RubyDay 2016 - Can we still innovate?

CAN WE STILL INNOVATE?
PIOTR SOLNICA RubyDay 2016, Florence

mojotech.com
rom-rb.org dry-rb.org

COMPLEXITY
ᕕ(˵•̀෴•́˵)ᕗ

“FAT MODEL” WAS
AN AMBIGUOUS IDEA

SEPARATING PERSISTENCE
FROM DOMAIN LOGIC

MAYBE A FULL-BLOWN
DATAMAPPER PATTERN?

NEW ABSTRACTIONS
▸ Form objects
▸ Value objects
▸ Service objects
▸ Query objects
▸ Policy objects
▸ View objects

ROM-RB INNOVATIONS
▸ Read/Write separation
▸ Relations as ﬁrst-class citizens
▸ Data transformations instead of object mutations
▸ Explicit relation composition instead of ambiguous query
DSLs
▸ Command API for custom db-speciﬁc database operations
▸ Externalized data validations

cfg = ROM::Configuration.new(:sql, 'sqlite::memory')
cfg.gateways[:default].create_table(:users) do
primary_key :id
column :name, String, null: false
column :age, Integer, null: true
end
cfg.gateways[:default].create_table(:tasks) do
primary_key :id
foreign_key :user_id, :users, null: false, on_delete: :cascade
column :title, String, null: true
end

class Users < ROM::Relation[:sql]
schema(infer: true)
end
class Tasks < ROM::Relation[:sql]
schema(infer: true)
end
SETS UP RELATION
ATTRIBUTES BASED ON DATABASE
SCHEMA SCHEMA
INFERS TABLE NAME FROM CLASS
NAME

rom = ROM.container(cfg)
NO GLOBAL
POLLUTION

class UserRepo < ROM::Repository[:users]
commands :create
end

user_repo = UserRepo.new(rom)
user_repo.create(name: 'Jane', age: 33)
# => #<ROM::Struct[User] id=3 name="Jane" age=33>

commands :create, update: :by_pk
end

user_repo = UserRepo.new(rom)
user = user_repo.create(name: 'Jane', age: 33)
changeset = user_repo.changeset(user.id, name: 'Jane Doe')

changeset.diff
# => {:name=>"Jane Doe"}
user_repo.update(user.id, changeset)
# => #<ROM::Struct[User] id=1 name="Jane Doe" age=33>

class Users < ROM::Relation[:sql]
schema(infer: true) do
associations do
has_many :tasks
end
end
end

relations :tasks
def create_with_tasks(user)
command(:create, aggregate(:tasks)).call(user)
end
end

user_repo.aggregate(:tasks)
# [#<ROM::Struct[User] id=3 name="Jane" age=33
tasks=[#<ROM::Struct[Task] id=5 user_id=3 title="Jane task 1">,
#<ROM::Struct[Task] id=6 user_id=3 title="Jane task 2">]>>

require 'dry-struct'
class Task < Dry::Struct
attribute :title, Types::Strict::String
end
class User < Dry::Struct
attribute :name, Types::Strict::String
attribute :age, Types::Strict::Int
attribute :tasks, Types::Strict::Array.member(Task)
end

user_repo.aggregate(:tasks).as(User).to_a
# => [#<User name="Jane" age=33 tasks=[#<Task title="Jane
task 1">, #<Task title="Jane task 2">]>, #<User
name="John" age=32 tasks=[#<Task title="John task 1">,
#<Task title="John task 2">]>]

DRY-RB INNOVATIONS
▸ Advanced data validation
▸ Flexible type system with coercions and constraints
▸ Composable transactions with error handling based on
monads
▸ Simple yet powerful Dependency Injection
▸ Application state management with component lifecycles

attribute :age, Types::Strict::Int
end

module Types
include Dry::Types.module
Age = Types::Strict::Int.constrained(gt: 18)
end
attribute :age, Types::Age
end

User.new(name: 'John', age: 17)
# [User.new] 17 (Fixnum) has invalid type for :age
(Dry::Struct::Error)

UserSchema = Dry::Validation.Schema do
required(:login).value(:str?, size?: 2..64)
required(:age).maybe(:int?, gt?: 18)
end
IS IT A STRING WITH A SIZE
BETWEEN 2 AND 64 CHARS?
CAN BE NIL BUT
IF IT’S NOT THEN IS IT AN
INTEGER AND IS IT GREATER
THAN 18?

UserSchema.(login: "j", age: 19).errors
# {:login=>["length must be within 2 - 64"]}
UserSchema.(login: "jane", age: nil).errors
# {}
SHORTCUT FOR `#CALL` :)

AddressSchema = Dry::Validation.Schema do
required(:street).value(:str?)
required(:city).filled(:str?)
required(:zipcode).filled(:int?)
end
UserSchema = Dry::Validation.Schema do
required(:login).value(:str?, size?: 2..64)
required(:age).maybe(:int?, gt?: 18)
required(:address).schema(AddressSchema)
end

UserSchema.(login: 'jane', age: 'oops').errors
# {:age=>["must be an integer"]}

UserSchema.(login: 'jane', age: 'oops').hints
# {:age=>["must be greater than 18"]}

require 'dry/system/container'
class Application < Dry::System::Container
configure do |config|
config.root = Pathname.new('./my/app')
end
end
# now you can register a logger
require 'logger'
Application.register('utils.logger', Logger.new($stdout))
# and access it
Application['utils.logger']

require 'dry/system/container'
class Application < Dry::System::Container
config.root = '/my/app'
config.auto_register = 'lib'
end
load_paths!('lib')
end

# under /my/app/lib/logger.rb we put
class Logger
# some neat logger implementation
end
# we can finalize the container which triggers auto-registration
Application.finalize!
# the logger becomes available
Application[‘logger']

# system/boot/persistence.rb
Application.finalize(:persistence) do
init do
require '3rd-party/database'
end
start do
container.register('database') do
3rdParty::Database.new
end
end
stop do
container['database'].disconnect
end
end

class Application < Dry::Component::Container
config.root = '/my/app'
end
end
# requires needed files and boots your persistence component
Application.boot!(:persistence)
# and now `database` becomes available
Application['database']

# lib/worker.rb
require ‘application/import'
class Worker
include Import['database']
def do_stuff
users = database.read('select * from users')
# ..
end
end
THIS WILL INJECT “DATABASE”
OBJECT AND MAKE IT AVAILABLE

LET’S CONTINUE
EXPLORING & INNOVATING

RubyDay 2016 - Can we still innovate?

More Related Content

Viewers also liked (12)

Similar to RubyDay 2016 - Can we still innovate? (20)

Recently uploaded (20)

RubyDay 2016 - Can we still innovate?