#clojure log - Sep 25 2015

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

2:16 tianyu: this is test

2:17 Trouble of eval clojure block within org-mode by cider :

2:17 org-babel-execute:clojure: Wrong number of arguments:

2:17 ((cider-buffer-ns cl-struct-nrepl-response-queue-tags t) (input

2:17 connection session &optional ns) "Send the INPUT to the nREPL server

2:17 synchronously.The request is dispatched via CONNECTION and SESSION.

2:17 The code is :

2:17 #+BEGIN_SRC clojure

2:17 (+ 1 2)

2:17 clojurebot: 3

2:17 tianyu: #+END_SRC

2:17 Plz help.

5:32 anti-freeze: Hi everyone. I need some help with ui.router. For some reason, every time I refresh the browser, ui.router redirects me to main.home, which is a route on top of the main abstract route. If I remove that route, everything works as normal. Here's the source code: http://plnkr.co/edit/WEvXWkpPolm7fMN54PxS?p=catalogue . Any ideas?

5:34 Wrong channel

5:34 Sorr

5:34 Sorry

8:10 wasamasa: now I know why it took so long to port pprint to cljs

8:11 it contains an implementation of CL's format

10:15 Guest24: good morning folks

10:31 triss: hey all, is there a reason the definition of clojure.core/fn is such a dense blob of code?

10:32 for example would pulling out the :pre and :post related stuff to some private functions/macros?

10:35 TimMc: I don't think there's a *good* reason, if that's what you're asking.

10:35 Some parts of core are a crawling horror because they haven't been touched since Clojure was still in very early development, and I think the Core team is reluctant to clean up code that is working.

10:36 There's something to be said for that.

10:41 sdegutis: When you use Components for your web app, do you even wrap your routes within a component?

10:43 kungi: sdegutis: I do. But you don't have to.

10:44 sdegutis: I'm having a hard time figuring out which things should be Components and which should just continue to be normal Clojure functions which Components call.

10:44 kungi: sdegutis: It depends on how you would like to inject your dependencies into your routes

10:46 sdegutis: Oh good point.

10:49 kungi: so the only practical difference is whether the functions can implicitly access record fields vs. having parameters to receive them?

10:49 There's no difference in terms of code-reloading?

10:53 kungi: sdegutis: As far as I see. Yes

10:54 sdegutis: You could also inject the Components into the http request through a middleware. This way you don't have to use function parameters.

10:56 craftybones: Are sets ordered by default in clojure? If so, is there a canonical way of finding the index of a member?

10:56 kungi: craftybones: no they are not ordered

10:56 craftybones: Thanks @kungi

10:56 bja: I have a component that takes an argument of things that will exist at start time on the ring-app component to be injected into the request

10:56 kungi: craftybones: there is sorted-set for that

10:57 craftybones: @kungi I don't want it sorted. I guess I'll just use a vector

10:57 xeqi: sdegutis: I usually just wrap the server, db, etc. Then I have the server build a ring handler passing in the components needed, but the ring handler/routing is just functions

10:57 bja: something like that: https://gist.github.com/b45d2dbf2fbf5ea8afe1

10:58 actually need to finish the blog post that code belongs to, it's a demo of using component to manage hot settings pushes from zookeeper

11:11 sdegutis: xeqi, kungi: thanks, very informative and helpful

11:12 I wonder what the downsides are of making *everything* be a Component, no random functions floating around.

11:12 I guess the main downside would be that it can't easily just say "okay, reload this function definition" while developing live (e.g. via CIDER), instead you have to shut down and restart the entire System component.

11:16 xeqi: sdegutis: I imagine the boiler plate for getting the setup/teardown for tests could get annoying

11:18 though if you pushed it far enough you'd basically have immutable namespace

11:18 sdegutis: xeqi: actually that would be the easiest part: there would be a (test-system) constructor function, which creates real versions of everything, except if there are any services which leave the system (e.g. email-service-component) then it would use an in-memory version.

11:19 xeqi: and each in-memory component would use atoms for state so you could easily setup state and test their side-effects in tests

11:20 E.g. my email-service takes an atom of emails, which I have a helper-macro that resets it to an atom that it lets me name and inspect after my test runs.

11:26 xeqi: sounds like what i do, though the recent cognicast has some mention of use cases for async channels instead of atoms

11:26 was overthinking the amount of isolation needed between tests when i mentioned the boilerplate

11:27 sdegutis: To this day I still don't understand when I need core.async.

