#clojure – 2012-10-30

2012-10-30

00:18	jonasen	bbloom: about cljs-412, I think the issues I raised here https://groups.google.com/d/msg/clojure-dev/Ge5uEqlUcJk/_b9xEVrKha8J might be related?
00:20	bbloom	jonasen: yes. there are some interrelated issues
00:20	bbloom	jonasen: var resolution is a bit of a mess at the moment
00:21	bbloom	jonasen: there's a bunch of janky tricks like (resolve … (dissoc env :locals) …)
00:21	bbloom	jonasen: and then resolve-var vs resolve-existing-var
00:22	bbloom	jonasen: and then even more weirdness in that resolve-var can throw a warning, but is sometimes used to generate a different warning, so some odd cases do or don't warn
00:22	jonasen	bbloom: I fixed resolve-var/resolve-existing-var in my fork
00:22	jonasen	bbloom: at least I think I did :)
00:23	jonasen	bbloom: but that's several months ago... ClojureScript has moved on since then
00:23	bbloom	jonasen: heh, yeah and new complexity has crept in to the resolution code paths
00:23	jonasen	bbloom: yes
00:24	bbloom	jonasen: but it looks like there is some good ideas in your branches… could you try to break it up a bit? for example the js/ prefixes
00:24	bbloom	jonasen: that would be a great patch on it's own
00:25	jonasen	bbloom: If I remember correctly, the defprotocol (or was it deftype?) did some very strange things also.. duplicating the resolve-var
00:25	bbloom	jonasen: something is funky in there....
00:26	jonasen	sorry, I mean the defprotocol macro dupicates the logic (badly) of resolve-var
00:26	bbloom	jonasen: a good goal would be to eliminate the cljs-warn-on-undeclared flag and to have it always be enabled
00:26	bbloom	jonasen: we should talk to dnolen and get his buy in on that
00:27	bbloom	jonasen: also, i'd like to see a more robust "warnings" system in general. as it is, warnings are a side effect. it would be nice to have them be represented in the AST or :env
00:31	bbloom	jonasen: regarding the "statics", i think that would be a small and welcome patch to replace them with declare/defs. I see no reason why cljs.core.List/Empty shouldn't just be empty-list
00:31	bbloom	and empty-vector
00:31	bbloom	same goes for the static methods
00:31	jonasen	bbloom: I totally agree with that
00:32	bbloom	jonasen: you're more likely to get your patches accepted if you distill them down to the essence & open separate patches for each. sadly that sometimes means one patch at a time & bugging dnolen :-)
00:33	jonasen	bbloom: The reason for cljs.core.PersistentVector/EMPTY is that it's a straight port from the java source
00:33	bbloom	jonasen: unsurprising :-)
00:33	jonasen	bbloom: but it's the wrong way to do it IMO
00:33	jonasen	bbloom: and it confuses the analyzer
00:34	bbloom	jonasen: yup. that seems like a reasonable starting place. could you work up a patch for just that? open a ticket
00:35	bbloom	jonasen: and if there is a way to get the analyzer to detect that situation: have it emit a deprecated warning!
00:36	jonasen	bbloom: I talked to dnolen about it (a few months ago) and he wasn't too keen on accepting such a patch
00:36	bbloom	jonasen: what did he say about it?
00:36	jonasen	bbloom: I
00:36	jonasen	bbloom: I'd have to search the IRC logs :)
00:38	bbloom	jonasen: http://clojure-log.n01se.net/date/2012-06-20.html
00:39	jonasen	bbloom: that's the one!
00:39	jonasen	bbloom: sorry, gotta run!
00:40	jonasen	bbloom: But I'd love to talk about this stuff more.. there is a lot of fun work with the analyzer still!
00:40	bbloom	jonasen: i don't really buy the "pollution" argument :-P there are LOTS of top levels that should be ^:private
00:40	bbloom	but aren't
00:40	bbloom	jonasen: but cljs doesn't really respect private yet… heh
01:42	tomoj	hmm https://www.refheap.com/paste/3e2aa9f9ae734ba5ac89aa71b clj->el would be nice
03:38	tomoj	why does pmap look at the number of available processors?
03:39	ivan	what else would it do?
03:43	tomoj	https://www.refheap.com/paste/bb37b0cf096f905d8ca995f9c
03:43	tomoj	er
03:43	tomoj	(+ 2 (.. Runtime getRuntime availableProcessors)) is 6, here
03:44	tomoj	yet it appears to use 32 threads at a time due to the chunkiness of range
03:44	tomoj	maybe that's just cus it's chunky..
03:44	tomoj	ah yes
03:44	tomoj	with s/(range)/(iterate inc 0)/ the behavior is as expected
03:56	ebaxt	Hi, I'm trying to extract some information from an XML-file produced by SAP. I'm using clojure.data.zip.xml and trying to figure out how I can do something like the following xpath //ATNAM[text() = "Z_BRAND"]/../ATWRT. I've tried (xml-> l descendants :ATNAM "Z_BRAND" ancestors :ATWRT text) but I get nothing, could someone please point me in the right direction?
04:19	tomoj	datomic.api/touch is confusingly named
04:20	brainproxy	tomoj: how so?
08:20	clgv	,(type {:a 1})
08:20	clojurebot	clojure.lang.PersistentArrayMap
08:20	deg	Does metadata get passed along when, e.g. I create a new map by assoc'ing into an existing one?
08:21	clgv	deg: yes
08:21	deg	very cool!
08:22	clgv	&(let [m (with-meta {:a 1} {:type :some-map})] (type (assoc m :b 2)))
08:22	lazybot	⇒ :some-map
08:22	chouser_log	one of the few ways lists differ from lazy seqs
08:24	clgv	chouser: you mean that lists keep metadata and lazy seqs do not, right?
08:24	chouser	well, each step of a lazy seq keeps its own metadata
08:25	clgv	yes. the implementation would have to pass on the metadata from the head if that was desired
08:25	chouser	&(let [L (with-meta '(a b) {:foo :bar})] (meta (next L)))
08:25	lazybot	⇒ nil
08:25	chouser	ha!
08:25	chouser	&(let [L (with-meta '(a b) {:foo :bar})] (next L))
08:25	lazybot	⇒ (b)
08:26	chouser	oh, next. That returns a lazy seq
08:26	chouser	&(let [L (with-meta '(a b) {:foo :bar})] (pop L))
08:26	lazybot	⇒ (b)
08:26	chouser	&(let [L (with-meta '(a b) {:foo :bar})] (meta (pop L)))
08:26	lazybot	⇒ nil
08:26	chouser	man, maybe I'm just plain wrong.
08:26	chouser	&(let [L (with-meta '(a b) {:foo :bar})] (type (pop L)))
08:26	lazybot	⇒ clojure.lang.PersistentList
08:27	chouser	&(let [L (with-meta '(a b) {:foo :bar})] (meta (conj L 'c)))
08:27	lazybot	⇒ {:foo :bar}
08:27	chouser	&(let [L (with-meta '(a b) {:foo :bar})] (meta (cons 'c L)))
08:27	lazybot	⇒ nil
08:27	chouser	&(let [L (with-meta (seq '(a b)) {:foo :bar})] (meta (conj L 'c)))
08:27	lazybot	⇒ {:foo :bar}
08:28	chouser	&(let [L (with-meta (map identity '(a b)) {:foo :bar})] (meta (conj L 'c)))
08:28	lazybot	⇒ nil
08:28	chouser	ok, enough spam, sorry.
08:29	clgv	humm interesting though
08:29	deg	chouser: Ok, meta-question here... What is the lazybot that seems to be executing your code? Is that magic in some IRC client?
08:30	clgv	the question is if you really need to use metadata on lists in practice
08:30	clgv	deg: it's an irc bot
08:30	deg	Triggered by the "& in front of the expression??
08:30	lazybot	deg: What are you, crazy? Of course not!
08:30	deg	&(+ 40 2)
08:30	lazybot	⇒ 42
08:30	deg	Aha
08:31	clgv	deg: but do try to not spam the channel much. it's good for examples to questions
09:09	brainproxy	tyring again to get slime-ritz / lein ritz working ... anyone successfully using it?
09:26	Cubic	Is there anything in core that would let me grab all fields from a class and expose them in my namespace?
09:26	Cubic	*all static final fields
09:33	clgv	Cubic: no, there is not. maybe there is some lib to do that
09:33	antares_	Cubic: no
09:34	Cubic	Eh, well I guess it's too localized to belong there
09:34	clgv	&(loaded-libs)
09:34	lazybot	⇒ #{bultitude.core cd-client.core cemerick.url cheshire.core cheshire.factory cheshire.generate cheshire.parse clj-config.core clj-http.client clj-http.cookies clj-http.core clj-http.util clj-time.core clj-time.format clojail.core clojail.jvm clojail.testers cloju... https://www.refheap.com/paste/6257
09:35	Cubic	What's the difference between lazy bot and clojure bot?
