the tooling figwheel depends on (clojure.tools.namepsace, I think) has decided you have a circular dependency in your namespaces, which clojure doesn't allow either, but clojure isn't as good at detecting it

So I'm requiring something circularly I guess?

yes a -> a, or a-> b -> a or a-> b -> ... -> a

I haven't noticed it in my :requires, can you help me find it from the files in question? it's just 3 files

repl

(ns voke.repl
  (:use voke.handler
        figwheel-sidecar.repl-api
        ring.server.standalone
        [ring.middleware file-info file]))

(ns voke.middleware
  (:require [ring.middleware.defaults :refer [site-defaults wrap-defaults api-defaults]]
            [prone.middleware :refer [wrap-exceptions]]
            [ring.middleware.params :refer [wrap-params]]
            [ring.middleware.reload :refer [wrap-reload]]
            [ring.middleware.json :refer [wrap-json-body wrap-json-response wrap-json-params]]
            [ring.middleware.keyword-params :refer [wrap-keyword-params]]
            [clojure.string :as string]
            [clojure.pprint :as pprint]))

handler

(ns voke.handler
  (:require [compojure.core :refer [GET defroutes routes POST context]]
            [ring.util.response :refer [response]]
            [compojure.route :refer [not-found resources]]
            [hiccup.page :refer [include-js include-css html5]]
            [voke.middleware :refer [wrap-api wrap-site]]
            [config.core :refer [env]]
            [voke.api.v1.utils :refer [defresponse]]))

oh, it isn't figwheel, it is wrap-reload

oh?

what is on (handler.clj:63:15)

(-> #'site ;; line 63 below (wrap-site))

It's where my routes are declared

(def app (routes (-> #'site (wrap-site)) (-> #'api (wrap-api))))

ah, ring reload doesn't use tools.namespace of course, but uses ns-tracker

what's that mean for me?

dunno, I haven't used ns-tracker so I have less of an idea what is going on

what are you doing where you see that exception happen?

well shit

It happens when I try to (start-server)

have you restarted your repl?

Many times

and my editor

it looks like it scans the entire src/ directory by default, so if you have any other files in there, even unused, they could trigger this

aw, shit

is this a known bug in clojure 1.9.0? it seems if :clojure.spec.alpha/invalid is within the body of certain clojure macros, everything blows up... for example, using clojure.test/is:

(is (= ::s/invalid (s/conform string? "1")))
=>
 Call to clojure.core/let did not conform to spec: In: [0 3] val:
  :clojure.spec.alpha/invalid fails spec: :clojure.core.specs.alpha/bindings
  at: [:args :bindings :init-expr] predicate: any?

(let [conformed? #(not= ::s/invalid? %)]
 (is (conformeded? (s/conform string? "1"))))
=> false

any refactor tips?

This macro got out of hand quick

the idea was to take some java pojo and call (pull-keys obj [:id [:name :getName]]) and produce the map {:id (.id obj) :name (.getName obj)}

when I started to add logic for doing this recursively I ended up with that letfn

and there still isnt logic for nested renaming

then I had to deal with some vavr Options so I figured I could provide a function to be called for everything extracted

that worked pretty well, but I definitely feel like im committing design sin

have you looked at bean?

it does something similar to this

yeah, unfortunately it doesnt seem to work with my objects

because?

Im consuming classes produced from https://immutables.github.io/

So im not quite sure why not

but I also have the good fortune to not fully comprehend what a bean is

good fortune inded

looks like that lib can produce different code styles - stuff like name() is not the “standard java bean” style and won’t be picked up by bean, but the getName() style should be

not sure if you can ensure you’re using that style

https://github.com/arohner/clj-wallhack/blob/master/src/wall/hack.clj#L38

I could I suppose but Im kinda torn

I have 2 client libs I wrote for two of the the apis we talk to

And my team writes in java primarily

so .name() and the immutables library make working with that data on the java side fairly pleasant

but at the same time, when I get the opportunity to work on something by myself, its hard to use clojure because I have to undo that strict typing to get idiomatic maps

not to mention stuff like Option and Try which are in there, which make for some clunky code on the clojure end

I almost want to make a .untyped() endpoint just for me and use cheshire or something

well, not trying to talk you out of it :)

it’s fun to get stuff like this working too

Its kinda ironic really

the more functional the java code, the harder it is to work with in clojure

I wonder if there's a way to tie the immutables library into the clojure protocols to make it play nicer

or would it even be worth it? It's been a while since i've done java interop stuff, let alone tying into clojure's core protocls

immutables seems like a generator, seems like you could generate the Clojure implementation of the interface and that would be ideal

a defrecord that implemented the interface and also exposed data as keyword accessors for example

That seems like an interesting road to go down

My instinct would be to add some code that just straight up converts the immutable into a clojure map

unfortunately that would mean writing a good amount of java code in an area of java I dont really understand

generating the clojure code is also an option, but I would probably end up writing a completely separate code generator that just reads the annotations

Not sure which would make the least hacky code

with me behind the wheel that is

Hi, I was looking for a thread-safe version of with-redefs, it seems to be working for me in a toy example when I use with-local-vars, are there any caveats here? Original

