I'm surprised/confused by the source argument to cljs.build.api/build. It takes one source path but I have many. In cljs.main it's set by resolving the source path of the :main namespace. I can do this although that's different between dev & prod environments which makes it an extra thing to twiddle... Is it really needed? How is it used?

Ah, you can pass a vector. The doc string An optional source parameter may be provided. and the cljs.main example led me to conclude that wasn't the case. source needs to implement -find-sources

(defprotocol Compilable
  (-compile [this opts] "Returns one or more IJavaScripts.")
  (-find-sources [this opts] "Returns one or more IJavascripts, without compiling them."))

There are other examples in the cljs.build.api where a compilable is passed in. The docstring for them is more sensible to me, for example:

compile
function
Usage: (compile opts compilable)
       (compile state opts compilable)
Given a Compilable, compile it and return an IJavaScript.

@bhauman May I please contact you via DM? I want to start contributing to figwheel/core and need some help in getting started.

@bhauman I isolated the problem and I am on my way to fix it 🙂 The problem is that figwheel_server.clj tries to parse the data and in the data there is #js literal. My question is how can I parse it to map when using clojure?

Imo I should slice the string to not contain #js but maybe you got better solution.

Hello, I'm using core.async cljs and can't figure out why the x function blocks when it tries to put true in the txt-chan

(defn x
  [txt-chan]
  (go
    (if true
      (do (prn "a")
          (>! txt-chan "true"))
      (>! txt-chan "fail"))))

(defn x-caller
  []
  (go (let [y (<! (x (chan)))]
        (prn y))))

hi guys

when i try this i got the error:

TypeError: (intermediate value).ih is undefined

what is it meaning?

@abdullahibra that is about missing externs and the hide fn call getting renamed

@thheller what actually i'm trying to do is using dateDropper plugin within my app: https://felicegattuso.com/projects/datedropper/

i got jquery hide function working properly now

that's the code

for using datedroppper which is jquery plugin

can anybody help me in this?

by trying jquery hide function, i'm sure jquery is loaded successfully globally

@thheller can you help ?

@abdullahibra what do you need help with?

i'm trying to use dateDropper inside my code but failed

and what i have done so far is: https://clojurians.slack.com/files/U145X0BPE/FC71H6JBD/-.js

@lilactown that's what i have tried

do you have any suggestions ?

remove the js/ in front of dateDropper

#(.dateDropper (js/$ "hello"))

TypeError: $(...).hh is not a function

that's what i got

but .hide works?

yeah

it's working correctly

but not working with .dateDropper

do you have externs inference on?

yeah

:foreign-libs [{:file "resources/public/js/date/datedropper.min.js" :provides ["ui.datedropper"]}]

then using (:require [ui.datedropper]) in the ns i want to use functions at

https://felicegattuso.com/projects/datedropper/

how are you including jquery in your project?

globally

<script src="https://code.jquery.com/jquery-3.3.1.min.js"></script>

in base.html

so it's available using js/$

hm. I’m not sure why it’s fine with (.hide ...) but not (.dateDropper ...)

add (set! *warn-on-infer* true) to the top of your file

WARNING: Adding extern to Object for property dateDropper due to ambiguous expression (. (js/$ "#hello") dateDropper) at line 42

referred line: [:input {:placeholder "input..." :on-change #(.dateDropper (js/$ "#hello"))}]

strange that it says i’ts adding the extern but when it compiles it isn’t right.

when you change it to .hide does it give the same error?

anyway, the solution is to write an externs file for datedropper

WARNING: Adding extern to Object for property hide due to ambiguous expression (. (js/$ "#hello") hide) at line 42

and it's working

what do you mean with externs file?

an externs file helps the Closure compiler know what external JS you have in your project so it can optimize correctly

https://clojurescript.org/guides/externs

so it can work without optimization true?

your current code should work without optimizations

ah ok

i'll try dev mode now

@bhauman I am here if you want to discuss further that figwheel parsing problem. 🙂

I replied to the ticket

and you did as well i see

so if the problem is a problem with reading a clojure string you don't want to adhoc change the string

If you prefer communicating via Github then I don't mind 🙂

you want to fix the whats reading the string

Yes you got right.

