#clojure log - Jan 13 2014

The Joy of Clojure
Main Clojure site
Google Group
IRC
List of all logged dates

0:05 amalloy: ah yes, nicola's patch fixes the issue that i carefully checked for but still failed to notice

0:26 logic_prog: is the semantics for "throwing an uncaught exception inside a go block" well defined ?

0:26 if so, what is it defined as?

0:49 cerealklr: Hey folks. Is there any way to have comments at the end of lines in multiline strings?

0:49 I've got a kind of gnarly regexp that I'd like to document

0:52 amalloy: cerealklr: well, a regex isn't a string. i don't think there's an answer for strings, aside from just writing (str "foo" ;; blah <RET> "bar" ;; duh)

0:53 cerealklr: My bad, I'd assumed the same rules would apply.

0:53 amalloy: but java's regular expressions support extended mode with the (?x) embedded flag (most regex engines do, even if they may have a more convenient syntax like /foo/x)

0:53 so #"foo # comment RET bar # comment2 ..."

0:53 ought to work

0:53 er

0:53 #"(?x)foo # comment RET bar # comment2 ..."

0:55 for more details about how extended mode changes the way regular expressions are compiled...well, look up extended mode, i guess

0:55 cerealklr: Dang. JS Regexps don't suppport extended mode, and I'm working in clojurescript. Didn't stop to consider the difference.

0:56 I suppose I can shim out to JS's Regexp ctor if there isn't a more elegant in language way.

1:00 amalloy: well, js does support extended mode, but it doesn't support inline flags

1:00 whereas clojure has no way to specify flags other than inline (or interop)

1:04 deadghost: hmm ordinarily you'd use vectors instead of lists except when writing macros right?

1:05 I have a few places where I'm using lists as tuples and vectors seem more appropriate

1:07 seangrove: cerealklr: Sounds like a good opp for a library

1:07 cerealklr: seangrove: I ended up using re-pattern with str

1:08 a macro would probably bundle that nicely

1:08 amalloy: cerealklr: well, not very nicely, really, right? because you have to double-escape things in a string but not in a regex

1:09 eg, #"\d{4}" vs (re-pattern (str "\\d" "{4}"))

1:09 cerealklr: fair enough. Didn't consider that

1:09 it's a tradeoff either way

1:10 amalloy: yeah

1:10 ruzu: `(println "yay")

1:12 amalloy: deadghost: huh? you use vectors and lists pretty much the same in macros as you would otherwise

1:14 deadghost: probably worded that poorly

1:14 you'd typically use vectors instead of lists right?

1:15 except in macros where you need to build an s-exp

1:15 which requires lists

1:16 amalloy: i mean, your output will include lists (seqs, really), but so does the output of many functions; and you'll probably type as many () and []s as you do when writing any other function

1:17 and the output will include vectors too, eg for emitting let-blocks

1:53 mischanix: Trying to pprint a seq that's basically the result of filter and getting ClassCastException: clojure.lang.PersistentHashMap cannot be cast to java.util.Map$Entry

1:54 Here's relevant code/stack trace: https://www.refheap.com/23162

2:07 amalloy: the formatting of that code is really weird, mischanix - it makes it very hard to read what is going on. but here it looks like the problem is that you're calling (vals (:heroes s)), and (:heroes s) is a list of maps, rather than a single map

2:07 Raynes: Holy shit.

2:08 Gotta say, I couldn't have come up with that layout myself.

2:10 mischanix: Ah, nice find, thanks.

2:11 amalloy: if i were going to make a whitespace-only rewrite of your code, it would look like https://www.refheap.com/23176

2:13 mischanix: Yeah. First project in clojure, never touched a lisp before either.

2:17 Got 600 more lines of stuff that looks exactly like that if you're into the horror genre.

2:20 amalloy: haha no thanks, i saw that a newcomer to the language was trying to write hearthstone and that one snippet was enough for me

2:21 mischanix: Might end up redoing it in C once I know my data structures, heh

2:23 logic_prog: with http://clojure.github.io/core.async/#clojure.core.async/&gt;! , is there any defined behavior for using >! / >!! on a closed channel?

2:23 i.e. is the behavior undefined, throw, or nil ?

2:28 proteneer_: anyone play around with ClojureC to be able to comment on its maturity?

2:33 amalloy: i would be astonished if clojurec were mature, proteneer_. aside from the readme itself saying it's experimental and not-polished, one of the two commits made in december was "make things compile again" after a two-month period of no commits, in which presumably things did not compile

2:33 proteneer_: hm

2:33 ok

2:34 thanks. I've been itching to dive into a FP language, but I don't know any java

2:39 TEttinger: proteneer_, there's a few people in here who are users of haskell, it's a solid choice if you are averse to java stuff

2:41 seangrove: Anyone have good suggestions for parsing a string as html in clojurescript?

2:52 bitemyapp: seangrove: jquery

2:52 wait, lemme think about that for a moment.

2:52 seangrove: bitemyapp: I suppose actually I should rephrase

2:52 bitemyapp: seangrove: http://api.jquery.com/jquery.parsehtml/

2:52 seangrove: I would like to parse a string into an html representation, but not actual html (could be unsafe), and then pull out the text content of the nodes.

2:53 bitemyapp: seangrove: then my link seems a good fit.

2:53 seangrove: So they never need to become dom nodes

2:53 * bitemyapp keeps pointing earnestly at his link

2:53 seangrove: bitemyapp: heh, "jQuery.parseHTML uses a native DOM element creation function to convert the string to a set of DOM elements, which can then be inserted into the document.

2:53 Raynes: lol

2:53 bitemyapp: seangrove: but you skip the "inserted into the document" part

2:53 seangrove: and just iterate over the contents

2:54 seangrove: bitemyapp: Hrm, I need to look up when a resource is downloaded - at tag cration time, or dom insertion time

2:55 Google is also epicly failing me at that query

2:55 I'll ask in an html channel tomorrow

2:55 bitemyapp: seangrove: what sort of resource?

2:56 seangrove: src tag, img tag, a bgaudio tag, etc.

2:56 Anything that could trigger an external resource and be used to track a load

2:56 bitemyapp: seangrove: it depends on where they're located and what sort of tag they are.

2:56 for example, there's a reason you put src tags at the tail of the body, rather than at the top.

2:57 mischanix: fwiw just doing document.createElement().innerHTML= does trigger a load with an <img> tag

2:57 bitemyapp: mischanix: without inserting it into the DOM?

2:59 mischanix: bitemyapp: no insertion: http://i.imgur.com/vB8B4uu.png ; I think the fragments API prevents this, though

3:01 actually no, I think I'm thinking of something else but I don't remember

3:04 * bitemyapp shrugs

3:04 bitemyapp: DOM Stuff.

3:13 seangrove: bitemyapp: Yeah, DOM + JS... it's enough to shake one's faith in the idea of progress and understanding what came before

3:13 But enough for tonight, off to dream of the doms, in whatever form they take

3:13 Xb

3:16 ddellacosta: seangrove: based on testing, seems pretty clear that for an image tag, at least, the resource is requested on creation of the element.

3:17 seangrove: If that's the case, I'll have to find a nice library for parsing strings into clj data structures

3:17 ddellacosta: seangrove: seems like this may have something to do with being a "replaced element:" http://www.whatwg.org/specs/web-apps/current-work/#replaced-elements

3:17 seangrove: yah, I was gonna say, may be better (and faster) to generate strings

3:18 I'm curious if this behavior varies based on the browser, however. I wouldn't put it past IE to do something different.

3:18 seangrove: Anyways, don't let me keep you from DOM dreams. ;-) 'night!

3:19 seangrove: ddellacosta: Oh, I have html strings already. I want data structures, then some filtering on them, then turn them back into (possibly different) dom nodes

3:19 But yes, tomorrow!

3:19 ddellacosta: gotcha.

3:38 logic_prog: given that (>! ... ...) always returns nil, regardless of whether the channel is open or closed

3:38 is htere anyway to quer whether a core.async channel is open or closed?

3:49 alew: what's the difference between drop and nthnext?

3:55 rurumate1: logic_prog: according to the doc string, (<!! ch) will only ever return nil when the channel is closed

3:55 logic_prog: rurumatel, yeah, this doesn't work well since I don't want to be reading from channels

3:55 rurumate1: oh, and !< has the same behaviour

3:55 logic_prog: to actually figure out if they're closed

3:55 rurumate1: I see because that would change them

3:56 logic_prog: i'd prefer soething that didn't alter the state ot the channel, say

3:56 close?

3:58 rurumate1: maybe you can use core.async/filter or core.async/remove on the channel, to prevent it from returning anything non-nil?

3:59 scape: channels should not pass nil

3:59 rurumate1: except when closed, no?

4:00 (!< (filter nil? ch)) can return nil

4:00 I'd wager

4:02 alew: maybe one is more lazy, like the difference between next and rest?

4:02 logic_prog: lol

4:02 what a hack

4:04 rurumate1: yeah, there should be a better way. maybe the core.async developers can give a better answer?

4:07 amalloy: alew: drop can return (), nthnext can return nil

4:08 em-dash: where can I put code (e.g., protocol extensions) that I want loaded everywhere? do I have to require it in to every namespace?

4:08 amalloy: uh, protocol extensions only have to be required once

4:13 em-dash: then I'm likely doing something (else) stupid and not yet aware of it

4:15 amalloy: hmm, does that hold for clojurescript (I would suppose so)?

4:15 amalloy: indeed

5:09 mattyw: real newbie question here - I've got a list of maps that I'm trying to flatten into one map - but it doesn't work as I expect:

5:09 (into {} (map (fn [x] {:a x}) '(:b :c :d)))

5:09 ^^ Must be because I've got a lazy list from doing the map call?

5:10 kzar: ,(into {} (map (fn [x] {:a x}) '(:b :c :d)))

5:10 clojurebot: {:a :d}

5:11 kzar: ,(map (fn [x] {:a x}) '(:b :c :d))

5:11 clojurebot: ({:a :b} {:a :c} {:a :d})

5:11 mischanix: map keys must be unique

5:11 kzar: (apply into {} (map (fn [x] {:a x}) '(:b :c :d)))

5:11 ,(apply into {} (map (fn [x] {:a x}) '(:b :c :d)))

5:11 clojurebot: #<ArityException clojure.lang.ArityException: Wrong number of args (4) passed to: core/into>

5:12 andyf: (apply merge sequence-of-maps) might be what you want

5:12 kzar: ,(apply merge (map (fn [x] {:a x}) '(:b :c :d)))

5:12 clojurebot: {:a :d}

5:12 kzar: ,(mapcat (fn [x] {:a x}) '(:b :c :d)))

5:12 clojurebot: ([:a :b] [:a :c] [:a :d])

5:12 andyf: See also merge-with

5:17 alew: amalloy: thanks, I also discovered nthnext returns a chunkedcons while drop returns a lazyseq

5:18 amalloy: i guess that's truee, but it's too specific

5:18 nthnext is eager, while drop is lazy; the particular classes returned are implementation details (and not always the same)

5:21 mattyw: kzar, mischanix thanks very much for your help

5:21 kzar: yw

5:23 alew: amalloy: by eager you mean it realizes the entire sequence? because it doesn't

5:23 amalloy: no, why would i mean that? i mean it eagerly does work the moment you call it

5:24 alew: I misunderstood eager

5:24 amalloy: (nthnext n xs) immediately calls next n times on xs, whereas (drop n xs) returns a lazy sequence that will call next, or rest, as many times as needed

5:25 alew: Ah ok

5:25 amalloy: &(time (do (drop 1e7 (range)) nil))

5:25 lazybot: ⇒ "Elapsed time: 26.189066 msecs" nil

5:25 amalloy: &(time (do (nthnext 1e7 (range)) nil))

5:25 lazybot: java.lang.IllegalArgumentException: Don't know how to create ISeq from: java.lang.Double

5:25 amalloy: &(time (do (nthnext (range) 1e7) nil))

5:25 alew: arg order is swapped

5:25 lazybot: ⇒ "Elapsed time: 1887.946738 msecs" nil

5:25 alew: right

5:51 honza: Looking for some ideas on how to move from the intermediate Clojure dev to something higher... anyone?

6:05 fredyr: honza: build cool things? :)

6:06 Southy: contribute to an open source project that you think is cool?

6:08 Look at ClojureScript if you are into Web stuff?

6:09 mischanix: Is there a function for this?

6:09 ,(let [l '(1 2 3 4 5)] (concat (take 2 l) (nthrest l 3)))

6:09 clojurebot: (1 2 4 5)

6:10 mischanix: Or a proper way to do it?

6:11 Sfiv: mischanix, you mean remove an element?

6:12 mischanix: So to speak, yeah. Get the list without the nth element

6:23 amalloy: mischanix: generally if you want to remove an element by index, your algorithm isn't very functional; if it really is the way you need to do things, you should pick a data structure for which random removal is more efficient than a list or a vector

6:26 for example, a set of objects, or a sorted map of some kind, like a map from index to object

6:31 mischanix: amalloy: I'm choosing randomly between all possible user actions. Some of these actions can lead to an invalid state, and that ends up returning nil up to the function that's choosing; when this happens, I'm simply recurring the choose function with the nil-resulting function removed from the possible choices. If there are no choices, choose also returns nil

6:31 amalloy: so use a set

6:32 mischanix: mmkie

6:36 Though, how do I pick randomly from a set? Do I end up making it a seq so I can use nth on it?

6:37 Is that not wasteful?

6:39 Also, there theoretically can't be more than ~20 choices

6:44 rurumate1: I need a shingle-producing function; the input would be a vector like ["the" "big" "black" "box"], output would be all 2-shingles: ["the big" "big black" "black box"]

6:45 is there something to help it, or should just write a loop?

6:45 ddima: rurumate1: partition is your friend

6:45 rurumate1: ddima: thx

6:46 DerGuteMoritz: how does partition help here?

6:46 ddima: ,(partition 2 1 ["the" "big" "black" "box"])

6:46 clojurebot: (("the" "big") ("big" "black") ("black" "box"))

6:46 DerGuteMoritz: ah!

6:47 nice

6:47 forgot about those arguments

6:49 AeroNotix: In a doto form, can I conditionally apply some of the functions?

6:50 hmm, I could just put a fn there

7:00 hm, doesn't seem to call it

7:00 rurumate1: AeroNotix: I've asked myself the same thing, also inside a -> form. Please paste your solution if you find it

7:02 AeroNotix: maybe -<> from swiss-arrows can help?

7:02 AeroNotix: wat

7:02 rurumate1: but the diamond can only appear once per form..

7:02 AeroNotix: oh I see

7:02 it's a real thing

7:02 rurumate1: I've started a discussion there, wanna join? https://github.com/rplevy/swiss-arrows/issues/22

7:03 clgv: AeroNotix: but there is `as->` as well

7:03 and that one is in clojure core since 1.5

7:04 &(doc as->)

7:04 lazybot: java.lang.RuntimeException: Unable to resolve var: as-> in this context

7:04 clgv: ,(doc as->)

7:04 clojurebot: "([expr name & forms]); Binds name to expr, evaluates the first form in the lexical context of that binding, then binds name to that result, repeating for each successive form, returning the result of the last form."

