https://clojure.org/reference/metadata

I don't think it's possible. Why not use a comment?

the section Metadata Reader Macros has a method to add docs

> ^{:doc "How it works!"} - adds the metadata map to the metadata of the next value read

The difference between doc strings and comments is you can use doc in the repl to read doc strings

You cannot use doc to read the doc string on something that is not stored in a var (a top level def)

Just write a comment

would adding it to the metadata and then assigning it to a car work?

I don't believe so, if I recall def copies metadata from the symbol to the var, but not from the value

ahhh ok

boolean

if you are used to 0 being false, in clojure it is treated logically true. I trip up on it from time to time :) - https://clojuredocs.org/clojure.core/boolean#example-54f09be3e4b0b716de7a652d

true?

That would return false for (true? 1).

Thanks, @U2FRKM4TW, I guess I misunderstood the question.

debug might have many meanings: special environment, nuances of the state, explicit intent. So you have to have your way to say “execute this form only when debug-mode”

you can also do it with an environment variable

java System properties are particularly suitable: • can be set from Lein / t.deps • can be toggled at runtime • are not erased by Clojure code reloading

(clj-otel does not do Metrics, only Traces. We want metrics too, so a "quick Java interop" and a few hours later):

This is the failure:

;; com.github.steffan-westcott/clj-otel-api {:mvn/version "0.1.4"}
(ns journal.2022-07-15-interop-for-metrics
  (:require
    [clojure.reflect])
  (:import
    (io.opentelemetry.api GlobalOpenTelemetry OpenTelemetry)))

(def global ^OpenTelemetry  (steffan-westcott.clj-otel.api.otel/get-global-otel!))
(def meterBuilderBuilder (.build (.meterBuilder global "wot")))
(def longBuilder (.counterBuilder meterBuilderBuilder "wotwot"))
(def by-method-name (group-by :name (:members (clojure.reflect/reflect longBuilder))))
(by-method-name 'setUnit)
; =>
;[#clojure.reflect.Method{:name setUnit,
;                         :return-type io.opentelemetry.api.metrics.LongCounterBuilder,
;                         :declaring-class io.opentelemetry.sdk.metrics.SdkLongCounter$Builder,
;                         :parameter-types [java.lang.String],
;                         :exception-types [],
;                         :flags #{:public :bridge :synthetic}}]
(.setUnit longBuilder "unit")
; => No matching method setUnit found taking 1 args for class io.opentelemetry.sdk.metrics.SdkLongCounter$Builder

I don't think we have this problem: https://stackoverflow.com/questions/58968216/cannot-call-particular-java-method-from-clojure/58969526#58969526

@U2C6SPLDS did you get somewhere with Metrics? And also thank you, loving clj-otel for traces!

If you can find the method with reflection, can you invoke it reflectively?

(.setUnit ^io.opentelemetry.api.metrics.LongCounterBuilder longBuilder "unit")

(.setUnit longBuilder "unit") works directly for me, too

looks like it fails on java8 but succeeds on 11

it = (.setUnit longBuilder "unit")

correction: it works on both

Can confirm the type hint works. Thank you!

Without it, not: $ java --version openjdk 11.0.11 2021-04-20 OpenJDK Runtime Environment AdoptOpenJDK-11.0.11+9 (build 11.0.11+9)

I'm still unclear on why it failed for you

In general though, trying interop calls with objs deffed to vars is challenging because the Vars are hinted to Object by default

all interop calls will reflect if the compiler doesn't know anything more specific than Object

Ok, reproducing is a weird one: • I had to add more dependencies • And move to def binds. In a let inside a fn it all still worked.

https://github.com/danieroux/builder-bridge-method-repro (fail to reproduce is the first commit)

(that repository now with a README that shows what I see)

@U9E8C7QRJ Sorry for the late reply, I'm on holiday (vacation). I'm glad you are enjoying the traces support in clj-otel 🙂 As noted in the README, there's no support yet for metrics or logs. I plan to start work on this when I return. Feel free to drop in #clj-otel

Encountered the same here, along with the author: https://ask.clojure.org/index.php/12362/environment-variables-interactive-process-processbuilder (.put env "a" "b") fails on openjdk version "11.0.16" but (.put ^java.util.Map env "a" "b") works. Found a similar ticket: https://clojure.atlassian.net/browse/CLJ-1243

merge only?

nice idea, but google doesn’t turn up anything

I guess CRDT

but that again is getting pretty technical

if you want people to be able to google it, you have to use the name for it, which is going to be technical, if you want to describe it in plain language it is going to be generic and not something you can google

indeed

Thanks, I'll take a look. I've heard of it but haven't familiarized myself with it. Interesting to see that it was your post back in 2016 that originally got the ball rolling on adding this functionality to core!

Here's a demo using SCI + nextjournal clojure-mode: https://nextjournal.github.io/clojure-mode/

You can find more projects using SCI on the SCI README.md in github, to see if there's anything like what you want to build. Feel free to post questions in #sci if you're interested

I forgot why that doesn't work; but what I normally do is wrap the interop in a function: (map #(.toUpperCase %) ["saku" "monu"])

(.toUpperCase x) is syntax sugar for (. x (toUpperCase))

.toUpperCase itself is not a function and isn't a symbol that evaluates to some function object that can be invoked, so cannot be passed as an argument

Another case is (map Long/valueOf ["1"])

same thing, just once is a static method one is an instance method

neither is reified as a first class value by default on the jvm

(Long/valueOf x) is syntax sugar for (. Long (valueOf x))

this is actually something we've done some work on for 1.12, but not sure it will survive

https://clojure.org/reference/java_interop#_the_dot_special_form the . and new special forms don't get used a ton since the syntax sugar was added on top of them, but they can be easier to generate in macros

"this" = using Java static methods as clojure fns

To me the static one is more surprising, since you could otherwise think of clojure functions kind of like static functions

it's more complicated than you think due to arities and overloads

Then how come these are valid?

(.toUpperCase "monu")

(Long/valueOf "123")

those get rewritten to the . as mentioned above

(and true) is valid, but (map and [true]) is not

the reason it doesn't work for add is and is a macro, which is expanded away at compile time, and sort of doesn't have a runtime value

.toUpperCase and Long/valueOf are similar

So in general no macro will not work inside map filter etc? Is this right?

you cannot take the value of a macro

similar to calling java methods, with a macro if you wrap it in a function, you can take the value of that

it's like in say some language with a C like syntax (js, java, C, C++, C#, etc) you cannot invoke && the and operator as function/method/whatever

js might be the best example because it has first class functions

[].reduce(function (x, y) { return x && y}, true) is fine, but [].reduce(&&, true) is not