11:30 bja: sdegutis that's only actually a downside of the current implementation of component (needing to restart everything)

11:30 sdegutis: bja: well it's not even really that much of a downside :)

11:30 bja: I've seen multiple private forks that take a new and current system and diff+ walk the system and only restart changed things

11:30 yeah, it's not too bad

11:32 I've actually seen people (other than myself) using com.stuartsierra.component/Lifecycle to build components but starting/managing them with heavy forks of the original codebase

11:33 sdegutis: bja: hmm I wonder why they haven't been merged back in?

11:34 bja: all of the ones except mine are private that I'm aware of

11:35 and mine isn't that heavy (it just uses destructure to process dependency injection maps)

11:35 there's a PR on component where stuart rejected mine, and I don't necessarily disagree with his reasons, but I find my fork useful

11:41 justin_smith: bja: I have a cljc fork

11:42 it's not "heavy" though, just cljc

11:42 one of these days I'll port all the tests to cljs.test and make a PR

11:43 bja: by heavy, I mean that almost none of the implementation details for starting/stopping are the same

11:43 I think my fork is something like a 40 line diff where I add a couple functions

11:43 justin_smith: bja: but if I am using component to provide lexically closed resources with a lifespan, all components using a changed component need restarting

11:43 (reading scrollback now)

11:48 tianyu: An error of eval clojure within org-mode: org-babel-execute:clojure: Wrong number of arguments: ((cider-buffer-ns cl-struct-nrepl-response-queue-tags t) (input connection session &optional ns) "Send the INPUT to the nREPL server synchronously.

11:48 The request is dispatched via CONNECTION and SESSION.

11:48 If NS is non-nil, include it in the request." (nrepl-send-sync-request (nrepl--eval-request input session ns) connection)), 1

11:48 PLZ help

11:49 justin_smith: tianyu: cider changes constantly, is org-babel even supposed to be working these days?

11:52 or, maybe a better way to put it - which version of cider is your version of org-babel's clojure support compatible with, and are you using that version?

11:54 sdegutis: Aw, this PR could have been useful https://github.com/stuartsierra/component/pull/25

11:55 justin_smith: yeah, I just made my own stupid helper that does that

11:55 (as something outside component proper)

11:56 sdegutis: Are you supposed to get System components via keyword-access?

11:57 justin_smith: yeah, the system passed into your start or stop method has all your dependencies associated on as keywords

11:57 or, with your selected keywords as keys

11:59 sdegutis: Right.

12:00 So it's idiomatic to destructure them, like (let [{:keys [db router time]} sys] ...) and use them like (get-today time) and (handle-req router req) ?

12:02 justin_smith: yeah, that seems straightforward, though it's usually nested deeper because I make components that each provide their own map of things

12:02 maybe you are doing it more modular so each component only provides one thing?

12:03 sdegutis: justin_smith: I'm not sure what you mean.

12:03 justin_smith: I have a flat list of components, but each component may depend on other components (no circular dependency though).

12:04 justin_smith: sdegutis: my components each provide a map with multiple things you might want from it

12:04 sdegutis: justin_smith: so I have [db router time-service email-service ...] but db may depend on time-service

12:04 justin_smith: wait, where is this destructuring happening again?

12:04 sdegutis: justin_smith: oh you mean each component has getters?

12:04 justin_smith: oh good point... I meant at the top-level entry point

12:05 justin_smith: each component associates multiple keys, and a component using it has to pick out the keys it wants

12:05 sdegutis: justin_smith: within a component, it just has the depended-upon components as record fields which are always visible

12:05 justin_smith: OK, what does destructuring at the top level entry point do for you?

12:05 sdegutis: justin_smith: oh good point.. I guess I don't need it

12:05 Okay, solved!

12:05 Thanks justin_smith.

12:05 justin_smith: tbh I'm mapping this out, I don't actually have code yet just a diagram

12:06 justin_smith: it was a legitimate question, but I guess "nothing" is an answer too

12:06 haha

12:06 sdegutis: I'm preparing for transforming my services into components (services were my ad-hoc solution I've been using before component was written)

12:15 elika: hello

12:16 blkcat: x1/win 32

12:16 ...wow

12:16 sdegutis: blkcat: what?

12:16 blkcat: don't mind me :D

12:16 scriptor: anyone know where LEIN_PASSWORD is coming from in https://github.com/technomancy/leiningen/blob/master/sample.project.clj#L103 ?

12:17 sdegutis: blkcat: plz do tell

12:17 scriptor: its an env variable

12:17 scriptor: yes

12:17 that much is clear

12:17 sdegutis: scriptor: you're supposed to set it

12:17 scriptor: that's how they work

12:18 scriptor: you misunderstand, I'm trying to figure out how leiningen is able to get LEIN_PASSWORD in the first place

12:18 I know how to set a bash env variable

12:18 but how does lein know that's the env variable?