7:04 rurumate1: I have no idea what that docstring means. Where are the examples?

7:07 ,(as-> 1 x (inc x) (inc x))

7:07 clojurebot: 3

7:08 rurumate1: brilliant

7:08 ,(as-> 1 x (inc x) (inc (inc x)))

7:08 clojurebot: 4

7:08 rurumate1: it's a dream come true

7:09 locks: rurumate1: if only any of the docstrings had examples

7:13 rurumate1: locks: I've heard the clojure maintainers are trying to keep them short. It's okay as long as you can find examles easily via google, but could not find easily for as->

7:14 hcumberdale: (into {} (map (fn [[a b]] (hash-map a b)) '(["1" "a"] ["2" "b"])))

7:14 is there a better way to do this?

7:15 DerGuteMoritz: ,(into {} '(["1" "a"] ["2" "b"]))

7:15 clojurebot: {"1" "a", "2" "b"}

7:15 DerGuteMoritz: hcumberdale: ^

7:15 hcumberdale: DerGuteMoritz: thx, but why does it work?

7:16 ahh it uses conj and that is working with pairs?

7:16 amalloy: (into x [a b c]) is (conj x a b c)

7:16 and conj on a map wants vector pairs

7:16 DerGuteMoritz: indeed

7:16 rurumate1: (conj {} ["a" "b"])

7:16 ,(conj {} ["a" "b"])

7:16 clojurebot: {"a" "b"}

7:22 hcumberdale: Also a better way to do: (into {} (map #(hash-map (str (:fieldname1 %)) (:fieldname2 %)) (j/query db [selectstatement]))) ?

7:22 jcromartie: how can I make use of Clojure STM while dealing with external transactions?

7:22 hcumberdale: where j is clojure.java.jdbc

7:23 jcromartie: i.e. outside APIs

7:23 locks: rurumate1: yeah, I guess clojuredocs could serve as an example repo, but it does seem a bit oudated?

7:24 DerGuteMoritz: hcumberdale: one way would be (into {} (map str (map (juxt :fieldname1 fieldname2) (j/query db [selectstatement])))) -- you might consider using ->> here

7:25 hcumberdale: sorry, that wasn't exactly the same actually

7:26 or was it :-)

7:26 jcromartie: I obviously can't do external calls inside a dosync

7:26 (HTTP etc)

7:26 DerGuteMoritz: hcumberdale: nah, it's buggy, sorry

7:27 hcumberdale: also thought about using juxt

7:27 but can't create anything with it that works :(

7:28 DerGuteMoritz: hcumberdale: how about something like (apply hash-map (map str (map (j/query db [selectstatement]) [:fieldname1 :fieldname2])))

7:29 rurumate1: locks: searching for as-> on clojuredocs' search engine gives me only one less-relevant result

7:29 maybe the poor search performance is a factor why it's not being used so much?

7:30 hcumberdale: DerGuteMoritz: changes everything to strings, doesn't it?

7:30 rurumate1: or maybe it really has no example for as->, which would be sad

7:31 hcumberdale: (map (j/query db [selectstatement]) [:fieldname1 :fieldname2])) << don't know that this is actually working

7:31 amalloy: (into {} (map (juxt (comp str :field1) :field2) (j/query ...))) looks equivalent to the originalm hcumberdale

7:31 DerGuteMoritz: hcumberdale: oh yeah sorry, I misread your original version

7:32 hcumberdale: thx amalloy!

7:32 amalloy: hashmap is a function you should almost never apply - into {} just works better so often

7:33 DerGuteMoritz: ah, right, query returns a list of maps

7:33 zing

7:33 ok I should not do this on the side :-)

7:34 rurumate1: ,(as-> 1 x (inc x) (inc (inc x)) (or (odd? x) (as-> 1 y y)))

7:34 clojurebot: 1

7:34 rurumate1: smashing

7:36 sm0ke: guys how do i compare this?

7:36 ,(= {:a 0.4} {:a 0.4})

7:36 clojurebot: true

7:36 sm0ke: ,(= {:a 0.4 :b 0.7} {:a 0.4 :b .7})

7:36 clojurebot: #<CompilerException java.lang.RuntimeException: Unable to resolve symbol: .7 in this context, compiling:(NO_SOURCE_PATH:0:0)>

7:36 sm0ke: ,(= {:a 0.4 :b 0.7} {:a 0.4 :b 0.7})

7:36 clojurebot: true

7:36 sm0ke: err

7:37 rurumate1: you're comparing it well imho

7:38 sm0ke: well wtf

7:38 let me check again

7:40 rurumate1: recur will never consume the stack, even when used without loop, right?

7:42 (when the enclosing function is used as recursion point)

7:42 ,((fn [x] (if (< x 10000) (recur (inc x)) x)) 1)

7:42 clojurebot: 10000

7:43 sm0ke: yes recur is optimized

7:43 so i was asking this ##(= 0.7 (float 0.7))

7:43 lazybot: ⇒ false

7:43 sm0ke: so i was asking this ##(== 0.7 (float 0.7))

7:43 lazybot: ⇒ false

7:43 sm0ke: so i was asking this ##(=== 0.7 (float 0.7))

7:43 lazybot: java.lang.RuntimeException: Unable to resolve symbol: === in this context

7:43 sm0ke: hmm weird

7:44 rurumate1: floats are weird

7:44 sm0ke: jeez this will make unit testing hard

7:44 whats the dfault thingy?

7:44 ,(type 0.1)

7:44 clojurebot: java.lang.Double

7:45 rurumate1: smoke, usually when unit testing with floats, you specify a margin of error, e.g. 0.000001

7:46 see http://junit.sourceforge.net/javadoc/org/junit/Assert.html#assertEquals%28double,%20double,%20double%29

7:46 clgv: sm0ke: comparing floating points numbers via equality is seldom right.

7:46 maybe deterministic testing is the only exception.

7:47 sm0ke: clgv: what do you mean?

7:47 clgv: otherwise you always want to know if two floating point numbers are approximately the same , i.e. have only a difference of at most epsiolon (being a small constant, e.g. 1E-10)

7:50 sm0ke: well i cannot have a epsilon based testing as i am comparing deeply nested maps

7:52 gnzh: hi guys, is there any way to stub function behaviour in midje using another function? I want function in my tests to behave like identity function.

7:52 clgv: gnzh: you can do that via (provided ...) for regular functions

7:52 good1: Hi there, what's the proper way to accomplish this CL code in clojure?

7:53 (let ((foo '())

7:53 (bar '(1 2 3)))

7:53 (dolist (x bar)

7:53 (setf foo (cons x foo)))

7:53 foo)

7:53 gnzh: clgv, can you provide simple example? (provided (a s) => s) doesn't work

7:54 clgv: gnzh: but it should. is `a` defined via `defn` ?

7:55 gnzh: `provided` does not work for protocol methods on regular deftypes/defrecords afair

7:55 gnzh: clgv, it just function, nothing fancy

7:58 clgv, I'm not sure that what I wan't is possible to implement

8:02 clgv: gnzh: this works (fact (f 1 2) => 3 (provided (a 1) => 1 :times 1 (a 2) => 2 :times 1)), but I am not sure if you can achieve a general identity function

8:03 gnzh: clgv, well i don't want to stub manually all possible inputs

8:03 clgv: yeah I guessed that

8:04 gnzh: there is clojure.core/with-redefs

8:08 gnzh: clgv, thanks, that works. I was hoping that midje can provide cleaner way.

8:08 fredyr: good1: this is pretty similar perhaps

8:09 good1: making a gist instead of pasting

8:09 good1: https://gist.github.com/fredyr/8399958

8:10 good1: fredyr: You're right sorry!

8:10 clgv: good1: what is the task in general? build a list from a list? concatenation of two lists?

8:11 good1: ##(concat [1 2 3] [])

8:11 lazybot: ⇒ (1 2 3)

8:12 fredyr: clgv: if i understand correctly, the CL code was reverse with an accumulator

8:12 pop and cons

8:14 good1: I am trying to get a new list (foo in this case) so that i can us dstinct to remove duplicates!

8:14 fredyr: good1: oh in that case

8:14 you alway get new lists in clojure

8:15 since they're immutable

8:16 ,(distinct [1 2 3 3 4 6 4])

8:16 clojurebot: (1 2 3 4 6)

8:16 good1: fredyr: Thank you!

8:17 fredyr: good1: np

8:29 nones: hi

8:29 I_play_2_win: hi

8:30 nones: is there ways to get current imported namespaces?

8:30 clgv: good1: you should read about Clojure's immutable/persistent datatypes. you never need to copy clojure datastructures to be able to use the unmodified and modified structure after that

8:34 rurumate1: immutability is a no-brainer unless you're used to doing it wrong

8:35 clgv: well he obviously is not used to it and seems to not know about it, so no reason to insult him

8:37 good1: clgv: Exactly! That was my problem, i see clearly now.

8:37 rurumate1: sorry I didn't want to sound rude

8:37 it's just that people seem to think it's a hard concept simply because there is this funny word

8:38 clgv: good1: as always, I'd recommend to read one of the clojure books to get you fast on track

8:38 rurumate1: where in reality there should be a complicated word for NOT using immutable structures

8:38 clgv: rurumate1: well, it's different from what most are used to from industry/mainstream (C/C++, Java)

8:38 rurumate1: like, "destructive programming"

8:39 clgv: I'm painfully aware of that

8:45 ddima: nones: maybe you can try something like (.getAliases (find-ns 'clojure.core)) (or .getMappings, depending on what you want to see)

8:48 nones: ddima: thanks, I'll search in that direction

8:48 ddima: nones: though in the case of current namespace you might use (the-ns *ns*)

8:49 oh

8:49 you dont even need the-ns then

8:49 k

8:49 nones: ,*ns*

8:49 clojurebot: #<Namespace sandbox>

8:50 nones: yes

8:52 ddima: nones: thought, forget the manual method-calling, there's ns-imports, ns-aliases, ns-refers etc

8:52 -t

8:52 AeroNotix: I want to use assert-args, is there a public replacement?

8:52 since assert-args is private.

8:53 nones: ddima: yes, I already found them

8:53 ddima: ahk. I was just poking around myself, just to see what exists, so

9:00 tbaldridge: AeroNotix: it's open source, so you could always copy what's in core.clj

9:01 clgv: AeroNotix: you can create a public `def` to that private function, e.g. (def assert-args #'clojure.core/assert-args)

9:01 or macro...

9:09 gtrak: et voila.

9:22 AeroNotix: hmm

9:33 edw: Anyone happen to know the currently least deprecated, least painful way of recursively deleting a directory?

9:37 edbond: edw, commons io has deleteDirectory - http://commons.apache.org/proper/commons-io/apidocs/org/apache/commons/io/FileUtils.html#deleteDirectory%28java.io.File%29

9:38 edw: edbond: I was reading in the list archives that delete-file-recursively got the axe because Java 6 cannot tell between a symlink and a real file. I'm guessing the Commons code has the same problem.

9:39 I'm going to non-recursively delete a single level of files, since what I'm working with (Cascading local textline files) have only a depth of one.

9:39 edbond: it seems commons.io skips symlinks, see source http://commons.apache.org/proper/commons-io/apidocs/src-html/org/apache/commons/io/FileUtils.html#line.1550

9:40 edw: edbond: Huh. Let me check that out then!

9:40 edbond: edw, did you look at file-seq? You can list files, filter and call #delete

9:41 edw: edbond: That's what I'm doing. Just reversing the sequence and deleting each file.

9:45 edbond: https://www.refheap.com/23322

9:46 edbond: edw, looks good. It does remove recursively, right?

9:46 edw: edbond: Yeah, for better or worse.

9:46 edbond: edw, I don't understand how you want to improve that

9:47 edw: Scary having a very easy-to-use "rm -rf...".

9:47 edbond: Oh, I just wrote that since asking my original question.

9:48 edbond: you can add a global filter to remove only in *working-directory*

9:49 or not remove at all, move somewhere to /tmp/<timestamp> :)

9:50 edw: edbond: I was looking at the code you referred me to, and, yeah, if you filtered based on the canonical name, removing only those files that are inside the canonical name of the directory you're deleting.

9:50 ...then you'd be a lot safe.

9:50 s/safe/safer/

9:55 deadghost: good morning #clojure

9:55 how are you today

9:56 hyPiRion: silent, mostly.

10:05 mikerod: I'm guessing that a .cljx source file cannot be compiled via clojure-maven-plugin ( from https://github.com/lynaghk/cljx ) .

10:14 edbond: om question: what is the difference bw set-state! and transact! ?

10:17 fredyr: edbond: transact and update work on the the om state via cursors

10:17 edbond: set-state updates the react state

10:18 edbond: the latter is same as doing setState on a normal react component

10:18 edbond: cursor is like pointer in react state?

10:18 fredyr: yes pretty much

10:18 edbond: denoting a particular sub state

10:19 edbond: how can I (should I) convert map to cursor? I have a go block that set-state! and om says 'cannot build from non-cursor'

10:20 I call set-state in go block and got that 'non-cursor' error

10:20 fredyr: edbond: not exactly sure

10:20 edbond: but do you create your om state in the om/root

10:21 edbond: yes, initial state is defined in om/root. Let me try refheap

10:21 fredyr: edbond: sure

10:24 edbond: fredyr, here is a code - https://www.refheap.com/23335

10:24 I misuse build-all I think

10:25 added w-row functions to refheap

10:26 I'm totally lost in components and state :)

10:28 fredyr: edbond: im not really following where load-watchlist come into play here

10:28 edbond: and that's the only place i can see a set-state call?

10:28 edbond: ah, it's called from will-mount

10:29 fredyr: ah okay, in watchlist-table?

10:29 edbond: yes, I updated paste

10:30 fredyr: so in your code now, you only have the top level cursor

10:30 `state` coming into watchlist-table

10:31 edbond: yes, how can I create cursors to some key in that state?

10:31 fredyr: you create new components using build

10:32 with `build` you can take the current state and specify a path to say which "substate" corresponds to that component

10:33 edbond: here's an example, https://github.com/swannodette/todomvc/blob/gh-pages/labs/architecture-examples/om/src/todomvc/app.cljs#L160

10:34 edbond: oh, i might be wrong here

10:34 edbond: the api has changed a bit since last i tried some code on it

10:38 edbond: fredyr, Thanks, I'll dig more into todomvc

10:38 fredyr: edbond: sure, i would've digged more into it as well, but i have to get back to work unfortunately :)

10:39 before you could use it like this: (om/build comment-list app {:path []})

10:40 where the last part was a path that was used similar to get-in and update-in

10:40 but now im not sure

10:43 :s

10:43 mrhanky: anybody used the servants library? does every function used in the worker need a defservantfn ?

11:00 gfredericks: what's the current situation for round-tripping clojure code?

11:01 (in the sense of parsing the code into a data structure and printing it back out)

11:05 nDuff: gfredericks: ...okay, now I'm curious too. I'm sure I've seen a library built for the purpose (retaining reader metadata &c), but I'm having trouble finding it.

