Can anyone point me to any good example of client side routing and rum?

not sure if i should post this in here or in the clojure chat

but

i wrote a script that looks for namespaces using clojure.tools.namespace.find

it works fine with .cljc files until i try to use #?( syntax in the ns at the top

then the namespace mysteriously disappears from what clojure.tools.namespace.find/find-namespaces-in-dir can see

anyone seen any behaviour like this at all?

@thedavidmeister Is it possible to post the ns form under discussion? Heard similar issues in the ns form.

@iku000888 yup, i’ll put something up in a sec

I'm refactoring an app I wrote about a year ago (my first web app !) to use the reagent atom instead of cljs.core/atom, and hopefully clean up a lot of the state management logic.

(It's this app here: http://gregor-samsa.herokuapp.com/register-machine.html)

ok so

I guess I'm a little unclear on how to accomplish my goal: ideally I'd like the input atom to result in a stack of inputs. Is the "correct" way to do this in reagent to keep the stack in a separate component?

i tried this

(ns styles.card
  (:require [garden.units :as u]
            [styles.colours]
            [styles.spacing]))
            
#?(:cljs (require '[garden.def :refer-macros [defstyles defrule]]))
#?(:clj (require '[garden.def :refer [defstyles defrule]]))

cljs didn’t like that WARNING: Use of undeclared Var styles.card/require at line 6

then i tried

(ns styles.card
  (:require [garden.units :as u]
            [styles.colours]
            [styles.spacing]
    #?(:clj [garden.def :refer [defstyles defrule]]))
  #?(:cljs (:require-macros [garden.def :refer [defstyles defrule]])))

i don’t get any more errors

but also

styles.card no longer appears in the list from clojure.tools.namespace.find/find-namespaces-in-dir

https://clojure.org/guides/reader_conditionals#_namespaces

Not too familiar with reader conditionals, but seems like you need to wrap and then splice the clj branch

Plus I think you still need to require the namespace for cljs? (Crossing my fingers here...)

@iku000888 i can use defstyles in this file and then prn the result of that in the browser

so at least the cljs side of things seems to be working ok

actually i can also require styles.card in clj and prn the result of defstyles there too

so i can use the macros in styles.card just fine in both clj and cljs

it’s just that when i run

(prn (clojure.tools.namespace.find/find-namespaces-in-dir ( "src/cljc")))

every namespace except styles.card appears

I would then try to debug the ns form with (read-string {:read-cond :allow} "ns form as string" ) in both cljs and clj

To see how the reader conditionals turn out.

oh ok, i haven’t done that before

i’ll try that

gotta duck out for a bit

thanks for the tip though

i’ll let you know how i go 🙂

Good luck 👍

Hey, I wanna test some cljc code but I think I have the structure wrong. For example this simple test fails

(ns ns-test
  (:require-macros [cljs.test :refer (is deftest async)]
                   [cljs.core.async.macros :refer [go]])
  (:require [cljs.test :as test]
            [cljs.core.async :refer [<!]]))
(defn adder
  []
  (go (+ 2 3)))

(deftest test-get-stats
  (async done
         (go (is (= (<! (adder)) 5))
             (done))))

So yeah what am I missing? I’m reading https://github.com/clojure/clojurescript/wiki/Testing

Oh man will you people please wake up?

Perhaps it would help to post what kind of error you get?

Based off http://stackoverflow.com/a/30781278

No error, just says test failed

(ns milia.api.submissions-test
  (:require-macros [cljs.test :refer (is deftest async)]
                   [cljs.core.async.macros :refer [go]])
  (:require
   [cljs.test :as test]
   [cljs.core.async :refer [<! >! chan take!]]))

(defn test-async
  "Asynchronous test awaiting ch to produce a value or close."
  [ch]
  (async done
         (take! ch (fn [_] (done)))))

(deftest test1
  (let [ch (chan)]
    (go (>! ch "Hello"))
    (test-async
     (go (is (= "Hello" (<! ch)))))))

Not familiar with core.async, but [cljs.test :refer (is deftest async)] should be [cljs.test :refer [is deftest async]] ?

