#clojure log - Aug 19 2014

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

0:01 andrewchambers: greenznc:

0:02 I dont understand why you are sending and reading from same coroutine

0:02 greenznc: andrewchambers, I'm new to core.async

0:02 why not the same (go)?

0:03 andrewchambers: I haven't really used it. But... if its an unbuffered channel i dont understand how that can work

0:03 since it will send then pause waiting for someone to recieve

0:04 trying reading from 1 go, and sending from another go andsee what happens

0:05 greenznc: andrewchambers, it still alerts six times

0:06 quite strange

0:06 andrewchambers: xD im confused as to how it alerted one time

0:06 haha

0:06 I dont have clojure setup on this machine otherwise i'd try it out xD

0:07 greenznc: well there's a site called cljsfiddle

0:08 andrewchambers, http://cljsfiddle.net/

0:08 no more setup

0:08 andrewchambers: does that have core.async ?

0:09 greenznc: yes

0:09 clojurebot: Huh?

0:13 andrewchambers: greenznc: server doesnt seem to be working

0:14 greenznc: andrewchambers, what do you mean by 'server'?

0:14 andrewchambers: well, the ckjsfiddle run button does nothing

0:14 thats all i mean

0:15 greenznc: mine works

0:15 andrewchambers: must be my version of firefox

0:17 greenznc: http://pastebin.com/4Z0fTVNu

0:17 i was just gonna try that and see what hapens

0:18 hopefully i havent made an error, since i havent checked it

0:18 oh yup i made a bunch

0:19 http://pastebin.com/wfmEXugm

0:19 greenznc: let me see

0:21 andrewchambers: core.async is pretty sweet when i understood how it works

0:21 I think its alot simpler than some of the tutorials made it seem

0:22 greenznc: andrewchambers, it calls those three times which is correct

0:22 andrewchambers: I just took your code

0:22 hmm

0:22 greenznc: but I have no idea what's wrong with my code

0:23 andrewchambers: if you print in your doseq

0:24 does it print 3 times or 6 times?

0:24 thats my question haha

0:24 btw, im not sure if that code is a memory leak, it might be. Since the while true loop never terminates

0:25 after the first 3 sends

0:25 it will just sit there forever

0:26 greenznc: andrewchambers, I know it

0:26 andrewchambers: cool

0:26 :P let me know if you work out why its sending 6 times haha

0:27 im curious now

0:27 maybe your code is just running twice

0:27 greenznc: in the future I'd love to re-queue timeout-ed requests

0:27 andrewchambers: is it in your top level?

0:27 or you just copied out of a function

0:28 greenznc: no, in an om's (will-mount []) component

0:28 I meant in an compont's will-mount

0:29 andrewchambers: yeah i understood that

0:29 I dunno, i think the code is just running twice for some reason

0:30 but i think having two go statements is needed just because sending can block until it succeeds, and it cant succeed if there are no live go blocks reading it

0:30 and by block, i mean cause the javascript to idle or whatever.

0:32 greenznc: andrewchambers, I move the code to the top level and it works as expected

0:32 andrewchambers: must just be called twice

0:33 greenznc: yup

0:33 andrewchambers: cool, now work out how om works :P

0:33 hehe

0:33 greenznc: well, I execute the component twice for experiment purpose

0:34 and forget it XD

0:34 andrewchambers: lol

0:34 hmm, put! might not actually block ever

0:34 according to the docs

0:34 im not sure

0:34 :P

0:34 i dont know how much space there is in a default chan

0:35 greenznc: I'm thinking of how to close the channel when three requests are complete

0:36 andrewchambers, any suggestions?

0:36 andrewchambers: just send :done

0:36 or something

0:36 then terminate loop

0:36 greenznc: but how to figure out all of them are complete?

0:37 andrewchambers: oh, i understand

0:37 greenznc: count them and store somewhere?

0:38 or is there any existing mechanism?

0:38 andrewchambers: hmm

0:38 well you know how many you expect

0:39 because of the length of the map

0:39 so you could replace the while true

0:39 with a loop / recur

0:39 where you decrement a value

0:39 until it hits zero

0:39 greenznc: andrewchambers, yeah

0:39 many thanks

0:40 andrewchambers: yeah, you might need code to handle if there is an error

0:40 but i suppose an error can just be passed through

0:40 in :status or whatever

0:41 I want to build something with core.async and om

0:42 seems really powerful

2:23 rei: If I wanted to build a hash-map from varying levels of another hash-map, some form of multiple (assoc my-hash) be best? An example hash-map would be {:one :a :two ["hello" "world] :three #{1 2}}.

2:25 amalloy: rei: huh?

2:28 rei: amalloy: Well, to be more explicit, I'm scraping a website and I'm getting a hash-map of tags with the content I need. I want to build a final hash-map with only the relevant information at the top-level.

2:28 I'm trying to find the most idiomatic way of doing that.

2:31 amalloy: okay, so something like ##((fn [m keypaths] (into {} (for [[k path] keypaths] [k (get-in m path)]))) {:foo 1, :a 2, :b {:c 3}} {:x [:a], :y [:b :c]})

2:31 lazybot: ⇒ {:y 3, :x 2}

2:36 rei: amalloy: It took me a minute to wrap my head around it. Yes, that seems not only ideal but reusable.

2:36 I didn't think to be able to list off keypaths, as you've declared them.

2:48 amalloy: rei: anytime you're considering writing out multiple calls to the same function, consider whether you can write out a list of arguments to it, reformulate it into a reduce or a map or something

3:07 rei: amalloy: Thanks. I'll definitely look at that way in the future. Sorry for the delay, I was trying to amend your snippet to my code, and I realized I need to apply an arbitrary function and not always a path.

3:07 But it worked like a charm.

4:35 zanes: technomancy: Would you advise against using Robert Hooke at this point?

5:52 lvh: Hi :)

5:53 I have some recursive structured data (specifically, a "spec" which is a map with some stuff OR an unordered collection of specs OR an ordered collection of specs)

5:54 I was wondering if there's any tools for actually checking that based on a declarative description of what it looks like

5:54 I know of prismatic/schema, and typed clojure

5:54 also, importantly, these specs will be stored (they're immutable) and then read/executed later

5:55 so it's important that I can verify them separately from actually running the code that uses them

5:55 (because it has side effects)

5:55 (does that make any sense?)

5:57 llasram: If they're just data, then I believe you can use prismatic schema

5:57 I've only played with it, but that's what it does -- verify some data matches an expected structure

5:58 They have convenient ways to wire that up to common times you'd like to check such structure, but the base checking is available on its own as well

6:14 spradnyesh: what is "Ljava.io.PrintStream" in "ClassCastException java.io.PrintStream cannot be cast to [Ljava.io.PrintStream"

6:14 llasram: spradnyesh: array of PrintStreams

6:15 You'll most frequently see that when trying to invoke a Java varargs method. Java varargs is a fiction -- really you need to create an array to pass as the last argument

6:16 spradnyesh: llasram: where can i find documentation to understand more about it? and thanks for the help :)

6:18 llasram: Good question. The official documentation doesn't seem to cover it.

6:18 Here's a relevant StackOverflow question, but doesn't say much more than I've already said: http://stackoverflow.com/questions/11702184/how-to-handle-java-variable-length-arguments-in-clojure

6:20 I think Clojure considers that an interop issue, with the attitude "if you're doing interop, you need to understand the platform you're interoping with"

6:20 spradnyesh: llasram: thanks for the pointer; i'll go through it

6:20 llasram: Not optimal, but there you go

6:21 spradnyesh: i (mildly put) don't like Java. but in my current project am forced to use a java lib (not on clojars :( )

6:24 llasram: Well at least (hopefully) beats needing to re-write the features of the library yourself from scratch

6:25 spradnyesh: agreed

6:25 how can i add java-source to project so that i can "jump-to-source" (from emacs + cider) on exception?

6:47 kitallis: does 'lein repl' or 'lein test' know to pick up the right profiles from profiles.clj if you have a map like: {:dev {:env {:database-url-foo "jdbc:postgres://localhost/dev"}}

6:47 :test {:env {:database-url-foo "jdbc:postgres://localhost/test"}}}

7:24 spradnyesh: kitallis: try https://github.com/weavejester/environ

7:29 laurio: I'm having an issue requiring clj-time.coerce namespace in my project because the project at the same time also depends on amazonica library - lein repl prints "Exception in thread "main" java.lang.NoClassDefFoundError: clj_time/core/DateTimeProtocol" because amazonica declares joda-time dependency as [joda-time "[2.2,)"], i think.. This is my namespace declaration: (ns myproject.core (:gen-class) (:require [clj-tim

7:29 e.coerce]))

7:29 Is there an easy workaround?

7:30 drbobbeaty: laurio: Look at the ":exclusions" in the leiningen project file. That might help you.

7:31 laurio: it's a way to exclude certain dependencies in the jars/libs that are being pulled in. You only need one copy of Joda, and it may be conflicts in the different versions that are hurting you.

7:32 laurio: I tried to declare the dependency in project.clj like so: [amazonica "0.2.24" :exclusions [joda-time]] but unfortunately it doesn't seem to help even after lein clean.

7:32 Bronsa: laurio: I don't see what joda-time has to do with that error

7:32 laurio: amazonica and clj_time both depend on joda-time library

7:32 Bronsa: laurio: try to replace your require line with (:require [clj-time] [cljs-time.coerce])

7:34 laurio: ok bu the error only talks about an issue finding a clj-time protocol, it never mentions joda-time. what made you think it hasto do with your error?

7:37 drbobbeaty: laurio: When you do a 'lein deps :tree' - does it have any suggestions at the top about conflicts in the dependencies?

7:41 laurio: drbobbeaty, unfortunately 'lein deps :tree' doesn't seem to suggest about any conflicts

7:42 hyPiRion: laurio: ah, that's problematic. Which version of clj-time are you using?

