@george.w.singer: I’d try :refer-macros as in [cljs.core.async :refer-macros [put! chan <! >!]]

@mfikes: thanks for the suggestion, but unfortunately it didn't work.

@george.w.singer: Hrm. Well. Hopefully this helps, for inspiration as to something that works: https://github.com/clojure/clojurescript/blob/c3899acf797eb6779c53b313f5606c5018360b83/src/main/clojure/cljs/repl.cljc#L778-L779

@george.w.singer: I also wrote about it here, but I don’t think it adds anything to what you have already tried, honestly: http://blog.fikesfarm.com/posts/2015-12-02-clojurescript-repl-auto-require.html

Wow -- I came across that before asking on Slack from Googling

Crazy coincidence lol

@george.w.singer: Oh, wait. The macros for core.async are in a different ns right?

yes

cljs.core.async.macros

Dang. Well, that’s why it doesn’t work. Seems like you might need a :repl-require-macros option added. Hmm.

I just tried :repl-require-macros '[[cljs.core.async.macros :refer-macros [go]]] and that also didn't work

Is there a way to get the previous item(s) in a seq when iterating over it?

No. If so, lists (which are singly-linked) would not be usable as a seq.

@puppybits: that would be a reduce, where you collect what you need for the "next" step I guess

just an idea

You can iterate over the list and the list offset by 1

(def items '(1 2 3 4 5))

(map #(vector %1 %2) (cons nil items) items)

=> ([nil 1] [1 2] [2 3] [3 4] [4 5])

Anybody pretty familiar with re-frame that could give me some assistance? I've been stuck for a while on this: I have a reaction that I expect to resolve to a vector, but it resolves to cljs.core.ValSeq which has been super troublesome and difficult to debug

cljs.core.ValSeq doesn't function inside of a for loop I believe

Let me giiist...

https://gist.github.com/bpicolo/4de274cc4c6f49cc72d1

My actual code unfortunately appears nowhere in the stack trace

=/

ahh, well actually it does in the for loop

what does your subscription look like?

@oli that's in the gist

also added what @tags looks like at that point

I've been stuck on this bug for quite a long time, not really sure how to debug what's going on because I don't really know what to expect either

(Very new to cljs)

i mean the register-sub decl.

like

uhh

(rf/register-sub :company (fn [db] (reaction (-> @db :company))))

I do not have one

should I?

Maybe that's what I missed

yeah - your handler updates the db

correct

your subscription listens for changes

ahh

oh I do have that uhh

but let me see if it makes sense

that's good... otherwise i'd have no idea where you're getting values from

@oli added that too : )

try getting rid of vals...

trying it

...wat

@oli that worked perfectly

but uhh

https://github.com/Day8/re-frame/blob/master/examples/todomvc/src/todomvc/subs.cljs#L25

I was just copying the todomvc ;_;

I get it now

https://clojuredocs.org/clojure.core/vals

It's because they are k:v

aha

@oli you are the best ❤️

pity that that function works at all on a vector

no prob. meanwhile I have this joy...

_ is not ISeqable

Isn't transpiling great

@bpicolo: Looking at your Gist:

Handlers have parameters of the format [db [event-name & args]]. So unless you use the trim-v middleware, your parameters needs to add the event name.

i.e., [db [event-name text]] or [db [_ text]] in the case of your add-tag handler.

Or, you could just use the trim-v middleware.

@cky: I was just hitting that issue 😄

https://github.com/Day8/re-frame/blob/master/examples/todomvc/src/todomvc/handlers.cljs

does that mean this is using middleware?

ahh

Most of the handlers there is using todo-middleware, which includes trim-v.

aha

cky do you recommend trim-v or explicit

I'll make it explicit for now I suppose

Sure, whatever works for you.

now to figure out how to get event.target.value

heh

aha

@cky where should I handle cleanup for events?

like uhh

actually, in this case it's trivial

nvm ^_^

all works now, marvelous

is it possible to partially apply a javascript object method with its this?

e.g., from bunyan lib. want to do: (partial (.-info log) log)

[or something like it]

[also if I'm able to do that first step, can apply then operate on it? [bunyan logging methods want the args expanded inline but I will have a vector of them]

Does anyone know if Onyx works on Clojurescript? I saw some archived posts where that was mentioned as a possibility.

Quick couple of questions: I want to use https://github.com/google/closure-compiler/blob/master/contrib/externs/maps/google_maps_api_v3.js in a luminous project. I've dropped it into `/target/cljsbuild/public/js/. Is that the right place for it and how do I properly import that into my core.cljs file to use it?

@bwstearns: You’re usually better off getting the dependency from clojars if you can...

https://clojars.org/rowtr/google-maps-web-api

https://clojars.org/cljsjs/google-maps

Sorry, the second is what you wanted… not a wrapper, just the javascript in the right place with the ‘externs'

http://cljsjs.github.io/

@smw You're the best! thanks. New to clojure/cljs. I saw cljsjs earlier but I misunderstood what it was.

