# Get Started If you are reading this on hexdocs, you may notice a few strange things. Primarily, that there are a lot of templates, like {{link:...}} This documentation is best viewed at [ash-hq.org](https://ash-hq.org) ## Goals In this guide we will: 1. Create a new Elixir application and add Ash as a dependency 2. Create a simple set of resources and show they can be used 3. Illustrate some core concepts of Ash 4. Point you to good next resources so you can explore Ash further ## Things you may want to read first - [Install Elixir](https://elixir-lang.org/install.html) - {{link:ash:guide:Philosophy}} ## Requirements If you want to follow along yourself, you will need the following things: 1. Elixir and Erlang installed 2. A text editor to make the changes that we make 3. A terminal to run the commands we show using `iex` ## Steps For this tutorial, we'll use examples based around creating a help desk. We will make the following resources: - `Helpdesk.Support.Ticket` - `Helpdesk.Support.Representative` The actions we will be able to take on these resources include: - Opening a new ticket - Closing a ticket - Assigning a ticket to a representative ### Create a new project We first create a new project with the `--sup` flag to add a supervision tree. This will be necessary for later steps. ```bash # In your terminal mix new --sup helpdesk && cd helpdesk ``` It is a good idea to make it a git repository and commit the initial project. You'll be able to see what changes we made, and can save your changes once we're done. ```bash # Run in your terminal git init git add -A git commit -m "init" ``` Open the project in your text editor, and we'll get started. ### Add Ash to your application Add the ash dependency to your `mix.exs` {{mix_dep:ash}} Add `:ash` to your `.formatter.exs` file ```elixir [ # import the formatter rules from ash import_deps: [:ash], inputs: [...] ] ``` And run `mix deps.get` If you are using ElixirLs (if you are using VScode, it is likely that you are), then add the following dependency to use Ash's custom autocomplete plugin. ```elixir {:elixir_sense, github: "elixir-lsp/elixir_sense", only: [:dev, :test]} ``` ### Creating our first resources The basic building blocks of an Ash application are resources. They are tied together by an API module (not to be confused with a web API), which will allow you to interact with those resources. Lets start by creating our first resource along with our first API. We will create the following files: - The API - `lib/helpdesk/support.ex` - A registry to list our resources - `lib/helpdesk/support/registry.ex` - Our tickets resource - `lib/helpdesk/support/resources/ticket.ex`. We also create an accompanying registry, in , which is where we will list the resources for our Api. To create the required folders and files, you can use the following command: ```bash # Run in your terminal mkdir -p lib/helpdesk/support/resources && touch $_/ticket.ex touch lib/helpdesk/support/registry.ex touch lib/helpdesk/support.ex ``` Your project structure should now look like this: ``` lib/ ├─ helpdesk/ │ ├─ support/ │ │ ├─ registry.ex │ │ ├─ resources/ │ │ │ ├─ ticket.ex │ ├─ support.ex ``` Add the following to the files we created ```elixir # lib/helpdesk/support/resources/ticket.ex defmodule Helpdesk.Support.Ticket do # This turns this module into a resource use Ash.Resource actions do # Add a set of simple actions. You'll customize these later. defaults [:create, :read, :update, :destroy] end # Attributes are the simple pieces of data that exist on your resource attributes do # Add an autogenerated UUID primary key called `:id`. uuid_primary_key :id # Add a string type attribute called `:subject` attribute :subject, :string end end ``` ```elixir # lib/helpdesk/support/registry.ex defmodule Helpdesk.Support.Registry do use Ash.Registry, extensions: [ # This extension adds helpful compile time validations Ash.Registry.ResourceValidations ] entries do entry Helpdesk.Support.Ticket end end ``` ```elixir # lib/helpdesk/support.ex defmodule Helpdesk.Support do use Ash.Api resources do # This defines the set of resources that can be used with this API registry Helpdesk.Support.Registry end end ``` ### Try our first resource out Run `iex -S mix` in your project and try it out To create a ticket, we first make an `Ash.Changeset` for the `:create` action of the `Helpdesk.Support.Ticket` resource. Then we pass it to the `create!/1` function on our API module `Helpdesk.Support`. ```elixir Helpdesk.Support.Ticket |> Ash.Changeset.for_create(:create) |> Helpdesk.Support.create!() ``` This returns what we call a `record` which is an instance of a resource. ```elixir #Helpdesk.Support.Ticket< ..., id: "c0f8dc32-a018-4eb4-8656-d5810118f4ea", subject: nil, ... > ``` ### Customizing our Actions One thing you may have noticed earlier is that we created a ticket without providing any input, and as a result our ticket had a `subject` of `nil`. Additionally, we don't have any other data on the ticket. Lets add a `status` attribute, ensure that `subject` can't be `nil`, and provide a better interface by making a custom action for opening a ticket, called `:open`. We'll start with the attribute changes: ```elixir # lib/helpdesk/support/resources/ticket.ex attributes do ... attribute :subject, :string do # Don't allow `nil` values allow_nil? false end # status is either `open` or `closed`. We can add more statuses later attribute :status, :atom do # Constraints allow you to provide extra rules for the value. # The available constraints depend on the type # See the documentation for each type to know what constraints are available # Since atoms are generally only used when we know all of the values # it provides a `one_of` constraint, that only allows those values constraints [one_of: [:open, :closed]] # The status defaulting to open makes sense default :open # We also don't want status to ever be `nil` allow_nil? false end end ``` And then add our customized action: ```elixir # lib/helpdesk/support/resources/ticket.ex actions do ... create :open do # By default you can provide all public attributes to an action # This action should only accept the subject accept [:subject] end end ``` Now we can play with these changes in iex: We use `create!