(defn ten [] 10)
(def x (doall (pmap (fn [_] (with-redefs [ten (constantly nil)]
                              (ten))) (range 20 100))))
(ten) ;; gives nil

(defn ten [] 10)
(def x (doall
        (pmap
         (fn [_] (with-local-vars [ten (constantly nil)]
                   (ten)))
         (range 20 100))))
(ten) ;; returns 10

This^ won’t work for fully qualified vars. 😞

Hi! What is the best way in Clojure to “transform a deep data structure from leaf to root”? e.g. apply some function to the leaves, then one level up, and so on

ah nevermind, I think that’s exactly what clojure.walk/postorder does

@hmaurer isn't that what postwalk does? (clojure.walk/postwalk-demo {1 {2 {3 4} 5 6} 7 8})

@pesterhazy yep, it is!

does anybody know what's happened with http://planet.clojure.in and their fb group?

no activity for more than a week

domain expired

original owner doesn't respond

and "they" are planning to relaunch the site on a different domain

at least, that's what i heard

GraalVM has been officially released! 🎉 http://www.graalvm.org/

I’m going to try to run some Clojure on it, if anyone else tries it out please share your results 🙂

Hi, i have a map like this {:type :type1 :children [{:type :type2 :children [..] ..]}, and i want to write a spec for it . there is 4 types and an item with a certain type can only exists inside a specific type too (like :country :city :street ), how do you suggest i can achieve that ?

@schmee Are you gonna be writing a Clojure frontend for it?

I’m going to try 🙂

as a smart person once said: “don’t try anything, just do it”!

awesome 😄 I tried a simple frontend with Graal sometime back, found this: http://epub.jku.at/obvulihs/content/structure/501665

but didn’t see any source code for it anywhere

If you’re giving it a shot then I’m up for it too 😄 Had been following Graal for sometime

> Are you working on a Java application? Do you want a faster runtime? Enhance your code with JavaScript! Either this is bad wording or I want some proves. Since when does javascript speed up Java applications?

@sveri lol what's that from?

@tbaldridge from the graalvm intro page: http://www.graalvm.org/ Under "Get Started"

I assume they speak of the native image capability, but how that relates to javascript, it still doesnt make sense 😄

>since when does javascript speed up Java applications? since graal can optimize js code faster than javac+hotspot can optimzie java code :)

I just wanted to give it a try, but no open source prebuilt binary is available for windows. A pity. Maybe someone else wants to do some comparisons.

I’m curious about the “Enterprise edition”, seems to boast “Improved performance and smaller footprint”

wonder how big the difference is

Would a truffle clojure be faster in startup?

No

with regular Graal it would be even slower

but if it could be made to run on SubstrateVM somehow, it would be almost instant

I’m not sure if that is technically possible since SVM has a lot of restrictions around reflection and dynamic class loading

Yeah you wouldn't want to build a Clojure/JVM on SubstrateVM. You'd want a "clojure native" on SVM. Something with good C interop, and all the good/bad that comes with that

you could build a Java bytecode interpreter in Truffle and use that for interop like they’ve already done with LLVM bytecode

I think that is already on the roadmap

That's also possible. Someone did that with PyPy, they wrote a Prolog interpreter in RPython and then integrated it into a PyPy build. So you could have Python and Prolog talking to eachother, but each with their own VM semantics.

PyPy does that as well with regexs (they have their own JIT/runtime)

cool, in fact a regex JIT just landed in Graal as well 🙂

this whole new generation of JITs is incredibly exciting 😄

^long , ^Long and (long ...) which is preferred?

use (long ...) for local variables and ^long for function arguments

what about for bindings for loop

those are local variables 🙂

so something like this makes no sense (loop [^long somelocal foo] ... ) and should be (loop [somelocal (long foo)] does the type hint persist on different iterations of the loop? perhaps type hints have no meaning outside the context of a function arg?

the compiler detects primitive long locals and tries to retain them in a loop

ok

Do type hints work inside a destructured form in a function arg?

generally yes

since functions are data in clojure, can i send a function in a network request? does anyone have an example?

no

unless you send it as code then eval it on the other side

https://github.com/technomancy/serializable-fn is a thing some people use for this

@drawls sending code over the wire raises a lot of security questions and you should think twice before doing that. Or three, four times maybe.

👋

Is there a better way to interactively redefine proxy functions from a repl than this?

(defn make-proxy []
  (proxy [Parent] []
    (function1 [])
    (function2 [])))
(defonce p (make-proxy))
(update-proxy p (proxy-mappings (make-proxy)))

The only alternative I can think of is to make each proxy function call out to an actual clojure function of the same name. Then I can easily redefine that clojure function at runtime. That doesn't feel very clean though.

have you tried https://github.com/clojure/tools.namespace? I’m not sure if that refreshes proxy bindings, but it handles most other things so I assume it will

Huh, I didn't think about that. It's been a while since I've used it. I've always been more in favor of manually evaluating things from Emacs. It's worth a shot though.