.
18:02	justin_smith	process management under windows in general is "fun"
18:03	{blake}	For the "Dwarf Fortress" definition of "fun".
18:03	justin_smith	does lein officially support windows now? at least at one point the attitude was that none of the lein devs used windows, and if it worked there it was basically a happy accident but not intentional
18:05	{blake}	justin_smith: Honestly? It's great under Windows. I don't know if it's officially supported or not, but I switch between Linux/Windows all the time, with no trouble.
18:07	justin_smith	that's cool
18:07	justin_smith	my info is likely out of date
18:09	{blake}	Honestly, having gone from Ruby to Clojure, I appreciate the crap out of lein. Although the last Ruby install I did was nice, with RVM.
18:10	justin_smith	(inc lein)
18:10	lazybot	⇒ 4
18:10	tcrayford____	(inc lein)
18:10	lazybot	⇒ 5
18:10	tcrayford____	you deserve more than that leinininininingen
18:11	{blake}	(inc lein)
18:11	lazybot	⇒ 6
18:11	{blake}	(inc ALL-THE-LEINS!)
18:11	lazybot	⇒ 1
18:11	TimMc	banananananagen
18:12	uptown	what's the sanest way to mock environment variables as in (with-environment-variables {"foo" "bar"} (fact "foo should be bar" (env :foo) => "bar))
18:13	uptown	(assuming midje and environ)
18:13	hiredman	the sanest way to do anything starts with not using midje
18:13	uptown	oh um
18:13	justin_smith	(inc hiredman)
18:13	lazybot	⇒ 68
18:13	{blake}	uptown: Isn't the point of having environment variables not caring what's in them?
18:16	AeroNotix	environment variables do seem a weird thing to want to mock
18:21	uptown	what, then, should i
18:22	TimMc	AeroNotix: Uh, why?
18:22	TimMc	You need to thread config into a program one way or another. Maybe I'm misunderstanding the midje question...
18:23	AeroNotix	TimMc: if your application's behaviour vastly changes with different environment variables, it's probably time to rethink.
18:23	TimMc	AeroNotix: I don't think that's a particularly unusual way to configure an app!
18:24	uptown	sorry, had a coffee accident here ha
18:24	AeroNotix	It's not, I'm just saying that needing to test each individual setting in tests seems like the envvars control too much
18:24	AeroNotix	e.g.
18:24	AeroNotix	DB_HOST=foo
18:24	AeroNotix	is a good envvar
18:24	AeroNotix	SHOULD_USERS_HAVE_ACCESS_TO_FEATURE=true
18:24	AeroNotix	also good
18:25	AeroNotix	WHEN_X_HAPPENS_DO_Y_AND_THEN_Z=true
18:25	AeroNotix	bad envvar
18:25	AeroNotix	namsayin?
18:25	uptown	DB_HOST might want a different value for testing purposes
18:25	uptown	DB_HOST=foo_test
18:25	AeroNotix	sure, and you set that before running the test stuite
18:26	AeroNotix	DB_HOST=foo lein midje
18:26	AeroNotix	run your test like that
18:26	uptown	in a repl, autotesting
18:26	uptown	if we agree that starting a repl is the way to go then okay
18:26	AeroNotix	I don't do that
18:26	AeroNotix	but whatevs
18:28	TimMc	uptown: I don't think your tests should assume/test for any particular default for environment variables.
18:28	justin_smith	my typical pattern is to have a DEPLOYMENT variable in env, and then based on its value, load resources/config/development.clj or resources/config/staging.clj, resources/config/production.clj etc.
18:28	justin_smith	and you just decide at startup which is the right config file to load
18:28	justin_smith	actually .edn is better than clj, but whatever
18:29	justin_smith	oh, and also resources/config/testing.clj too
18:30	uptown	TimMc: I'm not really testing the value itself, i just want to mock it for downstream purposes
18:33	uptown	justin_smith: i'm trying to be 12factor compliant, which means trying to keep all the config in env
18:34	uptown	if env let me load a properties file programmatically just for test purposes i'd be happy that way
18:35	TimMc	uptown: So when your program goes out to get an env var, it demands that it be there? It can't fall back to a default?
