#clojure log - Mar 06 2013

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

0:00 TimMc: dcolish: futures implement j.u.c.Future

0:01 dcolish: ok thats what i thought

0:01 adu: what does j.u.c stand for?

0:01 dcolish: java.util.concurrent

0:01 adu: ah

0:02 dcolish: althought it looks like future does a dispatch to some thread pool or executor too

0:03 TimMc: ~j.u.c.

0:03 clojurebot: excusez-moi

0:03 TimMc: Raynes: You can make it up to me by explaining how clojure.core/promise gets away with only implementing one arity of invoke for IFn.

0:04 Raynes: *shrug*

0:04 dcolish: ah it submits to an actor

0:04 having the source in the docs is a nice feature

0:05 i've wanted that in javadoc over xrefs

0:08 its too bad clojure-guava only wraps the guava collections

0:09 TimMc: Raynes: I guess the JVM allows instances of classes that have abstract methods, even though Java doesn't.

0:14 amalloy: TimMc: i think java allows you to use instances of such classes too, just not compile them

0:14 TimMc: amalloy: Create them via reflection? Sure.

0:15 But I would be surprised if you could link against them.

0:15 amalloy: TimMc: well, i meant "invoke methods on them when passed them from somewhere else", but i think that's actually a pretty silly distinction

0:17 TimMc: I'm not even sure how to look this up or test it out.

0:17 (The linking question.)

0:18 ,((promise) 1 2)

0:18 clojurebot: #<AbstractMethodError java.lang.AbstractMethodError: clojure.core$promise$reify__6354.invoke(Ljava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;>

0:21 dcolish: hmm, i think the jvm will enforce instantiation of abstract classes

0:21 its one of the access flags for the class file format

0:29 amalloy: TimMc: https://www.refheap.com/paste/9ed6c6cdbad1b47596d9be371 compiles fine, and runs fine with java -cp .:clojure.jar tmp.core - there's no special "clojure-specific" linking going on at that point, just the jvm linking with a .class file that claims to implement j.u.List but doesn't have all the methods

1:00 technomancy: clojurebot: java.util.concurrent?

1:00 clojurebot: java.util.concurrent is "When I find myself in times of trouble / Prof. Doug Lea comes to me / Coding lines of wisdom / j.u.c."

1:00 technomancy: better late than never, eh

1:00 jeremyheiler: That's great.

1:01 technomancy: I stole it from Ted Nyman

1:01 dcolish: amalloy: neat

1:02 technomancy: there's also a version that goes "When I find myself in times of trouble / Col. Sanders comes to me / Cooking breasts of chicken / K.F.C." but that one is less relevant to this channel

1:03 jeremyheiler: technomancy, I went to Oswego, so it makes me laugh because it's so true.

1:04 technomancy: oswego?

1:04 jeremyheiler: Sorry. Doug Lea teaches at Oswego.

1:04 technomancy: oh, nice

1:05 jeremyheiler: I stayed an extra semester so I could take is Parallel Computing class. Worth every dime.

1:05 his*

1:05 technomancy: he may not be aware that he has a copyright-infringing theme song, but maybe you can advise him of the fact the next time you visit your alma mater

1:06 huh: http://gee.cs.oswego.edu/dl/html/jcp22oct10.html

1:06 jeremyheiler: haha

1:06 oh, yeah, that was interesting

1:07 technomancy: go openjdk

1:09 jeremyheiler: yea seriously

1:11 technomancy: rockin' stache too

1:11 jeremyheiler: usually has the full Einstein look.

1:15 it's funny. he once said he '

1:15 "programs in java-ese and then translate it into java so we all can understand it"

1:16 speaking about juc of course.

2:26 noidi: ambrosebs, it doesn't work for me now either. somehow it did work for me on monday, which is odd. maybe I had my repl in some weird state that made it work :)

2:27 ambrosebs: noidi: It should be easy to support, probably just needs a flag that says which namespaces are currently being checked.

2:28 noidi: that would be nice

2:28 otherwise you'd need a unit test running check-ns for each namespace to get automatic type checking

2:29 ambrosebs: noidi: namespace management is also something that's lacking.

2:29 I might include a "typed-ns" macro which names dependencies and automatically has check-ns invoked in its mexpansion.

2:31 noidi: so you'd use typed-ns instead of ns?

2:31 ambrosebs: Um, probably not.

2:31 I was thinking of both.

2:31 Might not be the best idea.

2:31 amalloy: you can't really avoid having ns

2:31 ambrosebs: kinda.

2:31 noidi: yeah, that's what I thought

2:32 you'd still have to get a hold of the custom ns-replacement somehow :)

2:33 ambrosebs: perhaps you could provide leiningen with which namespaces should be type checked.

2:36 There's also some interesting contract wrapping that could be done. I basically want all references to a particular var untyped.ns/foo from a namespace typed.ns to be wrapped in a particular contract. But it does not affect other namespaces. Any ideas?

2:37 Maybe the easiest would be to create a new def typed.ns/foo which is untyped.ns/foo wrapped in a contract.

2:38 noidi: so you'd use the typed-ns macro to rewrite typed code (to add contracts, etc.)?

2:39 ambrosebs: Something along those lines. We don't really need the ns macro, just provide a typed-require.

2:39 Two ns macros kinda sucks.

2:47 noidi: maybe the contract checking should happen on the caller side. then it would also work when calling functions passed in as arguments.

2:48 otherwise untyped code could call typed code and pass in a function with the wrong contract

3:42 dEPy: What is reify in clojure? O_ô

3:47 illiux: DEPy: It creates an instance of an anonymous type

3:48 err dEPy

3:50 dEPy: am, naonymous type is what? like an interface?

3:52 tsdh: Is there a way to define the order of execution of deftests in a test file when run by "lein test"? It seems the order is arbitrary.

3:53 Foxboron: ambrosebs: yeah i have. Interesting tbh. But i don't really know how the whole type system works, so kinda unknown territory ^^

3:56 ambrosebs: Foxboron: Well if you feel it might be useful, I've started some tutorials http://clojure-doc.org/articles/ecosystem/core_typed/home.html

4:06 illiux: dEPy: Not an interface, but closely related. riefy you provide implementation methods for a set of protocols and interfaces, it returns and instance of the type defined by that implementation.

4:06 *an instance

4:21 ciphergoth: Inside a running web server process (Aleph) I want to poll for something every second. Is the right thing to fork off a thread that calls (Thread/sleep 1000) after every poll?

4:23 mpenet: ciphergoth: use scheduled executors in j.u.c

4:24 or is it tied to the requests/responses? (ex: long polling)

4:24 ciphergoth: As a server, we allow our clients to long poll so they don't have to worry about this sort of nonsense. Sadly the services we're using don't offer that.

4:25 hmm, https://github.com/overtone/at-at

4:25 mpenet: ok, so java.util.concurrent or a wrapper ex: https://github.com/mpenet/knit#scheduledfuture

4:25 or at-at yes

4:27 ciphergoth: I'm strongly influenced in my choice of library by things like nice colours in the README.md and the words "build:passing" after the name :)

4:27 mpenet: :)

4:27 ciphergoth: they tend to correlate with being the good library!

4:27 it's not quite as shallow as it seems!

4:27 mpenet: well the tests in knit are weak, but it works (I am using it in a few libs in production)

4:27 mpfundstein: are you guys using classes in clojure?

4:29 clgv: mpfundstein: deftypes and records when needed

4:29 mpenet: at-at should be just as good, it's just a different api, matter of personal preference

5:23 mpfundstein: can i somehow pass a sequence like this as an argument to assoc? (:id 33333 :time 1362565416258 :priority 3 :action "hehe" :args nil :group nil)

5:24 Kototama: mpfundstein: (apply assoc m s)

5:24 mpfundstein: Kototama: thx

5:24 Kototama: mpfundstein: see also (merge

5:49 augustl: with clojure.xml/parse, is it possible to manually specify encoding? We get an utf-8 encoded file but the xml file says its encoding is utf-16 so parsing fails

6:34 squidz: does anybody know what my javascript file is being read in as a html document? I am using compojure with clostache. I have two <script type="text/javascript" src="..."></script> calls. The first refers to a url which loads fine. The second call refers to a js file in my resources/public folder. I also added (route/resources "/") to my defroutes

6:35 I am getting a 404 not found while trying to load the second local js file

6:39 faust45: squidz: you can use firebug to debug your app

6:41 squidz: faust: yes I think it is a problem with my setup on the compojure side. I now see that I might have to specify the resources-path in my project.clj

6:43 maybe not

6:51 okay when I change my route addresses that preced my (resources "/") call then it works. I just changed a previous GET call from (GET "/p" [] (page-call)) to (GET "/" [] (page-call)). Can anybody explain why this affected my (route/resources ...) call?

6:57 faust45: squidz: i think this kind of questions you can debug by self

7:01 squidz: faust45: Yeah, I already have it solved, but I should state my solution, and out of curiosity ask to get a deeper understanding

7:46 justinleitgeb: rubygems

9:01 TimMc: amalloy_: (java.lang.reflect.Modifier/isAbstract (.getModifiers (class (reify java.util.List (size [this] 10))))) -> false

9:02 amalloy_: The question, is what would *javac* do upon encountering such a class file that a .java file was referencing. I know the JVM itself is all hunky-dory with it.

9:03 Is there some way I can persist a runtime class object to disk as a .class?

9:08 jcidaho: hi - anyone using elastisch or clj-elasticsearch - am wondering if either of these give the ability to limit the fields that come back from es, using http://www.elasticsearch.org/guide/reference/api/search/fields.html...

9:43 clgv: TimMc: yes set *compile-files* to true

10:01 fbernier: How could I split a sorted-map into two collection of keys and values while keeping the order

10:01 clgv: fbernier: do `keys` and `vals` not work?