7:43 drbobbeaty: laurio: Hmmm... OK... when you look at the tree, what does it say for the version of clj-time? Where is it in the tree? If you read the tree, it should make a lot of sense about how things depend on one another.

7:43 laurio: hyPiRion, [clj-time "0.8.0"]

7:43 Tried with lower versions also but this issue remained

7:44 hyPiRion: laurio: Could you try to check if [clj-time "[0.8.0]"] works? Also, would love to get the output from `lein deps :tree` too – might help

7:46 laurio: hyPiRion, [clj-time "[0.8.0]"] didn't help. here is the `lein deps :tree` output: [amazonica "0.2.24" :exclusions [[joda-time]]]

7:46 [com.amazonaws/amazon-kinesis-client "1.1.0"]

7:46 [com.amazonaws/aws-java-sdk "1.8.5"]

7:46 [com.fasterxml.jackson.core/jackson-annotations "2.1.1"]

7:46 [com.fasterxml.jackson.core/jackson-core "2.1.1"]

7:48 drbobbeaty: laurio: there should be an entry for clj-time in the output - I don't see it...

7:50 laurio: drbobbeauty, probably the text was too long - the relevant part is: [clj-time "0.8.0"] [joda-time "2.3"]

7:52 drbobbeaty: laurio: Well, if clj-time is using joda-time 2.3 - is there a problem with the other wanting to use 2.2? I don't know the other library, but I've used clj-time a lot.

7:52 llasram: laurio: Generally for pasting in content longer than one line it's best to use a paste site, such as refheap.com

7:55 laurio: thanks, llasram, I'll send a link instead next time

7:56 drbobbeaty: laurio: The 'lein deps :tree' should have reported the conflict in the different versions of joda-time.

7:56 llasram: Nah, they've got an exclusion

7:57 But as Bronsa mentioned a while ago, the exception mentioned is a problem finding the backing interface for a clj-time protocol, so probably has zilch to do with the joda-time dependency

7:57 laurio: Could you refheap a full exception

7:57 + stacktrace of what you see when you get your error?

7:58 laurio: llasram, how can I print that?

7:59 llasram: Oh, are you just getting the exception in your REPL, w/ no stack trace?

7:59 If so, try: (clojure.repl/pst)

8:07 laurio: llasrm, this is the output: http://pastebin.com/20LVNC4a

8:16 llasram: laurio: You've definitely got something weird going on with AOT. You said you did a `lein clean` already? Could you post your project file?

8:17 The problem is that there's somehow an AOT-compiled version of clj-time.coerce on your classpath, and that AOT-compiled version is being loaded & expecting to find the AOT-compiled classes for clj-time.core, but they somehow are not there

8:21 laurio: I thik you are right, llasram - think i didn't have this issue yesterday.. will investigate further. Thanks a lot for the help!

8:43 martinklepsch: I'm running a large-ish job on my macbook and would like to save the result directly to S3 — would I save that as a stream to S3 on the fly or would it be better to save it locally and then do a multipart upload to S3?

8:48 seems to me that stream uploading might be better since I don't know how big exactly the result is going to be and uploading can be started while it's still running

9:39 justin_smith: martinklepsch left, but he should really just do multipart from a stream (and use a lib that supports that)

9:43 lvh: what's the good url parsing thing? https://github.com/michaelklishin/urly

9:43 ?

9:59 justin_smith: if java.net.URL doesn't work for you, yeah, looks like urly is easier to use

9:59 I'm not sure what is best though, I have typically found java.net.URL to be good enough for my usage

10:06 lvh: justin_smith: since Date I just kind of assume the java thing is terrible

10:06 justin_smith: probably usually wrogn though :)

10:06 justin_smith: I have had no issues with URL

10:06 virmundi: Hello. I'm new to Clojure in general. I'd like to web develop with it. I'm not sure what libs/frameworks are current. I want to use Ring as the core to replace adapter.jetty with Parallel Universe's COMSAT when it comes out. I want to use bootstrap with it. Not horribly familiar with that. Where should I start?

10:07 llasram: Well, the java.net.URL hashCode implementation requires network access

10:07 So that's not good

10:07 justin_smith: virmundi: you would make html that serves bootstrap, that part is identical regardless of the server you use

10:07 llasram: lvh: What's your particular use-case?

10:07 lvh: llasram: what? that's ridiculous!

10:08 justin_smith: llasram: huh, I never would have known - my entire app is dead in the water without network access :)

10:08 lvh: llasram: well, amongst other things randomly store stuff in hash maps ;)

10:08 justin_smith: so's mine but that doesn't mean hashCode doing IO is acceptable :(

10:08 justin_smith: sure

10:08 lvh: llasram: I am parsing a URL and checking the scheme

10:08 virmundi: justin_smith: which frameworks support plugging into the bootstrap html? I'm use to the JSP/.NET fragment concepts.

10:09 llasram: lvh: If all your URLs are well-formed (and/or you can reject once which are not), then you can use the java.net.URI class

10:09 justin_smith: virmundi: most people in clojure land don't use frameworks - we prefer small composible libs for the most part

10:09 llasram: It's just a URI as a structured identifier, vs a handle to the representation of a resource the URL class tries to be

10:10 lvh: llasram: huh, interesting

10:10 llasram: my things are definitely UR*L*s though

10:10 llasram: Any URL is also a URI

10:10 URLs are the subset of URIs which also describe how to locate a representation of the identified resource

10:10 virmundi: justin_smith: that's a fair assessment. Is there a lib that supports that? My understanding is the noir and mustache are dead. Compojure looks to have a really steep learning curve.

10:10 lvh: sure; no problems with using the URI class :)

10:11 Glenjamin: compojure is pretty simple

10:11 which is both a shallow and steep curve, dependning on what you're doing

10:11 justin_smith: virmundi: compojure has nothing to do with bootstrap - it is simple, all it does is route requests to request handling functions

10:13 virmundi: there are a few templating options, which is where the bootstrap integration would come in. For the caribou project (which is actually kind of a framework) we made a templating lib called antlers, which can use arbitrary helper functions embedded in the markup, and is similar to mustache.

10:14 but I am sure you will get a bunch of other good templating recommendations

10:14 if the person making the templates knows clojure, enlive is great

10:14 virmundi: I'm looking at caribou now

10:14 thanks

10:15 justin_smith: we made caribou as an in-house migration target from using ruby

10:15 so we wanted something that matched the RoR workflows as much as possible for our agency

10:15 but it isn't neccessarily the best way to do a webapp in clojure

10:15 that said, it makes it very easy to make a database backed dynamic site with a cms

10:18 virmundi: justin_smith: I'm trying to piecemeal my libs, as you said. I want to use arangodb for the backend. Ring/* for routing and a template engine for the view. Bootstrap is just to get a simple mobile site without having to work. I'm purposely trying to avoid dynatism in the site. Full page reloads and all that.

10:19 justin_smith: ring doesn't do routing at all btw

10:19 it gives you a request as clojure data, some other code can use that for routing

10:20 (compojure for example, another alternative is caribou/polaris, there are other good ones too)

10:24 virmundi: justin_smith: so looking at polaris, it sits on Ring and provides routing; antler provides templating. But each is fairly independent. If I wanted to use Compojure I could with antler.

10:24 ?

10:24 justin_smith: absolutely

10:24 or you could use polaris plus some other routing lib

10:24 err. templating lib :)

10:25 though in practice I have been able to antlers for templating, but also pass in some pre-rendered enlive generated html (effectively using both templating engines at once - it actually made things simpler in a couple cases, believe it or not)

10:26 lvh: how do I figure out which exception prismatic/schema raises when validation fails? the traceback says ExceptionInfo Value does not match schema: (not (some (check % a-clojure.lang.PersistentHashMap) schemas)) schema.core/validate (core.clj:165)

10:26 is it an ExceptionInfo? that doesn't sound like an exception type

10:27 justin_smith: lvh: ExceptionInfo is used to pass extra data along with an exception

10:27 lvh: justin_smith: right

10:27 justin_smith: I'm writing a test though, and want to see that it fails

10:27 justin_smith: it could be that something you wanted to validate threw an exception and it's like "maybe we can validate the exception" then it's all like "nope"

10:27 just speculating

10:28 llasram: lvh: Or to elaborate: it is an Exception subclass provided by Clojure as an alternative to the Java RuntimeException, allowing delivery of a map of additional information about the exception

10:28 justin_smith: lvh: are you using the thrown / thrown-with-error features of clojure.test?

10:28 llasram: ,ex-info

10:28 clojurebot: #<core$ex_info clojure.core$ex_info@187514a>

10:28 llasram: Er

10:28 (doc ex-info)

10:28 lvh: justin_smith: I'm trying to use thrown?, but it requires me to tell it which exception is thrown

10:28 clojurebot: "([msg map] [msg map cause]); Create an instance of ExceptionInfo, a RuntimeException subclass that carries a map of additional data."

10:28 lvh: justin_smith: so that's what I'm trying to discover :)

10:28 justin_smith: lvh: just give it Throwable

10:28 it's all subclasses of Throwable :)

10:29 then you can check the actual class inside a catch on Throwable for example

10:29 btw do not catch Throwable in real (non test) code, that is way too expansive

10:29 lvh: justin_smith: I was hoping to discover the real class from the error :)

10:29 llasram: lvh: You know it -- ExceptionInfo

10:32 lvh: huh, why isn't that available by default

10:32 Unable to resolve classname: ExceptionInfo

10:36 johnwalker: justin_smith: i might use caribou for my next project. it looks outstanding

10:38 justin_smith: johnwalker: it's very big, but if you want a framework, it's definitely a framework

10:41 johnwalker: it seems to have aws built in

10:42 is there a lot involved in using other providers ?

10:42 seems like this could be abstracted away with jclouds

10:51 sdegutis: arrdem: wow you're one for holding a grudge aren't you?

