wow - and I thought that clojure build errors could be indecipherable: https://www.refheap.com/106544 anyone got a clue what could cause this?

Using clojurescript 0.0-3308 and clojure 1.7.0

or should I be asking this stuff in a different channel

I’d like to use an external React component in a ClojureScript project. It’s an ES6 module that’s been transpiled to UMD. Can anyone offer suggestions on how to generate externs? This is the component: https://github.com/orgsync/react-list/

@lucien.knechtli: are you requiring korma.core in a cljs file?

No, I use it on the server

The cljs file is basically torn from the default reagent lein template, except I replaced secretary with bidi

@lucien.knechtli: is cider-nrepl pulling an outdated version of tools.reader? https://www.reddit.com/r/Clojure/comments/3295j6/clojurescript_003196_and_cidernrepl/

Oh.. If that's the case, is there any way to resolve it other than getting rid of cider-nrepl?

Disregard that, I'll try what is recommended in the thread

yay! i'm getting errors that make sense now!

Thanks

Any tips on FRP(-like) libs that play nicely with Reagent? https://www.reddit.com/r/Clojure/comments/3d87vt/clean_concise_frp_lib_integration_for_reagent/

@dkvasnicka: re-frame and zelkova are your two obvious options. Hoplon (not reagent) is also FRP.

What is the correct way of using the google clojure stuff? I am trying to go through the Secretary stuff but am coming against this error: WARNING: Use of undeclared Var xyz.routes/History at line 7 xyz/routes.cljs

I am trying to do something like this: (defn hook-browser-navigation! [] (doto (History.) (events/listen EventType/NAVIGATE (fn [event] (secretary/dispatch! (.-token event)))) (.setEnabled true)))

Basically it is the history stuff that messes up

do you have (:import [goog.history History]) ?

I have tried a number of different variations, I have (:import [goog.history]) which doesn't complain when loading the namespace form

your version doesn't work though, as far as I can tell

So this is basically what my ns looks like (ns sagsbehandler.routes (:require [secretary.core :as secretary :refer-macros [defroute]] [goog.events :as events] [goog.history.EventType :as EventType]) (:import [goog.history])) (History.)

And I get an error with ExceptionInfo #<TypeError: undefined is not a function> clojure.core/ex-info (core.clj:4403)

You saw the example here? https://github.com/gf3/secretary#example-with-googhistory

nvm, looks like you did

Yeah, so the thing is that I think that might be a bit out of date

The include macros thing doesn't work for me at least, so I have to change the imports to (:require [secretary.core :as secretary :refer-macros [defroute]] [goog.events :as events] [goog.history.EventType :as EventType]) (:import goog.History)

(History.) still fails

So I don't know what I am missing 😞

we use (:import [goog.history History]) and it’s fine

Maybe it is something with the version of cljs maybe,

I am on [org.clojure/clojurescript "0.0-3211"]. Can I ask for your version?

@casperc: your :import is still not right, (:import [goog History]) is probably what you want

Yeah that one doesn't complain, but (History.) is still giving the error: ExceptionInfo #<TypeError: undefined is not a function> clojure.core/ex-info (core.clj:4403)

@casperc: then you need to put on your debugging hat.

somehow goog.History is not getting required

examine the generated source, you should see a goog.require(“goog.History”) at the top of the file.

if you don’t something is wrong.

@dnolen: Right The Source looks like this: goog.provide('sagsbehandler.routes'); goog.require('cljs.core'); goog.require('goog.History'); goog.require('goog.history.EventType'); goog.require('goog.events'); goog.require('secretary.core'); (new goog.History()); //# sourceMappingURL=routes.js.map

Which seems ok

I am thinking it could be a repl problem

in your browser JS console try new goog.History()

That works fine, and I just tried calling it when loading a page and it does it to work fine there.

So I think I got fooled by a repl problem actually

Are JS classes and the REPL not friends?

in your REPL wrap the expression in (fn [] …) to see the generated code

Hi, what’s the forward looking way to include a pure JS lib in a clojurescript project? CLJSJS still a good option? Or is there a something little bit less ceremonious?