09:36	clgv	Cubic: lazybot is more advanced ^^
09:36	clgv	right lazybot???
09:36	lazybot	clgv: How could that be wrong?
09:37	clgv	Cubic: the clojure.reflect namespace might help you to write a macro that imports those static final fields to local symbols.
09:38	Cubic	I already wrote that macro, I just wanted to know if there already was one out of pure interest. Though I did it with java.lang.reflect, I wonder if I could do it better with clojure?
09:44	Cubic	Just for reference, it's here: http://ideone.com/AHYm . It's my first nontrivial (for me) macro, so I probably did something wrong. It kinda does what I want it to, but you have to fully qualify the class names. Is there some way to avoid that?
09:45	clgv	"This page does not existPage not found."
09:45	Cubic	clgv: Sorry http://ideone.com/AHYmr8
09:47	clgv	Cubic: you would need to import the class to avoid full qualified class name
09:47	clgv	but since the code is generated why bother?
09:48	clgv	or did you mean the class you give as param to the macro must be full qualified?
09:48	Cubic	No I mean, you have to use the maco like this: (export-static-final-fields java.lang.Integer) rather than (export-static-final-fields Integer)
09:48	clgv	,(resolve 'Integer)
09:48	clojurebot	java.lang.Integer
09:49	Cubic	oh
09:49	clgv	there is the trick ^^
09:49	Cubic	clgv: thanks I'll try that
09:50	clgv	but you could easily skip that. replace (get-static-final-fields (Class/forName (str class))) by (get-static-final-fields ~class)
09:50	clgv	then it should be resolved automatically
09:52	clgv	oh I overlooked your unquoting. well then it's not that easy^^
09:52	clgv	s/overlooked/missed/
09:53	clgv	(get-static-final-fields (resolve class)) should work
10:12	tolsen	maybe this is more of a java question, but why does (/ (Double. 4.0) (Double. 0.0)) give a divide-by-zero error while (/ 4.0 0.0) gives Infinity. I understand the latter is by IEEE standard, but why don't boxed Doubles do the same?
10:14	clgv	tolsen: my guess: the operation on the boxed doubles can check the operands whereas the native operation is just executed
10:15	tolsen	clgv: that makes sense
10:26	tolsen	hmm.. if I write the equivalent boxed double program in java, I get Infinity: http://pastebin.com/rTvhBBMf . Could this inconsistency be considered a bug in clojure?
10:27	nDuff	tolsen: Would you mind using a different pastebin (gist.github.com, refheap, ideone) in the future? pastebin.com is full of flashy, animated ads for anyone not running a blocker.
10:27	tolsen	nDuff: sure. sorry
10:32	TimMc	clgv: People who work in or with the advertising industry have to.
10:32	nDuff	clgv: My last several employers, present included, have made their money off of tasteful advertisements used to promote valuable services. Seeing someone using tasteless ads to promote subpar services still make it big by virtue of their domain name rubs me the wrong way.
10:32	clgv	TimMc: for real? I'd argue that my job performs suffers from it^^
10:33	nDuff	clgv: Anyhow -- if I run adblock, and it prevents me from seeing a bug on a customer site, it's a problem. Been there, done that.
10:33	nDuff	s/to promote/to fund/
10:34	mpan	&(type 4.0)
10:34	lazybot	⇒ java.lang.Double
10:34	clgv	tolsen: you have to have a look at where the exception comes from: clojure.lang.Numbers.divide
10:34	mpan	(= 4.0 (Double. 4.0))
10:34	clgv	tolsen: so it's definitely a clojure bug or feature ;)
10:34	goracio	hi there - suppose we have func with (let ) form let returns nil but if i need to return last expr in let how i can do that ?
10:34	mpan	&(= 4.0 (Double. 4.0))
10:34	lazybot	⇒ true
10:35	nDuff	goracio: let doesn't return nil
10:35	nDuff	goracio: it returns the last thing within it.
10:35	mpan	goracio: put what you want to return in the expr after all the bindings
10:35	goracio	hmm will check ..
10:35	mpan	like, you have let, then a bunch of pairs of names and bindings, and then the expr you want
10:36	mpan	and the overall let-expr will evaluate to the expr
10:36	mpan	for example, ##(let [a 1] 1.0)
10:36	lazybot	⇒ 1.0
10:36	_ulises	##(let [a 1])
10:36	lazybot	⇒ nil
10:36	mpan	I mean, usually you use let because you want to use some of the newly-bound names in the expr, but that's a silly example that doesn't
10:37	_ulises	so let can result in nil
10:37	nDuff	_ulises: I don't think anyone said it _can't_
10:37	nDuff	...not with statements taken in context, anyhow.
10:37	_ulises	nDuff: 'course; I was just curious if it'd fail :)
10:37	_ulises	nDuff: *fail with an exception along the lines of "let needs a body" or something
10:38	_ulises	not being cheeky
10:38	mpan	_ulises: that is very interesting but I don't know what it would be used for
10:39	mpan	I recall someone mentioned that / is treated specially?
10:40	mpan	a while back I was wondering why / existed yet I wasn't allowed to define a / in the current ns
10:43	Bronsa	user=> (defn / [])
10:43	Bronsa	WARNING: / already refers to: #'clojure.core// in namespace: user, being replaced by: #'user//
10:43	Bronsa	#'user//
10:43	Bronsa	user=> (in-ns 'a)
10:43	Bronsa	#<Namespace a>
10:43	Bronsa	a=> user//
10:43	Bronsa	RuntimeException Invalid token: user// clojure.lang.Util.runtimeException (Util.java:170)
10:44	jkkramer	,((fn [& ∕] (let [[/ ∕] ∕] (#'/ / ∕))) 12 3)
10:44	clojurebot	4
10:46	mpan	I'm not sure what to make of that
10:46	mpan	is this generally not advisable to do, then>
10:46	mpan	?
10:47	Bronsa	fwiw it's just a one-line patch to LispReader.java to make that work
10:47	goracio	##(let [a 1] 1.0)
10:48	lazybot	⇒ 1.0
10:48	goracio	is this form valid only here :) ? ##(let [a 1] 1.0)
10:48	lazybot	⇒ 1.0
10:48	mpan	the use of ## is a special instruction to the bot
10:48	goracio	ah i see now :)
10:49	mpan	it says, read-eval this, regardless of its position in the line
10:49	mpan	whereas # must be at the start
10:49	mpan	the rest of that is a standard s-expr
10:50	mpan	btw perhaps I picked bad example values
10:50	mpan	that could have just as easily been ##(let [a 1] "HELLO WORLD")
10:50	lazybot	⇒ "HELLO WORLD"
10:50	mpan	or ##(let [a 1] (+ a 1))
10:50	lazybot	⇒ 2
10:50	mpan	and usually, you'd want to do something with the bindings, as such ^
11:39	sunkencityryleh	I'm wondering how I can do an inject in clojure. I'd like to pick out an element from an array of maps and create a new map with that element as the key and the map as value.
11:40	sunkencityryleh	In ruby i'd use each_with_object or inject if the other method is not available
11:40	sunkencityryleh	(sorry if this is too long) [{id:1,name:'foo'}, {id:2,name:'bar'}].each_with_object({}) { \|a,o\| o[a[:id]] = a } results in: {1=>{:id=>1, :name=>"foo"}, 2=>{:id=>2, :name=>"bar"}}
11:41	sunkencityryleh	now I do it like this in clojure (apply hash-map (flatten (map get-keypair acts))))
11:41	augustl	sunkencityryleh: inject === reduce basically
11:42	sunkencityryleh	augustl: ok I'll take a look at reduce
11:46	augustl	sunkencityryleh: (reduce (fn [curr prev] (assoc curr (:id prev) prev) ) {} [{:id 1 :name "foo"}, {:id 2 :name "bar"}])
11:47	augustl	err a bit weird naming but yeah
11:48	technomancy	is each_with_object new?
11:48	technomancy	looks kind of like a gross non-functional reduce
11:49	technomancy	if you're going to bash around something like that why not just close over it?
11:49	technomancy	oh ruby...
11:51	augustl	technomancy: haha
11:51	augustl	as I suspected, it is added by Rails
11:51	augustl	so it's not in Ruby
11:52	augustl	Rails brings out the best of Ruby, doesn't it
11:53	sunkencityryleh	it's from rails, it's inject but you don't have to do ;memo in the end :)
11:54	technomancy	yeah but what's the point of taking it as a block arg instead of using a closure?
11:55	sunkencityryleh	hm, it's a one-liner for fold-l ?
12:00	sunkencityryleh	I've been dabbling with compojure, one thing I don't get is how middleware is wrapped. doesn't (-> foo bar baz) expand to (bas (bar foo)), but it seems like the middleware is expanded run the other way. i.e. middleware that's last in the (-> expression is run first)
12:01	augustl	sunkencityryleh: correct
12:04	sunkencityryleh	augustl: it seems that there's something funky going on behind the scene there!
