Liking cljdoc? Tell your friends :D

EDN Config (formerly Aero)

A small library for explicit, intentful configuration.


The library was originally developed and maintained till early 2020 by JUXT LTD as just/aero but since then seems abandoned and not reflecting proposed fixes and changes. Since the library is widely used in many clojure projects this fork is intended to be a drop-in replacement.

Please, pay attention that the fork is not of a main goal of development and supporting the library itself but to unlock further development and improvement of other dependent services and to avoid future hardcoding of local fixes of original library into the services itself.

The artefact name change intended to avoid misuse and any of possible confusions and also to make a migrations footprint as small as possible: migration from aero versions of 1.1.4-1.1.6 to edn-config of version 1.0.0 should be as easy as just proper replacing dependency and referred :requires sections.


Add the following dependency to your project.clj file

GitHub edn config GitHub tag (latest by date) GitHub last commit


Please note that being a beta version indicates the provisional status of this library, and that features are subject to change.

Getting started

Create a file called config.edn containing the following

{:greeting "World!"}

In your code, read the configuration like this

(require '[edn-config.core :refer (read-config)])
(read-config "config.edn")

or to read from the classpath, like this

(read-config ( "config.edn"))

Keep in mind that even though (read-config "config.edn") will work in your Repl and when running tests, it’s very likely to catastrophically fail if you run your application from the generated .jar file.

So to avoid surprises it’s better to always use io/resource which works in all scenarios.

Design goals

Explicit and intentional

Configuration should be explicit, intentful, obvious, but not clever. It should be easy to understand what the config is, and where it is declared.

Determining config in stressful situations, for example, while diagnosing the cause of a production issue, should not be a wild goose chase.

Avoid duplication …

Config files are often duplicated on a per-environment basis, attracting all the problems associated with duplication.

… but allow for difference

When looking at a config file, a reader will usually ask: "Does the value differ from the default, and if so how?". It’s clearly better to answer that question in-place.

Allow config to be stored in the source code repository …

When config is left out of source code control it festers and diverges from the code base. Better to keep a single config file in source code control.

… while hiding passwords

While it is good to keep config in source code control, it is important to ensure passwords and other sensitive information remain hidden.

Config should be data

While it can be very flexible to have 'clever' configuration 'programs', it can be unsafe, lead to exploits and compromise security. Configuration is a key input to a program. Always use data for configuration and avoid turing-complete languages!

Use environment variables sparingly

We suggest using environment variables judiciously and sparingly, the way Unix intends, and not go mad. After all, we want to keep configuration explicit and intentional.

Also, see these arguments against.

Use edn

Fortunately for Clojure developers like us, most of the tech to read configuration in a safe, secure and extensible way already exists in the Clojure core library (EDN).

Tag literals

EDN Config (formerly Aero) provides a small library of tag literals.


Use #env to reference an environment variable.

{:database-uri #env DATABASE_URI}

It is considered bad practice to use environment variables for passwords and other confidential information. This is because it is very easy to leak a process’s environment (e.g. via ps e -f or to your application monitoring tool). Instead you should use #include - see here.


Use #envf to insert environment variables into a formatted string.

{:database #envf ["protocol://%s:%s" DATABASE_HOST DATABASE_NAME]}


Use #or when you want to provide a list of possibilities, perhaps with a default at the end.

{:port #or [#env PORT 8080]}


#join is used as a string builder, useful in a variety of situations such as building up connection strings.

{:url #join ["jdbc:postgresql://psq-prod/prod?user="
             #env PROD_USER
             #env PROD_PASSWD]}


Use profile as a kind of reader conditional.

#profile expects a map, from which it extracts the entry corresponding to the profile.

  {:port #profile {:default 8000
                   :dev 8001
                   :test 8002}}}

You can specify the value of profile when you read the config.

