#clojure – 2015-10-07

2015-10-07

00:34	Jaood	'
00:35	wmealing	'
00:35	wmealing	i couldnt let that go unmatched
00:35	justin_smith	' doesn't need a match in clojure
00:36	wmealing	oh NOW people talk
00:36	wmealing	;P
00:36	amalloy	justin_smith couldn't bear to see your correction go uncorrected
00:36	justin_smith	things have been getting quiet around here
00:36	amalloy	justin_smith: is everything in slack now, or what?
00:37	justin_smith	it's pretty active... but I can't use it like I do IRC
00:37	justin_smith	too much of an "app"
00:37	wmealing	what is slack ?
00:37	wmealing	https://slack.com/?r=1 ?
00:37	justin_smith	it's an app that's a lot like IRC, and there's a clojurians channel there that a lot of people are using now
00:38	justin_smith	it even has an IRC gateway that I may eventually set up, my workplace uses slack
00:42	justin_smith	last I checked, having an archive / searchable log / on slack cost money, so clojurians doesn't have a searchable backlog
01:13	Frozenlock	Gotta love all those new shiny stuff. Remember when email was broken? Now we have Facebook messages. Wheeeeee
01:16	wmealing	i didnt think that any of it was broken
01:17	Frozenlock	It wasn't, nor is IRC.
01:18	TEttinger	being able to post snippets of code inline with a preview is pretty nice though, on slack
01:18	mungojelly	i think irc is still awesome but it could also be improved, like why don't we have any irc clients that understand richer text, we wouldn't have to change anything about the infrastructure just put <i> or w/e and parse it on the other side
01:18	TEttinger	I like Gitter better than Slack, less signing up :P
01:19	mungojelly	signing up prevents abuse which is why the internet has been such a sparlking fantastic positive place since we started having to sign up everywhere /s
01:19	amalloy	yeah, i use slack at work, but i am not willing to sign up for slack in order to try the clojure slack
01:42	devangshah	how can i use java functions inside the anonymous functions as part of map or filter?
01:42	justin_smith	devangshah: by using java methods inside anonymous functions
01:43	devangshah	i am trying to access them using (doall (filter #(.isDirectory) (file-seq (io/file dirpath))))
01:43	justin_smith	devangshah: you need % to indicate the argument
01:43	justin_smith	#(.isDirectory %)
01:44	devangshah	silly me…
01:44	devangshah	thanks anyway..
01:44	devangshah	newbie to clojure
01:45	justin_smith	devangshah: #(foo) is a function of no arguments that calls foo, the argument count expected is determined by the usage of %, %2, etc. you can also use %& for varargs
01:46	devangshah	yes.. i knew it.. even i used it in my project.. but somehow i missed here.. thanks anyway..
02:11	sm0ke	,(clojure.java.shell/sh "pwd")
02:11	clojurebot	#error {\n :cause "clojure.java.shell"\n :via\n [{:type java.lang.ClassNotFoundException\n :message "clojure.java.shell"\n :at [java.net.URLClassLoader$1 run "URLClassLoader.java" 366]}]\n :trace\n [[java.net.URLClassLoader$1 run "URLClassLoader.java" 366]\n [java.net.URLClassLoader$1 run "URLClassLoader.java" 355]\n [java.security.AccessController doPrivileged "AccessController.java" -2]\n ...
02:11	justin_smith	,(System/getenv "PWD")
02:11	clojurebot	#error {\n :cause "access denied (\"java.lang.RuntimePermission\" \"getenv.PWD\")"\n :via\n [{:type java.security.AccessControlException\n :message "access denied (\"java.lang.RuntimePermission\" \"getenv.PWD\")"\n :at [java.security.AccessControlContext checkPermission "AccessControlContext.java" 372]}]\n :trace\n [[java.security.AccessControlContext checkPermission "AccessControlContext.java...
02:11	justin_smith	:P
02:12	sm0ke	,(do (require 'clojure.java.shell) (clojure.java.shell/sh "pwd"))
02:12	clojurebot	#error {\n :cause "denied"\n :via\n [{:type java.lang.SecurityException\n :message "denied"\n :at [clojurebot.sandbox$enable_security_manager$fn__883 invoke "sandbox.clj" 69]}]\n :trace\n [[clojurebot.sandbox$enable_security_manager$fn__883 invoke "sandbox.clj" 69]\n [clojurebot.sandbox.proxy$java.lang.SecurityManager$Door$f500ea40 checkExec nil -1]\n [java.lang.ProcessBuilder start "Process...
02:12	sm0ke	ok but does (sh/sh "echo foo") works for anyone?
02:13	wmealing	,(do (require 'clojure.java.shell) (clojure.java.shell/sh "echo $ENV"))
02:13	clojurebot	#error {\n :cause "denied"\n :via\n [{:type java.lang.SecurityException\n :message "denied"\n :at [clojurebot.sandbox$enable_security_manager$fn__883 invoke "sandbox.clj" 69]}]\n :trace\n [[clojurebot.sandbox$enable_security_manager$fn__883 invoke "sandbox.clj" 69]\n [clojurebot.sandbox.proxy$java.lang.SecurityManager$Door$f500ea40 checkExec nil -1]\n [java.lang.ProcessBuilder start "Process...
02:13	sm0ke	ah its supposed to be used like (sh/sh "echo" "foo")
02:13	sm0ke	weird
02:14	justin_smith	also it won't expand globs and doesn't really call sh
02:15	sm0ke	(sh "echo" "-n foo") (sh "echo" "-n" "foo") are different
02:16	sm0ke	interesting implementation
02:19	opqdonut	sm0ke: no, a very standard implementation
02:19	opqdonut	sm0ke: works just like the C exec
02:19	opqdonut	though I have to admit it's a bit weird to call it sh and not do sh-like command handling
02:21	sm0ke	there is a `sh` in C which is completely different
02:21	sm0ke	so you can compare it to `exec` and call it normal
02:21	opqdonut	do you mean `system` in C?
02:22	sm0ke	its like saying my cat is absolutely fine just that it looks like a elephant
02:22	opqdonut	yeah, I do agre that it is surprising
02:22	opqdonut	*agree
02:27	mungojelly	(clojure.java.shell/sh "clojure" :in "(clojure.java.shell/sh \"clojure\" :in \"(print \\\"hi\\\")\")")
03:05	nXqd	hi guys how can I pass the keys parameter in function call as a variable. For example : (def base-keys [username password]) (defn x [& {:keys base-keys} ())
03:08	sm0ke	,( (fn [& {:keys [foo]}] (prn foo)) :foo 1)
03:08	clojurebot	1\n
03:08	amalloy	~mapply
03:08	clojurebot	You have to do something like (defn mapply [f & args] (apply f (apply concat (butlast args) (last args)))), which just goes to show why unrolled keyword args are a bad idea
03:08	sm0ke	is this what you meant?
03:08	amalloy	oh, that. ignore mapply
03:08	amalloy	you can't do that at all
03:08	amalloy	somewhere i ahve a stackoverflow answer that is semi-relevant
03:09	nXqd	sm0ke: if i understand correctly, the example you show does show the normal usage of parameter destructuring
03:10	amalloy	basically you don't want to do that: instead define the function as taking a map, and write a simple macro like (with-user user body) that destructures username and password out of the user map
03:10	nXqd	what I want is to pass the key as variable
03:10	sm0ke	,(let [key :foo] ( (fn [& {:keys [foo]}] (prn foo)) key 1))
03:10	clojurebot	1\n
03:10	nXqd	yeah it would be a better way :P
03:11	amalloy	you can't
03:12	nXqd	amalloy: and btw, how to insert multiple value to a record in datomic. For example : I have email, password, name in the function, then I can insert some how all of the dynamically
03:12	nXqd	instead of specify each like :user/email email, :user/password password
03:12	amalloy	beats me. i can't even insert one anything into datomic
03:13	nXqd	hmm @(d/transact conn [{:db/id (d/tempid :db.part/user) :user/email email :user/password password}])
04:03	Fender	Hi guys, I want an async web server in clojure (such as httpkit or ring-async), however, most servers are not able to deal with an async channel for the complete response (instead, the only the body is an async channel). This means, I cannot set the HTTP status code of the response and I'll probably have to respond with 200 all the time. Then there's jet, that is able to do it but was obviously compiled against Java 8. So - is it usual to
04:05	TMA	Fender: your question was chopped off after "So - is it usual to"
04:06	Fender	thx - "So - is it usual to always respond in async servers with HTTP 200, am I missing something or is there no good server for my problem?"