Sorry for not being able to help directly 😞

iku000888 I don’t see the difference in what you wrote

Oh a vector

Thanks

@urbanslug the (done) is called before the is executes

the vector vs list doesn't matter

thheller How do I make it block?

The code I am seeing from other people is so cryptic

uhm you can only call it once per test

Some answer on SO says "Tests are executed synchronously, so if you go async the test-runner won't. In Clojure you need to block the test runner via <!!, in ClojureScript you have to return an async test object. "

(deftest test1
  (async done
    (let [ch (chan)]
      (async/put! ch "Hello")
      (go (is (= "Hello" (<! ch)))
          (done)))))

the deftest should do all the coordination and decide when to call done

hmmm I don’t see what you did so differently from this

(defn adder
  []
  (go (+ 2 3)))

(deftest test-get-stats
  (async done
         (go (is (= (<! (adder)) 5)
                 (done)))))

which fails thheller ☝️

@thheller

I messed up the done when you said it should be once per test

(<! (adder))

one less form

ah wait nvm

😄

@iku000888 not sure how to get read-string working in the way you suggest in cljs

but in clj:

why is adder a go?

(read-string {:read-cond :allow} "(ns styles.card
  (:require [garden.units :as u]
            [styles.colours]
            [styles.spacing]
    #?(:clj [garden.def :refer [defstyles defrule]]))
  #?(:cljs (:require-macros [garden.def :refer [defstyles defrule]])))")

results in

(ns styles.card (:require [garden.units :as u] [styles.colours] [styles.spacing] [garden.def :refer [defstyles defrule]]))

which seems fine

Even your code fails for me

@thheller

(ns milia.api.submissions-test
  (:require-macros [cljs.test :refer [is deftest async]]
                   [cljs.core.async.macros :refer [go]])
  (:require [milia.api.submissions :as sub]
            [cljs.test :as test]
            [cljs.core.async :refer [<! >! chan put! take!]]))



(deftest test1
  (async done
         (let [ch (chan)]
           (put! ch "Hello")
           (go (is (= "Hello" (<! ch)))
               (done)))))

sec, door

@thheller I just wanted a simple async function

sec? door?

adder is async because it is simple

btw I gave up and ended up asserting the type https://github.com/onaio/milia/blob/48ce90f7a7b070228c60ff9f8bfea1381218b5f6/tests/cljs/milia/api/submissions_test.cljs

Which doesn’t do anything really

any go block will pass that test

(deftest test1
  (let [ch (async/chan)]
    (async/put! ch "Hello")
    (async done
      (go (is (= "Hello" (<! ch)))
          (done))
      )))

this works for me

@thedavidmeister http://stackoverflow.com/questions/20046717/clojurescript-reader-read-string-returns-null

Apparently you need to explicitly require cljs.reader to use the read-string function

(cljs.reader/read-string "(ns styles.card
    (:require [garden.units :as u]
              [styles.colours]
              [styles.spacing]
      #?(:clj [garden.def :refer [defstyles defrule]]))
    #?(:cljs (:require-macros [garden.def :refer [defstyles defrule]])))”)

@thedavidmeister cljs.reader does not support conditional reading

Error: Could not find tag parser for ? in ...

Yeah... Just figured that out... @thheller

but even so, as i’m running (prn (clojure.tools.namespace.find/find-namespaces-in-dir ( "src/cljc"))) in clj, does it matter?

sorry I missed the context of what you are trying to do

@thheller for some reason the #? stuff in my cljc ns is causing the ns to not appear when i run clojure.tools.namespace.find/find-namespaces-in-dir

works for me?

@thheller can you paste your code?

its your code

(ns styles.card
  (:require [garden.units :as u]
            [styles.colours]
            [styles.spacing]
    #?(:clj [garden.def :refer [defstyles defrule]]))
  #?(:cljs (:require-macros [garden.def :refer [defstyles defrule]])))

like that?

yep

you sure your file is in src/cljc not src/clj or src/cljs by accident? 😛

yeah it is

i’m now wondering if the garden boot task is doing something...

(clojure.tools.namespace.find/find-namespaces-in-dir ( "src/main"))
=> (styles.card)

hurrr

@thedavidmeister: do you have an explicit dependency on clojure?

