DSL extensions for configuring ScyllaDB-specific options on Ash resources.

Supports Ash Framework 3.0+ features including base_filter, identities, aggregates, calculations, preparations, changes, validations, pipelines, multitenancy, code_interface, and extended action options.

Usage

defmodule MyApp.MyResource do
  use Ash.Resource,
    data_layer: AshScylla.DataLayer

  ash_scylla do
    table "my_table"
    keyspace "my_keyspace"
    consistency :quorum
    ttl 3600
    lwt true

    base_filter [status: "active"]
    default_context %{tenant: "org_123"}
    description "My resource description"

    secondary_index :email
    secondary_index [:name, :age]
    secondary_index :status, name: "idx_user_status"

    materialized_view :users_by_email,
      primary_key: [:email, :id],
      include_columns: [:name, :age]

    pagination :token
    per_action_consistency read: :one, create: :quorum
  end

  attributes do
    uuid_primary_key :id
    uuid_v7_primary_key :vid
    integer_primary_key :seq_id
    create_timestamp :inserted_at
    update_timestamp :updated_at
    attribute :name, :string, public?: true, writable?: true
    attribute :email, :string, sensitive?: false
    attribute :status, :string
    attribute :age, :integer
  end

  identities do
    identity :unique_email, [:email]
  end

  aggregates do
    count :total_count
    count :active_count, filter: [status: "active"]
  end

  calculations do
    calculate :display_name, :string, expr(name)
  end

  preparations do
    prepare build(:load, [:email])
  end

  changes do
    change fn changeset, _context -> changeset end
  end

  validations do
    validate attribute_equals(:status, "active")
  end

  pipelines do
    pipe_through :read
  end

  multitenancy do
    strategy :attribute
    attribute :org_id
  end

  code_interface do
    define :create_user
    define_calculation :active_count
  end

  relationships do
    belongs_to :organization, MyApp.Organization
    has_one :profile, MyApp.Profile
    has_many :posts, MyApp.Post
    many_to_many :tags, MyApp.Tag
  end

  actions do
    create :create do
      accept [:name, :email, :status]
      argument :organization_id, :uuid
      change fn changeset, _context -> changeset end
      validate present([:name])
    end

    read :read do
      prepare build(:load, [:email])
      pagination offset?: true, max_page_size: 100
      metadata :total_count, :integer
      filter [status: "active"]
    end

    update :update do
      accept [:name, :status]
      change fn changeset, _context -> changeset end
      validate present([:name])
    end

    destroy :destroy do
      soft? true
      change fn changeset, _context -> changeset end
    end
  end
end

Options

• :table - The table name in ScyllaDB (overrides default)
• :keyspace - The keyspace to use (overrides repo default)
• :consistency - The consistency level for reads/writes
• :ttl - Default TTL for inserted records (in seconds)
• :lwt - Enable Lightweight Transactions (LWT) for atomic upserts using INSERT ... IF NOT EXISTS (default: false)
• :secondary_index - Define secondary indexes for non-primary key columns
• :materialized_view - Define materialized views with different primary key structure
• :pagination - Pagination mode: :offset (default) or :token for token-based pagination
• :per_action_consistency - Per-action consistency overrides as a keyword list, e.g. [read: :one, create: :quorum]
• :base_filter - A filter expression applied to all queries on this resource (Ash 3.0)
• :default_context - Default context map merged into every query/changeset (Ash 3.0)
• :description - Human-readable description of the resource (Ash 3.0)
• :identity - Define unique identity constraints for upsert operations
• :aggregate - Define aggregate queries (count, sum, avg, min, max)
• :calculation - Define expression-based calculations
• :preparation - Define query preparations
• :change - Define changes applied to changesets
• :validation - Define attribute validations
• :pipeline - Define action pipelines via pipe_through
• :multitenancy - Configure multitenancy strategy (:context or :attribute)
• :code_interface - Define code interface functions
• :relationship - Define relationships (belongs_to, has_one, has_many, many_to_many)
• :action - Define actions with extended options (accept, argument, change, validate, prepare, pagination, metadata, filter, soft?)

Features

The data layer supports:

• Upsert (:upsert) - Insert-or-update semantics with optional LWT
• Atomic updates ({:atomic, :update}) - LWT-based conditional updates
• Atomic upserts ({:atomic, :upsert}) - LWT-based insert-or-update
• Bulk update/destroy - update_query and destroy_query for filtered operations
• Distinct - On partition-key columns only
• Aggregates - COUNT via :count aggregate
• Expression calculations - In-memory post-processing
• Boolean filter - With OR-to-IN rewriting
• Base filter - Automatic filter applied to all queries (Ash 3.0)
• Default context - Context merged into all queries (Ash 3.0)
• Identities - Unique constraints for upsert operations (Ash 3.0)
• Multitenancy - Context-based and attribute-based strategies (Ash 3.0)