10 KiB
Getting Started
Creating an application
For information on creating a new Elixir application, see this guide.
mix new my_app
For the finished example, see this example.
With Phoenix
The next guide will show you how to create a new Phoenix application and copy your resources and APIs over. However, if you know that you will be building a web application and would like to start with Phoenix, you can replace the above command with:
mix phx.new my_app --no-html --no-webpack --no-gettext
Add Ash
Add ash
to your dependencies in mix.exs
. The latest version can be found by running mix hex.info ash
.
# in mix.exs
def deps() do
[
{:ash, "~> x.x.x"}
]
end
If you want to have a more idiomatic formatting (the one used in this documentation) of your Ash resource and APIs,
you need to add Ash (and any other extensions you use) to your .formatter.exs
otherwise the default Elixir formatter will wrap portions of the DSL in parenthesis.
import_deps: [
:ash
]
Without that, instead of:
attribute :id, :integer, allow_nil?: true
the Elixir formatter will change it to:
attribute(:id, :integer, allow_nil?: true)
Create an Ash API
Create an API module. This will be your primary way to interact with your Ash resources. We recommend lib/my_app/api.ex
for simple setups. For more information on organizing resources into contexts/domains, see the "Contexts and Domains" guide.
# lib/my_app/api.ex
defmodule MyApp.Api do
use Ash.Api
resources do
end
end
Create a resource
A resource is the primary entity in Ash. Your Api module ties your resources together and gives them an interface, but the vast majority of your configuration will live in resources. In your typical setup, you might have a resource per database table. For those already familiar with Ecto, a resource and an Ecto schema are very similar. In fact, all resources define an Ecto schema under the hood. This can be leveraged when you need to do things that are not yet implemented or fall outside of the scope of Ash. The current recommendation for where to put your resources is in lib/my_app/resources/<resource_name>.ex
. Here are a few examples:
# in lib/my_app/resources/tweet.ex
defmodule MyApp.Tweet do
use Ash.Resource
attributes do
uuid_primary_key :id
attribute :body, :string do
allow_nil? false
constraints max_length: 255
end
# Alternatively, you can use the keyword list syntax
# You can also set functional defaults, via passing in a zero
# argument function or an MFA
attribute :public, :boolean, allow_nil?: false, default: false
# This is set on create
create_timestamp :created_at
# This is updated on all updates
update_timestamp :updated_at
# `create_timestamp` above is just shorthand for:
# attribute :created_at, :utc_datetime,
# writable?: false,
# default: &DateTime.utc_now/0
end
end
# in lib/my_app/resources/user.ex
defmodule MyApp.User do
use Ash.Resource
attributes do
attribute :email, :string,
allow_nil?: false,
constraints: [
match: ~r/^[\w.!#$%&’*+\-\/=?\^`{|}~]+@[a-zA-Z0-9-]+(\.[a-zA-Z0-9-]+)*$/i
]
uuid_primary_key :id
end
end
Add resources to your API
Alter your API in api.ex
like so:
resources do
resource MyApp.User
resource MyApp.Tweet
end
Test the resources
Now you should be able to create changesets for your resources
iex(7)> change = Ash.Changeset.new(MyApp.User, %{email: "ash.man@enguento.com"})
#Ash.Changeset<
action_type: :create,
attributes: %{email: "ash.man@enguento.com"},
relationships: %{},
errors: [],
data: %MyApp.User{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
calculations: %{},
email: nil,
id: nil
},
valid?: true
>
If you try to use an invalid email (the email regex is for demonstration purposes only) an error will be displayed as shown:
iex(6)> change = Ash.Changeset.new(MyApp.User, %{email: "@eng.com"})
#Ash.Changeset<
action_type: :create,
attributes: %{},
relationships: %{},
errors: [
%Ash.Error.Changes.InvalidAttribute{
class: :invalid,
field: :email,
message: {"must match the pattern %{regex}",
[
regex: "~r/^[\\w.!#$%&’*+\\-\\/=?\\^`{|}~]+@[a-zA-Z0-9-]+(\\.[a-zA-Z0-9-]+)*$/i"
]},
path: [],
stacktrace: #Stacktrace<>
}
],
data: %MyApp.User{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
calculations: %{},
email: nil,
id: nil
},
valid?: false
>
Add your data_layer
To be able to store and later on read your resources, a data layer is required. For more information, see the documentation for the data layer you would like to use. The currently supported data layers are listed below:
Storage | Datalayer | Documentation | Storage Documentation |
---|---|---|---|
postgres | AshPostgres.DataLayer | Documentation | Storage Documentation |
ets | Ash.DataLayer.Ets | Documentation | Storage Documentation |
mnesia | Ash.DataLayer.Mnesia | Documentation | Storage Documentation |
csv | AshCsv.DataLayer | Documentation | Storage Documentation |
To add a data_layer
, add it to the use Ash.Resource
statement. In this case we are going to use ETS
which is a in memory data layer good enough for testing purposes. Also we will make the ETS private so Read/Write limited
to owner process.
