i'm defining a new client/server protocol in which the client sends EDN to the server in the shape of ... {:kind :tx :data {:name :foo/bar :args [13 "baz" :fizz]}}

basically i want to "route" the request based on the :kind and :name

would using a multimethod seem like an obvious choice for this?

iirc defmulti can handle one pattern at a time so you'd have to write it separately for kind and name

defmulti takes a function in parameter, you can write (fn [x] [(:kind x) (get-in x [:data :name]) in defmulti and match it with e.g. (defmethod foo [:tx :foo/bar] ...)

Any way of setting *warn-on-reflection* true only for our own files? I'm flooded with warnings from dependencies

what's the right incantation to make (s/def (f "foo") any?) work? spec complains that k must be fully qualified keyword so i'm trying to make a macro that returns essentially ::foo but no luck so far

Hi, people. I need UI component I can use in my browser which allows to edit clojure code (does some syntax highlight, bracketing, idention) and thats all In general from REPL I neeed R part... as E part will be handled by clojure on the server

Do we have any editors yet that can do this: https://www.youtube.com/watch?v=d8-b6QEN-rk ?

Hi peepz, I am trying to install tailwindcss webjar, but lein deps fails with

Could not find metadata org.webjars.npm:caniuse-lite/maven-metadata.xml in local (/home/naiali/.m2/repository)
Failure to find org.webjars.npm:caniuse-lite/maven-metadata.xml in  was cached in the local repository, resolution will not be reattempted until the update interval of clojars has elapsed or updates are forced
Could not find metadata org.webjars.npm:caniuse-lite/maven-metadata.xml in local (/home/naiali/.m2/repository)

where do I start looking/fixing?

is zipmap expected to keep the order of keys in the map generated to be same as the keys coll arg ?

The map zipmap creates is a hashmap, which are not ordered

In general the default for maps in clojure is unordered

thanks , just noticed that it is ordered to keys arg for colls of size 8 or less , anything larger the result is unordered

That's an implementation detail you shouldn't rely on (array-map vs hash-map)

(type {1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8})
;; => clojure.lang.PersistentArrayMap

(type {1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9})
;; => clojure.lang.PersistentHashMap

:thumbsup:

Interestingly, the explicit hash-map constructor with 0 args returns the wrong type

(type (hash-map))
;; => clojure.lang.PersistentArrayMap

I wonder if that is a bug

its not what it says on the tin , so probably it is a bug

Hash maps with small numbers of elements (including zero elements) are implemented as array maps. Not a bug.

user=> (type {:a 1 :b 2 :c 3 :d 4 :e 5 :f 6 :g 7 :h 8})
clojure.lang.PersistentArrayMap
user=> (type {:a 1 :b 2 :c 3 :d 4 :e 5 :f 6 :g 7 :h 8 :i 9})
clojure.lang.PersistentHashMap
user=> 

Ah, I see you already posted that above. Anyways, not a bug -- by design.

I've added an example to zipmap just in case someone else mistakes it to be ordered like I did

Yeah, it's tempting when tutorials use small hash maps and they seem to maintain order.

There are also things like these: https://github.com/clj-commons/ordered

I would avoid such things, personally, because it has the potential to be very confusing: built-in hash maps and sets being unordered, mixed with code that expects "special" hash maps and sets to be ordered. To say nothing of the issue of predictable performance on lookup/traversal...

Yep, I understand the whole implementation/interchangeability thing, but I would think the hash-map and array-map functions were supposed to explicitly return those respective types

from the source code:

(defn hash-map
  "keyval => key val
  Returns a new hash map with supplied mappings.  If any keys are
  equal, they are handled as if by repeated uses of assoc."
  {:added "1.0"
   :static true}
  ([] {})
  ([& keyvals]
   (. clojure.lang.PersistentHashMap (create keyvals))))

(type (hash-map :a 1)) ;; => clojure.lang.PersistentHashMap

also, (self-hosted) clojurescript has the "correct" behaviour:

$ lumo
ClojureScript 1.10.520
cljs.user=> (type (hash-map))
cljs.core/PersistentHashMap

damn

It does seem a bit surprising, perhaps, that hash-map can return an object of type PersistentArrayMap, but yeah, if you try writing code that treats the concrete type of map returned by a function as an implementation detail, with no ordering guarantees (unless it is one of the ordered map types that explicitly guarantee ordering), then you will be happier long term.

I’m pretty new to Clojure and I am wondering about declaring variables in the leiningen project.clj file. I want to implement some gRPC services and the dependencies look like this:

[io.grpc/grpc-core "1.13.1"]
   [io.grpc/grpc-netty "1.13.1"]
   [io.grpc/grpc-protobuf "1.13.1"]
   [io.grpc/grpc-stub "1.13.1"] 

@gbson You can but it relies on a "quirk" of Leiningen and you wouldn't be able to do that if you switched to CLI/`deps.edn` at some future date.

You could put (def grpc-version "1.13.1") above the defproject form in project.clj and then use ~grpc-version in the dependency vector to get the version value substituted as I recall (been a long time since I've used Leiningen!).

Ah - should I perhaps use some other build tools now?

I’m only building a really simple application now to move some data around, to get familiar with the language

Leiningen is a "classic" and most tutorials/books still use it but more and more people are using the "official" CLI (`clj` / clojure scripts) with deps.edn.

Are you on Windows or macOS/Linux?

Develop on mac and deploy on linux

'k, so you can brew install clojure to get the new CLI tools.

(assuming you already have Homebrew on your Mac!)

Yep!

Ooh git dependencies!

Git and local deps, yeah. Lots of nice stuff with the new tooling.

Check out https://github.com/seancorfield/dot-clojure as a guide for what you can do with your ~/.clojure/deps.edn file.

Nice - currently we provide our api stubs as git projects so it may be very easy to use them here then

Oh clojure works with java 11 now? That’s neat

Clojure 1.10.x works with all JDK versions 8 and later.

Clojure is tested against quite a few JDKs https://build.clojure.org/job/clojure-test-matrix/

There are a few issues with warnings/etc. that you might experience on later JDKs that do not occur with earlier JDKs, but those are mostly because of tightening security restrictions in later JDKs, less because of Clojure. e.g. as described here: https://clojure.org/guides/faq#illegal_access

which you can mostly ignore, and/or handle via techniques described in the answer there, if you care to.

Is there a simple way of generating a random boolean?

(= (rand-int 2) 0) should return true or false with equal probability.

You can use rand-int with a larger range, and whatever condition on the return value you want, if you want non-equal probabilities for true vs. false.

Perfect. Thanks @andy.fingerhut!

@darrell I normally just do: (rand-nth [true false])