That can be controlled with https://clojure.github.io/clojure/clojure.core-api.html#clojure.core/*print-namespace-maps*

works wonders

(def foo ( "/var/tmp/foo.txt"))
(def parent (.getParentFile foo))
(def sibling ( parent "bar.txt"))
(def sibling2 ( foo ".." "bar.txt"))
(.getCanonicalPath sibling2) ;; "/private/var/tmp/bar.txt"
(.getCanonicalFile sibling2) ;; #object[.File 0x1d74ad00 "/private/var/tmp/bar.txt"]
(def multiple-levels ( parent "abra" "cadabra" "log.txt"))
(.getAbsolutePath multiple-levels) ;; "/var/tmp/abra/cadabra/log.txt"

could this be helpful too? https://docs.oracle.com/javase/7/docs/api/java/nio/file/Path.html#relativize(java.nio.file.Path)

Hmm, I'm not sure I understand phronmophobic. There already exists two files in the filesystem and I want to calculate a path from one of them to the other. But that http://java.nio.fi

No, it's calling the Java meta method of the clojure.lang.IMeta interface.

(. ^C x (m)) in Clojure is the same as x.m() in Java where x is of type C.

Also, FYI this is not the official Clojure source, which is at https://github.com/clojure/clojure

IIRC Clojure doesn't have a direct analog to import x.*. You will have to provide each class' name.

correct

(import '[com.braintreegateway BraintreeGateway])
(import 'com.braintreegateway.BraintreeGateway)
(import [com.braintreegateway BraintreeGateway])

(import (com.braintreegateway BraintreeGateway))

Works for me. Are you using it by itself or in an ns form? Are you sure you have added the dependency to wherever you add dependencies?

doesn't work == what happens?

Ahh, Braintree integration. Much fun lies ahead. You will probably need this subset for a reasonable minimum:

(:import (com.braintreegateway
            BraintreeGateway Environment Result
            ClientTokenRequest CustomerRequest AddressRequest PaymentMethodRequest SubscriptionRequest
            SubscriptionSearchRequest
            Subscription
            Subscription$Status
            WebhookNotification)

In my experience (20+ Clojure apps, small and big-ish) things like Component are incredibly helpful to separate "application" structure from the biz logic

@U08716WQP If you like, you could post a github link to your code, and I can leave some comments when I have time.

Interestingly enough - I cannot recall many articles about structuring Clojure applications

I don’t know of any great articles on it. The best general advice I could give is: Flatten your logic. Stop creating functions.

@U07S8JGF7 I’m intrigued. What do you mean by stop creating functions?

At a literal level, I mean “stop using defn” 😄

That’s obviously exaggerated.

But it’s very important that every name you introduce carries a cost: the cost of remembering what it means. Clojure often allows you to be so concise, that inlining those functions is both more precise and less overhead.

I’ve found that difficult-to-read clojure programs often suffer from this problem.

(the problem of “too many names”)

I think it’s a holdover from OOP where “short methods = good.”

What does flatten logic mean?

When you stop creating nested function calls, your logic naturally gets pushed to the top level.

So you end up with longer, flatter functions in general.

Again, this is the opposite of a lot of OOP advice. But when you’re dealing primarily with immutable values, longer functions become easier to manage.

Because, for example, you cannot silently tweak state on line 27 that affects a switch on line 124.

here’s an example: https://github.com/potetm/advent-of-code/blob/master/src/advent_2019/day_5.clj#L15-L106

very long function

But I heard multiple people complain about this state machine during advent of code, and, for me, it was extremely easy to make changes because all of the logic was bundled together.

Note that I could have broken this up and used multimethods or something. But 1. the op codes are a closed set and 2. that would have required figuring out how parse a bunch of values and pass them into the multimethod.

You can avoid all those problems by cheating: Just putting everything you need in one scope.

@U07S8JGF7 I wouldn't have thought to ask this question, butI like appreciate this insight - thanks

regarding the organization of clojure code in general, I know of nothing better than Elements Of Clojure https://leanpub.com/elementsofclojure

I think I get it.

