@alun Cloverage doesn’t do ClojureScript, but should be not that hard to port to it since it’s a source transformation

@alun As its maintainer I’d be happy to do code review 🙂

[lein-figwheel 0.5.9] has just been released

https://github.com/bhauman/lein-figwheel/blob/master/CHANGES.md

I have been playing with the new externs inference feature. Does anyone know how to typehint this correctly?

My best guess was this, but that didn't seem to work:

So I learned in my pursuit of secure sockets and such for my cljs application..... always use HTTPS. That way, all communication is within an encrypted layer. On top of that, you can do logins and the initial over-the-wire username/pw combination can be considered secure. Once you have passed a token back to the client that they send on every request, you can also encorporate some "is this the same IP as before?" and make sure tokens expire over time so that your site stays secure even if token info is somehow compromised

I just upgraded to cljs 1.9.456 and my conformer spec started throwing errors

error is cljs.spec.spec_impl.call(...).call is not a function

;; conformer
(defn coll->set [x] 
  (cond
    (set? x) x
    (coll? x) (set x)
    :else :clojure.spec/invalid))

(s/def ::my-spec-coll (s/coll-of ::my-spec :kind (s/conformer coll->set) :into #{}))

I'm seeing a problem with array-seq in advanced compilation mode. Works fine with no optimizations but in production I keep getting Cannot read property 'length' of undefined exceptions.

It's within a this-as macro that I'm calling array-seq

@danielstockton the array you call array-seq on: does it come from a javascript option which might need externs? ie. (array-seq (.getThatArray someJsObject))

that is missing externs?

@thheller yes, thats probably it.

:pseudo-names true is very useful for such situations

What does that do exactly? Emit certain warnings during compilation?

I'll just try it out..

it'll mangle names, but in a predictable way

The lib i'm using is a cljsjs package, does that mean they missed an extern?

@pesterhazy Does that mean it's not something I want turned on all the time for my production builds? Are there disadvantages to it being true?

it may not mean that, e.g. if it's in a callback

no you only turn it on during dev time, temporarily, to figure out what's going on

Ah right, ok. It is in a callback, would adding ^:export metadata help?

no I don't think so

(aget this "field")?

probably best to just use goog.object/get instead of .- notation

mhm ok, thanks

IMO every call to .- needs to vetted to see if it'll continue to work during adv comp

personally I mostly resort to goog.object/get by default these days

Ok, thanks for the help, that sorted it.

👍

How to find all vars in some namespace? Clojure has ns-publics

@isaac there’s no way to do those kinds of Clojure var things at runtime

you can do it at compile time with macros and the cljs.analyzer.api namespace

The cljs.analyzer.api/ns-publics returns is not vars. I want to filter vars by meta data.

argh ... cljsbuild fails on travis-ci with Can't redefine a constant at line 61 ..., but it works locally. I've fought this issue once, but it went away, now it came back out of the blue 😞

back then I assumed this was a windows issue, but travis runs linux

hi, in a regeant project template, I've added [org.clojure/core.async "0.2.395"] in my project.clj file, and then in core.cljs I require [cljs-http.client :as http] [cljs.core.async :refer [<!]]

running lein figwheel in project root, I get this error: No such namespace: cljs-http.client, could not locate cljs_http/client.cljs, cljs_http/client.cljc, or Closure namespace "cljs-http.client"

@tomaas you didn’t declare your dependency on cljs-http in your project.clj

sorry, thanks @dnolen

@isaac you can use it to return vars, cljs.test does this trick as well - you might want to look there for ideas

you could also just filter on the meta at compile time

FYI: I just built ClojureScript app with React directly from NPM and there seem to be a bug, you can read more about it here https://github.com/facebook/react/issues/8887

@roman01la: have you tried your cljs-npm-test under none or simple optimizations?