# in both lib/my_app/resources/user.ex
# and lib/my_app/resources/tweet.ex
use Ash.Resource, data_layer: Ash.DataLayer.Ets
Add actions to enable functionality
Actions are the primary driver for adding specific interactions to your resource. You can read the actions section to learn how to customize the functionality. For now we will enable all of them with default implementations by adding the following block to your resources:
# in both lib/my_app/resources/user.ex
# and lib/my_app/resources/tweet.ex
actions do
create :default
read :default
update :default
destroy :default
end
Test functionality
Now you should be able to use your API to do CRUD operations on your resources.
Create resource
iex(1)> user_changeset = Ash.Changeset.new(MyApp.User, %{email: "ash.man@enguento.co
m"})
#Ash.Changeset<
action_type: :create,
attributes: %{email: "ash.man@enguento.com"},
relationships: %{},
errors: [],
data: %MyApp.User{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
calculations: %{},
email: nil,
id: nil
},
valid?: true
>
iex(2)> MyApp.Api.create(user_changeset)
{:ok,
%MyApp.User{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
calculations: %{},
email: "ash.man@enguento.com",
id: "2642ca11-330b-4a07-83c7-b0e9ef391df6"
}}
List and Read a resource
iex(3)> MyApp.Api.read MyApp.User
{:ok,
[
%MyApp.User{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
calculations: %{},
email: "ash.man@enguento.com",
id: "2642ca11-330b-4a07-83c7-b0e9ef391df6"
}
]}
iex(4)> MyApp.Api.get(MyApp.User, "ash.man@enguento.com")
{:ok,
%MyApp.User{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
calculations: %{},
email: "ash.man@enguento.com",
id: "2642ca11-330b-4a07-83c7-b0e9ef391df6"
}}
Add relationships
With our resources stored in a data layer we can move on
to create relationships between them. In this case we will
specify that a User
can have many Tweets
- this implies that
a Tweet
belongs to a specific User
.
# in lib/my_app/resources/user.ex
relationships do
has_many :tweets, MyApp.Tweet, destination_field: :user_id
end
# in lib/my_app/resources/tweet.ex
relationships do
belongs_to :user, MyApp.User
end
Test relationships
Now we can use the new relationship to create a Tweet
that belongs to a specific User
:
iex(8)> {:ok, user} = Ash.Changeset.new(MyApp.User, %{email: "ash.man@enguento.com"}) |> MyApp.Api.create()
{:ok,
%MyApp.User{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
calculations: %{},
email: "ash.man@enguento.com",
id: "0d7063f8-b07c-4d02-88b2-b671f1aa0ad9",
tweets: #Ash.NotLoaded<:relationship>
}}
iex(9)> MyApp.Tweet |> Ash.Changeset.new(%{body: "ashy slashy"}) |> Ash.Changeset.replace_relationship(:user, user) |> MyApp.Api.create()
{:ok,
%MyApp.Tweet{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
body: "ashy slashy",
calculations: %{},
created_at: ~U[2020-11-14 12:54:06Z],
id: "f0b0b9d5-832c-45c9-9313-5e3fb9f1af24",
public: false,
updated_at: ~U[2020-11-14 12:54:06Z],
user: %MyApp.User{
__meta__: #Ecto.Schema.Metadata<:built, "">,
__metadata__: %{},
aggregates: %{},
calculations: %{},
email: "ash.man@enguento.com",
id: "0d7063f8-b07c-4d02-88b2-b671f1aa0ad9",
tweets: #Ash.NotLoaded<:relationship>
},
user_id: "0d7063f8-b07c-4d02-88b2-b671f1aa0ad9"
}}
Add front end extensions
Now that the Elixir API is complete, you can move on to the next section to learn how to change the data_layer to PostgreSQL and expose it via a JSON API.
AshJsonApi
- can be used to build a spec compliant JSON:API.AshPostgres.DataLayer
- can be used to persist your resources to PostgreSQL.
See Ash documentation for the rest
Ash.Api
for what you can do with your resources.Ash.Query
for the kinds of queries you can make.Ash.Resource.Dsl
for the resource DSL documentation.Ash.Api.Dsl
for the API DSL documentation.