04:41	nXqd	anyone here using component, please explain me what is map->ComponentName , is map-> a macro or something. how is it defined ? thanks
04:42	ianhedoesit	map->* is a constructor created by defrecord
04:44	ianhedoesit	I suppose factory functions might be easier to distinguish due to the OO hold on the term "constructor"
04:44	ianhedoesit	->TypeName takes positional parameters to field values and map->TypeName takes a map of keywords to field values
05:02	tdammers	so I'm having a bit of trouble adjusting to clojure (coming from Haskell)
05:02	tdammers	my main gripe is that every fn reads as something like "Anything -> Anything -> ... -> IO (Maybe Anything)
05:03	tdammers	and the only strategy I've found so far for understanding what it does is reading the code
05:03	tdammers	including all of its dependencies
05:03	tdammers	but I can't help but think that there must be a better way
05:05	ianhedoesit	you mean you don't know how to understand functions in Clojure because there's no type hints or anything?
05:05	ianhedoesit	(I'm not trying to sound rhetorical or anything, I'm trying to understand your problem)
05:07	tdammers	pretty much, yes
05:07	tdammers	sometimes, there's documentation, which helps a lot, but still
05:08	nXqd	ianhedoesit: hey thanks ! I didn't notice Clojure has bunch of -> factory function :)
05:08	tdammers	e.g., I'm working with kioo here, and there are all these macros that take an optional emit-opts argument
05:08	tdammers	there is no documentation on that argument
05:09	ianhedoesit	nXqd: it actually (I think?) only has the two that are generated from defrecord (->TypeName and map->TypeName)
05:09	ianhedoesit	they're pretty handy.
05:09	tdammers	in order to figure out how to use it, I have to descend about 4 or 5 levels into the call chain
05:09	tdammers	several paths, even, because each call dependency looks at a different key
05:10	tdammers	this is really not a complaint, I just want to know where the "easy" button it
05:10	tdammers	*is
05:12	ianhedoesit	tdammers: I think that's one of the common complaints (and something I worry about with dynamic languages) about all dynamic languages, no?
05:12	tdammers	ianhedoesit: yes
05:12	ianhedoesit	one of the huge niceties of Haskell, for example, is that you can just look at a function signature and know, at least, exactly what types you need to pass to a function to do _something_
05:13	tdammers	exactly
05:13	ianhedoesit	in dynamic languages you don't have that option. however, core.typed and Prismatic/Schema try to help with that.
05:13	tdammers	also, even with nothing but type signatures in place, the generated documentation is still useful in that you can easily click through to anything any function references
05:17	ianhedoesit	well to attempt an answer of your initial problem/question, you have to rely on good code, for the most part. also making assumptions like (defn foo [m k v] ,,,) probably takes a map, a key, and a value, etc.
05:17	ianhedoesit	or add type hints and whatnot.
05:17	tdammers	so it takes a map; but which keys does it expect, which values are allowed for them, etc.
05:21	ianhedoesit	that's typically where you need to look at the docstring
05:22	ianhedoesit	I don't know if that's all that different even for statically typed languages though? just because you have Map -> Key -> Anything -> Map doesn't mean you know what exactly to pass it either
05:23	ianhedoesit	you can have the same questions ("What key? What Anything? What's the Anything used for?" etc.)
05:25	tdammers	except that typically you wouldn't be using `Map Anything Anything -> Anything -> IO Anything`, but rather something like `Map UserID User -> UserID -> Maybe User`
05:26	ianhedoesit	true, but then you need to define all those things, no?
05:26	ianhedoesit	(I have virtually 0 experience with Haskell, so excuse my ignorance in this situation)
05:27	tdammers	you have to do that anyway, either explicitly (through data types) or implicitly (through the code that handles them)
05:27	tdammers	can be as simple as newtype UserID = UserID Int, even
05:27	tdammers	or even just type UserID = Int, which is less useful, but still acts as documentation at least
05:28	tdammers	but yes, defining your data types upfront does form a bit of overhead
05:36	mungojelly	couldn't you also just type a program strongly enough that you don't need anything but types, that the whole flow of the program can only go one way b/c types, has anyone done that yet?
05:36	amalloy	yes, you can just run your entire program inside the compiler
05:37	amalloy	but then it's no easier to analyze than a program that runs ar tunrime
05:37	tdammers	GHC's type checker itself is turing complete, so you can write anything in it, yes
05:37	mungojelly	like you'd have a function processed-some -> processed-more, and then processed-more -> processed-most, and processed-most -> done, and so it's just in the types how to compute the result
05:38	mungojelly	turing complete, everything's turing complete. :/ don't we have any better standards
05:39	tdammers	not everything is turing complete
05:39	tdammers	most of the older SQL dialects aren't
05:39	tdammers	many type checkers aren't either
05:39	mungojelly	but i mean people say something's turing complete like that proves you could do anything useful with it, but what did anyone make with a turing machine lately
05:40	tdammers	the thing about TC is that if you can implement an algorithm in one TC language, you can implement it in all of them
05:40	tdammers	theoretically speaking at least; practical considerations often make the implementation unfeasible
05:41	tdammers	(e.g., it'd require way more memory than we can provide, or it would take very long to run)
05:42	mungojelly	i want to bound my queues but i have unreasonable amounts of data coming into them, what's my problem called
05:42	opqdonut	see also "turing tarpit"
05:42	emauton	Or it'd be a nightmarish inverted tower of Babel spiralling down into the nine hells of computing. ;o)
05:42	emauton	mungojelly: "overload"
05:43	mungojelly	i know some strategies like (a) forget oldest unprocessed requests (b) drop new requests (perhaps applying backpressure)
05:44	emauton	Have a read of chapter 3 of "Erlang in Anger" by Fred Hebert (it's free online).
05:44	mungojelly	idk i just thought of (c) putting new things into random spots once the queue's full
05:44	mungojelly	emauton: ok thanks will do :)
05:44	opqdonut	dropping random stuff is good
05:44	opqdonut	has all sorts of nice theoretical properties
05:44	tdammers	"load shedding"
05:44	tdammers	^ useful google incantation
05:44	mungojelly	yeah i felt that as soon as i started running it in my mind, i was like huh well then at least it does its best like statistically
05:45	tdammers	yes
05:45	tdammers	if you take great care to shed in "atomic" units, then that's probably the best you can get
05:45	mungojelly	ooh, "load shedding" is a good term for it. that's the opposite of the hell of naming i guess, the heaven of beautiful names.
05:54	ianhedoesit	tdammers: I stepped away for a minute, so this is backtracking a couple screens, but that is interesting. however, for complex types I fear the overhead would be much more significant, no? also, it's much less composable, is it not?
05:55	ianhedoesit	I don't know if I can make a point for the composability of either form, though
05:57	tdammers	I'd argue that it's more composable, because the mental overhead stays constant as your codebase grows
05:57	tdammers	also, you usually use type variables and constraints rather than actual types in your signatures, so your code stays generic and composable
05:58	ianhedoesit	as in Int instead of UserID, but a UserID could still be passed?
05:58	ianhedoesit	or what do you mean by that?
05:58	tdammers	no, more like how, for example, fmap can map over any functor, not just lists
05:59	tdammers	instead of writing :: (String -> String) -> [String] -> [String]
05:59	mungojelly	did y'all see that talk from strangeloop this year about chemical computing? that's like there's just types or just types and environments and the program makes the specific connections for you
05:59	tdammers	we write it as :: Functor f => (a -> b) -> f a -> f b
05:59	tdammers	and with that, you can use fmap on lists, vectors, maps, maybes, eithers, etc.
06:00	amalloy	tdammers: it might be easier to start with map instead of fmap, and explain how the types for that work, because it's more familiar
06:00	tdammers	amalloy: yeah, probably
06:00	tdammers	amalloy: was using fmap because that's more reusable and generic
06:01	tdammers	map is kind of inbetween, in that it takes lists of a variable type, but they still have to be lists
06:01	tdammers	:: (a -> b) -> [a] -> [b]
06:01	amalloy	well yes, but if you don't understand how map can be typed generically (i didn't, when i was starting to figure out functional programming) thenfmap is hopeless
06:02	tdammers	I'm assuming an audience that is familiar with generic functions here
06:08	dragon-irl	Is there a way to get intellij to resolve js calls when using cursive?
06:17	mungojelly	everything about backpressure and granting resources says to make a request queue of things that want the resource but what if you get too many requests to even handle those, how do i like pass out a finite number of request tickets
06:22	ianhedoesit	tdammers: is `map :: (a -> b) -> [a] -> [b] really more simple than `map ([f coll])`?