18:35	justin_smith	uptown: looking at 12factor's website, you can use resources/config/x.edn as long as they are not checked into version control
18:35	AeroNotix	ugh, 12factor
18:35	justin_smith	uptown: and given that this is a security concern, you can also use an encrypted properties file
18:36	justin_smith	(with the key coming in from env of course)
18:39	uptown	TimMc: i'm wrapping a service that wants a config map so i'm just passing along whatever i find at the appropriate names, including nothing
18:39	uptown	the failure is handled
18:41	cbp	#join #clojure-offtopic
18:41	cbp	er
18:42	uptown	justin_smith: i was hoping to use env's property file support as the file-based solution, which works fine except for the testing build / teardown cycle
18:42	justin_smith	uptown: it wouldn't work to have a testing specific config?
18:43	Glenjamin	I normally put a thin layer between ENV vars and the app (eg. Config function/map)
18:43	Glenjamin	Then just use a different one for tests
18:43	uptown	i can certainly get it to but was hoping to keep the codepath and resource count at zero beyond the test module
18:44	Glenjamin	Turning env vars into config values is an app init concern
18:45	Glenjamin	Oh, that's basically what Tim said
18:45	mgaare	can also run whatever you want inside of a with-bindings on environ/env and merge in whatever you want to change from other things
18:46	uptown	aha that works. thanks all
18:47	uptown	say but if i can return to the previous topic: midje is evil?
18:47	uptown	what's the current flavor?
18:50	justin_smith	uptown: clojure.test, with the assistance of libs like expectations and test.check
18:51	justin_smith	that's the impression I get at least, I just use regular clojure.test
18:52	{blake}	Some people feel very strongly about midje, on both sides.
18:52	{blake}	There are probably people who are very strongly ambivalent about it, too.
18:53	justin_smith	uptown: the problem I see with midje is that it seems in many ways to carry a specific strongly opinionated approach to how tests should be done, that doesn't neccessarily match the general philosophy or style of clojure itself
18:54	TimMc	uptown: I use midje. It is mostly nice, and when it isn't, I can fall back to building my own test utilities.
18:54	justin_smith	eg. in clojure.test the tests are functions, and I can define a function that calls test/is and call it from inside a test, it uses general functional programming concepts
18:55	uptown	i inherited some code that uses midje and just kept using it.
18:55	uptown	hm i'll keep an eye on that
18:55	uptown	thanks
18:59	TimMc	Midje locks you in a bit more but comes with a larger library of testing utilities.
19:03	{blake}	I've been using this logger at times (http://rjevans.net/post/2655430930/adding-custom-middleware-to-ring-compojure) to help me grasp compojure/ring/etc better.
19:04	{blake}	And I notice, when I use it, only the user's request is logged, i.e., "myapp/whatever/page". Even if that page requests all kinds of resources.
19:05	{blake}	At the same time, I note that I have middleware (rsrc/wrap-resource "/public") which I know is being used to serve resources.
19:05	{blake}	I'm not getting how that's handled.
19:05	amalloy	your logging middleware must be applied inside of the wrap-resource handler insread of around it
19:06	amalloy	change from (-> app (wrap-logging) (wrap-resources)) to (-> app (wrap-resources) (wrap-logging))
19:07	{blake}	I've been moving it up and down the chain.
19:08	{blake}	I go to "/" and I see that request followed by the redirect to "/login"...
19:09	{blake}	Wait...app vs....I'm doing "handler/api app"...
19:15	{blake}	All my middleware: https://www.refheap.com/96607 and I still only see the main request, no resources, no AJAX.
19:36	notbrent	I feel like I would prefer if put/take macros for core.async were defined as >! and !< rather than the take being <! - I somehow still get tripped up on that
19:36	notbrent	I believe that would be easier to visually parse
19:37	notbrent	what do you folks think?
19:38	notbrent	functions rather*
19:45	{blake}	Man, I hate it when I think I've figured out something, made it work, then can't make it break.
19:47	xyz_	Hello, I'm back. I made a lot of progress thanks to your help, but I need help with my conversion. Here is what I have so far : https://www.refheap.com/96610 . Any help ?
19:48	{blake}	xyz_: Are you trying to convert to a Number now?
19:49	xyz_	blake: well the first ones are broken as well and I can't seem to figure why. I have a "cannot be cast to clojure.lang.IFn" error
19:50	{blake}	desired-type
19:50	{blake}	Wouldn't "(case (desired-type)..." cause "desired-type" to be executed as function?