I’m still pretty new too, hope it helps.

@cjmurphy: it doesn't work on clojurescript. There's nothing technically stopping it, but it would require a pretty big refactoring.

has anyone run into something like: Uncaught TypeError: Cannot read property 'mount_components' of undefined? I think it's saying that myapp.core.mount_components is undefined, but it wasn't a while ago.

does "extern only" mean that you don't explicitly import google maps cljsjs in the .cljs file where you're using it?

Any hints on how to define how a type made with deftype prints with pr-str? In Clojure I am implementing a method on the print-method multimethod, but it seems to be different in Clojurescript unfortunately.

@casperc: you must implement IPrintWithWriter https://funcool.github.io/clojurescript-unraveled/#printing

if you want to customize printing for your type

@dialelo: Nice thanks alot. Cool book too, didn’t know about that one.

you're welcome! thanks for the kind words, we have to update some stuff and finish the project but we did it all in our spare time

Wierd that it isn’t aligned between Clojure and Clojurescript though

yep, is one of the differences between Clojure and ClojureScript

Ah, you’re one of the writers too?

yes, me and @niwinz started the project and lots of people have contributed

@bwstearns: Google naps api files have to be loaded from their js server and because of that the cljsjs package can only provide the externs

I think you still need to require cljsjs/Google-maps though to ensure the externs are loaded for compilation

@dialelo: @niwinz wondering if you have investigated goog.Promise as underlying promise library for promesa... Would be curious to hear why you went with bluebird

@martinklepsch: is mainly explained here https://github.com/funcool/promesa/issues/11

is not about goog.Promise but the reasons are pretty the same

however, it shouldn't be too difficult to adapt goog.Promise for using it with promesa's API https://github.com/funcool/promesa/blob/master/src/promesa/protocols.cljc

question about Promesa for Clojure. When would I use that instead of Clojure's own promise, when developing in Clojure (not Clojurescript)?

@daan them ara pretty different and has different purposes. But the main difference is that the clojure promise and future has blocking api

promesa in CLJ uses CompletableFuture, that is highly composable and has proper async api.

it also gives you more control on how they run; you can provide your own j.u.c.Executor for running callbacks

is promesa compatible with manifold by any chance? I recall manifold extended CompletableFutures

Is not directly compatible, but is easy to make the promesa api work in the same way with manifold deferreds

@niwinz: when you say "composable", do you mean you can chain them in a way? Monad-like? That is what I sometimes miss in Clojure coming from Scala. To be able to map over promises and apply functions over the results without actually having to wait for the result.

@daan, yes

you can chain them, apply function without waiting the result.

in fact, CompletableFuture has monadic api.

nice!

@dialelo: @niwinz thanks for the pointers will give it a try when I'll use promesia

you're welcome, happy to hear any feedback

Has anyone played with core.match and ClojureScript ? I would like to know if it increases the size of the output JS files. My understanding after reading the wiki it that it will generate efficient conditions code via a macro and not embed some kind of engine (therefore a reasonable size increase). Is that correct ?

I wonder, I'm using cljsjs/chartist, and the extern file only lists the most used functions. Would there be a specific reason the whole Chartist api isn't included, or is it just likely that whoever wrote it only included what they personally needed?

@nha should be fine but be careful when using it inside core.async go blocks, that might lead to an explosion in size

besides that yes it is all done in a macro, no engine

@thheller: Thanks do you have a link as to why with core async it changes ? (just curious)

not really related to core.match just a general thing with core.async in cljs

since core.match might generate a bunch of nested if and such

core.async doesn't deal well with that

Oh ok - so match should be fine for me I am worrying about core async though now 😕

no link, just a general obversation I made over the years 😉

Ok thanks for sharing

just keep your go blocks small

move as much code into functions as you can

keep only the concurrency related stuff in the go

ok thanks for the tip

there is very little information about the size of libraries in cljs (I realise the closure compiler complicates things of course)

don't worry about it for most cases 😛

Do you happen to have a workflow that allows to monitor the size ?

size doesn't really matter too much due to the closure compiler

In my case I do care since part of it will be distributed as a js lib

things that aren't used get removed

not like normal js world 😉

right I get that - I just think that JS devs are more sensitive to size than we are generally

aha right well that's the thing for me ^^

believe me I'm a size nazi

I optimize more than I probably should

but cljs + closure is just hard to beat

it is close to perfect

Really good to know (I heard about rollup.js as a new contender for size by the way)

nah still won't come close

they have to deal with too much undefined crap in the js world

and nobody writes the strict style the closure compiler requires by hand

cljs does that for us 😉

Perfect even if it wasn't the case cljs brings other amawing things to the table

Thanks again !

yeah it is pretty perfect

> and nobody writes the strict style the closure compiler requires by hand well, some google devs do, for example the chrome/webkit devtools team does

well yes the closure lib is huge as well

but still .. even most of the new stuff coming from google isn't written in closure style

Is there something like https://github.com/technomancy/slamhound for ClojureScript?