If not it might be possible that the pod the garden task uses ends up loading a different clojure version

i added clojure in my dependencies, yeah

1.8.0 btw

nah i don’t think it’s the garden boot task

i did this

(deftask garden
  "Wraps the garden task provided by boot-garden"
  [p pretty-print? bool "Pretty print the CSS output."]
  (comp
    (watch)
    (with-pass-thru [_]
      (require '[styles.compile])
      (styles.compile/foo))))

(ns styles.compile
  (:require [garden.core]
            [clojure.tools.namespace.find]))

(defn foo []
  (prn (clojure.tools.namespace.find/find-namespaces-in-dir ( "src/cljc"))))

(styles.animation styles.border styles.box-model styles.box-shadow styles.colours styles.compile styles.neat styles.spacing styles.typography)

no styles.card 😞

@martinklepsch on a side note, i don’t really understand the pods thing. What is it doing for the garden task?

@thheller what version of clojure are you using?

1.9.0-alpha13 but that shouldn't matter

tried 1.8.0 same result

Have to go now but basically it helps isolate dependencies by providing a construct "pod" that can have a different classpath than the rest of the application

oh ok

i’m thinking i want my garden to sit alongside my application

maybe that’s wrong

@thheller so weird

have you tried in a REPL without boot?

i tried with boot repl

is there another way to get a repl that might find it?

find src/cljc

I somehow can't believe that the file is getting filtered 😛

is it really there?

and what happens if you (require 'styles.card)?

oh i can require it just fine

i checked that i can require it and prn the output of defstyles in both clj and cljs

if I see that right (styles.animation styles.border styles.box-model styles.box-shadow styles.colours styles.compile styles.neat styles.spacing styles.typography)

is missing styles.breakpoints as well?

yea

well that has the same issue

strange indeed

(ns styles.breakpoints
  (:require [garden.stylesheet :refer [at-media]]
            [garden.units :as u])
  #?(:cljs (:require-macros [styles.breakpoints :refer [defbreakpoint]])))

#?(:clj
    ; 
    (defmacro defbreakpoint [name media-params]
      `(defn ~name [& rules#]
         (at-media ~media-params
           [:& rules#]))))

but they appear if you remove the #??

any file i stick the cljc reader thingo into the ns for disappears from the list

yeah

(ns styles.breakpoints
  (:require [garden.stylesheet :refer [at-media]]
            [garden.units :as u]))
  ; #?(:cljs (:require-macros [styles.breakpoints :refer [defbreakpoint]])))

#?(:clj
    ; 
    (defmacro defbreakpoint [name media-params]
      `(defn ~name [& rules#]
         (at-media ~media-params
           [:& rules#]))))

=

boot.user=> (prn (clojure.tools.namespace.find/find-namespaces-in-dir ( "src/cljc")))
(styles.animation styles.border styles.box-model styles.box-shadow styles.breakpoints styles.colours styles.compile styles.neat styles.spacing styles.typography)

i can use #? in the file

but not in ns

I know nothing about boot but did you check if the classpath is clean?

no conflicting versions of tools.reader on there?

not sure how to do that

i can set a dependency though

what version are you using?

actually it does look like ring depends on [org.clojure/tools.namespace "0.2.10”]

org.clojure/tools.reader "1.0.0-beta1"

via clojurescript

+ org.clojure/tools.namespace "0.2.11"

[org.clojure/tools.reader "1.0.0-beta3”]

omg that worked

boot.user=> (prn (clojure.tools.namespace.find/find-namespaces-in-dir ( "src/cljc")))
(styles.animation styles.border styles.box-model styles.box-shadow styles.breakpoints styles.card styles.colours styles.compile styles.neat styles.spacing styles.typography)

hopefully it’s an upgrade in tools.namespace and not a regression in tools.reader...

i’ll check

yay

so i think it was just a bug in 0.2.10

@thheller thanks

hehe good, i was out of ideas 😉

in fairness, this is probably the first thing i should have tried

i just assumed i was doing something wrong in cljc

I would expect some kind of error in that case

this is the first time i’ve used cljc

all this so i can pass my data from garden to js libs 😛

Hi! I am new to clojurescript (and clojure). I'm trying to work out an architecture for an isomorphic web app (clojure back-end). I know there are some examples of server-rendering in reagent. Examples aren't so good for rum because it's unopinionated. That's one issue. The other is how to handle CSS. There are some suggestions out there to use sass, but that's not enough. What about autoprefixer, removing unused CSS, etc.? It seems like there aren't any lein or boot tools for this yet. Is my only option to use node for post-processing CSS?

