Funx - Functional Programming Patterns for Elixir

⚠️ Beta notice: Funx is in active development. APIs may change until version 1.0. Feedback and contributions are welcome.

As a rule, functional libraries consider teaching functional programming as out of scope. This one does not.

Funx was built alongside Advanced Functional Programming with Elixir, which helps you build lasting mental models of functional programming—showing not just how these abstractions work, but how to apply them in real systems using Elixir.

Funx is a functional programming library for Elixir providing protocols and combinators for equality, ordering, monoids, monads, and other core abstractions.

Elixir doesn’t have a static type system, so it can’t enforce functional patterns through the compiler. Instead, it uses pattern matching, protocols, and structs to model abstractions at runtime.

Funx brings functional programming patterns into Elixir using the language’s own tools—without sacrificing its dynamic nature.

Equality

The Eq protocol defines how two values are compared, making equality explicit and adaptable to your domain.

• Define what “equal” means—compare by ID, name, or any derived attribute.
• Compose multiple comparisons—require all to match or just one.
• Implement for structs, built-in types, or custom comparators.

Ordering

The Ord protocol defines ordering relationships in a structured way, without relying on Elixir’s built-in comparison operators.

• Define comparisons based on properties like size, age, or priority.
• Chain orderings to create fallback tiebreakers.
• Implement for any type, including custom structs.

Monads

Monads encapsulate computations, allowing operations to be chained while handling concerns like optional values, failures, dependencies, or deferred effects.

• Identity: Wraps a value with no additional behavior—useful for organizing transformations.
• Maybe: Represents optional data using Just for presence and Nothing for absence.
• Either: Models computations with two possibilities—Left and Right.
• Effect: Encapsulates deferred execution with error handling, similar to Task.
• Reader: Passes an immutable environment through a computation for dependency injection or configuration.
• Writer: Threads a log alongside a result using any monoid—useful for tracing, reporting, or accumulating metadata during computation.

Monoids

Monoids combine values using an associative operation and an identity element. They are useful for accumulation, selection, and combining logic.

• Sum: Adds numbers (0 is the identity).
• Product: Multiplies numbers (1 is the identity).
• Eq.All: Values are equal only if all comparators agree.
• Eq.Any: Values are equal if any comparator agrees.
• Predicate.All: All predicates must hold.
• Predicate.Any: At least one predicate must hold.
• Ord: Defines ordering compositionally.
• Max and Min: Select the largest or smallest value by custom ordering.
• ListConcat: Concatenates lists ([] is the identity).
• StringConcat: Concatenates strings ("" is the identity).

Predicates

Predicates are functions that return true or false. Funx provides combinators for composing them cleanly.

• p_and: Returns true if both predicates pass.
• p_or: Returns true if either predicate passes.
• p_not: Negates a predicate.
• p_all: Returns true if all predicates in a list pass.
• p_any: Returns true if any predicate in a list passes.
• p_none: Returns true if none pass.

Folding

The Foldable protocol defines how to reduce a structure to a single result.

• fold_l: Reduces from the left, applying functions in order.
• fold_r: Reduces from the right, applying functions in reverse.

Useful for accumulating values, transforming collections, or extracting data.

Filtering

The Filterable protocol defines how to conditionally retain values within a context.

• guard: Keeps a value if a condition is met; otherwise returns an empty context.
• filter: Retains values that satisfy a predicate.
• filter_map: Applies a transformation and keeps results only when the transformed value is present.

Sequencing

Sequencing runs a series of monadic operations in order, combining the results.

• concat/1: Removes empty values and unwraps the present results from a list.
• concat_map/2: Applies a function to each element and collects only the present results.
• sequence/1: Converts a list of monadic values into a single monadic value containing a list. Short-circuits on the first failure or absence.
• traverse/2: Applies a function to each element and sequences the resulting monadic values.
• sequence_a/1: Applicative version of sequence—combines all and collects results.
• traverse_a/2: Applicative version of traverse—applies a function to each element and collects results.

Lifting

Lifting functions promote ordinary logic into a monadic or contextual form.

• lift_predicate/3: Wraps a value in a monad if a condition holds; returns an empty or failed context otherwise.
• lift_eq/1: Adapts an Eq comparator to work within a monadic context.
• lift_ord/1: Adapts an Ord comparator to work within a monadic context.

Interop

Funx integrates with common Elixir patterns like {:ok, value} and {:error, reason}.

• from_result/1: Converts a result tuple into a monadic context that distinguishes success from failure.
• to_result/1: Converts a monadic value back into a result tuple.
• from_try/1: Wraps a function call in a monad, capturing exceptions as failures.
• to_try!/1: Extracts the value from a monad or raises if it represents a failure.

Installation

To use Funx, add it to the list of dependencies in mix.exs:

def deps do
  [
    {:funx, "~> 0.1.5"}
  ]
end

Then, run the following command to fetch the dependencies:

mix deps.get

Documentation

Full documentation is available on GitHub Pages.

Usage Rules

Each module includes usage rules designed for humans and LLMs:

• When to use a given abstraction
• How to compose and combine it
• Examples of domain modeling with Eq, Ord, Maybe, etc.

➡️ Browse Usage Rules

Contributing

1. Fork the repository.
2. Create a new branch for the feature or bugfix (git checkout -b feature-branch).
3. Commit changes (git commit -am 'Add new feature').
4. Push the branch (git push origin feature-branch).
5. Create a pull request.

License

This project is licensed under the MIT License.