Elixir SDK Reference
ConfigCat Elixir SDK on GitHub
This documentation applies to the v3.x version of the ConfigCat Elixir SDK. For the documentation of the latest release, please refer to this page.
Getting started
1. Add configcat
to your list of dependencies in mix.exs
def deps do
[
{:configcat, "~> 3.0.0"}
]
end
2. Add ConfigCat
to your application Supervisor tree
def start(_type, _args) do
children = [
{ConfigCat, [sdk_key: "#YOUR-SDK-KEY#"]},
MyApp
]
opts = [strategy: :one_for_one, name: MyApp.Supervisor]
Supervisor.start_link(children, opts)
end
3. Get your setting value
isMyAwesomeFeatureEnabled = ConfigCat.get_value("isMyAwesomeFeatureEnabled", false)
if isMyAwesomeFeatureEnabled do
do_the_new_thing()
else
do_the_old_thing()
end
Setting up the ConfigCat Client
ConfigCat Client is responsible for:
- managing the communication between your application and ConfigCat servers.
- caching your setting values and feature flags.
- serving values quickly in a failsafe way.
{ConfigCat, options}
returns a client with default options.
Properties | Description |
---|---|
sdk_key | REQUIRED. SDK Key to access your feature flags and configurations. Get it from ConfigCat Dashboard. |
base_url | Sets the CDN base url (forward proxy, dedicated subscription) from where the SDK will download the config JSON. |
data_governance | Describes the location of your feature flag and setting data within the ConfigCat CDN. This parameter needs to be in sync with your Data Governance preferences. Defaults to :global . More about Data Governance. Available options: :global , :eu_only . |
cache_policy | CachePolicy.auto/1 , CachePolicy.lazy/1 and CachePolicy.manual/0 . Defaults to: CachePolicy.auto/0 See See below for details. |
cache | Caching module you want configcat to use. Defaults to: ConfigCat.InMemoryCache . More about cache. |
http_proxy | Specify this option if you need to use a proxy server to access your ConfigCat settings. You can provide a simple URL, like https://my_proxy.example.com or include authentication information, like https://user:password@my_proxy.example.com/ . |
connect_timeout_milliseconds | Timeout for establishing a TCP or SSL connection, in milliseconds. Default is 8000. |
read_timeout_milliseconds | Timeout for receiving an HTTP response from the socket, in milliseconds. Default is 5000. |
flag_overrides | Local feature flag & setting overrides. More about feature flag overrides |
default_user | Sets the default user. More about default user. |
offline | Defaults to false . Indicates whether the SDK should be initialized in offline mode. More about offline mode. |
hooks | Used to subscribe events that the SDK sends in specific scenarios. More about hooks. |
name | A unique identifier for this instance of ConfigCat . Defaults to ConfigCat . Must be provided if you need to run more than one instance of ConfigCat in the same application. If you provide a name , you must then pass that name to all of the API functions using the client option. More about multiple instances. |
Anatomy of get_value()
Parameters | Description |
---|---|
key | REQUIRED. The key of a specific setting or feature flag. Set on ConfigCat Dashboard for each setting. |
default_value | REQUIRED. This value will be returned in case of an error. |
user | Optional, ConfigCat.User Object. Essential when using Targeting. Read more about Targeting. |
client | If you are running multiple instances of ConfigCat , provide the client: :unique_name option, specifying the name of the instance which you want to access. |
value = ConfigCat.get_value(
"keyOfMySetting", # Setting Key
false, # Default value
ConfigCat.User.new("#UNIQUE-USER-IDENTIFIER#") # Optional User Object
)
Anatomy of get_value_details()
get_value_details()
is similar to get_value()
but instead of returning the evaluated value only, it gives more detailed information about the evaluation result.
