A subjective on Functional Programming

My points on why FP needs to spread.

Transcript

1. A subjective on Functional Programming Robin Palotai @bootcode http://ron.shoutboot.com August 13, 2012 Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
2. Foreword Have the following ever occured to you? Software we write is buggy Errors are rather logged than handled Similar patterns of code are written over and over again Existing code is not easily adapted We use frameworks with tons of xml conguration and go insane Are you fed up? Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
3. Types are powerful Theorem CurryHoward isomorphism: A proof is a program, the formula it proves is a type for the program. Notes Applies for subprograms (functions) too. Informally means if the code compiles, it is proved that from the input types the output type can be produced. Average response: So what? Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
4. Types are underrated Programmers don't appreciate types, partially because types they encounter are usually shallow in information. primitives, string (no semantic) void (can't do anything with it) Object (not too informative) MemberProcessorStrategy (name doesn't tell us anything) There are some refreshing exceptions. Collection types (clear concept + expectable operations) Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
5. A contrieved (?) example Object getItem(String id) Doesn't tell much detail about data or the underlying computation, so we resort to the Apidoc, if any. Apidoc throws if id is invalid syntactically returns null if no such id normally returns either Foo or Bar type by performing IO request and blocking the thread until the request is complete Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
6. Types can be expressive Object getItem(String id) Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
7. Types can be expressive Object getItem(String id) getItem: String => Any Note In Functional Programming, functions are also rst-class values and have a type signature. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
8. Types can be expressive Object getItem(String id) throws if id is invalid syntactically getItem: String => Any Note In Functional Programming, functions are also rst-class values and have a type signature. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
9. Types can be expressive Object getItem(String id) throws if id is invalid syntactically String => Validation[E, Any] Note Validation holds a successful result or a description of the errors encountered during execution of the computation. While throw/try/catch is hard to compose, functions returning a Validation compose easily. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
10. Types can be expressive Object getItem(String id) throws if id is invalid syntactically returns null if no such id String => Validation[E, Any] Note Validation holds a successful result or a description of the errors encountered during execution of the computation. While throw/try/catch is hard to compose, functions returning a Validation compose easily. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
11. Types can be expressive Object getItem(String id) throws if id is invalid syntactically returns null if no such id String => Validation[E, Option[Any]] Note Option is either None or Some(value), marking nullness explicitly (no more NPE). Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
12. Types can be expressive Object getItem(String id) throws if id is invalid syntactically returns null if no such id normally returns either Foo or Bar type String => Validation[E, Option[Any]] Note Option is either None or Some(value), marking nullness explicitly (no more NPE). Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
13. Types can be expressive Object getItem(String id) throws if id is invalid syntactically returns null if no such id normally returns either Foo or Bar type String => Validation[E, Option[Either[Foo, Bar]]] Note Either carries one of two alternatives. If more alternatives are needed, it is common to use a custom Algebraic Data Type. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
14. Types can be expressive Object getItem(String id) throws if id is invalid syntactically returns null if no such id normally returns either Foo or Bar type by performing IO request String => Validation[E, Option[Either[Foo, Bar]]] Note Either carries one of two alternatives. If more alternatives are needed, it is common to use a custom Algebraic Data Type. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
15. Types can be expressive Object getItem(String id) throws if id is invalid syntactically returns null if no such id normally returns either Foo or Bar type by performing IO request String => Validation[E, IO[Option[Either[Foo, Bar]]]] Note IO holds the execution plan of an I/O action, making it possible to defer IO actions until necessary. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
16. Types can be expressive Object getItem(String id) throws if id is invalid syntactically returns null if no such id normally returns either Foo or Bar type by performing IO request and blocking the thread until the request is complete String => Validation[E, IO[Option[Either[Foo, Bar]]]] Note IO holds the execution plan of an I/O action, making it possible to defer IO actions until necessary. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
17. Types can be expressive Object getItem(String id) throws if id is invalid syntactically returns null if no such id normally returns either Foo or Bar type by performing IO request and blocking the thread until the request is complete String => Validation[E, Future[IO[Option[Either[Foo, Bar]]]]] Note Future (alias Promise) describes a computation that is running concurrently. With it blocking threads can be avoided. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
18. FP lets you manipulate types Functional Programming has the concepts and tools to dene, use and compose types capturing both the description of data and context of computation, resulting in correct and reusable software. Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming
19. Where to now? Start studying FP. It is not just a new API to skim overnight, but a new concept that will change the way you interpret and create software. Start now. Haskell Learn You a Haskell: http://learnyouahaskell.com Real World Haskell: http://book.realworldhaskell.org Scala Functional Programming in Scala: http://www.manning.com/bjarnason Scalaz 7: https://github.com/scalaz/scalaz/tree/scalaz-seven Robin Palotai@bootcodehttp://ron.shoutboot.com A subjective on Functional Programming