10:51 arrdem: look, your choice, but if I were you I'd give it up; grudges are poison

10:52 justin_smith: johnwalker: I am sure it could be - I put aws in as an option because that was the service we used (so it is set to go for s3 / beanstalk / aws instance integration)

10:52 sdegutis: arrdem: anyway I done my apologizing, and I meant it sincerely; it's all on you now

10:52 justin_smith: johnwalker: but I frequently talked about wanting something more general / openstack compatible. I am sure it would be possible.

10:53 johnwalker: and further, if you have specific questions about how that would work, what the gotchas would be, feel free to contact me (I am noisesmith on github / caribou repo, or you can find me here)

10:53 lvh: I have a schema; I want RequestSpecs that are either SimpleRequests, or ordered collections of RequestSpecs, or unodered collections of RequestSpecs https://github.com/lvh/icecap/blob/master/src/icecap/schema.clj

10:53 it appears to work, but the error handling is not what I want it to be

10:55 here's the github gist explaining why I'm unhappy: https://gist.github.com/lvh/f2d743b02c8cccd4410b

10:56 mdeboard: MELPA's down :(

10:56 johnwalker: ok, thanks :)

11:01 clgv: lvh: wild guess defrecord vs deftype/java class?

11:02 lvh: clgv: the code is here: it's just (def SimpleRequest {:target ... something} )

11:02 err, https://github.com/lvh/icecap/blob/master/src/icecap/schema.clj

11:02 clgv: lvh: ah wait I confused the camelcase name with classnames ;)

11:02 lvh: there ^ ;)

11:02 llasram: lvh: Is that just the error you get when none of the cases match?

11:02 lvh: llasram: but the first case is supposed to match

11:02 llasram: Oh, I see

11:03 johnwalker: mdeboard: melpa is up

11:03 lvh: llasram: and if I use a thing that does match, it works fine: (s/validate RequestSpec {:target "http://www.google.com"})

11:03 so it is really the error case

11:03 johnwalker: i'm using it now, and there have been no tears in #emacs

11:03 lvh: I just don't know why the error reporting is so bad

11:04 llasram: lvh: Wait, what -- no, it doesn't match the first case. The scheme is `ftp` still.

11:04 mdeboard: Yeah melpa is up

11:04 I fail at the internet

11:04 I was typing .com

11:04 johnwalker: ahh

11:04 lvh: llasram: oh, derp

11:04 johnwalker: ;)

11:04 lvh: llasram: right, so it doesn't match the first case

11:04 llasram: I agree the error could be better, but it can't give you anything terribly specific, because all it knows is none of them matched

11:05 lvh: llasram: I wish it could give the error messages for *why* it didn't match those cases though

11:05 llasram: lvh: But it could be a different reason for each case

11:07 If you look at the code, the error means that the value being checked at that point was a map (it reports the type of the value the `either` failed to validate)

11:10 hugod: Is it possible to get a Var to pprint without its deref'ed value? i.e. #<Var@478f4a71> instead of #<Var@478f4a71: {:a :b}>

11:16 lvh: llasram: s/conditional apparently does more of what I want :)

11:18 Bronsa: hugod: with-pprint-dispatch & defining a dispatch for Var

11:35 AWizzArd: I have those two files: src/footest/core.clj and test/test/footest/core.clj In my project.clj I want to specify :jar-exclusions so that the test/ folder will be excluded, but not the file in the src folder. #"test/" as arg fails. Ideas?

11:37 With #"test/(?=footest)" I was partially successful, but it still included the empty folder "test/" in the resulting .jar.

11:37 hugod: Bronsa: thanks

11:40 Bronsa: np

12:07 greghendershott: If Racket can pollinate Clojure (e.g. Typed Racket), what's the biggest to-do for the reverse? Clojure's immutable lazy collections?

12:08 technomancy: greghendershott: applicable associative data types

12:08 and persistent vectors, I guess

12:09 lvh: I'm using prismatic/schema; (s/check RequestSpec []) ;; => [(not (present? "at least one"))]

12:09 greghendershott: technomancy: Also as I work on racket-mode I notice I have most of the same open items as you do, here: https://github.com/technomancy/dotfiles/blob/master/.gripes.org#racket :)

12:09 lvh: actually, writing it down made me figure it out.

12:09 greghendershott: The ones not already addressed by rackjure ;)

12:10 technomancy: greghendershott: great minds think alike =D

12:10 greghendershott: Arg lists, doc strings, doc sigs...

12:10 lvh: actually, wait, nope, that's still not it

12:10 greghendershott: The good news is DrR is awesome. Maybe the only downside is there's been less motivation to make that stuff for 3rd party tooling. (Just my guess why)

12:10 lvh: is there an easy way to get... something.. that allows me to put that value in a test, without guesswork?

12:11 technomancy: greghendershott: it's a bit of a category killer

12:11 but emacs will find a way

12:11 greghendershott: Oh god, I'm sorry, I thought I was on the #racket channel.

12:11 technomancy: ^ imagine that said in David Attenborough's accent

12:11 lvh: actual: (not (= [(not (present? "at least one"))] [(not (present? "at least one"))]))

12:11 greghendershott: I'm so sorry. Switching over.

12:11 lvh: not the greatest error message :(

12:11 technomancy: it's cool; we're pro-racket here

12:12 * greghendershott honestly I wasn't trying to "recruit" or "promote". slinks away

12:12 lvh: egads; that is a ValidationError!

12:12 TEttinger: greghendershott: there's already clj data structures for racket

12:13 greghendershott: TEttinger: Where?

12:15 TEttinger: greghendershott, at least some are here https://github.com/technomancy/rackjure

12:15 technomancy: hehe

12:15 TEttinger: that's a fork of greghendershott's lib =)

12:15 TEttinger: indeed!

12:15 I just saw

12:17 verma: hey guys, can anyone please review my mini blog post https://gist.github.com/verma/a15dea0689089e1287e3

12:19 lvh: Okay, so, I'm writing tests, and I want to compare the output of prismatic.schema.core.validate against something. How can I get the output of s/check in my test source?

12:19 It looks something like: {:target (not ("supported-scheme?" "ftp://example.test"))}

12:20 quoting the list doesn't work.

12:23 rweir: clojre.test?

12:23 lvh: yeah

12:23 rweir: apparently it's an exception made to look like a map that doesn't really work like a map, or something

12:23 rweir: oh, that's the 'repr' of the exception?

12:26 jakecraige: verma: Then only thing I notice off a quick look is that you just throw in route/resources "/" and route/not-found but don't mention them

12:26 milos_co1agen: verma: happy to give feedback but without knowing intended audience, it's hard. for now, mis-spelled convenience and json inconsistently capitalized.

12:28 TEttinger: jsON all the way

12:28 milos_cohagen: lol

12:29 justin_smith: TEttinger: differentiates it from its dual, jsOFF

12:30 jakecraige: hahah

12:31 TEttinger: javascript? Old, Fugly, Fetid.

12:31 milos_cohagen: verma: i'd include some links / references to where some of the magic comes from.. eg the lein new template, the embedded jetty, maybe how to start/stop via repl, the middleware code, etc. so a reader can explore more.

12:34 dnolen_: core.async release today w/ cljs tranducers support

12:35 tvanhens: Anyone experienced with riddley or clojure.walker? I'm trying to make a function that transforms edn into a validation and I'm running into a bit of trouble

12:41 mdeboard: Kind of random question, I thought nREPL was a general network protocol-type thing that was language agnostic

12:41 is that totally wrong?

12:42 mgaare: not totally wrong, no. but I'm not aware of nrepl servers for any other language

12:44 tvanhens: how could I convert all vectors in a nested data structure to lists?

12:44 gfredericks: clojure.walk/postwalk

12:45 verma: jakecraige, milos_cohagen thanks for the feedback :)

12:45 gfredericks: with #(cond-> % (vector? %) (list*))

12:45 tvanhens: giving it a go, had been trying with riddley before

12:45 verma: jakecraige, the route/resources stuff was generate by lein template stuff so I didn't think I need to explain that, but you're right, a little explanation won't hurt

12:46 tvanhens: wow, perfect thanks gfredericks

12:47 verma: milos_cohagen, I will include some more links and some repl stuff as well. Thanks for catching typos, I guess the intended audience is beginner to intermediate level clojure people

12:53 mdeboard: mgaare: I'm interested in creating one

12:54 mgaare: for which language?

13:18 PigDude: mgaare: erlang software runs from a networked repl

13:18 mgaare: i'm sure there are others

13:19 mdeboard: mgaare: I had to restart my IRC bouncer, I missed any response you may have given

13:20 PigDude: Elixir is actually the language I'm asking about nREPL for

13:20 not nREPL specifically, maybe, but just where to learn more

13:21 I'd like to have a `jack-in` command for elixir-mode

13:22 technomancy: mdeboard: are you aware of distel?

13:23 mdeboard: technomancy: negative

13:23 technomancy: mdeboard: supposedly it's possible to get emacs to speak the erlang distribution protocol natively

13:23 mdeboard: Interesting

13:24 how does that help here? I'm clueless.

13:24 starting from scratch here knowledge-wise

13:24 PigDude: mdeboard: I can't speak to elixir but it uses the erlang vm and creates beam files, so erlang emacs tooling should work

13:24 justin_smith: technomancy: btw, I saw your tweet about Midas Flesh, bought the bundle, found it to be excellent. Also the "ben franklinstein" gag from the bravest warriors comic in the bundle caused much amusement

13:24 mdeboard: PigDude: True, I guess I need to look into exactly what tooling they have.

13:24 for emacs, that is.