06:22	tdammers	not by itself
06:23	ianhedoesit	I realize that since I have very little understanding of Haskell, it's more difficult for me to parse it mentally, but seeing that my parameters are "f" and "coll", I assume it wants a function and a collection
06:23	amalloy	the point is that you can write (map 5 *) in clojure, and the compiler doesn't know that's wrong
06:23	tdammers	but checking whether String -> Text is a valid first argument for (a -> b) -> [a] -> [b] is much easier than judging whether (fn [x]) is a valid first argument for (fn [f c])
06:23	ianhedoesit	I know what a and b should be in Haskell's map, but if I didn't know what the map function does from an abstract point of view, I would have no idea
06:24	amalloy	with a static type the compiler can know enough to say "that's not how you use map"
06:24	tdammers	actually, there's not an awful lot the map function can do, given its type signature
06:24	ianhedoesit	amalloy: I know, and I agree that it's a valid complaint, and I worry about that too when thinking about committing time to a dynamic language
06:25	tdammers	it could reverse the list while processing it, or it could completely ignore the function argument and always produce an empty list, or it could stop mapping after reaching a hard-coded limit
06:25	tdammers	but if you add the convention that the function does not add any hard-coded information, nor ignore any of its input, then map's behavior is pretty much completely described by its type signature
06:25	clojurebot	A nod, you know, is as good as a wink to a blind horse.
06:27	ianhedoesit	I really should try to spend more time with Haskell. but I love my lisps the more I use them.
06:29	ianhedoesit	TMA: why only in the past?
06:29	TMA	but then I got to a point where I realized that it was not the static typing that made me uneasy, it was rather the unexpressiveness and burden thereof (C/C++/Java type systems)
06:30	tdammers	those languages have done a lot to set back statically typed languages
06:31	TMA	ianhedoesit: in a sense, dynamic typing is like global mutable state but applied to the metaprogram
06:31	ianhedoesit	I agree to some degree.
06:32	TMA	it can bite you if you do not pay attention (like passing a (fn [a b] ...) to first argument of map)
06:32	mungojelly	every paradigm has a mixture of typing strategies depending on where you look, you're always dynamically determining the meanings of some things
06:33	TMA	but it is hidden (not readily apparent)
06:34	TMA	I caught myself adding (declaim (ftype (function (...) ...) ...)) in my CL programs as a form of documentation
06:35	ianhedoesit	TMA: what are your feelings on things like core.typed, Schema, or Typed Racket, for example?
06:38	TMA	ianhedoesit: I think that gradual typing is somehow THE way; however, I have not yet managed to immerse myself in any of those
06:39	ianhedoesit	hm, thanks.
06:39	ianhedoesit	TMA: what would you say your language of choice is today?
06:39	TMA	like: bang together the prototype and then add the type annotations to hunt down some of the bugs
06:39	ianhedoesit	as far as static typing goes
06:40	TMA	ianhedoesit: it depends on the problem [and the environment]
06:41	TMA	but mostly I write my code in any language at hand, while designing (in my head) the thing in lisp code with haskell types
06:41	ianhedoesit	fair enough
06:42	TMA	unfortunately I forego most of the benefits of the compiler in this way :(
06:42	ianhedoesit	I suppose the answer I was hoping for was something like "try Haskell to feel the power of a proper statically typed language"
06:42	tdammers	"lisp code with haskell types" sounds like a great mode of operation
06:42	TMA	but the interoperability is much more importatnt to ship the result
06:43	TMA	sometimes prolog lends its hand in too :)
06:44	TMA	ianhedoesit: by all means, try it out
06:51	TMA	yesterday, I was hunting down a persistent NullPointerException in Java code, when I finally found the place that was causing it, I really missed my Maybe monad
06:55	spacepluk	hi, I need to build a very big JSON string incrementally by appending small pieces very quickly over a period of time. What would be an efficient clojureish solution for this?
06:55	spacepluk	should I just use interop with StringBuffer or is there a better way?
06:57	tdammers	you could write the JSON to a stream
06:58	tdammers	that way you could avoid having it all in RAM at once
06:58	clojurebot	It's greek to me.
06:59	ianhedoesit	TMA: I do really like Maybe (from what I understand of it)
07:00	tdammers	it's really just null checks lifted to the type level
07:00	ianhedoesit	right, but that's powerful.
07:00	tdammers	although, because Maybe is also a Monad, you get a bunch of things for free
07:01	tdammers	such as short-circuiting Maybe calculations as soon as one of the operands is Nothing
07:01	tdammers	do { x <- Just 1; y <- Just 2; return (x + y) }
07:01	tdammers	or, in applicative syntax, (+) <$> Just 1 <*> Just 2
07:02	ianhedoesit	Haskell's syntax also scares me every time I try to get involved with it.
07:02	ianhedoesit	I also find myself wanting to resort back to simple sexps
07:04	tdammers	Lisp syntax scares me... everything looks the same, even when it's something radically different
07:04	tdammers	;)
07:07	ianhedoesit	I have the opposite fear. having (according to you) `do { x <- Just 1; y <- Just 2; return (x + y) }` and `(+) <$> Just 1 <*> Just 2` (I assume the return statement was left out, but would be after `Just 2`) be the same confuses and upsets me.
07:11	TMA	tdammers: the pointless style scares me; and it even does not solve the problem of getting rid of the parentheses
07:11	ianhedoesit	I'd rather have everything look alike. I like that def, -> and + are three entirely different things, but (def a) (-> a) and (+ a) are all valid
07:12	ianhedoesit	TMA: pointless style?
07:12	TMA	ianhedoesit: pointfree
07:12	TEttinger	ianhedoesit: I think you missed a winky emoticon <tdammers> ;)
07:14	TMA	ianhedoesit: let's assume for a while an exotic clojure dialect where there are partial applications
07:14	ianhedoesit	TEttinger: I saw it but I'm not sure what inflection that's supposed to add
07:14	TEttinger	not serious?
07:15	TEttinger	that inflection
07:16	TMA	ianhedoesit: (f a b c) is just a shorthand for (((f a) b) c)
07:17	ianhedoesit	TEttinger: oh, I assumed he was serious. I figure it's actually pretty common that people don't like that everything looks the same.
07:19	TMA	ianhedoesit: so instead of (defn x [a b c] (f a a b c)) you write (def x (join f)), never mentioning the arguments at all
07:19	tdammers	ianhedoesit: there is no return statement in Haskell. return is a function that "wraps" a value in a monad. it's not a statement, it doesn't return from a function, it "returns a value into a monad"
07:20	tdammers	re everything looks the same: I was half serious there, in that I do think having different things look different is good, and in case of doubt, I prefer sacrificing "same things look the same" for that
07:20	TMA	ianhedoesit: "return" is a haskelese for unit :: Monad m => a -> m a
07:21	TMA	tdammers: in haskell everything (but the type declarations) looks the same
07:21	tdammers	TMA: I would say in Haskell, you have a lot of room for making things look different
07:21	ianhedoesit	TMA: so then with x what do you do? how do you use it if it's defined as just (join f)?
07:22	tdammers	applicative syntax is done entirely in Haskell itself, no special cases in the compiler are needed
07:22	TMA	ianhedoesit: you call it as usual: (x a b c)
07:23	tdammers	re point-free style:
07:23	tdammers	consider map
07:23	tdammers	(map f items)
07:23	TMA	point-free style is awesome to write, but hell to read :-L
07:23	tdammers	now, "map f over something" is very common, it's basically a specialization of map
07:24	tdammers	you could wrap that in a function like so:
07:24	tdammers	(fn [x] (map f x))
07:24	tdammers	the "x" in there is often called a "point" in Math textbooks
07:25	tdammers	"point-free", then, just means that we write the function such that no points appear
07:26	tdammers	which, if clojure functions were automatically curried, would just be (map f)
07:26	ianhedoesit	oh okay.
07:26	tdammers	and in Haskell, we can do exactly that: passing too few arguments produces a new function that takes the remaining arguments
07:27	ianhedoesit	so currying is key
07:27	tdammers	that's one way to look at it, yes
07:27	ianhedoesit	I was mostly just wondering what "(+)", "<$>" and "<*>" were.
07:27	tdammers	oh, that
07:28	tdammers	(+) is just the + operator, but the parentheses allow us to use it in prefix notation
07:28	TMA	"(+) a b" is just another way to write "a + b"
07:28	tdammers	<$> and <*> are library-defined operators from applicative
07:28	tdammers	<$> is an alias for fmap
07:29	TMA	tdammers: like (<$>) = fmap ?
07:29	tdammers	yes, that's exactly how it's defined
07:30	ianhedoesit	what is <*>?