12:18 sdegutis: scriptor: according to the comments above it, when it sees :env, it asks for the environment variable

12:19 scriptor: probably via (System/getEnvironmentVariable "LEIN_PASSWORD") or whatever the static function name is

12:19 justin_smith: sdegutis: System/getenv

12:19 sdegutis: yeah that

12:19 justin_smith: ,(System/getenv "PWD")

12:19 scriptor: but how does lein know to ask for LEIN_PASSWORD, there's no other reference to that anywhere els

12:19 clojurebot: #error {\n :cause "access denied (\"java.lang.RuntimePermission\" \"getenv.PWD\")"\n :via\n [{:type java.security.AccessControlException\n :message "access denied (\"java.lang.RuntimePermission\" \"getenv.PWD\")"\n :at [java.security.AccessControlContext checkPermission "AccessControlContext.java" 372]}]\n :trace\n [[java.security.AccessControlContext checkPermission "AccessControlContext.java...

12:19 justin_smith: :P

12:19 sdegutis: I get the naming conventions between that and System/getProperty mixed up

12:19 scriptor: it's hard-coded in

12:20 scriptor: I'm really having trouble understanding a reason for your confusion on this

12:20 scriptor: ah, so LEIN_PASSWORD is a hardcoded variable that lein looks for

12:20 sdegutis: scriptor: it's a reasonable convention that Leiningen is using, where it gives you a name of an env variable which you can then set if you choose, and gives you a way to tell it to look for it

12:21 scriptor: the comment says that an "environment variable will be used based on the key"

12:21 since the key is :password

12:22 justin_smith: scriptor: hmm, maybe it is constructing the variable to look up based on the key?

12:22 scriptor: I was wondering if lein simply uses that key and prepends 'LEIN_' to it to create the actual environment variable name

12:22 sdegutis: scriptor: right, it's saying: if the value is a string, it uses that literally; if it's literally :env, it looks in the environment variable "LEIN_PASSWORD" for a literal string to use

12:22 scriptor: maybe

12:23 sdegutis: the value is :env, but the key is :password, are you saying that somewhere inside lein the :password key is hardcoded to map to LEIN_PASSWORD?

12:24 justin_smith: scriptor: I think it is more likely that :foo is hardcoded to be LEIN_FOO (somewhere) but I am not finding evidence of any of this searching the lein github repo for getenv usage

12:24 sdegutis: justin_smith, scriptor: it's probably literally this code: (let [m {:password "foo"} {:keys [password]} m password (if (= password :env) (System/getenv "LEIN_PASSWORD") password)] ...)

12:24 Ore more clearly: (let [{:keys [password]} {:password "foo"}, password (if (= password :env) (System/getenv "LEIN_PASSWORD") password)] ...)

12:24 scriptor: so, you're not certain?

12:24 justin_smith: sdegutis: the weird thing is LEIN_PASSWORD only shows up in the readme, nowhere in code

12:24 scriptor: ^

12:25 sdegutis: scriptor: I'm telling you that's what the documentation claims; I haven't looked at the code so I don't know, but that's what the docs are saying

12:25 justin_smith: ooooh

12:26 scriptor: ah so yeah it's probably a convention like justin_smith says, where it probably has this: (defn get-value [m k] (or (get m k) (System/getenv (format "LEIN_%s" (name k)))))

12:26 In which case it's doing (get-value your-map :password)

12:27 I bet a grep for "LEIN_" would turn that function up.

12:29 scriptor: https://github.com/technomancy/leiningen/blob/1bc9d42c430b0c94c0a25e8f869c94bb41d9b2cf/leiningen-core/src/leiningen/core/user.clj#L144

12:29 sdegutis: there you go

12:30 scriptor: so my assumption was right, it takes the key and simply prepends LEIN_ to it

12:30 sdegutis: you got it bud

12:31 justin_smith: sdegutis: you forgot the part where it converts to SHOUTY_CASE

12:31 but yeah

12:31 sdegutis: touché

12:31 * justin_smith is off to implement string/SHOUTY_CASE as a function.

12:31 sdegutis: wha? why?

12:32 xemdetia: much like strfry

12:32 justin_smith: because it's silly of course

12:32 xemdetia: this is a vital part of libraries

12:32 justin_smith: xemdetia: exactly

12:37 TimMc: Wow, that's gross.

12:37 I don't like it when variable names are constructed, it makes them hard to grep for. :-)

12:38 sdegutis: TimMc: meh it has a time and place

12:38 TimMc: now /function names/ on the other hand...