13:25 PigDude: mdeboard: erlang provides an official emacs distribution

13:25 mdeboard: I am laboring under the impression that Clojure is kind of cutting edge on the integrated REPL/code evaluation front. Maybe I'm wrong

13:25 PigDude: mdeboard: http://www.erlang.org/doc/apps/tools/erlang_mode_chapter.html

13:25 technomancy: justin_smith: the bundle is a bit intimidating; barely scratched the surface beyond midas =)

13:25 PigDude: mdeboard: as an end-user, it isn't particularly cutting edge, but i'm sure it has some nice things i don't notice

13:26 mdeboard: PigDude: It's cutting edge among the languages I use, anyway. I'm speaking specifically about emacs integration, not general tooling

13:27 technomancy: mdeboard: afaict erlang's distribution protocol is more advanced than clojure's, but the repl-specific parts of it tend to be total junk

13:27 mdeboard: I see

13:28 I refactored elixir-mode very heavily yesterday and now I can worry about niceties like better REPL & code evaluation... instead of syntax highlighting & indentation.

13:29 justin_smith: technomancy: on linux, the "comix" program is handy, (if you use the cbz files)

13:29 technomancy: http://p.hagelb.org/antipathy.jpg <- re: the fact that erlang just yells at you when you do ctrl-d

13:29 mdeboard: lol

13:30 PigDude: i hope that isn't the part you think is total junk ...

13:30 technomancy: PigDude: no, just the highest wtf level due to the frustration:difficulty-to-fix ratio

13:31 PigDude: heh

13:32 mdeboard: I actually don't even use Elixir at all now, I get way more satisfaction out of working on the language mode :P

13:32 PigDude: Tools Developer Problems

13:35 justin_smith: mdeboard: some kind of Monty Python skit "you want the elixer toolchain development toolchain developer, down the hall on the left, I am the elixer toolchain documentation toolchain developer"

13:35 mdeboard: haha

13:36 I'm just waiting to find a project that's actually a good fit for OTP

13:36 "generic website" ain't it

13:36 technomancy: definitely the hardest thing about learning erlang

13:36 justin_smith: just think how many telephone line switchboards we could have by now if erlang was more popular, heh

13:37 mdeboard: I have a computer vision thing I think could be sped up via paralellization w/ Elixir

13:37 it would actually be a good fit too, I think

13:37 but I know nothing

13:39 splunk: Anyone know any tools for unit testing that a function has reasonably constant memory usage, e.g. no head-holding?

13:41 justin_smith: ,(. (Runtime/getRuntime) freeMemory) ; splunk - maybe this helps?

13:41 clojurebot: #<CompilerException java.lang.SecurityException: Reference To Runtime is not allowed, compiling:(NO_SOURCE_PATH:0:0)>

13:41 SegFaultAX: Is there an easy way to override the formatter that c.j.jdbc uses for a specific type?

13:41 Specifically, jodatime datetimes

13:42 justin_smith: splunk: I would think you could check the freeMemory, run the code, then check that the delta in freeMemory is small enough?

13:43 though you may also need to check availableMemory (to see if it bumped up the allocated size in the meantime)

13:48 splunk: justin_smith: thanks

14:21 TimMc: I generally use WeakReference ot solve this problem.

14:21 ahoenigmann: if a function has this parameter list: [executor {:keys [fn interval]}]

14:21 what does it mean

14:21 it takes a executor yes

14:21 justin_smith: ahoenigmann: that's a destructuring

14:22 ahoenigmann: but {:keys [fn inteval]}

14:22 TimMc: It's not guaranteed, but I've always been able to use it to demonstrate that an object is GC'able

14:22 ahoenigmann: thanks justin

14:22 justin_smith: ,((fn [{:keys [:a :b]}] (+ a b)) {:a 40 :b 2})

14:22 clojurebot: 42

14:23 justin_smith: ,((fn [{:keys [a b]}] (+ a b)) {:a 40 :b 2}) ; identical, closer to your case

14:23 clojurebot: 42

14:24 ahoenigmann: hmm not getting it

14:24 justin_smith: the args to :keys are turned from keywords in the map, into bindings in the local scope

14:24 gfredericks: ahoenigmann: do you know what [executor {fn :fn, interval :inteval}] means?

14:24 without the misspelling

14:24 joshuafcole: ahoenigmann: destructuring is a way of taking arguments in one form and converting them into an easier to use form.

14:25 ninja'd

14:25 justin_smith: so {:keys [x y z]} is like (let [x (:x m) y (:y m) z (:z m)] ...) inside your function

14:25 ahoenigmann: ok so keys will take the keys from the map passed in

14:25 :keys will return a list of keys

14:25 justin_smith: no

14:26 it's just a syntax for binding locals

14:26 (keys m) returns all keys

14:26 but that's different

14:27 gfredericks: ,(macroexpand '(fn [executor {:keys [fn interval]}] ...))

14:27 clojurebot: (fn* ([executor p__83] (clojure.core/let [{:keys [fn interval]} p__83] ...)))

14:27 gfredericks: ,(macroexpand '(let [executor e {:keys [fn interval]} m] ...))

14:27 clojurebot: (let* [executor e map__108 m map__108 ...] ...)

14:27 gfredericks: ,(macroexpand '(let [{:keys [fn interval]} m] ...))

14:27 clojurebot: (let* [map__133 m map__133 (if (clojure.core/seq? map__133) (clojure.lang.PersistentHashMap/create (clojure.core/seq map__133)) map__133) interval ...] ...)

14:27 gfredericks: geez whiz.

14:28 justin_smith: yeah, the expansions don't help much here do they

14:28 gfredericks: I can map destructure any seq?

14:28 ,(let [{:keys [x y z]} '(:x 7 :z 10)] [x y z])

14:28 clojurebot: [7 nil 10]

14:28 gfredericks: ha

14:28 justin_smith: gfredericks: I guess that makes sense, for like & key args

14:28 amalloy: gfredericks: yes, that's how & keys works

14:28 gfredericks: OH I SEE.

14:28 amalloy: you're not really *supposed* to do it

14:29 gfredericks: so every map destructuring involves that check just in case

14:29 amalloy: yes

14:29 gfredericks: okay okay okay okay okay

14:30 ahoenigmann: ok

14:30 seeing [fn through me off

14:30 joshuafcole: Man. The more I learn the more I love macroexpand.

14:31 ahoenigmann: its not the same as (fn

14:31 (fn …

14:31 gfredericks: ahoenigmann: yeah it coulda been [interval fn] and meant the same

14:31 justin_smith: yeah it is a confusing local binding name

14:31 amalloy: ahoenigmann: yeah, i don't recommend using fn as the name of a local

14:31 ahoenigmann: :)

14:31 amalloy: then someone writes (fn x) and i'm like...augh, what is going on

14:31 justin_smith: it's more common to see f used (and then g for the next one)

14:32 so we can pretend we are doing math and shit

14:32 gfredericks: in this case it's the name of a key, so you have to destructure differently to name it something else

14:32 ahoenigmann: its a neat destructuring of being able to call a function on the destructed map

14:33 gfredericks: huh?

14:33 ahoenigmann: The fact that you can call :keys on the {}

14:34 justin_smith: :keys isn't a call there, it's a binding expression

14:34 ahoenigmann: ah

14:35 gfredericks: ,(let [{:fake []} {}] {})

14:35 clojurebot: {}

14:35 gfredericks: ,(let [{:keys [x]} {:x 1}] x)

14:35 clojurebot: 1

14:35 gfredericks: ,(let [{:keyss [x]} {:x 1}] x)

14:35 clojurebot: #<CompilerException java.lang.RuntimeException: Unable to resolve symbol: x in this context, compiling:(NO_SOURCE_PATH:0:0)>

14:36 gfredericks: I guess it's a difficult typo to make

14:36 (without noticing)

14:38 pyrtsa: ,(let [x :foo, {:keyss [x]} {:x 1}] x) ;; Unless you tried to shadow previous bindings

14:38 clojurebot: :foo

14:38 gfredericks: yeah that's why it's not impossible :)

14:45 amalloy: &(let [{:keyss [x]} {:x 1}] x)

14:45 lazybot: java.lang.Exception: Unsupported binding form: :keyss

14:45 amalloy: an unfortunate side effect of the change to namespaced-keys destructuring made in 1.6

14:45 it used to be a typo that was impossible to make

14:46 TimMc: ahoenigmann: (let [{:keys [a b]} m] ...) is macroexpanded to something like (let [val m, a (get val :a), b (get val :b)] ...)

14:48 ahoenigmann: is see

14:48 i see

14:49 TimMc: The let macro uses the binding macro to do all the heavy lifting, there; the fn macro does the same thing.

14:50 joshuafcole: ,#({:key %})

14:50 clojurebot: #<sandbox$eval26$fn__27 sandbox$eval26$fn__27@457235>

14:50 joshuafcole: ,#(hash-map :key %)

14:50 clojurebot: #<sandbox$eval53$fn__54 sandbox$eval53$fn__54@1166179>

14:51 TimMc: ahoenigmann: Err, rather, the destructure macro. Here it is: https://github.com/clojure/clojure/blob/master/src/clj/clojure/core.clj#L4237

14:51 amalloy: TimMc: the destructure function

14:52 TimMc: Oh, so it is!

14:52 gfredericks: ,(let [{:x :x} {:x 1}] x)

14:52 clojurebot: 1

14:52 TimMc: Makes sense.

14:52 gfredericks: &(let [{:x :x} {:x 1}] x)

14:52 lazybot: java.lang.Exception: Unsupported binding form: :x

14:53 * TimMc squints at clojurebot

14:55 amalloy: TimMc: i said, bad change that got introduced in 1.6

14:55 or, at least, good change with bad side effects

14:56 it came along with the ability to

14:56 ,(let [{:keys [foo/bar]} {:foo/bar 1}] bar)

14:56 clojurebot: 1

15:01 TimMc: I don't see how the :x thing works.

15:02 * TimMc grumbles and digs into the code

15:02 amalloy: TimMc: it calls clojure.core/name on the left-hand side of the pair

15:02 so as to discard the namespace part

15:03 and in doing so, fails to notice it was erroneously a keyword instead of a symbol

15:05 TimMc: Ouch.