Elements of Clojure does a good job of putting words to this. I hesitate to recommend it because it’s so abstract I think it would be hard to understand if you don’t already have a pretty good idea.

Interesting thread. I am at a stage where I'm re-designing my application for the 5th or 6th time, as I discover new better ways to structure the code. One interesting thing I found is that the split into layers (UI, Application, Business Logic, Infrastructure) doesn't work that well. Specifically, Business Logic (which I call Domain) should not be a "layer" above infrastructure, but something entirely separate, because it is useful both in server-side Clojure and client-side ClojureScript.

But I'd love to read some books about designing large systems with Clojure. I recently read "Domain-Driven Design" by Eric Evans and found out that I've been practicing Domain-Driven Design all these years. Good ideas in that book, although it gets somewhat mired in the object-oriented tarpit.

Yeah I’ve tried similar splits, and I found them equally unhelpful.

Hope you’re not serious. This statement is obviously a bit of a joke.

It’s one person who’s obviously very, very biased talking about his own feelings. No reason to count it as anything more than that.

Is he really joking? If so, it went over my head, lol. And pleasantly so….

No, I don’t think he’s joking. But it’s clearly a joke of a statement 😄

are you saying typing has no tradeoffs? :)

it's a pretty good deal for the typer :)

If you use test.chuck’s checking macro (https://github.com/gfredericks/test.chuck#alternate-clojuretest-integration) you should be able to use normal clojure.test fixtures

Nice, yeah I was thinking of using a fixture but good to know that it doesn’t work out of the box

I'd be surprised if it doesn't work out of the box

what is even being mreasured in this graph?

https://snyk.io/wp-content/uploads/jvm_2020.pdf

page 11

alternative source: https://snyk.io/blog/kotlin-overtakes-scala-and-clojure-to-become-the-2nd-most-popular-language-on-the-jvm/

"For this survey, we teamed up with conferences and communities across the JVM ecosystem" - an incredibly biased data source.

the repl doesn't use to-string, it uses the system underlying prn, which is based on a multimethod

iirc `*print-method*` but it shouldn't be hard to verify

yeah, print-method

(ins)user=> (source pr)
(defn pr
  "Prints the object(s) to the output stream that is the current value
  of *out*.  Prints the object(s), separated by spaces if there is
  more than one.  By default, pr and prn print in a way that objects
  can be read by the reader"
  {:dynamic true
   :added "1.0"}
  ([] nil)
  ([x]
     (pr-on x *out*))
  ([x & more]
   (pr x)
   (. *out* (append \space))
   (if-let [nmore (next more)]
     (recur (first more) nmore)
     (apply pr more))))
nil
(ins)user=> (source pr-on)
Source not found
nil
(ins)user=> (source clojure.core/pr-on)
(defn pr-on
  {:private true
   :static true}
  [x w]
  (if *print-dup*
    (print-dup x w)
    (print-method x w))
  nil)
nil

usually you want to override both print-method and print-dup

So print-dup is being called when I use the REPL rather than the print-method I've defined?

you didn't define print-method, you defined toString I didn't read carefully

OK, so here's the complete definition:

(defrecord langStr [s tag]
  Object
  (toString [_] s))

(defmethod print-method langStr
  [literal ^.Writer w]
  (.write w (str "#langStr \"" literal "@" (:tag literal) "\"")))


(defmethod print-dup langStr [o w]
  (print-method o w))

(defn read-langStr [form]
  (let [langstring-re #"^(.*)@([-a-zA-Z]+)" 
        m (re-matches langstring-re form)
        ]
    (when (not= (count m) 3)
      (throw (ex-info "Bad langString fomat"
                      {:type ::BadLangstringFormat
                       :regex langstring-re
                       :form form})))
    (let [[_ s lang] m]
      (langStr. s lang))))

Plus I have

langStr defined in my data_readers.clj

But when I evaluate it I still get the original reader defined for a defrecord:

> #langStr "jail@en-us"

{:s "jail", :tag "en-us"}

@alexmiller is there something else I need to do to override the defrecord print-method?

You’re using langStr there - might want to check that’s evaluating to the full class?