10:03 fbernier: clgv: no, they return a hash which is not guaranteed to keep the order

10:04 user=> (keys {:a 1 :b 2 :c 3})

10:04 (:a :c :b)

10:05 rplaca: fbernier: that's not a sorted map

10:05 clgv: ,(keys (sorted-map :a 1 :c 3 :b 2))

10:05 clojurebot: (:a :b :c)

10:06 clgv: ,(vals (sorted-map :a 1 :c 3 :b 2))

10:06 clojurebot: (1 2 3)

10:06 fbernier: sweet

10:06 thanks

10:28 jcromartie: so I found a very simple solution to the "no println output when running a compjure server in Emacs/nREPL" problem

10:29 simply wrap your handler with bound-fn when starting Jetty...

10:29 now what are the potential problems with this?

10:40 jtoy: can anyone point me to a webpage that explains how to interpret this or can someone explain it to me: [& {:keys [file user]}]

10:40 that is options for a function i see

10:41 andreaslundahl: jtoy: http://clojure.org/special_forms#Special%20Forms--Binding%20Forms%20(Destructuring)

10:42 mikerod: jtoy: Another link I like for clojure destructuring is http://blog.jayfields.com/2010/07/clojure-destructuring.html

10:44 jtoy: thanks

10:51 deg: I've just moved from Clojure 1.4 to 1.5, and the (pst) function seems to have disappeared??

10:51 lazybot: deg: Uh, no. Why would you even ask?

10:52 aroemers: heh, there's your answes :)

10:52 *answer

10:52 deg: ?? Why did Lazybot trigger on that line?

10:53 aroemers: Just randomly

11:08 rahcola: anyone around who knows about the internals of nrepl?

11:13 jtoy: how do I create a sequence with one item in it? (map ( fn[x] '(x)) [1 2 3]) doesnt seem t owork

11:13 llasram: jtoy: ##(map vector [1 2 3])

11:13 lazybot: ⇒ ([1] [2] [3])

11:16 jtoy: llasram: i think i need an actual seq, im testing now though

11:17 llasram: &(seq? [1])

11:17 lazybot: ⇒ false

11:17 llasram: Oh, heh

11:17 mpfundstein: i love clojure

11:17 period

11:17 llasram: jtoy: Then ##(map list [1 2 3])

11:17 lazybot: ⇒ ((1) (2) (3))

11:17 gtrak: ,(seq [1])

11:17 clojurebot: (1)

11:18 llasram: Yeah. Vectors are seqable, but don't directly implement ISeq (apparently)

11:18 gtrak: quoting doesn't work because it quotes what's inside the s-exp

11:19 does the :parent key in a lein2 project map have practical use in a multi-module project? How would I go about doing something like that?

11:22 jtoy: parenedit vim question, if I am at the |: [|(re-pattern x)] , how do I add parens so it becomes: ( [(re-pattern x)] )

11:34 gadget: jtoy: i put my cursor over the left bracket and type <leader>W(

11:39 jtoy: sorry, i should have typed it out first before respondign. you can either do <leader>W or <leader>w(

11:39 jtoy: gadget: cool, im still learning paredit

11:40 babilen: Speaking of paredit ... Is there a nice cheat sheet somewhere?

11:40 spligak: Making a request using clj-http with :throw-exceptions false and :as :json, when the response would normally throw an exception (say, status 400+), it doesn't honor the :as :json and decode the body. Anyone else experience this?

11:41 I don't mind decoding the body manually, just seemed odd.

11:41 ToBeReplaced: babilen: first google result http://emacswiki.org/emacs/PareditCheatsheet

11:41 jtoy: gadget: that only puts around the current symbol, that doesnt go around the whole piece im trying to add it around, do I then just need to move it manually?

11:43 babilen: ToBeReplaced: ... for vim (as was previously discussed)

11:44 gadget: jtoy: i'm not sure i understand, you wanted it to look like ([(re-pattern x)]) right? what does it look like to you after wrapping it in the extra parens?

11:44 babilen: ToBeReplaced: sorry, I should have stated that more clearly rather than relying on the assumptions that people will interpret that question in the context of the ongoing conversation

11:45 gadget: babilen: i just use the paredit.txt file in the ./vim/bundle/paredit-0.9.7/doc directory

11:45 if you're using pathogen that's where it would be at least

11:46 babilen: gadget: yeah, sure. I can read that easily with ":help paredit.txt", but Emacs' paredit has a very helpful cheatsheet that might have been translated for the vim version already

11:46 gadget: babilen: oh sorry i don't know of one

11:46 babilen: gadget: Take a look at the cheat sheet ToBeReplaced linked earlier and you might realise how that comes in handy during the initial learning period

11:46 NeedMoreDesu: I'm still interested in hot code swapping on .class/.jars. Any help would be appreciated.

11:50 http://stackoverflow.com/questions/15145721/clojure-hot-code-swapping-for-uberjars-classes

11:53 dakrone: spligak: let my see if I can reproduce it

11:55 spligak: dakrone, I'm using the 'request' function directly, passing in a handful of options in addition to those two. happy to throw up a gist if you'd like.

11:56 TimMc: clgv: Thanks! *compile-files* is exactly what I needed.

11:57 clgv: TimMc: good. :) you have to use alter-var-root though

11:57 TimMc: `set!` won't work...

11:59 deg: I've just moved from Clojure 1.4 to 1.5, and the (pst) function seems to have disappeared?? Is this a known issue?

12:02 kittylyst: Anyone recommend a good Clojure project for parsing Apache logs?

12:03 spligak: dakrone, hah. get it? "throw" up a gist? *rimshot*

12:03 S11001001: deg: clojure.repl/pst is still defined in clojure-1.5.0

12:03 zodiak: deg, yeah, pst is definitely still in 1.5

12:03 dakrone: spligak: heh, a gist would be helpful if you could get one to me

12:04 TimMc: amalloy_, clgv: Yep, you can totally compile Java against a class with abstract methods: https://gist.github.com/timmc/5100962 Color me surprised.

12:04 spligak: not a problem at all. one minute.

12:05 zodiak: kittylyst, surely you could jst open up a clojure.java.io/reader and run over it

12:05 kittylyst, if it's really big, and always appending, you could always play around with storm ;)

12:05 clgv: TimMc: duck typing with java ^^

12:05 args with jvm ;)

12:06 kittylyst: zodiak: Homebrewing with regexps was precisely what I was trying to avoid :)

12:07 zodiak: regexps ? surely you jst use clojure.string/split, and then decompose what you want out from there

12:07 your regex would be " " :)

12:07 owengalenjones: does anyone use clj-soap? is this project dead forever?

12:07 zodiak: or rather, #"\s+" (if you prefer)

12:09 clj-soap is dead ?

12:09 gfredericks: So (let [nums (range)] (#(dorun nums))) leaks, but (let [nums (range)] ((^{:once true} fn* [] (dorun nums)))) does not

12:09 owengalenjones: sean corfield has stated he has given up on it

12:09 gfredericks: is the :once metadata a supported feature? should it be encouraged in macros that create thunks that are only called once, e.g. clojure.java.jdbc/transaction ?

12:10 zodiak: owengalenjones, must have missed that.. do you have a problem with the implementation, or is missing something ?

12:11 owengalenjones: zodiak: I have the same issue listed in this thread (which I using as a test) https://groups.google.com/forum/#!topic/clojure/dewY8-Ao9AE, sean's last comment appears to be him giving up on it :/

12:12 zodiak: interesting. reading

12:12 kittylyst: zodiak: Your regex fails on the first line of my log file, e.g.:

12:13 - - [03/Mar/2013:06:31:19 +0100] "GET / HTTP/1.1" 302 299 "-" "Mozilla/5.0 (compatible; Baiduspider/2.0; +http://www.baidu.com/search/spider.html)"

12:13 mikerod: If I want return a function that has a to-be-determined number arguments/formal parameters, I have determined this works for me (eval `(fn [~@params] ~expr)). I do not think this is a clean solution (it uses eval). What I'm wanting to be able to do is to spread out some list of params into the param vector of the anonymous fn.

12:14 kittylyst: zodiak: And I can make it fail in pretty much arbitrary ways if I can't control the date format

12:14 zodiak: kittylyst, the apache regex wholly and totally depends on how ~you~ have configured apache and it's log format

12:15 perl-mongers have some pretty nice regex's on this mind you. don't see why a regex (posix compliant) shouldn't jst map across (no pun intended)

12:15 clojurebot: map is slightly retarded

12:15 zodiak: +1 internets clojurebot

12:15 kittylyst: zodiak: Yes, I'm an old perl monkey, so I'm aware of some of them I just hoped there were nicely wrapped versions

12:16 E.g. "^([\\d.]+) (\\S+) (\\S+) \\[([\\w:/]+\\s[+\\-]\\d{4})\\] \"(.+?)\" (\\d{3}) (\\d+) \"([^\"]+)\" \"([^\"]+)\"" - which is not quite right, but isn't a bad start

12:16 \

12:16 spligak: dakrone, https://gist.github.com/jasonjohnson/5101078 -- let me know if that helps.

12:16 owengalenjones: zodiak: anyway what I was attempting to do is use magento's soap api as their rest api only is available to half their customers, everything seems to work using their examples but clj-soap chokes: http://d.pr/n/fbKI

12:16 spligak: appreciate you checking it out. love the library.

12:16 S11001001: mikerod: it's not meaningful to build a non-variadic function if you don't know the arity until runtime, so you might as well just use & args and apply

12:16 hiredman: kittylyst: there are a number of parser combinator type libraries around on github, but none of them seems to be very good, and they often end up abandoned

12:17 spligak: dakrone, I'm using clojure 1.4.0 - that's just what lein spit out when I created the sample. updating that.