15:12 amalloy: TimMc: now, i hadn't actually looked at the code; i'm just presuming this based on its behavior. i'm looking now, and the code is hard to follow, but i see (cond ... (keyword? b) (-> bvec (conj (symbol (name b))) (conj v)))

15:35 martinklepsch: I'm running a large-ish job on my macbook and would like to save the result directly to S3 — would I save that as a stream to S3 on the fly or would it be better to save it locally and then do a multipart upload to S3?

15:35 seems to me that stream uploading might be better since I don't know how big exactly the result is going to be and uploading can be started while it's still running

15:37 hiredman: martinklepsch: I would do something where you chunk up the stream as comes in and upload each chunk as an object to s3

15:40 martinklepsch: hiredman: so sort of like partition-all the incoming sequence?

15:41 justin_smith: martinklepsch: well, it won't be a sequence, it will be a byte stream right?

15:42 akhudek: martinklepsch: consider what happens when your connection inevitably goes down for a moment

15:42 martinklepsch: hiredman: when in general should one use regular put/stream uploading/multipart uploading? I don't fully understand except that multipart is resumable

15:42 *understand the differences

15:42 justin_smith: martinklepsch: stream and multipart are orthogonal

15:43 you can do multipart stream or or you can do a file all in one put

15:44 martinklepsch: justin_smith: so when would you use what?

15:44 justin_smith: multipart for things that are big or unknown total size

15:44 regular put for small things

15:45 martinklepsch: ah, I thought multipart would requrie the size to be known

15:45 justin_smith: no, the opposite in fact

15:45 for a regular put, it complains if you do not specify a size

15:45 but with multipart, all it needs to know is the chunk size

15:48 martinklepsch: justin_smith: ah thats cool. so mutlipart with clj-aws-s3 requires the uploaded object to be a java.io.File object

15:49 justin_smith: right - that's just a limitation in the wrapper lib

15:49 you can use hiredman's lib, or mine, to get around that

15:49 martinklepsch: justin_smith: do I remember correctly that you forked it for that reason?

15:49 justin_smith: right

15:49 caribou/clj-aws-s3

15:49 martinklepsch: yeah, looked at hiredman ones

15:51 justin_smith: I actually implemented the feature because I was doing async upload (a resized image) where I did not want to wait until I knew the total size of the output image before uploading (and I didn't want to put it on disk either - I don't need it on local disk at all, I just want to put it on s3)

15:52 I may not have fully gotten the "upload bytes before image is done resizing" part working iirc, but I definitely got "upload without knowing the total size first, and without creating a file" working

16:08 SagiCZ1: hey guys do you know if the author of braveclojure.com is #clojure channel regular?

16:08 justin_smith: I know he posts on reddit/r/clojure

16:10 SagiCZ1: justin_smith: thanks

16:35 anyone using LightTable? just stumbled upon it..

16:38 joegallo: do you have a question about it, or are you just looking for a show of hands?

16:39 SagiCZ1: joegallo: i am looking for someone who uses it to tell me if its any good

16:39 joegallo: before i investigate more

16:41 TEttinger: SagiCZ1, I used it but never tried too hard to learn the intricacies

16:41 I ended up having an easier time learning in NightCode

16:41 SagiCZ1: TEttinger: NightCode?

16:42 TEttinger: https://nightcode.info/

16:42 SagiCZ1: looks reasonable

16:42 lighttable has the instant code eval.. not sure if i need it

16:43 TEttinger: I could never get it working

16:43 it seemed temperamental for Swing and other graphical apps

16:43 nightcode has a repl in a pane at the botton

16:43 bottom

16:43 clojurebot: Huh?

16:43 TEttinger: and another to the side, actually

16:44 SagiCZ1: i see.. well so far i am using Cursive but its just a plugin which might be not-free in the future.. looks very similar to nightcode. also has REPL on the

16:44 bottom

16:45 btw are you happy with Paredit? i dont know how to learn it.. seems so alien to me

16:45 amalloy: SagiCZ1: it took a little while to get used to, but it's hard to imagine living without it

16:45 TEttinger: I turned it off in nightcode, it's easy to switch

16:45 you can turn it back on with a checkbox

16:45 martinklepsch: hiredman: I'm giving your S3 lib a spin but running this https://gist.github.com/mklappstuhl/c5cfef977252455fefb9 results in a few 400 errors without much more information. is there any way to get more information out of it?

16:46 SagiCZ1: amalloy: so were you just trying really hard? or did you study some kind of tutorial? i will do anything to reduce keyboard to mouse transtitions

16:46 TEttinger: i see.. cool stuff.. i might give NightCode a shot then

16:46 martinklepsch: SagiCZ1: I used lighttable for a few months and I think it's pretty awesome if you're starting out with clojure

16:47 amalloy: SagiCZ1: i referred to the paredit cheat sheet a lot at first

16:47 $google paredit cheat sheet

16:47 lazybot: [EmacsWiki: Paredit Cheatsheet] http://www.emacswiki.org/emacs/PareditCheatsheet

16:47 SagiCZ1: how is the integration of leiningen.. light table or night code both?

16:47 hiredman: martinklepsch: there is definitely more, you get a whole stacktrace (your paste is a partial stacktrace)

16:47 SagiCZ1: amalloy: ok that might be helpful, thank you for the link

16:47 TEttinger: SagiCZ1, yeah both are fully integrated to lein

16:48 SagiCZ1: TEttinger, i see

16:48 amalloy: wait and these are emacs commands?

16:48 hiredman: martinklepsch: likely you are using some tool that is hiding information about the exception in a misguided attempt to be friendly to beginners

16:48 TEttinger: I liked the rainbow parens option in lighttable, which isn't available in nightcode

16:48 martinklepsch: hiredman: using cider in emacs & showed all frames

16:48 amalloy: SagiCZ1: of course; paredit is an emacs mode

16:49 TEttinger: it's also the same in nightcode

16:49 SagiCZ1: amalloy: i know but its implemented into many other editors, such as nightcode or cursive

16:49 amalloy: if you're using some other implementation of paredit, the commands probably mostly exist but will have different bindings

16:49 hiredman: martinklepsch: but the message in the exception will tell you about the error from s3

16:49 SagiCZ1: amalloy: so they would be like keyboard shorcuts or something.. depending on the implementation

16:49 TimMc: SagiCZ1: Warning: Once you get used to paredit, it will be infuriating to try to write Clojure without it.

16:50 kind of a red pill sort of thing

16:50 SagiCZ1: TEttinger, rainbow parens are cool and all.. except it was quite a challenge to find enough colors i can safely distinguish... being colorblind and all... :D

16:50 martinklepsch: hiredman: this is everything I've got in the stacktrace buffer: https://gist.github.com/mklappstuhl/6ce2a1f4caef0a2f6cc5

16:50 SagiCZ1: TimMc: you use paredit in emacs too?

16:50 martinklepsch: did I oversee something?

16:50 TEttinger: I type my parens with a weird stroke, shift 9 0, let go of shift, left arrow

16:51 and it's such a habit now

16:51 amalloy: SagiCZ1: i don't really find rainbow parens very compelling. they make you focus on matching up the parens, when you really should be letting your editor deal with that and just pay attention to indentation

16:51 TEttinger: i just type 9

16:51 TEttinger: 9 or shift 9?

16:51 amalloy: (having remapped that key to produce the ( character)

16:51 TEttinger: oh ok

16:51 SagiCZ1: amalloy: well paren matching is a breeze with paredit right

16:52 amalloy: SagiCZ1: sure. which is why you don't need rainbow parens

16:52 SagiCZ1: got it

16:52 mgaare: rainbow parens do make your source look a bit less monotonous at least

16:52 TEttinger: amalloy, heh you should see some of my code that DID need some visual indicator (it was terrible)

16:52 SagiCZ1: as i am typing without paredit i have the feeling there must be a better way

16:52 hiredman: martinklepsch: you need to get the contents of the body of the error

16:53 martinklepsch: so *e in the repl will be the last uncaught exception

16:54 (doc ex-info)

16:54 clojurebot: "([msg map] [msg map cause]); Create an instance of ExceptionInfo, a RuntimeException subclass that carries a map of additional data."

16:54 hiredman: (-> *e ex-data :object :body slurp) or something like that

16:55 martinklepsch: hiredman: oh! that perfectly shows me an error

16:55 hiredman: like magic

17:01 martinklepsch: hiredman: hehe

17:07 If I have a long lazy sequence and want to split that into chunks of bytes to upload them to S3, could I do this? I assume calling .getBytes on a lazy seq will first fully realize it? Should I use something like partition-all and call .getBytes on this?

17:08 gfredericks: it's a lazy sequence of what?

17:09 martinklepsch: gfredericks: maps

17:10 justin_smith: so you want the pr-str representation of those maps?

17:10 or maybe json?

17:10 martinklepsch: justin_smith: pr-str

17:11 but that would best come after the splitting, right?

17:11 gfredericks: depends on what you're trying to split into

17:11 e.g., some max number of bytes?

17:12 martinklepsch: gfredericks: Oh, right. For multipart uploading I'd need to split at exact bytes so splitting with partition-all is not really an option I guess

17:13 justin_smith: seems like you want something that takes a lazy seq of strings and turns it into an OutputStream

17:13 once it's an OutputStream, the lib you are using can do the chunking part

17:14 I wonder if that function exists? if not you can likely make it via proxy or reify

17:15 gfredericks: are you sure an inputstream wouldn't work?

17:15 justin_smith: the right way to make it of course would be so that it does not consume more of the lazy-seq coming in than is needed for the number of bytes read

17:15 gfredericks: because there is ##java.io.SequenceInputStream

17:15 justin_smith: gfredericks: sorry, I mean inputstream

17:15 gfredericks: ,(def my-strs ["foo" "bar" "baz"])

17:15 clojurebot: #'sandbox/my-strs

17:16 gfredericks: ,(def is (->> my-strs (map #(java.io.ByteArrayInputStream (.getBytes %))) (java.io.SequenceInputStream.)))