11:08 stuartsierra: I think cgrand had some experimental stuff for that.

11:10 hyPiRion: Are you guys talking about rewrite-clj?

11:10 (https://github.com/xsc/rewrite-clj)

11:10 karls: a question about protocols: say i define a protocol in a namespace A and then want to use a method declared in that protocol in a different namespace B. i don't understand the mechanism that allows me to require the protocol namespace A and then do A/some-desired-method. can someone explain this quickly?

11:13 `cbp: karls: you want an explanation on how it works or how you're supposed to use it?

11:13 karls: how it works

11:14 does defprotocol somehow make that method available in the namespace the protocol was declared in?

11:14 gfredericks: stuartsierra: hyPiRion: thanks!

11:14 `cbp: karls: you might wanna look at the source? :P

11:15 matt444: lein-cljsbuild documentation is a little week, how do you get it to bundle your external javascript dependencies?

11:15 I'm doing: :libs ["lib/base64.js"]

11:15 stuartsierra: karls: Protocol methods are real Clojure functions. You can require/refer them just like any other function defined in that namespace.

11:15 matt444: and it doesn't work

11:19 nDuff: matt444: :libs takes a list of directories, not files.

11:19 karls: aha! that clears things up. thanks `cbp and stuartsierra

11:19 nDuff: matt444: (at least, as I follow the example docs)

11:19 matt444: ...beyond that, (1) are you sure the 3rd-party library in question survives the Closure Compiler? (2) Is a reproducer available?

11:19 matt444: nDuff: are those javascript files or clojurescript?

11:20 nDuff: lein-cljsbuild doesn't complain

11:20 nDuff: matt444: the former.

11:21 matt444: "doesn't complain" isn't the same as "leaves your code in a runnable state when it's done".

11:21 matt444: my code is in a runnable state

11:21 nDuff: matt444: keep in mind that it's not cljsbuild that does the (potential) mangling here, it's Google Closure.

11:21 matt444: uhh. If your output included a runnable copy of the third-party library in question, you wouldn't be complaining (if I correctly understand the question).

11:22 matt444: ahh, misunderstood the question

11:22 No, my output doesn't contain the third-party library

11:22 nDuff: matt444: Which level of optimization are you using?

11:22 matt444: none

11:22 nDuff: Then I'd need a reproducer to investigate with.

11:23 matt444: k

11:23 nDuff: (someone like dnolen might have an off-the-top-of-their-head answer, but I'm not...)

11:23 matt444: nDuff: do you happen to know the diff between :extern and :libs

11:23 :externs i mean

11:23 nDuff: matt444: :externs takes a set of headers and uses them to avoid name-mangling.

11:23 matt444: it doesn't actually include those libraries into the output

11:24 matt444: it just makes sure your output will be able to successfully call things having those headers.

11:24 matt444: that's actually how I've always done interaction with third-party libraries -- I've never used :libs

11:24 matt444: so then how do you include them?

11:24 nDuff: matt444: ...from the same place I'm including the javascript that cljsbuild emits?

11:25 matt444: ok, i don't understand what you mean

11:25 cljsbuild emits javascript that it compiles from clojurescript

11:26 nDuff: the what and why of :externs is documented at https://developers.google.com/closure/compiler/docs/api-tutorial3#mixed

11:28 matt444: ok, I have a example project, where should I upload it?

11:29 nDuff: matt444: github is always good. refheap, gist, &c.

11:31 TimMc: hyPiRion: Do you think you can get ahold of eval in swearjurescript?

11:31 hyPiRion: TimMc: wish I could

11:31 nDuff: matt444: ...to restate what I was trying to communicate above -- I *don't* directly include external libraries into cljsbuild's output, and don't try to. I pull them into the document or runtime being invoked the same way I'm pulling in cljsbuild's output.

11:31 TimMc: hyPiRion: I feel like there should be a way to iterate over the properties of something.

11:32 hyPiRion: TimMc: Only way I've discovered yet is by iterating the metadata of a ns

11:32 nDuff: matt444: AFAIK, the mechanisms for including raw javascript into cljsbuild's output all require that javascript to be written in such a way as to be compliant with the Closure Compiler's specifications.

11:32 matt444: ...and, well, rather a lot of 3rd-party code isn't.

11:32 matt444: nDuff: ok thanks, script tags it is

11:32 Uruk: Question from a new user - what's the best idiomatic way to loop across/map across a java object array []? I'm looking at the amap docs and that looks right, but not sure what the ^arguments are about.

11:33 nDuff: matt444: ...if you're going to turn on advanced-mode compilation eventually, you'll need to use :externs.

11:33 llasram: Uruk: Just use `map` unless you have specific need (known performance issue)

11:33 matt444: ok, i will when i get there, thanks again

11:33 llasram: Uruk: All JVM arrays are `seq`able, so the normal Clojure sequence functions work on them just fine

11:34 sveri: hi, how would i add n-times the same element to a collection? (defn addXtoXsNTimes [xs x n]...

11:34 llasram: sveri: (into xs (repeat n x)) ?

11:35 dnolen: matt444: if you have some random JS that won't play with Google Closure you can use :preamble now

11:36 sveri: llasram: thx, i try thaqt

11:36 nDuff: Ooh, nice.

11:37 matt444: dnolen: what does preamble do?

11:37 dnolen: matt444: it just concatenates files to the beginning of the final output

11:37 matt444: perfect, thanks

11:38 TimMc: hyPiRion: To clarify, this also applies to CLJS?

11:38 hyPiRion: TimMc: oh, that really depends I guess?

11:38 You need to be able to use ns-map on namespaces

11:39 TimMc: I think JS exposes some new things that CLJ doesn't provide, like being able to iterate over the properties of things.

11:39 dnolen: matt444: CLJS React users share this problem - you might find the dev and prod build settings described here useful http://github.com/swannodette/om

11:40 matt444: i'm a cljs react user :)

11:40 thanks

11:41 hyPiRion: TimMc: hm, interesting

11:42 anyway, the way I get eval in clj is by doing something like this

11:42 ,(-> #'+ meta (#((`[~@%](*))(*))) ns-map (#((`[~@%](*(+(*)(*))(+(*)(*))(+(*)(*))(+(*)(*))(+(*)(*))(+(*)(*))(-(*(+(*)(*))(+(*)(*))(+(*)(+(*)(*))))(*))))(*))))

11:42 clojurebot: #'clojure.core/eval

11:42 hyPiRion: it works better in clojure 1.0 because of the meta reader (^), so I only need ns-map by then

11:43 TimMc: No vars in cljs, I think.

11:43 hyPiRion: TimMc: right, ns stuff is compile time? Not worked with cljs

11:43 bbloom: hyPiRion: right

11:43 TimMc: Neither have I. :-P

11:43 bbloom: probably can do some hackery with aget and window

11:43 matt444: @dnolen using cloact right now but plan on trying om

11:43 hyPiRion: heh

11:44 TimMc: Alright, we have a decent challenge then :p

11:44 TimMc: I think I can do it all with just "window".

11:45 dnolen: matt444: nice Cloact looks pretty cool.

11:46 TimMc: Hmm, maybe not. I can make "eval", but I can't just grab it off of window as easily as I could a map in CLJ.

11:49 ro_st: cemerick: thanks for all your awesome work on cljsbuild and cljx. can you tell me if it's possible to using both plugins' hooks in a project to get the auto build mode for each of them? or do i have to `lein cljx auto` and `lein cljsbuild auto` separately?

11:50 i'm hoping i can somehow get a repl and both auto modes in a single jvm

11:53 Uruk: What am I doing wrong here? I have a java array; (alength my-array) returns 28. I'm doing this: (map my-array (defn blah [x] (do-some-stuff-with x))) -- and the inner function blah is not getting called.

11:53 klrr: are there any intro to clojure for people new to java?

11:54 ro_st: you need to realise the result; map is lazy. wrap the (map) in (doall)

11:54 also

11:54 chronno: Uruk: also, defn is intended to be used as a top-level form

11:54 ro_st: instead of defn, just fn

11:55 defn also places a var in the current ns

11:55 Uruk: klrr: try this -- http://en.wikibooks.org/wiki/Clojure_Programming/Tutorials_and_Tips#Clojure_for_Java_Programmers

11:55 ro_st: so (fn [x]) rather than (defn blah [x])

11:55 Uruk: ro_st: so fn is same form as defn, but without method name (anonymous fn)?

11:55 ro_st: correct

11:55 Uruk: gotcha. Old hat to the concepts, completly new to clojure syntax

11:55 ro_st: defn is actuall just sugar around (def blah (fn …))

11:56 Uruk: so but the syntax is (map array function) and *NOT* (map function array) correct?

11:56 ro_st: http://www.braveclojure.com/ http://aphyr.com/tags/Clojure-from-the-ground-up

11:56 Wild_Cat: just checking, is there a difference between (defn- foo ...) and (defn -foo ...) ?

11:56 ro_st: that's right Uruk

11:56 Wild_Cat: defn- marks the fn as ^:private metadata

11:56 -foo is just a var name starting with -

11:57 klrr: so there are no resources for people who want to learn clojure and utiliaze java interop but with no previous java knowledge?

11:57 ro_st: conventionally used in 'internal' protocols and such

11:57 Wild_Cat: ro_st: OK, so what's the purpose of the latter? I've seen it a couple times, but can't grasp what it's used for?

11:57 ro_st: klrr: java interop is pretty minimal. clojure.org/java_interop

11:58 klrr: do you need to know java in-depth to utiliaze it?

11:58 ro_st: no java knowledge: there must be loads of info out there :-) it's a fairly popular language

11:58 i'm ok and i've never written any java. lots of other similar things, but not java itself

11:58 klrr: okey

11:59 nDuff: Wild_Cat: the cases you've seen it probably have been situations where gen-class is used with its default settings

11:59 ro_st: pick one of the links above and go through it. you'll learn enough to get moving

11:59 klrr: do you need to know OOP to utilize it is it just to call functions from java?

11:59 Wild_Cat: nDuff: probably, yeah

11:59 nDuff: Wild_Cat: ...where the prefix it's expecting for functions exposed to Java is -

11:59 ro_st: gotta run

12:00 Wild_Cat: nDuff: ah, I see. Thanks.

12:00 Uruk: Clojure java interop -- good documentation: http://clojure.org/java_interop

12:12 mikerod: Hmm, looks like the latest counter clockwise 0.22.0 stable doesn't upgrade in (my) Eclipse; with error concerning: Missing requirement: Counterclockwise Fragment for Eclipse 4 0.22.0 (ccw.core.e4 0.22.0) requires 'bundle org.eclipse.e4.core.contexts 1.3.1' but it could not be found Cannot satisfy dependency:

12:13 anyone familiar with that one?

12:13 clgv: mikerod: upgrade worked just fine this morning

12:15 mikerod: clgv: What Eclipse version? I am dealing with version 4.2.1 here I believe

12:16 I came across a similar error on 0.21.0 @ http://marketplace.eclipse.org/content/error/report/1414545

12:16 clgv: from 0.20

12:17 I upgraded on two computers so far with no problems from 0.20.0

12:20 mikerod: I'm coming form 0.12.3 stable

12:20 I don't see where this version fits in the pattern

12:20 clgv: mikerod: oh. maybe thats the reason

12:21 mikerod: does this need to be deleted perhaps?

12:21 clgv: clean install, or try to install 0.20.0 before 0.22.0

12:21 mikerod: hmm ok

12:21 thanks for advice, I'll try it out

12:21 clgv: I'd go for "clean eclipse install"

12:21 there were changes on the update site which might be the reason for your problem

12:22 mikerod: ok, that's what I am planning on trying

12:22 I noticed the site seems to have changed

12:22 clgv: yes indeed

12:23 good luck (should work) I gotta go

12:23 mikerod: cool, thanks

12:25 zerokarmaleft: what's with the inconsistency URL-parsing between java.sql.DriverManager/getConnection and clojure.java.jdbc/with-connection?

12:28 seems like I have to dance b/w strings...e.g. "jdbc:postgresql:<subname>" vs "jdbc:postgres://<host>/<subname>"

12:32 turbopape: Hi guys,

12:33 I have a project that litterally makes heavy use of java interop,

12:33 any bad or good vibes you mind sharing in doing projects that basically are like writing java in clojure ?

12:34 technomancy: turbopape: if you can isolate all the interop to a handful of namespaces it makes things a lot more tidy

12:34 possibly not always practical, but it's more pleasant

12:35 turbopape: okay, but can we fairly say, that this is not where clojure shines technomancy ?

12:35 I mean, all of this importing and implicit casting,...

12:36 Do I pay a huge price for that ?

12:36 technomancy: it's still a hell of a lot nicer than doing it in Java if that's what you mean

12:37 gfredericks: turbopape: more opportunity for refactoring via macros compared to java

12:37 turbopape: technomancy I am sure, that's why I don't want to leave Clojure world :)

12:38 gfredericks: areduce is a good example of using macros to sane-ify low-level things

12:38 technomancy: I guess the question is whether it's practical to do the bulk of your work in terms of immutable values

12:39 turbopape: Yes technomancy cc gfredericks .

12:39 It is basically a mahout project

12:39 it uses their arrays, etc...

12:40 And I'm afraid of the shift, so... besides, after googling, you don't see much people do mahout in clojure

12:40 but you find couple using mahout via scala, for instance,

12:40 so that's why I asked !

12:40 llasram: turbopape: I'm using it some. Mostly the vector and matrix impl

12:41 turbopape: llasram and are you okay with the shift code-wise and performance-wise ?

12:41 llasram: turbopape: I've got a 95% finished core.matrix implementation in terms of Mahout, but honestly mostly am using the Mahout API fairly directly, because startup time for core.matrix is a *beast*

12:42 turbopape: Yeah, it isn't too bad. I've got a small wrapper lib for doing a few of the most-common operations w/o hinting and letting you use IFns and Mahout Double/DoubleFunctions

12:43 I mostly do mutation w/in `reduce` loops, so I'm getting the benefits of mutable vectors, but the mutation is hidden across interface boundaries

12:43 turbopape: How heavy java projects compare, say, to scala , llasram ? cc technomancy ?

12:44 llasram: I'm afraid I don't understand your question :-)

12:44 technomancy: eh I've never actually written java; no idea

12:44 rurumate: turbopape: I don't know what mahout does exactly, but I've made friends with cascalog, which is pure clojure

12:45 maybe it's worth a look for you

12:45 llasram: rurumate: Er. It's built on Cascading, which shows through if you work with it a lot / deeply. For something much closer to "pure Clojure on Hadoop" check out https://github.com/damballa/parkour/

12:46 rurumate: hmm that's news

12:46 turbopape: llasram, technomancy rurumate : it's okay guys :) I need mahout for the machine learning algorithms, hadoop is just a backend there...

12:47 llasram: turbopape: For that it can vary quite a bit. Some things like the SSVD implementation are pretty flexible -- get your data in the right shape and you can essentially just invoke it

12:47 rurumate: llasram: methinks it's a "good thing" being based on cascading

12:48 turbopape: I ask about scala as a fall back option : if it is a better java, and if I unfortunately need to write some raw java, just to benefit from "simpler interop", can something like scala save me ?