12:17 dakrone: spligak: looking through the tests, apparantly this is on purpose right now, because many errors don't come back as json when a 503 or whatever is hit, so I'm thinking of adding a {:coerce #{:always :unexceptional :exceptional}} setting to force behaviour

12:18 spligak: what version of clj-http?

12:18 zodiak: owengalenjones, yeah.. I don't think the "cool kids" use soap much these days (sorry). It's kinda strange/worrying that the author seemed to hit a bump and then say 'eff it'

12:18 hiredman: there is always http://meshy.org/2009/12/13/widefinder-2-with-clojure.html

12:18 owengalenjones: I dont blame him, this is my first experience with soap, its giving me an ulcer after only a day of looking at it. thanks for your time

12:19 zodiak: owengalenjones, no worries. Yeah. soap is .. pretty nasty (especially all that automagically wsdl stuff).

12:19 spligak: dakrone, using 0.6.4 in my project. and that's totally cool - just making sure I wasn't using it improperly.

12:19 dakrone: spligak: okay, I'll work on a fix

12:22 mikerod: S11001001: I understand what you are saying. I have a strange use-case for this. I writing a DSL in Clojure and at what ends up happening in this case is that I need a function with unknown number of args and unknown formal parameter names, however, this function will end up being called from java with the IFn#invoke(args).

12:23 zodiak: "Magento Core API supports the SOAP and XML RPC protocols. It does not support REST" .. how old IS this magento thing ?! xml-rpc ? mercy (that maybe a LOT easier to get running in clojure btw)

12:23 owengalenjones, I suspect you can find a fair few xml-rpc libs

12:23 S11001001: mikerod: if your DSL is, by nature, a builder of fresh clojure code, then perhaps it ought to do so by way of macroexpansion?

12:24 eriko: kittylyst: this rough parser is for cachefly logs, but they're derived from the Apache format. YMMV: https://gist.github.com/erikols/5101134

12:25 noidi: is it OK to leave the project name empty on the Contributor Agreement, if I want to be able to contribute to any project requiring a CA?

12:25 kittylyst: eriko: Thx. I've arrived at this for general Extended Common Log Format:

12:25 #"([^ ]*) ([^ ]*) ([^ ]*) \[([^]]*)\] \"([^\"]*)\" ([^ ]*) ([^ ]*) \"([^\"]*)\" \"([^\"]*)\""

12:25 You can do better for the apache case, by doing better for the case of first-column-is-an-IP

12:25 noidi: I'd hate to send a new CA every time I want to send a patch to a new core project...

12:26 mikerod: S11001001: I agree, macroexpansion is usually the best place to do something like this. I will take a look at what I'm doing a bit more closely and see if I can restructure it into a macro. Thanks for the help.

12:27 eriko: kittylyst: Cool, thx!

12:36 owengalenjones: zodiak: thanks, Im looking into https://github.com/brehaut/necessary-evil (its got an awesome name)

12:36 zodiak: *laughs* that's all sorts of good naming :D

12:43 owengalenjones: zodiak: omg thank you, it appears to have worked for my tests, work can continue. I never would've tried that, I dont even know the difference between soap xml-rpc all this junk...

12:44 zodiak: owengalenjones, nw. yeah. soap is the heavier cousin of xml-rpc. it's "technically" better (betamax vs vhs - which is an urban myth, I know ;) but. meh.

12:44 whatever works. works.

12:46 back to simcity 2013 for me :D

12:47 owengalenjones: zodiak: enjoy your tiny 2k*2k city plots :)

12:48 * technomancy gets nostalgic for simcity 2000

12:52 rplaca: noidi: just put "Clojure and Contributed Libraries" and you'll be set

12:52 noidi: rplaca, thanks!

12:53 rplaca: noidi: np

12:53 * rplaca ruminates on a life which has involved too many contracts

12:54 noidi: :)

12:57 I wish Clojure used something like echosign, like the Chef project http://wiki.opscode.com/display/chef/How+to+Contribute

12:58 technomancy: it's been discussed many times, but there doesn't seem to be much motivation to actually make contribution easier

12:58 noidi: sending information by moving physical objects seems so archaic :)

13:00 I don't think I even have any envelopes or stamps at hand, so that means a trip to the post office

13:01 technomancy, I've seen the discussions, and I've always wondered why Clojure/Core does not see this as an issue

13:02 technomancy: http://p.hagelb.org/mystery.gif

13:04 llasram: Maybe because the current process is already yielding more patches than rhickey wants personally review?

13:06 technomancy: #firstwordlproblems

13:06 noidi: llasram, that's a good problem to have and should be solved by scaling the approval process, not by making it harder to contribute

13:07 llasram: noidi: I don't disagree -- just generating hypotheses

13:40 thearthur: arohner: I sent a pull request porting clj-r53 to 1.4.0, It was crashing for me with 1.4.0 :-/

13:41 arohner: If I buy you beer sometime will ya release it as 1.0.2 soon?

13:41 arohner: thearthur: great!

13:43 thearthur: I tested it on create and update record

13:45 arohner: thearthur: cool. releasing soon

13:49 thearthur: 1.0.2 is released

13:49 thanks for the patch

13:51 TimMc: noidi: You're not sending information by physical object in the CA process -- you're applying Color to existing information: http://ansuz.sooke.bc.ca/entry/23

13:51 thearthur: arohner: and thanks for the quick turnaround, I hate releasing with forks... much appreciated

13:52 TimMc: It's a legal ritual that "blesses" the data.

13:53 ,(dorun (range))

13:53 clojurebot: Execution Timed Out

14:01 amalloy: so i noticed something interesting about delays yesterday. if you delay an expression that throws an exception, then forcing it multiple times does not always yield the same result

14:02 specifically, the first time you get the exception you threw; after that you get a NPE in the middle of clojure.lang.Delay

14:02 llasram: That seems to have a low probability of being intended/desirable behavior

14:02 Raynes: Mhm

14:03 amalloy: indeed. it's possible to work around it, either in your code or in c.l.Delay, so that it saves the first exception and rethrows it. it's not possible to make it actually retry in case some mutable state has changed

14:03 llasram: Yeah. Re-throwing is what I'd expect to happen

14:06 amalloy: well, i should be more specific. not *every* exception-throwing delay expression causes this. it has to be one that refers to some lexical locals in the scope of the delay

14:20 tomoj: is it me or are url locals bound in a strange place by moustache?

14:20 I mean, the locals it gives you for url segments

14:21 it seems you have to pass them into functions that return handlers

14:21 Raynes: Those words don't make sense to me.

14:22 tomoj: (app ["foo" id] (foo-handler-generator id))

14:22 (fn foo-handler-generator [id] (fn [req] ...))

14:22 that's the only way I can see to make use of the 'id binding

14:23 Raynes: Oh.

14:23 tomoj: Yeah, I thought you were talking about mustache.

14:23 I'm not a fan.

14:25 Man, I love hating on cgrand's stuff.

14:25 tomoj: I don't think I am either

14:25 Raynes: I'm waiting for him to challenge me to a duel.

14:26 danneu: if you /quit

14:26 whoops

14:27 rabbit_airstrike: lisp duel

14:28 tomoj: on the other hand compojure seems to make a sort of opposite error..

14:30 I want (route :get ["foo" :id]) to be a middleware that routes and adds an :id param or url-param to the request

14:30 maybe compojure can do something very close to that?

14:35 ah, make-route

14:38 amalloy: tomoj: it's called context. i forget what compojure namespace it's in

14:38 (context "/foo" [id] some-route)

14:40 tomoj: oh wait

14:40 I guess moustache assoc's onto the request too

14:42 not usefully afaict

14:43 yogthos: Raynes: speaking of I've got a quesiton ;)

14:44 Raynes: can't seem to get wrap-file-info to work with uberjar

14:45 Raynes: I did (wrap-resource "public") and in the uberjar you end up with a public directory under it, so what am I missing there

14:45 Raynes: works fine in wars

14:46 Raynes: yogthos: No idea.

14:46 yogthos: Raynes: I guess I'll have to find weavejester for this mystery :)

14:47 Raynes: I'm going to be so glad when I have that app-handler working properly, nesting middleware can be so brittle

14:47 Raynes: nobody should have to do that again :P

14:48 Raynes: yogthos: So the problem is that your resources aren't being seen at all?

14:48 yogthos: Raynes: they are but the mime type isn't set

14:49 Raynes: Does that even make sense?

14:49 yogthos: Raynes: if I don't add wrap-resource or wrap-file-info then mime types are set

14:49 Raynes: So you *can* access the resources?

14:49 It just doesn't set the mime type?

14:49 yogthos: Raynes: yeah

14:49 Raynes: Yeah, that's way above my pay grade.

14:49 yogthos: Raynes: hehe

14:50 Raynes: what's weirder is that it *sets* the mime type if you don't put wrap-file-info at all

14:50 Raynes: so I wonder if there's some alternate mechanism that gets invoked

14:53 sujeet: devn, where u b dawg

14:55 jrobie: for what operations does nil NOT raise a n exception if used as an operand - just for conditions and for functions specifically designed to handle it?

14:59 amalloy: jrobie: lots of things. collection operations, especially

15:00 jrobie: amalloy, thanks - is there an easy way to look up the collection operations for which nil is defined?

15:00 amalloy: i don't think so. it's basically all of them

15:01 jrobie: thanks. but i assume all math functions, string functions, aggregate functions, etc. raise exceptions for nil, are there any other big groups of functions besides the collection functions that i should think of?