` with an exclamation point here because that will raise the error which gives a nicer view of the error in `iex` ```elixir # Use this to pick up changes you've made to your code, or restart your session recompile() Helpdesk.Support.Ticket |> Ash.Changeset.for_create(:open, %{subject: "My mouse won't click!"}) |> Helpdesk.Support.create!() ``` And we can see our newly created ticket with a subject and a status. ```elixir #Helpdesk.Support.Ticket< ... id: "3c94d310-7b5e-41f0-9104-5b193b831a5d", status: :open, subject: "My mouse won't click!", ... > ``` If we didn't include a subject, or left out the input, we would see an error instead ```text ** (Ash.Error.Invalid) Input Invalid * attribute subject is required ``` ### Updates and validations Now lets add some logic to close a ticket. This time we'll add an `update` action. Here we will use a `change`. Changes allow you to customize how an action executes with very fine-grained control. There are built-in changes that are automatically available as functions, but you can define your own and pass it in as shown below. You can add multiple, and they will be run in order. See the {{link:ash:guide:Actions}} guides for more. ```elixir # lib/helpdesk/support/resources/ticket.ex actions do ... update :close do # We don't want to accept any input here accept [] change set_attribute(:status, :closed) # A custom change could be added like so: # # change MyCustomChange # change {MyCustomChange, opt: :val} end end ``` Now we can try it out in iex, opening a ticket and closing it: ```elixir # parenthesis so you can paste into iex ticket = ( Helpdesk.Support.Ticket |> Ash.Changeset.for_create(:open, %{subject: "My mouse won't click!"}) |> Helpdesk.Support.create!() ) ticket |> Ash.Changeset.for_update(:close) |> Helpdesk.Support.update!() #Helpdesk.Support.Ticket< ... status: :closed, subject: "My mouse won't click!", ... > ``` ### Querying without persistence So far, there is no persistence happening. All that this simple resource does is return the record back to us. You can see this lack of persistence by attempting to use a `read` action: ```elixir Helpdesk.Support.read!(Helpdesk.Support.Ticket) ``` Which will raise an error explaining that there is no data to be read for that resource. In order to add persistence, we need to add a data layer to our resources. Before we do that, however, lets go over how Ash allows us to work against many different data layers (or even no data layer at all). Resources without a data layer will implicitly be using `Ash.DataLayer.Simple`, which will just return structs and do no persistence. The way that we do this is by leveraging `context`, a free-form map available on queries and changesets. The simple data layer looks for `query.context[:data_layer][:data][resource]`. It provides a utility, `Ash.DataLayer.Simple.set_data/2` to set it. Try the following in iex. We will open some tickets, and close some of them, and then use `Ash.DataLayer.Simple.set_data/2` to use those tickets. ```elixir # Ash.Query is a macro, so it must be required require Ash.Query tickets = for i <- 0..5 do ticket = Helpdesk.Support.Ticket |> Ash.Changeset.for_create(:open, %{subject: "Issue #{i}"}) |> Helpdesk.Support.create!() if rem(i, 2) == 0 do ticket |> Ash.Changeset.for_update(:close) |> Helpdesk.Support.update!() else ticket end end # Show the tickets where the subject contains "2" Helpdesk.Support.Ticket |> Ash.Query.filter(contains(subject, "2")) |> Ash.DataLayer.Simple.set_data(tickets) |> Helpdesk.Support.read!() # Show the tickets that are closed and their subject does not contain "4" Helpdesk.Support.Ticket |> Ash.Query.filter(status == :closed and not(contains(subject, "4"))) |> Ash.DataLayer.Simple.set_data(tickets) |> Helpdesk.Support.read!() ``` The examples shown here could be implemented easily using things like `Enum.filter`, but the real power here is to allow you to use the same tools when working with any data layer. If you were using AshPostgres, the above code would be exactly the same, except for the call to `set_data/2`. ### Adding basic persistence Before we get into working with relationships, lets add some actual persistence to our resource. This will let us add relationships and try out querying data. There is a built in data layer that is good for testing and prototyping, that uses [ETS](https://elixir-lang.org/getting-started/mix-otp/ets.html). To add it to your resource, modify it like so: ```elixir use Ash.Resource, data_layer: Ash.DataLayer.Ets ``` Now we can slightly modify our code above, by removing the `Ash.DataLayer.Simple.set_data/2` calls, and we can see our persistence in action. Keep in mind, ETS is in memory, meaning restarting your application/iex session will remove all of the data. ```elixir require Ash.Query for i <- 0..5 do ticket = Helpdesk.Support.Ticket |> Ash.Changeset.for_create(:open, %{subject: "Issue #{i}"}) |> Helpdesk.Support.create!