I get an error which is also of the form of the issue you opened

this._instantiateReactComponent is not a function

@anmonteiro No, I didn’t. But I’m setting up dev build at the moment, so we’ll see

It's not specific to your repo btw, happens whenever you consume React from NPM

ok, I’m doing that on the app at work

No idea why it works in advanced mode

And not in other modes

hihi. can you point me to routing examples in case where I have subcomponents? (menu, submenu) (still learning react)

not sure how to choose which component from root menu to render, when I have subcomponent of the submenu =/

@anmonteiro for some reason I couldn’t even make figwheel see a foreign lib js module which requires stuff from NPM cc @bhauman I’m using [figwheel-sidecar "0.5.9"]

for the infer-externs feature of 1.9.456, It seems failed to infer my project. https://github.com/DogLooksGood/HiccupMacroParser Is there any tips for how to write code that can get correct externs inference?

It seems like the problem is related to this line.

(.setState this# new-state#)

@doglooksgood externs inference isn’t automatic in that way - have you read through the guide?

it works via warning on missing type-hints and uses those type hints to compute externs

http://clojurescript.org/guides/externs

No yet, I'm going to read the guide.

I walk through the guide and start adding type hints. all warning solved except this:

(fn [db]
    (let [f-type (:filter db)
          f (case f-type
              :all identity
              :active (complement :completed?)
              :completed :completed?)]
      (->> (get db :todos)
           (filter f))))

fairly certain that is the case accessing the fqn of the keyword(s)

so safe to ignore

@doglooksgood hrm thanks will take a look at that- but yes safe to ignore

quick question: In :cljsbuild configuration I had to prepend :asset-path "js/compiled/out" with leading "/", otherwise when my server url is not root, I was getting figwheel warning, and it tried to find my js modules within current url subpath. so now my :asset-path is "/js/compiled/out". Am I doing it right?

@leov Yes

(but official documentation says "js/compiled/out", as well as chestnut configuration sets it so? Sure it was corrected before, and I couldn't encountered the problem first?)

@leov No you're good, that's one of the first things I change. I've seen this question a few times and this is the correct answer to your problem

omg thanks. I guess we should update both figwheel and chestnut >_<

(and probably clojurescript wiki/site)

If anyone is interested, I am trying initialize an open source project that facilitates building campaigns for public office in a transparent and less costly manner. There are more details in the #political-canvas channel

https://github.com/dmillett/political-canvas

did anybody else get the cljs.spec conformer errors on 1.9.456 that I mentioned last night?

for reference: I just upgraded to cljs 1.9.456 and my conformer spec started throwing errors [11:04] error is cljs.spec.spec_impl.call(...).call is not a function [11:06]

;; conformer
(defn coll->set [x] 
  (cond
    (set? x) x
    (coll? x) (set x)
    :else :clojure.spec/invalid))

(s/def ::my-spec-coll (s/coll-of ::my-spec :kind (s/conformer coll->set) :into #{}))

seems like there were breaking changes

@jiangts need more information and something more minimal

what you have is not enough to go on

@dnolen hmm, ok. In my actual codebase it’s basically

(defn coll->set [x] 
  (cond
    (set? x) x
    (coll? x) (set x)
    :else :clojure.spec/invalid))

(s/def ::my-spec int?)
(s/def ::my-spec-coll (s/coll-of ::my-spec :kind (s/conformer coll->set) :into #{}))
(s/def ::my-spec-wrapper (s/keys :req-un [::my-spec-coll]))

(s/valid? ::my-spec-wrapper {:my-spec-coll [1 2 3]})

and on the call to s/valid? I get the cljs.spec.spec_impl.call(...).call is not a function error

is that more helpful?

@jiangts I don’t see this issue

@jiangts also :clojure.spec/invalid should be ::s/invalid

aha, didn’t realize that. And I’ll play around a bit more in my repl to see if I can isolate the problem a bit better

yeah - I copied and pasted your code and I don’t see that error

ok so playing in the figwheel REPL

(defn coll->set [x]
  (cond
    (set? x) x
    (coll? x) (set x)
    :else ::s/invalid))