In this case you want to make sure that #js tags are not causing things to blow up

but still the string that is in the ticket is not a proper clojure string

"{:value {#js {:prop1 blabla}}"

because if you remove the #js you are left with a map with one item in it

Did not want to overhelm the ticket with to much information

oh ok

so "{:value #js {:prop1 blabla}}"

https://github.com/bhauman/lein-figwheel/issues/711

I edited my comment

So as you can see it's not a problem with the #js

it's a problem that the parser tries to evaluate (or read ((clojure noob here, don't punch plx))) the #object #js as reading macros

so it is a reader problem because its happening in the reader

You mean edn/read-string

https://clojure.org/reference/reader

the reader is the clojure parser

yea but how the Clojure parser can know what #js means. It means something in Clojurescript environment not in Clojure env

so on that page of particular interest is the tagged literal section

and in that section is describe the *default-data-reader-fn*

and the default data reader fn

Ok sorry let me read that

will handle any missing tagged literals

@kwcharllie379 you also don't seemed to have responded to me stating that figwheel.main has the exact same edn/read-string and so I'm very suspicious about what is going on here

Let me check that again

https://github.com/bhauman/figwheel-repl/blob/master/src/figwheel/repl.cljc#L693

we should move this conversation to #lein-figwheel

Ok @bhauman

I am trying to see what it would look like to write CLJS tests with Jest. The problem is that Jest defines global functions. These globals are only available are runtime so there is no easy way to import them. Is it possible to compile my CLJS without importing the actual modules, perhaps they are mocked in some way, to get through the compilation step? The idea being that when these tests are run, I will have access to the globals because Jest will provide them.

Does js/someGlobal works?

Yup, it totally does. I have no idea why I was thinking that the compiler would try to see if that global exists and just fail because its not there. Thanks!

No problem 😛

what http-lib do you guys usually use for Cljs? I'm thinking about using https://github.com/r0man/cljs-http for my project

How would you guys handle it if you had an array of values and wanted to make a network request for each one?

trying to use reduce or for but all my attemps are just a fail

@urbanslug Assuming your network requests are async, you’d probably want to return an array of promises or core.async channels onto which the responses are placed, and then turn that into a collection of resolved values.

yes they're async

I don’t know if there is a function already in core.async like this, but if you have a seq of channels you could use something like

(defn take-chans [chans]
  (go
    (loop [results []
           chans   chans]
      (if (seq chans)
        (recur (conj results (<! (first chans))) (rest chans))
        results))))

How you form the seq of channels is another half of the problem, depending on your network request lib, but you’d have it put the responses onto each of a collection of channels.

This is a fairly naïve way to do it, I suspect, but it gets you in the right direction of something that works.

One sec

@urbanslug Try

(go (prn (<! (take-chans (map mock-request [1 2 3])))))

This returns a vector of the values produced by mock-request, so your mock-request implementation could be revised to simply put the value on the channel. (It appears you have a vector containing a value, so that vector wrapper could be removed, essentially.)

I suppose that what I posted here is lazy, do you could wrap the map with doall or use mapv (to force requests to be initiated eagerly)

I’m curious if there is a better solution than the take-chans posted above, but it works.

hmmm I tried it out with the mock and it works. with your take chans returning a single channel of requests

I wonder how this handles failure and retries

:thinking_face:

Yeah, none of the above has any robustness built into it.

🙂

I would expect that people have ran into this kind of problem in the past

Hmm it depends on your problem.

Do you run the code in backend environment (Node.js)

or in the browser?

browser

Do you prefer to handle your errors in a single place

or you would rather to handle the error in the place where you send the request?

what?

you mean client side?

Yes. For instance you can take the value from the channel check whether it's an error and then pass it to your error handler. But if you prefer some kind of FRP then promesa might be a good choice https://github.com/funcool/promesa.

Thanks @kwcharllie379 I'll eval both options

I think that if we could take a look into the real sample of code then we would give you better and more accurate advices.

if I create an array of promises though it's hard to handle errors because that means looking through an array of results

ya I just need to take a break and get back to it