fittings¶ ↑

This is a fork of ModCloth's mc-settings gem which has gone unmaintained and has deprecation warnings. This modernizes that and will be maintained.

Install¶ ↑

Add to your Gemfile:

gem "fittings"

For Rails, create config/settings/default.yml as well as +config/settings/environments/{development,testing,production}.yml+.

Then, in config/environment.rb, before the call to initialize!:

Setting.load(path:  "#{Rails.root}/config/settings",
             files: ["default.yml", "environments/#{Rails.env}.yml"])

Why is it `Setting`?¶ ↑

This is a fork of the mc-settings gem that is intended to be backwards-compatible.

Summary ¶ ↑

This gem provides an easy and Capistrano-friendly way to manage application configuration across multiple environments, such as development, QA, staging, production, etc.

Applications typically rely on configuration settings, such as host names, URLs, usernames and many more. Some change between environments, some do not. This gem assumes that application configuration is represented by a Hash of arbitrary depth, and provides convenient and compact syntax to access the settings through a singleton instance inside Setting class.

Configuration is stored in one or more YAML files with the top-level data structure being a Hash, with keys being the names of individual settings. For example, consider the following sample application configuration file:

tax:
  default: 0.0
  california: 7.5
states:
  default: 
    - 'CA'
    - 'WA'
    - 'NY'
  ship_to:
    - 'CA'
    - 'NY'
math_pi: 3.14159526

Setting Gem provides Setting.load(..) method to load configuration from files in a way that allows some configuration files to override previously loaded values, and then offers a simple method API to access the values, for example Setting.tax(:california) or Setting.tax. Supporting default values in 2nd, 3rd, .. - level hashes is one of the advantages of using this gem.

By loading configuration from YAML files, Setting gem is inherently compatible with Capistrano deployment methodology, where a certain set of files may become “activated” by simply sym-linking them into the appropriate settings folder.

Note: using example above, “1st level” hash is the one with keys “tax”, “states” and “math_pi”.

2nd-level hash is, for example, the tax definition one, with keys “default” and “california”.

Usage in Code¶ ↑

Once configuration is initialized using Setting#load or Setting#reload methods (see below), they can be used in code in the following way:

Setting.key_name is optimized to return default value if available instead of a Hash.
Setting.key_name(:sub_key_name) returns a value from the 2nd level hash.
Setting.key_name(:sub_key_name, :sub_sub_key_name) returns value from the 3rd level hash if available. The algorithm is recursive, so only the maximum method stack depth will limit the number of nested hash values you can access this way.
Special syntax Setting, Setting[:sub_key_name], etc also supported. This method, however, does not support default values (see below).

Method notation is recommended over square bracket notation for accessing single values. However, square bracket notation may be useful when you want to fetch the entire 2nd level hash that includes the default value, instead of the default value itself.

For example, given the above YAML file, you can access the settings in your code as follows:

Setting.tax => 0.0
Setting.tax(:california) => 7.5
Setting.math_pi => 3.14159526
Setting[:math_pi] => 3.14159526
Setting.states => [ 'CA', 'WA', 'NY' ]
Setting.states['ship_to'] => [ 'CA', 'NY' ]

Method-calling notation allows passing an array of keys to fetch a value from a nested hash.

This method also supports returning a default value, stored against the “default” key.

Setting.tax => 0.0

Square bracket syntax returns the actual nested hash, without any regard for the default value:

Setting[:tax] => { 'default' => 0.0, 'california' => 7.5 }

Loading Settings¶ ↑

The gem should be initialized in your environment.rb (if using Rails), or in any other application initialization block. Setting.load() method is provided for loading settings, and it can be called only once in application lifecycle, or it will throw an exception. If you need to reload settings completely, you can use reload() method with similar arguments.

Consider an example:

Setting.load(:path  => "#{Rails.root}/config/settings",
             :files => ["default.yml", "environments/#{Rails.env}.yml"],
             :local => true)

The argument is an options hash that configures which YAML files to load, in what order, and from where.

path specifies the “root” folder where settings files will be loaded from
files is an array that lists file names relative to the :path. In the example above, settings folder contains subfolder “environments” where Rails-specific environment files are located (such as “development.yml”, “staging.yml”, “production.yml”, etc)
local can be optionally specified as a true value, and if specified Setting gem will load all *.yml files that live under the :path/local folder.

Below is list of YAML files loaded in order specified in the above example, assuming that “development” is the Rails environment, and “local” folder exists with 3 additional YAML files in it:

config/settings/default.yml
config/settings/environments/development.yml
config/settings/local/authorize-net.yml
config/settings/local/paypal.yml
config/settings/local/other.yml

Each YML file defines a ruby Hash. During file loading, the hashes are merged, so that values loaded in early files may be overwritten by values in subsequent files. This is deliberate and by design: it allows you to create small “override” files for each environment, or even each machine you want to deploy to. Exactly how you split your application settings in files is up to you.

Nested Hashes and Default Values¶ ↑

MC Setting gem provides a convenient way to access nested values, including full support for the default values within nested hashes (as of 0.1.1).

Consider the following nested hash example:

default.yml:

services: 
  inventory:
    url: http://ims.mycompany.com:3443/inventory_manager
    name: Inventory Management
  shipping:
    url: http://ship.mycompany.com:3443/shipper
    name: Shipping

Setting.load(:files => ['default.yml'], :path => ...)

Setting.services(:inventory)       => { :url => "http://localhost:3443/inventory_manager" :name => "Inventory Management"}
Setting.services(:inventory, :url) => "http://localhost:3443/inventory_manager"

staging.yml

We are changing URLs for services in staging.yml, so they work in the staging environment. Service URLs have been updated to use localhost:

services: 
  inventory:
    url: http://localhost:8009/inventory_manager
  shipping:
    url: http://localhost:8008/shipper

Setting.load(:files => ['default.yml', 'staging.yml'], :path => ...)

Setting.services(:inventory)       => { :url => "http://localhost:8009/inventory_manager" :name => "Inventory Management"}
Setting.services(:inventory, :url) => "http://localhost:8008/inventory_manager"

Capistrano Recommendations¶ ↑

Assume the directory structure of your Rails application is as follows:

config/settings/default.yml
config/settings/environments/development.yml
config/settings/environments/staging.yml
config/settings/environments/production.yml
config/settings/local
config/settings/systems/reporting.yml
config/settings/systems/admin.yml

Note that the “local” directory is empty, and that “systems” directory contains several YAML files that provide alternative configuration for a reporting server, and an admin server (both of which run in “production” rails environment).

When deploying to the main production site, neither YAML files inside “systems” folder are activated or used.

But upon deployment to the admin server, Capistrano could symlink “admin.yml” from config/settings/local folder, so the Setting gem would load these values. So for each Capistrano role, you can define which files need to be symlinked into local, thus creating a very flexible configuration scheme that's easily managed by Capistrano.