(read-config "config.edn" {:profile :dev})

which will return

  {:port 8001}}

(#profile replaces the now deprecated #cond, found in previous versions of Aero)


Use when config has to differ from host to host, using the hostname. You can specify multiple hostnames in a set.

  {:port #hostname {"stone" 8080
                    #{"emerald" "diamond"} 8081
                    :default 8082}}}

int, long, double, keyword, boolean

Use to parse a String value into a Integer, Long, Double, keyword or boolean.

{:debug #boolean #or [#env DEBUG "true"]
  {:port #long #or [#env PORT 8080]
   :factor #double #env FACTOR
   :mode #keyword #env MODE}}


#user is like #hostname, but switches on the user.


Use to include another config file. This allows you to split your config files to prevent them from getting too large.

{:webserver #include "webserver.edn"
 :analytics #include "analytics.edn"}
By default #include will attempt to resolve the file to be included relative to the config file it’s being included from. (this won’t work for jars)

You can provide your own custom resolver to replace the default behaviour or use one that edn-config provides (resource-resolver, root-resolver). For example

(require '[edn-config.core :refer (read-config resource-resolver)])
(read-config "config.edn" {:resolver resource-resolver})

You can also provide a map as a resolver. For example

(read-config "config.edn" {:resolver {"webserver.edn" "resources/webserver/config.edn"}})


Merge multiple maps together

#merge [{:foo :bar} {:foo :zip}]


To avoid duplication you can refer to other parts of your configuration file using the #ref tag.

The #ref value should be a vector resolveable by get-in. Take the following config map for example:

{:db-connection "datomic:dynamo://dynamodb"
  {:db #ref [:db-connection]}
  {:db #ref [:db-connection]}}

Both :analytics and :webserver will have their :db keys resolved to "datomic:dynamo://dynamodb"

References are recursive. They can be used in #include files.

Define your own

EDN Config (formerly Aero) supports user-defined tag literals. Just extend the reader multimethod.

(defmethod reader 'mytag
 [{:keys [profile] :as opts} tag value]
  (if (= value :favorite)

Hide passwords in local private files

Passwords and other confidential information should not be stored in version control, nor be specified in environment variables. One alternative option is to create a private file in the HOME directory that contains only the information that must be kept outside version control (it is good advice that everything else be subject to configuration management via version control).

Here is how this can be achieved:

{:secrets #include #join [#env HOME "/.secrets.edn"]

  #profile {:test #ref [:secrets :aws-test-key]
            :prod #ref [:secrets :aws-prod-key]}}

Use functions to wrap access to your configuration.

Here’s some good advice on using EDN Config (formerly Aero) in your own programs.

Define a dedicated namespace for config that reads the config and provides functions to access it.

(ns myproj.config
  (:require [edn-config.core :as aero]))

(defn config [profile]
  (aero/read-config "dev/config.edn" {:profile profile}))

(defn webserver-port [config]
  (get-in config [:webserver :port]))

This way, you build a simple layer of indirection to insulate the parts of your program that access configuration from the evolving structure of the configuration file. If your configuration structure changes, you only have to change the wrappers, rather than locate and update all the places in your code where configuration is accessed.

Your program should call the config function, usually with an argument specifying the configuration profile. It then returned value passes the returned value through functions or via lexical scope (possibly components).

Using EDN Config with Plumatic schema

EDN Config (formerly Aero) has frictionless integration with Plumatic Schema. If you wish, specify your configuration schemas and run check or validate against the data returned from read-config.

Using EDN Config with components

If you are using Stuart Sierra’s component library, here’s how you might integrate EDN Config.

(ns myproj.server
  (:require [myproj.config :as config]))

(defrecord MyServer [config]
  (start [component]
    (assoc component :server (start-server :port (config/webserver-port config))))
  (stop [component]
    (when-let [server (:server component)] (stop-server server))))

(defn new-server [config]
  (->MyServer config))
(ns myproj.system
  [com.stuartsierra.component :as component]
  [myproj.server :refer [new-server]])

(defn new-production-system []
  (let [config (config/config :prod)]
      (component/system-map :server (new-server config))

However, another useful pattern you might consider is to keep your system map and configuration map aligned.