12:48 llasram: turbopape: For the classification and clustering algorithms, you end up needing to descend into a deep morass of partially-documented interfaces

12:48 turbopape: and can I interop with Clojure ?

12:48 llasram, ok. any pure Clojure options you heard about ?

12:48 for ML ?

12:49 llasram: rurumate: Depends on what you're doing. If you have a lot of joins, it helps. If not, in my experience it is a liability

12:49 dnolen: I don't suppose anyone's done a CSS grammar for Instaparse?

12:50 llasram: turbopape: It really depends on your needs. Weka has reasonable implementations of some algos if your data isn't too large. There's special-purpose libraries for doing SVMs like libsvm. Clojure-specific, I'm not aware of much

12:51 rurumate: llasram: I do some joins, yes. Have you tried a join in parkour?

12:51 llasram: But ML algorithms are just algorithms -- rolling your own is always an option

12:52 turbopape: yeah llasram ... that would also be an option...

12:52 llasram: rurumate: It depends on the join. Doing a "replicated" join is trivial just by directly loading your replicated data from the distributed cache. Doing a reduce-side join where you know one or both sides will fit in RAM is simple.

12:53 Parkour has no direct support for Cascading-style reduce-side joins with arbitrary numbers of records on both sides, but IMHO those are a huge anti-pattern, and you should re-write any job depending on them

12:55 rurumate: llasram: well in the joins I do, usually both inputs are over 10 GB so would not try to fit any of them in memory / distributed cache

12:55 but access to distibuted cache is a plus for parkour, yes

12:55 llasram: rurumate: For the reduce-side "fit in RAM" I meant for an individual grouping key

12:56 rurumate: oh. well then, got any example code?

12:56 llasram: For joins? Not yet, but I should add one to the Parkour examples

12:56 riley526: For somebody familiar with Backbone for typical JS applications, is it worth trying to shoehorn Backbone in to ClojureScript for the REST-y models? Better alternatives?

12:56 rurumate: yes please

12:57 joins are super useful

12:57 they are what drove me from clojure-hadoop to cascalog

13:03 bbloom: riley526: i'd definitely not use backbone w/ cljs, since what backbone does is so hilariously simple you can just recreate it yourself in a few dozen lines using our proper immutable data structures

13:03 dnolen: riley526: there's going to be significant impedance mismatch

13:04 riley526: integration with React and Angular.js seems to involve less friction

13:05 riley526: dnolen: I've never used Angular though. You think the data model meshes with cljs?

13:05 dnolen: riley526: I can't say myself, but people seem to like using it with CLJS

13:05 katox: I'd recommend to start with something like react ;)

13:06 riley526: dnolen: Ok I'll look into it

13:06 bbloom, dnolen: thanks guys

13:06 katox: The question is really about syncing models via REST, which Backbone does and React has nothing to do with.

13:07 katox: riley526: angular is a mutation in a box, I can't image how to make that simple

13:07 dnolen: riley526: but routing isn't a particularly complex thing, there's an opportunity to construct something generic here for the ClojureScript community

13:08 katox: riley526: you need to put everything into angular scope and broadcast + watch everything

13:09 riley526: the mismatch is right in the core

13:09 riley526: katox: yeah I hear you

13:09 katox: riley526: that doesn't mean that angular isn't _very_ productive in js itself

13:09 dnolen: riley526: I personally prefer React myself, but I don't know Angular.js well enough to have an informed opinion - from the outside seems overly complex

13:11 katox: it is complex once you dive into directives

13:12 I've never wrapped my head around when I'm inside and when I'm outside of angular processing

13:12 riley526: dnolen: how do you typically handle syncing data if all you're using is React? or do you even have that need?

13:13 cespare: Is visualvm the state of the art for clojure profiling?

13:13 riley526: I mean I could always right an ajax request for every endpoint

13:13 But that's been solved before for JS.

13:17 nDuff: cespare: YourKit is also pretty awesome.

13:18 dnolen: riley526: I would probably write a simple router

13:19 nDuff: cespare: ...unfortunately, they only occasionally offer personal licenses (at less than full price) -- they had that running for a bit this month, but it looks like the window closed.

13:19 cespare: nDuff: would it supplant or supplement visualvm?

13:20 nDuff: ...waitaminute, today's the 13th, so that should still be open.

13:20 cespare: full replacement.

13:20 cespare: nDuff: s'ok, i'm sure my company will buy it if it's useful

13:21 Wild_Cat: dakrone: just submitted a clj-http pull request for the {:as :json-strict} we talked about the other day.

13:21 dakrone: Wild_Cat: great!

13:22 cespare: nDuff: can it operate from the commandline? Or at least collect a profile that way? Right now I'm using visualvm over ssh -X and it's quite painful

13:22 Wild_Cat: dakrone: this is my first non-basic-typo pull request to a Clojure project, so be sure to review it thoroughly before merging :p

13:22 dakrone: Wild_Cat: will do

13:22 nDuff: cespare: yes, collection can happen via CLI only, or over a socket.

13:22 cespare: ...I'm surprised you can't do that with VisualVM too.

13:24 cespare: nDuff: alright, thanks for the suggestion. I'll check it out.

13:24 katox: does anyone know what's the current state of cider regarding cljs?

13:24 dnolen: katox: seems like people are working on improving it

13:26 cespare: nDuff: are you aware of a way to generate a blocking profile?

13:26 the profilers i've found generate cpu and memory profiles only

13:26 katox: dnolen: good to hear, what'll be next? code completion or even evals?

13:27 dnolen: katox: gtrak is working on code completion, evals would be really simple

13:27 katox: dnolen: awesome!

13:28 cemerick: katox: evals? Like, a REPL?

13:28 dnolen: katox: the ClojureScript analyzer exposes a lot of information, cider support could be awesome

13:28 katox: cemerick: I got REPL working, thank to your austin

13:29 cemerick: katox: ok, good. What do you mean by 'evals', then?

13:29 katox: cemerick: buffer-to-repl connection

13:30 cemerick: I can eval stuff from clj into the repl, not the same from cljs

13:30 nDuff: cespare: "blocking" meaning lock contention?

13:31 cespare: nDuff: yes

13:31 nDuff: cespare: intrinsic locks it handles. Not sure about higher-level primitives.

13:33 katox: another super cool thing would be inspectable repl results

13:34 light table has this gratis because of chrome developer tools

13:34 but I'm on debian so I'll have to wait for 0.6 a bit to try it out

13:36 cemerick: katox: still not sure what you mean; you should be able to evaluate files, top-levels, and individual expressions directly into ClojureScript from cider

13:36 worksforme, anywa

13:37 katox: cemerick: ok, then I have my setup broken in some ways

13:37 cemerick: katox: Either that, or something has changed in cider (I'm still using nrepl.el, and C-M-x and C-c C-k both work well)

13:39 katox: cemerick: not sure, but cider is trying to send the eval to clj not cljs repl, so it bombs out

13:40 cemerick: the repl itself is fine though I can get new stuff into it via cljsbuild auto

13:45 cemerick: katox: yeah, that's broken somehow

13:45 I'll bet cider is wrapping the expressions to be evaluated with its own (Clojure-only) code. Filing a bug would be appropriate.

13:47 katox: cemerick: yeah, it could be something like that... I need to update to latest cider before filing a bug report though ;)

13:47 cemerick: did you try with the latest cider or was that a guess?

13:48 cemerick: katox: total guess; sorry, I (foolishly) assumed you were already running the latest

13:49 katox: cemerick: I'm running some very recent, but not the very latest

13:57 cemerick: katox: ok; well, I'll be rebuilding my environment later this month, so cider will be fully cljs / Austin / piggieback-capable one way or the other. :-)

14:06 katox: cemerick: okay, I updated and it certainly did something

14:07 cemerick: I got one command through (defn eval) but now I'm looking at frozen emacs ;)

14:07 cemerick: but at least it can differentiate between clj and cljs buffer

14:10 loliveira: Can I use http-kit without calling httpkit/run-server? I am using lein ring server-headless to start the server and I believe I can't change it.

14:10 technomancy: uuuuuuse lein-run with a -maaaaaaaaain

14:12 shep-werk: technomancy: instructions not clear; stuck in ceiling

14:12 loliveira: technomancy: I *need* to create a war file. I can't run "lein-run" in our production. =(

14:13 katox: cemerick: but it seems to be (sort of) working, good, certainly not worse than before

14:13 technomancy: loliveira: oh, bummer =(

14:14 loliveira: =(

14:15 technomancy: so.. no http-kit inside a war file?

14:15 technomancy: sorry, I've never used http-kit or war files

14:17 wink: WAR! Hoo! What is it good for? YAY! Absolutely.. nothing!

14:17 katox: dnolen: what's the intent of saving state on every change in wrap-form-element in om?

14:17 squidz: how can I apply a funciton with variable arguments to a number of vectors and that returns a vector of each corresponding result? Example: (+ args) with [[1 2 3] [1 2 3] [1 2 3]] ===>>> [3 6 9]

14:18 loliveira: technomancy: thank you.

14:18 dnolen: katox: our requestAnimationFrame model doesn't play well with form elements, this is just a hack to fix it

14:18 katox: "doesn't play well with *React* form elements"

14:19 squidz: nvm I found you just pass map to apply as a second argument

14:20 katox: dnolen: I though that it's there just to force a refresh if there is no change on app-state atom

14:20 dnolen: katox: no

14:21 loliveira: wink: We are java software house. They decided that all apps must be delivered as war file. Corp stuff. =/

14:22 wink: loliveira: https://www.youtube.com/watch?v=r-bA9FYB8HY#t=0m55s in case oyu missed the joke ;)

14:22 loliveira: wink: =)

14:22 tbaldridge: "You-all"

14:25 loliveira: wink: thank you. I believe the reason for such restriction is that war files are "web scale". =) http://www.youtube.com/watch?v=b2F-DItXtZs

14:25 wink: lol

14:26 egghead: lol war file deployments

14:37 justin_smith: loliveira seems to be gone, but you don't need a war file on production, an uberjar plus java -jar works just fine

14:37 and yes, lein on production should not be done

14:37 TimMc: I've got the build gremlins from hell over here.

14:38 When I run against a dependency pulled from a jar repository, I get one behavior, but when I lein install the same git tag I get different behavior.

14:38 I compared the JARs and everything is identical except for some minor stuff in pom.xml and manifest.mf.

14:39 ipostelnik: can someone explain the difference between (nil?) function and #(= nil %)

14:40 joegallo: the difference is a few characters of typing

14:40 Wild_Cat: ipostelnik: they do the same thing

14:40 (nil?) is nicer because it tells whoever's reading your code your intent much more succintly.

14:41 ipostelnik: (nil?) calls (clojure.lang.Util/identical x nil)

14:41 TimMc: I've confirmed that I'm not running with lein checkouts, and I'm trying again with a fresh m2 and source checkout -- what else should I check?

14:41 ipostelnik: and it's 5-7x slower on my machine

14:41 TimMc: ipostelnik: According to what test?

14:42 rurumate: not sure if in clojurescript, (= nil (aget x "foo")) (nil? (aget x "foo")) would be the same if x does not have property x

14:42 property foo I mean

14:42 dobry-den: noHistory is used in datomic for any facts that have no semantic of accreting, right?

14:42 technomancy: TimMc: slurp the .clj file's resource and diff that

14:42 ipostelnik: TimMc, according to my unscientific test on my machine

14:42 TimMc: ipostelnik: Try with criterium

14:42 gfredericks: I'm trying right now

14:43 TimMc: technomancy: As in, find a file that *could* have changed and make sure the REPL sees the same thign in both cases?

14:43 technomancy: right

14:44 if you can isolate it to one ns

14:45 TimMc: technomancy: I can't, but I'll give it a shot.

14:45 gfredericks: 3.8 ns for nil? and 5.2 ns for (= nil x)

14:46 nDuff: hrm.

14:49 ipostelnik: gfredericks, that makes sense reading the code

14:52 dnolen: rurumate: the same

14:53 TimMc: technomancy: I cloned a new repo, I temporarily moved m2, could not reproduce it -- and then I switched everything back and *it is working now*.

14:53 Maybe I should remove [lein-gremlins "0.0.1-SNAPSHOT"] from my profiles.clj.

14:54 technomancy: yeah that one has a dependency on [lein-hallucinations "LATEST"] which you really don't want

14:55 TimMc: Note that I didn't even run 'lein clean' at any point in this process.

14:56 stuartsierra: dobry-den: Not exactly sure what you mean. noHistory is usually for things like counters where you only care about the latest value.

15:10 ro_st: cemerick: that was me about cljx + cljsbuild on twitter. thanks for the responses. if you don't use lein-cljsbuild, how do you use cljsbuild ? via some nrepl middleware?

15:10 sw1nn: hi, anyone have opinions about the oldest version of clojure a library could/should legitimately depend on? I'm looking to update clj-aws-s3 and it's currently referencing 1.2.1? One of the deps [clj-time "0.5.0"] depends on 1.5.1 tho.

15:11 ro_st: austin is great but you don't want to manually evaluate all your code into the browser with it when resuming work on a project

15:11 just curious what your workflow is, because i really don't want to use `lein blah auto` if i can automate things better than that

15:13 technomancy: sw1nn: 1.2 is old enough that you could see subtle weirdness around dynamic vars that it's probably not worth bothering with; no one's on anything older than 1.3 these days

15:13 mrhanky: how do i write this in clojurescript? if(self.document === undefined)

15:13 technomancy: or anyone still on 1.2 is already going to expect nothing to work with their code

15:13 mrhanky: what's "===" and "undefined" in cljs?

15:14 sw1nn: technomancy: that was my thoughts, 1.3 + profiles for 1.4 and 1.5. sound right?

15:15 technomancy: sw1nn: sounds good to me

15:15 rurumate: mrhanky: I think there's no way to write or obtain undefined

15:15 cemerick: ro_st: generally only for automated testing, and build prior to deployment; piggieback handles all REPL use cases.

15:15 noonian: i think nil is equal to undefined in cljs but not sure

15:15 rurumate: aget will never return undefined, it will convert it to nil for you

15:16 gfredericks: nil is null I think

15:16 ro_st: cemerick: just so i'm clear, you start your repl up, start austin up, and then evaluate your top-level cljs ns and that gets everything loaded into the browser

15:16 bbloom: mrhanky: === is clojure.core/identical?

15:16 gfredericks: last time I checked which was likea year ago

15:16 rurumate: (.-x y) is probably the same as (aget x

15:16 logic_prog: is there a more efficient way of doing (into #{} (filter ... set)) ?

15:16 mrhanky: i'll try (nil? self/document)

15:16 bbloom: rurumate: mrhanky: you can get undefined with js/undefined

15:17 logic_prog: i.e. I want to filter on the set, but right now, it's set -> list -> filter -> list -> set

15:17 gfredericks: logic_prog: reduce

15:17 logic_prog: whereas, I'd prefer set -> filter -> set

15:17 noonian: yeah, but if you have a value returned from some js call that is undefined, i bet (= undefined-obj nil) would return true or (nil? undefined-obj)

15:17 rurumate: bbloom: oh

15:17 gfredericks: logic_prog: or apply disj I guess

15:17 logic_prog: how does reduce help?