13:14 justin_smith: sdegutis: TimMc: worst is when entire namespaces are made that way.

13:15 sdegutis: justin_smith: never heard of that

13:15 justin_smith: there's one amazon api wrapping project in particular, turns each class in the api into a namespace, each static method gets a function wrapper

13:16 sdegutis: so if you go into the code looking for the code for the thing you are using, you find nothing, just the general odor of magic in the air

13:17 sdegutis: this macro, on line 5, creates a whole namespace full of vars and functions https://github.com/mcohen01/amazonica/blob/master/src/amazonica/aws/cloudformation.clj

13:18 s/macro/function - here is the function with its doc https://github.com/mcohen01/amazonica/blob/master/src/amazonica/core.clj#L884

13:19 sdegutis: so someone is like "how do I use cloudformation/foo" and I go look at the source, and I'm like "wat"

13:21 jstew: Offer to rwrite it :)

13:21 justin_smith: haha

13:27 bja: amazonica is the one that manages client objects in some var (global? thread-bound?) for you right?

13:28 justin_smith: bja: sounds about right - it puts all kinds of things into vars

13:29 bja: if this is the one I'm thinking of, it was fine for devops scripts but made it worthless to my service since I was putting everything in components

13:29 jstew: vars are like violence? :)

13:29 justin_smith: jstew: vars are global, which can be inflexible

13:30 bja: I mean, it's fine if an api just handles managing the client for you, as long as there's some way to break out and pass in a particular client instead of the global thing

13:31 and by some way, I don't mean reflecting/binding private stuff

13:31 jstew: Is there any value to it over just calling out to java, since that's what it's doing for you, albeit in a messy way?

13:31 bja: it has a few higher level abstractions so you don't have to paginate certain things yourself

13:31 justin_smith: bja: it was a big revelation to me that when I was working on a lib (that could potentially use globals vs. dynamic vars vs. explicit value passing) that every other method of sharing a value can be derived from explicitly passing a value to a consumer. But that's the only flexible one, the others are not inter-compatible.

13:32 bja: justin_smith: that's an excellent way of putting it

13:32 justin_smith: bja: so by using explicit passing of an arg, you leave the most flexibility for the developer making their own app

13:33 bja: I did that for clj-aws-s3 here: https://github.com/curiosity/clj-aws-s3

13:33 I think there are 3 or 4 different forks that do essentially the same thing w.r.t to removing the implicit dynamic vars

13:34 justin_smith: they can use globals or dynamic vars or atoms or deserialized objects from a db or whatever the hell they want, and it will all be compatible with explicit arg passing in the end

13:34 bja: hah, makes sense

13:35 bja: actually, looking at that, I probably need to type-hint all those client arguments

13:36 I should profile that next week

13:37 oh, forgot, clj-aws-s3 wasn't even a dynamic var, it was just a private global

13:37 that was memoized so you couldn't even call it again

13:37 justin_smith: haha, nice

13:37 bja: but it had nice abstractions for s3

13:37 so it was worth forking and maintaining

14:37 preyalone: Which plugin { id ... } ... apply plugin: ... should I drop into build.gradle for Clojure projects?

14:41 RedNifre: good evening

14:42 snowell: preyalone: I don't know that there would be a build.gradle. Generally we use leiningen to build clojure projects, so you'd have project.clj instead

14:44 RedNifre: So I started with Clojure three days ago and the language looks fine to me. Now I wonder about the practical side: Is Cursive any good, how would it fit with Android Studio, is there some clojure library repository somewhere (or are clojure libraries just jars that you add to your gradle file?) how does it all fit together?

14:45 ToxicFrog: RedNifre: I can answer the clojure library repository part, at least

14:45 RedNifre: great :)

14:46 ToxicFrog: The de facto standard library repository is Clojars; if you're using Leiningen, the Clojure build tool, it knows how to download libraries from there automatically if you list them in the deps

14:46 justin_smith: we use leiningen instead of gradle - or at least 99% of us do

14:46 ToxicFrog: It also has support for fetching libraries from github and I think a few other sources

14:46 I have no idea what gradle is

14:46 justin_smith: ToxicFrog: it's like leiningen for groovy or something

14:47 preyalone: snowell: Aye, Leiningen is an optino. But I'd prefer to use one build manager for all my altJVM projects, rather than a medley of Maven + Rake + Leiningen + Grapes

14:47 RedNifre: I heard of leiningen, but have no idea what it is yet. Does it only fetch clojure libraries or can I also fetch any other jar, e.g. from maven central?

14:47 justin_smith: RedNifre: clojure libraries are hosted in any jar repo, there are some on maven central too

14:47 RedNifre: gradle is just a build tool. you have a gradle file like you have a pom file or an ant file I guess.

