Programming language: Go
License: Apache License 2.0
Tags: Configuration    

piper alternatives and similar packages

Based on the "Configuration" category.
Alternatively, view piper alternatives based on common mentions on social networks and blogs.

Do you think we are missing an alternative of piper or a related project?

Add another 'Configuration' Package


piper - Simple Wrapper For Viper

example workflow Go Report Card

  • Single Source of Truth
  • Generated Key Structs, No Typo
  • Config Inheritance
  • Multiple Config Strategies Support
  • Cache For Better Performance

Why Piper

If you are familiar with Django, this is how Django settings module looks like:

```console {12-20} โ””โ”€โ”€ settings โ”œโ”€โ”€ base.py โ”œโ”€โ”€ dev.py โ”œโ”€โ”€ stage.py โ””โ”€โ”€ prod.py

`dev.py` will inherit from `base.py`, also `stage.py` will inherit from `dev.py`.

Start Django with selected setting:
export DJANGO_SETTINGS_MODULE=mysite.settings.prod
django-admin runserver

And this is how you access settings in Django:

from django.conf import settings

author = settings.Author

I want to have similar experience with Viper, so here comes this wrapper:

```console {12-20} โ””โ”€โ”€ config โ”œโ”€โ”€ base.toml โ”œโ”€โ”€ dev.toml โ”œโ”€โ”€ stage.toml โ””โ”€โ”€ prod.toml

This is how you access config using piper:

import "your_project/config"

func main() {
    author = piper.GetString(config.Author)

Check example folder for more details.


go get github.com/Yiling-J/piper/cmd

Add Config Files

Add your config files to your config folder, usually your config folder should be under project root folder.

toml example, you can also use yaml or json. ```console {12-20} project โ””โ”€โ”€ config โ”œโ”€โ”€ base.toml โ”œโ”€โ”€ dev.toml โ”œโ”€โ”€ stage.toml โ””โ”€โ”€ prod.toml

To support inheritance, you need to add a special key to your config file called `pp_imports`

pp_imports = ["base.toml", "dev.toml"]

Piper will resolve that automatically.

## Config Key Generation
Run code generation from the root directory of the project as follows:
go run github.com/Yiling-J/piper/cmd your_config_folder

In this step piper will load all files in your config folder and merge them together. Then piper will generate config.go under your config folder, include all your config keys. Also you will see the config structure when pipe generating code.

After code genertation, your config folder should look like: ```console {12-20} โ””โ”€โ”€ config โ”œโ”€โ”€ base.toml โ”œโ”€โ”€ dev.toml โ”œโ”€โ”€ stage.toml โ”œโ”€โ”€ prod.toml โ””โ”€โ”€ config.go

## Use Piper

### Strategy I - Embed
embed your config folder into your code, single executable when you deploy.
import (

//go:embed config/*
var configFS embed.FS

author := piper.GetString(config.Author)

Strategy II - Embed with Env

embed your config folder into your code, single executable when you deploy, and replace secret with env.

import (

//go:embed config/*
var configFS embed.FS

os.Setenv("SECRET", "qux")
// make sure turn on AutomaticEnv first then loading config,
// this way the env vaiable is also cached, so IGet* methods can work properly
secret := piper.GetString(config.Secret)

Strategy III - Copy config directory

copy config folder when building docker image, so the true config folder exists.

import (

author := piper.GetString(config.Author)

Strategy IV - Mix embed and copy directory

Embed your config folder, but keep some secret keys in a real config file.

import (

//go:embed config/*
var configFS embed.FS

// "config/stage_with_secret.toml" is not in source code,
// may come from docker build or k8s ConfigMap
author := piper.GetString(config.Author)

True directory will take precedence over embeded directory.

Access Viper

Piper is just a wrapper, so you can always get the wrapped viper instance:

v := piper.V()

Be careful when using viper directly to load config, piper may not work properly.

Piper or Pipers?

Piper comes ready to use out of the box. There is no configuration or initialization needed to begin using Piper. Since most applications will want to use a single central repository for their configuration, the piper package provides this. It is similar to a singleton.

In all of the examples above, they demonstrate using piper in its singleton style approach.

Working with multiple pipers

You can also create many different pipers for use in your application. Each will have its own unique set of configurations and values. Each can read from a different config file. All of the functions that piper package supports are mirrored as methods on a piper.


x := piper.New()
y := piper.New()


// access viper
// p := x.V

When working with multiple pipers, it is up to the user to keep track of the different pipers.


Sometimes you may find viper a little slow, because viper need to check in the following order on Get: flag, env, config file, key/value store. If you have confidence some of your configs won't change, you can use piper's IGet* methods. Piper will build a configs cache on Load, and those IGet* methods will get config from cache directly.

goos: darwin
goarch: amd64
pkg: github.com/Yiling-J/piper/integration
cpu: Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz

BenchmarkGet-12           895533              1278 ns/op
BenchmarkIGet-12        19141876                61.35 ns/op