@arnaudsj: I can’t think of anything less ceremonious than CLJSJS, you completely avoid manual configuration

@dnolen: (fn [] (History.)) becomes: #<function (){ return (new goog.History()); }>

@dnolen: I meant that the lib in question does not exist in the CLJSJS repo, so it appear I need to write an extern file for it (or am I wrong?)

@casperc: so something is wrong with your assumptions. I would enable stop on breakpoints in your browser to see what is actually happening. I would disable source maps for this.

er "break on exception"

@arnaudsj: the first thing you should determine if Closure does not have the functionality of the library you are interested in.

@dnolen: I am looking at https://github.com/jstat/jstat which is neat stats library, and I don’t think I have seen this in Clojure land (incanter does have some of the things I want, but I did not think incanter could be brought into CLJS land)

in particular I am interested in the Beta distribution implementation for sampling.

@arnaudsj: ah cool, yeah you need externs to use that.

@casperc: does this working code help? https://github.com/Day8/re-frame/blob/master/examples/todomvc/src/todomvc/core.cljs No wait. My Slack session has just caught up and I can see a LOT of discussion since you asked. So ignore this.

@dnolen: ok, let me give it a shot then, if I am successful i will send a PR. Thanks!

How do you create a uuid in clojurescript ? I'm getting (random-ui id) var not found errors in figwheel/0.0-3211

@dnolen: So I am pretty sure it is a REPL problem. I am pausing on all exceptions in the browser and it doesn't break. But I tried defining the funtion to print the History object and it is actually printing it while throwing the exception in the REPL.

That should be (random-uuid)

@casperc: what REPL are you using?

Figwheel's REPL

@adwelly: sounds like you don’t have the right ClojureScript dependency

@casperc: I would try this with a regular ClojureScript REPL a la Quick Start

if it works then it’s a Figwheel bug

It does seem like it, because everything else seems to work, but I'll give it a go in a clojurescript REPL and make a bug report fir figwheel if it ends up being the problem

Thanks for the assistance

@casperc: np

I would create a sample reagent project using the lein template and compare my core.cljs with their core.cljs.

I've tried requiring cljs.core and also (core.cljs/random-uuid) but no joy

Figwheel seems to be working perfectly in all other respects though

Largish web app running in it without problems

@adwelly: what version of ClojureScript? pretty sure random-uuid is 3308 only

@dnolen 0.0-3211 - aha!

yeah random-uuid is brand new

@dnolen, @roberto, @casperc Thanks for the advice. I will give this a try

https://github.com/cljsinfo/cljs-api-docs/tree/catalog#latest-changes

https://github.com/cljsinfo/cljs-api-docs

cljsinfo is doing an amazing job documenting exactly when things are added or removed.

...and there was me thinking I had the latest...

brain explosion https://twitter.com/mayaneise/status/620746001599582208

💥

waoahahahahahah

Not enough 👍 to @maria

👏 @maria

https://twitter.com/mfikes/status/620972615784636417

And, if you are interested in helping verify patches: http://dev.clojure.org/jira/secure/IssueNavigator.jspa?mode=hide&amp;requestId=10707

is there a recommended best practice for adding “vendor” js libraries to a project? (e.g. foundation.js, jquery, etc)?

@curtosis: http://cljsjs.github.io/

ahh, yes. now I remember. sadly, no foundation.js and family 😞

I’m transitioning from a prototype I did several months ago to a piece of a luminus project, and I haven’t made all the mental mappings yet.

https://github.com/cljsjs/cljsjs.github.io/issues/3

heh

so if there’s something not in cljsjs yet, how should I integrate/refer to it?

fwiw I’m porting over a page template from liftweb, so I’m hoping to avoid rewriting it completely on this pass.

@curtosis: I use this to work with some JS libraries that I wrapped myself. Allows me to keep them in git. https://github.com/LonoCloud/lein-voom

hmm… I suspect it’s the wrapping bit that I need more than the artifact bit… I already have a local repository in my git project for those two 3rd-party java libraries that aren’t online. lein-voom seems to solve that problem, differently?