14:48 justin_smith: RedNifre: leiningen uses the maven spec (but not maven itself0

14:48 right, leiningen is a build tool also

14:48 RedNifre: okay, so I would write some clojure code, add dependencies with leiningen, both clojure jars and regular jars and in the end I compile it into one jar that will just work like a java jar?

14:48 justin_smith: exactly

14:48 ToxicFrog: That said, clojure libraries are either jar files, or clojure source you can turn into jar files, so if you're using some non-lein build tool that consumes jars you can just feed them to it without issue.

14:48 If you're using lein it also has an "uberjar" command that combines the jars for all your libraries with the program/library you're building so that it's self-contained, which is handy for self-contained deployment.

14:50 schmudde: Yep. Leiningen user here. I'm consistently surprised at its power. Right now I need a little ClojureScript functionality as well... and I'm running that compile along with the a hosted Clojure web app.

14:50 RedNifre: ah, I think that was called "fat jar" in eclipse.

14:50 schmudde: Works great with Clojars.

14:50 snowell: RedNifre: I think you're right. Uberjar is a much cooler name

14:50 ToxicFrog: justin_smith: I don't know what groovy is either soooooo

14:50 justin_smith: ToxicFrog: yet another jvm language

14:51 ToxicFrog: so gradle uses groovy to build things in the way that leiningen uses clojure to build things

14:52 RedNifre: It sounds like building a simple jar that can run on a server is easy. Though I still wonder how to make use of clojure on android, either by having an app that is all clojure or by having the bulk of your code in a clojure jar but handle all the android specifics in java...

14:53 With gradle, your build file is a groovy program. If leiningen build files are written in edn then it should be the same concept.

14:53 ToxicFrog: aah

14:53 justin_smith: RedNifre: I'd look at one of the existing android clojure projects. Interop from clojure to java is very easy for the most part (unless you need to inherit from a specific class, that can be annoying from clojure sometimes)

14:53 RedNifre: they are!

14:53 RedNifre: I don't know Groovy the language, but the syntax is fine so I like gradle files :)

14:53 justin_smith: RedNifre: lein inputs being edn / clojure code, that is

14:55 bja: /join #pixie-lang

14:58 ToxicFrog: RedNifre: as far as android goes, I've only dabbled in it very briefly, but since you can call java directly from clojure very easily, there was no need for java compatibility layer or anything

14:58 The drawback is that the resulting app takes forever to load as it dynamically loads/initializes the clojure runtime and stdlib

14:58 RedNifre: oh

14:59 ToxicFrog: There is some work on improving that, not sure how far along it is -- this was back in late 2013, IIRC

14:59 xemdetia: yeah that was what I was mentioning yesterday

14:59 RedNifre: I remember trying Ruboto (JRuby for Android), but the app took 10 seconds to start back then (I think it's down to 3 seconds now, but still...).

14:59 xemdetia: part of the android system is that there is one JVM that never gets unloaded

14:59 ToxicFrog: If you can get away with running as a Chrome App or similar, you can use cljs instead of cljvm

14:59 RedNifre: hm.

14:59 ToxicFrog: Which will probably load a lot faster, although I haven't tried this outside of the test environment

15:00 RedNifre: so how long is "takes forever to load"? 1 second? 10 seconds?

15:02 ToxicFrog: RedNifre: on a Galaxy Nexus, two years ago -- ~15 seconds.

15:02 For hello world.

15:02 But like I said, there has been a lot of active work since on improving this.

15:03 E.g. some quick googling finds http://blog.ndk.io/2015/04/23/state-of-coa.html and http://clojure-android.info/skummet/

15:03 bja: There is a lein plugin that uses a forked compiler and some tooling that does some additional optimization to make clojure on android slightly better

15:04 ToxicFrog found it

15:04 I think the plugin is `lein-droid` or something like that

15:05 RedNifre: yeah, I'm reading about those.

15:08 bja: I wonder if React Native boots fast enough to make it worth using cljs and something like Reagent/Om for an Android app

15:08 the demo here looks interesting: https://github.com/Gonzih/reagent-native

15:12 RedNifre: What's the meaning of ^Something again?

15:20 bja: typehint

15:25 schmudde: I haven't done any mobile development work, and I don't plan on leaving emacs, but Nightcode looks like an interesting way to hit the ground running for Clojure across platforms: https://sekao.net/nightcode/

15:26 looks like it uses lein-droid...

15:26 Anybody use it?

15:27 bja: RedNifre: technically it's a tag

15:27 which doesn't have to be a typehint

15:35 dillap: Is there any writeups for adding reagent to an existing (non-cljs) clojure webapp?