15:17 gfredericks: (apply disj set (remove ... set))

15:17 or

15:17 logic_prog: is this known to be faster

15:17 or we are premature optimizing ?

15:17 mrhanky: both are working, (nil? self/document) and (identical? self/document js/undefined)

15:18 thx

15:18 gfredericks: logic_prog: it was your question :P I'm sure it's gota be assymptotically faster; dunno about small sets

15:18 noonian: what is self?

15:18 bbloom: mrhanky: rurumate: undefined is kinda a leaky interop thing. it gets coerced to nil in many cases

15:18 gfredericks: logic_prog: well asymptotically better on GC at least

15:18 mrhanky: noonian, self = this in this case

15:19 dsrx: does anyone know of any idiomatic-ish cljs bindings for webgl? or if not, is three.js relatively painless to use from cljs?

15:19 gfredericks: logic_prog: if you want to get really premature, use a transient since disj doesn't do that for you

15:19 logic_prog: lol

15:19 screw this

15:20 let's write jvm bite code

15:20 and hack fht ejit to make sure this hotpath is optimized for

15:20 mrhanky: just tried to find out if my code is really running as webworker noonian

15:20 logic_prog: gfredericks: this was very useful

15:20 I'm going to stop worrying about this micro optimization until it becomes a bottleneck

15:25 dsrx: I tell myself that all the time and it never works :)

15:25 dobry-den: I have a dynamic Compojure app running on Ubuntu/Nginx. What kind of solutions should I start looking at if I want to get a better picture of what's actually going on? For instance, if someone is brute-forcing my /login path, what's the sort of solution people use to visualize that kind of stuff

15:26 I guess it would start with nginx's access logs

15:26 gfredericks: logic_prog: I usually spring for things that at least honor how data structures were intended to be used

15:43 dnolen: dsrx: https://github.com/Asakeron/cljs-webgl ?

15:43 dsrx: Three.js could probably work too.

15:45 dsrx: oh, as it turns out I had accidentally blocked github.com with the google personal blocklist extension

15:46 dnolen: thanks!

15:50 piranha: dnolen: I'm a bit confused concerning update!/transact! in om. What's the right way to update nested map? (transact! cursor [:my :path] assoc :q) or (update! cursor assoc-in [:my :path] :q)? I'm just unsure what transact! is.

15:51 dnolen: piranha: transact! and update! are conceptually the same, just different signature

15:51 piranha: ok, I was somehow confused by transact's name

15:51 so I guess transact is intended to be used here, in other case it just wouldn't exist? :)

15:51 dnolen: piranha: transact! is conceptually the important one

15:52 piranha: update! is just a sugar for the case where you don't need to provide a path

15:52 piranha: I see

15:52 dnolen: thanks for clarification!

15:52 dnolen: piranha: transact! name was intended to give a clue, you're changing the global application state, not component local state

15:53 piranha: it's a bit important as the global application state may have changed by the time an event handler fires

15:53 piranha: so in some cases you need to recheck your assumptions

15:54 piranha: makes sense...

15:56 sveri: hi, i just had a look at the repeat function in clojure and i dont get how this works, does someone know a blog post or something that explains repeat? why does it call itself? why doesnt it use recur?

15:58 hyPiRion: sveri: It has to do with laziness. I'll see if I can find a blogpost with the topic

15:58 nDuff: sveri: look into what lazy-seq does.

15:58 `cbp: sveri: it doesn't use recur because recur must be in tail position

16:00 sveri: hyPiRion: that would be nice

16:00 noonian: basically, you can write recursive functions that return lazy seqs without blowing the stack if you wrap your fn in (lazy-seq ...); it's like magic on crack

16:00 sveri: noonian: ok, that explains it a little bit

16:06 gfredericks: sveri: a lot of lazy-seq-producing functions are structured like that

16:06 hyPiRion: sveri: Yeah, I have some issues finding a good explanation of it, but it's essentially like this: lazy-seq is actually a macro, which makes a LazySeq object containing an anonymous function with the the body. If you try to do recur inside that anonymous function, you won't call the function you intended to cal

16:06 *call

16:07 gfredericks: that's half; the other half is why it doesn't blow the stack when you _don't_ user recur

16:07 hyPiRion: ,(macroexpand-1 '(lazy-seq (this is my body)))

16:07 clojurebot: (new clojure.lang.LazySeq (fn* [] (this is my body)))

16:10 gfredericks: ,(format "foo" 42)

16:10 clojurebot: "foo"

16:10 gfredericks: ,(format "foo %s")

16:10 clojurebot: #<MissingFormatArgumentException java.util.MissingFormatArgumentException: Format specifier 's'>

16:10 sveri: ok, i think i get it, at least enough to use it

16:10 thank you all very much :-)

16:24 bmath: ,(+ 1 2)

16:24 clojurebot: 3

16:25 bmath: ,"clojurebut does what I probably think it does"

16:25 clojurebot: "clojurebut does what I probably think it does"

16:32 matt444: Does anyone have a minimal http script that just serves static files?

16:33 nm, there's one in the guides

16:33 technomancy: `python -m SimpleHTTPServer 3991`

16:34 TimMc: clojurebot: Are you just gonna take that?

16:34 clojurebot: No entiendo

16:41 gfredericks: (inc clojurebut)

16:41 lazybot: ⇒ 1

16:43 justin_smith: (inc clojurebutt)

16:43 lazybot: ⇒ 1

16:43 gfredericks: but...could we do it with 3 t's???

16:43 lazybot: gfredericks: Yes, 100% for sure.

16:43 gfredericks: lazybot: I'm sure you're enjoying this game

16:44 justin_smith: *lazybutt

17:04 bitemyapp: akurilin: http://www.meetup.com/SF-Types-Theorems-and-Programming-Languages/events/155960462/

17:04 justin_smith: excellent name.

17:05 slpsys: china basin? too far

17:06 bitemyapp: slpsys: that isn't China Basin.

17:06 slpsys: it's SOMA

17:07 slpsys: is china basin just across the channel? i've actually been unclear on that

17:07 bitemyapp: slpsys: yes.

17:07 slpsys: 164 Townsend is solidly in SOMA.

17:08 slpsys: having checked the map, all but the NE tip of townsend is in SOMA.

17:08 slpsys: how long have you lived here?

17:09 slpsys: haha. five years, but i've only worked in the city for about 7 months, and i didn't really think about what to call that neighborhood when i was living in the city

17:27 Node_347: Would someone with experience in Ragtime and Korma mind taking a look at this gist to see if I'm performing migrations correctly? https://gist.github.com/arlandism/8408370

17:28 mrhanky: anybody here used servant library?

17:30 bitemyapp: Node_347: tried it with something that isn't h2?

17:33 Node_347: <bitemyapp> No, I haven't

17:34 daydreamt: Damn you encog.

17:35 Is anyone using neural networks in clojure?

17:35 bitemyapp: Node_347: well.

17:36 mrhanky: (:require [foo.bar :as bar] [foo.baz :as baz] [foo.bat :as bat])

17:37 how do i write this shorter? something like [foo [bar :as bar] [baz :as baz] [bat :as bat]] ?

17:38 justin_smith: that is valid, but not many people use that syntax

17:38 bitemyapp: Node_347: that's cool, I can tell you the same thing on the mailing list.

17:38 Node_347: if you check the mailing list, you'll see I've suggested you try using something other than H2 to see if there's still a repro.

17:38 stuartsierra: mrhanky: That feature exists as "prefix lists" — see the docs for `require` — but I find it hard to read and grep for.

17:39 mrhanky: good point with grep stuartsierra

17:40 Node_347: bitemyapp: Thanks!

17:40 rasmusto: wtb gsexpp

17:42 AWizzArd: daydreamt: yes, I do.

17:42 technomancy: mrhanky: check out slamhound for stuff like that

17:43 daydreamt: What libraries do you use AWizzArd? Everything I looked at, I looked at pybrain, then I looked back, then I cried.

17:43 AWizzArd: daydreamt: same here, so I implemented it myself :-)

17:43 I have implemented an MLP and an RBM in Clojure which runs on nvidia’s CUDA.

17:44 ruzu: what about NFL or NBA

17:44 daydreamt: AWizzArd how easy would it to extend it to create arbitrary structures?

17:44 I wanted to do RNNs.

17:45 AWizzArd is it opensource?

17:45 AWizzArd: Well, probably this year I will add an LSTM.

17:45 Not yet open, but not much speak against opening it.

17:45 rasmusto: CUDA MLP, RBM w/ LSTM?

17:46 wikipedia disambiguation isn't helping me much :<

17:46 daydreamt: rasmusto they are types of neural networks

17:46 AWizzArd: rasmusto: cuda means it runs my code on the gpu. An MLP is a Multilayer Perceptron network, and an RBM is a restricted Boltzmann machine.

17:46 technomancy: you get more ROI with an NIH TLA tohugh

17:46 rasmusto: gotcha, didn't realize this was part of the neural network convo still :D

17:47 AWizzArd: daydreamt: which rnn are you interested in?

17:47 rasmusto: I tried to do a game of life on cuda once, with limited success

17:47 AWizzArd: And how you train it? :-)

17:48 bitemyapp: I don't think clj would be my first pick for CUDA stuff.

17:48 * nDuff is vaguely considering playing with neural networks for echo cancellation (tricky part of which is prediction), and perks up ears.

17:48 daydreamt: AWizzArd, probably backprop

17:49 AWizzArd, not sure about the structure. That's why I wanted the library to be extendable

17:49 bmath: PCMCIA

17:49 AWizzArd: Well, total flexibility doesn't mix with gpus.

17:49 ruzu: MLB

17:49 AWizzArd: For example, all neurons use the same propagation function, which is just the weighted sum for example.

17:50 This way everything can be expressed in Matrix operations and run on the GPU with massive speed.

17:50 daydreamt: The fewer types of operations the better?

17:50 AWizzArd: It can be many operations, but in principle every neuron needs to be treated in the same way.

17:51 In our brains each neuron has its own propagation and activation function.

17:51 But that beast also has a hundred billion processors <3

17:51 vs 2k on gpu ;-)

17:51 TimMc: brb, mining bitcoin on brains

17:51 AWizzArd: haha

17:52 rasmusto: yeah, lots of neurons are wasted thinking about lunch in a real human brain, its not apples to apples

17:52 daydreamt: TimMc: whose brains?

17:52 TimMC: You just opened the door to a very dystopian future

17:52 jjido: TimMc: always the pragmatic :)

17:52 AWizzArd: daydreamt: do you think backprop would work on RNNs?

17:53 ruzu: won't be long before the nsa will require access to all of humanity's brains "because terrorism"

17:53 AWizzArd: ruzu: you are aware that your words were logged, yes?

17:53 bitemyapp: TimMc: Da Matrix.

17:54 daydreamt: AWizzArd, I'm just getting into RNNs, but the paper I read, I could swear they used some form of backprop.

17:54 ruzu: should i be concerned?

17:55 AWizzArd: Well, it’s planned to have such. I have one candidate that seems it can be trained.

17:55 daydreamt: read Kurzweil’s book “How to create a mind”.

17:56 bitemyapp: people take Kurzweil seriously?

17:56 AWizzArd: This book is actually pretty good.

17:56 bitemyapp: that's like asking a scientologist for advice on your psych medication.

17:56 ruzu: i'm notifying the scientology administration of your slander.

17:57 daydreamt: AWizzArd: they do use backprop. "Backpropagation through time" or something like that

17:57 bitemyapp: ruzu: spookay.

17:57 logic_prog: what is the clojure.core.async stance on ">!" can't be used across functions (i.e. has to be within a lexical async/go block) ? Is the stance: (1) this is the way we like it and it will always be this way , (2) this is due to sucikness of the JVM and we will fix it , or (3) none of the above ?

17:58 AWizzArd: bitemyapp: just have a look at what guys wrote reviews to his book. Some top brain researchers. And even more important: I say it’s good \o/

17:59 bitemyapp: AWizzArd: well, if you cherry-pick the right subset, you can find climatologists that deny global warming too

17:59 AWizzArd: I used to work alongside and live with a neuroscientist, we're a long ways away but making steady progress. Kurzweil is not the person you want status updates from, he's an enthusiast, not an authority.

18:00 AWizzArd: This however is his bestest book.

18:00 Nothing new for a neuroscientist though.

18:00 bitemyapp: AWizzArd: I liked L Ron Hubbard's Battlefield Earth the most, out of them all.

18:00 you all should definitely read that one, since it was his bestest book.

18:03 Always nice to see my beautiful compact little handler code get destroyed and sullied by the arrival of reality.

18:03 dnolen: logic_prog: shallow yield is a good solution that sits well with Clojure, any more requires far, far more compiler support

18:03 "well with Clojure's compilation model"

18:04 logic_prog: dnolen: https://www.google.com/#q=clojure+shallow+yield is not very useful

18:04 dnolen: can you point me at api docs to read ?

18:04 dnolen: logic_prog: http://www.cs.indiana.edu/~sabry/papers/yield.pdf

18:05 turbofail: if you want to get around that you'll need delimited continuation support

18:05 logic_prog: dnolen: does this mean "use https://github.com/swannodette/delimc " or is there another library you have in mind?

18:05 dnolen: logic_prog: no

18:06 logic_prog: core.async gives you a weaker form of delimited continuations that's good enough for most problems w/o complicating the compilation model

18:06 logic_prog: dnolen: let me rephase my question. how do I use shallow yield in clojure? what are good examples of shallow yifled in clojure?

18:06 dnolen: logic_prog: core.async gives you shallow yield

18:06 logic_prog: err, so the ansewr is "shallow yifled does NOT provide what was asked" but is the best clojure does ?

18:07 it's infuriating :-) that I can't put >! in functions unless I wrap them in go blocks ....

18:07 go blocks encourage me to write giant functions, whereas for all these years, I've been trying to get better at writing smaller, composable functions

18:07 dnolen: logic_prog: you break things about into smaller go blocks

18:08 s/about/apart

18:08 noonian__: logic_prog: can't you just call >! on a channel in the go block and return the value on a new channel then return the channel?

18:09 logic_prog: yeah, but then I end up putting <! everywhere

18:09 basically I have to sprinkled <! all over the palces, becuase functions on longer return values, but instead, return go channels that I have to read from

18:10 jjido: logic_prog: sounds like continuations

18:10 logic_prog: dnolen: what is the best continuations library for clojure right now?

18:10 dnolen: logic_prog: core.async

18:10 logic_prog: lol

18:11 bitemyapp: logic_prog: call/cc was a bad idea anyway.

18:11 dnolen: logic_prog: it sounds like you need to think about your problem domain more, or organize things at level higher then the abstraction core.async gives you out of the box.

18:11 jjido: bitemyapp: agreed

18:11 powrtoc: dnolen: does anyone know how core.logic LCons works? I know it implements cons from scheme, but how does it work?