17:16 clojurebot: #<CompilerException java.lang.ClassCastException: Cannot cast clojure.lang.LazySeq to java.util.Enumeration, compiling:(NO_SOURCE_FILE:0:0)>

17:16 gfredericks: ,(def is (->> my-strs (map #(java.io.ByteArrayInputStream (.getBytes %))) (clojure.lang.SeqEnumeration.) (java.io.SequenceInputStream.)))

17:16 clojurebot: #<CompilerException java.lang.reflect.InvocationTargetException, compiling:(NO_SOURCE_FILE:0:0)>

17:17 justin_smith: gfredericks: it takes a collection of inputstreams and returns their concatenation

17:17 so you probably need the string->stream conversion there too

17:17 gfredericks: I forgot a dot

17:17 justin_smith: (mapped)

17:17 gfredericks: ,(def is (->> my-strs (map #(java.io.ByteArrayInputStream. (.getBytes %))) (clojure.lang.SeqEnumeration.) (java.io.SequenceInputStream.)))

17:17 clojurebot: #'sandbox/is

17:17 gfredericks: ,(slurp is)

17:17 clojurebot: "foobarbaz"

17:17 justin_smith: there we go

17:17 (inc gfredericks)

17:17 lazybot: ⇒ 82

17:18 martinklepsch: and so this is lazy?

17:18 gfredericks: should be

17:19 because of map, SeqEnumeration, and SequenceInputStream all being lazy

17:19 let's make an infinite one

17:19 ,(def is (->> (range) (map str) (map #(java.io.ByteArrayInputStream. (.getBytes %))) (clojure.lang.SeqEnumeration.) (java.io.SequenceInputStream.)))

17:19 clojurebot: #'sandbox/is

17:20 gfredericks: ,(-> is (clojure.java.io/reader) ((fn [r] (apply str (map char (repeatedly 20 #(.read r)))))))

17:20 clojurebot: "01234567891011121314"

17:20 justin_smith: yup, that would have hung if it wasn't lazy, I'd reckon

17:21 gfredericks: "(apply str (map char (repeatedly 20 #(.read r))))" rhymes a bit if you read it out loud

17:21 Clarice: Only the char and r.

17:22 martinklepsch: thats awesome

17:22 thanks gfredericks & justin_smith, would have probably taken me ages to figure that out :)

17:22 gfredericks: Clarice: I'm looking forward to your better clojure poem

17:23 Clarice: gfredericks: See you at the poetry slam later tonight

17:24 justin_smith: http://en.wikipedia.org/wiki/Shakespeare_(programming_language) why not use this instead of clojure

17:25 the hello world is awesome http://en.wikipedia.org/wiki/Shakespeare_(programming_language)#Example_code

17:26 "Thou art as disgusting as the quotient between Romeo and twice the

17:26 difference between a mistletoe and an oozing infected blister! Speak

17:26 your mind!" <- that statement has arithmatic and IO in it

17:34 ahoenigmann: in emacs how do you get start a new clojure project?

17:34 I would like the path to get: cider repl + all files in project loaded + paredit mode started on all .clj buffers

17:35 justin_smith: (add-hook 'clojure-mode-hook 'paredit-mode) ; for the paredit part

17:36 for the rest, I typically just open one of the files in the project, and then run jack-in, then open others as I need them

17:37 afhammad: in core.asyc, when do you use a (put! c "hello") vs (go (!> c "hello")) ?

17:37 justin_smith: ahoenigmann: you can use speedbar, but you have to configure speedbar to pay attention to .clj files (it ignores them by default)

17:38 that's done by setting speedbar-show-unknown-files to a truthy value

17:39 ahoenigmann: speedbar looks cool kind of sublime like

17:39 justin_smith: it's much older, and a bit weird

17:39 but it works, if you want a file tree kind of thing

17:39 martinklepsch: one last question :) is there anything specific I need to be aware of when splitting up a SequenceInputStream or do I just use partition-all kinda like this to get 5mb chunks? (map #(apply str %) (partition-all (* 5 1024 1024) (repeat "xxx ")))

17:39 justin_smith: for some modes, it gives clickable targets for individual functions

17:40 I don't know if there is anything that intigrates that with clojure awareness so you could get hyperlinks to jump to def / defn though

17:40 martinklepsch: don't partition it, let the uploader partition the bytes itself, it will take as many bytes from your stream as it likes at a time

17:41 martinklepsch: justin_smith: in propS3t?

17:41 justin_smith: it should - or do you have to do that part manually?

17:42 clj-aws-s3 does the chunking, you just tell it how big the chunks would be

17:42 if you have to create the chunks yourself, definitely don't use partition, just read the amount of bytes you need for each chunk into a byte array or whatever the lib expects

17:42 martinklepsch: justin_smith: I think in propS3t you need to write parts manually

17:43 justin_smith: OK. so like I said, don't use partition.

17:44 martinklepsch: unless I am misreading the code, you should just pass the stream to write part

17:45 write-part will read as much of the stream as it needs

17:45 and streams are stateful

17:45 so just repeatedly call write-part until the stream has no more data left to give you

17:45 then call end-multipart

17:46 martinklepsch: justin_smith: write-part is done manually in the tests or am I misreading that? https://github.com/hiredman/propS3t/blob/master/test/propS3t/test/core.clj

17:47 justin_smith: yes, you need to call it manually, unless he has a higher level function that abstracts over the multi-part process

17:47 martinklepsch: ah, yeah, got it

17:47 justin_smith: this is all done auto-magically in clj-aws-s3 - you give it a stream and a chunk size, it does the rest

17:48 also, mind that each call to write-part returns a "part", and you need to keep those "parts" as a sequence to pass to end-multipart

17:48 though the test probably makes that clear

17:48 martinklepsch: justin_smith: yeah, saw that

17:48 maybe I'll play with clj-aws-s3 again as well

17:49 It's sort of surprising that there is no de-facto S3 lib out there

17:52 ahoenigmann: so i just added a new dependency in the project.clj

17:52 is there any way to pull it in from with emacs? or do I have to do lein ....

17:53 justin_smith: you can add new deps without restarting if you use alembic/still

17:53 otherwise, you need to restart your repl

17:53 ahoenigmann: what is that?

17:53 justin_smith: (or use pomegranate, but alembic is much easier)

17:53 it's a lib that adds deps at runtime

17:53 including reloading your project.clj to add new deps

17:53 ahoenigmann: whoa

17:54 justin_smith: or just finding and loading arbitrary whatever into your repl

17:54 ahoenigmann: the tooling is harder than the language x10

17:54 justin_smith: https://github.com/pallet/alembic

17:54 ahoenigmann: thanks

17:54 justin_smith: ahoenigmann: easier - how many other languages give you reliable per-project dep management?

17:54 ahoenigmann: hmm ok

17:54 justin_smith: and reliable deployment to an environment with exactly the deps you have locally

17:55 it's not hard, it is just different

17:55 but it's much better

17:55 ahoenigmann: does restarting/starting a cider repl jack in download dependencies?

17:55 justin_smith: sure, if you have new ones

17:55 but you can also use alembic, as I mentioned

17:57 I think clojure tooling is harder if you expect it to act like tooling does in other languages. But if you let it be what it is, in the long term you realize it's much better.

18:05 turbofail: i dunno i found the clojure tooling way simpler than all the other tooling i'd used previously

18:05 brand0: same, I really like it

18:06 justin_smith: ahoenigmann: it sounds like you are new to emacs? if so that may be the extra complexity

18:06 emacs is definitely weird

18:08 martinklepsch: +1 on finding clojure tooling good (esp. lein)

18:11 shanemhansen: emacs and cider-jack-in are a revelation.

18:12 brand0: now if oxcart would just hurry up ....

18:12 shanemhansen: But not necessarily the simplest thing to learn.

18:17 gfredericks: (inc justin_smith) ;; for shakespeare PL

18:17 lazybot: ⇒ 60

18:17 danielcompton: There was a time when I'd use alembic so infrequently that I had to look up the docs every time I needed to use it, but now it's a daily tool for me

18:18 Is there an emacs equivalent of https://github.com/blog/1877-folder-jumping ?

18:18 martinklepsch: shanemhansen: lighttable makes you not even need to think about that sort of stuff

18:19 danielcompton: martinklepsch: The last time I used light table with beginners it was easy to start them but every now and then they would run into a weird bug and the simplest solution was to just tell them to restart it

18:19 joshuafcole: Yeah

18:19 well, it's open sourced now

18:19 So I can track down the bugs when I run into them

18:20 justin_smith: danielcompton: in dired, you can hit the "i" key while the cursor is on a directory to expand it inline

18:20 joshuafcole: but it's still kind of wild west, versus the polish of time on the more venerable editors of yore

18:20 martinklepsch: I agree that you might want to try emacs+cider later though, but for beginners things are incredibly simple with lighttable. I mean even with ruby, python and other kind of beginner friendly languages it's not that simple

18:20 justin_smith: danielcompton: not as automatic and convenient, but reaches a similar end

18:20 martinklepsch: good night everyone!

18:20 joshuafcole: ciao

18:20 danielcompton: justin_smith: that's pretty cool

18:22 justin_smith: danielcompton: yeah, I had to figure out a project with an extremely hairy directory structure (where for each feature I had to change a template, some js, and some backend controller code, and each was under a totally separate path), and that feature was a total gamechanger

18:22 I was just lucky I didn't need to update the css too...

18:24 danielcompton: justin_smith: sounds painful

18:24 justin_smith: yeah, but being able to expand things inline in dired made it at least possible

18:25 also, of course, it was one of those file-heirarchy heavy projects where an overdone OO inheritance tree like thing is modeled using dirs and subdirs

18:25 waaaaay over-engineered

18:31 ThinkRusty87: If anyone needs help with anything let me know.

18:40 CaptainLex: I seem to be having an issue where Compojure is munging special characters

18:41 justin_smith: in the body?