I don’t understand the output of (type #js [1 2 3]):

#object[Array "function Array() { [native code] }”]

I'm new to CLJS, so I'm not sure what the hash means, but "Array" is the right constructor. The 2nd part looks like a string representation of the Array constructor in JS. "[native code]" is binary code.

@borkdude type in cljs gives you the constructor of the object. in your case that is the Array function which is native code

I see

cljs.user=> (source type)
(defn type
  "Return x's constructor."
  [x]
  (when-not (nil? x)
    (.-constructor x)))

@borkdude you might be looking for goog/typeOf

cljs.user=> (goog/typeOf #js [])
"array"

the #object... is just the printer trying to print edn compatible output

https://twitter.com/nolanlawson/status/793113537908400129

Unsurprisingly Closure dominates load time and run time against all other JS tooling as project complexity increases

Is ^^^ a bug introduced with CLJS-1658?

@keeds probably file a bug - you should be able to reproduce this on js/String though, I would leave anything that involves a browser out of the issue

@keeds actually I cannot replicate on js/String so will need more information

Ok. Will do.

When dealing with cljs/js interop and libraries that use this-context method chaining, what’s the best way to invoke that from cljs? for example:

someLibrary(a_div).doSomething({ args: true}).doAnotherThing()

If I try a thread macro the this context always comes up null/undefined, which I assume is due to the rewriting of the macro not knowing/caring about this?

My other suspicion is that perhaps I'm misusing closure’s object.get, expecting it to do something it’s not…

(-> (js/someLib node)
        (#((oget % "foo") (clj->js {:some :args})))
        (#((oget % "on") "resize" (clj->js {:some :more-args}))))

@lwhorton just use method chaining

(.. js/foo (bar a b) (baz c d) (woz e f))

(.foo x) is just sugar for (. x (foo))

(.. x …) is just repeated method calls using each preceding return value as the first argument

@dnolen FWIW I can repro with js/String in 1.9.293:

cljs.user=> (extend-type js/String ISeqable (-seq [x] x))
            ⬆
WARNING: Extending an existing JavaScript type - use a different symbol name instead of js/String e.g string at line 1
#object[Function "function (x){
var x__$1 = this;
return x__$1;
}"]
cljs.user=> (first "asd")
No protocol method ISeq.-first defined for type object: asd

@anmonteiro try it with a custom protocol

that’s what I tried

aha... sugar gave me semi-colon cancer? I’ll have to investigate how .. handles advanced compilation to avoid munging issues, but thanks for the tip @dnolen ~

@lwhorton there’s no implications for advanced compilation

the only time you should ever be thinking about advanced compilation is when you’re including some random JS library

and the only thing you ever should be doing is providing externs for that stuff

I was hoping to use object.get and object.set and strings to avoid having to write an externs entirely

I don’t know why

a custom protocol works.. still don't understand why the above wouldn't?

since I’m only using a tiny piece of the library, and a complete externs is a much larger committment

@anmonteiro well there’s our bug 🙂

@lwhorton you never need complete externs

I write one line externs files all the time

FWIW it doesn't work on 1.9.229, so I could also be doing something wrong

this externs fear needs to stop 🙂

my $0.02 https://github.com/binaryage/cljs-oops#externs-from-hell

oh… I thought I was being a bad cljsjs citizen by not writing a full externs

🙂

@lwhorton if you’re writing an app and you’re not publishing something who cares how complete your externs are

all this time trying to avoid writing externs is just a huge waste of time - IMHO

eventually we will automate what we can - but it’s really not that hard to write a couple of line for whatever you are using from some random library if that’s all you need

@anmonteiro the issue as @keeds reported it is really a red herring anyway

@anmonteiro if this is failing it’s most certainly going wrong at satisfies? or implements?

should (extend-type js/String ISeqable (-seq [x] x)) work though?