18:11 turbofail: you wouldn't need full call/cc to provide a better yield, in fact it'd be easier to implement it using delimited continuations

18:11 dnolen: powrtoc: the implementation is not that long :)

18:12 bitemyapp: logic_prog: there are better, less opaque, more reliable abstractions that can be composed in programs. Functions.

18:12 turbofail: that said, supporting even delimited continuations would be difficult

18:12 ruzu: anyone happen to have read Clojure Data Analysis Cookbook, and have an opinion on it?

18:12 jjido: turbofail: the trampoline works fine.

18:12 * arrdem starts the monad countdown timer

18:12 bitemyapp: arrdem: god fucking dammit I was typing it

18:12 powrtoc: dnolen: yes I've looked at it :-) I guess the likes of cache are confusing me...

18:12 turbofail: jjido: sure but you'd have to recompile all of clojure's standard library in that style

18:12 technomancy: arrdem: you jinxed it

18:13 bitemyapp: I was going to say, a better abstraction is continuations from monads. Like: http://hackage.haskell.org/package/mtl-1.1.0.2/docs/Control-Monad-Cont.html

18:13 turbofail: and you pay a performance cost for all code even if you only use delimitd continuations in a small portion

18:13 bitemyapp: Not least bececause it doesn't need to be reified to the compiler.

18:13 jjido: turbofail: it's slow yes

18:13 bitemyapp: But also because it's fairly simple to understand algebraically.

18:13 dnolen: powrtoc: cache is just to cache hashCode

18:14 turbofail: jjido: and eventually you realize that core.async is an acceptable performant tradeoff

18:14 powrtoc: ahh so that's why it's mutable

18:14 bitemyapp: having dnolen ignore me is proving advantageous so far.

18:14 dnolen: Erik Meijer said so, so it must be true :)

18:14 bitemyapp: I can teach without being interrupted, except by arrdem.

18:14 Erik Meijer is an apostate.

18:14 turbofail: anyway many other languages have decided that a shallow-yield is good enough, and i agree for the most part

18:15 jjido: bitemyapp: away with monads

18:15 bitemyapp: jjido: no

18:15 jjido: waah

18:15 technomancy: there we go

18:16 bitemyapp: Using monads to teach a concept doesn't fall under convincing others to use Haskell, they're the just the only ones with documentation for things like Cont.

18:16 not my fault >:)

18:17 technomancy: just saying; got my eye on you

18:17 * bitemyapp tap dances stage right

18:18 bitemyapp: technomancy: you use monads too. Async. :P

18:19 technomancy: I was kind of amazed that when they introduced it, they were like "(by the way, this is technically called a monad, but you don't have to care about that right now)"

18:19 I have never heard anyone introduce monads in a way that wasn't shouting from the rooftops about how amazing they are; it was quite unexpected.

18:19 * logic_prog is a secret Haskell agent, trying to get people to use Monads, by asking questions about things Core.async can't quite do.

18:19 technomancy: (in the ocaml book)

18:20 jjido: technomancy: nice closure

18:20 bitemyapp: technomancy: it's not as much about amazingness as that they're the natural next-step from the relatively trivial HOFs Clojure leverages.

18:21 you could hold your nose and pretend protocols were typeclasses to accomplish the same in Clojure.

18:21 not that you'd enjoy it.

18:21 ie, Fluokitten.

18:23 technomancy: you can do monads in OCaml but I haven't seen anybody mix different kinds of monads in a single function.

18:23 technomancy: I'm actually headed to a type theory meetup tonight where at least one or two people that worked at Jane Street will be present, I'll ask them about it.

18:24 technomancy: bitemyapp: mostly talking about people who blog about "I finally understand monads and now you can too"

18:24 bitemyapp: technomancy: every single one of those blogs I've seen have been unconscionably horrible.

18:24 technomancy: followed by incomprehensible rambling, right

18:24 jjido: Jane Street?

18:24 bitemyapp: Monads are too generic to be explained like that. It's better to move into practical implementation and the algebra directly.

18:24 technomancy: jjido: Jane Street Async, yeah

18:24 bitemyapp: jjido: Minsky's crew.

18:25 technomancy: "We don't have concurrency, but through the Power of Monads we can fake it!"

18:25 bitemyapp: that's pretty much it right there.

18:25 technomancy: functional inversion of control goes a long way :)

18:26 technomancy: it's much nicer than any other callback system in single-threaded runtimes I've used

18:26 mrhanky: are there any additional steps when installing jayq? i get this: Uncaught ReferenceError: jQuery is not defined

18:26 powrtoc: Wow core.logic has a datomic dependency? What's that for?

18:27 bitemyapp: technomancy: I'd tend to agree. I'd still rather have 1:1 and N:M threading though.

18:27 powrtoc: datalog?

18:28 powrtoc: bitemyapp: Why would core.logic need that?

18:28 arrdem: technomancy: the main reason that I lurk here isn't that I like clojure, it's that bitemyapp is an unending font of cheap shot targets

18:28 technomancy: the clojure is an added benifit

18:28 akurilin: bitemyapp: cool. I'm afraid I can't go tonight, demo day tomorrow, gotta wake up at 5 :|

18:29 bitemyapp: Btw I'd love to know more about the ACL library/pattern you mentioned on the mailing list.

18:29 technomancy: arrdem: yah, I haven't written clojure in months

18:29 akurilin: bitemyapp: Is there anywhere I could read up more about on it?

18:29 technomancy: I'm here for the lulz

18:30 arrdem: technomancy: can't blame you. this is a really sharp croud :D

18:32 dnolen: powrtoc: it doesn't have a datomic dep

18:35 powrtoc: dnolen: on master datomic-free is listed as a dependency in project.clj... just wondering why... am I right in thinking datomic.clj extends core.logic unification etc... to datomic types?

18:35 dnolen: powrtoc: "provided"

18:35 powrtoc: there's no hard dependency on datomic at all

18:36 powrtoc: the datomic namespace won't even compile if datomic isn't on your classpath

18:36 powrtoc: yes, it just an experimental namespace, Datomic still doesn't provide the right hooks for it to work

18:36 powrtoc: dnolen: ah cool... I think leiningen has moved on a bit since I last did Clojure in anger...

18:36 dnolen: interesting...

18:41 dnolen: is the nominal logic and finite domain stuff complete these days?

18:42 dnolen: powrtoc: more or less, there's more work to do w/ finite domain w/ respect to perf

18:43 powrtoc: dnolen: cool

18:44 technomancy: powrtoc: Central won't accept poms with dependencies on non-OSS code

18:44 unfortunately clojars is not as careful about that =\

18:44 powrtoc: technomancy: ahh

18:44 bitemyapp: technomancy: proper way to interact with a jar from Clojure code is lib/ right?

18:44 confirming for a coworker that apparently needs to fiddle with some NLP whatevers.

18:44 technomancy: bitemyapp: proper way to interact with a jar is from a repository

18:45 bitemyapp: technomancy: is lein-localrepo a good idea? We don't have our own maven repo.

18:45 technomancy: if it's just for local exploration you can use lein deploy to get a jar into ~/.m2 for your own machine

18:45 otherwise all it takes is a directory served over HTTP

18:46 bitemyapp: technomancy: it's a flat standalone jar provided by a third party.

18:47 technomancy: thanks!

18:48 technomancy: you can tell your co-worker it is totally OK to yell at anyone who thinks it's reasonable to distribute a jar that's not in a repository in 2014

18:48 tell them technomancy sent you

18:48 matt444: I'm following the ClojureScript getting started guide and when I attempt to compile I get this: http://pastebin.com/guqrk4Ue

18:48 technomancy: I mean if you're getting this from a 3rd-party

18:48 bitemyapp: technomancy: I did, actually. I said who you were in the email :)

18:48 technomancy: nice

18:50 powrtoc: Is alphaKanren described at all in reasoned schemer?

18:50 or is it only in the alpha kanren paper?

18:51 matt444: following these steps: https://gist.github.com/ninjudd/1098120

18:51 java 1.6.0_65

18:52 noonian__: matt444: what version of cljs are you trying to compile?

18:53 matt444: version of cljs?

18:53 i'm not compiling cljs

18:53 i'm compiling a .cljs file

18:53 https://github.com/clojure/clojurescript

18:53 master

18:54 noonian__: matt444: ah, from your gist you are using the cljs compiler directly and i've never tried that

18:54 matt444: then how do you compile?

18:55 noonian__: matt444: most people use the leiningen plugin lein-cljsbuild to compile cljs in a leiningen project

18:55 matt444: yes, i use that

18:55 but i have client side and node

18:55 so I use lein-cljsbuild for client side

18:55 noonian__: you can use it for both

18:56 matt444: hm, example project.clj?

18:57 just have a second :build

18:57 ?

18:58 so how do you arrange your folder?

18:58 my client stuff is in src/

18:58 noonian__: yeah, second build

18:58 something like this: https://www.refheap.com/23531

18:58 bitemyapp: "actual: java.lang.ClassCastException: datomic.db.Datum cannot be cast to clojure.lang.Associative" dATOMIC rage.

18:58 noonian__: just arrange your src directories accordingly

18:59 matt444: so if I move my files in src/ it's still going to be able to find the references?

18:59 where does your package.json go?

18:59 src/cljs/node ?

19:00 noonian__: i'm not sure what that is, i haven't really used node

19:00 if you want to share code then you should have 2 src-paths for each cljsbuild build where one is shared between the builds

19:00 bitemyapp: "java.lang.IllegalArgumentException: Don't know how to create ISeq from: datomic.db.Datum" I am going to kill something.

19:00 matt444: i don't want to share code

19:01 arrdem: bitemyapp: core.typed talk in an hour.. let you know how it goes.

19:01 noonian__: i mean, share code between both your builds, like share a common set of cljs utilities for instance

19:02 matt444: yeah, i don't need that

19:02 bitemyapp: arrdem: I am atomic-raging at some type error / stupid data-type stuff in Datomic, let you know how many bodies I have to dispose of later.

19:02 arrdem: type theory meetup in ~3 hours.

19:02 matt444: that example doesn't have a output-dir

19:02 how do i handle that?

19:03 noonian__: :output-to "some/path/for/node/node-cljs.js"

19:03 will stick your compiled js wherever you specify

19:04 matt444: yes, but there is also a :output-dir

19:06 noonian__: just make it different for each build

19:06 matt444: I don't think that project.clj is right, it doesn't have :target :nodejs

19:06 noonian__: https://www.refheap.com/23532

19:07 yeah, i don't know about any node specific settings

19:09 bitemyapp: (bean dat)

19:09 ClassCastException java.util.Date cannot be cast to java.lang.Number datomic.db.Datum (db.clj:219)

19:09 breathe. Breathe Chris. Find your center.

19:11 mrhanky: why does jayq.core gives me this: Uncaught ReferenceError: jQuery is not defined

19:11 :/

19:11 bitemyapp: the answer for the Daily Double was: (let [{:keys [e a v tx]} dat] (into {} [[:e e] [:a a] [:v v] [:tx tx]]))

19:12 noonian__: mrhanky: you probably need to include jquery in your page first

19:12 mrhanky: i have, i tried 2.x and now 1.9.x

19:13 ivan: is jQuery defined in the console?

19:13 mrhanky: yes

19:14 Bronsa: bitemyapp: wouldn't (select-keys dat [:e :a :v :tx]) do?

19:14 matt444: I've had so many problems with lein-cljsbuild

19:14 ivan: mrhanky: are you using some Google API loader that will asynchronously load jQuery after your other JS runs?

19:15 matt444: Does anyone know where, with lein-cljsbuild I should put :target :nodejs

19:15 noonian__: stick it inside the :compiler map of each build

19:15 bitemyapp: Bronsa: lets find out!

19:16 Bronsa: berossus.api-test> (select-keys dat [:e :a :v :tx])

19:16 ClassCastException datomic.db.Datum cannot be cast to java.util.Map clojure.lang.RT.find (RT.java:733)

19:16 Bronsa: no dice grandma.

19:16 matt444: noonian__ thanks for your patient help, i'll pastebin my config

19:16 technomancy: bitemyapp: man; things that aren't maps

19:16 bitemyapp: Bronsa: confer with my previous angry messages to see why that didn't work.

19:16 matt444: noonian__: https://www.refheap.com/23536

19:17 bitemyapp: technomancy: I was spittin' mad but my tests are passing now so just this once: http://i.imgur.com/J9NLYcI.gif

19:17 noonian__: matt444: np, here a blog post for nodejs setup with an example config that shows the compiler option: http://dannysu.com/2013/01/14/clojurescript-for-nodejs/

19:17 matt444: looks good to me hehe

19:17 technomancy: nice

19:17 matt444: hm

19:17 Why do they do :optimizations :advanced for node?

19:18 there's no reason to do that

19:18 noonian__: it will make your code compile down much smaller

19:18 matt444: code size doesn't matter in node

19:18 it's not going over the wire

19:18 mrhanky: ivan, seems like codemirror produces this ._.

19:18 matt444: is :builds a vector or a map? I've seen both

19:19 mine is currently a map

19:19 it seems that almost no one is using clojurescript with Node

19:19 noonian__: i've seen both also, mine's a map

19:20 matt444: The word node only appears twice in lein-cljsbuild's project, both issues

19:21 noonian__: i've seen a few posts on people doing it but mainly for fun

19:22 `cbp: lol @ those tweets

19:22 matt444: thanks for your help, i'll log an issue

19:23 bitemyapp: `cbp: I'm glad you enjoy my tweets of sorrow.

19:25 mrhanky: ivan, i hade cljs code using (set! (.-onload js/window) (fn [] ...)), switched to jqueries ready

19:25 solved it

19:26 Bronsa: bitemyapp: ouch, that sucks. what's wrong with {:e e :a a :v v :tx tx} anyway?

19:27 bitemyapp: Bronsa: it's absurdly tedious. I shouldn't have to create maps by specifying each attr/value pair and destructuring accordingly.

19:27 Bronsa: datoms are printed as if they're maps, they should behave like ACTUAL maps.

19:27 there is no reason at all for them to do otherwise except for projected Hickeysian torment.

19:28 the pedantry in pretending they're not maps is doubly pointless in an untyped language where there's no real safety to be had anyway.

19:28 if you're in the mud already, play fuckin' ball.

19:29 technomancy: bitemyapp: but inheritance is bad always

19:31 bitemyapp: I'm anticipating my next resolved ticket to involve some (= (class ...) blah.WhoCares)

19:31 kill for a pattern match.

19:33 technomancy: I don't understand why every clojure codebase out there isn't using core.match now that the AOT bugs are gone

19:33 is it still in the middle of a rewrite?

19:33 * bitemyapp sighs and shrugs

19:34 technomancy: not having pattern matching is pretty terrible

19:34 bitemyapp: I gave up on it after having it breaking my code multiple times at different points in time.