19:50	xyz_	blake: hopefully it's just something silly, I'm still learning Clojure syntax
19:50	xyz_	blake: oh maybe that's why. Let me try :)
19:51	xyz_	blake: you are right it's getting better. I have a "No matching clause: class java.lang.Boolean", so it went further ^^
19:52	{blake}	There ya go.
19:52	xyz_	blake: I'm a bit ashamed now not to have found it... Thank you !
19:53	{blake}	xyz_: Heh. No worries. I'm a relative novice, too.
19:53	amalloy	xyz_: you can't really use a 'case to match on Class literals
19:54	amalloy	because Boolean is really a symbol, not a class, and 'case doesn't evaluate its test clauses
19:54	xyz_	amalloy: I just saw that now by googling for my error : http://stackoverflow.com/questions/12028944/clojure-case-statement-with-classes
19:54	xyz_	amalloy: does that seem correct to you ?
19:55	amalloy	yes
19:56	xyz_	amalloy: ok thanks. Time to learn about defmulti and defmethod then ^^
19:56	amalloy	i mean you can just replace (case x ...) with (condp = x ...) for a simple fix
19:57	xyz_	amalloy: oh intersting. I'll try that first then.
20:00	bodie_	notbrent, I think <! >! is clearer since the ! always comes after mutating functions, and the functions are < and >
20:00	notbrent	that makes perfect sense
20:00	notbrent	thanks bodie_
20:00	bodie_	it is a little ugly though ;)
20:01	notbrent	indeed - I like with >! and !< how you can think of ! as a channel in the middle of two processes
20:01	notbrent	one on the left puts it on, one on the right takes it off
20:01	notbrent	but the convention of ! after mutating functions should take precedence there for consistency
20:01	notbrent	unfortauntely
20:02	notbrent	unfortunately*
20:02	bodie_	hello all -- I'd like to use a lein plugin from its latest upstream on github, rather than from clojars. how would I go about doing that?
20:03	bodie_	I just want to put the dep in my project.clj as usual, and have it be usable by other members of my team. but it seems kinda un-kosher.
20:05	notbrent	also curious about bodie_'s question
20:05	amalloy	you can't run anything off of like "latest github commit"
20:05	amalloy	it needs to be a jar somewhere in the world
20:05	amalloy	you can jar it up yourself and publish it, if you want a particular version to be available
20:06	bodie_	hmm, ok. then how would I go about including it as a dependency?
20:09	notbrent	https://github.com/ato/clojars-web/wiki/About#do-i-have-to-use-leiningen-with-clojars-and-clojars-with-leiningen
20:09	notbrent	in particular: https://github.com/technomancy/leiningen/blob/preview/doc/DEPLOY.md
20:09	{blake}	We use a local archiva server.
20:11	bodie_	notbrent, cool, thanks
20:12	notbrent	np
20:15	R0B_ROD	Hello
20:15	R0B_ROD	Currently exploring the Clojure Koans
20:21	n0n3such	moan
20:22	TEttinger	loan
20:22	n0n3such	loon
20:22	TEttinger	goon
20:22	n0n3such	bufoon
20:22	TEttinger	heh
20:22	n0n3such	lol
20:33	notbrent	(= ___ true)
20:42	dangit	I’m trying to use compojure with SS’s components. I can’t figure out how to get the component instance to the routes.
20:43	dangit	Can someone give me some guidance?
20:49	noonian	dangit: instead of using defroutes, write a function that takes the components it depends on as arguments and returns a call to compojure.core/routes with your routing code that depends on the component. Then in your server component or however you structure it (maybe have a RingHandler component) you call that function with any components it needs
20:49	zacts	huh, I think I may try the koans myself too
20:49	zacts	what do they teach mainly?
20:49	zacts	or are they just like euler problems?