12:10	TimMc	foo is the request?
12:12	sunkencityryleh	TimMc: another example: (def app (-> (handler/api app-routes) debug-session1 wrap-session debug-session2) debug-session2 is the middleware that gets run first
12:12	TimMc	gets run first == outermost?
12:12	sunkencityryleh	yes, the last one
12:12	sunkencityryleh	wouldn't -> wrap the other way?
12:13	TimMc	No, this makes sense to me.
12:14	TimMc	If foo were the request, you'd be right.
12:14	sunkencityryleh	hm
12:15	sunkencityryleh	'(-> '(1 2 3 4) rest rest rest first)
12:15	hiredman	sunkencityryleh: if you wrap a box in red, then green, then white paper, when you go to open it what order do you encounter the papers in?
12:15	sunkencityryleh	4
12:15	sunkencityryleh	(first (clojure.core/-> (clojure.core/-> (clojure.core/-> (quote (1 2 3 4)) rest) rest) rest))
12:15	TimMc	It's not about ->.
12:15	sunkencityryleh	ok so what am I missing?
12:15	sunkencityryleh	I
12:15	sunkencityryleh	m completely lost :)
12:15	TimMc	sunkencityryleh: If each of those middleware calls is returning a fn, then the last one in the chain is returning the "outermost" fn, which then gets the reqeust object first.
12:16	sunkencityryleh	TimMc: ah, thanks
12:17	TimMc	In hiredman's analogy, each of those fns is a piece of wrapping paper.
12:17	sunkencityryleh	I need to meditate on this
12:19	sunkencityryleh	hiredman: Ok, now I see it (almost)
12:22	sunkencityryleh	when I macroexpand an expression with -> why is there still clojure.core/-> left
12:22	sunkencityryleh	in the expression
12:22	hiredman	because macroexpand is not recursive
12:23	sunkencityryleh	hiredman: ok tnx
12:26	TimMc	sunkencityryleh: ##(let [incw (fn [f] #(f (inc %))), firstw (fn [f] #(f (first %))), stack (-> identity incw firstw)] (stack [0 2 4 6]))
12:26	lazybot	⇒ 1
12:27	TimMc	Here, identity stands in for the core of your app.
12:29	TimMc	Here's a version that does the wrapping in the opposite order:
12:29	sunkencityryleh	TimMc: tnx
12:29	TimMc	&(let [incw (fn [f] #(inc (f %))), firstw (fn [f] #(first (f %))), stack (-> identity firstw incw)] (stack [0 2 4 6]))
12:29	lazybot	⇒ 1
12:30	bhenry	how can i take everything but the last three items in a vector?
12:31	bhenry	reverse drop 3?
12:31	TimMc	butlast?
12:31	bhenry	butlast only takes the coll
12:31	hiredman	,(doc drop-last)
12:31	clojurebot	"([s] [n s]); Return a lazy sequence of all but the last n (default 1) items in coll"
12:31	bhenry	boom. thanks hiredman
12:31	TimMc	Ah, right.
12:31	TimMc	,(apropos 'last)
12:31	clojurebot	(last butlast drop-last take-last)
12:31	hiredman	checkout the impl of drop-last, it is pretty cute
12:32	hiredman	~def drop-last
12:32	TimMc	Oh, is that the one that uses drop and ternary map?
12:32	hiredman	dunno if that works
12:32	hiredman	TimMc: yeah
12:32	TimMc	Very cute.
12:33	technomancy	oh that is clever
12:41	goracio	can we save 2 values in session ( memory store) ? something like this (session/put! :key val key2 val2)
12:44	thorbjornDX	haha, drop-last :P
13:01	jcromartie	sets are functions
13:01	jcromartie	but functions are sets!
13:02	jcromartie	or close enough
13:02	_ulises	yes! \o/
13:04	bhenry	can things inside of a future start other futures?
13:05	ohpauleez	bhenry: Yes
13:05	ohpauleez	they pull from the same pool though
13:19	technomancy	lynaghk: mind if I pm?
13:25	dnolen	bbloom: ping
13:25	bbloom	dnolen: what's up?
13:26	dnolen	bbloom: any thoughts on CLJS-411, I keep coming up with more cases that are broken. Not feeling so hot about the CLJS-369 changes.
13:26	dnolen	bbloom: there are many more broken cases that have nothing to w/ the top level.
13:27	dnolen	bbloom: part of the issue is that CLJS-369 assumes that lexical scope exists in JS, it does not
13:27	dnolen	bbloom: even if not at the top level if you have multiple lets in the same fn scope and say you conj some fns into atom, it's going to be broken
13:27	technomancy	(require '[clojure.dev.jira :as jira]) (add-watch (jira/issue 369) (partial email bbloom))
13:28	dnolen	bbloom: nested top level dos w/ lets are broken, very bad for macros
13:28	bbloom	dnolen: bah. sorry about this.
13:28	dnolen	bbloom: I'm kicking myself for not seeing this issue much earlier.
13:29	bbloom	dnolen: give me a few minutes to investigate/think
13:29	dnolen	bbloom: k
13:31	dnolen	bbloom: an ideal fix would be perhaps somehow always gem-syming let bindings only in compiler.clj - thus punting the issue to backends - some languages will have sane scoping, some won't.
13:31	bbloom	dnolen: yup. my primary motivation was to allow analysis to happen BEFORE introducing insane semantics :-)
13:32	bbloom	dnolen: by that i mean downstream analysis, like my CPS transform
13:33	dnolen	bbloom: yes, I understand the utility of the fix, but we absolute need to gem-sym bindings for JS unless we introduce significantly more complicated backend machinery - I'd like to have a fix by Friday since this is a pretty nasty issue for all users of CLJS
13:33	bbloom	dnolen: ok. are you going to be around irc for a little while? i've got a couple random things to take care of, but i'll find some time in the next few hours to dig into this
13:34	dnolen	bbloom: I'll be around, thx much.
13:59	uvtc	Why doesn't the clojars page for quil https://clojars.org/quil display its project url? Quil's project.clj file has :url set, and the version is 1.6.0.
13:59	uvtc	And 1.6.0 is what's displayed at clojars.
14:02	uvtc	Sorry, perhaps I should ask that on a different channel.
14:03	TimMc	Was :url added after the last release?
14:04	TimMc	Check the project.clj in the JAR artifact from clojars.
14:04	uvtc	Oh, right, TimMc. Good eye dear.
14:08	bbloom	"One's sentiment toward JavaScript flips between elegance and disgust without transiting intermediate states." -- http://matt.might.net/articles/javascript-warts/
14:08	bbloom	truer word have never been spoken :-P
14:09	Frozenlock	I'm storing a java class in a `let' and using it further in my code, but I get an error. Is there something that prevents me from doing it? Some kind of special form behavior? (let [my-class some-class] (new my-class some-argument))
14:10	bbloom	dnolen: does Herwig's patch fix the issue? http://dev.clojure.org/jira/secure/attachment/11606/0002-CLJS-401-gensym-let-names-in-statement-context.patch
14:11	bbloom	dnolen: maybe go with that and back out aea65a693 ?
14:12	TimMc	Frozenlock: I think new takes a literal symbol.
14:12	TimMc	&(let [foo StringBuffer] (new foo))
14:12	lazybot	java.lang.IllegalArgumentException: Unable to resolve classname: foo
14:13	bbloom	dnolen: hacky gensyms seems like the only quick fix. anything more elegant would require deeper JS analysis
14:13	bbloom	"el-cheapo gensyming" as he describes it, heh
14:13	Frozenlock	TimMc: Is there a workaround? I don't know in advance what will be the class...
14:14	TimMc	j.l.Class has some appropriate constructor methods.
14:14	Frozenlock	Thanks!
14:14	TimMc	&(let [foo StringBuffer] (.newInstance foo))
14:14	lazybot	⇒ #<StringBuffer >
14:15	Frozenlock	Magic :)
14:15	TimMc	Frozenlock: And I would bet that there are some nice helper methods in the Clojure compiler that you could use for non-nullary constructors.
14:17	dnolen	bbloom: yeah that patch actually look ok
14:18	tgoossens	Hi
14:18	tgoossens	I'm here to epxress my frustrations
14:18	dnolen	bbloom: thanks for giving it a look, i'll give it a shot and add some more tests cases later.
14:18	tgoossens	At university I have to do (obligated) teamwork in java
14:19	bbloom	dnolen: then let's go with that so we un-break everybody. revert aea65a693 and apply that patch.
14:19	tgoossens	And
14:19	uvtc	hi tgoossens .