15:35 Everything I can find assumes you are starting from scratch, and configuring the compiler and setting up figwheel is a bit intimidating.

15:43 RedNifre: hm, nightcode sure looks interesting.

15:46 sdegutis: blkcat: React Native? as opposed to Reagent?

15:46 Oh, Android stuff.

15:57 irctc: Is there a performance difference between running a compiled jar and using 'lein run'?

15:58 sdegutis: irctc: lein run will compile the jar and do some other fancy things, and spin up two JVMs

15:59 irctc: it's faster to run a jar directly, but the you don't have access to any Leiningen plugins (unless they were used to compile the jar)

15:59 irctc: gotcha

15:59 sdegutis: irctc: and it's recommended to build an uberjar (via 'lein uberjar' maybe) and deploy that jar to production for running

15:59 irctc: right, I'm doing that, just was curious about the other way

15:59 thnx

16:00 Anybody using docker to deploy clojure apps?

16:09 schmudde: irctc - I'm haven't used Docker, but my experience with Clojure and Heroku has been good so far.

16:10 codefinger: irctc: schmudde: use both! http://jkutner.github.io/2015/08/17/deploy-clojure-immutant-heroku-docker.html

16:11 schmudde: codefinger: interesting. Why would you want to use both?

16:12 codefinger: schmudde: having the locally development env is a plus. but if you are happy with your current heroku setup... i would recommend sticking with that.

16:12 you can mix and match too. so if you set it up for Docker, you can still deploy with Git

16:13 the addons is really key though. being able to run a "local heroku" with addons can help you debug stuff

16:13 schmudde: codefinger: I think I understand what you're saying. Don't I get the same features out of Leiningen's Environ? I have a dev environment and a production deployment.

16:14 codefinger: Yeah, debugging is tough. I cross my fingers whenever I deploy b/c I cannot test as a "local heroku". Interesting.

16:14 codefinger: schmudde: i agree. lots of people seem to want docker support though... because reasons :)

16:15 schmudde: wrt debugging http://jkutner.github.io/2015/05/19/heroku-remote-debug-java.html

16:15 schmudde: codefinger: Thanks for the tip. I'll dig a little deeper online. I'm new to the auto app deployment game. I've always done it manually, but my hosting provider isn't going to add JVM support any time soon.

16:16 codefinger: that's a bit of a hack. there is going to be some good/better stuff coming soon. can't really say (I work at heroku) but think nrepl :)

16:16 justin_smith: also, it's possible to go the opposite direction of docker too - something like OSv where there is no operating system, the hardware vm loads a jvm directly, and the jvm runs your code

16:16 schmudde: codefinger: Cool!

16:27 TEttinger: justin_smith: I can't remember, do you work for heroku or was it another long-time resident of #clojure ? Raynes?

16:28 justin_smith: TEttinger: technomancy once did

16:28 TEttinger: ah that would maybe explain it

16:28 justin_smith: TEttinger: being mistaken for technomancy is quite flattering

16:28 TEttinger: I know someone did it a few days ago with me

16:28 te*

16:29 (identity justin_smith) ;; BOT FAILURE

16:30 justin_smith: TEttinger: clearly what happened is my karma finally exceeded that of amalloy and so amalloy took the bot down to hide his shame

16:30 TEttinger: haha

16:30 justin_smith: (I don't think any of these things actually happened)

16:31 TEttinger: yeah amalloy stepped up his game when your karma started approaching his

17:28 devth_: my clojure's a little rusty. easy way to pick the first function that returns non-nil when applied to a value?

17:28 justin_smith: somefn

17:28 (doc somefn)

17:28 clojurebot: It's greek to me.

17:29 justin_smith: some-fn but, err, that doesn't pick a function, that picks a non-nil result

17:29 devth: (doc some-fn)

17:29 clojurebot: "([p] [p1 p2] [p1 p2 p3] [p1 p2 p3 & ps]); Takes a set of predicates and returns a function f that returns the first logical true value returned by one of its composing predicates against any of its arguments, else it returns logical false. Note that f is short-circuiting in that it will stop execution on the first argument that triggers a logical true result against the original predicates."

17:29 devth: justin_smith: it works in my case. thx!