(s/def ::set-ints (s/coll-of int? :kind (s/conformer coll->set) :into #{}))
(s/def ::vec-ints (s/coll-of int? :kind vector? :into []))
(try
  (s/valid? ::set-ints [1 2 3])
  (catch js/Object e
    (js/console.error e)))
(s/valid? ::vec-ints [1 2 3])

prints

TypeError: cljs.spec.spec_impl.call(...).call is not a function
    at eval (eval at figwheel$client$utils$eval_helper (utils.cljs?rel=1485763033743:57), <anonymous>:2:151)
    at cljs.spec.every_impl.cljs$core$IFn$_invoke$arity$4.cljs.spec.t_cljs$spec92144.cljs$spec$Spec$conform_STAR_$arity$2 (spec.cljs?rel=1485763028944:844)
    at cljs$spec$conform_STAR_ (spec.cljs?rel=1485763028944:40)
    at Function.cljs.spec.valid_QMARK_.cljs$core$IFn$_invoke$arity$2 (spec.cljs?rel=1485763028944:357)
    at cljs$spec$valid_QMARK_ (spec.cljs?rel=1485763028944:353)
    at eval (eval at figwheel$client$utils$eval_helper (utils.cljs?rel=1485763033743:57), <anonymous>:1:137)
    at eval (eval at figwheel$client$utils$eval_helper (utils.cljs?rel=1485763033743:57), <anonymous>:9:4)
    at eval (eval at figwheel$client$utils$eval_helper (utils.cljs?rel=1485763033743:57), <anonymous>:17:3)
    at eval (eval at figwheel$client$utils$eval_helper (utils.cljs?rel=1485763033743:57), <anonymous>:22:4)
    at figwheel$client$utils$eval_helper (utils.cljs?rel=1485763033743:57)

@jiangts remove Figwheel from your tests if you’re trying to isolate the bug

try to recreate the problem with the Quick Start uberjar thanks

ok, made a new project from mies template and updated the deps to cljs 1.9.456

and ran the same code as above and got the following:

TypeError: cljs.spec.spec_impl.call(...).call is not a function
    at core.cljs:10
    at cljs.spec.every_impl.cljs$core$IFn$_invoke$arity$4.cljs.spec.t_cljs$spec9434.cljs$spec$Spec$conform_STAR_$arity$2 (spec.cljs:844)
    at cljs$spec$conform_STAR_ (spec.cljs:40)
    at Function.cljs.spec.valid_QMARK_.cljs$core$IFn$_invoke$arity$2 (spec.cljs:357)
    at cljs$spec$valid_QMARK_ (spec.cljs:353)
    at core.cljs:13

the contents of the file are simply

(ns repro.core
  (:require [cljs.spec :as s]))

(defn coll->set [x] 
  (cond
    (set? x) x
    (coll? x) (set x)
    :else ::s/invalid))

(s/def ::set-ints (s/coll-of int? :kind (s/conformer coll->set) :into #{}))
(s/def ::vec-ints (s/coll-of int? :kind vector? :into []))
(try
  (s/valid? ::set-ints [1 2 3]) 
  (catch js/Object e
    (js/console.error e)))
(s/valid? ::vec-ints [1 2 3]) 

@jiangts just do it with the Quick Start please

how do I do that?

that’s the only thing we’ll take for bug reports

https://clojurescript.org/guides/quick-start

so download that cljs jar file and run it like java -cp cljs.jar:src clojure.main release.clj?

sorry I’ve only really used lein + figwheel so far

exactly

when trying to figure out bugs need to remove all variables except ClojureScript

ok. Did that and got the same error:

TypeError: cljs.spec.spec_impl.call(...).call is not a function
    at core.cljs:10
    at cljs.spec.every_impl.cljs$core$IFn$_invoke$arity$4.cljs.spec.t_cljs$spec860.cljs$spec$Spec$conform_STAR_$arity$2 (spec.cljs:844)
    at cljs$spec$conform_STAR_ (spec.cljs:40)
    at Function.cljs.spec.valid_QMARK_.cljs$core$IFn$_invoke$arity$2 (spec.cljs:357)
    at cljs$spec$valid_QMARK_ (spec.cljs:353)
    at core.cljs:13

@jiangts great open an issue with exact steps http://dev.clojure.org/jira/browse/CLJS

and I will take a look later - if I can repro will definitely get resolved by the next release

there are some other things that need fixing that have come up so it won’t be long

awesome, thanks!

Anyone got a suggestion for a library to do selectors on Hiccup? I know of https://github.com/cjohansen/hiccup-find, but it doesn’t work with long form elements like [:div {:class “foo”}], only CSS-selector style like [:div.foo]. I suspect that this functionality has to exist in some library somewhere, even as a support function...