14:19	tgoossens	the rest of my team just doesn't get it. How to write simple code
14:19	tgoossens	apparently they like
14:19	tgoossens	instanceof
14:19	tgoossens	and switch case
14:19	bbloom	dnolen: that also means that the :shadow keys aren't necessary in the analyze phase, but we can address that later
14:19	tgoossens	so now. We have monstercode
14:19	tgoossens	that 1) doesn't work
14:20	tgoossens	2) is hell to debug
14:20	tgoossens	But I can't seem to convince them
14:20	tgoossens	it seems like "if its not complicated it is not smart / good"
14:20	uvtc	tgoossens: Do you have a specific question about Clojure?
14:20	dnolen	bbloom: I think we still need that for namespaces right?
14:21	technomancy	tgoossens: that's Java 101. anything simple is viewed with suspicion.
14:21	bbloom	dnolen: oh ugh yeah, duh, right
14:21	tgoossens	lol
14:21	tgoossens	No. Just implicitly expressing that I like the way things go in clojure :D
14:22	uvtc	Aye, she's a fine ship she is.
14:22	mpan	tgoossens: not inherently wrong to use instanceof or switch, though I'm sure -- like most tools -- they can be misused
14:23	mpan	and if your model for hierarchical relationships is expressed using inheritance, well you've got to put distinguishing logic somewhere, whether as overloaded methods, externally in some control flow block, etc
14:25	Frozenlock	TimMc: You example works fine, but for my class I get an error saying that it doesn't have the .newInstance method o_O
14:25	TimMc	Then you don't have a Class.
14:26	TimMc	or you're trying to give it more args
14:26	mpan	just curious what you get if you ask it for its type
14:26	bbloom	dnolen: when you're satisfied that this issue is resolved, i'd like to discuss my experiences working with the compiler with regard to :env, state, codeine, etc. i think i understand the issues better now & this lets issue helped my understanding a bit too
14:27	bbloom	s/codeine/codegen/
14:27	tgoossens	mpan: Please don't tell me this is good: http://pastebin.com/edNSBXFX
14:28	dnolen	bbloom: sure, I'm pretty sure Herwig's patch will fix the issue. It just pushes gensym into the compiler and out of analysis.
14:28	bbloom	dnolen: yup, let me know how it goes
14:28	dnolen	bbloom: rather you patch removed it, and Herwig's patch put it back in the backend.
14:28	mpenet	Raynes: can lazybot run on password protected servers? (I would like to run it on a grove.io server)
14:28	mpan	tgoossens: meh, not particularly emotional about it one way or the other
14:28	tgoossens	hehe
14:28	mpenet	Raynes: it seems it is not the case after a quick look, but I prefer to ask before diving into the source
14:28	dnolen	bbloom: if you're referring to some of the thoughts you were working out earlier go for it.
14:29	bbloom	dnolen: that's what i get for trying to be clever :-)
14:29	tgoossens	but it's out of context of course
14:29	mpan	you could perhaps convince them to refactor it to be more OO if you shift a lot of the responsibility on the individual objects themselves
14:29	tgoossens	problem: time damnit
14:29	bbloom	dnolen: so i ran into an issue with :env in a few odd cases. for example, it's perfectly valid to have let statements that don't have a body (let [x 1])
14:30	bbloom	dnolen: in such a case, there are no body statements to get the :env from
14:30	bbloom	dnolen: if you need the :env in which x is a local, you don't have an easy way to get that
14:30	mpan	tgoossens: I figure, just for me personally, my past employment has drastically relaxed my standards for the sort of code which will offend me personally, but I suppose that's getting off-topic at this point
14:30	TimMc	switch statements are always trouble.
14:30	bbloom	dnolen: one possible solution would be to say that there's really an implicit do there: (let [x 1] (do))
14:30	TimMc	Sometimes there's not much you can do, though.
14:31	bbloom	dnolen: however, a similar situation applies to def
14:31	tgoossens	oh
14:31	tgoossens	don't forget
14:31	tgoossens	the code isn't working :p
14:31	mpan	well, that's a separate issue which you all must work hard to resolve
14:31	bbloom	dnolen: so my thought is that there really needs to be a before and after :env
14:31	TimMc	If Java had syntactic abstraction, you could write "xcase", which would be like "case" with an implicit break at the end of each clause.
14:31	mpan	it will occur regardless of your choice of language or style
14:32	bbloom	dnolen: @namespaces is really just the after-env
14:32	bbloom	dnolen: does that make sense so far?
14:33	bbloom	dnolen: i need to step out for a minute, so i'll just say the last bit and hopefully that makes sense without clarification
14:34	bbloom	dnolen: if the codegen phase emitted AST nodes instead of having the side effect of printing code to out, then we could have a similar thing for JS
14:34	bbloom	dnolen: emitting a var could have an after-env with a binding in it
14:35	bbloom	dnolen: which would have made this whole issue trivial: we could have looked at the js-env side-by-side with the cljs-env to compute the shadowed symbol
14:36	dnolen	bbloom: yeah not completely following. you should probably write this up somewhere.
14:37	dnolen	bbloom: also probably needs a more interesting use case then gen-syming let bindings for me to see the point.
14:37	bbloom	dnolen: eh, ok. i'll put that on the list of things to write up when i release my CPS transform (assuming i ever get it working robustly enough to announce it)
14:37	dnolen	then avoiding gen-syming I mean.
14:38	hiredman	emitting strings is pretty meh
14:38	bbloom	hiredman: yeah, i looked into emitting GClosure AST nodes & it looks promising, but i've got too many random projects going on at once already :-P
14:39	bbloom	gotta run
14:41	hiredman	emitting gclosure ast nodes directly is not an alternative that occured to me, very interesting
14:43	tufflax	I'm doing (apply conj a-coll a-possibly-empty-list) but when the list is empty it's an error, but I simply don't want to add anything. Is there a nice way to work around that problem?
14:45	ro_st	does (partial) evaluate its args?
14:45	TimMc	It's a fn, so yes, the args are evaluated.
14:45	jonasac	is there are reasonable way to tame the stacktraces from running lein test, everytime a test fails my screen fills up
14:46	ro_st	ok so if i want to cache a result locally, i needn't do so separately if i'm using it in a call to partial
14:47	technomancy	jonasac: yeah, but it involves convincing chouser to release longbottom
14:48	mpenet	Raynes: nevermind, I managed to get it working
14:48	jonasac	technomancy: btw thanks for leiningen, new to clojure, was having a really rough time with everything until i met leiningen last night :)
14:49	technomancy	heh; cool =)
14:49	zerokarmaleft	tufflax: i think you want reduce instead
14:50	tufflax	zerokarmaleft yeah that works thanks
14:51	dnolen	hiredman: it could be interesting, tho it's not at all clear if AST rep is something you can expect to be stable.
14:51	technomancy	wow, sometimes you just don't realize how good you have it: https://mobile.twitter.com/jreichhold/status/263338857397907456?p=v
14:52	AdmiralBumbleBee	java programmers have it good
14:52	egghead	hey Raynes, I'm playing with conch for a little process manager i'm working on, is there any way to get it to find things in the project/resources? (sh/proc "interpreter" "somefile.script") searches project/somefile.script instead of project/resources/somefile.script
14:52	technomancy	job security?
14:52	AdmiralBumbleBee	they're not using C, that's a pretty massive thing to be happy about
14:54	ro_st	technomancy: find-file-in-project is now precisely what i need. thanks for your help yesterday
14:54	ro_st	and you, mpenet
14:54	Frozenlock	TimMc: http://stackoverflow.com/questions/9167457/in-clojure-how-to-use-a-java-class-dynamically?lq=1 The constructor fn solved all my problems :)
14:54	hiredman	dnolen: yeah going to gclosure's ast had not occured to me, I was thinking more a long the lines of delaying generating strings as long as possible, generate some clojurescript defined js ast that can be further manipulated before being turned in to strings (or the gclosure ast)
14:54	Frozenlock	/s/constructor/construct
14:55	ro_st	AdmiralBumbleBee: see if you can find a way to use juxt to improve your mood. i'm pretty sure that's one of the things you can use it for.
14:55	technomancy	I don't use it very often
14:55	ivan	surely it's not a leak if you lose your references to the map?
14:56	uvtc	juxt?
14:56	clojurebot	juxt is a little hard to grok but it's the best thing ever
14:56	ro_st	i still don't understand juxt
14:57	ro_st	macros are easier
14:57	normanrichards	anyone know how to invoke a db function using clojureql? for example, something to the effect of: (conj! events {:name "blah" :occured_at "now()"})
14:57	bbloom	,((juxt inc dec) 5)
14:57	clojurebot	[6 4]
14:57	bbloom	that's the best example i've come up with for juxt
14:58	ro_st	ok, so return a seq of the results of each successive fn to the args
14:58	uvtc	The example that the just docs include makes sense to me: ((juxt a b c) x) => [(a x) (b x) (c x)]
14:58	ro_st	yeah
14:58	technomancy	ro_st: no, return a function that calculates such
14:59	ro_st	oh, yes
14:59	technomancy	,(map (juxt inc dec) (range 5))
14:59	clojurebot	([1 -1] [2 0] [3 1] [4 2] [5 3])
14:59	tomoj	&((juxt min max) 1 2 3)
14:59	lazybot	⇒ [1 3]
14:59	bbloom	uvtc: that makes sense to me too, but not everybody has mastered what i'd call "haskell signature reading skills"
14:59	ro_st	where have you found good use for it, technomancy?