15:03 * jrobie is trying to design something like nil for another language

15:04 amalloy: jrobie: what language? if it's one you're building from scratch, i recommend not including nil

15:04 technomancy: inc

15:04 jrobie: amalloy, we need something like nil for xquery to support json, imho

15:04 technomancy: if you want inspiration, look at a language with a proper type system like OCaml or Haskell

15:04 jrobie: we have empty sequences

15:04 but we need something that occupies a position

15:05 amalloy: no you don't. you can use a special null value for the json library; it doesn't have to be part of the language

15:05 Tony Hoare, inventor of the null reference, has a presentation called: "Null References: The Billion Dollar Mistake"

15:05 jrobie: amalloy, good to know

15:05 * jrobie googles that

15:06 jrobie: i do need a special null value

15:06 amalloy: for example, haskell's json library defines what is basically an enum for the six-ish kinds of values json can have

15:06 and for operations outside of json, which in clojure or java might return either a value or nil, it instead returns Maybe value

15:06 S11001001: Maybe is cool

15:07 jrobie: ok. xquery doesn't have a maybe value

15:07 zodiak: jrobie, well, now, null was only "required" in the case that you had missing data. C.J Date proposed everything having a "default value" (which I am sure has Codd spinning in his grave). maybe that's another solution ?

15:07 but yeah.. a maybe monad could be the ticket

15:08 jrobie: we don't have anything like maybe in xquery, and what we mostly need is an item that can occupy a position and represents the lack of a vlaue

15:08 i'd like to keep this as simple as we can

15:09 xquery uses empty sequences as the rough equivalent of a sql null, but an empty sequence can't occupy a position in a sequence

15:09 solussd_: is it possible to extend a keyword's ability to look itself up in a map? How is that ability defined?

15:10 jrobie: zodiak, i've spent much too much of my life debating how sql nulls relate to xquery ;->

15:11 codd, date, and chamberlin definitely disagreed on that issue

15:11 zodiak: jrobie, I believe you :)

15:11 jrobie: chamberlin was a co-inventor for xquery, so guess whose view won in xquery?

15:11 zodiak: oohh.. tough q ;)

15:13 it could be you have an impedence mismatch that's not easily solved.. in which case you may jst be better off either one of (a) not taking null's in the first place, (b) default value if null or (c) maybe monad (although then you are pushing the complexity to the user/next fn to deal with)

15:13 probably a few more solutions but.. it's not going to be pretty ;)

15:14 Apage43: solussd_: Keywords can look themselves up in anything that implements ILookup

15:14 -or- anything that clojure.lang.RT.get works on

15:14 https://github.com/clojure/clojure/blob/master/src/jvm/clojure/lang/Keyword.java#L132

15:14 solussd_: thanks

15:15 jrobie: zodiak, we're trying to minimize the damage ...

15:16 Apage43: which means java.util.Map works too

15:16 zodiak: jrobie, minimize and pass upstream or wrap and deal with it ?

15:16 jrobie: hmmmm, Maybe is looking better than i had expected at first blush ...

15:17 zodiak: monads ftw (jst don't tell #haskell that ;)

15:18 arkh: what's the next best option to a multiple bind (when-let [ ... ]) ?

15:18 because when-let only allows for a single binding

15:18 * jrobie tunes out to go learn about things said already here

15:20 jimkcar: does anyone in here work for Relevance?

15:21 rabbit_airstrike: one of the things I like about R is the granularity of NULL, NaN, NA, and the Infs

15:22 zodiak: arkh, well.. you could always jst pass in a hashmap into the vector

15:22 (when-let [data {:a 9, :b nil}] (println (:a data) (:b data)))

15:22 jrobie: zodiak, amalloy - opinions on Python's None?

15:22 amalloy: rubbish. it's no different from nil

15:22 zodiak: jrobie, not much really, jst another form of nil, null from my understanding

15:23 arkh: zodiak: I like it - thanks!

15:23 zodiak: arkh, more than welcome

15:36 siscia: Hi, I am having problem to understand protocols, namespace, and implementation of the protocols themselves...

15:36 do you suggest any good resource ?

15:39 hiredman: a protocol is a named group of named functions that are polymorphic on their first argument's type

15:40 the names of protocols and the functions that are part of protocols are vars

15:40 vars exist in namespaces

15:41 the impl is not something to concern yourself with

15:42 (vars themselves also have names, which are namespaced symbols)

15:54 solussd_: anything wrong with extending a protocol to a clojure.lang Interface? E.g. IPersistentMap

15:55 siscia: hiredman, sorry i didn't hear the bip of xchat

15:55 solussd_: wait… dont records have marker interfaces now?

15:56 siscia: hiredman, still something no completely clear

15:57 hiredman, I define a protocol in a namespace (call it "A")

15:57 hiredman, (in-ns B (:use A))

15:58 (extend-type java-class somenthing-in-A (function [_]) (f2 [_]))

15:59 hiredman, if now I (:use A) in another namespace i should be able to call (f2 java-class)

15:59 it is right ?

16:02 frommorf: (filter #(zero? (mod n %)) (range 1 n)) ; Why does this result in a outofmemory for large n? isnt it lazy?

16:03 nDuff: frommorf: You're clearly evaluating it somewhere.

16:03 frommorf: If I define it without evaluating or printing it, I get no memory error.

16:04 TimMc: solussd_: If you own the protocol, that's fine.

16:04 frommorf: nDuff: i evaluate it at the repl

16:05 solussd_: TimMc: I want to implement a protocol on all things that act like maps, so I'm thinking about extending MyProto to clojure.lang.ILookup

16:05 nDuff: frommorf: Exactly.

16:05 frommorf: I imagine that if you only grab a few items of the beginning with take, you have no problem.

16:05 solussd_: too bad I cant extend a protocol to a protocol.. :)

16:06 amalloy: http://dev.clojure.org/jira/browse/CLJ-1175 is the jira issue i just created for the weird behavior of clojure.core/delay, if anyone wants to look at it or, say, vote for it

16:07 frommorf: nDuff: yes, but i need all the items. is there a way to optimize it?

16:08 siscia: frommorf, how big is n ?

16:08 frommorf: 600851475143

16:08 siscia: 600851475143

16:08 siscia: frommorf, are you looking for the divisor of that number ?

16:09 frommorf: siscia: all the factors of that number, I know there's a better algorithm. But i am wondering why it so slow this way

16:09 ninjudd: everyone should go vote for amalloy's issue

16:09 nDuff: Err. The slow part doesn't seem surprising at all.

16:10 siscia: frommorf, me neither...

16:10 nDuff: Running out of memory, on the other hand, was surprising (if not holding onto the head).

16:11 siscia: frommorf, a quick fix could be something like (filter #(zero? (mod n %)) (range 1 (int (sqrt n))))

16:11 frommorf: I'm very new to clojure so I dont know what is going on.

16:11 edw: I'm looking to write some code that needs to be evaluated eventually, but I really want to treat it as "fire and forget"; is it kosher to do something like (dorun (for [batch (partition 10 1 nil xs)] (future (frobinate batch))))?

16:12 siscia: frommorf, you are making a lazy seq (you know what that is?) and you try to print it all...

16:13 frommorf: siscia: yes i do. but i thought since it was lazy it wouldnt need to store the whole thing in memory?

16:14 siscia: frommorf, yes that is the point of the lazyness, however if you try to print it

16:15 the sequence has to be somewhere...

16:15 it has to be realized in order to print it...

16:16 frommorf: siscia: right, sorry my brain is not working today.

16:17 siscia: frommorf, :-)

16:19 TimMc: Who's the main Clojure JIRA wrangler? It looks like the version tag "Release 1.5" is still marked as Unreleased.

16:22 * bbloom imagines a guy with a whip and a cowboy hat... "yehaa! ima create one of them newfangled issues queries!"

16:24 ohpauleez: TimMc: I'm not 100% sure but Sierra or Andy

16:26 nDuff: Hrm; in clojure-mode, (avout/dosync!! *avout-client* <NEWLINE> indents the next line to be with the *avout-client*; is there a straightforward way to configure emacs to only indent one level (two spaces) there, rather than aligning with the first parameter?

16:26 TimMc: stuartsierra: FYI, the CLJ Jira project still has "Release 1.5" marked as Unreleased.

16:27 nDuff: Something about M-x customize-variable RET clojure-defun-indents

16:29 stuartsierra: I disavow all responsibility for JIRA. ;)

16:31 nDuff: TimMc: useful; thanks

16:39 amalloy: whew! it's nice to not be the only one who recommends defun-indents, TimMc

16:40 TimMc: amalloy: I still haven't used it, and if nDuff breaks his leg and comes running to me, I'm going to point him at you.

16:41 nDuff: TimMc: ...for what it's worth, my leg appears to be intact

16:41 (and my code well-formatted).

16:41 TimMc: stuartsierra: Andy is the one, then?

16:41 jimkcar: running with a broken leg would be impressive

16:41 TimMc: amalloy: Is there some sort of modeline thingum that can be used to make it portable?

16:41 amalloy: if anyone is interested, https://www.refheap.com/paste/0bccc08fbe2924fd38fa38b88 sets up git to provide custom hunk headers for clojure files in diffs. so the diff header tells you what function the diff is in the middle of

16:41 stuartsierra: TimMc: Probably. Email clojure-dev.

16:41 amalloy: TimMc: huh?

16:42 TimMc: amalloy: 404

16:42 Interlocutor Not Found

16:42 amalloy: this world makes no sense any longer

16:42 i resign

16:43 TimMc: amalloy: 404 on your refheap link, actually.

16:43 amalloy: really? it works fine for me, even clicking on it now

16:44 TimMc: Oh, huh!

16:44 jimkcar: I get "Insert fancy 404 image here."

16:44 TimMc: gnome-terminal was including the space after the URL

16:44 Got a weird char in there?