18:42 or wait, you said compojure, so you mean routing strings are being munged?

18:42 because that would make sense, it should be a valid URL after all

18:43 CaptainLex: justin_smith: Maybe I'm misunderstanding what Compojure is doing. I mean special characters in the body

18:43 justin_smith: OK, that would be ring

18:43 CaptainLex: Oh okay thanks

18:43 So I guess

18:43 justin_smith: first off, check that your content types are right all around

18:43 CaptainLex: Ring is munging special characters :)

18:43 Okay, one moment

18:43 justin_smith: ie. are you setting the right character encoding on your requests

18:45 CaptainLex: Chrome says text/html

18:46 justin_smith: character encoding?

18:46 CaptainLex: (For the record, I've tried replacing the special characters with their HTML encondings to the same effect)

18:46 I'll check the template

18:46 one moment

18:47 The charset in the meta is utf-8, if that was what you meant

18:47 justin_smith: CaptainLex: oh, and just to be clear - is it the data going from server to browser that is corrupted, or coming back?

18:47 CaptainLex: yeah, sorry, I had the wrong term

18:48 CaptainLex: justin_smith: No worries. These are characters that are in raw HTML.

18:48 I've tested the various enlive functions so I know the characters are still valid after that stage

18:48 justin_smith: so is it getting corrupted in the server->browser direction or in the browser->server direction?

18:48 CaptainLex: server->browser

18:48 justin_smith: OK

18:48 you may want to check what the result looks like in curl

18:49 CaptainLex: Hmmmm

18:49 good idea!

18:49 justin_smith: to rule out weird behavior in the browser

18:49 and simplify verifying the actual data going through

18:49 CaptainLex: Well, actually

18:49 woops!

18:49 justin_smith: ?

18:49 jumblemuddle: Anyone around here worked with titanium? If so, do I need to get a titan server up and running, which it connect to, or does it do that itself? Also, do I need to choose which storage backend to use? Right now embeded-cassandra looks the best.

18:50 CaptainLex: I forgot that last time I debugged this issue (some months ago, and then I put thew project on hold)

18:50 I figured out my console didn't display those characters either

18:50 I could redirect curl though

18:50 I'll do that

18:50 justin_smith: or wget

18:50 see if it looks right in your $EDITOR

18:50 emacs has rest-client-mode which lets you specify a request and then see the reply below it

18:51 (it's a package you can install)

18:51 it could also be that you are telling the other end that it is utf8, but ring is creating some other content-type

18:52 CaptainLex: justin_smith: I guess it's a browser thing! I curled > test.txt and open that in Geany and it worked

18:52 Lemme test Firefox

18:52 justin_smith: what does Geany think the encoding of the file is?

18:52 or does it have an option to show you that

18:53 the linux "file" utility should also show you encoding

18:54 CaptainLex: I wasn't able to read that setting, but I was able to tell it to use UTF-8 and it didn't break it

18:54 Oooh I'll try that

18:54 danielszmulewicz: Am I killing my 1gig VPS with 5 JVM instances?

18:54 justin_smith: CaptainLex: I think I misremembered that actually

18:55 danielszmulewicz: It looks like starting one instance kills another one. Is that even possible?

18:55 CaptainLex: justin_smith: the output of running file actually did tell me an enconding

18:55 justin_smith: danielszmulewicz: I am sure it could handle 500 simultaneous helloworld.java instances

18:55 CaptainLex: ahh, I must have double checked on the wrong file here :)

18:55 CaptainLex: ISO-8859

18:56 justin_smith: so yeah, you need to tell your ring adaptor that you want to send utf8, or tell the client you are sending ISO-8859

18:56 danielszmulewicz: justin_smith: Good to know. I must look elsewhere then. I don't get it.

18:57 justin_smith: danielszmulewicz: that was sort of a joke - point being, what you are running matters more than whether it is jvm or not

18:57 CaptainLex: justin_smith: Thanks! I'll look into it

18:57 justin_smith: CaptainLex: unless you are serving mostly Chinese text or something, utf8 will likely be your best option

18:58 danielszmulewicz: justin_smith: Fair enough.

18:58 justin_smith: I mean what can you know without profiling all the apps?

18:59 Jaood: danielszmulewicz: five lein repls will definite kill it ;)

18:59 danielszmulewicz: justin_smith: I suck at devops. What tool should I use to measure memory on my VPS and get a picture of how much mem each instance uses?

18:59 justin_smith: danielszmulewicz: I like htop, for real time monitoring at least

18:59 danielszmulewicz: you can get snapshots using ps

18:59 danielszmulewicz: Jaood: Yeah. Lein is a beast. But I'm running the standalone jars.

19:00 amalloy: danielszmulewicz: if you start up 5 JVMs without doing any configuration of them, then yeah, that probably takes up more than 1g

19:00 danielszmulewicz: My instances are your typical ring web applications. Nothing fancy.

19:00 justin_smith: danielszmulewicz: also, since we are using uberjars, you can get really good profiling data locally using something like jvisualvm or yourkit, and it should mostly reflect the behavior online (modulo load)

19:00 amalloy: but you should tell the jvm not to use more than X amount of memory, which for a simple ring webapp can be just a couple dozen megs

19:01 danielszmulewicz: amalloy: Should I use a small maximum heap setting (XmX)?

19:01 amalloy: yes. -Xmx40m is reasonable for a simple webapp

19:01 i think that's what lazybot has. maybe he has 60m these days. either way it's more than he really needs

19:06 danielszmulewicz: (inc amalloy)

19:06 lazybot: ⇒ 160

19:06 danielszmulewicz: (inc justin_smith )

19:06 lazybot: ⇒ 3

19:07 justin_smith: that makes three times someone has added an extraneous space after my name in an inc

19:09 TEttinger: (inc justin_smith)

19:09 lazybot: ⇒ 61

19:12 cbp: (inc)

19:13 justin_smith: (inc ✈)

19:13 lazybot: ⇒ 1

19:17 technomancy: is clojure-doc.org's vim coverage good? http://clojure-doc.org/articles/tutorials/vim_fireplace.html

19:25 gfredericks: hey, what's your favourite joke library?=

19:26 besides lein-lein

19:26 (or anyone really)

19:27 mdeboard: Sorry but programming is a very serious matter

19:27 I do not download or even talk about "joke" libraries

19:27 amalloy: technomancy: https://github.com/mattdiamond/fuckitjs ?

19:28 technomancy: amalloy: sorry, has to be a clj lib that I can use from a trivial web app

19:28 oh, maybe that kebab snake one

19:28 amalloy: camel snake kebab

19:28 technomancy: mmmm delicious https://github.com/qerub/camel-snake-kebab

19:28 justin_smith: https://github.com/rodnaph/clj-php (more for the concept than the execution, which I am ignorant of)

19:28 joshuafcole: There isn't a fuckit.cljs yet?

19:28 If not we should get on that.

19:28 mdeboard: I started to but I was like

19:28 "fuck it"

19:28 joshuafcole: :)

19:28 justin_smith: clearly fuckit.hs would be the best

19:29 joshuafcole: Honestly, you should probably just bootstrap it

19:29 jumblemuddle: Let me know when you've got it finished. I'd be happy to use it... :)

19:29 joshuafcole: get it close enough to working and then run it on itself until it's "good"

19:29 justin_smith: a "guess what I mean and suppress errors" haskell extension, who would not love that

19:29 joshuafcole: "guess what I mean"? Whoah, fuckit.xyz doesn't /care/ what you mean. It just wants the compiler/runtime checker to shut up.

19:30 justin_smith: heh

19:30 amalloy: justin_smith: i dunno, i think it would go the opposite direction. fuckitjs works because that's the javascript mindset. a haskell version would take the opposite approach: "this code looks a little weird. maybe it would compile if i tried, but...fuck it, let's throw a type error"

19:31 justin_smith: file doesn't end with a newline? type error

19:31 joshuafcole: haha

19:31 justin_smith: mixed tabs and spaces? type error

19:31 amalloy: exactly

19:31 justin_smith: I love it

19:31 bridgethillyer: technomancy: on clojure-doc.org’s vim coverage: I just reviewed it quickly

19:31 joshuafcole: Don't stop there!

19:31 terribly cliche variable names?

19:31 bridgethillyer: It’s not perfect, but it’s generally accurate based on my quick read

19:32 joshuafcole: Shadowing of any variable names?

19:32 :P

19:32 justin_smith: joshuafcole: variables with more than one letter in the name: type error

19:32 technomancy: bridgethillyer: thanks!

19:32 joshuafcole: Though, now it's getting complex. Maybe just throw a type error on anything that's not "Hello World" to be safe.

19:33 justin_smith: bbl

19:33 joshuafcole: ciao

19:33 bridgethillyer: sure thing. Now that I have done that, I’d like to go back and review it for current accuracy. Thanks for the inspiration

19:37 aperiodic: technomancy: re: joke libraries: https://github.com/fxn/i-told-you-it-was-private

19:40 xeqi: harsh

19:41 joshuafcole: It might seem harsh, but down the road of projects like this and fuckit lies the zen of 0 sloc

19:43 technomancy: aperiodic: <3

19:50 gfredericks: technomancy: is system-slash-exit a joke?

19:50 I also have https://github.com/gfredericks/flexnum but it's rubby

19:50 kind of surprised that one doesn't have a readme

19:51 needs an example of truncating your database table via MyModel.all.length.zero!

19:58 danielcompton: Does Spring qualify as a joke library?

20:00 mdeboard: zing

20:00 ice cold, sick burn

20:07 talios: danielcompton - surely you mean scala

20:07 danielcompton: Is that a gas torch from a flame war starting up that I hear?

20:17 technomancy: have things like esk-cleanup-buffer been moved to another emacs package now that emacs-starter-kit has been sort of deprecated?

20:25 amalloy: esk sounds like some kind of fantastical forest creature

20:27 blimps: If I evaluate "(and true true)" I get "true", but if try "((fn [& x] (and x)) true true)" I get "(true true)". What's happening here?