@curtosis: Yeah it allows me to treat my git repos like clojars, at least that's how I'm using it.

so I think I need something kind of like this: https://github.com/clojure/clojurescript/wiki/Packaging-Foreign-Dependencies with the caveat that right now I don’t actually need to access them from . I just need the foundation calls in the page template to work. cljsifcation to come later

@curtosis: I used this to help me get started generating the externs. It's not perfect, but it helped. http://www.dotnetwise.com/Code/Externs/index.html

do I even need the externs, though? I think at this point all I’m trying to do is get bog-standard js libraries copied over into the resources/public/js directory controlled by cljsbuild.

so perhaps it’s just as simple as a :libs entry under the :compiler key? hmmm….

Those would have to be closure compatible though.

because it’s going to try to compile it all together?

even though it’s not referenced from clojurescript?

https://www.youtube.com/watch?v=DPHkBp9Mkzk

If it's not referenced, what is it being used for? I got by without externs at first, but once I tried to do advanced compilation I had to generate some. Otherwise the Closure compiler gets cranky.

in the simplest case, the page template loads up foundation.alert.js, and calls $.foundation() to bind to the alert box onClick.

very basic old-school.

another way to think about it is that this is (currently) a ring app that includes JavaScript. The side comes later.

so the question could be better framed as "how do I include static js resources for html pages served by ring"

short of just including links to cdn versions. :p

@curtosis: Ring has static resources middleware

@dnolen: right... I think luminus hijacks the default resources/js directory for cljsbuild. But thinking about it now I don't see why I can't use a different directory for the static (non-closured) stuff.

you can use compojure.route/resources

http://weavejester.github.io/compojure/compojure.route.html#var-resources

@roberto: there’s one of those in the default luminus template (route/resources “/“) that is presumably mapped somewhere to resources/public?

i don’t know. I’ve never used Luminus

if route == compojure.route, then it should be

it is, yes

this is less helpful than I'd hoped: http://www.luminusweb.net/docs/routes.md#static_resources

>By default, any resources located under the resources/public directory will be available to the clients. This is handled by the compojure.route/resources handler found in the <app>.handler namespace:

(defroutes base-routes
  (route/resources "/")
  (route/not-found "Not Found"))

so, you want to have access to a js library, for example a vendors/jquery.js ?

exactly

you should put that folder under resources/public

then you can refer to it as /vendors/jquery.js in your html

so, the ‘/‘ is resources/public

I think that's where the discussion has led me today, yep

yeah... that's the mapping I'm not clear on, though. where the "/" -> "resources/public" mapping is stored. I can work around it, but I'd like to know where the magic is.

If I wanted untraceable magic, I'd be using rails.

http://weavejester.github.io/compojure/compojure.route.html#var-resources

it states there

the root prefix path of the resources, defaults to ‘public’

The luminus docs also says it :

By default, any resources located under the resources/public directory will be available to the clients. This is handled by the compojure.route/resources handler found in the <app>.handler namespace:

i thought that route/resources itself wasn’t working for you

ok, sure. I guess it could be clearer, since to properly make all those connections you have to know that: a) resources/route has a default :root parameter b) that path is relative to resources/

and, specifially wrt luminus, that c) resources/public/js is managed (and cleared) by the cljsbuild process

looks like b) comes from lein-ring :resources-path defaults to resources/

or rather, that's leiningen's default; lein-ring just uses it as the base dir.

definitely re-thinking my choice to start from luminus rather than building it up library-by-library. 🍸

but yes, dropping things into a resources/public/vendor directory worked like a charm. thanks for the many pointers leading me there.

I haven’t used luminus before. It does look odd to me that it re-writes public/js

yeah... here's a bit from project.clj:

:clean-targets ^{:protect false} ["resources/public/js"]
  
  :cljsbuild
  {:builds
   {:app
    {:source-paths ["src-cljs"]
     :compiler
     {:output-dir "resources/public/js/out"
      :externs ["react/externs/react.js"]
      :optimizations :none
      :output-to "resources/public/js/app.js"
      :pretty-print true}}}} ...