07:31	TMA	holy whatever! why? why? why would you take a fine function with established name and rename it to a haphazardly named operator of unknown precedence and associativity?
07:32	tdammers	(<*>) :: f (a -> b) -> f a -> f b
07:33	tdammers	i.e., it takes a pure function and converts it into a function that works on the same types wrapped in some applicative
07:33	hyPiRion	TMA: Because f <$> a <*> b is easier to read sometimes.
07:33	ianhedoesit	that's another question. so in the above example of `(+) <$> Just 1 <> Just 2`, is `(+) fmap Just 1 <> Just 2`?
07:33	ianhedoesit	equivalent*
07:33	TMA	ianhedoesit: no
07:34	tdammers	fmap (+) Just 1 <*> Just 2
07:34	ane	<$> is defined as an infix operator
07:34	TMA	you would need to wrap the fmap in `fmap` backticks
07:34	ane	if you did (+) `fmap` ... that would work
07:34	ianhedoesit	oh, I see.
07:34	ane	(backticks make anything act as infix)
07:34	ianhedoesit	yeah, I remember that now.
07:34	tdammers	the reason why applicative syntax is so popular is because of the similarity between pure function application and applicative function application:
07:34	tdammers	f x y ==> f <$> x <*> y
07:35	tdammers	the first calls f on two pure values; the second one calls the same function on two values in some applicative functor and returns the result back into it
07:35	tdammers	the function itself is blissfully ignorant of the functor's plumbing
07:36	tdammers	and you can write the whole thing almost like you would write pure code, except that the functor takes care of handling effects for you
07:36	tdammers	e.g., I/O, failure, iteration, etc.
07:36	tdammers	it's kind of like taking function application to a higher abstraction level
07:37	tdammers	(hence the name "applicative functor")
07:37	TMA	and then you come across ((.) <*>) and you give up on haskell
07:37	tdammers	well, of course all of these features can be abused into academic obscurity
07:37	MasseR	TMA: you can always overdo stuff, but that syntax isn't really recommended :)
07:41	TMA	I have learned haskell in 2001 or 2002 IIRC; the simple typing rules have stuck in my head [the not so simple have gone]. I have tried to get back to Haskell since the last year, but I consistently hit the wall
07:42	TMA	most of the "new" stuff seems to be heavily fond of avoiding alphabetic characters and parentheses, except to turn the operators into their prefix form
07:44	MasseR	TMA: Ignore everything about lens and it will be much, much easier
07:44	mungojelly	yeah i like the ideas about typing but i don't quite understand or even believe that the perl syntax has anything to do with them
07:49	MasseR	someString & allVersions %~ succ
07:49	MasseR	as an example
07:50	MasseR	But really, avoid them :)
07:56	ianhedoesit	well I've learned a lot about everything other than Clojure tonight.
07:56	ianhedoesit	so thanks everyone for that. :)
07:59	clgv	aleph question: My IdleStateHandler does not fire IdleStateEvents on the client side - but on the server side it does as expected. What might be wrong there? The pipeline is set up the same way on client and server.
08:01	clgv	$seen ztellman
08:11	tdammers	anyway
08:11	tdammers	the thing I'm missing the most, I think, is the ability to step in at any point in the code and have the compiler figure out the shape of the data I can expect at that point
08:29	crocket	Does clojure enable mediocre programmers to maintain programs of 10k lines or even 100k lines?
08:29	opqdonut	no programming language enables that
08:30	opqdonut	development processed do
08:30	opqdonut	*processes
08:30	damethos	+1
08:31	muhuk	I'd expect 100 kloc programs written in blub to be 5~10k when written in Clojure
08:32	mavbozo	can clojure enable mediocre programmers to create programs of 10k lines before they give up?
08:32	clgv	mavbozo: yeah, copy & paste ;)
08:33	clgv	that's language agnostic
08:33	crocket	What is the economic tradeoff?
08:34	crocket	muhuk, If you're correct, would a C++ program of 140k lines can be written in 7-14k lines?
08:34	crocket	ouch
08:34	mavbozo	i assume mediocre programmers just sprinkle defs everywhere even inside function, or even using atoms as plain-old variables
08:34	crocket	If you're correct, would a C++ program of 140k lines be written in 7-14k lines?
08:34	mavbozo	effectively create imperative programs with clojure
08:35	crocket	mavbozo, That would be terribly programmers.
08:35	crocket	I am a mediocre programmer, and I don't write that kind of code.
08:35	crocket	terrible*
08:36	crocket	I couldn't have written such a beautiful snippet of code as https://github.com/crocket/clj-ddns-client/blob/master/src/clj_ddns_client/core.clj in java or javascript.
08:36	crocket	Even, a mediocre programmer can write such beautiful codes in clojure.
08:37	muhuk	crocket: I'm generalizing. But yes.
08:37	crocket	muhuk, If it includes test codes, wouldn't it double?
08:37	crocket	7-14k -> 14-28k
08:37	crocket	Or, does it include tests?
08:37	muhuk	possibly more than double
08:38	crocket	Test code is code
08:38	muhuk	sometimes my tests are 5 times LOC
08:38	muhuk	I wouldn't included test code when I'm sizing up my codebase.
08:38	crocket	I could probably get less LOC in haskell or OCaml if you include tests.
08:39	crocket	You have to maintain tests as well
08:39	crocket	Since I'm a mediocre programmer, I don't know what functions I should test in clojure.
08:40	xcv	I think it's more a matter of the paradigms people are used to. Imperative + OO is pretty complex to wrap your head around, compared to FP.
08:40	crocket	Some of them just look like internal details
08:40	xcv	It's easy to forget what it was like to learn imperative programming.
08:40	xcv	Just my two cents.
08:40	crocket	Imperative programming is straightforward.
08:40	crocket	OO isn't.
08:41	xcv	Yeah, it is. And if the approach is "hack at it until it works" instead of writing something you basically understand, it can seem quite a bit easier.
08:43	crocket	14*5 = 60k
08:43	crocket	YOu don't save much LOC
08:44	crocket	30k? maybe
08:44	tdammers	test code is code, period.
08:45	crocket	period
08:45	tdammers	in fact, I believe that to get a fair comparison, one should also include the necessary documentation
08:45	crocket	If a C++ program has 140k lines of code, how many LOC would a clojure equivalent including tests have?
08:45	crocket	oh, documentation
08:46	tdammers	which, granted, would be favorable for underdocumented codebases, so one would have to come up with a metric for "sufficient documentation"
08:46	tdammers	but the reason I'd want to include documentation is similar to why I'd include tests
08:47	mavbozo	"everything is an expression" and immutability effectively forces clojure programmers to write code in less imperative style and more towards functional style
08:47	tdammers	mavbozo: it's very easy to opt out of either though
08:47	spacepluk	crocket: wild guess 1/10th of that if you don't need to call a lot of native code
08:48	tdammers	mavbozo: clojure has immutable values, but almost all the other side effects are completely uncontrolled
08:48	muhuk	remove the test code and your program works fine. Similar to generated code/files, test code is second class to actual code.
08:48	crocket	spacepluk, including tests?
08:48	spacepluk	crocket: and assuming it's written in a functional way not just copy the c++ code in clojure
08:49	spacepluk	crocket: do you need to mock a lot of things?
08:49	crocket	mock what kind of things?
08:49	spacepluk	crocket: servers, databases, etc.
08:49	crocket	I don't know
08:49	crocket	yet
08:50	jsabeaudry	Is there a way to not resolve a symbol in a backquote context?
08:50	spacepluk	either way pure functional code is easier to test
08:50	muhuk	oh, wait. I forgot to say "period".
08:50	muhuk	period.
08:50	crocket	Someone wrote a set-theoretic database in C++ of 140k lines including tests. I was just wondering how many LOC I can save with clojure.
08:51	TMA	jsabeaudry: `(... ~'a ...)
08:51	crocket	It's like NoSQL.
08:51	jsabeaudry	TMA, oh sweet, thanks!
08:52	crocket	spacepluk, ^^
08:52	mavbozo	tdammers, what's the state of the art of controlling side effects? static typing?
08:52	TMA	crocket: it depends heavily on the style it is written in
08:52	spacepluk	crocket: I'm sure it will be shorter and be simpler if done right
08:53	crocket	hmm....
08:53	taylanub	mavbozo: usually just not using side-effecting operations, I believe. :-) and monads I guess.
08:53	spacepluk	crocket: at least you get rid of all the c++ boilerplate, build system etc.
08:53	crocket	His code is full of templates which he was proud of because templates fit his needs.