15:00	ro_st	wait, i can use juxt to turn (map fn1 (map fn2 coll)) into (map (juxt fn1 fn2) coll) right?
15:00	technomancy	it works great when paired with destructuring
15:00	technomancy	you can turn 3 lines in a let into one
15:00	technomancy	ro_st: no, that's comp
15:00	ro_st	oh. right
15:01	TimMc	francis-: clojure.lang.Reflector/invokeConstructor looks good.
15:02	TimMc	francis-: Sorry, that was meant for Frozenlock.
15:02	dnolen	hiredman: it could be interesting, but it just sounds like writing a lot more code and adding another step since we still probably want to go through Closure. hard to see a big benefit.
15:03	bbloom	dnolen: i pseudo-code-ported a few of the uglier emit methods to pure functions & they saw substantial reductions in size.
15:03	bbloom	dnolen: it also lets you reuse more logic b/c you are more free to re-arrange the order of evaluation without fear of side effects. that means even more savings when applied across multiple emit methods
15:03	bbloom	dnolen: clearly very low priority, but IMO worth exploring
15:06	hiredman	it is kind of amusing that compilers are often pointed to as an example of functional programming, but clojurescripts compiler is not a function
15:06	dnolen	bbloom: yeah, side effect aspect of emitting strings sucks. tying in to closure AST might be worth it if the win is big enough - tho I suspect it will means we won't be able to track closure compiler so nimbly.
15:08	bbloom	dnolen: there is a helper method interface, IIRC called IR.java, which seems to be reasonably stable, but i really didn't dig in too much
15:08	dnolen	bbloom: http://code.google.com/p/closure-compiler/source/list?path=/trunk/src/com/google/javascript/rhino/Node.java&start=2249, this is the relevant file no?
15:09	bbloom	dnolen: one of them
15:09	bbloom	dnolen: the Node class has lots of pseudo nodes for optimizations, etc
15:09	bbloom	dnolen: see also http://code.google.com/p/closure-compiler/source/browse/trunk/src/com/google/javascript/rhino/IR.java?spec=svn1996&r=1996
15:10	dnolen	bbloom: so that file the only one we would need to deal w/?
15:11	bbloom	dnolen: maybe? i looked at this a long while ago
15:11	bbloom	dnolen: this is also why i asked about the js* form
15:12	bbloom	dnolen: b/c there is no obvious way to do the js* form in a glosure AST world
15:12	bbloom	dnolen: however, i did come up with some hack to emit a js fragment, so that you could incrementally move towards a functional approach (i.e. one emit method at a time)
15:14	dnolen	bbloom: I agree this looks intriguing. But probably a lot of potholes. I'm curious if this would make compile times w/o optimizations slower / faster.
15:14	bbloom	dnolen: my guess? initially, slower, but ultimately faster
15:14	bbloom	dnolen: slower since you go from cljs ast -> js ast -> string -> js ast -> string
15:15	bbloom	dnolen: but once all of the emitter is converted, you can call the compiler directly
15:15	bbloom	dnolen: you won't need to run the js parser
15:15	dnolen	bbloom: the slowest thing about closure isn't the parser - it's advanced optimization. it dominates everything.
15:16	bbloom	dnolen: i believe that :-)
15:16	bbloom	dnolen: we could probably also get free source maps
15:16	bbloom	dnolen: i believe the gclosure Node class is semi-mutable
15:16	bbloom	dnolen: with the only mutable bits being things like line/column metadata
15:16	dnolen	so in the worst case everything everything else gets slower, advanced optimization roughly the same.
15:17	dnolen	bbloom: yes free source maps would be nice.
15:17	bbloom	we could pretty easily wrap the static public methods if IR.java to call .setLine on each node
15:18	bbloom	dnolen: i don't think worst case is slower. i think slower is just an intermediate step
15:18	bbloom	once you're a 100% AST step, you cut out the disk completely
15:18	bbloom	and there is no to string or from string intermediate step
15:18	bbloom	which, intuitively, has to be faster? right :-P
15:19	bbloom	but like you said, advanced optimizations remain the bottleneck
15:19	bbloom	so the primary advantages would be functional purity, code size reduction, ease of reasoning, and free-ish source maps
15:19	bbloom	clear disadvantage would be the lose of js*
15:19	bbloom	however, it could be replaced by js** which returns AST nodes ;-)
15:20	goracio	my compilation time of cljs file with 200 loc takes 20-25 sec usually with simple or whitespaced optm
15:20	dnolen	bbloom: right
15:20	tomoj	goracio: with a fresh jvm?
15:20	bbloom	i have a branch somewhere that i attempted to eliminate js*
15:20	bbloom	let me look...
15:20	goracio	but cljbuild auto take 4-5 sec
15:20	tomoj	ah right
15:20	dnolen	goracio: you mean 25000+200 lines of JS
15:21	goracio	dnolen: i guess so don't understand why 25k needed
15:21	bbloom	dnolen: see https://github.com/brandonbloom/clojurescript/commits/no-js-form
15:21	aaelony	n
15:22	hiredman	I have a 50 line cljs file that takes 17-20 seconds in andvanced mode
15:22	goracio	tomoj: fresh jvm ? 7 version i use
15:22	dnolen	goracio: because you 200 lines of CLJS wouldn't do anything w/ the other 6500 lines of cljs.core
15:22	dnolen	w/o
15:22	bbloom	dnolen: it would also be nice to be able to memoize the output of a compilation unit
15:22	bbloom	dnolen: so in the case of core, we could keep the AST in memory :-)
15:23	dnolen	bbloom: that's true
15:23	bbloom	see those last 3 commits on that branch
15:24	goracio	dnolen: 6500 ok and what for others ? _))
15:24	dnolen	goracio: sorry, not following. others?
15:24	TimMc	dnolen: 6500 < 25000
15:25	TimMc	15:20 < dnolen> goracio: you mean 25000+200 lines of JS
15:25	goracio	dnolen: 25k lines minus 6500 18,5k left
15:25	dnolen	goracio: sorry just mistyped.
15:25	bbloom	dnolen: step one is minimizing usage of js*
15:26	dnolen	goracio: every CLJS app needs to load the other >6500 lines of cljs.core. w/o any optimization this will become 25000 lines of JS
15:26	bbloom	dnolen: which is already a goal :-)
15:26	TimMc	goracio: It's just like how your Clojure program won't run without the Clojure JAR file being included.
15:26	bbloom	although js* to js/ isn't much of a win, heh
15:26	goracio	actually it's google closure code if we look at it in browser
15:26	dnolen	bbloom: yeah, I think there's like 5 places in core, and the only other place we use it for inlining low level ops.
15:26	bbloom	dnolen: yup, i think we reaaaallly need better inlining support in cljs
15:26	TimMc	Except with CLJS it can't be linked.
15:27	bbloom	dnolen: we do a lot of inlining hacker via macros, where true inlining could help
15:27	dnolen	bbloom: why?
15:27	bbloom	dnolen: b/c i feel like a human compiler
15:27	bbloom	dnolen: isn't that reason enough to do anything in the lisp world? :-)
15:27	hiredman	silly bbloom, humans are computers, not compilers
15:28	dnolen	bbloom: have you needed to do much inlining w/ when writing CLJS?
15:28	bbloom	dnolen: no, just in core
15:28	TimMc	hiredman: Nonsense, I've been asked to compile a report before at work.
15:28	dnolen	bbloom: so my question remains, why?
15:28	ivan	V8 inlines small functions
15:28	hiredman	TimMc: as a computer you where asked to run a compiler routine
15:28	hiredman	were
15:29	dnolen	I haven't heard of anyone doing writing their own inlining macros in CLJS since we cover the bases
15:29	bbloom	right, core takes the suffering so that real apps don't have to
15:29	bbloom	but i want to eat the cake too: we shouldn't need a macro-version of every major bit operation in core :-)
15:30	dnolen	bbloom: in so many ways not a priority. That likes 100 lines of code in core.clj
15:30	bbloom	dnolen: i wasn't proposing it as a priority :-)
15:30	goracio	so with clojure it appears that we can make pretty big apps without classes and objects stuff ))
15:31	bbloom	dnolen: i think minimizing js* usage should be, however
15:31	goracio	just wrote pretty big app from scratch took about 2 weeks and it was pretty easy and fun
15:32	dnolen	bbloom: yes, I think it appears in only 5? 7? places in core.cljs now. happy to take patches the remove the remaining cases that can be removed.