19:34 break*

19:34 similarly to how I stopped bothering with c.j.j after the massacre of 0.3.x

19:34 `cbp: rip

19:35 bitemyapp: (inc `cbp)

19:35 lazybot: ⇒ 2

19:35 bitemyapp: (inc cbp)

19:35 lazybot: ⇒ 1

19:35 bitemyapp: for good measure.

19:35 `cbp: :p

19:42 scape: bitemyapp: cjj?

19:42 rasmusto: java jdbc I assume

19:42 coventry`: Is this really the cleanest way to prepare a jar for submission to a storm cluster? https://github.com/nathanmarz/storm-starter/issues/51

19:44 bitemyapp: I write some really dense test code.

19:45 Wild_Cat: quick question: What's the accepted way in a Clojure test to check that some returned value is a specific vector (and specifically *not* the equivalent list)?

19:45 bitemyapp: Wild_Cat: vector?

19:45 Wild_Cat: =?

19:45 Wild_Cat: bitemyapp: I want the test to pass when the returned value is ["foo" "bar"] but not when it's '("foo" "bar")

19:45 bitemyapp: ,(= [0 1 2] '(0 1 2))

19:45 clojurebot: true

19:45 bitemyapp: well that's fucking stupid.

19:45 Wild_Cat: yup, that's the issue :p

19:46 bitemyapp: :|

19:46 Wild_Cat: (and (= ...) (vector? ...))

19:46 * bitemyapp grumbles

19:46 Wild_Cat: right now I'm checking (= (type ret) clojure.lang.PersistentVector) but that's horrible

19:46 hyPiRion: ^

19:46 bitemyapp: Wild_Cat: dude no. use vector?

19:46 Wild_Cat: oh, I didn't know that function existed. Thanks.

19:47 scape: :)

19:47 noonian__: Wild_Cat: (and (vector? ret) (= ret v))

19:47 bitemyapp: scape: I'm still scowling at the badness of that equality.

19:47 Cr8: ,(map vector? [1 [] '() {} #{}])

19:47 clojurebot: (false true false false false)

19:47 bitemyapp: Cr8: good paranoia.

19:47 lol.

19:47 hyPiRion: or eventually (identical? x coll) if you know that a specific vector will be returned

19:47 bitemyapp: That makes = untrustworthy because vector is associative and '() is not.

19:48 Wild_Cat: hyPiRion: oh! Even easier. Thanks.

19:48 noonian__: most of the ops you do on vectors return lazy seqs though so you want that equality

19:48 bitemyapp: so we still haven't learned anything from the egal paper. Sweet.

19:48 noonian__: it's not really about that.

19:48 Cr8: = "do these things have similar insides"

19:48 bitemyapp: Cr8: sounds like JavaScript's == to me.

19:49 Cr8: are you coming to the type theory meetup

19:49 SegFaultAX: ^^

19:49 Wild_Cat: ,(= '(0 1 2) [0 1 2])

19:49 clojurebot: true

19:49 Wild_Cat: (phew, thankfully that's commutative :p )

19:50 Cr8: Cr8: prolly not, I didn't sleep much

19:50 bitemyapp: commutatively...wrong.

19:50 Cr8: be a man. Do the right thing.

19:50 technomancy: bitemyapp: it's a bit weird, but I can understand the stance that laziness would be unbearably tedious without that quirk

19:50 TEttinger: ok uh... my clojure game is really fast when run with lein, which defaults to a server JVM (an example command that does the same thing: java -server -jar myGame.jar). it's also really fast if I have a JDK installed and used for the java command. but... the java that is installed by default for users is the client vm, which seems at least 3x slower

19:50 technomancy: but it's definitely not egal

19:50 bitemyapp: technomancy: Right.

19:50 Wild_Cat: bitemyapp: yeah, sure, that sucks, but it's still better than it not being commutative or transitive

19:50 bitemyapp: technomancy: if the lazy semantics were universal instead of ghettoized in a coll it wouldn't be necessary to fuck up the equality.

19:51 Wild_Cat: (not that I'm implying that Clojure is influenced by PHP or anything :p )

19:51 bitemyapp: technomancy: one wrong decision causes another...

19:51 Wild_Cat: well they're both untyped.

19:51 technomancy: bitemyapp: right; because that doesn't cause *any* weird problems

19:51 TEttinger: (it takes a half second to process a keypress on client, it is almost instantaneous on server)

19:51 bitemyapp: technomancy: I'm sorry, I can't hear you from my shining ivory tower.

19:51 where things just work

19:51 Bronsa: Wild_Cat: what if I told you == isn't commutative in 1.5.1?

19:52 ,*clojure-version*

19:52 clojurebot: {:interim true, :major 1, :minor 6, :incremental 0, :qualifier "master"}

19:52 Bronsa: &*clojure-version*

19:52 lazybot: ⇒ {:major 1, :minor 4, :incremental 0, :qualifier nil}

19:52 bitemyapp: that's old.

19:52 Bronsa: &[(== 1 1N 1.0) (== 1 1N 1.0 1M) (== 1 1N 1.0 1.0M) (== 1 1.0 1N 1.0M)]

19:52 lazybot: ⇒ [true true true false]

19:52 TEttinger: does anyone have any tips for speeding up JVM clojure without the trick lazybot uses to launch another java command?

19:52 bitemyapp: Bronsa: oh sweet.

19:52 Wild_Cat: eew.

19:52 bitemyapp: PHP-jure.

19:53 technomancy: TEttinger: are you sure your users don't have access to the server JVM?

19:53 Bronsa: bitemyapp: it's fixed since 1.6.0-alphasomething though

19:53 TEttinger: technomancy, kinda

19:53 Wild_Cat: all right, thanks for helping me make my pull request prettier.

19:53 bitemyapp: Bronsa: not really comforting.

19:53 TEttinger: I tried it on a JRE only java 7 install, up to date

19:53 no -server option there

19:53 technomancy: TEttinger: who are these mysterious users running on 32-bit boxes?

19:53 TEttinger: my mom.

19:53 hyPiRion: bitemyapp: that's just because loljava bigdecimals

19:54 ,(== 1.0M 1.00M) ;; etc

19:54 TEttinger: technomancy, is it really a 64-bit only thing?

19:54 clojurebot: true

19:54 hyPiRion: oh, that's fixed on master

19:54 &(== 1.0M 1.00M)

19:54 lazybot: ⇒ false

19:54 technomancy: TEttinger: no, it's only that 64-bit machines are server-only

19:54 bitemyapp: hyPiRion: Datomic already drained my ability to care today.

19:54 hyPiRion: bitemyapp: hurray.

19:55 technomancy: most users don't even have access to client hotspot

19:55 TEttinger: on JRE, not JDK???

19:55 lazybot: TEttinger: Oh, absolutely.

19:55 Wild_Cat: TEttinger: yeah, even on JRE

19:55 technomancy: TEttinger: pretty sure it's the same on both

19:55 they never bothered implementing -client

19:55 Wild_Cat: AFAIK the JVM installed by the JRE and the JDK is the same.

19:55 bitemyapp: TEttinger: betcha wish ya had native binaries right about now >:)

19:56 Wild_Cat: the main difference being that one comes with the Ask toolbar -.-

19:56 TEttinger: bitemyapp, yeah. except haskell doesn't have any game libs.

19:56 technomancy: TEttinger: never heard of -server being missing

19:56 you could try tiered compilation maybe

19:57 bitemyapp: TEttinger: well that's not true at all.

19:57 TEttinger: there's Helm, HGame3d, there's a whole channel devoted to it at #haskell-game

19:57 TEttinger: oh neat

19:57 bitemyapp: TEttinger: you can use any dyn-libs via the FFI you want.

19:57 technomancy: RainCat!

19:57 bitemyapp: TEttinger: most people use SDL for 2d, OpenGL for 3d.

19:57 technomancy: is adorable

19:57 Cr8: java hates me today

19:57 TEttinger: ouch bitemyapp

19:57 bitemyapp: technomancy: yes it is.

19:58 TEttinger: ouch?

19:58 TEttinger: I mean, if you want something like Unity3d, you're not going to get it.

19:58 TEttinger: raw OpenGL is not something I have had a good time with

19:58 I usually have used wrappers

19:58 bitemyapp: TEttinger: so use GLUT or GLEW

19:59 TEttinger: http://www.haskell.org/haskellwiki/OpenGL http://www.haskell.org/haskellwiki/OpenGLTutorial1 http://hackage.haskell.org/package/GLUT http://www.reddit.com/r/haskell/comments/1tay08/modern_opengl_in_haskell/

19:59 TEttinger: OpenGL and GLUT are included in the Haskell platform by default.

19:59 * technomancy looks at the clock

19:59 bitemyapp: at most, you'd install some headers.

19:59 technomancy: dammit bitemyapp

19:59 bitemyapp: technomancy: I'm helping him!

19:59 technomancy: you're like one hour early

19:59 Wild_Cat: Haskell has no games. It has generalized abstract game engine transformer transformer combinator transformers.

19:59 TEttinger: yeah, total rewrite is not helping

19:59 Wild_Cat: :p

20:00 bitemyapp: :(

20:00 TEttinger: is it on Github?

20:00 TEttinger: yeah, not the latest yet, because my current local copy is kinda broken ATM

20:00 bitemyapp: TEttinger: link please.

20:00 technomancy: TEttinger: does tiered compilation work?

20:00 TEttinger: https://github.com/tommyettinger/dungeon-kingpin

20:01 technomancy, I don't know what it is yet

20:01 and it seems to be Java 7 only?

20:01 technomancy: no, it's been around a long time

20:01 it's faster in 7 though

20:01 Cr8: COME ON

20:01 bitemyapp: TEttinger: your dangling ))'s are unholy in the eyes of god

20:01 Cr8: i keep getting outofmemoryerror java heap space

20:01 bitemyapp: Cr8: jack dat Xmx yo.

20:02 technomancy: TEttinger: it's a way of telling -server to start out as fast-launching and shift to high-perf over time

20:02 an attempt at getting best of both worlds of -client and -server

20:02 Cr8: bitemyapp: I jacked it up to 8G :[

20:02 bitemyapp: Cr8: hahahahaha ouch.

20:02 Cr8: and all i'm trying to do is start a repl

20:02 TEttinger: technomancy, but it needs to be the other way around...?

20:03 bitemyapp: Cr8: wat.

20:03 Cr8: lein repl; empty project; empty profiles.clj; out of heap

20:03 technomancy: TEttinger: not sure; it might be a way to force it into high-perf mode even if you can't invoke -server

20:03 Cr8: not really ..

20:03 okay now it works

20:03 TEttinger: http://docs.oracle.com/javase/7/docs/technotes/guides/vm/server-class.html

20:03 Cr8: wait i lied

20:03 it started a repl

20:04 then i typed a character and i got it again

20:04 amalloy: https://github.com/tommyettinger/dungeon-kingpin/blob/master/src/dk/herringbone.clj -- best source file ever, TEttinger?

20:04 TEttinger: amalloy,I know right

20:05 code is data!

20:05 Cr8: https://www.evernote.com/shard/s17/sh/95196df9-343c-4170-ae86-214aeb84fd53/04497dbef41ff7e8b1c904a35d0fb1f4/deep/0/1.-apage43@Tyrathect----proj-tiny-(zsh).png

20:05 TEttinger: Cr8, heh

20:05 bitemyapp: Cr8: what in god's name did you do to your JVM?

20:05 TEttinger: it is probably trying to load a ns

20:05 Cr8: i think i'm going to reboot my computer

20:06 bitemyapp: Cr8: Tyrathect!

20:06 Cr8: I just read that book!

20:06 Cr8: fantastic.

20:06 TEttinger: and then dying because you have a massive def or something

20:06 Cr8: that's my computer's name

20:06 noonian__: Cr8: if you have a :main key in your project.clj try commenting it out

20:06 clojurebot: It's greek to me.

20:06 Cr8: bitemyapp: cool :D

20:06 noonian__: I don't. This is a fresh lein new

20:06 bitemyapp: Cr8: reboot and try again?

20:07 Cr8: doing so

20:07 noonian__: lol, that sucks then

20:08 arrdem: technomancy: I love the new topic

20:08 bitemyapp: arrdem: I'm very proud of that.

20:08 * bitemyapp wipes tear

20:09 gtrak: hahaha

20:09 Cr8: well

20:09 everything's fine now

20:09 llasram: Another hour to go!

20:09 Cr8: user=> 1

20:09 1

20:09 success!

20:09 arrdem: llasram: that means I have an hour to give my talk and get bitemyapp on dota so he doesn't steal more souls...

20:09 gtrak: vote for cljs compiler dumps! http://dev.clojure.org/jira/browse/CLJS-748

20:10 technomancy: bitemyapp has set his alarm

20:10 two hours after ALLBALLS

20:10 llasram: heh

20:11 Cr8: ,(vector? (first {:a 1}))

20:11 gtrak: once we have that, we'd be about an hour away from an autodoc.

20:11 clojurebot: true

20:11 bitemyapp: Cr8: seq of vecs

20:12 Cr8: I have to memorize the types because I use an untyped language.

20:12 Cr8: ,(class (first {:a 1}))

20:12 clojurebot: clojure.lang.MapEntry

20:12 bitemyapp: pseudo-vecs.

20:12 fine.

20:12 inheritance?

20:12 Cr8: &(-> {:a 1} first class supers)

20:12 lazybot: ⇒ #{java.lang.Iterable clojure.lang.Sequential java.io.Serializable clojure.lang.IPersistentStack clojure.lang.IPersistentVector java.util.Collection java.lang.Comparable clojure.lang.AFn java.util.concurrent.Callable java.lang.Runnable clojure.lang.ILookup clojure.lan... https://www.refheap.com/23557

20:13 bitemyapp: A and I anyway.

20:13 Cr8: has APersistentVector

20:13 yeah

20:20 bitemyapp: for datomic.db.Datum: #{clojure.lang.ILookup java.lang.Object clojure.lang.Counted clojure.lang.Indexed clojure.lang.IType datomic.impl.db.IDatum datomic.Datom datomic.db.IDatumImpl}

20:20 well that's not nice.

20:20 Cr8: that's..

20:20 short

20:20 bitemyapp: Cr8: did you see my smashing my head into them earlier?

20:20 Cr8: I did

20:26 speaking of books, Don't Sleep, There Are Snakes was good, I read it on a plane a bit back

20:28 w.r.t. your rt there

20:29 though it doesn't actually go into much detail on the piraha language, the story was interesting

20:35 insamniac: TEttinger: Is dungeon-kingpin supposed to work? I can't pull down all the deps

20:35 TEttinger: insamniac, that's odd

20:35 which deps can't it get?

20:36 bitemyapp: insamniac: I really like your name.

20:37 insamniac: please tell me your given name is Sam.

20:37 seangrove: Deploying Clojure to Heroku in what seem like off-hours? Grab a drink, sit back, relax. You got loads of time.

20:38 arrdem: seangrove: really? it's always been pretty fast for me..

20:38 seangrove: arrdem: And you uberjar?

20:39 bitemyapp: uberjars start faster, but are fatter.

20:39 arrdem: seangrove: I don't think so. I just used their git target with an uberjar->slug for their end.

20:39 *configuration

20:39 bitemyapp: but I don't use Heroku.