08:54	TMA	mavbozo: or a pure language like haskell
08:54	crocket	Is monad common in clojure?
08:56	muhuk	nope
08:56	tdammers	mavbozo: in clojure? I don't know; "discipline" and "naming conventions", I believe
08:56	tdammers	mavbozo: Haskell uses the type system... the traditional approach is to use monads (or, as the case may be, functors or applicatives)
08:56	spacepluk	you don't really need monads as bad as in haskell
08:56	tdammers	spacepluk: you do, but you can keep them implicit
08:57	crocket	Since I don't know monads, I don't know if I need them.
08:57	tdammers	crocket: you are already using them, they're just not formalized
08:57	spacepluk	tdammers: yeah, I mean for side-effects and that kind of stuff
08:57	tdammers	spacepluk: monads and effects are somewhat orthogonal
08:58	spacepluk	tdammers: what I mean is that you don't need monads to wrap side-effects in clojure
08:58	tdammers	haskell doesn't have side effects at all
08:58	spacepluk	which makes your life easier if you're careful hehe
08:58	tdammers	just effects
08:59	tdammers	but anyway
08:59	crocket	Has any of you used typed clojure? How was it?
08:59	tdammers	you don't need monads for that in haskell either
08:59	tdammers	you do need types that encode effects, though
08:59	tdammers	and those types almost always happen to be monads
09:00	mavbozo	tdammers, state of the art in programming world. In clojure, I resort to some patterns that is described here http://www.infoq.com/presentations/Clojure-Design-Patterns
09:00	crocket	Ok
09:00	tdammers	mavbozo: ah, ok... I'm not sure about that, but I believe most approaches treat effects individually, building separate features for each into their languages
09:01	spacepluk	crocket: are you familiar with other lisps?
09:01	tdammers	e.g., exception handling for failures, comprehensions for lists, "something" for I/O, const-ness checks for mutability, thread control for concurrent mutability
09:01	crocket	spacepluk, no
09:02	crocket	I know a little of emacs lisp, though
09:02	tdammers	but there are a few approaches that try to generalize effects into a uniform pattern
09:02	spacepluk	crocket: I would keep it simple and skip types at the beginning
09:02	crocket	Is there a better way to handle logging than mutating the global logging configuration in a namespaced var?
09:03	crocket	It is not pure
09:03	crocket	I tried to separate logging from business logic, but you can't attach hooks onto protocol methods and multimethods.
09:04	spacepluk	crocket: a macro might help there
09:05	crocket	how
09:05	crocket	https://github.com/technomancy/robert-hooke/ doesn't work on protocols
09:05	crocket	I tried to add logging as a post hook.
09:06	crocket	spacepluk, Can you show me how macro helps?
09:06	spacepluk	I'm not familiar with that library
09:07	spacepluk	but I guess you can make a macro that calls the logger to define the implementations
09:26	crocket	Oi
09:28	dstockton	Oi
09:42	clgv	io
09:45	jonathanj	is there something like readthedocs.org for Clojure projects?
09:45	muhuk	no. But RTD is language agnostic.
09:53	patrkris	hi there. is it possible somehow to make leiningen change the settings in logback.xml when building for production?
09:55	justin_smith	patrkris: for one thing, you could have a different :resources-paths setting for the build
10:02	patrkris	justin_smith: oh yeah, good idea. how come you always have the time to help? :D
10:08	visof	hi guys
10:12	justin_smith	patrkris: I'd probably find the answer to that question depressing
10:12	crocket	I realized there are only a few common types to go with in clojure.
10:12	crocket	Map, list, vector, and set
10:12	crocket	So, you don't worry much about types
10:12	visof	what is the best solution for this: i have a record Foo with :x and :y fields, and i want to apply function f on all elements in this Data structure, #Foo{:x (#Foo{:x (p1 p2 ...) :y (p1 p2 p3....)} ...) :y (#Foo{:x (p1 p2 p3...) :y (#Foo{...})})
10:13	visof	?
10:13	crocket	Types generated by defrecord are sort of treated like maps
10:13	crocket	If defrecord types are not treated like maps, they are type-checked during compile time.
10:14	crocket	How convenient.
10:14	visof	special_map function (special_map function datastructure) ?
10:14	crocket	Partially typed
10:14	justin_smith	crocket: they are maps, as much as any of the other clojure types implementing the interface
10:14	crocket	Dynamic typing is not as painful when all types are map, list, vector, and set.
10:14	crocket	sort of
10:16	taylanub	I find dynamic typing never painful. static typing on the other hand...
10:17	ToxicFrog	crocket: the problem there is that "map" covers infinity types. If you pass a map that is a lookup table of userIDs to user info structs to a function that expects a map that is a user info struct, you're going to have a bad time.
10:17	ToxicFrog	But they're both the same type!
10:18	ToxicFrog	Dynamic typing in and of itself isn't painful, but debugging type mismatch errors that only occur at runtime far from the point of error absolutely is, and if core.typed actually worked I'd be using it for all of my clojure work.
10:18	crocket	ToxicFrog, yes
10:18	crocket	I guess that's where haskell shines.
10:18	crocket	or Ocaml
10:23	tdammers	interesting... I find dynamic typing painful as soon as the codebase reaches a fairly benign complexity level
10:57	skeuomorf	Does core.match allow function applications when matching? e.g. execute `f` if the pattern is "x"?
10:57	skeuomorf	s/applications/application
10:58	clgv	skeuomorf: on the selected branch? why not?
10:59	skeuomorf	clgv: Yeah, on the selected branch, skimming the examples and I didn't see similar usage, so was wondering
11:00	clgv	skeuomorf: just try it ;)
11:00	skeuomorf	clgv: can't afford to spawn a repl currently :)
11:00	clgv	lein try core.match x.y.z
11:19	spacepluk	is it just me or package management in java is completely broken? :(
11:20	spacepluk	every time I add a new dependency I need to track down incompatibilities with joda-time jackson or that kind of libraries
11:20	spacepluk	npm feels like alien technology compared to this
11:20	muhuk	spacepluk: package management in _ is completely broken
11:20	rhg135	It's pretty nice compared to say c
11:20	spacepluk	rhg135: that's true
11:22	spacepluk	muhuk: this wouldn't happen in node/npm because each module can depend on a different version of the same library
11:23	spacepluk	I don't think you can do that in the jvm
11:23	rhg135	As for npm, try upgrading/downgrading node and watch the native module build failures
11:23	muhuk	spacepluk: I don't think that's a good idea.
11:23	spacepluk	rhg135: yeah, you need to wipe the cache node_modules in that case, but you don't need to do package manager work yourself, tracking down versions, etc.
11:24	rhg135	Sometimes they just won't build
11:24	spacepluk	muhuk: why not? I'm having this problem because of implementation details of different libraries
11:25	spacepluk	rhg135: I guess you can't fix c package management hehe
11:25	rhg135	Sad but true
11:27	spacepluk	I'm trying to use storm/kinesis-storm-spout/amazonica/monger together, I looked great on paper
11:28	skeuomorf	spacepluk: oh npm is horrible, maybe not as horrible as Java though, I like nix and guix and in terms of languages, Cargo is great (Rust's de-facto package manager)
11:28	muhuk	spacepluk: because there's no guarantee the effects of having two versions of the same lib will be isolated like that.
11:28	spacepluk	muhuk: yeah, but that's a jvm limitation right?
11:28	spacepluk	skeuomorf: yeah, cargo is very nice
11:30	rhg135	Aren't classloaders meant to isolate classes?
11:31	spacepluk	rhg135: I thought about that, but in this case with storm/thrift's serialization I don't think I can
11:32	rhg135	I've never used either, please explain
11:32	spacepluk	so I'm trying to find a combination of versions that can work together, hence my frustration
11:33	spacepluk	storm is a distributed realtime computation framework
11:33	muhuk	spacepluk: why would you say that?
11:33	spacepluk	the computations must be serializable so that storm can send the code through the cluster
11:34	spacepluk	I don't think I can control how storm loads the classes
11:35	spacepluk	muhuk: what? sorry
11:35	rhg135	Doesn't it use Class/forName?
11:36	muhuk	spacepluk: why do you think it's a jvm limitation?
11:36	rhg135	Unless it does weird stuff with Unsafe
11:36	muhuk	spacepluk: suppose you call f & g in your code. Both of which in turn calls different versions of h
11:36	spacepluk	muhuk: yeah, you're right you can load the same thing in different classloaders, is a limitation of my usecase I guess
11:37	muhuk	spacepluk: how would you compare the results of h?
