Usually just include big.js, but this wiki describes some of the differences: https://github.com/MikeMcl/big.js/wiki If you're looking for a CLJS example, Fulcro RAD has a namespace worth checking out: https://github.com/fulcrologic/fulcro-rad/blob/develop/src/main/com/fulcrologic/rad/type_support/decimal.cljc

There's js/BigInt that exists in browsers and on Node. And it supports arithmetic just fine, so you don't have to treat it in a special way at least there. But (number? ...) returns false, so it might not work in some cases, and some libraries might not know about it.

=> (def a (js/BigInt "900719925474099100"))
#'cljs.user/a
=> (def b (js/BigInt "900719925474099101"))
#'cljs.user/b
=> (+ a b)
#object[BigInt 1801439850948198201]
=> (number? a)
false

I saw there's a library called bn.js. Is it necessary to be used in ClojureScript? Or js/BigInt is exact what I want?

There are so many libraries in js world

js/BigInt by itself does not need any libraries, it's built-in. But hard to say whether you need something in addition to that because we don't know your exact requirements.

Okay, thank you!

The warning is not from meta but from (A. ...). The text of the warning is rather clear IMO - you have defined a record that expects no arguments (`[]`), and yet you pass one argument when creating it. If you want {:tag 1} to be the metadata of a newly constructed record, then you have to extract that map from under (A. ...) and replace meta with with-meta.

the thing is that when defining a record, the constructor can take 3 additional arguments on top on the positional fields: __meta, __extmap and __hash. Setting the __meta does seem to work despite the warning, I'm properly getting it back when calling meta on the result. If that's discouraged I would except a warning about that instead of the arity.

Ah, I see what you mean. The docstring says that there are two additional arguments though, not 3:

Two constructors will be defined, one taking the designated fields
  followed by a metadata map (nil for none) and an extension field
  map (nil for none), and one taking only the fields (using nil for
  meta and extension fields).

I'm guessing it's this line ? https://github.com/clojure/clojurescript/blob/master/src/main/clojure/cljs/analyzer.cljc#L2558

Yeah, I think so.

Are you using (require ...) from a REPL?

no, in an ns macro

does it matter?

There's a difference between require and :require. The ns form should use only the latter, and its inner vector should not be quoted.

So in the end, should be like

(ns my.ns
  (:require [moment]))

yes, that’s what I’m using. Just for context, I’ve used clojure for several years, but new to clojurescript.

expanding on your snippet

(ns my.ns
  (:require [moment]))

(println moment)

I see in the output folder that the js file is generated with a line that says require("moment") and the bundler webpack is re-run. but still nothing until I refresh the browser

Oh, so you're using some non-trivial setup then, given that webpack is involved. Sorry, no clue then. I use only shadow-cljs and have never had such issues.

I probably can't help, but the next question someone will probably ask you is how you're compiling/building your clojurescript (lein, boot, shadow-cljs, other) and where you're running it (in a browser, in nodejs, other)

and also how you installed it - npm install? yarn add? local? global?

I installed with npm. Compiled with shadow-cljs. It is a react-native project.

I hate to do it as the guy is a saint but you'll probably get a faster and more specific answer in the shadow-cljs slack, but also the best information you'll find without asking someone will be in the shadow-cljs user manual

it sounds like what you've done should work, shadow-cljs respects/integrates NPM and normally finds npm modules, have you killed your cache (in .shadow-cljs) and done a full rebuild?

also can you see the package in your node_modules subdirectory, with a reasonable package.json?

if you look into your .shadow-cljs/builds/<your-build-name> directory, you can see where shadow-cljs is physically moving your dependant packages

that might give you a clue

Why js/require though instead of require?

@U2FRKM4TW that's a better question thanks, js/require works right but if shadow-cljs hasn't bundled the dependency then it won't, and is it the (ns (:require... part that actually prompts shadow-cljs to bundle in the npm package?

I think so - I'm not an expert here. And maybe it's common to use js/require in React Native, which I myself don't use, but you definitely don't do that if you target a browser.

did you reload the app after adding the package? in react-native metro is responsible for loading packages and as such is only available if metro has seen it

if you load the namespace with the require via the REPL first then metro won't see it

and as such the package won't be avialable

Hi @U63D7UXJB when is the files in .shadow-cljs/builds/<your-build-name> directory, generated, and how are the moved? I see they are all named in the format of .cljs.cache.transit.json

Just saw this section: https://shadow-cljs.github.io/docs/UsersGuide.html#_compiler_cache which answers what the files.