For example, imagine you have a config file:

{:listener {:port 8080}
 :database {:uri "datomic:mem://myapp/dev"}}

Here we create a system as normal but with the key difference that we configure the system map after we have created using merge-with merge. This avoids all the boilerplate required in passing config around the various component constructors.

(defrecord Listener [database port]
  Lifecycle …)

(defn new-listener []
  (using (map->Listener {}) [:database])

(defrecord Database [uri]
  Lifecycle …)

(defn new-database []
  (map->Database {}))

(defn new-system-map
  "Create a configuration-free system"
   :listener (new-listener)
   :database (new-database)))

(defn configure [system profile]
  (let [config (aero/read-config "config.edn" {:profile profile})]
    (merge-with merge system config)))

(defn new-dependency-map [] {})

(defn new-system
  "Create the production system"
  (-> (new-system-map)
      (configure profile)
      (system-using (new-dependency-map))))

Also, if you follow the pattern described here you can also ensure accurate configuration is given to each component without having to maintain explicit schemas. This way, you only verify the config that you are actually using.

Feature toggles

EDN Config (formerly Aero) is a great way to implement feature toggles.

Use a single configuration file

If at all possible, try to avoid having lots of configuration files and stick with a single file. That way, you’re encouraged to keep configuration down to a minimum. Having a single file is also useful because it can be more easily edited, published, emailed, watched for changes. It is generally better to surface complexity than hide it away.

(Alpha) Define macro tag literals

edn-config.alpha.core defines a new experimental API for tagged literals. This API allows you to define tagged literal "macros" similar to macros in Clojure. It is intended for use in creating your own conditional constructs like #profile and #or.

case-like tag literal

The easiest kind of tagged literal to create is a case-like one. A case-like tagged literal is one which takes a map of possible paths to take. An example of this in EDN COnfig is #profile.

Here’s how you can define your own version of #profile:

(ns myns
  (:require [edn-config.alpha.core :as edn-config.alpha]))

(defmethod edn-config.alpha/eval-tagged-literal 'profile
  [tagged-literal opts env ks]
  (edn-config.alpha/expand-case (:profile opts) tagged-literal opts env ks))

eval-tagged-literal allows you to define macro tagged literals. expand-case is a function which forms the common behaviour beneath #user, #profile, etc.

Other conditional constructs

#or is very different from #profile in implementation, and doesn’t have a convenience function. The source for #or in edn-config.core is a good example of doing custom partial expansion from a tagged literal.

The primitives you will need to understand are: edn-config.alpha.core/expand, edn-config.alpha.core/expand-coll, edn-config.alpha.core/expand-scalar. And helpers: edn-config.alpha.core/expand-scalar-repeatedly. These vars have docstrings which explain their specific purpose.

All expand-* functions take parameters opts, env, and ks. opts are the same opts that are passed to edn-config.core/read-config. env is a map of ks to their resolved values in the config, being absent from this map means the value is not yet resolved. ks is a vector representing the current key path into the location of this tagged literal.

Your implementation of eval-tagged-literal must assoc the ks into env if it is successfully resolved.


EDN Config (formerly Aero) is built on Clojure’s edn.

EDN Config (formerly Aero) is influenced by nomad, but purposely avoids instance, environment and private config.


Thanks to the following people for inspiration, contributions, feedback and suggestions.

  • Gardner Vickers

The MIT License (MIT)

Copyright © 2015-2022 JUXT LTD (as just/aero) Copyright © 2022 UPD Consulting Ltd (as com.dkdhub/edn-config)

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.


Can you improve this documentation?Edit on GitHub

cljdoc is a website building & hosting documentation for Clojure/Script libraries

× close