20:28 Bronsa: ,((fn [& x] x) 1 2 3)

20:28 clojurebot: (1 2 3)

20:28 Bronsa: ,((fn [a b & x] x) 1 2 3)

20:28 clojurebot: (3)

20:28 Bronsa: ,((fn [a b & x] [a b x]) 1 2 3 4)

20:28 clojurebot: [1 2 (3 4)]

20:30 danielcompton: while not strictly a joke library, https://github.com/ztellman/proteus has a funny readme

20:31 * technomancy opens an issue that just says "You monster."

20:33 Bronsa: "That's it. That's the end of the library." Just imagine if there was more to it

20:35 tuft: any idea how this could possibly come up? java.lang.IllegalArgumentException: Don't know how to create ISeq from: clojure.core$vector

20:35 ,(seq [1 2 3])

20:35 clojurebot: (1 2 3)

20:35 justin_smith: tuft: seq knows how, the jvm doesn't I guess?

20:35 aperiodic: tuft: I think it's saying it doesn't know how to make a seq from that function

20:35 amalloy: ,(seq vector)

20:35 clojurebot: #<IllegalArgumentException java.lang.IllegalArgumentException: Don't know how to create ISeq from: clojure.core$vector>

20:35 justin_smith: ahh

20:35 tuft: ahh i see, thanks

20:36 blimps: Ok, I think I see what you're saying. How could I get a function to apply "and" to an argument list resulting from "&"?

20:37 tuft: just needed a ->> instead of -> =)

20:37 AimHere: blimps, sadly, 'and' and 'or' are macros, so you can't.

20:38 amalloy: AimHere: well, it's not exactly "sadly". the point is you're supposed to use a function that operates on a list, such as `every?` or `some`

20:38 blimps: Yeah, I just read about that. Is there some sort of workaround for this sort of situation?

20:39 AimHere: You could write some sort of function that does the same job - even a mere wrapper around 'and' would do

20:40 justin_smith: amalloy: is there an alternative to every? that returns the last arg like and would?

20:40 of course that would be easy to write...

20:40 amalloy: AimHere: that function already exists. it is every?. you do not want to wrap and

20:41 gfredericks: ,(every? inc [1 2 3])

20:41 clojurebot: true

20:41 gfredericks: er nm

20:41 justin_smith: ,(and 1 2 3)

20:41 clojurebot: 3

20:41 gfredericks: I thought he meant like some

20:41 justin_smith: ,(every? identity [1 2 3])

20:41 clojurebot: true

20:41 brehaut: ,(and 4)

20:41 clojurebot: 4

20:42 amalloy: justin_smith: i mean, i guess you could do it. there's not much point; i don't imagine you want to get the last answer very often, when you've got a list of arbitrary size

20:42 brehaut: ,(+ 4)

20:42 clojurebot: 4

20:42 amalloy: ,(partial 4)

20:42 clojurebot: 4

20:42 brehaut: hmm i thought suffusion of yellow occured with results greater than 3?

20:42 ,(+ 1 2)

20:42 clojurebot: 3

20:42 amalloy: (+ 2 2)

20:42 clojurebot: 4

20:42 brehaut: ,(+ 1 2 3)

20:42 clojurebot: 6

20:42 justin_smith: amalloy: I often use and in cases where I only want to use the last value if the preceding are all truthy

20:42 brehaut: now its just being devious

20:42 amalloy: brehaut: he doesn't mess with results of actual eval requests starting with ,

20:43 brehaut: (+ 1 2 3)

20:43 clojurebot: *suffusion of yellow*

20:43 amalloy: he has/had some joke responses if you tried to use him for math without starting with ,

20:43 brehaut: amalloy: aha

20:43 amalloy: for some reason i thought it was a both situation

20:43 gfredericks: justin_smith: do you have an example use case with a non-static set of things?

20:43 amalloy: justin_smith: sure, and that's very common. but that was a fixed set of tests/values

20:43 with a list that you get from somewhere else? it's not at all common

20:44 justin_smith: right, with varargs thats a bit weirder, yeah

20:44 gfredericks: clojurebot: amalloy |says| your use case is contrived.

20:44 clojurebot: You don't have to tell me twice.

20:44 amalloy: (inc gfredericks) ; very good

20:44 lazybot: ⇒ 83

20:45 amalloy: i look forward to the roulette of ~amalloy generating things i don't remember saying

20:45 ~amalloy

20:45 clojurebot: amalloy was right again in a blazing show of total predictability

20:45 Bronsa: hah

20:45 joshuafcole_: bahaha

20:45 justin_smith: ~amalloy

20:45 clojurebot: amalloy has an (obsessive) posse

20:46 amalloy: huh, i don't even remember that one

20:46 gfredericks: your posse does

20:46 * gfredericks tries to figure out how to make a joke out of the phrase "the empty posse"

20:47 amalloy: lazybot doesn't have it anywhere in his logs. i wonder if someone taught clojurebot that via pm

20:47 gfredericks: it feels familiar to me; maybe only because it's randomly come up before

20:47 amalloy: oh, no, i see

20:47 i forgot it would look like |has| in the logs

20:47 justin_smith: gfredericks: the intersection of posse commitatus and the insane clown posse is the empty posse

20:48 amalloy: llasram taught him in june

20:48 http://logs.lazybot.org/irc.freenode.net/%23clojure/2014-06-20.txt

20:49 gfredericks was apparently the posse member in question

20:49 blimps: Ok, I was about to do what I needed with some and every? Thanks. :)

20:50 able to, that is

20:50 gfredericks: that was back in the day when amalloy would sometimes use the same string of words twice in one day just to see what would happen

20:50 * amalloy just forgot the order of args to `contains?`. how did *that* happen

20:51 xeqi: must be an imposter

20:52 Bronsa: I can never remember teh order of args to isa?/instance?

20:52 the*

20:53 * gfredericks is having a hell of a time figuring out how to spell "gord" using google & wikipedia

20:53 justin_smith: gourd?

20:53 PigDude: Bronsa: the solution is clearly `infix`

20:53 aperiodic: ~gourds

20:53 justin_smith: as in pumpkins and such?

20:53 clojurebot: SQUEEZE HIM!

20:53 gfredericks: aperiodic: PHEW

20:54 that was exactly what I wanted to do and now it is done.

20:54 aperiodic: I got your back

20:54 gfredericks: (inc aperiodic) ;; back possession

20:54 lazybot: ⇒ 8

20:54 amalloy: reminds me of the classic gourdian knot

20:55 Bronsa: ,(isa? String Object)

20:55 clojurebot: true

20:55 Bronsa: ,(instance? Object String)

20:55 clojurebot: true

20:55 Bronsa: this doesn't help

20:56 amalloy: man, http://en.wikipedia.org/wiki/Gordian_Knot is an amazingly bad description of the myth

20:57 i guess the words are all there, it's just weirdly abrupt

23:24 mdeboard: with nREPL, am I understanding correct that the "cider-jack-in" buffer is sending forms to be evaluated out to the socket over which it is communicating with the nREPL server?

23:24 and then the server is doing the computation, and responding with the result?

23:25 Jaood: yes, its a server-client model

23:25 cider is an nrepl client just like the lein repl is too

23:25 mdeboard: Yeah, I just put 2 and 2 together about what exactly the traffic is between the two. In hindsight it's obvious

23:27 Oh wow Github just pushed a really nice UI/UX improvement live

23:27 lets you skip over empty directories

23:28 https://github.com/clojure/tools.nrepl/tree/master/src/main

23:28 Clarice: Define "just"

23:28 mdeboard: idk, in the past 24 hours

23:28 Clarice: Oh

23:28 Yes, I see what you're saying

23:28 mdeboard: maybe sooner

23:28 Clarice: That's niiiiiifty

23:29 mdeboard: yarly

23:46 Clarice: How are clojure vectors constant time if they're represented by tries?

23:48 mdeboard: Clarice: What do you mean by "how are they constant time"? What operations do you mean?

23:48 Clarice: Getting the value at an index.

23:48 mdeboard: (I do not know right off the top of my head which ops are O(1))

23:48 Clarice: I imagine that becomes a pointer chasing exercise for values near the tail of the vector

23:49 danielcompton: What's the difference between comp and -> ?

23:49 Clarice: danielcompton: One is the composition of two or more functions.

23:50 danielcompton: which returns a new function. The other is a macro which operates on forms and pipelines the results.

23:50 danielcompton: Clarice: where could you use one but not the other?

23:51 Clarice: danielcompton: I encourage you to macroexpand -> and see what you get

23:51 mdeboard: Clarice: I don't kow the answer to your question but I'd guess it has some kind of hashmap like {idx: val} no?

23:52 Clarice: mdeboard: That defeats the purpose of structure sharing.

23:53 danielcompton: (-> x (form) (form) (form)) is essentially the same as (apply (compose f f f f) x). But they're slightly different.

23:53 danielcompton: Clarice: hmm, think I'm getting it

23:54 Clarice: danielcompton: You'd use compose if you want to talk about "the function you get when you compose these other functions". You want to use -> when you want to describe a pipeline process.

23:54 mdeboard: wouldn't the hashmap of positions also be updated when the vector is "updated"

23:54 Clarice: danielcompton: Have you used F#? There's a |> operator, which basically means "take whatever's on the left and shove it as the input of the function on the right." -> does the same thing, but as a macro and by rearranging the list structure of your code.

23:54 mdeboard: I'm talking out of my ass for the most part.

23:55 s/ for the most part.//

23:55 danielcompton: Clarice: never used F# but it makes sense

23:55 rei: Clarice: I believe it's just due to really shallow tries, but this post may explain it better. http://hypirion.com/musings/understanding-persistent-vector-pt-1

23:56 Clarice: rei: Oh! I remember bookmarking that and never getting around to reading it. Thank you for reminding me it exists.

Logging service provided by n01se.net