20:40 because I build my own lightsabers.

20:40 seangrove: bitemyapp: Yeah, but I can compile them quickly locally, but letting whatever scraps heroku uses to compile them is painful

20:40 technomancy: bitemyapp: whatever; I saw your keyboard

20:40 _eric: I have some tests where I want to slurp in some files to use in the tests

20:40 is there a convention for where to put the files and how to read them?

20:41 bitemyapp: technomancy: that's @work, I use a MX cherry thingamajig at home with Linux.

20:41 technomancy: seangrove: we just opened up the API to accept 3rd-party slugs though so you can do it from a CI server or something

20:41 please don't build on your laptop =(

20:41 bitemyapp: we build uberjars on our dev machines here, lol.

20:41 arrdem: _eric: $PROJECT_ROOT/resources/*

20:41 seangrove: technomancy: That sounds good, but I'm wondering if we get much benefit from heroku/clojure over just using ansible + uberjar

20:41 arrdem: _eric: leiningen will put that stuff on the classpath for you.

20:41 bitemyapp: seangrove: very neutral source.

20:42 arrdem: _eric: (slurp <filename>) should work.

20:42 * bitemyapp smiles

20:42 seangrove: going on ~6 minutes on this push...

20:42 _eric: cool

20:42 technomancy: bitemyapp: http://p.hagelb.org/disappointed.gif

20:42 _eric: is it only $PROJECT_ROOT/resources/ or is there a $PROJECT_ROOT/test/resources?

20:42 seangrove: technomancy: The API sounds nice though, it's the next piece of the workflow I'd like to setup - jenkins + push-button deploys of passing uberjars

20:43 bitemyapp: technomancy: I knew what the gif was going to be before I clicked.

20:43 hyPiRion: you know what

20:43 arrdem: _eric: you can put files wherever you want on the classpath

20:43 technomancy: bitemyapp: damn; too predictable

20:43 hyPiRion: clojurebot: mystery |is| http://p.hagelb.org/mystery.gif

20:43 clojurebot: Roger.

20:43 bitemyapp: technomancy: it's the first time you haven't surprised me though.

20:43 other than the rather common mystery.gif, but that one is always a delight.

20:43 technomancy: bitemyapp: http://p.hagelb.org/tastes-mature.png

20:44 * technomancy makes up for it with a hasty non sequitur and departs for dinner

20:45 hyPiRion: bitemyapp: hey, I bet you've not seen this one before: http://hypirion.com/imgs/lenanign.png

20:45 seangrove: Go home hyPiRion, you're drunk

20:45 bitemyapp: hyPiRion: why?

20:45 seangrove: francophones.

20:46 hyPiRion: bitemyapp: people misspell leiningen in many ways

20:46 gtrak: hyPiRion: she's in every image processing textbook.

20:47 seangrove: technomancy: Admit it, Clojure is so despised at Heroku that the only build servers allotted to clojure projects are the old coffee-grinders thrown out running in the back of McGranaghan's station wagon.

20:47 hyPiRion: gtrak: What do you want me to do? Attach a mustache to every single one of those images?

20:48 r00k_: Anybody here make much use of :pre and :post conditions for your functions? Do you feel like it pays off?

20:48 seangrove: r00k_: I use :pre in quite a few places, it's pretty handy for sanity checks

20:48 gfredericks: hiredman: oh man I did not realize sneakyThrow was actually sneaky

20:48 technomancy: seangrove: I could show you some horrifying graphs that involve scala

20:48 hyPiRion: Maybe I should make an image detection plugin for my browser which automatically adds a mustache to Lena

20:48 gfredericks: I suddenly feel like I don't understand the java type system

20:48 technomancy: seangrove: we get tickets for 15-minute scala builds timing out on a regular basis

20:50 r00k_: seangrove: And you find those asserts blowing up often enough to feel worth the cost (time writing them, visual weight)?

20:50 bitemyapp: seangrove: hahahahaha

20:50 r00k_: well in the absence of a sensible type system...

20:50 llasram: gfredericks: http://james-iry.blogspot.com/2010/08/on-removing-java-checked-exceptions-by.html

20:50 seangrove: technomancy: Well, we'll port over to RoR ASAP ;)

20:50 bitemyapp: whew, just typed defun in Clojure code.

20:50 amalloy: gfredericks: in fairness, neither does java

20:50 seangrove: r00k_: They're generally tiny

20:51 e.g. {:pre [(not (empty? api-key))]}, {:pre [(integer? org-id)]}

20:51 gfredericks: llasram: thanks!

20:51 r00k_: bitemyapp: Yeah, I was going to ask if maybe we're just adding a poor man's type system.

20:51 seangrove: Poor, poor-man's type-system

20:51 hyPiRion: gfredericks: yeah, sneakythrow is generic abuse to the max

20:52 arrdem: bitemyapp: 10 minutes till you can start pushing haskell

20:52 amalloy: seangrove: "tiny...(not (empty? x))" :( (seq x)

20:52 seangrove: ,(seq "")

20:52 clojurebot: nil

20:52 seangrove: ffs

20:53 r00k_: seangrove: Do they tend to blow up just when you're first writing the code that calls your function, or later when your data gets screwy?

20:53 seangrove: r00k_: I put them in in the beginning when I don't want to write a full test, and take them out when things start looking solid

20:53 r00k_: Hah. (seq x) makes it *really* look like you just want a type system.

20:53 amalloy: seangrove: (empty? x) is literally implemented as (not (seq x)). there's never a reason to write (not (empty? x))

20:53 gfredericks: I guess the suggested idiom of "throw sneakyThrow(...)" is just to make it syntactically obvious that you're throwing something?

20:53 seangrove: amalloy: What about (not (not (seq x?))) ?

20:54 ;)

20:54 r00k_: Hah.

20:54 amalloy: maybe if someone put a gun to your head and said "start your preconditions with (not...or DIE"

20:54 gfredericks: ,'not...or

20:54 clojurebot: not...or

20:54 r00k_: Thanks for the input, seangrove. I appreciate it!

20:55 Wild_Cat: hey, that's pretty cool: http://blog.travisthieman.com/write-you-a-starcraft-ai-in-clojure/

20:55 gfredericks: ,(let [not...or identity, DIE 42] (not...or DIE))

20:55 clojurebot: #<CompilerException java.lang.ClassFormatError: Illegal field name "not...or" in class sandbox$eval73, compiling:(NO_SOURCE_PATH:0:0)>

20:55 hyPiRion: gfredericks: close, but no cigar

20:56 insamniac: this game should be called dancing with ogres

20:57 gfredericks: ,(def not...or identity)

20:57 clojurebot: #'sandbox/not...or

20:57 gfredericks: ,(let [DIE 42] (not...or DIE))

20:57 clojurebot: #<CompilerException java.lang.ClassNotFoundException: not///or, compiling:(NO_SOURCE_PATH:0:0)>

20:58 gfredericks: huh.

20:58 llasram: I'm really amazed you can `def` that

20:58 gtrak: ,(read-string "clojure//")

20:58 gfredericks: ,(let [DIE 42] (sandbox/not...or DIE))

20:58 clojurebot: clojure//

20:58 42

20:58 gtrak: ,(read-string "cljs//")

20:58 clojurebot: cljs//

20:58 gfredericks: llasram: ooh you can use it fully qualified too

20:58 llasram: niiiice

20:58 gtrak: ,(read-string "cljs.core//")

20:58 clojurebot: cljs.core//

20:58 TEttinger: insamniac, haha

20:58 gtrak: damn... I forget which one was broken :-)

20:59 gfredericks: ,(def .......... 7)

20:59 clojurebot: #'sandbox/..........

20:59 llasram: gtrak: It's been fixed

20:59 gtrak: oh really?

20:59 gfredericks: ,(* sandbox/.......... 2 3)

20:59 clojurebot: 42

20:59 llasram: ,*clojure-version*

20:59 clojurebot: {:interim true, :major 1, :minor 6, :incremental 0, :qualifier "master"}

20:59 gtrak: aha

20:59 gfredericks: oh man we can make unary helpers

20:59 ,(def .. 2)

20:59 clojurebot: #<CompilerException java.lang.SecurityException: denied, compiling:(NO_SOURCE_PATH:0:0)>

20:59 gfredericks: ,(def ... 3)

20:59 clojurebot: #'sandbox/...

20:59 gfredericks: ,(def . 1)

21:00 clojurebot: #'sandbox/.

21:00 gfredericks: waaat

21:00 llasram: ,`.

21:00 clojurebot: .

21:00 gfredericks: what is going on with 2

21:00 ,(def .... 4)

21:00 clojurebot: #'sandbox/....

21:00 hyPiRion: gfredericks: lol

21:00 gfredericks: ,(* sandbox/... sandbox/. sandbox/....)

21:00 clojurebot: 12

21:00 bitemyapp: yum. more arity type errors.

21:00 hyPiRion: ,(+ . .)

21:00 clojurebot: 2

21:01 darthdeus: for all the light table fans out there http://blog.jakubarnold.cz/light-table-plugin-tutorial :P

21:01 bitemyapp: hyPiRion: dafuq.

21:01 gfredericks: ,#'clojure.core/..

21:01 clojurebot: #'clojure.core/..

21:01 gfredericks: oh maybe that is why

21:01 llasram: Ohhhhh

21:01 Yes

21:01 That makes "sense"

21:01 gfredericks: :)

21:01 ,(+ ... ...)

21:01 clojurebot: 6

21:01 Cr8: wat

21:01 hyPiRion: oh, I got this

21:01 gfredericks: ,(def ...... 6)

21:01 clojurebot: #'sandbox/......

21:01 hyPiRion: ,(* ... (+ . ...))

21:01 clojurebot: 12

21:02 gfredericks: clojure is a DSL for fans of unary who are allergic to 2

21:02 hyPiRion: "And today, in Swearjure part 2, we'll have a look at periods."

21:02 TEttinger: ,(. toString 4)

21:02 clojurebot: #<CompilerException java.lang.RuntimeException: Unable to resolve symbol: toString in this context, compiling:(NO_SOURCE_PATH:0:0)>

21:02 TEttinger: no!

21:02 ,(.toString 4)

21:02 clojurebot: "4"

21:02 amalloy: not so much clojure as clojurebot, gfredericks

21:02 TEttinger: ,(. 4 toString 4)

21:02 clojurebot: #<IllegalArgumentException java.lang.IllegalArgumentException: No matching method found: toString for class java.lang.Long>

21:02 TEttinger: hm

21:03 there is one though

21:03 amalloy: ,(. 4 toString)

21:03 clojurebot: "4"

21:03 gfredericks: amalloy: oh good point

21:03 amalloy: you can define .. for real in a repl you own

21:03 TEttinger: ,(Long/toString 4 4)

21:03 clojurebot: "10"

21:03 gfredericks: you mean I can own my very own repl?

21:03 TEttinger: there we go, it's static

21:07 hyPiRion: oh, that's an interesting identity

21:07 ,(map #(->> % (Long/toString %) read-string (Long/toString %) read-string) (range 1 20))

21:07 clojurebot: (1 2 3 4 5 ...)

21:09 gfredericks: ...it is?

21:09 akurilin: If I wanted to keep my site's static html/css/assets separate from the .jar, it doesn't really matter where I place them, right? As long as I have a path to them and file system permissions, I should be able to keep those outside of the jar resources?

21:10 I just feel dumb for putting assets in the jar, I have to build the ring app and restart it every time I need to make the smallest change.

21:10 abaker: sure, but you can also build a war, explode the war, put in tomcat, and then mess with stuff all day long

21:11 Cr8: ^~

21:11 akurilin: Is that still pretty simple? Because right now my setup is jetty daemonized by runit, which is as bare-bones as it gets and works reallyw ell.

21:12 Cr8: jetty should be able to run exploded WARs as well

21:12 hyPiRion: akurilin: You can place it separate from the .jar. That's what I've done so far.

21:12 rovar: I'm working through http://www.learndatalogtoday.org/ I see that they solve a combinatorial problem rather elegantly, and I also see that there was a bit of discussion on the mailing list about core.logic and datalog..

21:13 I guess my question is has anyone tried to tackle more complex problems with datalog.. like the travelling salesman or other optimization problems?

21:13 I'm actually interested in vehicle routing with constraints..

21:13 akurilin: hyPiRion: good to know that people do it.

21:13 rovar: or some variant therein

21:14 hyPiRion: akurilin: For me, I just placed `(route/files "/" {:root "my-directory"})` at the bottom of my main compojure defroutes, where my-directory are the static files

21:14 er, the directory containing the static files

21:16 akurilin: hyPiRion: ok so it's just a regular unix path, I imagine through an env var or something of that sorts?

21:16 hyPiRion: akurilin: yeah, it's an env var for me, (and it can be relative)

21:17 abaker: rovar: sure, with SPARQL, which can map 1-to-1 with datalog

21:17 akurilin: hyPiRion: ok perfect, I love to shove everything into environ through our CM.

21:18 I guess for some reason I thought there's be special rules to accessing files outside of the jar due to some JVM permission rules, but I guess an app is an app.

21:19 hyPiRion: akurilin: nah, standard ACL rules apply

21:21 akurilin: hyPiRion: great, thanks a bunch for clarifying that :)

21:21 hyPiRion: no problem

21:21 :)

21:22 afhammad: Hi, what is the base footprint for latest clojurescript projects after advanced compilation?

21:24 rovar: abaker, so you're saying you've done it on a graph database?

21:26 abaker: rovar: yeah - I work on Stardog, where sparql + reasoning/rules and such can be used to build all sorts of fun stuff

21:28 rovar: abaker, neato. I do big data infrastructure for a day job, but have been poking around with scheduling and routing optimization for fun..

21:30 abaker: rovar: yeah fun stuff, upcoming release we're adding higher level services for graph analytics, shortest path computations, finding cliques, etc

21:37 _eric: is there an efficient way to say "if ... else if ... else if ... else ..."?

21:38 riley526: _eric: ,(doc cond)

21:38 _eric: sort of like (case) works

21:38 riley526: ,(doc cond)

21:38 clojurebot: "([& clauses]); Takes a set of test/expr pairs. It evaluates each test one at a time. If a test returns logical true, cond evaluates and returns the value of the corresponding expr and doesn't evaluate any of the other tests or exprs. (cond) returns nil."

21:38 gregorstocks: hey bitemyapp should i use haskell

21:38 _eric: sweeeet

21:38 yay!

21:38 thanks

21:39 riley526: np!

21:39 _eric: I just spent 10 minutes trying to figure out what I was asking for

21:40 riley526: _eric: Yeah some things can be tough to find if you don't know what it's called

21:47 akurilin: gregorstocks: lol.

22:36 logic_prog: what's a good name for something that is a message + a return channel

22:38 Raynes: logic_prog: A portable post office.

22:38 logic_prog: i'm trying to come up with good names for deftype / defrecord

22:39 Raynes: PortablePostOffice

22:39 :p

22:44 Cr8: SASE

22:45 llasram: I was just thinking that too, Cr8 :-)

Logging service provided by n01se.net