I'm not too far in, so I'm still considering a ground-up, one-library-at-a-time rebuild.

the benefit of knowing where all the plumbing is is nontrivial

@tonsky: "In ClojureScript #datascript/DB handler is installed by default, just use … ” really? That sounds strange to me. Is datascript/DB something that’s explicitly handled in CLJS?

No, if you’re using DataScript, of course

DS just populates global CLJS tags table

Ah, ok. I misread then

Reformulated: > In ClojureScript #datascript/DB handler is installed globally when using DataScript, so read-string just works:

hi all - do the optimisations make any difference to run-time performance for ClojureScript or is it just about filesizing/combining? Objective or anecdotal experience is fine 😉

×2 improvment

Yes, advanced is faster

It’s strange, but almost always when I measure it’s about ~2 times faster under advanced

the number is very consistent

OK, two out of two for faster - thanks!

I also remember @dnolen mentioned that that’s mostly thanks to CLJS compiler optimizations, not because of Google Closure

gclosure is just minification

but I believe it should do some inlining too

No, gclosure under advanced mode does more than just minimization. It does inlining as well as several other algebraic code transforms.

In certain pathological cases it can be dramatically faster. It's about 6x faster than identical code when running Enlive in cljs.

thanks @luke

In an intense HTML canvas project I was hacking on a few months back, advanced made the difference between sadness and 30fps

Ok, shameless self-promotion: DataScript now has a Tips & Tricks wiki page https://github.com/tonsky/datascript/wiki/Tips-&amp;-tricks

@luke: advanced really doesn’t buy you anything at all in terms of perf

most closure elimination happens in simple

inlining does matter at all, and a lot advanced inlining is hurtful because callsites become megamorphic

after having benchmarked tons and tons of code

:simple + :static-fns true is the fastest combination across all major JS runtimes

previously JavaScriptCore and Firefox didn’t do property caching so the namespace pattern hurt you, but far as I can tell this hasn’t been the case in a long time

the primary advantage of :advanced is code size

this said :static-fns and :optimize-constants don’t kick in until :advanced because they interact badly with REPL driven development

so you have to use them explicitly when using :simple

just clarifying that the real ClojureScript perf optimizations are handled by ClojureScript not Closure.

is there a way to turn off inlining in advance mode but keep minification?

@dnolen then how come I have observed 2-6x perf increases between :simple and :advanced?

@luke: I already explained that above

:advanced permits ClojureScript to generate more aggressive code.

ah, sorry, missed that. Thanks.

so much to keep up with in CLJS. Turn my back for a month and everything's different 😛

@luke a lot of these optimizations are actually years old

but I never explained them anywhere

actually StrangeLoop once, but not sure how many people have seen that

but that was 2 years ago

yeah, that stuff should probably be documented at some point. I need to sit down with you at some point (prior to the next release of the CLJS book) and get a brain dump on that kind of thing.

@luke more than happy to oblige

@dnolen: yoyo! any suggestions on how to compare qualifiers? i only compared major and minor and got bitten by the qualifier part. alpha versions of 1.7.0 wont work with cljs but beta will. I can compare strings by mapping them to numbers same way maven does, but that will add a few lines more and Alan Kay would not be happy about that 😉

@annapawlicka: I think we should just assume the known qualifiers that have been used, “alpha”, “beta”, “RCN"

there’s never really been others

cool, that’s only 3 instead of 7

Hey there, sort of a beginner question here. We (company I work for) are transitioning over to CLJS/Om. Looks like the standard Chestnut app boilerplate calls (sh "lein" "less" "auto") inside the start-less function, meaning it never kills off the child lein processes and they build up over time. NBD to kill them but I am wondering 1. if any of you have experienced this before and there is a generally-accepted solution 2. if any of you are aware of a relatively clean way to check if the lein less auto process is already running before invoking it again (iow not accepting the output of ps and testing a regexp against the string returned from the sh function) Sorry if this is old/newbish/etc, I wasn't able to figure this out by googling but assumed it must be a relatively common thing in CLJS dev environments