Hello. I am learning clojure, lambda calculus, computation theory and architectural design

@trevor there is built-in feature to re-use connections: http://clojure-doc.org/articles/ecosystem/java_jdbc/reusing_connections.html#using-connection-pooling also, check for special libraries that manage connection pooling. We use https://github.com/tomekw/hikari-cp in our Postgres-driver Clojure project.

@trevor moreover, take a look at conman library. It’s a mixture of HugSQL + connection pooling + small details that make everything work fine. https://github.com/luminus-framework/conman

Hey guys, 4Clojure problems updated on Clojurecademy - https://clojurecademy.com/courses/17592186045426/learn/syllabus

hi guys

what is the wrong with this code https://gist.github.com/aibrahim/f48b8d35eaa4e63013dd66e07fce47cc StackOverflowError java.math.MutableBigInteger.divide (MutableBigInteger.java:1153)

why clojure breaks for recursion like this

?

You're not using tail recursion, so the stack will explode. And I think you're using "normal" integers alongside bigints. It may be out of date, but this could be helpful https://dev.clojure.org/display/design/Documentation+for+1.3+Numerics

I doubt it will fix it but you could try adding an M suffix to your numbers like (/ n 2M). But you may want to redesign your algorithm to use the recur form in the tail position. https://clojuredocs.org/clojure.core/recur

why recursion like this not working in clojure?

@abdullahibra also your code will logically never return, if you run (fast-fib 5), you will never reach the the case n=0

you always divide the argument by two and call it again

works

user=> (fast-fib 3) [2 3] user=> (fast-fib 10) [55 89] user=> (fast-fib 5) [5 8]

only added 2M as olical suggested

user=> (fast-fib 100) ArithmeticException integer overflow clojure.lang.Numbers.throwIntOverflow (Numbers.java:1501)

should not work...

Because you're not using recur in the tail position though, it will blow the stack at some point.

(defn fast-fib [n]
  (if (== n 0)
    [0, 1]
    (let [[a, b] (fast-fib (int (/ n 2)))
          c (* a (- (* b 2) a))
          d (+ (* a a) (* b b))]
      (if (zero? (mod n 2))
        [c, d]
        [d, (+ c d)]))))

you have to do integer coercion

otherwise you never reach the end, tail recursion is not the main problem here

if I have :profiles :dev in my project.clj, under what conditions does that profile take effect by default without my explicit use of with-profile ?

Why doesn’t run! take an init argument?

because run! is about consuming each element individually, not accumulating

ok, I needed some state in my proc (alternating boolean), so I just used reduce

could have used map-indexed, but doesn’t matter I guess

@borkdude If it's a vector you can use reduce-kv and check for even? / odd? to get alternating booleans

in hindsight I couldn’t use the index at all, because it depends…

so I need reduce for sure, but don’t need the end result.. hence my question 🙂

then again, the end result is a boolean, so who cares.

I have a strange situation. A function works as expected in repl (freshly started, no old state build up) but not from a compiled uberjar.

(defn print-router-config [rcfg]
  (log/info "printing router config")
  (println rcfg)
  (let [cfg (:customer-router-config rcfg)
        fname (str (:region account) "_" (:alias account) "_" (:virtual-interface-id rcfg) "_" (:location rcfg) ".txt")]
    (spit fname (xml->json (xml/parse (java.io.ByteArrayInputStream. (.getBytes cfg)))))))


(defn download-router-config [account]
  (log/info "doing the thing")
  (let [cfgs (get-router-configs account)]
    (println "should map print over cfgs now")
    (println (type cfgs))
    (println (count cfgs))
    (println (type (first cfgs)))
    (map print-router-config cfgs)))

map is lazy

You can simply replace map with run!

thanks @hiredman, @madstap! obviously i don't really understand laziness. reading https://stuartsierra.com/2015/08/25/clojure-donts-lazy-effects now. any other resources/books that may help with clarity?

run! did the trick!! yay. thank you so much.

Apologies if this is the wrong channel, but is anyone else having trouble accessing http://jarkeeper.com/? I'm getting an "Origin DNS error" message from CloudFlare.

@deactivateduser10790 Maven Central reported a DNS issue that seems to have screwed a lot of build systems up...

https://twitter.com/sonatype_ops/status/930138391005589505

Ah ok - jarkeeper uses central for some of its info, I presume?

Jarkeeper is down at the moment, it's been down for a few weeks now

I was seeing Travis builds failing all over the place this morning (while trying to release a new version of clj-time).

Oh, so this is unrelated? OK, I'll get back in my box... 😈

@deactivateduser10790 I just announced https://versions.deps.co/ which provides the same service as Jarkeeper (and was forked from it)

Thanks @danielcompton!

I'm a big fan "public self shaming" as a motivation technique. 😉

Jarkeeper-type tools are great for this.

When I deploy my release to clojars, printing Failed to deploy artifacts/metadata: Cannot access releases with type default using the available connector factories: BasicRepositoryConnectorFactory`

What is that mean ?

What tool are you deploying with?

leiningen.

which profile in project.clj is the one that is distributed in a clojars dependency?