16:44 Raynes: I don't see any 404/

16:45 llasram: non-breaking space between the URL and 'sets'

16:45 amalloy: you're right, it looks weird to me too. must be some weird character

16:45 jimkcar: <space> sets was coming up as part of the url

16:45 amalloy: try https://www.refheap.com/paste/0bccc08fbe2924fd38fa38b88

16:45 TimMc: http://lab.brainonfire.net/whatchar/identify.php?data=https%3A%2F%2Fwww.refheap.com%2Fpaste%2F0bccc08fbe2924fd38fa38b88%C2%A0

16:45 amalloy: my browser (or my irc client?) was apparently stripping out that extra character at the end

16:45 TimMc: Non-breaking space.

16:46 amalloy: Sp, re: clojure-defun-indents, is there a way to make it so that collaborators get the same indentation rules?

16:46 *So

16:46 amalloy: $google dir-locals.el

16:46 lazybot: [Directory Variables - GNU Emacs Manual] http://www.gnu.org/s/emacs/manual/html_node/emacs/Directory-Variables.html

16:47 stuartsierra: You can also use file-local variables.

16:47 amalloy: i found it a bit fiddly to get everything lined up just right so that those work, TimMc, but it's definitely possible

16:47 TimMc: Oh, so it's a modeline thingy.

16:47 amalloy: no it's not. i don't know why you keep talking about the modeline

16:48 TimMc: I think that's what vim calls that.

16:48 amalloy: well, emacs has something called the modeline, and it's something other than whatever you're talking about :P

16:48 TimMc: Oh right, that term is overloaded.

16:49 Polymorphic across editors. :-P

17:03 gfredericks: So (let [nums (range)] (#(dorun nums))) leaks, but (let [nums (range)] ((^{:once true} fn* [] (dorun nums)))) does not

17:03 is the :once metadata a supported feature? should it be encouraged in macros that create thunks that are only called once, e.g. clojure.java.jdbc/transaction ?

17:07 edw: My app is not exiting after creating agents (and awaiting them too). Is there something I need to do to let the JVM die a natural death?

17:07 amalloy: gfredericks: thunks that are called only once are delays

17:07 ie, don't write them yourself, use the one that already exists

17:08 hiredman: amalloy: the problem with delays is they memoize their results

17:08 well, I guess not a problem in gfredericks's example

17:08 nDuff: edw: there's a call to shut down the agent pool.

17:09 amalloy: right, i think for anything you only call once that memoization won't be a problem

17:09 but i could be wrong

17:09 hiredman: "the problem with using delays with lazy-seqs is ..."

17:09 nDuff: edw: (shutdown-agents)

17:09 edw: Ah, I was just staring at the cheat sheet on my iPad.

17:09 Thanks.

17:12 amalloy: hiredman: if you let the delay get GCed, that shouldn't be a problem, right?

17:12 ,(let [d (delay (range))] (dorun @d))

17:13 clojurebot: Execution Timed Out

17:14 amalloy: ie, if it's really called only once, you can arrange to call it in a place where the delay itself is GCable, and the memoization doesn't matter then

17:15 hiredman: amalloy: sure, but if you do something like (let [d (delay (produce-seq))] some-complicated-logic-with-d) it is very easy to end up holding on to the head of the seq by holding on to d

17:19 amalloy: gfredericks: speaking of delays, did you see my jira issue earlier today about (what is arguably) a bug in c.l.Delay? http://dev.clojure.org/jira/browse/CLJ-1175 - vote early, vote often!

17:19 llasram: Isn't it vote early, vote only twice?

17:19 Of course, I have no idea how many tickets I voted on

17:20 amalloy: llasram: rich suggested that policy, but since it's based on rich believing that clojure developers will follow the honor system, of course it can never work

17:21 hiredman: I think the reports andy generates do some kind of fractional weighted voting

17:21 amalloy: last week andy fingerhut started counting how many issues you've voted on, so that if you've voted five times then each vote counts as 1/5 of a vote

17:21 llasram: Ah, interesting

17:21 amalloy: thus everyone "counts" as one vote, distributed as thinly as you like

17:21 llasram: That sounds like a better approach

17:21 amalloy: much

17:21 technomancy: time to start creating sock puppets

17:22 amalloy: yeah, i know

17:22 llasram: And have them submit CAs?

17:22 amalloy: technomancy: but sock puppets can't fill out CAs

17:22 so this is 100% secure

17:22 llasram: Sock Q Puppet III

17:22 cemerick: has anyone yet abused the BigDecimal implementation in gwt-math with ClojureScript? The references to it aren't exactly encouraging.

17:22 technomancy: amalloy: good point, that would involve signing something which is false, which as we know is a logical impossibility.

17:23 hiredman: technomancy: well for bots

17:23 llasram: Mechanical Turk

17:23 hiredman: ~truth

17:23 clojurebot: Titim gan éirí ort.

17:24 trptcolin: more like mechanical JERK

17:24 ok, i'm done, sorry

17:24 augustl: what's the simples way to turn a map into a map with a type I can use to extend a protocol?

17:25 llasram: augustl: Existing protocol, or your own?

17:26 augustl: llasram: my own, so I can pass in a map to a protocol function and have it run my implementation for my specific type

17:26 essentially I just want a map with a type I specify so I can implement a protocol for that type

17:26 llasram: If it's your protocol, could you just use multimethods instead?

17:27 &(map type [{} {:type :foo})

17:27 lazybot: java.lang.RuntimeException: Unmatched delimiter: )

17:27 llasram: &(map type [{} {:type :foo}])

17:27 lazybot: ⇒ (clojure.lang.PersistentArrayMap clojure.lang.PersistentArrayMap)

17:27 llasram: fail, me

17:27 &(map type [{} ^{:type :foo} {}])

17:27 lazybot: ⇒ (clojure.lang.PersistentArrayMap clojure.lang.PersistentArrayMap)

17:28 llasram: Huh. That should have worked

17:28 &(map type [{} (vary-meta {} assoc :type :foo)])

17:28 lazybot: ⇒ (clojure.lang.PersistentArrayMap :foo)

17:28 llasram: Anyway

17:29 So there's a builtin approach to using multimethods dispatching on the type of a value, or on some other arbitrary token you specify via metadata

17:30 augustl: here's an example of what I want to do: https://www.refheap.com/paste/12180

17:31 the idea is that I have some routing code but I don't want it to be ring specific, i.e. I don't want to assume that request contains ":request-method", but also allow for just :method (which I actually use, since I have a system where I use HTTP-like routing over a message queue, no Ring

17:31 )

17:31 so the default implementation assumes a Ring map, but I want uses to be able to extend it

17:31 not sure if protocols are a good fit though :)

17:32 llasram: Protocols dispatch on concrete JVM types. If you don't already have concrete types or need the speed their implementation in terms of interfaces provides, then they probably aren't the right fit

17:33 You could create a macro which generates new `deftypes` which are basically just trivial map implementations, but it doesn't seem like the right solution to me

17:33 augustl: I do have just a normal map, yeah

17:33 llasram: Then multimethods seem like a much better fit

17:34 augustl: not sure what I should dispatch on since they're always just normal maps

17:35 llasram: If you were already contemplating wrapping them in some other type, you can just assoc in a dispatch key instead, or add :type metadata

17:35 :type metadata has other implications for e.g. printing though

17:35 augustl: hmm

17:36 types seems cleaner.. Not that I know what I'm talking about, though

17:36 and a protocol also means the methods will be neatly grouped

17:36 brehaut: augustl: concrete implementation types are only clearer if they already exist; if you are going to map all your data to a concrete type just for dispatch, multimethods are better

17:37 llasram: Types are only cleaner if you really need a concrete type for some reason, which I don't believe is the case here

17:37 As for grouping together, you can create a macro which lets you define groups of related multimethods

17:37 I don't have it published publicly anywhere ATM, but I use such a macro pretty extensively in my work codebase

17:38 For ex: https://www.refheap.com/paste/12181

17:39 augustl: why do that when protocols work just fine? (Except I'm not sure they do, since I'm still not sure how to create a special map that's a map except a different type)

17:40 llasram: augustl: Basically, you could implement protocols in terms of multimethods, but not vice versa

17:40 Grouping together related functions is just a nicety, making code a bit easier to read by making the relationship clearer

17:41 But the fundamental machinery of multimethods is more powerful and flexible

17:41 I'd even argue that protocols only define functions together because they *need to*, due to their JVM implementation in terms of interfaces, which requires that all the functions be known at once

17:41 augustl: you can defmethod for defmultis from another namespace, right?

17:41 llasram: Yes

17:47 jimkcar: is there a white paper or something that shows how Datomic uses the key-value storage underneath. i.e. what do the keys and values look like

17:48 or is that proprietary?

17:51 bbloom: jimkcar: it's likely proprietary, but it's not that difficult to imagine what it looks like if you understand a bit about how datomic indexes work

17:52 jimkcar: in short, key-value or other storage is used like a big hash table

17:52 jimkcar: you don't store one datum per key, you store many

17:52 gfredericks: amalloy: hiredman: whoops I ran away; thanks for the discussion :)

17:53 bbloom: jimkcar: so basically, you name the key after some identifier for a page of data

17:53 jimkcar: yeah that's kind of what I figured. I work a lot with the Apache Accumulo project and was wondering how difficult it would be to adapt Datomic to use it.

17:53 augustl: jimkcar: all I know is that it's a 3 level deep tree structure

17:53 bbloom: jimkcar: i assume not that difficult, considering they already support other sorted key value stores

17:53 augustl: the actual data they put in storage, that is

17:54 hiredman: it sounds like the main requirement is conditional writes

17:54 bbloom: augustl: what are those levels?

17:54 augustl: supporting key/value stores is not necessarily easy. For riak, they need to use zookeeper as well, to keep track of things

17:54 bbloom: augustl: i assume the correspond to elements of EAVT

17:55 augustl: bbloom: I'm not sure..

