Is there a standard way of storing a Directed Acyclic Graph in an Entity-Attribute-Value store ?

an adjacency list would be a simple way to do it

do you store the child list as a single vector, and when you add/del one child, overwrite the entire vector ?

that's an implementation decision, the totally denormalized version would be to make each child relation a separate entry

I think

hello every one, I'm starting to learn clojure (my first real functional language) after doing some functional JS

do you recomend any sql abstractions to use in my project with postgres?

I saw korma, but looks like it is abandoned, and I didn't like the yesql and hugsql approach

@matheusashton I'd recommend HoneySQL which lets you compose query fragments, on top of clojure.java.jdbc which is the Clojure Contrib library for JDBC operations.

Also @matheusashton since you're new to Clojure and interested in SQL, I'll recommend the #beginners and #sql channels.

Is there a clojure tool that will take a codebase, and show me all defns / defs taht are not used? It can assume I'm not doing something weird via string->symbol->resolve-in-ns

Yagni.

https://github.com/venantius/yagni

is boot-check the closest for boot ?

yagn ieems to depend on lein

$ boot check/with-yagni latest report from yagni.... [You Rock!]

Yes, boot-check runs Yagni

Gorilla REPL crashes every time when I use it on my Mac.

Does anyone else have the same problem?

@konanki.sivaram Use :plugins [[lein-gorilla "0.4.0"]] instead of :plugins [lein-gorilla "0.4.0"]

How do people feel about doing things like http requests inside of dosync, for example?

I find myself in a situation where I am receiving many (kafka) messages asynchronously, any of which triggers a particular http call. But I only want to make that call once.

So, I receive the message, dosync to assoc something under a key and trigger a request only if that key doesn’t exist yet

Can I do this without triggering a request inside the dosync?

Ah, dosync may retry anyway, so that doesn’t work

Hmmm, how do I avoid the race condition here: deref to see if key exists, if not dosync to add key and then (at some point) trigger request

Multiple threads could deref and see the missing key, both add the key and trigger request, unless those are somehow a single transaction

I guess in a single dosync it can check the key and add it in, no need for a separate deref

the question is when the request is triggered to guarantee it only happens once

Maybe this is finally a use case for agents

Ah that is actually a quite elegant solution

I finally understand what agents are for

generic question. Say I’m building a library and in the default namespace I need an instance of a parser. I will need this parser for every call into the library. I could have a (def parser ...) directly in the main namespace of the library, but for some fuzzy reason this makes me feel slightly dirty and I would feel more pure with lets say a memoized function returning the parser. Opinions? Is it ok and good style to just (def things or is it indeed somehow better to keep things in functions? One difference I see is that the def gets potentially ’unnecessarily` evaluated on load. The user of my lib might elect to not call any functions in my namespace etc...

@mbjarland Could use https://github.com/stuartsierra/component to avoid re-defining global variables on load.

That's what was recommended to me instead of global atoms

cloc dumps ttl lines, ttl white space, ttl comments is there some other tool that will dump ttl + also summaries for each file ?

i.e. I want a table of four columns: file name, comments, whitespace, line-count // though two clumns of just filename and line-count is fine, if it's better than wc -l

I do that. The repl starts and works for some time and then it stops working. Now I don’t seem to have that problem.

Good Morning Everybody 🙂

besides fn -> \lambda is there a standard set of "prettify these clojure keyword" setup ?

Sorry guys but I still cannot understand what those cmd line tool just released https://clojure.org/reference/deps_and_cli here are for. What gap did I have that they are trying to fill?

they are for people who have never tried clojure and want something minimal to fire it up

@alexmiller pointed me to https://clojure.org/guides/deps_and_cli. But if I have leiningen, what do i need those for?

you don't, and it's not for you

it's for people who haven't used clojure and want a simple way to fire up a clojure repl. Or people who want something ligher weight / more targetted than lein or boot for running a repl.

I also get the impression that they provide an official solution from the clojure team for launching a clojure repl. Previously it was sufficiently simple to not matter.

in the past we could use clojure.jar for that, clojure.jar no longer runs standalone because of clojure.spec.alpha

Like people have a python repl

right, python just runs as a standalone command

dumb question: why does clojure.spec.alpha prevent clojure.jar from being standalone ?

because it's not part of clojure.core, it's a separate artifact

so you need to construct a classpath with at least three artifacts to make it work iirc

you can't just 'stuff it in the *.jar" ? I thought jars were just zips/tar files and you could throw stuff into it

@qqq That leads to a classpath conflicts.

And tools.deps.alpha?

deps.alpha knows how to resolve deps, that is why it is there

it helps construct a classpath and download jars that you might need

Isn't that what mvn is for?

but unlike lein or boot it isn't for packaging or deploying and doesn't have tooling for plugins

yes, you could use mvn for that, but then you need an mvn command for running clojure

and mvn is also a packaging and plugin and deployment tool like leiningen or boot are

Thanks @noisesmith I was afraid I am missing something here.

is there a builtin for (magic [f1 f2 f3 f4] [x1 x2 x3 x4]) => [(f1 x1) (f2 x2) (f3 x3) (f4 x4)] }

the closest thing would be (map #(%1 %2) ...)

(def a-z 
  (mapv int->char  (range (char->int \a) (inc (char->int \z))))) 
(def letters-lower "abcdefghijklmnopqrstuvwxyz")

Is there something like (binding [*ns* 'afsdfsd] ::a) but that works? ('work' -> result in :afsdfsd/a)

no

read time happens too early for that to work

yeah, guessed so... thanks for the confirmation! useful coming from tools.analyzer author 🙂

np

hey everyone, how can I lowercase the keys in map? 😄 I have the following:

{:CurrentGame 1
 :Score 30}

this response comes from the server as json and when it is converted from json using clj-http, that’s how it shows up

solved it like this

(defn lower-case-keys [result]
  (if (map? result)
    (into {}
          (for [[k v] result]
            [(keyword (lower-case (name k))) v]))
    (map lower-case-keys result)))

(reduce-kv #(assoc %1 (keyword (lower-case ( name %2))) %3) {} json-response) something this

I like that one too!

I just needed my fn to support vectors / lists too

didn’t know about reduce-kv, thanks!

personally I like to use my own syntax sugar

(defn map-keys [f m]
  (->> m
       (map (fn [[k v]]
              [(f k) v]))
       (into {})))

(defn map-vals [f m]
  (->> m
       (map (fn [[k v]]
              [k (f v)]))
       (into {})))

@bravilogy with specter: (transform [MAP-KEYS NAME] clojure.string/lower-case m)

also does not change the type of the map and much higher performance than other approaches, especially for small maps

you can also do it with transducers and avoid creating the intermediate seq like this @bravilogy

(defn lower-case-keys [m]
  (into {} (map (fn [[k v]] [(-> k name lower-case keyword) v])) m))