Parameters | Description |
---|---|
key | REQUIRED. The key of a specific setting or feature flag. Set on ConfigCat Dashboard for each setting. |
default_value | REQUIRED. This value will be returned in case of an error. |
user | Optional, User Object. Essential when using Targeting. Read more about Targeting. |
client | If you are running multiple instances of ConfigCat , provide the client: :unique_name option, specifying the name of the instance which you want to access. |
details = ConfigCat.get_value_details(
"keyOfMySetting", # Setting Key
false, # Default value
ConfigCat.User.new("#UNIQUE-USER-IDENTIFIER#") # Optional User Object
)
The details
result contains the following information:
Field | Description |
---|---|
value | The evaluated value of the feature flag or setting. |
key | The key of the evaluated feature flag or setting. |
default_value? | True when the default value passed to get_value_details() is returned due to an error. |
error | In case of an error, this field contains the error message. |
user | The User Object that was used for evaluation. |
matched_evaluation_percentage_rule | If the evaluation was based on a percentage rule, this field contains that specific rule. |
matched_evaluation_rule | If the evaluation was based on a Targeting Rule, this field contains that specific rule. |
fetch_time | The last download time (UTC DateTime) of the current config. |
User Object
The User Object is essential if you'd like to use ConfigCat's Targeting feature.
user_object = ConfigCat.User.new("#UNIQUE-USER-IDENTIFIER#")
user_object = ConfigCat.User.new("[email protected]")
Parameters | Description |
---|---|
identifier | REQUIRED. Unique identifier of a user in your application. Can be any string value, even an email address. |
email | Optional parameter for easier Targeting Rule definitions. |
country | Optional parameter for easier Targeting Rule definitions. |
custom | Optional Map of strings representing the custom attributes of a user for advanced Targeting Rule definitions. e.g. User role, Subscription type. |
user_object = ConfigCat.User.new("#UNIQUE-USER-IDENTIFIER#", email: "john@example", country: "United Kingdom",
custom: %{SubscriptionType: "Pro", UserRole: "Admin"})
Default user
There's an option to set a default User Object that will be used at feature flag and setting evaluation. It can be useful when your application has a single user only, or rarely switches users.
You can set the default User Object either on SDK initialization:
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
default_user: ConfigCat.User.new("[email protected]")
]}
or with the set_default_user
method of the ConfigCat client.
ConfigCat.set_default_user(ConfigCat.User.new("[email protected]"))
Whenever the get_value
, get_value_details
, get_variation_id
, get_all_variation_ids
, or get_all_values
methods are called without an explicit user
parameter, the SDK will automatically use the default user as a User Object.
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
default_user: ConfigCat.User.new("[email protected]")
]}
# The default user will be used in the evaluation process.
value = ConfigCat.get_value("keyOfMySetting", false)
When the user
parameter is specified on the requesting method, it takes precedence over the default user.
other_user = ConfigCat.User.new("[email protected]")
# otherUser will be used in the evaluation process.
value = ConfigCat.get_value("keyOfMySetting", false, other_user)
For deleting the default user, you can do the following:
ConfigCat.clear_default_user()
Polling Modes
The ConfigCat SDK supports 3 different polling mechanisms to acquire the setting values from ConfigCat. After latest setting values are downloaded, they are stored in the internal cache, then all get_value()
calls are served from there. With the following polling modes, you can customize the SDK to best fit to your application's lifecycle.
More about polling modes.
Auto polling (default)
The ConfigCat SDK downloads the latest values and stores them automatically every 60 seconds.
Use the poll_interval_seconds
option parameter to change the polling interval.
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
cache_policy: ConfigCat.CachePolicy.auto(poll_interval_seconds: 60)
]},
Available options:
Option Parameter | Description | Default |
---|---|---|
poll_interval_seconds | Polling interval. | 60 |
max_init_wait_time_seconds | Maximum waiting time between the client initialization and the first config acquisition in secconds. | 5 |
Lazy loading
When calling get_value()
, the ConfigCat SDK downloads the latest setting values if they are not present or expired in the cache. In this case get_value()
will return the setting value after the cache is updated.
Use cache_refresh_interval_seconds
option parameter to set cache lifetime.
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
cache_policy: ConfigCat.CachePolicy.lazy(cache_refresh_interval_seconds: 300)
]}
Available options:
Option Parameter | Description | Default |
---|---|---|
cache_refresh_interval_seconds | Cache TTL. | 60 |
Manual polling
Manual polling gives you full control over when the config JSON
(with the setting values) is downloaded. ConfigCat SDK will not update them automatically. Calling force_refresh()
is your application's responsibility.