Hi folks! I have a (maybe somewhat dumb) question about Figwheel. When I run lein figwheel in my terminal, it knows it has to pickup the dev build. How does it know that? And what happens when I add another build that uses Figwheel?

you can specify builds via command-line: lein figwheel build1 build2 … (this works the same as lein cljsbuild), other option is to configure it via figwheel’s config in project.clj

I understand that, but I'm wondering how Figwheel understand to pick up my dev profile when I don't specify a profile on the command line at all

Is dev the first in your lein project file?

my hguess would be that it picks all?

@cjmurphy: yes

Is there a recommended / idiomatic way of cancelling timeouts?

@jannis If you’re using js/setTimeout, pass the returned value to js/clearTimeout.

@exupero: Good point, why not just use the JS timeout instead of core.async's.

Ah, didn’t realize you were talking about core.async. As far as I know you just use alts to circumvent a timeout and let it run out without effect.

With core.async you could use a little component system, might be overkill thou.

.. wish I knew how to make it take up less space 😞

@cjmurphy: If you create it as a snippet, it’ll hide all but the first bit. You can edit the above post.

Yeah, a component is definitely too much overhead 😉

Background: this is a function that takes a time in ms and a callback and returns a function that collects callback arguments so the callback is only called once every X ms with all arguments provided in the current time window.

@jannis sounds like you want to core.async alt! over one timeout and one channel that collects the arguments

@cjmurphy: (async/close! poison-ch) instead of the put! would be more idiomatic

@thheller: Great thanks - I thought I might luck in and get some good feedback...

(go-loop [count 0]
  (alt!
    (timeout 1000)
    ([_]
       (println "In timer at " count)
       (recur (inc count)))
    poison-ch
    ([v]
       (println "[printer-component] stopping"))))

personally I like alt! in this case

@thheller: That creates a loop that keeps running. What I am trying to go for is a one-shot-reschedulable argument-collecting timeout. If you don't call the collecting function at all, it never runs. If you call it once, it schedules the timeout once. If you call it twice with <X ms in between, it reschedules the timeout with the second call. If you call it twice with >=X ms in between, it schedules the timeout once for the first call and then a second time for the second call. Does that make sense?

That loop was just improving the code snippet a bit.

you could use a transducer if you want to get really fancy 😉

I sure do want to get fancy 😉

but it would require using core.async and channels all the way

doesn't sound like you want to do that

ie. instead of calling a function you put something on a channel

I'd be ok with it if it doesn't involve a permanent loop (unless it sits and waits most of the time) and if timeouts can be pushed back / rescheduled.

oh wait you can't use a transducer

well it'd be hacky as hell if you did

so better not

@darwin Any idea if chromex can be used with a figwheel/boot-reload-like live updating dev workflow?

I have a question for integration testing with Phantom, in particular I would like to test reading (async Ajax fetching) served files

Now I am using lein-doo but I saw that Phantom can act as webserver

what is a good way to achieve this?

@richiardiandrea I'd say that's not really the user case for phantom's built in server

I'd use existing Clojure HTTP tooling

@pandeiro: thanks for answering, so you would boot up a stand alone server for that right?

Yeah definitely

@pandeiro: chromex works with figwheel, except for content scripts

See project.clj in chromex-sample

Cool, thanks @darwin

is it possible to partially apply a javascript object method with its this? e.g., from bunyan lib. want to do: (partial (.-info log) log) [or something like it] [also if I'm able to do that first step, can apply then operate on it? [bunyan logging methods want the args expanded inline but I will have a vector of them]

@jonathan: you can use javascript’s .call or .apply to pass this

(.call (.-info log) log …) or (.apply (.-info log) log (to-array extra-args))

@johanatan: ^

@darwin: sweet! thx!

so, the short answer is: ClojureScript's partial and apply only operate with ClojureScript functions?

@johanatan: You can also use it with JS functions that don’t care about the value of this. 😉

ahh, ok

@darwin: @cky: here's the final version (pretty handy tool for anyone using bunyan): https://gist.github.com/johanatan/2e6540e9e3e198bac875

in Javascript, 46245 << 16 = -1264254976

in Clojure, (bit-shift-left 46245 16) = 3030712320

Is this related to JavaScript number precision being only 53 bits?

anyone know how I can get the same result in Clojure?

FWIW, the solution is: (unchecked-int (bit-shift-left 46245 16))