15:32	bbloom	dnolen: what about core.clj ?
15:32	bbloom	dnolen: did you look at that random old branch i linked to?
15:33	dnolen	bbloom: yes we already talked about that - the inlining macros are the other case.
15:34	bbloom	dnolen: two separate issues. inlining vs js* usage. the js* operates on strings, not AST nodes
15:34	bbloom	dnolen: the inlining macros have various bugs, btw
15:34	bbloom	dnolen: max and min both evaluate their arguments multiple times
15:35	dnolen	bbloom: patches welcome ;)
15:35	bbloom	dnolen: instanceof evaluates it's arguments in reverse order
15:36	bbloom	dnolen: i apparently can't contribute any patches without simultaneously breaking everything ;-)
15:36	dnolen	bbloom: heh, low hanging fruit & compiler improvements not the same kind of patch :)
15:36	dnolen	bbloom: I'd happily take inlining macro fixes as one patch
15:37	bbloom	on a tangential personal note: i quit my job. anybody looking for a clojure dev in seattle? :-P
15:38	dnolen	bbloom: too bad you're not coming to the Conj
15:38	TimMc	bbloom: I'll keep an ear out. Does Seajure still exist?
15:39	emezeske	goracio: I'm not sure "pretty big app" and "in two weeks" are compatible facts :P
15:39	bbloom	TimMc: it does. i haven't been to it yet and can't go to the next one on thursday, but they apparently happen on the first thursday of each month. i'll try to make it to the next one
15:40	hiredman	next seajure meeting is this thursday at u. zoka
15:40	bbloom	hiredman: i've already volunteered to help out at the TechStars demo day on thursday; and i'm playing in the startup poker tournament that night too :-)
15:41	hiredman	I may or may not be there, and I may or may not have a a clojure powered robot
15:41	TimMc	bbloom: Is moving to Finland an option? :-P
15:41	goracio	emezeske: it' all depends of course but if consider that it is my first project in clojure and func lang then it's ok
15:42	bbloom	TimMc: no, but SF or NY are on the table
15:42	tomoj	bbloom: I'd love to hire you
15:42	TimMc	bbloom: I think ohpauleez's company does some Clojure stuff in NY.
15:42	tomoj	but can't afford you right now
15:43	bbloom	tomoj: heh :-)
15:43	bbloom	TimMc: paul moved to portland
15:43	TimMc	oop
15:43	emezeske	goracio: By all means, be happy with your first project! I'm not trying to trivialize it. I just wanted to point out that it's probably not really a data point for "clojure is suitable for large programs."
15:44	tomoj	bbloom: hopefully that will change in ~3 weeks
15:44	bbloom	tomoj: what's your startup?
15:44	bbloom	(i presume it's a startup, if you're expecting a financial event like that, heh)
15:45	TimMc	Why do you assume a startup? Maybe it's for homework help. :-P
15:46	uvtc	hehehe
15:46	bbloom	the ~3 weeks comment implied a financing to me, but maybe i've just been around too many startups
15:47	TimMc	Or maybe tomoj runs a turkey farm, and the Thanksgiving sales will provide money for the rest of the year.
15:48	bbloom	oh, this game is fun... maybe tomoj is planning a bank heist
15:49	hiredman	maybe he has been in contact with some helpful nigerians
15:49	bbloom	are there unhelpful nigerians?
15:49	goracio	emezeske: so i mean it appears that good apps can be written with only functions without classes and objects and inheritance stuff and so on
15:50	goracio	emezeske: i think oop is overcompicated and there is a choice like clojure that is good )
15:53	emezeske	goracio: I agree.
15:53	tomoj	bbloom: pm?
15:55	bbloom	tomoj: anytime
15:58	Frozenlock	I have a java class that requires a float. Is there a way to make it accept any number and convert it on-the-fly?
15:59	bbloom	,(doc float)
15:59	clojurebot	"([x]); Coerce to float"
16:00	Frozenlock	er.. yes.. I'm searching a more universal answer, I have many of those class requiring specific datatypes.
16:01	andrewmcveigh	If I want to use a worker queue in clojure, is the best option still a java BlockingQueue, or similar?
16:01	bbloom	Frozenlock: depends. do you have an interface to reify? can you just simply use a function instead of a method?
16:03	Frozenlock	Well I could wrap each class in a function, each making sure the type is correct and coercing them if needed.
16:03	hiredman	andrewmcveigh: nothing wrong with a nice linkedblockingqueue
16:03	Frozenlock	I have ~15 of these classes.
16:04	andrewmcveigh	hiredman: cheers. It's how I did it last time. Just wanted to make sure I'm not missing anything new/better/etc., in the Clojure world.
16:05	TimMc	Frozenlock: What is it?
16:05	Frozenlock	bacnet4j
16:12	jcromartie	every time my friend hits me up with some kind of programing problem, it's something Clojure solves already… I'll get him one day
16:24	TimMc	Frozenlock: Sounds like a Clojure wrapper might be really useful.
16:26	jonasac	can somone point me in the right direction for using (binding) im following the examples, but still getting an IllegalStateExeption in the repl
16:26	tomoj	bbloom: so cljs-cps could be used to implement something like http://taskjs.org/ , but is more powerful, right?
16:26	Frozenlock	TimMc: Yeah I think also... all these classes and subclasses are becoming really heavy in my code.
16:28	jcromartie	jonasac sure
16:30	bbloom	tomoj: precisely
16:32	bbloom	tomoj: the goal is more akin to C#'s async/await keywords
16:32	AntelopeSalad	if i do... (:require [compojure.response :as response]) , why doesn't it work when i call response/header later on?
16:32	tomoj	huh
16:32	tomoj	async/await seem like task.js to me
16:32	jonasac	jcromartie: if i do (def x0) and then call (binding [x 1] (var-get #'x)) i get an IllegalStateException saying "Can't dynamically bind non-dynamic var"
16:33	AntelopeSalad	note: the code above is taken directly from the official github page for compojure
16:33	tomoj	but I never did c# - you get the full power of delimited continuations?
16:33	bbloom	tomoj: yeah, pretty much the same thing
16:33	jcromartie	jonasac: yes, you need to add the :dynamic metadata to your dynamic vars
16:33	jcromartie	(def ^:dynamic x 0)
16:34	jonasac	ow, the book didnt mention that :p
16:34	tomoj	ah, so your main use case for cps is just async/await? me too
16:34	jcromartie	jonasac: it's new-ish
16:34	xeqi	AntelopeSalad: are you looking for ring.util.response/header ? http://ring-clojure.github.com/ring/ring.util.response.html#var-header
16:34	bbloom	tomoj: not really. c# special cases functions that are marked with the async keyword. those functions must return objects of type ITask or ITask<T>, which are just promises with optional continuations, and then turns the async function into a state machine
16:35	jcromartie	jonasac: the fundamental docs on clojure.org are good reading, and I don't mean that in a condescending way… I go there all the time http://clojure.org/vars
16:35	hiredman	difficult to see the difference
16:35	jonasac	i guess this while chapter falls out then cause it uses binding to bind to functions passed in (making a stubbing framework), unless i do that to all the functions in my code
16:35	AntelopeSalad	xeqi: yep, thanks -- then compojure.util :as response to call it as response/header?
16:35	jonasac	while == whole
16:35	bbloom	tomoj: that works great for C# b/c a state machine with a switch statement will outperform a whole bunch of lexical closures tangled together, but it's a less general approach. it really doesn't scale in terms of complexity to a language like closure
16:35	jcromartie	jonasac: I think it's since 1.3
16:35	jcromartie	jonasac: what book?
16:35	AntelopeSalad	oh wait, it's ring not compojure hah
16:35	jonasac	clojure in action
16:35	bbloom	tomoj: there are a few other use cases for a CPS transform
16:35	jonasac	chapter 8.2
16:35	bbloom	tomoj: tail call optimizations is one
16:36	bbloom	tomoj: another is composeable (aka delimited) continuations
16:36	bbloom	tomoj: that is shift and reset
16:36	Frozenlock	TimMc: I've often use java interop, but is there some rules of thumb one should follow when doing a wrapper?
16:36	jcromartie	yeah Clojure 1.3 came out just before Clojure in Action
16:36	jcromartie	I think
16:37	jcromartie	jonasac: typically you don't need a whole lot of dynamic vars
16:37	jcromartie	and you can also use let
16:37	bbloom	tomoj: continuations also let's you do some fancy server side cleverness like pg does with arc and HN, but there are a lot of folks who think that's not a great idea :-P
16:37	AntelopeSalad	xeqi: i'm actually not sure now, i was trying to implement some code that someone pasted here the other day -- i'm getting aliasing errors now where it doesn't make sense
16:37	AntelopeSalad	they make a call to response/header
16:38	tomoj	seaside-esque?