ConfigCat.force_refresh()
get_value()
returnsdefault_value
if the cache is empty. Callforce_refresh()
to update the cache.
value = ConfigCat.get_value("key", "my default value") # Returns "my default value"
ConfigCat.force_refresh()
value = ConfigCat.get_value("key", "my default value") # Returns "value from server"
Custom cache behaviour with cache:
option parameter
The ConfigCat SDK stores the downloaded config data in a local cache to minimize network traffic and enhance client performance.
If you prefer to use your own cache solution, such as an external or distributed cache in your system,
you can implement the ConfigCache
behaviour
and provide the cache
option when initializing the SDK.
This allows you to integrate ConfigCat with your existing caching infrastructure seamlessly.
To be able to customize the caching layer, you need to implement the ConfigCat.ConfigCache
behaviour:
defmodule MyApp.CustomConfigCache do
alias ConfigCat.ConfigCache
@behaviour ConfigCache
@impl ConfigCache
def get(cache_key) do
# here you have to return with the cached value
end
@impl ConfigCache
def set(cache_key, value) do
# here you have to store the new value in the cache
end
end
Then use your custom cache implementation:
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
cache: MyApp.CustomConfigCache
]}
The Elixir SDK supports shared caching. You can read more about this feature and the required minimum SDK versions here.
Multiple ConfigCat
instances
If you need to run more than one instance of ConfigCat
, there are two ways
you can do it.
Module-Based
You can create a module that uses ConfigCat and then call the ConfigCat
API functions on that module. This is the recommended option, as it makes the
calling code a bit clearer and simpler.
You can pass any of the options listed above as arguments to use ConfigCat
or specify them in your supervisor. Arguments specified by the supervisor take
precedence over those provided to use ConfigCat
.
# lib/my_app/first_flags.ex
defmodule MyApp.FirstFlags do
use ConfigCat, sdk_key: "sdk_key_1"
end
# lib/my_app/second_flags.ex
defmodule MyApp.SecondFlags do
use ConfigCat, sdk_key: "sdk_key_2"
end
# lib/my_app/application.ex
def start(_type, _args) do
children = [
# ... other children ...
FirstFlags,
SecondFlags,
]
opts = [strategy: :one_for_one, name: MyApp.Supervisor]
Supervisor.start_link(children, opts)
end
# Calling code:
FirstFlags.get_value("someKey", "default value")
SecondFlags.get_value("otherKey", "other default")
Explicit Client
If you prefer not to use the module-based solution, you can instead add
multiple ConfigCat
children to your application's supervision tree. You will
need to give ConfigCat
a unique name
option for each, as well as using
Supervisor.child_spec/2
to provide a unique id
for each instance.
When calling the ConfigCat API functions, you'll pass a client:
keyword
argument with the unique name
you gave to that instance.
# lib/my_app/application.ex
def start(_type, _args) do
children = [
# ... other children ...
Supervisor.child_spec({ConfigCat, [sdk_key: "sdk_key_1", name: :first]}, id: :config_cat_1),
Supervisor.child_spec({ConfigCat, [sdk_key: "sdk_key_2", name: :second]}, id: :config_cat_2),
]
opts = [strategy: :one_for_one, name: MyApp.Supervisor]
Supervisor.start_link(children, opts)
end
# Calling code:
ConfigCat.get_value("someKey", "default value", client: :first)
ConfigCat.get_value("otherKey", "other default", client: :second)
Hooks
With the following hooks you can subscribe to particular events fired by the SDK:
-
on_client_ready
: This event is sent when the SDK reaches the ready state. If the SDK is set up with lazy load or manual polling it's considered ready right after instantiation. If it's using auto polling, the ready state is reached when the SDK has a valid config JSON loaded into memory either from cache or from HTTP. If the config couldn't be loaded neither from cache nor from HTTP theon_client_ready
event fires when the auto polling'smax_init_wait_time_seconds
is reached. -
on_config_changed(config: map())
: This event is sent when the SDK loads a valid config JSON into memory from cache, and each subsequent time when the loaded config JSON changes via HTTP. -
on_flag_evaluated(evaluation_details: EvaluationDetails.t())
: This event is sent each time when the SDK evaluates a feature flag or setting. The event sends the same evaluation details that you would get fromget_value_details
. -
on_error(error: String.t())
: This event is sent when an error occurs within the ConfigCat SDK.