17:55 hiredman: augustl: zookeeper is used to enable conditional put on riak

17:56 jimkcar: Does Datomic have any sense of permissions like Accumulo? There can be a lot of data in Accumulo, but you may not be able to access it all.

17:56 augustl: no, none at all afaik

17:59 bbloom: yeah, i think lack of permissions is an intentional design decision

17:59 hiredman: jimkcar: you might check out cascalog, since it looks like Apache Accumulo plays well with hadoop

17:59 bbloom: there are many types of permissions systems and you can build them on top if you want

17:59 augustl: llasram: seems another reason not to use defrecords "in place of" maps is that records aren't callable, like (my-map :key)

17:59 hiredman: cascalog is a datalog like layer on top of serveral other layers on top of hadoop

18:00 jimkcar: yeah I've played with cascalog a bit. Full disclosure, this is me: http://hadoopsummit2013.uservoice.com/forums/196821-enterprise-data-architecture/suggestions/3714766-managing-hadoop-and-accumulo-with-clojure

18:00 bbloom: augustl: whoa, i didn't believe you, but you are right...

18:00 augustl: indeed, it's very common to build APIs (with authentication etc) on top of datomic peers.

18:00 bbloom: wut? :)

18:01 bbloom: augustl: i did (defrecord Point [x y]) and then ((Point. 5 10) :x) and was astonished to see an exception cannot be cast to clojure.lang.IFn

18:01 augustl: ah right

18:01 dakrone: spligak: just released 0.6.5 with this commit: https://github.com/dakrone/clj-http/commit/426049c93eba360c6eca92fc542b70627a7d2be3

18:02 jimkcar: augustl: the problem with that is that the peers have the data in hand, regardless of the permissions

18:02 augustl: jimkcar: indeed

18:03 I'm not sure about the specifics, but I guess that the fact that peers gets transactions pushed to them is one of the reasons there are no permissions on that level

18:03 to avoid fancy routing on the transactor to correctly distribute data based on permissions

18:05 jimkcar: I really need to grok Datomic a bit more.

18:05 bbloom: what fucking year is it that jira still breaks my fucking back button!?! argh.

18:05 augustl: jimkcar: feel free to ask :)

18:05 amalloy: bbloom: really? i've had many problems with jira, but not that one

18:06 bbloom: amalloy: if you issue a "quick search" query, you get an "Issue Navigator" results page

18:06 amalloy: the URL does not contain your search terms

18:06 if you click an issue and then press back, Chrome warns about content resubmission

18:06 amalloy: bbloom: well, there's your problem: you're using jira's search feature

18:06 jimkcar: augustl: well i don't really understand how it can scale. If the index is at the peer for searching, doesn't that limit your index size quite a bit? How do you deal with Petabytes of data in your store?

18:07 bbloom: amalloy: how dare i attempt to use software

18:07 augustl: jimkcar: datomic is not a good fit for that

18:07 jimkcar: if by petabytes of data you mean petabytes of data in use

18:07 if you need to average or count a petabyte of numbers, with datomic you need to have a petabyte of memory available. So you're better off with something like map/reduce in that case

18:08 bbloom: augustl: datomic can still be useful for that, but you just need a layer above it to coordinate the subprocesses

18:08 jimkcar: augustl: so what is the most common Datomic use?

18:08 augustl: jimkcar: however, datomic has a LRU cache for data on the peer. So say you're building a project management system. Projects that haven't been active for a while (for example, due to limiting number of active projects in the app to 10 and having an archive function that makes the project inaccessible from the app) makes the data for those projects disappear

18:08 from the cache

18:08 </randomexample>

18:09 bbloom: jimkcar: datomic is designed to be a better foundation for the use cases you'd traditionally use a SQL database for

18:09 augustl: partitions can also help, if you have completely unrelated domain objects that typically are used in separate peers for each partition

18:09 jimkcar: you need to have the active data set in RAM anyway for most databases, to get good performance ;)

18:09 spligak: dakrone, hey thanks! that's really helpful when working with things like couch. allows me to define that meaningful/exceptional line nicely (and in the correct layer, IMHO)

18:10 dakrone: spligak: thanks for the report, it's a nice feature to have

18:10 jimkcar: I guess good performance is relative.

18:10 augustl: bbloom, jimkcar: yeah, there's aggregation right?

18:10 Roxxi: Aside from Cascalog and Clojure-hadoop, are there any other popular clojure map-reduce libraries that people commonly use?

18:10 jimkcar: augustl: yes there is aggregation

18:11 Roxxi: not that I've used

18:11 augustl: don't have a good answer to your guess :)

18:18 jimkcar: augustl: so there is only ever a single transactor in play?

18:19 augustl: jimkcar: yes

18:19 technomancy: Roxxi: it's easy to roll your own over a queue: https://github.com/technomancy/die-roboter

18:19 augustl: jimkcar: you can have failover transactors ready and waiting to avoid having to wait for JVM startup etc

18:20 jimkcar: but only one is running at a time, in order to achieve ACID

18:20 jimkcar: so if you need arbitrary write scale (in other words, you can't have ACID), Datomic is not a good fit

18:21 note however that because the transactor only does writes, it is likely easier to scale than an ACID database where the same process/server/whatever does both reads and writes

18:22 jimkcar: augustl: yes I see. I wonder if you could have both. Accumulo is highly optimized for data ingest. A single Transactor wouldn't be able to keep up.

18:23 is it possible to make bulk updates to the transactor indices?

18:23 corecode: i think clojure will become my next favorite language

18:23 augustl: the actual indexing operation is eventual. The ACID transaction only writes a log to make it fully ACID, then it's eventually merged into the index tree

18:26 ..if that's what you mean

18:28 jimkcar: augustl: So I'm thinking it would just be an app service along side your main data store.

18:37 ayia_: Hi guys! Can I wrap some code in other code depending on a condition? I have a parameter to my function. If it is a seq then i should evaluate some code. If it is a string I should wrap this code in (apply str ...).

18:39 brehaut: ayia_: maybe https://github.com/flatland/useful/blob/develop/src/flatland/useful/fn.clj#L30-L48 is what you are after

18:41 janiczek: hi guys, having a trouble with namespaces :/ i want to use this https://github.com/clojure/java.jdbc - no trable so far, [org.clojure/java.jdbc "0.2.3"] in project.clj, (:use clojure.java.jdbc) in (ns), but when I want to use clojure.java.jdbc.sql, it screams at me... how should I refer to it? ( https://github.com/clojure/java.jdbc/blob/master/src/test/clojure/clojure/java/test_sql.clj )

18:44 technomancy: janiczek: that's a test namespace

18:45 ayia: brehaut: thanks... so, there is no "standard" way...

18:46 janiczek: technomancy: I'm not requiring clojure.java.test-sql, but clojure.java.jdbc.sql

18:46 brehaut: janiczek: i wasnt aware that the jdbc.sql was in a released version yet

18:47 technomancy: yeah, you're looking at the master branch rather than the release

18:47 janiczek: aha ... so it's just on a github but not in clojars yet?

18:47 brehaut: and its also pretty experimental

18:48 janiczek: can I refer to it as 0.2.4-SNAPSHOT, or is 0.2.3 the most I'll get from leiningen?

18:49 technomancy: janiczek: that varies by project. typically contrib projects don't deploy to clojars, so it might be more difficult to track down

18:50 janiczek: tried it, clojars don't have 0.2.4-SNAPSHOT. hmm ... so right now my chances are a) just don't use it, b) copy the file from github to my project?

18:50 augustl: jimkcar: that could work, yeah. As long as you don't try to alter what datomic puts in storage for itself :)

18:50 technomancy: janiczek: you could ask the maintainer to deploy a snapshot or install it locally from a checkout

18:50 brehaut: janiczek: https://oss.sonatype.org/index.html#nexus-search;quick~java.jdbc

18:51 janiczek: "0.2.4-20130227.030138-16" appears to be on the sonatype snapshot repo

18:51 technomancy: that's weird; I thought the next version was going to be 0.3.0

18:52 brehaut: technomancy: true, i thought so too

18:52 janiczek: how do I make leiningen see that repo? it says it can't find the artifact in clojars, so it's probably only searching that

18:53 or should I just download that jar?

18:53 technomancy: janiczek: see `lein help sample`; look for :repositories

18:54 brehaut: janiczek: 'just download the jar' is almost never the right idea with leiningen / maven

19:04 janiczek: YES, did it :) thank you!

19:17 augustl: hmm, I'm not able to sign stuff for "lein deploy" it seems. I generated a key, and "lein deploy" prompts me for "Enter passphrase:". I enter it, and nothing happens. If I hit enter one more time, I get "gpg: Invalid passphrase; please try again ...". Running OS X. Any suggestions?

19:21 are there any getting started guides for using leiningen to deploy to clojars from os x?

19:23 Frozenlock: seancorfield: Do you have some tips on references in mongo/congomongo? I have some collections that are becoming quite big (>Go) and was thinking about spliting them at each 'keyword'. For example {:main {:a data :b data :c data}} ----> {:main ref_a ref_b ref_c} {:a data} {:b data} {:c data}.

19:23 TheBusby: augustl: I did it last night for the first time and it as pretty simple

19:23 augustl: had to use macports to get gnupg installed, and then setup a gpg key

19:24 seancorfield: Frozenlock: I haven't had cause to use references in MongoDB yet, sorry

19:25 augustl: fixed it by not having a passphrase for the gpg key

19:29 how do I find my gpg public key to paste into the clojars profile page?

19:33 xeqi: augustl: gpg --armor --export $KEY_ID

19:33 amalloy: Frozenlock: have you read http://docs.mongodb.org/manual/applications/database-references/ already? it seems quite likely to me that you only need manual refs, which are not hard

19:34 augustl: xeqi: hmm, I don't have $KEY_ID in my env

19:35 xeqi: augustl: `gpg --list-keys`, find your 'pub' one, use 8 characters after "/"

19:36 augustl: xeqi: thanks!