@anmonteiro yes it should

I think it should, but there could be something preventing that

but at least we know some things are working and others aren’t - which is a bit more information

once again, much appreciated @dnolen

@lwhorton https://github.com/cognitect/transit-js/blob/master/resources/externs.js

this is what I did for transit-js, value and done weren’t in the Google Closure externs yet for the JS Iterator stuff

not exactly a lot of work

I’ve never seen an externs operate directly on Object.. always some variable = function () {}; variable.prototype.foo = ...

that must mean you can use a callback fn that’s handed a transit object and (.-tag/rep/done/etc. obj) becomes un-mangled (or rather never mangeld)?

@lwhorton all an externs file is a bunch of top level names

if you put a property on Object, then when Closure sees a property with that name and infers type Object, it knows not to rename

I see.. that’s much simpler. (— stupid redacted—)

A counter point could be that when you set some arbitrary wire-related externs on Object it can bite you a few months later when you forget about it and start wondering why some arbitrary names do not get renamed in advanced mode in cases where you would expect them to. It can be pretty hard to track down if such externs are bundled in some library or otherwise hidden from plain sight. There is no tool which would tell you “this happened because of Y”, at least I’m not aware of any.

the real solution is everyone should stop using javascript and write everything in cljs from here forward, obviously

@darwin “wire” related externs doesn’t make sense

you don’t need externs for wire stuff

that’s what goog.object is for

the more people keep conflating the reasons for this stuff the longer everyone stays collectively in the dark

but you have to know it, you have to know, that the object you are passed is “external data” so you should use string names to access its properties, the line can be pretty blurry here

there are static properties of your program - this is what externs are for

then there is runtime data - you don’t need externs

yes, I agree

the more you bang this drum

the faster newcomers will figure it out

more precisely "you don’t need externs and you must NOT use dot property names"

right don’t use dot property names for dealing with data

speaking about it this way also has wider implications even when we get externs inference

but still, I’m afraid, this sounds all clear and easy to you and me, but newcomers really struggle to distinguish these nuances without deeper understanding how closure compiler, javascript and clojurescript work

we can probably infer static properties of programs

we will probably not be able to infer dynamic properties of data

@darwin sounds like a guide worth writing for http://clojurescript.org for someone who is willing 🙂

🙂

your word choice makes me think this is work under way with the compiler? that will be exciting

@lwhorton there’s some prototype work I got started on and it will probably happen relatively soon

however I’m side tracked a bit on Node.js / ES6 integration work

Hello everybody, anyone know of a decent logging library that works for both clojure.tools.logging in JVM Clojure and (.log js/console) in browser-based ClojureScript?

timbre has adapters for c.t.l

and works in browser

@timgilbert Google Closure also has one if shared logging isn’t a goal

Oh, interesting, will look at that

Not sure if timbre will work or not, but I'll give it a look too

@timgilbert I started working on one, but my goal is not to make it 100% compatible, I will just mimic clojure.tools.logging api in cljs and make it cljs-devtools friendly, e.g. it will use pprint in advanced mode (if not elided)

@timgilbert if you have ideas to share, I would like to hear them - I’m still not 100% sure how I want to approach it, want something like https://github.com/adzerk-oss/cljs-console, but without string printing, and api similar to clojure.tools.logging, in dev mode it would just boil down to console.log calls, in advanced mode some or all log levels can be elided and those which are not would be presented with pprint or similar alternative code-path

also I want to support formatting from devtools, but in a different way, not inheriting inflexible C-format strings which don’t compose

That all sounds great @darwin. Right now I'm using shodan and its killer feature is that it uses macros, so that the log statements in the console all point back to the line in the actual source file where the log statement was, instead of all pointing to some line of code in a library somewhere

https://github.com/noprompt/shodan/

@timgilbert yes, I will support that as well

Personally my use-case is that I have a fairly small set of cross-platform code that I need to call from both Clojure and ClojureScript, and I'd like to be able to minimize the amount of #?(:clj :cljs) blocks in there, to zero if possible

Right now I'm looking at just writing a thin macro layer, which isn't hard to do but I thought I'd check if something already existed first. I did look at timbre briefly but some other folks on my team seem to dislike it