You can subscribe to these events either on SDK initialization:
def on_flag_evaluated(evaluation_details) do
# handle the event
end
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
hooks: [on_flag_evaluated: {__MODULE__, :on_flag_evaluated, []}]
]}
or with the Hooks
property of the ConfigCat client:
ConfigCat.Hooks.add_on_flag_evaluated({__MODULE__, :on_flag_evaluated, []})
A hook callback is either an anonymous function or a module/function name/extra_arguments tuple. Each callback is passed specific arguments. These specific arguments are prepended to the extra arguments provided in the tuple (if any). For example, you might want to define a callback that sends a message to another process which the config changes. You can pass the pid of that process as an extra argument:
def MyModule do
def subscribe_to_config_changes(subscriber_pid) do
ConfigCat.hooks()
|> ConfigCat.Hooks.add_on_config_changed({__MODULE__, :on_config_changed, [subscriber_pid]})
end
def on_config_changed(config, pid) do
send pid, {:config_changed, config}
end
end
Online / Offline mode
In cases when you'd want to prevent the SDK from making HTTP calls, you can put it in offline mode:
ConfigCat.set_offline()
In offline mode, the SDK won't initiate HTTP requests and will work only from its cache.
To put the SDK back in online mode, you can do the following:
ConfigCat.set_online()
With
ConfigCat.offline?
you can check whether the SDK is in offline mode.
Flag Overrides
With flag overrides you can overwrite the feature flags & settings downloaded from the ConfigCat CDN with local values. Moreover, you can specify how the overrides should apply over the downloaded values. The following 3 behaviours are supported:
-
Local only (
:local_only
): When evaluating values, the SDK will not use feature flags & settings from the ConfigCat CDN, but it will use all feature flags & settings that are loaded from local-override sources. -
Local over remote (
:local_over_remote
): When evaluating values, the SDK will use all feature flags & settings that are downloaded from the ConfigCat CDN, plus all feature flags & settings that are loaded from local-override sources. If a feature flag or a setting is defined both in the downloaded and the local-override source then the local-override version will take precedence. -
Remote over local (
:remote_over_local
): When evaluating values, the SDK will use all feature flags & settings that are downloaded from the ConfigCat CDN, plus all feature flags & settings that are loaded from local-override sources. If a feature flag or a setting is defined both in the downloaded and the local-override source then the downloaded version will take precedence.
You can set up the SDK to load your feature flag & setting overrides from a file or a map.
JSON File
The SDK can load your feature flag & setting overrides from a file.
File
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
flag_overrides: ConfigCat.LocalFileDataSource.new(
"path/to/the/local_flags.json", # path to the file
:local_only # local/offline mode
)
]}
JSON File Structure
The SDK supports 2 types of JSON structures to describe feature flags & settings.
1. Simple (key-value) structure
{
"flags": {
"enabledFeature": true,
"disabledFeature": false,
"intSetting": 5,
"doubleSetting": 3.14,
"stringSetting": "test"
}
}
2. Complex (full-featured) structure
This is the same format that the SDK downloads from the ConfigCat CDN. It allows the usage of all features that are available on the ConfigCat Dashboard.
You can download your current config JSON from ConfigCat's CDN and use it as a baseline.