17:29 not sure i ever used that even when i was writing a lot of clj

17:30 ((some-fn :no :yes :bad) {:yes 1})

17:30 my original question wording wasn't precise

17:30 justin_smith: ,((some-fn :no :yes :bad) {:yes 1})

17:30 clojurebot: 1

17:30 justin_smith: so there are multiple keys you might find, and the first non-falsey is good

17:30 sdegutis: Wow

17:31 some-fn looks incredibly useful in some cases

17:31 Can't figure out which ones tho

17:31 justin_smith: sdegutis: stupid apis that can't decide which key your data is under

17:31 sdegutis: ah interesting, for legacy stuff

17:31 TimMc: ,(filter #(% {:yes 1}) [:no :yes :bad])

17:31 clojurebot: (:yes)

17:31 sdegutis: what else?

17:31 devth: i'm actually using this to catch errors in stupid apis that name error message differently :)

17:32 sdegutis: TimMc: also juxt

17:32 justin_smith: sdegutis: but juxt does not short circuit

17:32 sdegutis: maybe outside the filter?

17:33 ,(filter #(if-let [x (% {:yes 1})] ((juxt identity %) {:yes 1})) [:no :yes :bad])

17:33 clojurebot: (:yes)

17:33 sdegutis: or something

17:33 I dunno anymore

17:33 justin_smith: oh, you are trying to get the first key that got a non-nil result

17:34 eg. what the question might have implied (but did not in fact want)

17:50 devth: would be neat if i could do (format ":foo is a :bar / :qux" {:foo 1 :bar 2 :qux 3})

17:51 justin_smith: devth: seems like an easy enough function to write!

17:51 devth: justin_smith: yep

17:54 but instead im resorting to: (apply (partial format "%s %s: %s") (map (% vpc) [:vpc :flavor :capacity]))

18:13 justin_smith: hmm... I think a clever regex could turn that into the version you want though

18:14 you could even get fancy and have :2d act like %2d etc...

18:49 amalloy: devth: (apply format "..." (map ...))). there's no need for this partial nonsense

18:49 devth: oh yeah. thanks

18:55 bitraider: any good recommendations on modern web app development in clojure... tutorials/books?

18:56 sdegutis: devth: actually yeah that function would rock

18:56 justin_smith: regex is *never* the answer

18:56 amalloy: that function has been written a lot of times. peronally i don't really see a big advantage over the existing format function

18:57 sdegutis: amalloy: it combines (format) with destructuring, with the added benefit of being able to see what's put in the string where relative to each other

18:57 devth: just a minor advantage :)

18:58 sdegutis: amalloy: so personally I think (format2 ":name, :email" user) is nicer than (format "%s, %s" (:name user) (:email user))

18:58 That said, I usually end up destructuring earlier, so it's (format "%s, %s" name email) which isn't /that/ bad

18:58 devth: y no string interpolation, clj?

18:58 even scala got it :-)

18:59 justin_smith: when we decide we don't have room for features, we give them to scala

18:59 scala is a feature hoarder

18:59 turbofail: there's a library that has a string interpolation macro

18:59 schmudde: lol

18:59 devth: justin_smith: ha ha

18:59 sdegutis: Scala is where our features go to die.

19:00 devth: clj where every imaginable lang feature exists as a lib

19:00 sdegutis: devth: except core.async is not as flexible as it could be in a static language, even with core.typed thrown in

19:00 devth: that's interesting

19:00 sdegutis: devth: so yes, features can be added via libs, but only /to an extent/

19:01 devth: so not exactly the same as e.g. haskell extensions

19:02 sdegutis: similar, but haskell extensions have their own limitations too.. like, it's really hard to dynamically find and refer to functions in files in the local tree

19:02 *local file tree

19:03 no one language can be everything, because some things within everything contradict each other

19:05 devth: i wonder if contradictory lang features can solved by importing features where need, like scala has done

19:06 sdegutis: There would be conflicts in libraries.

19:06 devth: but the imported feature is local to files

19:06 sdegutis: It'd be a bigger mess than it already is.

19:06 Consider the C++ ecosystem

19:06 devth: import scala.language.higherKinds

19:06 actually isn't that how haskell works too?

19:07 sdegutis: And what if one lib requires this feature to work, and another lib is incompatible with this feature, and you need both libs?

19:07 devth: the feature is only enabled for the local file

19:14 amalloy: that's what racket does

19:21 sdegutis: amalloy: weird

19:35 noncom|2: hmm, now thinking of the clojure and jvm startup time. could somebody tell me, why can't we have the java vm running as a service, without a restart and just wipe its memory to "restart" without actually restarting?

19:36 wink: noncom|2: have you had a look at drip or nailgun?

19:36 noncom|2: nope

19:38 wink: noncom|2: also, https://github.com/technomancy/leiningen/wiki/Faster

19:38 noncom|2: okay, these are practical solutions, i am going to try them out

19:39 wink: not saying there are real solutions, but people have spent thinking and tinkering ;)

19:39 noncom|2: but i wonder what is the explanation behind that JVM cannot be clean-wiped

19:40 in readme of drip it says that cake is taking this approach and it gets jvm overwhelmed by errors over time

19:42 also, drip is year old, does it fit today?