19:37 xeqi: technomancy: ^ has that been documented anywhere besides your inital mailing?

19:37 technomancy: xeqi: no, it should be added to clojars

19:37 but right now that's only used for promotion, which is basically on hold

19:38 I mean, it's documented as part of GPG of course

19:48 amalloy: technomancy: it looks to me like LEIN_IRONIC_JURE's error message doesn't match its behavior. its regex doesn't match "clojure", but its message says "clojure" isn't allowed. it also says it doesn't allow *jure, when really it means *jure* as far as i can tell: eg, jurey-duty is forbidden by that check

19:51 technomancy: amalloy: the check for "clojure" is outside the regex

19:51 but yeah, I'd take a pull req for the latter

19:51 we kept getting confused newbies who named their own projects "clojure" =\

19:52 I should probably go in and replace that message with something more sarcastic

19:52 amalloy: technomancy: that's my point. this cond-clause never "fails" for the name "clojure", so why does its error message tell you about it?

19:52 technomancy: oh, gotcha. dunno. I took a pull request for that and didn't read it carefully.

19:52 amalloy: the other clause, which matches specifically clojure, has the right error message

19:52 technomancy: story of my life

19:54 amalloy: technomancy: a friend notes: "words you can't use: abjure, adjure, conjure, injure, objure, perjure. so much for my lawyer-themed wizard-hospital simulator!"

19:54 technomancy: heh

19:54 brehaut: ha

19:54 technomancy: conjure and I think perjure are taken; my condolances

19:54 amalloy: abjure too

19:55 technomancy: there was one particularly outrageous one that prompted me to add that check where before I had only been thinking about it

19:55 TimMc: $dict objure

19:55 lazybot: TimMc: : To swear.

19:55 technomancy: an objective-C bridge IIRC

19:55 Couverjure argh

19:55 amalloy: ugh. already this discussion has induced semantic satiation for the syllable "jure". none of these look like words

19:55 TimMc: Oh man, that's what I should have called swearjure.

19:56 gfredericks: TimMc: crap.

19:56 TimMc: technomancy: http://static.tvtropes.org/pmwiki/pub/images/stopallthis_5242.png

19:56 amalloy: TimMc: abjure is a good name for some kind of security service

19:58 technomancy: TimMc: http://p.hagelb.org/outrage.jpg

19:59 gfredericks: technomancy: why your domain name gotsta be so cool :(

20:00 technomancy: that's the thing with domains

20:00 gotta act fast

20:00 gfredericks: I had lolwaffle.com for a year or so but never figured out how to make a million dollars with it.

20:01 TimMc: I own oddlyshapedrock.com for no good reason.

20:01 technomancy: I assume roflwaffle was taken?

20:01 brehaut: ahaha

20:01 gfredericks: technomancy: too explicit

20:02 lolwaffle's users are sophisticated and do not need to be spoonfed the joke

20:02 * technomancy nods solemnly

20:03 gfredericks: they are also X for any property X

20:03 TimMc: What joke? I don't get it!

20:03 * TimMc goes back to watching America's Funniest Explained Jokes

20:05 TimMc: There's also nextnewsound.com, which would probably be awesome for I dunno, maybe a music reviews site?

20:08 brehaut: TimMc: markov based music reviers using pitchfork.com's reveiws to generate fictional reviews of bands you havent heard of yet

20:08 (and pitchfork so that they are snarky)

20:08 TimMc: That's... that's a really good idea.

20:08 Look, if anyone wants to write that I will host the shit out of it.

20:08 technomancy: I always had a dream of putting together a compilation album where real artists would each submit a track as a fictional band.

20:09 brehaut: kind of like the gorillaz and new order

20:09 but more people

20:09 amalloy: if that's what living in seattle does to you, i think i'll stay in california

20:09 gfredericks: if a band makes real music in what sense could it still be fictional?

20:09 amalloy: right? seattle is where the music hipsters are? maybe that's portland

20:09 technomancy: gfredericks: no, I mean like musicians who are established from an existing fictional universe

20:10 TimMc: technomancy: Somehow this feels related: http://www.thesecretknots.com/2012/12/06/how-to-make-an-album-with-people-instead-of-songs/

20:10 technomancy: like the Tenmen or Jason Taverner

20:10 rabbit_airstrike: I think Seattle, Portland, and Brooklyn compete for music hipsters

20:10 gfredericks: technomancy: heck if I know of any fictional musicians but I get it now

20:10 brehaut: TimMc: if i was smart enough at the markov stuff, i'd totally be building that for you

20:10 TimMc: brehaut: The core of it is super easy. I just don't feel like doing all the plumbing.

20:10 brehaut: real musical hipsters come from towns you've never heard of

20:11 technomancy: TimMc: interesting

20:11 amalloy: portland still holds the crown

20:11 metellus: I'm sure someone else has done the Markov work somewhere

20:11 rabbit_airstrike: psychic tv did a double album where all of the listed artist credits were of people who were not psychic tv

20:12 TimMc: I'd probably just throw all the data together and use Dissociated Press, which is easier, albeit a bit slower.

20:12 gfredericks: seattle is the _____ of portland

20:12 technomancy: I had a bunch of other bands I would include too, but I've forgotten now

20:12 possibly Sex Bob-omb, but that's been done now.

20:12 rabbit_airstrike: "bigger, gloomier, more urban sister of"

20:12 gfredericks: "texas"

20:13 amalloy: technomancy: i think youo'd have to include at least one real band that everyone makes fun of, "because they've never released any real music"

20:13 TimMc: gfredericks: What do you suppose France's Paris is?

20:13 technomancy: amalloy: in this alternate fictional universe featured on this band's web site, $BAND are a creative, dynamic group of artists

20:25 ambrosebs: technomancy: I don't think I'll be using metadata as the primary storage for static types. The main worry I have is whether vars exist at all at the point I do the static type checking. A secondary worry is I want complete control over when and how my types are updated.

20:26 devn: hey all

20:27 technomancy: updating metadata after the fact is easy

20:27 brehaut: ambrosebs: I've started dabbling with c.typed on a new project; ive hit on some code thats very Any-full. Am i right in assuming that you can't do much with Any?

20:27 technomancy: I don't understand the use case for wanting to annotate non-existing vars though

20:29 devn: simplest way to do pagination in a compojure+ring+lib-noir web app?

20:29 ambrosebs: My point is I can freely put my type annotations wherever I want, and it doesn't matter whether I've compiled the relevant namespace for the var to exist.

20:29 devn: without writing the damn thing myself

20:29 TimMc: devn: Like, first/prev/next/last?

20:29 ambrosebs: So it can't be the primary storage, but I'm ok with adding redundant metadata.

20:29 technomancy: I understand what you mean, I just don't see the value.

20:30 hiredman: you can annotate non-existant vars by creating the vars

20:30 devn: TimMc: Yeah -- I'm using elastisch and right now I have it returning 25 results from the offset which I'm calculating

20:30 hiredman: ,(doc intern)

20:30 clojurebot: "([ns name] [ns name val]); Finds or creates a var named by the symbol name in the namespace ns (which can be a symbol or a namespace), setting its root binding to val if supplied. The namespace must exist. The var will adopt any metadata from the name symbol. Returns the var."

20:30 technomancy: to me, having to refer something at compile time is extra overhead

20:30 I would expect to be able to add annotations to a file that has none without touching my ns form

20:30 otherwise I'll be less inclined to actually do it

20:31 TimMc: devn: Well, this is a thing I wrote that computes some stuff based on page number + total results: https://github.com/timmc/handy/blob/develop/src/org/timmc/handy.clj#L137

20:31 devn: TimMc: but im not calculating the number of pages, and so on. I'm just looking for a simple way to have a "model-like" thing so I can get pudding-simple pagination without a lot of messing around.

20:32 TimMc: it would be nice if there was some kind of convention for adding pagination to "model-like" things in clojure web apps

20:32 TimMc: Yeah, pagination is surprisingly... involved.

20:32 devn: in rails (hold your sighs) things like will_paginate and so on have always been very popular

20:33 it would be nice if we could just allow people to define a basic pagination partial in their templating language of choice, and then let the person using the library provide it with a couple of required params

20:33 technomancy: devn: take and drop on lazy seqs are *so* close to getting you that

20:33 sadly there is no way for drop to signal that a given section of the lazy seq is going to be discarded

20:33 hiredman: technomancy: devn is talking about generating pages

20:33 devn: technomancy: yeah, but with elasticsearch you say: "give me 25 results, from the following offset"

20:33 technomancy: I know

20:34 TimMc: drop and give me 50

20:34 devn: you could get them all, but idk, just give me some simple pagination brothers

20:35 TimMc: technomancy: hiredman: Does it seem ill-advised or wrong to have a library, let's call it "paginatjure" just to irk technomancy

20:36 TimMc: paginature

20:36 ambrosebs: technomancy: Yes, that would be cool.

20:36 TimMc: oh, yes: paginatjure

20:36 devn: where you basically say: here is the basic paginate function. it takes a template you specify where prev, next, last, first, and so on are specified

20:36 qbg: pajinature

20:36 TimMc: devn: It's *hard*. There are three basic cases: No results, page out of bounds, and valid page.

20:36 devn: honestly i dont know what it looks like, im just trying to think of a way to provide a very general pagination library where someone using hiccup, enlive, whatever, can use it

20:37 ambrosebs: technomancy: I'm not sure how I'd recognise the type annotations though. Right now I check for dummy invocations of core.typed vars, which signify where type annotations are.

20:37 TimMc: devn: After that, you have to have or not have a link for each of first/prev/next/last, and maybe even swap out images for each of those.

20:37 The template looks like a goddamn mess when you're done with it.

