Is it possible to use local library in Clojurescript like we can do in Clojure? Will deps like below work?

{:deps
 {time-lib/time-lib {:local/root "../time-lib"}}}

From personal experience it does not seem to work. Just wanted to make sure its doable.

it should. because everything works on a classpath. should be no difference if its in a jar in ~/.m2/repository/... or ../time-lib

Got it working, the directory structure at the local library did not match the namespace. Thanks @dpsutton

glad to hear. simple tools mean that the bugs can't hide too much

https://2020.stateofjs.com/en-US/technologies/javascript-flavors/

How do I get rid of this error message in self-hosted CLJS?

ClojureScript 1.10.597
cljs.user=> (defmacro ctx-fn [ctx-bind & body]
       #_=>   `(with-meta (fn [~ctx-bind]
       #_=>                 ~@body)
       #_=>      {:sci.impl/op utils/evaluate}))
#'cljs.user/ctx-fn
cljs.user=> (ctx-fn ctx nil)
                    ^
WARNING: Use of undeclared Var cljs.user/ctx at line 1

cljs.user=> (defmacro foo [x] `(fn [~x]))
#'cljs.user/foo
cljs.user=> (foo x)
                 ^
WARNING: Use of undeclared Var cljs.user/x at line 1
(cljs.core/fn [nil])

:thinking_face:

can't define macros in CLJS like that. not sure about self hosted but the way above will just call it like a function

right

I did put a #?(:cljs (:require-macros [sci.impl.analyzer :refer [ctx-fn]])) in my ns decl

not actually sure you can do this in the REPL

don't care about the REPL, I just wanted to get rid of the > WARNING: Use of undeclared Var cljs.user/ctx at line 1 warning in my code for self-hosted during load time

it looks to me like the macro is being called as a function rather than a macro but thats where my self-host knowledge ends 😉

yet it does work in self-hosted when called

I used macrovich to solve similar issues, but this time around it doesn't seem to work

ah well, I got rid of the macro. less macros, less problems

Found a subtlety of CLJS this AM that I wanted to run by y’all - == calls out to -equiv, but it doesn’t seem to pick up on any NEW -equiv implementations. maybe it is bound at compile time? The var #'== on the other hand does do the right thing when applied. for example:

(deftype Wrapper [x]
  IEquiv
  (-equiv [_ that]
    (-equiv x that)))

(let [x (Wrapper. 10)]
  (== x 10)     ;;=> false
  (#'== x 10))  ;;=> true

@sritchie09 my guess would be that because of the 10 number a compiler optimization kicks in to use JS ===. guess that doesn't account for custom -equiv impls comparing to numbers

yup, exactly, which is what I was installing

Ratio support, js/BigInt for two clearcut cases

and for a slightly funkier one, an automatic differentiation implementation that needs to compare “dual numbers” to normal numbers using just the primal-part, not the tangent-part

@thheller actually, if I throw a println in the -equiv impl and call (== x x) it looks like my custom -equiv never gets called, even if no number is passed as an arg

well = is equiv

isnt' == just a numerics things?

(defn ^boolean ==
  "Returns non-nil if nums all have the equivalent
  value, otherwise false. Behavior on non nums is
  undefined."
  ([x] true)
  ([x y] (-equiv x y))
  ([x y & more]
   (if (== x y)
     (if (next more)
       (recur y (first more) (next more))
       (== y (first more)))
     false)))

not according to its implementation

Behavior on non nums is undefined.

oh 🙂

its probably a function with a macro inlining stuff

(core/defmacro ^::ana/numeric ==
  ([x] true)
  ([x y] (bool-expr (core/list 'js* "(~{} === ~{})" x y)))
  ([x y & more] `(and (== ~x ~y) (== ~y ~@more))))

as expected uses JS === directly

okay, got it

thanks for the help sleuthing!

another spot that seemed like it might beg a patch was compare, which has this branch for numbers:

(number? x) (if (number? y)
                 (garray/defaultCompare x y)
                 (throw (js/Error. (str "Cannot compare " x " to " y))))

garray/defaultCompare works great between numbers and goog.math.Long or goog.math.Integer , but we can’t get there with this branch

int? and pos-int? check for these types, so perhaps the right side should check for them here too?

(unfortunately I still had to make a new compare implementation in sicmutils, because with the special case for number? there is no way to get a number to compare to a js/BigInt instance. These also work well with garray/defaultCompare, but we don’t speak js/BigInt anywhere else so I wouldn’t expect that to come in. If the number? case were removed and fell through to a new -equiv implementation for numbers, however, then I could override that.

might be worth taking this to http://ask.clojure.org or direct jia issue. stuff tends to get lost in slack.

I didn't find anything about that so I don't think so, but is it possible to obtain a list of original names -> munged names after advanced compilation?