() if rem(i, 2) == 0 do ticket |> Ash.Changeset.for_update(:close) |> Helpdesk.Support.update!() end end # Show the tickets where the subject contains "2" Helpdesk.Support.Ticket |> Ash.Query.filter(contains(subject, "2")) |> Helpdesk.Support.read!() # Show the tickets that are closed and their subject does not contain "4" Helpdesk.Support.Ticket |> Ash.Query.filter(status == :closed and not(contains(subject, "4"))) |> Helpdesk.Support.read!() ``` ### Adding relationships Now we want to be able to assign a ticket to a representative. First, lets create the representative resource: ```elixir # lib/helpdesk/support/resources/representative.ex defmodule Helpdesk.Support.Representative do # This turns this module into a resource use Ash.Resource, data_layer: Ash.DataLayer.Ets actions do # Add the default simple actions defaults [:create, :read, :update, :destroy] end # Attributes are the simple pieces of data that exist on your resource attributes do # Add an autogenerated UUID primary key called `:id`. uuid_primary_key :id # Add a string type attribute called `:name` attribute :name, :string end relationships do # has_many means that the destination attribute is not unique, meaning many related records could exist. # We assume that the destination attribute is `representative_id` based # on the module name of this resource and that the source attribute is `id`. has_many :tickets, Helpdesk.Support.Ticket end end ``` And lets modify our tickets resource to have a relationship to the representative ```elixir # lib/helpdesk/support/resources/ticket.ex relationships do # belongs_to means that the destination attribute is unique, meaning only one related record could exist. # We assume that the destination attribute is `representative_id` based # on the name of this relationship and that the source attribute is `representative_id`. # We create `representative_id` automatically. belongs_to :representative, Helpdesk.Support.Representative end ``` Finally, lets add our new resource to our registry ```elixir # lib/helpdesk/support/registry.ex entries do ... entry Helpdesk.Support.Representative end ``` You may notice that if you don't add the resource to the registry, or if you don't add the `belongs_to` relationship, that you'll get helpful errors at compile time. Helpful compile time validations are a core concept of Ash. We focus as often as possible on ensuring that your application is valid before it compiles. ## Working with relationships There are a wide array of options when managing relationships, and going over all of them here wouldn't be reasonable. See the guide on {{link:ash:guide:Managing Relationships}} for a full explanation. For now, we'll show a simple example. Add the following action to allow us to assign a ticket to a representative. Here we also show the use of action arguments, the method by which you can accept additional input to an action. ```elixir # lib/helpdesk/support/resources/ticket.ex update :assign do # No attributes should be accepted accept [] # We accept a representative's id as input here argument :representative_id, :uuid do # This action requires representative_id allow_nil? false end # We use a change here to replace the related representative # If there is a different representative for this ticket, it will be changed to the new one # The representative itself is not modified in any way change manage_relationship(:representative_id, :representative, type: :append_and_remove) end ``` Lets try it out! ```elixir # Open a ticket ticket = ( Helpdesk.Support.Ticket |> Ash.Changeset.for_create(:open, %{subject: "I can't find my hand!"}) |> Helpdesk.Support.create!() ) # Create a representative representative = ( Helpdesk.Support.Representative |> Ash.Changeset.for_create(:create, %{name: "Joe Armstrong"}) |> Helpdesk.Support.create!() ) # Assign that representative ticket = ( ticket |> Ash.Changeset.for_update(:assign, %{representative_id: representative.id}) |> Helpdesk.Support.update!() ) ``` ### What next? What you've seen above constitutes some very simple usage of Ash, barely scratching the surface. In a lot of ways, it will look very similar to other tools that you've seen. If all that you ever used was the above, then realistically you won't see much benefit to using Ash over other options. Where Ash shines is in all of the additional turn-key tools that are built in, the ability to extend the framework yourself, and the consistent design patterns that enable unparalleled velocity, power and flexibility as your application and needs grow. #### Clean up your code that uses Ash? Creating and using changesets can be verbose. Check out the {{link:ash:guide:Code Interface}} to derive things like `Helpdesk.Support.Ticket.assign!(representative.id)` #### Persist your data See {{link:ash_postgres:guide:Get Started With Postgres:AshPostgres}} to see how to back your resources with postgres. This is highly recommended, as the postgres data layer provides tons of advanced capabilities. #### Add an API Check out the AshJsonApi and AshGraphql extensions to effortlessly build APIs around your resources #### Authorize access and work with users See the {{link:ash:guide:Policies}} guide for information on how to authorize access to your resources using actors and policies.