20:38 devn: TimMc: that's totally fine though! what im suggesting is that there is a library of disgusting templates written in various templating languages

20:38 TimMc: heh

20:38 Here's what I ended up doing for my photo gallery, UI-wise: http://gallery.brainonfire.net/filter/run?tag[Content][]=flower

20:39 devn: and consumers of the lib can say: (:use [paginatjure.templates.hiccup :only (pagination-controls)])

20:39 TimMc: It was easier to write. :-P

20:39 paje

20:39 devn: and then you just have [:html [:body [:div#pagination (pagination-controls)]]]

20:39 or something

20:40 pagination-controls takes a few args

20:40 but it offloads as much of the bs as it can

20:40 TimMc: (pagination-controls 50 0 7145)

20:40 + a link builder function, I guess

20:41 devn: TimMc: i know this is needed, just not sure what the right way would be to do it for consumers

20:41 maybe just assume you're using hiccup at first?

20:41 TimMc: And you'd have to tell it where to find images, if you're using images...

20:42 devn: that's extra puddin'

20:42 TimMc: I dunno, I'd want something pretty generic.

20:42 devn: at first it's just a damned (range 1 (calculated-max-num-pages))

20:45 qbg: Maybe you'd want a collection of page number to label mappings?

20:48 devn: http://getclojure.org/search?q=comp+AND+juxt&num=2

20:48 http://www.github.com/devn/getclojure

20:48 TimMc: you can see the nastiness im doing for pagination. basically it doesn't work for lots of things. i just unconditionally assume there's 10 pages of results.

20:49 it also needs some design love, but it's seeded with 32k records or so and returns results and plays nice with query strings

20:50 TimMc: Oh, this is that thing.

20:50 devn: TimMc: that thing :)

20:51 TimMc: patches very welcome. i'd like to make it really nice to use.

20:51 that includes adding some kind of a rating system so people can rate examples based on a couple of dimensions

20:51 TimMc: devn: Well, the org.timmc/handy thing will certainly tell you how many pages of results you have, and what the next record offset will be.

20:51 devn: like "idiomatic", "concision", etc.

20:52 TimMc: devn: "swearjure"

20:52 devn: haha

20:52 TimMc: "someone trying to break a bot"

20:52 devn: "abomination (but still fun anyway)"

20:52 TimMc: I guess those are more categories than ratings.

20:53 devn: "thus spake rhickey"

20:53 etc.

20:53 Sgeo: What's an abomination?

20:53 devn: like definitionally?

20:53 or what we were referring to?

20:53 Sgeo: What you were referring to

20:53 devn: we were talking about swearjure

20:53 Sgeo: Ah

20:54 devn: I kid the swearjure people. I think it's cool, just trying to guard poor defenseless newbs from running away screaming when they see it.

20:55 (#(% %) #(% %))

20:55 still my favorite

20:55 Sgeo: That's pretty famous (that concept, not the swearjure), I think

20:55 ghadishayban: arrgggh local clearing bug in 1.5: https://gist.github.com/w01fe/5088279

20:55 devn: yeah, i just like the symmetry in swearjure

20:56 it looks nice

20:56 ghadishayban: can't figure out whether it was my code that is blowing the heap or the bug...

20:56 Sgeo: The Haskell version of that uses letters

20:57 (\x -> x x) (\x -> x x)

20:57 Although that can be written as

20:57 app id id (app id id)

20:57 I think

20:57 (According to lambdabot)

20:58 yedi: how painful is the process of working with nodejs through cljs

20:59 Sgeo: erm, not app, ap

21:00 ghadishayban: yedi: check bodil stokke's wonderful work on github, and her video https://www.youtube.com/watch?v=MTxNnYfWHOw&list=PLZdCLR02grLoyWsKpovatiBYJyf-RKx0c

21:00 Sgeo: id<*>id$id<*>id

21:00 Should be the same

21:00 Don't know how to get rid of id there

21:00 yedi: ghadishayban: many thanks

21:00 Sgeo: Also, apparently this is not actually valid Haskell

21:01 ghadishayban: yedi: github/bodil/cljs-noderepl

21:01 Sgeo: Bot is manipulating things that don't actually type

21:01 My head hurts

21:04 Can write a function that can take itself

21:04 (id can take itself)

21:04 But can't write (\x -> x x) which can only take a function that can take itself

21:12 gfredericks: ghadishayban: I had a memory leak a week or two ago which was very similar to the bug but I think just regularly my fault

21:14 ghadishayban: gfredericks: yeah - it's hard to tell sometime

21:15 gfredericks: when you reduce over a large file under a with-open, then try/catch on the inside as well...

21:30 gfredericks: Raynes: dangit you got me using ^:private now

21:30 Raynes: gfredericks: Good, good.

21:31 gfredericks: but it will scare the n00bs

21:33 ambrosebs: technomancy: Sorry I'm slow. I finally figured out your beef was with the massive dependency that is all of core.typed. Is it you'd rather core.typed just didn't exist at runtime (production?) and just leaves a trace of type annotations on vars?

21:34 I was thinking about the actual type checking strategy.

21:36 bbl

21:43 gfredericks: is there a good local name for hiccup data?

21:46 devn: best JS syntax hilighter for clojure?

21:46 amalloy_: Raynes: opinions?

21:46 Raynes: syntaxhilighter

21:46 brehaut wrote a great lexer for it.

21:46 devn: is that still the one to use?

21:46 the one from a couple years ago?

21:46 Raynes: Probably.

21:47 devn: cool, didn't know if something had moved into its place

21:47 brehaut: i dont think anybody else has been insane enough to bother

21:47 devn: i just finished doing cleanup on (clojure.string/join "|" (map (comp str :name meta second) (ns-publics 'clojure.core)))

21:48 yogthos: syntaxhilighter is pretty fantastic :)

21:48 devn: for a giant regex

21:48 trying to use google prettify and am getting more and more disappointed

21:48 should have stuck with syntaxhighlighter, but it adds way more markup than prettify

21:48 probably for good reason

21:50 brehaut: the clojure brush is structurally aware though, which i like to think makes up for it

21:50 devn: brehaut: yeah, that's important actually

21:50 im getting some weird issues with prettify highlighting this:

21:51 (let [-' [10]] (clojure.core/seq (clojure.core/concat -')))

21:51 it misses the outer closing ] in the let binding form

21:51 brehaut: my clojure brush misses #= stuff

21:52 devn: brehaut: that could be an issue for me :X

21:52 brehaut: patches welcome ;)

21:52 devn: hahaha

21:52 brehaut: what's the link to your highlighter?

21:52 brehaut: mostly i jest; theres quite a large loop you'd have to tackle

21:52 devn: inc-clojure-brush?

21:52 brehaut: https://github.com/brehaut/inc-clojure-brush

21:52 yeah thts it

21:52 devn: k cool

21:54 brehaut: its probably actually not too hard to add; mostly it'd be like supporting #_

22:28 gfredericks: is anybody on the latest CLJS? I got an NPE when the first form in my file was (foo.bar "docstring")

22:28 would like to report it as a "bad error" bug

22:34 devn: brehaut: what's the proper way to wrap with your brush?

22:35 gfredericks: to the beat

22:35 devn: brehaut: is that a css override i should use? basically, in some cases the code is an extremely long line. I'd like to wrap it for viewing purposes. Is there a good way to wrap?

22:35 gfredericks: well played

22:35 (inc gfredericks)

22:35 lazybot: ⇒ 16

22:35 gfredericks: a hypercube!

22:36 brehaut: devn: yeah i think you can add an overflow declaration to something syntax highlighter adds. just a moment

22:36 devn: (-> with-wrapper the-beat)

22:36 brehaut: i can show you what i mean, one sec

22:37 brehaut: devn: sure. actually, if you want it to wrap rather scroll, you might be out of luck. thats outside the scope of the brush itself

22:39 devn: brehaut: deploying right now

22:39 brehaut: yeah. i figured. it doesn't *have to* be that way

22:40 im just looking for a way to display long lines in a way that doesn't suck

22:40 tomoj: &(reduce-kv #(assoc %1 %3 %2) {} [:foo :bar :baz])

22:40 lazybot: ⇒ {:baz 2, :bar 1, :foo 0}

22:40 tomoj: &(clojure.set/map-invert [:foo :bar :baz])

22:40 lazybot: java.lang.UnsupportedOperationException: nth not supported on this type: Keyword

22:40 devn: either that means changing to something like opening the code in a new window or something

22:41 brehaut: http://getclojure.org/search?q=comp+AND+juxt&num=0

22:41 see the first result for example

22:41 brehaut: devn: i think you are limited to scrolling horiz; syntax highlighter outputs a table with two columns, one for line numbers, one for code. it limits options :/

22:42 devn: but i could do a collapse and truncate the line or something

22:42 with jTruncate

22:42 brehaut: devn: if you are happy with not having line numbers, you'll probably just be able to use some css

22:42 devn: idk, requires some imagination

22:42 brehaut: just a moment

22:42 devn: just trying to get an idea of what is usable

22:43 one other unfortunate bit is that i think SyntaxHighlighter kills my pretty printing of clojure code using code-dispatch

22:43 :(

22:43 oops, no, nevermind, still works

22:44 see: ((comp (juxt (juxt first second) (juxt (comp first rest rest) (comp second rest rest ...))))

22:44 on the above URL

22:45 brehaut: i dont know what it should look like?

22:45 devn: heh, nor do i

22:45 im soliciting suggestions

22:46 basically, what would be a good thing to do for disgustingly long values or output?

22:59 pppaul: you juxt my heart away

23:01 Raynes: Oh stop, you're making amalloy blush.

23:36 devn: you (comp)lete me

23:42 Raynes: devn: Maaaaaaaaaaarry me.

Logging service provided by n01se.net