11:37	spacepluk	muhuk: I think that's ok if h-stuff is not exposed through f&g interfaces
11:37	muhuk	spacepluk: yeah, I was speaking in general terms. Having two versions of the same lib is a bad idea.
11:37	muhuk	spacepluk: it's not OK though, if it's exposed.
11:38	spacepluk	muhuk: I feels like the simplest solution for this specific case though, I just don't think I can do it
11:38	spacepluk	muhuk: agreed
11:39	muhuk	spacepluk: I too wish maintainers would keep releasing with updated deps often. It would make it a lot easier to align such dependencies then. (more options to choose)
11:39	spacepluk	how do you deal with this situations? is there any tool I can use to find combinations of versions?
11:40	spacepluk	muhuk: yeah, I didn't expect so much pain, I've been away from jvm development for quite a long time, and had high expectations about maven and friends
11:40	muhuk	spacepluk: I exclude the older lib, sacrifice the goat and keep my hands inside the vehicle at all times.
11:40	spacepluk	hahaha
11:41	spacepluk	is there something like pactree for maven?
11:45	muhuk	spacepluk: lein deps :tree ?
11:45	rhg135	https://github.com/flatland/classlojure
11:46	spacepluk	muhuk: that's great thanks
11:46	muhuk	yw
11:51	spacepluk	rhg135: that sounds cool thanks, but first I'm gonna try to avoid messing with the class loaders
11:52	spacepluk	hopefully there's a combination that can work
11:52	rhg135	Old versions of software make me cry at night
11:53	rhg135	It makes me feel if I have a bug people will not upgrade till much later
11:53	spacepluk	same here :(
11:54	spacepluk	I'm stuck with clojure 1.6 because of storm ¬¬
11:54	rhg135	It makes programming harder
11:55	rhg135	Have to be loose in your input
11:55	rhg135	Stableish api
11:56	spacepluk	there should be a very big "dependencies obsolete" flag in every package manager so that people feel guilty and upgrade
11:56	spacepluk	it's a lot easier when you do it often
11:56	rhg135	I agree, but sometimes you can't
11:57	muhuk	spacepluk: lein ancient
11:58	spacepluk	muhuk: I tried that with this libs it broke everything hehe
11:58	rhg135	But that's optional :-P
11:58	muhuk	spacepluk: define broke
11:58	spacepluk	the same problem I'm having now basically
11:59	spacepluk	it just breaks at different points
11:59	muhuk	spacepluk: I don't get it. It just reports libs that has newer versions.
11:59	spacepluk	I had a working combination, but then added amazonica and started to dance again
12:00	spacepluk	muhuk: yes, my problem is that the libs I'm using depend on different incompatible versions of the same thing joda-time, jackson, and utility stuff like that
12:02	muhuk	spacepluk: ok, you've probably already figured this out. But FWIW I set the latest versions, and downgrade one by one until I find a working combination.
12:02	spacepluk	in the end I might need to fork one, I'm not sure how to do that with lein/maven
12:03	spacepluk	muhuk: :_( yes
12:10	spacepluk	muhuk: rhg135: thanks for you help, at least now I don't feel stupid
12:10	rhg135	Np
13:05	mavbozo	i use mysql db and suppose I want to deliver this request "Give me a list of phone numbers of people who called in the last 4 days, whose call wasn’t answered and had never called us before"
13:05	mavbozo	usually, at first, I just create a function to wrap the query and parametrized the number of days
13:05	mavbozo	(defn phone_numbers_of_people_who_called_in_the_last_N_days_and_call_not_answered_and_never_called_before [db N] ...)
13:05	mavbozo	is it usual to wrap such query in a function with long name? or are there any other alternative?
13:09	sobel	tons of alternatives
13:09	sobel	use a shorter name
13:10	sobel	that is ridiculous. also, hyphenate not underscore, if you want to be fashionable.
13:10	sobel	that fn name seems to be the spec of the query
13:11	sobel	personally, if it's that important, i'll make a sql function/procedure and mirror it with a clj fcn of similar name.
13:15	mavbozo	oops, i still use old php, js function naming style. old habits die hard.
13:17	sobel	- is an operator in most languages. fortunately clojure has not operators.
13:17	sobel	so lisp(s) are gonna be the odd ones out, using -
13:20	mavbozo	still, i find it really hard to find shorter-meaningful names for query functions that satisfies many specific criteria
13:20	sobel	strict fn naming criteria sounds like like a self-injury to me
13:20	sobel	i don't seek meaning from fn names. just reference value.
13:27	mavbozo	sobel, how do you name your query functions then? do you have some rule of thumbs based on your experience?
13:28	oddcully	mavbozo: why don't you name it what biz calls them? these are just the where clauses put togehter. you could as well name it select_star_from_phonenumbers_where_...
13:39	spacepluk	mavbozo: you can also make a small dsl to build the queries via function composition
13:39	spacepluk	I'm sure somebody already did that
13:41	mavbozo	oddcully, in most cases the biz has these names, but when they don't have ones, i use long name for the functions
13:41	spacepluk	mavbozo: maybe name the function after the reason you need to make the query?
13:43	spacepluk	mavbozo: or use a shorter more general name and put filters as named parameters
13:43	spacepluk	I would hate running into a name like that hehe
13:48	sobel	mavbozo: i call them something short but related to their behavior
13:49	sobel	example: i had a recursive query that calculated and ordered script dependencies. it did a lot of different things but ultimately, it delivered ordered dependencies. it got named get-ordered-deps.
13:49	skeuomorf	,(defprotocol MyProtocol (mutate [this]))
13:49	clojurebot	MyProtocol
13:49	sobel	it's a workhorse query. i deleted SO much procedural code after that query took over its job.
13:50	sobel	but it doesn't need a fancy name.
13:50	mungojelly	is this shortness fetish really appropriate, code is read a lot more than it's written, i'd rather read code written in whole words instead of using my actual brain guessing what "ch" and "prob" and stuff mean.
13:51	skeuomorf	,(defrecord MyRecord [mymap] MyProtocol (mutate [this] (set! (. this mymap) (assoc mymap :hello 2))))
13:51	clojurebot	sandbox.MyRecord
13:51	sobel	fetish implies that it's an aesthetic (sexual!) preference. it's actually very practical.
13:51	skeuomorf	(def x (MyRecord-> {:nah 1}))
13:51	mungojelly	practical how? please don't say efficiency of typing :/
13:51	sobel	but it's optional, if you don't like it, go ahead and type those long names you love to read
13:51	skeuomorf	,(def x (MyRecord-> {:nah 1}))
13:51	clojurebot	#error {\n :cause "Unable to resolve symbol: MyRecord-> in this context"\n :via\n [{:type clojure.lang.Compiler$CompilerException\n :message "java.lang.RuntimeException: Unable to resolve symbol: MyRecord-> in this context, compiling:(NO_SOURCE_PATH:0:0)"\n :at [clojure.lang.Compiler analyze "Compiler.java" 6704]}\n {:type java.lang.RuntimeException\n :message "Unable to resolve symbol: MyR...
13:51	sobel	it's quicker to read
13:52	skeuomorf	,(def x (->MyRecord {:nah 1}))
13:52	clojurebot	#'sandbox/x
13:52	sobel	my editor can autocomplete them but i'll be damned if i want to LOOK at code with over-length identifiers
13:52	skeuomorf	,(x)
13:52	clojurebot	#error {\n :cause "sandbox.MyRecord cannot be cast to clojure.lang.IFn"\n :via\n [{:type java.lang.ClassCastException\n :message "sandbox.MyRecord cannot be cast to clojure.lang.IFn"\n :at [sandbox$eval157 invokeStatic "NO_SOURCE_FILE" 0]}]\n :trace\n [[sandbox$eval157 invokeStatic "NO_SOURCE_FILE" 0]\n [sandbox$eval157 invoke "NO_SOURCE_FILE" -1]\n [clojure.lang.Compiler eval "Compiler.java...
13:52	skeuomorf	,(mutate x)
13:52	clojurebot	#error {\n :cause "sandbox.MyRecord cannot be cast to compile__stub.sandbox.MyRecord"\n :via\n [{:type java.lang.ClassCastException\n :message "sandbox.MyRecord cannot be cast to compile__stub.sandbox.MyRecord"\n :at [sandbox.MyRecord mutate "NO_SOURCE_FILE" -1]}]\n :trace\n [[sandbox.MyRecord mutate "NO_SOURCE_FILE" -1]\n [sandbox$eval181 invokeStatic "NO_SOURCE_FILE" 0]\n [sandbox$eval181 ...