20:51	noonian	i haven't really done them but I think they are designed to teach an understanding of Clojure more than figure out mathematical problems ala project euler
20:51	dangit	noonian: thanks, that’s helpful
20:51	zacts	oh neat
20:51	noonian	dangit: np :)
20:52	zacts	"By following along the path set before you, you will touch on all the major aspects of the language, from simple datatypes to macros, from tail recursion to Java interoperability. "
20:52	zacts	that's cool, I'll do them then
20:53	zacts	they are cooler than euler for what I want right now
21:06	doky	hello does anybody know why is problem when i sending emails in future throught (map #(send-email %) emails) ? when i send only in repl all emails are send correctly
21:19	justin_smith	doky: map is lazy
21:19	justin_smith	it won't do anything unless you consume the result
21:24	TEttinger	doky: the more correct way is with doseq, but you can also force a lazy seq with doall after using map
21:25	doky	justin_smith: i have cycle, do stuff then wait 2 seconds and repeat but when i try to sending email in that cycle and in one function i have (map #(send-email %) emails) it don't send anything ... but in lighttable when i call that function manually it works ... it took about 3 second to send email probably while it gain a connection
21:25	TEttinger	light table evaluates lazy sequences, just like the REPL will
21:25	TEttinger	but a running program won't
21:25	TEttinger	(doc doall)
21:25	clojurebot	"([coll] [n coll]); When lazy sequences are produced via functions that have side effects, any effects other than those needed to produce the first element in the seq do not occur until the seq is consumed. doall can be used to force any effects. Walks through the successive nexts of the seq, retains the head and returns it, thus causing the entire seq to reside in memory at one time."
21:25	TEttinger	(doc dorun)
21:25	clojurebot	"([coll] [n coll]); When lazy sequences are produced via functions that have side effects, any effects other than those needed to produce the first element in the seq do not occur until the seq is consumed. dorun can be used to force any effects. Walks through the successive nexts of the seq, does not retain the head and returns nil."
21:26	justin_smith	TEttinger: better yet, dorun
21:26	justin_smith	yeah
21:27	doky	TEttinger: thanks doall works great :)
21:27	doky	TEttinger++
21:28	doky	justin_smith++
21:28	justin_smith	doky: it's a small difference for smaller collections, but in general dorun is the better choice if you don't need the return value
21:28	justin_smith	(inc TEttinger)
21:28	lazybot	⇒ 38
21:28	doky	(inc TEttinger)
21:28	lazybot	⇒ 39
21:28	doky	(inc justin_smith)
21:28	lazybot	⇒ 175
21:29	doky	justin_smith: i forgot
21:34	zacts	hum.. I guess I'm stuck with JSON for now, as edn isn't really stable, and is subject to change
21:34	zacts	Unless you guys know of a better JSON alternative
21:36	justin_smith	zacts: I have had good luck with using cheshire for edn / json conversion
21:37	justin_smith	and also, edn hasn't really change much so far - though I guess there are no promises
21:38	justin_smith	but for the subset of edn that converts directly to json, I wouldn't expect changes (strings, numbers, vectors, hash maps)
21:38	TEttinger	(inc justin_smith)
21:38	lazybot	⇒ 176
21:40	TEttinger	zacts: I like protocol buffers when I'm in .NET-land, but Kryo is very good at serializing Java stuff (and I think it works for Clojure too)
21:40	zacts	justin_smith: ah ok let me check out cheshire
21:40	zacts	justin_smith: huh, perhaps I may consider a subset of edn yeah
21:41	zacts	TEttinger: well serializing clojure is just having literal clojure expressions in a text file.
21:41	TEttinger	https://github.com/sritchie/carbonite
21:41	zacts	I mean clojure is basically already an AST
21:41	zacts	which is cool
21:41	justin_smith	zacts: also there is transit, if you want efficient transfer with json on the client end and clojure on the server end
21:41	TEttinger	right, but binary serialization is sometimes desirable (much smaller)
21:41	zacts	oh I didn't even consider that TEttinger
21:48	zacts	how about some good clojure general purpose parsing frameworks?
21:49	zacts	kind of like ANTLR
21:49	justin_smith	instaparse is popular
22:01	zacts	coolio
22:57	R0B_ROD	Seeking some basic introductory instructions on programming using Clojure.
22:57	R0B_ROD	~book
22:57	clojurebot	book is http://www.pragprog.com/titles/shcloj/programming-clojure
22:57	R0B_ROD	Thanks and good evening to all.
23:14	AlwaysBCoding	what is a good way to cache an api response with a clojure/ring web app? i.e. if someone goes to a given route serve a static response for an hour then have the first response after the hour is up go through the controller get the new response and cache that? I’ve done this before w/ Rails, just wondering what library / workflow I should use to do it in Clojure?
23:17	fairuz	AlwaysBCoding: how about core.cache?
23:20	fairuz	I would create a middleware to check for the request if there's a cache for it, and if it does, just send the cache instead
23:21	AlwaysBCoding	is there a good example anywhere of using core.cache with ring?