In the dispatch value

Well my data_reader has this clause:

langStr ont-app.sparql-endpoint.core/read-langStr

Data readers switch on the tag symbol

Printing switches on Java class

def jail #langStr "jail@en-us")
#'ont-app.sparql-endpoint.core/jail
(class jail)
ont_app.sparql_endpoint.core.langStr

Also ont_app.sparql_endpoint.core.langStr shows up in the output from (methods print-method)

I tried this:

(prefer-method print-method langStr clojure.lang.IRecord)
Execution error (IllegalStateException) at ont-app.sparql-endpoint.core/eval13294 (form-init4216778243462303189.clj:139).
Preference conflict in multimethod 'print-method': interface clojure.lang.IRecord is already preferred to class ont_app.sparql_endpoint.core.langStr

So there's just no way to get the REPL to print a defrecord except as {<k> <v>, ...} ?

Ah. Sure 'nuff deftype fixes the problem

Thanks guys!

you can totally do this with defrecord - you may need to add a prefer-method though

$ clj
Clojure 1.10.1
user=> (defrecord Rec [a])
user.Rec
user=> (prn (->Rec 1))
#user.Rec{:a 1}
nil
user=> (->Rec 1)
#user.Rec{:a 1}
user=> (defmethod print-method Rec [^Rec r ^.Writer w] (.write w ^String (str (:a r))))
#object[clojure.lang.MultiFn 0x5b022357 "clojure.lang.MultiFn@5b022357"]
user=> (defmethod print-dup Rec [^Rec r ^.Writer w] (print-method r w))
#object[clojure.lang.MultiFn 0x6ceb7b5e "clojure.lang.MultiFn@6ceb7b5e"]
user=> (->Rec 1)
1
user=> (prn (->Rec 1))
1
nil

I got it working a while back by using deps https://shadow-cljs.github.io/docs/UsersGuide.html#deps-edn

so just moving dependancies to deps.edn will make cursive work again?

The problem I had was that cursive doesn't recognize shadow-cljs.edn as a project configuration file. It does recognize deps.edn so if that's the problem it might fix it. Otherwise I suggest bringing up the question #cursive

Ah I see! Ok I will check it out! Thanks!

Unfortunatly it doesn't work. 😞 I am just getting this: https://gist.github.com/ShakoFarhad/4b73a62b2ed1dfe054473faa12d6e67e

it sounds like you have an error in your deps

it’s hard to help you troubleshoot without knowing what you’re doing

@U0121V7FXG8 The Shadow documentation has a section on how to work around this: https://shadow-cljs.github.io/docs/UsersGuide.html#_cursive

I am getting this problem: https://gist.github.com/ShakoFarhad/39ab6314891431f540e2ccbac2c351d7 It talks about the http://java.erxe not being found. It is weird because it is at the given path. Nothing is missing. :s

This is how my project.clj file looks

And this is the shadow-cljs.edn file

do you have a package.json with the same version of shadow-cljs installed into node_modules?

I will just keep using shadow-cljs and keep updating the project.clj file whenever I make a dependancy change in shadow-cljs.edn. That way cursive will keep working and I can still use shadow-cljs for building. It is perhaps a bit inefficient to keep two files up to date with one another, but at least everything works now 😄

I fixed it! Apparently I needed a specific version of jdk. I had 13.0.2 but it complained about not finding 13.0.1. So I installed 13.0.1 and it works now. 😮

“transit-encoded edn?” You mean you serialize to EDN and then serialize to transit? Surely not.

https://github.com/binaryage/cljs-devtools does this for EDN. I’m assuming you can use the same approach for transit.

Oh right I use that one, but it doesn't affect the network panel, so normal transit requests/responses are hard to read. Not sure if that is possible in a browser extension.

No, it doesn’t support transit. But I’m pretty sure it could, if you made it do so 😄

Ok, I'll look into it. I'm happy to do it if possible.

it isn’t possible to get the network panel to show transit decoded

note that it might vary depending on which version of transit your app is using

the best option I’ve found is to copy the response and decode it using jet

https://github.com/borkdude/jet

ah I see, thanks @U4YGF4NGM