13:52	mungojelly	it's quicker to read made up shorthand than actual words? hm kwl id, fn w/lgwg
13:52	skeuomorf	What's going on here? ^
13:52	spacepluk	I appreciate when code fits in 80chars
13:52	sobel	mungojelly: if you make up shorthand that lame, you deserve what you get
13:53	sobel	mungojelly: do you code?
13:53	spacepluk	names like that make that impossible
13:53	sobel	you're always here dumping on traditions evolved by years of practice and preference without any useful suggestions of what would actually be better
13:53	sobel	even less evidence of why you think these things
13:53	sobel	so i wonder
13:54	mungojelly	it seems to me a lot of these short names are "obvious" once you've simply steeped in this culture for a decade. i can usually guess them. but i use my brain doing that and it excludes newbies.
13:54	sobel	and making a ridiculous extreme out of a very reasonable idea, just so you can make fun of it, is pathetic and rude.
13:54	sobel	no one suggested unreadable or "made up" shorthand except you, for the purpose of belittling a reasonable idea of not making identifiers so long
13:55	sobel	so, i've been using clojure for commercial code less than 6 months, but the nomenclature was pretty obvious from day one. i'm not sure why you think it takes 10 years unless you're unfamiliar with coding.
13:56	sobel	which is ok, i'm just curious
13:56	sobel	this is a good place if you code and need a little support on the language or approaches. i come here for that. i don't understand why you're here, though.
13:56	mungojelly	obviously there's some balance between length and clarity somewhere. of course this level of shortness makes sense to people here, that's why it persists. but it's talking like this in a world of people who'd like to program with us but can't that feels rude to me.
13:56	hiredman	skeuomorf: a few things, but regardless of the other things, that will never work because mymap is not mutable so you cannot use set! to change it
13:57	sobel	mungojelly: it's not a big deal.
13:57	sobel	mungojelly: as matters of programming goes, this isn't. it's just a social question, unless you're trying to exceed the min or max bounds for identifier length.
13:58	skeuomorf	hiredman: hmm, I think? I should probably use an atom and swap its value
13:58	mungojelly	sure it's "just" a social question, just a question of how many and which kind of people are able to read the things we write.
13:58	sobel	mungojelly: bullshit.
13:58	skeuomorf	hiredman: An atom holding a map and swap that map with the new map
13:59	sobel	mungojelly: if you write something, you get to determine what the identifiers are. most people like their work product to be commonly acceptable.
13:59	sobel	mungojelly: let me know how those long ids work out for you when you code with an eye for passing review, peer acceptance, or maintenance by others.
14:00	sobel	until then you just don't get a say because you're proposing telling others how to do things you don't even yourself do
14:00	sobel	be done
14:00	hiredman	there is some syntax to mark fields as mutable
14:01	hiredman	but an atom is going to be nicer to use so I would recomend using that
14:02	hiredman	that compiler error isn't actually related to the mutable field thing, but if you solved the compiler error you would hit the mutable field thing, and atom would deal with both
14:04	kungi	I have the strangest of strange bugs. When I run my web application through the cider emacs repl I get each line of log output exactly once as it should be. When I run it through lein repl on the commandline I get each line exactly three times. The logger currently just uses println. O.o
14:05	hiredman	kungi: cider opens a few emacs buffers, and while some output gets directed to the repl buffer, other output can end up in other buffers
14:06	hiredman	so you maybe printing 3 times in cider too, and the output just ends up in one of the other buffers
14:07	mavbozo	sobel, oddcully spacepluk, mungojelly thank you for the perspectives and insights
14:07	kungi	hiredman: just had a look in every buffer. It doesn't look like there is any stray debug output there.
14:11	oddcully	mavbozo: naming it get-missed-opportunities-phone-numbers makes it easier to compose it internally and don't have it's name beeing wrong after each change in the underbelly
14:14	justin_smith	,(defmacro define-function [& body] `(defn ~(symbol (clojure.string/join \- (flatten body))) ~@body))
14:14	clojurebot	#'sandbox/define-function
14:14	justin_smith	,(define-function [x] (+ x x))
14:14	clojurebot	#'sandbox/x-+-x-x
14:15	justin_smith	,(x-+-x-x 2)
14:15	clojurebot	4
14:15	kungi	hiredman: https://github.com/clojure-emacs/cider-nrepl/issues/266
14:18	mavbozo	oddcully, that's a good name
14:19	mavbozo	justin_smith, :-D
14:23	skeuomorf	hiredman: I see
14:37	skeuomorf	,(defrecord MyRecord [mymap] MyProtocol (mutate [this] (swap! mymap (assoc @mymap :hello 2))))
14:37	clojurebot	#error {\n :cause "Unable to resolve symbol: MyProtocol in this context"\n :via\n [{:type clojure.lang.Compiler$CompilerException\n :message "java.lang.RuntimeException: Unable to resolve symbol: MyProtocol in this context, compiling:(NO_SOURCE_PATH:0:0)"\n :at [clojure.lang.Compiler analyzeSeq "Compiler.java" 6891]}\n {:type java.lang.RuntimeException\n :message "Unable to resolve symbol: ...
14:38	hiredman	that call to swap is incorrect
14:43	skeuomorf	hiredman: Yeah, hastiness in editing the weechat buffer :)
15:30	zxc	Hello! How can I change this so it won't give me a lazy seq? I want just a string
15:30	zxc	http://lpaste.net/142497
15:32	oddcully	zxc: (apply str...) e.g.
15:32	zxc	oddcully: Thank you!
15:56	sagax	hi all!
15:57	sagax	i new be in clojure
15:57	sagax	but interesting,
15:57	skeuomorf	sagax: Hi!
15:57	skeuomorf	sagax: Welcome :)
18:34	triss	when using ArrayMaps apparently "Subsequent assoc-ing will eventually cause it to 'become' a hash-map."
18:34	triss	that kinda sucks doesn't it?
18:34	triss	I really would like a nice array map...
18:35	triss	and when i say something's an array-map i want it to stay that way
18:37	ianhedoesit	I'm confused.
18:38	triss	http://clojure.org/data_structures#Data Structures-ArrayMaps
18:40	ianhedoesit	right, but it won't actually change the type of whatever collection you have. if you need to ensure it's an ArrayMap you can just do (array-map (assoc initial-array-map :keys [values])), no?
18:41	triss	um... yeah i think i could. might have to jus for safeties sake in what I'm p to
18:41	triss	thanks ianhedoesit
18:41	ianhedoesit	you certainly can.
18:42	ianhedoesit	(apologies for the long example to follow)
18:42	triss	maybe i should just use a vector of key value pairs though
18:42	ianhedoesit	,(type (assoc {} :a 1 :b 2 :c 3 :d 4 :e 5 :f 6 :g 7 :h 8 :i 9 :j 10))
18:42	clojurebot	clojure.lang.PersistentHashMap
18:42	ianhedoesit	,(type (assoc {} :a 1 :b 2 :c 3 :d 4 :e 5 :f 6 :g 7 :h 8 :i 9))
18:42	clojurebot	clojure.lang.PersistentArrayMap
18:42	ianhedoesit	,(type (array-map (assoc {} :a 1 :b 2 :c 3 :d 4 :e 5 :f 6 :g 7 :h 8 :i 9)))
18:42	clojurebot	clojure.lang.PersistentArrayMap
18:43	ianhedoesit	woops, the last one should add another kv pair, but either way, the explicit call to array-map will make it an ArrayMap.
18:43	ianhedoesit	a map with more than 10 key value pairs will be a HashMap because it's more efficient than an ArrayMap at that point.
18:44	triss	yup... i understand why HashMap might be implemented that way behind the scenes
18:45	triss	but this is a bit bonkers if you ask me...
18:45	triss	also if i cast the hash map i get given back by the assoc i'll have lost the ordering wont i?
18:45	triss	which is what I'm after
18:52	rhg135	on a tangent, if I wanted to use this logic to build at runtime (ala (apply f ...)) is that possible?
18:53	ianhedoesit	triss: why do you want a sorted map in the first place?
18:54	triss	ah so I've got these maps... and I want to derive values from other values if they aren't specified...
18:55	triss	i put functions in these maps to derive them...
18:57	triss	and some of these values depend on other values that derived via funcions
18:57	clojurebot	It's greek to me.
18:57	ianhedoesit	that sounds complicated and I don't understand why a sorted map is required.
18:57	triss	it kind of is...
18:57	ianhedoesit	but I'm not going to be here long enough to help, unfortunately. good luck.
18:57	triss	thanks man
18:58	triss	so am I the only one that thinks this behaviour is potentially confusing?
18:59	TEttinger	triss: would an ordered map be a better fit?