The URL to your current config JSON is based on your Data Governance settings:
- GLOBAL:
https://cdn-global.configcat.com/configuration-files/{YOUR-SDK-KEY}/config_v5.json
- EU:
https://cdn-eu.configcat.com/configuration-files/{YOUR-SDK-KEY}/config_v5.json
{
"f": {
// list of feature flags & settings
"isFeatureEnabled": {
// key of a particular flag
"v": false, // default value, served when no rules are defined
"i": "430bded3", // variation id (for analytical purposes)
"t": 0, // feature flag's type, possible values:
// 0 -> BOOLEAN
// 1 -> STRING
// 2 -> INT
// 3 -> DOUBLE
"p": [
// list of percentage rules
{
"o": 0, // rule's order
"v": true, // value served when the rule is selected during evaluation
"p": 10, // % value
"i": "bcfb84a7" // variation id (for analytical purposes)
},
{
"o": 1, // rule's order
"v": false, // value served when the rule is selected during evaluation
"p": 90, // % value
"i": "bddac6ae" // variation id (for analytical purposes)
}
],
"r": [
// list of Targeting Rules
{
"o": 0, // rule's order
"a": "Identifier", // comparison attribute
"t": 2, // comparator, possible values:
// 0 -> 'IS ONE OF',
// 1 -> 'IS NOT ONE OF',
// 2 -> 'CONTAINS',
// 3 -> 'DOES NOT CONTAIN',
// 4 -> 'IS ONE OF (SemVer)',
// 5 -> 'IS NOT ONE OF (SemVer)',
// 6 -> '< (SemVer)',
// 7 -> '<= (SemVer)',
// 8 -> '> (SemVer)',
// 9 -> '>= (SemVer)',
// 10 -> '= (Number)',
// 11 -> '<> (Number)',
// 12 -> '< (Number)',
// 13 -> '<= (Number)',
// 14 -> '> (Number)',
// 15 -> '>= (Number)',
// 16 -> 'IS ONE OF (Hashed)',
// 17 -> 'IS NOT ONE OF (Hashed)'
"c": "@example.com", // comparison value
"v": true, // value served when the rule is selected during evaluation
"i": "bcfb84a7" // variation id (for analytical purposes)
}
]
}
}
}
Map
You can set up the SDK to load your feature flag & setting overrides from a map.
map = %{
"enabledFeature" => true,
"disabledFeature" => false,
"intSetting" => 5,
"doubleSetting" => 3.14,
"stringSetting" => "test"
}
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
flag_overrides: ConfigCat.LocalMapDataSource.new(map, :local_only)
]}
Logging
In the ConfigCat SDK, we use the default Elixir's Logger so you can customise as you like.
Debug level logging helps to inspect how a feature flag was evaluated:
[debug] Evaluating get_value('isPOCFeatureEnabled').
User object: %ConfigCat.User{country: nil, custom: %{}, email: "[email protected]", identifier: "435170f4-8a8b-4b67-a723-505ac7cdea92"}
Evaluating rule: [Email:[email protected]] [CONTAINS] [@something.com] => no match
Evaluating rule: [Email:[email protected]] [CONTAINS] [@example.com] => match, returning: true
The following example shows how to set the Log Level on the internal ConfigCat logger. Set the log level of the module with put_module_level/2 function. Put the following code into your application.ex file and run it on start:
defp set_config_cat_log_level do
:configcat
|> Application.spec(:modules)
|> Logger.put_module_level(:debug)
end
On Elixir 1.13 or later you can use put_application_level/2 function which is equivalent to the code above.
get_all_keys()
You can query the keys from your configuration in the SDK with the get_all_keys()
method.
keys = ConfigCat.get_all_keys()
get_all_values()
Evaluates and returns the values of all feature flags and settings. Passing a User Object is optional.
values = ConfigCat.get_all_values(
ConfigCat.User.new("#UNIQUE-USER-IDENTIFIER#") # Optional User Object
)
get_all_value_details()
Evaluates and returns the detailed values of all feature flags and settings. Passing a User Object is optional.
all_value_details = ConfigCat.get_all_value_details(
ConfigCat.User.new("#UNIQUE-USER-IDENTIFIER#") # Optional User Object
)
Using ConfigCat behind a proxy
Provide your own network credentials (username/password), and proxy server settings (proxy server/port) by passing the proxy details to the creator method.
{ConfigCat, [
sdk_key: "#YOUR-SDK-KEY#",
http_proxy: "https://user@pass:yourproxy.com"
]}