16:38	xeqi	jonasac: if you're just wanting to redefing functions for a test you might want ##(doc with-redefs)
16:38	lazybot	java.lang.SecurityException: You tripped the alarm! with-redefs is bad!
16:38	xeqi	hah
16:38	bbloom	tomoj: yeah, that smalltalk framework
16:38	bbloom	tomoj: simplest case is pg's "arc challenge" http://paulgraham.com/arcchallenge.html
16:39	xeqi	AntelopeSalad: it should be that function but hard to tell without the paste showing how its used
16:39	AntelopeSalad	k, one sec
16:40	AntelopeSalad	what was that site that supported clojure highlighting for pastes?
16:40	AdmiralBumbleBee	refheap
16:40	AntelopeSalad	thanks
16:40	AdmiralBumbleBee	and gist
16:41	egghead	anyone itc going to the LA clojure meetup this thursday?
16:41	AntelopeSalad	xeqi: https://www.refheap.com/paste/6268 , the important bits are probably lines: 12, 16-21 and 35
16:41	AntelopeSalad	i also don't know if 34 will work, will worry about that later
16:41	AntelopeSalad	the repl is crying that compojure.response is already aliased
16:42	xeqi	AntelopeSalad: did you restart your repl after aliasing it?
16:43	xeqi	or remove the ns?
16:43	jonasac	xeqi: with-redef was just what i needed :)
16:43	AntelopeSalad	every time i make a change i run: (require 'hello-world.core :reload-all) in the repl
16:43	AntelopeSalad	that is when the alias problem comes up
16:43	xeqi	if I understand correctly, that doesn't remove the old ns and redo it
16:43	AntelopeSalad	"alias response already exists in the namespace, aliasing compojure.response"
16:44	AntelopeSalad	i don't even see how that's aliased though
16:44	AntelopeSalad	because i never aliased it
16:44	xeqi	you didn't load the file with it aliases as described in your original question?
16:45	AntelopeSalad	load it in what sense? it never properly gets loaded because of the alias error
16:45	AntelopeSalad	i am running lein repl manually
16:45	AntelopeSalad	start the repl, do the reload all command, start the server
16:46	AntelopeSalad	it seems you're right though, i really don't know what i'm doing haha
16:46	AntelopeSalad	someone recommended to :reload-all as the "right" way to reload code
16:47	AntelopeSalad	i just changed :as response to :as responsefoo and then changed my fn call to responsefoo/header and it failed with no such var
16:47	AntelopeSalad	(after doing a reload-all)
16:47	AntelopeSalad	that confirms that reload-all isn't pickuping up a NS change right?
16:47	AntelopeSalad	*picking
16:49	TimMc	Frozenlock: I don't have any specific recommendations, sorry.
16:49	TimMc	I guess just try to make something that hides all the insanity that it makes sense to hide.
16:56	AntelopeSalad	xeqi: i'm a bit more confused now, i googled to see if anyone is even using util.response and this code came back: https://github.com/ibdknox/noir/blob/master/src/noir/statuses.clj
16:56	AntelopeSalad	problem is, the guy doesn't reference ring-resp anywhere -- why is he even including it?
16:59	xeqi	AntelopeSalad: most likely was meaning to replace L28-29 and forgot about it
17:00	AntelopeSalad	any thoughts on how to solve this?
17:00	AntelopeSalad	i tried reloading with just (use 'ns) in the repl, same aliasing error as before
17:01	jcromartie	AntelopeSalad: sounds like :reload-all is broken
17:01	xeqi	did you restart your repl?
17:01	AntelopeSalad	no, i did not restart it
17:01	jcromartie	xeqi: the goal is to avoid restarting it
17:02	AntelopeSalad	i'll restart it though now and see
17:02	AntelopeSalad	same problem
17:02	AntelopeSalad	have to go, zzz... i didn't expect something so silly to be 2 hours of debugging haha -- will try again tomorrow
17:03	AntelopeSalad	if the guy who wrote ring is here, please update your docs to include samples on how to use the code you've written -- beginners to the language cannot figure this stuff out with no documentation
17:04	jcromartie	AntelopeSalad: which part of ring?
17:04	jcromartie	AntelopeSalad: sorry I haven't followed the whole discussion
17:06	jcromartie	all this talk about continuations in web apps… what about representing a HTTP session as a lazy seq?
17:07	jcromartie	that the session is a single sequence of requests -> responses
17:10	TimMc	Continuation IDs in URLs work great until you want to bookmark something.
17:11	TimMc	or use back/forward after being AFK for 15 minutes
17:12	bbloom	oh yeah, it's annoying as hell as an end user. i wouldn't actually recommend writing an app with the same bugs as HN :-P
17:13	jcromartie	heh
17:14	ivan	hey, if you have a durable lazy seq...
17:14	emezeske	Yeah, the only anoying thing is the non-durability
17:15	jcromartie	same problems as seaside though
17:15	jcromartie	and only slightly less debuggable
17:15	bbloom	i think it's a fundamentally failed approach b/c it's effectively the opposite of conservative garbage collection
17:15	technomancy	heh; "HREF considered harmful"
17:15	bbloom	you have external demands on resources that you don't know about
17:15	jcromartie	I dunno, I liked the idea of Seaside but I had a lot of issues
17:15	jcromartie	hm, yeah
17:16	dnolen	jcromartie: the continuation thing is actually good for something very specific that no-one bookmarks - complex form flows
17:16	bbloom	but i think that applies to sessions, full stop.
17:16	bbloom	dnolen: even complex form flows suck... what if i'm on page 4 of 10 in a survey and then close my laptop and go for a walk?
17:17	bbloom	the nature of http is stateless
17:18	dnolen	bbloom: I'm not saying there aren't other ways to deal w/ the issue, but it's improvement. I haven't messed w/ JS browser history to support enough to say whether it sucks more or not
17:18	bbloom	shrug session state is a hack and always will be
17:19	bbloom	the only way around it is to have a persistent connection (or some kind of long polling to simulate a persistent connection)
17:19	jcromartie	then might as well say all sessions such
17:19	jcromartie	suck
17:19	bbloom	but even then, you need to support disconnect/reconnect
17:19	bbloom	again, consider the long survey and a closed laptop
17:19	dnolen	bbloom: perhaps, in anycase DabbleDB seemed to get some good mileage out of the approach. Perhaps other people are catching w/ pure JS apps? I can't say. Haven't done complex forms in a long time.
17:20	dnolen	catching up
17:21	bbloom	i think that blurring the line between the server and the client is fundamentally a bad idea
17:22	bbloom	i much rather a rich client app that maintains some in memory state and a server that is either totally stateless or connection oriented
17:22	hiredman	fallacies of network computing?
17:22	bbloom	totally stateless --> session oriented
17:22	bbloom	er i mean totally stateless --> totally session-less
17:23	bbloom	hiredman: yup: http://en.wikipedia.org/wiki/Fallacies_of_Distributed_Computing
17:37	tomoj	oh, good, I can stop thinking about the arc challenge now :)
17:40	bbloom	tomoj: it's still a worthwhile mental exercise :-)
17:50	TimMc	Oh man, I just read that. So ridiculous!
17:51	TimMc	If you can have an arbitrary web lib off to the side, there's no challenge.
17:53	jcromartie	The Arc Challenge is silly: he's starting with a language that defines a form that creates an HTTP handler in a running server, presumably?
17:53	jcromartie	I mean, that's all I can gather from the (defop …) form
17:56	TimMc	Yeah, right after saying that it's general purpose.
18:01	Frozenlo`	Can I use extend-type on a class to control how a new instance is created?
18:01	gfredericks	like hook into the class's constructor?
18:03	Frozenlo`	gfredericks: Yeah, like that :P
18:04	Frozenlo`	I suppose I could make my own protocol with a method 'new2'
18:04	gfredericks	I do not believe the jvm would go along with that plan
18:05	gfredericks	Frozenlock: even then it'd be a method on instances rather than the class
18:05	gfredericks	maybe you could extend it to java.lang.Class if you don't mind the same behavior for all classes...
18:06	gfredericks	then you'd have to instantiate via reflection...I really don't know a lot about this.
18:07	bhenry	what is an alternative to partition where i can still have a smaller list at the end if the original collection doesn't divide evenly?
18:07	amalloy	~partition
18:07	clojurebot	partition is probably not what you want; see partition-all.
18:07	bhenry	thanks!
18:34	etherealG	hi, I noticed on 4clojure the order of the problems seems to be a lot harder if done as it takes you through them. simple stuff like recursion comes much later than more complicated ones like interleave. anyone know which order to do them in if learning clojure from scratch?
18:35	AimHere	The problems are rated Elementary, Easy, Medium and Hard or something
18:36	AimHere	You could go by that
18:38	tomoj	bbloom: yeah
18:42	tomoj	I'm thinking you could get enough of the good stuff that arc has, separate the client and server better, and get by with just async/await
18:47	antares_	Monger 1.3.3 is released: http://blog.clojurewerkz.org/blog/2012/10/31/monger-1-dot-3-3-is-released/
18:50	tomoj	async/await plus a little more..