18:59	triss	the array-map becoming a hash-map thing?
18:59	TEttinger	no
18:59	TEttinger	https://github.com/flatland/ordered
19:00	triss	TEttinger: BOOM! thanks man
19:01	rhg135	It's rather nice to have optimized maps
19:01	TEttinger	triss: if you were relying on either array maps or hash maps having a consistent iteration order, that's probably a bug. I haven't used flatland/ordered yet, but I now strongly prefer using ordered collections in Java
19:02	TEttinger	(try iterating through a java.util.HashMap in Java 7, and in Java 8. different order just about every time. then try with a LinkedHashMap, same order. the latter is the behavior of ordered maps)
19:02	triss	no array-maps are documented as having consistent order.
19:02	TEttinger	yes, but they also have a maximum size, I think
19:02	triss	yup ten....
19:04	triss	it seems quite an easy thing to get confused about... I homestly thought an array-map would just stay an array map when i first saw it...
19:04	triss	And stuff changing type based on size seems nuts to me.
19:05	triss	yes. do it behind scenes for performance....
19:05	triss	but just tell me you're a HashMap...
19:06	triss	do other types change like that in Clojure?
19:06	TEttinger	sorta. Integer can promote to BigInteger if you use the promoting math fns. * won't promote, but *' will
19:07	TEttinger	,(reduce * (range 20 200))
19:07	clojurebot	#error {\n :cause "integer overflow"\n :via\n [{:type java.lang.ArithmeticException\n :message "integer overflow"\n :at [clojure.lang.Numbers throwIntOverflow "Numbers.java" 1501]}]\n :trace\n [[clojure.lang.Numbers throwIntOverflow "Numbers.java" 1501]\n [clojure.lang.Numbers multiply "Numbers.java" 1867]\n [clojure.lang.Numbers$LongOps multiply "Numbers.java" 467]\n [clojure.lang.Numbers ...
19:07	TEttinger	,(reduce *' (range 20 200))
19:07	clojurebot	32416343309932862480078063450197751286360345254602064903101245078980649624910495919277583333729267927978044287550245689991180639587939195917412832736019999345923394479196148258074227897996791388337336270220011364475288934580843359607845024722273008686295523615766901687120288519406781073348991324061273932221644800000000000000000000000000000000000000000000N
19:07	TEttinger	note the N, which is a BigInteger
19:07	TEttinger	,(reduce *' 1.0 (range 20 200))
19:07	clojurebot	Infinity
19:07	TEttinger	ha
19:07	TEttinger	,(reduce * 1N (range 20 200))
19:07	clojurebot	32416343309932862480078063450197751286360345254602064903101245078980649624910495919277583333729267927978044287550245689991180639587939195917412832736019999345923394479196148258074227897996791388337336270220011364475288934580843359607845024722273008686295523615766901687120288519406781073348991324061273932221644800000000000000000000000000000000000000000000N
19:08	TEttinger	it doesn't need a promoting fn if it's already a BigInteger
19:08	TEttinger	,(reduce * 1M (range 20 200))
19:08	clojurebot	32416343309932862480078063450197751286360345254602064903101245078980649624910495919277583333729267927978044287550245689991180639587939195917412832736019999345923394479196148258074227897996791388337336270220011364475288934580843359607845024722273008686295523615766901687120288519406781073348991324061273932221644800000000000000000000000000000000000000000000M
19:08	TEttinger	M is BigDecimal
19:08	Frozenlo`	I'd prefer if * was automatically promoting. (Wasn't it like that before?)
19:08	TEttinger	it was. it caused performance problems
19:09	TEttinger	it was changed back in 1.1 or 1.2 or so?
19:09	TEttinger	there's also the unchecked-whatever fns, not sure when they were added
19:09	TEttinger	which do the standard java behavior of not throwing exceptions on arithmetic overflow
19:10	TEttinger	and just doing weird stuff to the bits
19:10	Frozenlo`	Screw perfomance :-p
19:12	TEttinger	Frozenlo`: just for you:
19:12	TEttinger	,(alter-var-root #'* *')
19:12	clojurebot	#error {\n :cause "Cannot cast clojure.core$_STAR_ to java.lang.Number"\n :via\n [{:type java.lang.ClassCastException\n :message "Cannot cast clojure.core$_STAR_ to java.lang.Number"\n :at [java.lang.Class cast "Class.java" 3176]}]\n :trace\n [[java.lang.Class cast "Class.java" 3176]\n [clojure.core$cast invokeStatic "core.clj" 338]\n [clojure.core$_STAR__SINGLEQUOTE_ invokeStatic "core.clj"...
19:12	TEttinger	,(alter-var-root #'* #'*')
19:12	clojurebot	#error {\n :cause "Cannot cast clojure.core$_STAR_ to java.lang.Number"\n :via\n [{:type java.lang.ClassCastException\n :message "Cannot cast clojure.core$_STAR_ to java.lang.Number"\n :at [java.lang.Class cast "Class.java" 3176]}]\n :trace\n [[java.lang.Class cast "Class.java" 3176]\n [clojure.core$cast invokeStatic "core.clj" 338]\n [clojure.core$_STAR__SINGLEQUOTE_ invokeStatic "core.clj"...
19:12	TEttinger	sigh
19:12	TEttinger	I don't remember how it works
19:14	Frozenlo`	For user-friendliness, I'd argue that * should do the 'obvious' layman stuff, while *' the more machine related perfomant stuff.
19:14	TEttinger	,(alter-var-root #'* (constantly *'))
19:14	clojurebot	#object[clojure.core$_STAR__SINGLEQUOTE_ 0x291e9ae9 "clojure.core$_STAR__SINGLEQUOTE_@291e9ae9"]
19:14	justin_smith	TEttinger: alter-var-root is the same api as swap basically
19:14	TEttinger	,(reduce * (range 20 200))
19:14	clojurebot	32416343309932862480078063450197751286360345254602064903101245078980649624910495919277583333729267927978044287550245689991180639587939195917412832736019999345923394479196148258074227897996791388337336270220011364475288934580843359607845024722273008686295523615766901687120288519406781073348991324061273932221644800000000000000000000000000000000000000000000N
19:16	TEttinger	Frozenlo`: I think also anything that returns different types, with one of them being a primitive or boxed primitive, could be trouble. arraymap and hashmap both implement the same interface, you just return Map or whatever more specific variant they use. I don't think Integer and BigInteger share that?
19:17	TEttinger	there's some specialization on certain primitives in a gnarly chunk of clojure's internals
19:36	Frozenlo`	I have a webserver doing some task each time it receives a POST. Some of them might take some time, and I'd like to organize them in a queue and possibly do some monitoring. Any suggestion for a library?
19:58	Frozenlo`	To answer my own question (logs), https://github.com/Factual/durable-queue seems to do most of what I asked for.
21:33	crocket	If I rewrite a C++ program of 140k LOC, which language between haskell and clojure will result in the least LOC including tests?
21:33	crocket	Or OCaml
21:34	wmealing	this always depends on the task at hand
21:35	wmealing	your C++ is likely heavy class oriented, and your clojure application will be data oriented.
21:36	wmealing	if you're a competent C++ programmer, and a first time other language programmer you likely wont get the advantage of the language unless you know the domain you're writing from very well.
21:37	wmealing	(trying not to be a downer, just from my personal experience, going from C -> (erlang \| clojure \| haskell)
21:37	wmealing	)
21:39	wmealing	crocket: what i found as a useful exercise is to find the core of your problem domain, and try to port part of that to your target language and do a comparison on it.
21:39	crocket	wmealing, Let's assume I have decent expertise in all languages involved.
21:39	wmealing	ok
21:39	crocket	Let's say I have coded with each of them for 2 years.
21:39	wmealing	including C++ right ?
21:39	crocket	yes
21:43	wmealing	i haven't forgotten, just give m e a moment
21:45	wmealing	ok, what i was -planning- to do was take an approach of LOC used in debians benchmark game
21:45	wmealing	http://benchmarksgame.alioth.debian.org/play.html
21:46	wmealing	i know that doesn't include test cases
21:46	wmealing	but it will give an idea of how many LOC for similar problems
21:46	wmealing	across multiple problems
21:57	wmealing	annoyingly there is no single github of source.. they keep it in an issue tracker
21:57	wmealing	https://alioth.debian.org/tracker/?func=browse&group_id=100815&atid=413122
22:22	crocket	How do I attach logging to protocol methods and multimethods?
22:39	neoncontrails	Say you're using Ring as the HTTP server for a figwheel app.
22:39	neoncontrails	Do you boot in figwheel?