About Cake Pattern

About Cake Pattern.md

About Cake Pattern

Below are the sources I took notes and hope to learn about Cake Pattern from.

Videos:

• SF Scala, Peter Potts: Cake Pattern in Practice
• Cake Pattern: The Bakery from the Black Lagoon
• Dead-Simple Dependency Injection

Dead-Simple Dependency Injection

Dead-Simple Dependency Injection

• setUserPwd - a Scala method that updates some data in a database
• Version 1 - everything's hardcoded
• Everything = SQL driver + database URL
• It's self-contained, reusable, fine for little toy tests
• It's not how you wanna write your actual applications
• Version 2 - global factory
• Provides database connections
• What about closing a database connection?! Who's in charge? It's not clear whose responsibility it is to close a connection.
• Bad idea
  • Hidden dependency on a global object
  • Requires magic initialization step = you can't call this method before that object
• Inversion of control = pretentious way of saying "taking an argument"
  • That reminds me the other presentation about functional programming...
• Connection taken from ...FIXME
• Version 3 - Inversion of Control
• taking an argument = injecting c which is a database connection
• I don't care where this connection comes from - just accept that I have one and can use it.
• I don't care how it's discarded
• I just assume that the connection is open
• You may object and say "What if the thing that knows how to obtain and a connection is way up here "
• In the call graph it needs the connection. Do I need to pass the connection through? It can get pretty easily cumbersome.
• Version 4 - Currying
• Instead of passing in the connection right away, simply partially apply
• Instead of returning Unit, I'm going to return a function that accepts a connection at convenient time
• The only difference is that I'm delaying the actual passing-in a connection
• Design patten - Connection Reader
• Capture this idea in a data type - just for convenience (as usual, isn't it?)
• case class DB[A](g: Connection => A)
• A wrapper around the type from Connection to A
• DB[A] = a database action = Given a Connection produce some value of type A
• apply to have a convenient call
• Lift existing functions using map
  • map[B](f: A => B): DB[B] ~ (A => B) => (DB[A] => DB[B])
  • Map (or promote) an existing function A => B to DB[A] => DB[B]
  • Implementation of map is just a function composition
• Combine two actions
  • I have an action that produces DB[A] and you call flatMap with a function that