19:42 xeqi: some problems might be external resources like file handlers and sockets, and stopping background threads/threadpools

19:54 tom``: I'm trying to set up nginx in front of ring. Netstat shows it listening on ":::8080" but when I proxy_pass nginx to "http://localhost:8080" it give a 502 error.

19:54 curl to localhost:8080 says connection refused

19:55 but curl to :::8080 works fine

21:39 sdegutis: noncom: Component allows that workflow you're talking about

21:43 bitraider: has anyone used Luminus for web development...any comparison to Compojure ??

22:07 retrogradeorbit: hi

22:08 hlolli: Im in a let function and I want to find the max value of list/vector with (apply max) but I get java.lang.Long cannat be cast to IFn. I assume its because the list derives from various seqs, butall wont fix. Is there an alternative finding max value of say '(1 2 3 4) list?

22:09 butall = but doall... yes bugs make my eyes blurry

22:13 retrogradeorbit: sure max has not been rebound as a number... a long?

22:15 hlolli: if I try to cast it as long or int I get; Expecting var, but Double is mapped to class java.lang.Double


22:15 ok this one was double, same for all types

22:15 retrogradeorbit: what code are you running? (apply max '(1 2 3 4)) works fine for me

22:16 croak.macros> (apply max (map double '(1 2 3 4)))

22:16 4.0

22:16 how do I get the clojure bot here to run code?

22:17 ~4

22:17 clojurebot: Titim gan éirí ort.

22:17 hlolli: (apply max (map val (doall (select-keys @t-ref-buffer (modulo-keywords (if (zero? (:no @t1-ref-buffer)) 3 (:no @t1-ref-buffer)))))))

22:17 ,(prn "code running")

22:17 it's crambled

22:17 retrogradeorbit: ,4

22:17 clojurebot: "code running"\n

22:17 4

22:18 retrogradeorbit: ,(apply max (map double '(1 2 3 4)))

22:18 clojurebot: 4.0

22:19 retrogradeorbit: whats val, and max?

22:19 reduce is the same here as apply

22:20 ,(reduce max '(1.0 2.0 3.0))

22:20 clojurebot: 3.0

22:20 hlolli: the val if I do prn it's (1 1 1) with mapv [1 1 1] very simple

22:20 retrogradeorbit: (1 1 1) is not callable

22:21 hlolli: it's just that this data structure is different [1 1 1] can be different from [1 1 1]

22:21 yes

22:21 retrogradeorbit: how can you map (1 1 1) as a function

22:22 hlolli: but I still get Don't know how to create ISeq from: java.lang.Long

22:23 retrogradeorbit: thats dirrerent to what you firs posted

22:23 java.lang.Long cannat be cast to IFn.

22:24 hlolli`: lost internet, so I lost if you said anything last 3 min

22:24 retrogradeorbit: < retrogradeorbit> thats dirrerent to what you firs posted

22:24 < retrogradeorbit> java.lang.Long cannat be cast to IFn.

22:25 hlolli`: yes, right, was looking at the error from high speed of errors

22:26 later one is correct

22:28 retrogradeorbit: does select-keys always return a sequence? always?

22:28 hlolli`: I got this to work, really really weird solution

22:28 amalloy: i would hazard a guess and say never, retrogradeorbit

22:29 hlolli`: I just made new let argument do the apply max

22:29 really strange I think

22:30 retrogradeorbit: hey does anyone here have chops working with connection errors in http-kit client?

22:30 it _returns_ the error exception, not raising it

22:30 how do I get the details inside?

22:30 https://www.refheap.com/109969

22:31 how would I get access to the cause?

23:32 justin_smith: retrogradeorbit: you can use .getStackTrace to see the stack (probably not helpful to you), if you want more information check the type, if you are lucky it will be exceptioninfo

23:37 retrogradeorbit: .getMessage might also be informative

23:37 sdegutis: hi jus

23:38 justin_smith: hello sd

23:38 sdegutis: hows it be

23:39 justin_smith: chill, keeping it real

23:39 sdegutis: w.

23:40 justin_smith: hey i found out about reddit btw

23:40 thats where i found this https://www.youtube.com/watch?v=IikaDKvC1as&feature=share have u reddit?

23:41 justin_smith: oh man, back when Ed McMahon did more than run sweepstakes scams.

23:54 sdegutis: ye that

Logging service provided by n01se.net