19:30	nDuff	Hrm.
19:30	nDuff	For some reason nrepl.el is trying to use complete.core, but not require'ing it
19:30	nDuff	so I have to explicitly do that myself.
19:34	brainproxy	nDuff: I had that problem; does it happen when you do nrepl-jack-in, or only when you're connecting from emacs to an already running nrepl server?
19:36	technomancy	hm; looks like it changed since I originally added completion
19:36	technomancy	my version requires it correctly =)
19:36	brainproxy	technomancy: any luck with your attempt to get namespace (and dependent namespace) reloading working?
19:36	brainproxy	i.e. when using nrepl.el
19:37	nDuff	brainproxy: I don't use nrepl-
19:37	nDuff	erm, nrepl-jack-in at all, so can only speak to the latter case.
19:37	technomancy	brainproxy: I got it working, but it was ugly. I don't remember if it got merged to master.
19:37	brainproxy	nDuff: okay sure; how are you invoking the nrepl server?
19:38	brainproxy	i mean, are you doing so from the command line w/ lein, or are you embedding it w/in some project
19:38	nDuff	Neither
19:38	nDuff	Using gradle because lein doesn't support Ivy
19:39	nDuff	...my build.gradle invokes clojure.tools.nrepl.cmdline/-main with args ['--port', '6678']
19:39	brainproxy	okay, maybe i'm confused; you are using nrepl.el to connect to an nrepl server?
19:39	brainproxy	ah gotcha
19:40	brainproxy	technomancy: have you had luck with ritz-swank?
19:41	technomancy	haven't tried it
19:41	brainproxy	I've made several attempts to get it working to no avail; but have no problems with swank-clojure
19:41	brainproxy	okay
19:44	scottj	brainproxy: I had trouble with it last time I tried. time before that it worked great. I'd contact author he's very responsive
20:46	holo	hi
20:46	mpan	hi
20:47	Scriptor	hey holo
20:49	holo	hey Scriptor
20:49	mpan	By any chance, does anyone have a recommendation for a genetic algorithm library? I imagine it would probably be in Java and work through interop, so the convenience of that would be a factor.
20:52	gfredericks	~
20:52	clojurebot	Titim gan éirí ort.
20:53	mpan	hm, internet connection flaking out on me
20:54	mpan	btw what's ~ meant to do? random quote from the bot's collection?
20:54	gfredericks	no I sent it accidentally
20:54	mpan	is there documentation of the bots' commands?
20:54	gfredericks	was playing with the ssh client interface
20:54	antares_	~maven
20:54	clojurebot	Help, I'm trapped in an XmlFactoryFactory!
20:54	gfredericks	I think "~" asks clojurebot to comment on the empty string about which he apparently doesn't know anything
20:54	gfredericks	clojurebot: is an empty string
20:54	clojurebot	Or, for instance, foo.toString().length() is (.. foo toString length)
20:54	mpan	~test
20:54	clojurebot	forget latest is 1382
20:55	gfredericks	clojurebot: \|is\| an empty string
20:55	clojurebot	read-string \|is\| as unsafe as eval unless read-eval is bound to false.
20:55	gfredericks	I apparently don't know how to make him respond to "~"
20:56	mpan	wait huh? when do you eval at read-time?
20:56	mpan	I'm blanking on any examples
20:56	gfredericks	when you use #=
20:57	gfredericks	I don't know of any use cases that aren't addressed by the newer data readers
20:57	mpan	sorry, but what's hash-equals?
20:57	gfredericks	reader eval :)
20:57	mpan	that sounds scary
20:57	gfredericks	you'll have to try it at the repl as I think the bots disable it
20:57	gfredericks	try (read-string "(1 2 #=(+ 3 4) 5)")
20:58	mpan	is it the same as eval except when?
20:58	gfredericks	I think so
20:58	mpan	what's the need to upset the usual semantics of read?
20:59	mpan	or rather, what was
20:59	gfredericks	I think just when you want to read in objects that don't have their own syntax
20:59	hiredman	try (read-string "#=(+ (+ 3 4) 5)")
20:59	gfredericks	hiredman: that's slick; so it just supports top-level function calls?
20:59	mpan	what are "objects that don't have their own syntax"?
21:00	gfredericks	mpan: e.g., records
21:00	gfredericks	queues
21:00	gfredericks	finger trees
21:00	mpan	wait huh?
21:00	hiredman	gfredericks: I think, I forget, it's been I while since I've paid attention to eval reader
21:00	gfredericks	I've never had a real use for #=
21:00	hiredman	records have a syntax now
21:00	gfredericks	hiredman: looks like it also calls macros as functions
21:00	mpan	even so, why would you want them at read-time?
21:00	mpan	rather than at eval-time?
21:01	mpan	like, why would you want the eval-ed representation?
21:01	gfredericks	maybe for data that you're only reading rather than evaling?
21:01	hiredman	mpan: because the reader can read other things besides code
21:03	gfredericks	so it looks like it reads everything in the list, expects the first item to be a symbol that it resolves to a function, and then calls that function on the other (uneval'd) arguments
21:03	gfredericks	e.g., #=((comp inc +) 4) doesn't work either
21:07	Cubic	How do I supply a type hint for a non-primitive array?
21:08	mpan	all right, thanks guys
21:09	amalloy	&(class (into-array ["x"]))
21:09	lazybot	⇒ [Ljava.lang.String;
21:11	Cubic	&(defn [x] ^[Ljava.lang.String x)
21:11	lazybot	java.lang.RuntimeException: Unmatched delimiter: )
21:11	Cubic	That's not valid clojure
21:11	mpan	why does (read-string "{:a {:b :c}}") produce what looks to be an exception at the repl yet still yields the expected result?
21:11	mpan	or, what does the bot say about ##(read-string "{:a {:b :c}}")
21:11	lazybot	⇒ {:a {:b :c}}
21:12	gfredericks	mpan: no idea what sort of exception you're seeing
21:12	gfredericks	mpan: works fine for me
21:12	mpan	uh, I can't even reproduce it
21:12	mpan	https://www.refheap.com/paste/6272
21:13	mpan	I would expect it to be deterministic, but, well... I guess something happened
21:13	gfredericks	I feel like it could have been some kind of stale exception from the previous command
21:13	gfredericks	I've had spooky repls like that before
21:13	gfredericks	not lately
21:14	mpan	previous line was a genuine exception produced
21:14	gfredericks	but anyhow that exception looks like the one generated by my last reader-eval example
21:14	mpan	yes that
21:14	gfredericks	did it print the exception twice then?
21:14	mpan	yes
21:14	gfredericks	heckiphino
21:14	mpan	oh but, the first time was printing it as the result
21:14	gfredericks	ah ha
21:14	mpan	and the other time was printing it as an error message
21:15	gfredericks	spooky repl still
21:15	mpan	i.e. the first one was prefixed by #< and the second time wasn't
21:15	mpan	suggesting one was the toString() and the other was... uh... something
21:20	mpan	gfredericks: wait, would that constitute a repl bug? if so, should I report it somewhere?
21:29	gfredericks	mpan: probably; repls are complicated. I wouldn't know what to suggest.
21:31	mpan	the repl that lein uses is at its core nrepl? so this gets reported to nrepl?
21:33	antares_	mpan: it uses REPLy
21:34	mpan	is REPLy on top of nREPL?
21:35	mpan	oh, hm
21:35	Cubic	lein repl does create an nrepl server if I'm not completely mistaken
21:35	xeqi	reply is a frontend to nrepl, think client server
21:35	mpan	so if there's what appears to be a bug in what appears to be the repl itself? is that in nrepl?
21:35	Cubic	amalloy_: thanks, got it to work
21:39	xeqi	mpan: I'd start by filling it with reply, trptcolin is good about tracking things down and pushing it to nrepl when needed
21:39	xeqi	though (read-string "{:a {:b :c}}") works fine in `lein repl` for me
22:03	amalloy	my suspicion is that mpan's repl was in some weird/broken state based on something totally unrelated, which cased the first eval-attempt to fail. later attempts worked fine because it was no longer in that state. ie, i don't think nrepl maintainers can ever repro this from just the paste above
22:05	gfredericks	at the very least his previous command, which actually generated the exception, would be needed
23:34	lil`nbv4	when i type `lein ring server` it says "That's not a task. Use "lein help" to list all tasks." i'm using lein 1.7, what could the problem be?
23:37	technomancy	lil`nbv4: best to get on leiningen 2
23:37	xeqi	lil`nbv4: do you have :plugins [[lein-ring "0.7.5"]] in your project.clj?
23:42	lil`nbv4	xeqi: I do now :) i am new to clojure, thanks
23:46	xeqi	then I also recommend using lein2